The α1a-Adrenergic Receptor Occupies Membrane Rafts with Its G Protein Effectors but Internalizes via Clathrin-coated Pits*

The α1a-adrenergic receptor (α1aAR) occupies intracellular and plasma membranes in both native and heterologous expression systems. Based on multiple independent lines of evidence, we demonstrate the α1aAR at the cell surface occupies membrane rafts but exits from rafts following stimulation. In non-detergent raft preparations, basal α1aAR is present in low density membrane rafts and colocalizes with its G protein effectors on density gradients. Raft disruption by cholesterol depletion with methyl-β-cyclodextrin eliminates these light rafts. To confirm the presence of the α1aAR in plasma membrane rafts, fluorescence resonance energy transfer measurements were used to demonstrate colocalization of surface receptor and the raft marker, cholera toxin B. This colocalization was largely lost following α1aAR stimulation with phenylephrine. Similarly, receptor stimulation causes exit of the α1aAR from light rafts within 3-10 min in contrast to the G proteins, which largely remain in light rafts. Importantly, this delayed exit of the α1aAR suggests acute receptor signaling and desensitization occur entirely within rafts. Interestingly, both confocal analysis and measurement of surface α1aAR levels indicate modest receptor internalization during the 10 min following stimulation, suggesting most of the receptor has entered non-raft plasma membrane. Nevertheless, activation does increase the rate of receptor internalization as does disruption of rafts with methyl-β-cyclodextrin, suggesting raft exit enables internalization. Confocal analysis of surface-labeled hemagglutinin-α1aAR reveals that basal and stimulated receptor occupies clathrin pits in fixed cells consistent with previous indirect evidence. The evidence presented here strongly suggests the α1aAR is a lipid raft protein under basal conditions and implies agonist-mediated signaling occurs from rafts.

Signaling cascades activated by cell surface receptors are tremendously complicated processes involving both acute and longer term signaling events. Although early studies did not generally address the membrane environment of receptors, a body of work has now demonstrated that heterogeneity exists in composition of the plasma membrane and that distinct lipid environments play many roles in receptor function (1)(2)(3). Cholesterol-rich microdomains, often referred to as lipid rafts and more recently as membrane rafts (4), have become a central facet of signaling research, as the function of many receptors and their downstream effectors are dependent on rafts (5)(6)(7)(8).
The presence of receptor and effector proteins in membrane rafts, as well as the ability of rafts to enhance receptor signaling (9), has led to the concept of a signalosome, where proteins are localized together to facilitate receptor signaling following agonist exposure (5,10). For some receptors, raft complexes appear to play an inhibitor role (9), although this need not conflict with signal facilitation if inhibition is directed at the unstimulated receptor. Activation-dependent receptor movement both into and out of membrane rafts has been reported for many G qcoupled G protein-coupled receptors (GPCR) 2 and members of the adrenergic receptor (AR) family (6,7,9). Although disruption of membrane rafts with the cholesterol chelator, methyl-␤-cyclodextrin (11,12), generally correlates with decreased receptor activity (9), a relevant exception is the ␤ 2 AR that is activated by cholesterol depletion (13,14).
Although the existence of membrane rafts is indisputable (15,16), biophysical characterization of rafts in live cells has proven to be far more difficult than in model membranes (17). This has led to the concept that cholesterol-rich rafts in live cells are small and unstable and may be organized around protein complexes (4,18). In this regard, it has been recognized for some time that actin plays an important role in raft function (19 -22), and indeed disruption of the actin cytoskeleton with cytochalasin B eliminates rafts from non-detergent raft preparations (23). The physical connection between rafts and the actin cytoskeleton is regulated by the plasma membrane-associated phospholipid, phosphatidylinositol (4,5)-bisphosphate (PIP 2 ) (3,22). As phospholipase C-␤ (PLC␤) cleaves PIP 2 following activation by G q , it is unsurprising that the G q -coupled ␣ 1a AR appears to control PIP 2 levels in caveolae (24,25) and is an important regulator of cytoskeletal architecture (26). For example, in both neonatal rat myocyte and heterologous expression systems, activation of the ␣ 1a AR also causes phosphorylation of cytoskeletal proteins (27,28) and morphological changes (29 -31).
Given the likelihood that membrane rafts play a central role in ␣ 1a AR function, it is surprising that even the primary issue of whether this receptor is a membrane raft protein is still open to question (24,32). The presence of ␣ 1a ARs in a wide variety of membrane environments, including the plasma membrane, intracellular vesicles, and perinuclear structures, complicates this issue (28,(33)(34)(35), particularly given that membrane raft isolation usually involves crude separation of lighter membrane rafts from all denser membranes. To investigate the role of membrane rafts in ␣ 1a AR function, we modified a non-detergent, neutral pH, membrane raft preparation procedure (36), and we used this protocol to demonstrate that ␣ 1a ARs occupy light rafts along with their primary G protein effectors. Localization of ␣ 1a AR to rafts in the plasma membrane was confirmed by FRET analysis between the receptor and the raft marker, cholera toxin B (CT-B). Stimulation of the ␣ 1a ARs with agonist results in a shift of the receptor out of membrane rafts, a process that appears to precede modest and gradual receptor internalization mediated by clathrin pits. The membrane raft preparation used here allows separation of the ␣ 1a AR in distinct membrane environments, providing a means of characterizing the environment of the receptor following agonist stimulation or other treatments.
Construction of YFP-and CFP-tagged ␣ 1a AR-To generate CFP-and YFP-tagged versions of the human ␣ 1a AR, an EcoRI-AgeI PCR product of the ␣ 1a gene, with a Kozak consensus sequence immediately prior to an HA epitope and with the stop codon replaced with the amino acid alanine, was generated via PCR using LA Taq DNA polymerase (Takara Mirus Bio Corp., Madison, WI) using the following primers: (forward) 5Ј-TTT-TGAATTCGCCACCATGTACCCATACGACGTCCCAG-ACTACGCCGTGTTTCTCTCGGGAAATGCTTCCGAC-3Ј and (reverse) 5Ј-CATGGTGGCGACCGGTGCGACTTCCT-CCCCGTTCTCACTGAGGGA-3Ј. The resulting gel-purified product was ligated into either the pECFP-N1 (catalog number 6900-1) or the pEYFP-N1 (catalog number 6006-1) vectors (Clontech). Vectors were a kind gift from Dr. Katharine Herrick-Davis, Albany Medical College, Albany, NY.
Cell Culture-Clonal cell lines of rat-1 fibroblasts expressing HA-␣ 1a AR (ϳ1 pmol/mg) have been described (37) as have rat-1 fibroblasts expressing HA-␣ 1a AR-EGFP (ϳ1 pmol/mg) (38). Cells were maintained in medium (10% heat-inactivated fetal bovine serum, penicillin/streptomycin, 0.5 mg/ml G418, and DMEM) at 37°C under 5% CO 2 . Two days prior to use, cells were split into medium without G418 at dilutions necessary to reach confluence at the time of use (ϳ50,000 cells/well for 12-well plates, ϳ150,000 cells/well for 6-well plates, and ϳ3 ϫ 10 6 cells for 15-cm plates). In all cases, cells were exposed to treatments under incubation conditions in medium at 37°C. 100ϫ or 1000ϫ stocks of phenylephrine (PE) in water were gently mixed into cell medium to a final concentration of 10 Ϫ5 M, whereas MCD treatments were initiated by replacing medium with MCD-containing medium at 37°C.
Crude Raft Extract Preparation-Cells in 15-cm plates were grown and treated as described above. To prepare rafts (36), medium was poured off the plates, which were placed on ice, and then washed twice with ice-cold base buffer (20 mM Tris, pH 7.5, at 25°C (pH ϳ7.8 at 4°C), 250 mM sucrose, 1 mM CaCl 2 , and 1 mM MgCl 2 ). For cell collection 1 ml of raft buffer (base buffer plus 1:500 dilution of PIC III) was added, and cells were scraped from the plates with a Teflon scraper. Scraped cells were centrifuged for 2 min at 250 ϫ g (4°C), the cell supernatant (CS) was removed, and the cells resuspended with 800 l of raft buffer. Raft extractions for no more than four samples were performed in a cold room using precooled equipment by passing the cell solution 15 times through a 1-ml syringe with a 22 gauge, 3-inch stainless steel needle (catalog number 511059, BD Biosciences). For the slow draw syringe procedure, the plunger was drawn back slowly so that about 100 -150 l of air volume was present in the syringe. For the fast draw syringe procedure, the plunger was moved immediately to the 950-l mark, and the solution was allowed to fill the syringe. In each case, samples were then centrifuged (10 min, 1000 ϫ g, 4°C), and the crude lipid raft extract was collected without disturbing the cell debris pellet. Generally, this extract was frozen with liquid nitrogen in two tubes (660 and 100 l for gel analysis) and stored at Ϫ80°C. A 100-l sample of the CS and the pellet (following resuspension with 760 l of raft buffer) were also frozen.
Density Separation of Crude Extract-Density gradients for ultracentrifugation were made from solutions of raft buffer and Optiprep as follows: 0% (0.5 ml), 5% (1 ml), 10% (1 ml), 15% (1 ml), and 20% (1 ml). Densities with 0, 10, and 20% Optiprep included 1 g/ml bromphenol blue. Gradient fractions were underlaid using a 1-ml syringe and a 3-inch needle to guarantee accurate fraction size. Immediately prior to loading, as many as four fractions were rapidly thawed without agitation in cool water until little ice remained, mixed with an equal volume of cold 50% Optiprep in raft buffer (650 l), and underlaid beneath ice-cold gradients with a syringe. Following ultracentrifugation in a swinging bucket rotor (90 min, 52,000 ϫ g, 4°C), 1-ml fractions crossing the density interfaces (as indicated by the bromphenol blue transitions) were carefully collected from the top (fraction 6 is only ϳ800 l). Aliquots (100 l and remainder) were frozen in liquid nitrogen and stored at Ϫ70 -80°C. For gel analysis 100-l samples were thawed combined with 2ϫ SDS-PAGE sample buffer with 20 mM dithiothreitol and heated 5 min at 95°C. Because the pellet was resuspended to the original volume, equal volumes of the extract and pellet were loaded, along with twice as much CS as this has about twice the volume. Because the extract was diluted 2-fold with Optiprep prior to density centrifugation, twice the volume of the gradient fractions was loaded compared with extract.
Fluorescence Resonance Energy Transfer (FRET)-COS-7 cells were grown in media at 37°C on 4-well (1 ϫ 10 5 cells/well) chambered coverglass (Lab-Tek). Cells were transfected with HA-␣ 1a AR-CFP plasmid DNA (250 ng/well) using GeneJuice (Novagen) accord to the manufacture's protocol. For control experiments, cells were transfected with pECFP-N1 vector DNA (250 ng/well). Lipid rafts were labeled with cholera toxin B (CT-B) using Vybrant lipid raft labeling kit (Molecular Probes) according to the manufacturer's protocol with minor modifications. Briefly, 24 h post-transfection, cells were labeled with cholera toxin unit B conjugated with Alexa Fluor 555 on ice for 10 min, washed three times with cold DMEM, and the CT-B-labeled lipid rafts cross-linked with anti-CT-B antibody for 10 min on ice. Cells were washed with cold DMEM three times. The cold-labeled cells were used immediately for FRET as zero time points or treated with pre-warmed medium with or without 10 Ϫ5 M PE for 20 min at 37°C before FRET. FRET was measured by acceptor photobleaching (39,40), with the following modifications. Photobleaches were done using Zeiss LSM 5 Live confocal imaging system with 40 ϫ 1.4NA objective. CFP and CT-B Alexa Fluor 555 were excited with the 440 nm line of a Diode laser and the 532 nm line of DPSS laser, respectively. Because Zeiss LSM 5 Live confocal imaging system is a designed line scanning model, a stripe with 512 ϫ 23 pixels was positioned across the field over the cells of interest. To obtain a prebleach value, CFP and CT-B Alexa Fluor 555 were collected twice simultaneously at low laser intensity (2%) using a 532 dichroic filter to split the beam, a 470/500 band-pass filter to collect CFP emission, and a 550 long pass filter to collect Alexa Fluor 555 emission, and then the cells were irradiated with the 532-nm DPSS laser at 100% intensity for 50 iterations. Following the bleaching, a single image of CFP and Alexa Fluor 555 was collected simultaneously at low laser intensity (2%) to obtain a post-bleach value. The interval of image collection was 0.5 s for both before and post-bleaching. Thanks to the high scanning speed of Zeiss LSM 5 Live confocal imaging system, the whole FRET process (pre-bleach image collection, 50-iteration bleach and post-bleach image collection) could be finished in less than 3 s, thus minimized the bleaching of the cells and the migration of the region of interest. FRET on the membrane region was measured as an increase in CFP fluorescence intensity following CT-B Alexa Fluor 555 photobleaching. FRET efficiency was calculated as 100 ϫ ((CFP post-bleach Ϫ CFP pre-bleach )/CFP pre-bleach ).
Western Blotting-Following SDS-PAGE using a Bio-Rad Criterion system, gels were equilibrated with Tris, glycine, 10% methanol for 5 min and transferred to polyvinylidene difluoride. Blots were blocked for 30 min with 10% dry milk in TBS-Tw (20 mM Tris, pH 8, 150 mM NaCl, and 0.1% Tween 20). All antibodies were incubated with blots for 1 h with the exception of 3F10-HRP, which was incubated for 2 h. Blots were individually washed four times with 40 -50 ml of TBS-Tw over 20 min. Binding was detected using Supersignal WestDura substrate (Pierce). Blots were reprobed many times following 5 min of room temperature treatment with Restore Western blot Stripping Buffer (Pierce) and re-equilibration with TBS-Tw. For presentation purposes, images may have had background subtraction and linear adjustments to scale intensity; however, images on a single raft preparation were adjusted simultaneously (for each antibody) to maintain signal intensity relationships. Quantification of relative band intensity for crossover analysis of the ␣ 1a AR was obtained from digitalized film images analyzed with ImageQuant. Primary antibody dilutions were as follows: 3F10-HRP (1:1000), G q (1:500), G␤ 1  Surface Receptor Measurement-Rat-1 cells expressing HA-␣ 1a AR and plated in 12-well plates were treated as indicated in medium at 37°C, prior to fixation for 10 min at room temperature with 3.7% formaldehyde in phosphate-buffered saline (PBS). Remaining steps were also done at room temperature. After two washes with PBS, 800 l of blocking buffer containing 5% nonfat dry milk in PBS was added to each well. For determination of nonspecific background, 3 l of 200 g/ml rat monoclonal 3F10 antibody was then added (final concentration, 75 ng/ml). After blocking for 30 min, anti-HA 3F10-HRP (stock concentration, 25 units/ml) was added to a final concentration of 0.125 units/ml. Following one wash with blocking buffer and two washes with PBS, cells were incubated for 1 h with 0.8 ml of ABTS solution (Roche Diagnostics). Thereafter the solution was transferred from each well to 48-well plates, and absorbance values were read at 405 nm. Each point was assayed in duplicate or triplicate. Specific absorbance values were obtained by subtracting nonspecific absorbance from total absorbance.
Transferrin Internalization Assay-Rat-1 cells expressing HA-␣ 1a AR plated in 6-well plates were treated with the indicated concentrations of MCD for 30 min in 800 l of medium at 37°C. 125 I-Tnf (ϳ0.6 Ci) was added to a final concentration of 10 nM and gently mixed prior to incubation at 37°C for 30 min. Internalization was terminated by removal of medium and placement on ice/water followed by three washes with cold Tnf wash solution (20 mM acetic acid and 500 mM NaCl), which also removes surface-bound Tnf (41). Cells were solubilized with 800 l of room temperature 1% SDS with 50 mM Tris, pH 8, for Ͼ5 min and transferred to tubes for gamma counting. Each point was assayed as a singlet. Background was determined by adding 8 l of cold 200 M Tnf to each condition. Internalization values were then obtained by subtracting nonspecific binding from each experimental value.
Confocal Microscopy-Rat-1 cells expressing HA-␣ 1a AR or HA-␣ 1a -EGFP were grown on 4-well (1 ϫ 10 5 cells/well) or 8-well (5 ϫ 10 4 cells/well) chambered cover glasses (Lab-Tek). Cells were transfected with EGFP-human clathrin light chain A (for HA-␣ 1a AR) or mRFP-human clathrin light chain A (for HA-␣1aEGFP) using GeneJuice (Novagen) according to the manufacturer's protocol. EGFP-clathrin (42) and mRFP-clathrin (43) were obtained from Dr. Lois E Greene, National Institutes of Health, Bethesda. 24 h post-transfection, cells were pre-labeled with 3F10 (1:200 dilution) at 4°C for 15 min in DMEM. Cells were washed with ice-cold DMEM and immediately fixed or returned to pre-warmed medium for 60 min at 37°C. After fixation for 10 min with 3.7% formaldehyde in PBS, cells were permeabilized in the same solution with 0.1% Triton X-100 for 10 min. Secondary antibody (goat anti-rat IgG Alexa Fluor 647) was applied at a 1:500 dilution for 30 min in PBS with 10% fetal bovine serum plus 0.05% Triton X-100. Cells were washed with PBS and used for confocal imaging.
All imaging was performed using a Zeiss LSM 5 Live confocal imaging system with a ϫ63, 1.45 NA oil-immersion objective. EGFP, mRFP, and Alexa Fluor 647 were excited with the 488 nm line of a Diode laser, the 532 nm line of a DPSS laser, and 635 nm line of a Diode laser, respectively. Green, red, and blue emissions were collected simultaneously using a 635 dichroic filter to spilt the beam, a 525/20 bandpass filter to collect green emission, a 615/550 bandpass filter to collect red emission, or a 650 long pass filter to collect blue emission. Images were generally collected as Z-stacks with 0.5-m thick slices. Images represent individual Z-slices from above the basal membrane (which is next to the coverslip) but still close enough to the slide that the full width of cell remained in focus. Images were processed with Adobe Photoshop software.

Substantial Amounts of the ␣ 1a AR Are Present in Membrane
Rafts-Recently, Macdonald and Pike (36) have developed a simple, fast, neutral pH method of creating non-detergent membrane rafts that appears to allow consistent distribution of raft and non-raft membrane proteins into distinct structures of varying density. The membrane raft locations of an amino-terminal, HA-tagged version of the ␣ 1a AR stably expressed in rat-1 fibroblasts were investigated by adapting this procedure to allow analysis of receptor membrane environment following agonist stimulation and other treatments. In this method (Fig. 1A), cells are scraped and collected releasing most cytosolic proteins into the CS, followed by syringe shearing and subsequent low speed centrifugation to remove unextracted cellular debris as a pellet and producing a crude raft extract. When cells were extracted using a slow draw of the syringe plunger (Fig. 1B), a fraction of the ␣ 1a AR (ϳ20 -25%) was released from the scraped cells (Fig. 1B, compare Pel with Ext). When these extracts were separated by density centrifugation with an Optiprep step gradient (Fig. 1B), the highest concentrations of the ␣ 1a AR were present in fraction 2 representing the 5-10% Optiprep interface, with a lower maxima sometimes occurring in fraction 4 at the 15-20% interface. As has been observed during gel analysis of the ␣ 1a ARs (44), both dimers and under-glycosylated forms are present, although the glycosylated monomer predominates in our system.
Adaptation of the Lipid Raft Protocol to Gain Reproducibility-Early in this project it became clear that the distribution of ␣ 1a ARs between light and moderate density fractions varied somewhat in experiments done in different weeks. Given our ultimate goals it was essential to compare treatment conditions with a protocol that produced reproducible results for each condition. To obtain near quantitative reproducibility, we found it necessary to freeze crude raft extracts so that many treatments could be performed on the same day and the raft extracts subjected to density centrifugation at a later date. Concurrent replicate experiments in which one extract was frozen FIGURE 1. Membrane raft procedure and ␣ 1a AR localization in rafts. A, after the cells are scraped off the plates and pelleted, the CS, which contains most cytosolic proteins and some cell debris, is removed. Following resuspension the cells are subjected to shearing using multiple syringe strokes. Heavier and unextracted membrane structures are pelleted (Pel), and the supernatant is recovered as the crude lipid raft extract (Ext), which was generally frozen for later use. This extract is made dense with Optiprep and underlaid beneath a five-step Optiprep density gradient. Ultracentrifugation then causes the lighter lipid rafts to float upward in the gradient after which fractions are collected from the top down. B and C, Western blot analysis of non-detergent membrane rafts show the ␣ 1a AR is present in light and putative moderate density rafts. Cells were subjected to shearing using a slow (B) and fast (C) syringe draw (see "Experimental Procedures"). Cell supernatants (CS), unextracted pellets (Pel), and crude raft extracts (Ext) are shown along with the density gradient fractions. Marker lanes are indicated by m. Loads allow direct comparison of fractions (see "Experimental Procedures"). HA-␣ 1a AR detected using 3F10-HRP reveals bands representing monomeric and dimeric receptors some of which is underglycosylated (indicated by u). Results are representative of the slow draw (n ϭ 11) and fast draw (n ϭ 7) procedures. and one was not indicate the distribution of the ␣ 1a AR has not been significantly altered by freeze/thawing (supplemental Fig.  s1). Freezing crude extracts prior to density separation has been incorporated in all experiments.
Extraction Efficiency Is Controlled by Shearing Pressure-Also while adapting the protocol, it became clear that the force used in the syringe shearing was a critical variable not only in determining the quantity of ␣ 1a AR released into the lipid raft extract but also in the distribution of ␣ 1a AR between light (fraction 2) and moderate density fractions (fraction 4). When the extraction was performed using a fast draw of the syringe (see "Experimental Procedures"), as much as 50% of the ␣ 1a ARs now entered the crude raft extract (Fig. 1C, compare Pel with Ext), resulting in a more even distribution of receptor in fractions 2-4 ( Fig. 1C and supplemental Fig. s2, B and C). Although membrane structures that float into the gradient may represent non-raft membranes (36), the lightest vesicles (fraction 2) have a density and protein composition (see below) consistent with cholesterol-rich rafts (36). Despite the extraction efficiency of the fast draw, most experiments reported here use the slower draw as the light rafts produced are purer and more similar to high pH carbonate raft preparations (36).
Cholesterol Depletion Disrupts ␣ 1a AR-containing Membrane Structures-To further investigate the nature of the ␣ 1a ARcontaining membranes, lipid rafts were prepared using the slow draw from cells treated with the cholesterol chelator, MCD (11,12). As shown in Fig. 2A, increasing MCD concentrations and treatment times caused a shift of the ␣ 1a AR out of lightest gradient fractions. The presence of the receptor in the moderate density vesicles ( Fig. 2A, fraction 4) was more resistant to cholesterol depletion; however, two behaviors suggest these vesicles may also represent a raft environment. First, the ␣ 1a AR begins to appear predominantly in fraction 5, which represents the interface between the load (25% Optiprep) and the first gradient step (20% Optiprep) rather than fraction 4. Second, the shift does not appear to be a transfer as much as a faster loss from the light rafts and a slower loss from the moderate density structures. To conclusively demonstrate that ␣ 1a AR was being lost from the raft extracts, we used crossover analysis to compare the amount of ␣ 1a AR in the extract to that remaining in the unextracted pellet). As is visually evident and confirmed by quantitation (Fig. 2B), increased cholesterol depletion results in less receptor extracted into the crude raft extract. Because cholesterol depletion can almost eliminate moderate density vesicles (fraction 4) containing the ␣ 1a AR, these vesicles may also be derived from membrane rafts. Thus to distinguish these two potential raft environments, the low density membrane structures (fraction 2) will be referred to as light rafts.
FRET Analysis Demonstrates the ␣ 1a AR Occupies Plasma Membrane Rafts and Exits upon Stimulation-To confirm that the ␣ 1a AR is in a raft environment at the plasma membrane, we used photobleaching-based FRET measurements between CFP-tagged ␣ 1a AR and the lipid raft marker, CT-B (45). Because rat-1 cells failed to bind appreciable CT-B, we used COS-7 cells transiently transfected with the HA-␣ 1a AR-CFP. CT-B was loaded onto the surface of these cells at 4°C to limit internalization and the cells immediately subjected to bleaching (Fig. 3A, large box). Focusing on peripheral surface regions (Fig. 3A, small box), we observed average energy transfer of about 12% between the CT-B and the plasma membrane-associated unstimulated ␣ 1a -CFP (Fig. 3B). Placing the cells at 37°C for 20 min, which allows some CT-B internalization, does not significantly impact energy transfer at the plasma membrane. In contrast, stimulation of the ␣ 1a AR with PE during this 20-min incubation causes a 70% decrease in energy transfer, strongly suggesting the receptor has moved away from the CT-B-associated raft environment (Fig. 3B). To appreciate the dramatic change in FRET efficiency that has occurred upon receptor stimulation, it is useful to compare the transfer efficiencies from individual cell measurements with and without agonist stimulation (Fig. 3C). Many cells with unstimulated receptors FIGURE 2. Cholesterol depletion causes loss of both light and putative moderate density rafts containing the ␣ 1a AR. A, cells growing in medium at 37°C were treated for the indicated times with various concentrations of MCD, prior to harvesting for crude raft extract preparation. Western analysis of the gradient fractions reveals increasing loss of the HA-␣ 1a AR with increasing cholesterol extraction. B, relative quantitation of the HA-␣ 1a AR in the crude raft extract (Ext) compared with unextracted receptor remaining in the pellet (Pel). The highest pellet concentration and the duplicated extract lanes contain 0.56% of each sample. Relative quantitation of HA-␣ 1a AR in the extract is indicated on the right and was obtained by interpolating between the appropriate higher and lower pellet values. FEBRUARY 1, 2008 • VOLUME 283 • NUMBER 5 display high FRET values (Ͼ12%) and almost none of these cells display near zero transfer efficiency (Fig. 3C, solid and hatched  bars). Directly contrary to this, few cells with stimulated receptor display high FRET efficiency although many of these cells display no energy transfer (Fig, 3C, open bars). This loss of energy transfer is strong evidence that colocalization of the ␣ 1a AR and CT-B-labeled rafts occurs on the plasma membrane prior to receptor activation.

The ␣ 1a -Adrenergic Receptor Is a Membrane Raft Protein
Identification of Proteins in the Basal ␣ 1a AR Raft Environment-Raft preparations were probed with antibodies to proteins associated with the ␣ 1a AR function, membrane rafts, or other membrane structures (Fig. 4). Colocalization on a gradient is poor evidence that two proteins occupy the same membrane environment. However, given the differences in the localization of the ␣ 1a AR with the slow and fast draws, proteins that colocalize with the ␣ 1a AR under both conditions are very likely to be present in a similar membrane environment.
Importantly, vesicles containing G q showed strikingly similar densities to those containing the glycosylated ␣ 1a AR monomer. Colocalization occurred with the slow draw procedure, where both receptor and G q were concentrated in light rafts (Fig. 4A, fraction 2, and supplemental Fig. s2A), and with the fast draw, where a more even distribution of these proteins was observed (Fig. 4B, fractions 2-4 and supplemental Fig. s2, B and C). The G proteins ␤ subunits, G␤ 1 and G␤ 2 (Fig. 4) as well as G␥ 5 (data not shown), also displayed similar distributions. These data strongly suggest the unstimulated ␣ 1a AR and its trimeric G protein effectors occupy a similar set of membrane environments, which include light and perhaps putative moderate density rafts.
Given the implication that colocalization could enhance receptor/G protein interaction, we wished to determine whether the G q effector, PLC␤, was also present, potentially providing all of the proteins necessary for cleavage of PIP 2 into diacylglycerol and inositol triphosphate. Surprisingly, Western analysis showed that PLC␤3 was present in load fractions (Fig.  4, slow and fast), consistent with a lack of movement during high speed centrifugation. Although this could indicate PLC␤3 is associated with dense non-raft plasma membranes, it is unlikely such plasma membranes would have exactly the same density as the 25% Optiprep load (as implied by the lack of movement) and more likely PLC␤3 simply disassociated from membrane structures during the extraction procedure. Similar results for PLC␤4 and PLC⑀, the other potentially relevant lipases (46) expressed at the mRNA level in rat-1 fibroblasts, 3 suggested even less stable membrane association (data not shown).
Based on indirect evidence, it has been suggested that ␣ 1a AR internalizes via clathrin-coated pits (28,34). Western analysis of the clathrin heavy chain and ␤-adaptin showed little colocalization with the ␣ 1a AR (Fig. 4, slow and fast). However, these proteins displayed a similar pattern to that of PLC␤3, again suggesting dissociation from membrane structures.
Interestingly, the raft marker, flotillin 2, displayed a profile distinctly different from the ␣ 1a AR and G proteins, with a max-3 B. Lei, unpublished observations. ima in fraction 3 (Fig. 4, slow and fast). The similarity of the flotillin 2 distribution with a slow or fast draw further demonstrates the membrane environment of this protein is different from that of the ␣ 1a AR. Although caveolin 1 showed some colocalization with the ␣ 1a AR, clear distinctions between the proteins distributions suggest limited ␣ 1a AR occupancy of caveolae. Specifically, almost no caveolin is extracted with the slow draw in contrast to a significant portion of ␣ 1a AR and G proteins (Fig. 4A, compare Pel and Ext). In addition, with the fast draw caveolin 1 consistently shows a maxima in moderate density fraction 4 ( Fig. 4B and supplemental Fig. s2, B and C) compared with more even distribution of the receptor and G proteins. Both of these observations suggest ␣ 1a AR occupancy of caveolae may be modest.
Markers for the nucleus, ER and Golgi suggest light rafts containing the ␣ 1a AR were not derived from these organelles (Fig. 4), consistent with identification of light rafts as plasma membrane proteins. Relative to the ␣ 1a AR and G proteins, the ER marker calnexin was poorly extracted and largely excluded from light rafts even with the faster draw (Fig. 4B, fraction 2). Note that the calnexin patterns produced by the slow and fast draw were different. It is probably not a coincidence that each pattern roughly parallels the distribution of the unglycosylated ␣ 1a AR monomer, as would be expected if unglycosylated receptor represents newly synthesized protein still in the ER. Unglycosylated ␣ 1a AR dimers also display very poor extraction and could be observed only in fraction 6 with long exposure times (Figs. 1C and 4B).
The ␣ 1a AR Exits Light Rafts upon Stimulation-As agonist activation of GPCRs has been shown to cause receptor movement both into and out of membrane rafts (6,7,9), we wished to determine whether stimulation induced ␣ 1a AR movement. Because ␣ 1a ARs and the G proteins were highly enriched in the lightest rafts, which are most similar to high pH carbonate raft preparations (36), we adopted a slow draw extraction as light rafts can more easily be analyzed with this procedure. To test the response of the ␣ 1a AR to agonist stimulation, receptor-containing cells were treated with PE for times from 0 to 90 min prior to extraction of the lipid rafts (Fig.  5A). This stimulation causes an obvious shift of receptor from light lipid rafts (fraction 2) into moderate density vesicles (fraction 4). Movement of the ␣ 1a AR occurs after 3 min  FEBRUARY 1, 2008 • VOLUME 283 • NUMBER 5 and involves most of the ␣ 1a AR present in light rafts, strongly suggesting that most of these receptors have been activated by agonist. Despite variability in the distribution of unstimulated ␣ 1a AR between light and moderate density membranes, the shift of ␣ 1a AR upon stimulation was qualitatively reproducible across many experiments (n ϭ 10). In contrast, G q displayed either modest or little shift out of light density rafts (supplemental Fig. s3). Although these slight movements of G q could be functionally important, we also find it significant that most G q did not shift with the ␣ 1a AR. The separation of the ␣ 1a AR from G q following stimulation suggests that ␣ 1a AR-containing fraction 4 membranes in stimulated raft preparations (Fig. 5A, 10 -90 min) are unlikely to represent the putative ␣ 1a AR containing fraction 4 rafts present in unstimulated cells (Fig. 5A, 0 min). The presence of receptor in fraction 5 also rises, potentially suggesting movement of ␣ 1a AR into denser plasma membranes.

The ␣ 1a -Adrenergic Receptor Is a Membrane Raft Protein
Exit of the Stimulated ␣ 1a AR from Light Rafts Does Not Lead to Immediate Internalization-Given the similar time frames, it was possible that exit from light rafts was directly related to ␣ 1a AR internalization. As the half-time of agonist-induced ␣ 1a AR internalization can be variable (data not shown), matched surface assays and raft preparations were performed. To quantitate the shift of ␣ 1a AR from light rafts to moderate density vesicles, crossover analysis was used to compare the amount of receptor in light rafts (fraction 2) to that in moderate density vesicles in fraction 4 (Fig. 5B). Although these experiments showed only modest enrichment of unstimulated ␣ 1a AR in light rafts (about 1.13-fold or 13% more receptor in fraction 2 than 4), the shift out of light rafts occurred in about the same time frame as the experiments in Fig. 5A. Clearly exit from the light rafts appears to be maximal within 10 min of ␣ 1a AR stimulation (Fig. 5, A and B). Parallel assays of surface receptor levels, performed using a horseradish peroxidase (HRP) conjugate of 3F10 against the surface-exposed HA tag of HA-␣ 1a AR, reveal receptor internalization to be a slower process (Fig. 5C). This suggests that exit from rafts precedes agonist induced internalization and that exiting receptors may move into bulk plasma membrane as least temporarily.
Effect of Cholesterol Depletion upon ␣ 1a AR Internalization-If membrane rafts are playing a central role in ␣ 1a AR behavior, cholesterol depletion should alter receptor behaviors such as signaling and trafficking. Although not the focus here, we do find that 5 mM MCD decreases maximal ␣ 1a AR-mediated inositol triphosphate production, 3,4 in agreement with a recent publication (24). Trafficking of the ␣ 1a AR is complex and involves both constitutive and agonist-mediated internalization with available evidence suggesting both processes are mediated by clathrin-coated pits (28,34). Unexpectedly, analysis of basal surface receptor density shows that MCD treatment by itself causes a time-and concentration-dependent loss of unstimulated ␣ 1a AR from the cell surface (Fig. 6A). An artifactual explanation for decreasing surface receptor could have been cell loss; however, concentrations of MCD at or below 10 mM caused no cell detachment (supplemental Fig. s4). To test the effect of cholesterol depletion on agonist-mediated internalization, untreated and MCD-treated cells were exposed to PE for 1 h and assayed for surface receptor (Fig. 6B). As observed previously, PE causes a modest loss of surface ␣ 1a AR coincidentally similar to the decrease in surface receptor density caused by 5 mM MCD. Importantly, cholesterol depletion did not block PE-mediated receptor internalization in this cell line (Fig. 6B) or in two additional cell lines stably expressing the ␣ 1a AR (data not shown). The effects of MCD on internalization of both unstimulated and agonist-stimulated receptor strongly suggests lipid rafts are not mediating ␣ 1a AR internalization. The ability of MCD to increase internalization was potentially confusing as cholesterol depletion can also inhibit clathrin-mediated internalization (16). To demonstrate that clathrin-mediated pits are still functioning in rat-1 cells in the presence of MCD, we looked at the classic, clathrin-mediated internalization of Tnf. Not only did Tnf internalization continue with cho-4 G. A. Michelotti, unpublished observations. FIGURE 5. Agonist stimulation of the ␣ 1a AR causes a shift of receptor from light rafts to moderate density vesicles. A, cells growing in medium at 37°C were treated for the indicated times with 10 Ϫ5 M PE, prior to preparation of crude raft extracts. Western analysis shows the distribution of HA-␣ 1a ARs shifting from light (fraction 2) to moderate (fraction 4) density membranes. Alternating time points are from independent experiments performed on the same day. B, quantitation of relative HA-␣ 1a AR density in light rafts (fraction 2) versus moderate density membranes (fraction 4) expressed as a ratio (f2/f4) following treatment with 10 Ϫ5 M PE for various times. Relative concentrations were determined by crossover analysis as in Fig. 2 except that fraction 4 was serially diluted so that samples of fraction 2 loaded in duplicate were within the range of the dilution series. Raft extracts were prepared from four independent time courses performed over 2 days. C, surface HA-␣ 1a AR levels were assayed using 3F10-HRP binding to exposed surface receptor following treatment with 10 Ϫ5 M PE for various times. Two of 4 assays were done in parallel with the raft preparations above and two on neighboring days. Values expressed as mean Ϯ S.E. (n ϭ 4).
lesterol depletion, but it displayed modest MCD dose-dependent enhancement (Fig. 6C), demonstrating at minimum that clathrin pit-mediated internalization can still occur.
Confocal Microscopy Localizes the ␣ 1a AR to Clathrin Pits in Fixed Cells, but Internalization Is Hard to Observe in Living Cells-Given that basal receptor occupies membrane rafts despite indirect evidence that the ␣ 1a AR constitutively internalizes via clathrin-coated pits, we wished to visually demonstrate the receptor was using this pathway. Attempts to employ confocal microscopy to show colocalization of ␣ 1a ARs and clathrin in living cells were fairly unsuccessful despite multiple strategic approaches. These included use of either HA-␣ 1a AR (with BODIPY/prazosin labeling) or HA-␣ 1a AR-EGFP under basal and stimulated conditions in cells cotransfected with clathrin-GFP or clathrin-RFP and exposed to various temperature and sucrose internalization blocks. In some cases, the colocalization could be observed with or without agonist stimula-tion, but only in a few clathrin pits in a fraction of the live cells (data not shown). These results may not be too surprising as agonist-induced receptor internalization is not only slow (Fig.  5C) but is difficult to observe in fixed cells (28) and difficult to observe by time-lapse photography in living cells (supplemental Fig. s5 and movie, 0 -10-min PE stimulation.). As an alternative strategy, designed to increase the relative signal from internalizing receptor, we used surface labeling of HA-␣ 1a AR with antibody (3F10) followed by incubation in medium at 37°C to allow internalization. As shown previously, antibody labeling at 4°C results in surface staining of the ␣ 1a AR (Fig. 7A, panel i) with no apparent staining of intracellular receptor (28). As expected, this surface staining displays no colocalization with clathrincoated pits made visible with transiently transfected GFPclathrin LC (Fig. 7A, panels ii and iii). In contrast, incubation in medium at 37°C allows the antibody-receptor complexes to enter the cell (Fig. 7B, panel i) resulting in intracellular punctates, many of which partially or fully colocalize with clathrin pits (Fig. 7, B, panel iii and C, panels i-iii). Similar colocalization is observed for HA-␣ 1a AR-EGFP and RFP-clathrin (supplemental Fig. s6). In addition, colocalization of receptor and pits is also observed following agonist stimulation and appears qualitatively the same as in unstimulated cells (data not shown).

DISCUSSION
A current problem in the study of membrane raft proteins is the rudimentary nature of available techniques, including raft preparation (9,17,36). Detergent-based membrane raft preparations alter raft protein and lipid composition (7) and are probably based on an unnatural aggregation phenomenon (47). For these reasons, non-detergent membrane raft preparations are believed to better reflect in vivo conditions (7). The most commonly used non-detergent membrane raft preparations are derived from a high pH (ϳ11) NaCO 3 extraction that removes peripheral membrane proteins (48,49). These protocols use various combinations of homogenization, sonication, and French press-like syringe shearing to break up the membranes; however, these processes can produce inconsistent preparations (36). Furthermore, extracts are commonly separated on a discontinuous 5:35% sucrose step gradient (48,49), which concentrates membrane raft proteins at the interface hiding preparation inconsistency. We presume that the separation that many laboratories observe in fractions around this interface reflects a gradient that is formed when pipetting the 5% sucrose fractions directly onto the denser lower fraction, a process that inevitably results in some mixing.
To investigate the role of lipid rafts in ␣ 1a AR signaling and trafficking, it was essential to have a method that allowed consistent distribution of proteins so that multiple conditions could be compared. To limit gradient nonreproducibility, we modified the linear Optiprep gradient used by Macdonald and Pike (36) to a five-step gradient with fraction collection across density interfaces (see Fig. 1). Even so, it was necessary to freeze the lipid raft extracts prepared in multipoint experiments and perform density centrifugation on a subsequent day to achieve sufficient reproducibility for each treatment condition to allow comparison of multiple conditions. Using this modified protocol, we show that the ␣ 1a AR occupies light rafts and that the receptor exits these rafts following stimulation. Importantly, FRET analysis provides independent conformation that the ␣ 1a AR associates with membrane rafts at the cell surface. The exit of ␣ 1a ARs from CT-B-bound rafts following agonist stimulation provides a clear demonstration that surface receptor is present in rafts prior to stimulation. Although FRET efficiency did not fall to zero following stimulation, the absence of cells displaying high transfer efficiency (Ͼ12%) and the gain of cells with near zero transfer efficiency suggest most receptors have left surface membrane rafts.
The challenge of localizing the ␣ 1a AR to membrane rafts is complicated by the presence of substantial intracellular pools of receptor in rat-1 fibroblasts and a variety of other cell types (28,(33)(34)(35). Although it is difficult to quantitate the relative amount of intracellular ␣ 1a AR in rat-1 cells; confocal analysis suggests less than half of stably expressed HA-␣ 1a AR-EGFP (28) or HA-␣ 1a AR 3 occupies intracellular vesicles. To some extent, proteins in these intracellular membranes are extracted and do enter the raft gradient as exemplified by the ER protein cal-nexin. However, extraction of calnexin is poor even with the hard draw, and the protein is excluded from the light rafts. Even so, the presence of membranes from many sources in moderate density fractions (36) precludes identification of the moderate density membranes containing basal ␣ 1a AR as rafts. Nevertheless, loss of most of these membranes upon MCD treatment and colocalization of the ␣ 1a AR with the lipid raft-associated protein G q suggests these structures are probably membrane rafts although they may be from internal sources. In any case, internal structures are poorly extracted by the relatively mild syringe extractions and thus it is unsurprising that most intracellular ␣ 1a AR, including unglycosylated forms, would remain with the pellet.
Two previous reports examining the location of native ␣ 1 ARs in lipid rafts appear contradictory despite the fact that both studies used whole rat heart and are based on a similar application of the high pH carbonate extraction procedure. In these studies Cav3, G q , and PLC␤1 were in the light lipid raft fractions of the 5:35:45% sucrose step gradient. However, Fugita et al. (32) observed ␣ 1a ARs and ␣ 1b ARs to be enriched at FIGURE 7. Direct evidence that ␣ 1a AR internalizes through clathrin-coated pits. Rat-1 cells stably expressing the HA-␣ 1a AR were transiently transfected with EGFP-tagged clathrin light chain to identify clathrin pits. Living cells were surface-labeled with 3F10 at 4°C (A) or returned to medium for 1 h at 37°C (B and C) prior to fixation. In both cases cells were then permeabilized and the 3F10 labeled with Alexa Fluor 647-conjugated goat anti-rat IgG secondary (displayed as red). Composite images (panel iii) accent the lack of receptor colocalization with clathrin pits in surface-labeled cells (A) and the presence of receptor in pits in receptor-internalized cells (B and C). Probable clusters of clathrin pits containing the HA-␣ 1a AR (B, panels i-iii) are indicated (arrowheads), as are individual pits in the enlarged boxed area (C, panels i-iii). the 5:35% interface, whereas Lanzafame et al. (24) observed ␣ 1 ARs spread across the gradient. Importantly, a recent study has demonstrated that EGFP-tagged ␣ 1a ARs transfected into neonatal cardiomyocytes are predominantly intracellular (35). Consequently, most ␣ 1a ARs occupy membrane structures that are expected to remain in denser fractions, consistent with the behavior observed by Lanzafame et al. (24). The probable explanation for the absence of denser membrane structures in the work by Fujita 5 is the use of a low speed centrifugation step (500 ϫ g for 10 min) prior to loading of the lysate (extract) on the density gradient. Having removed the denser structures (as we have), Fujita observed a predominant lipid raft localization for the ␣ 1 ARs. We believe the available evidence is internally consistent and points to ␣ 1a ARs occupying plasma membrane rafts as well as denser intracellular membranes in both rat-1 and neonatal cardiomyocyte models. Another study that identified a putative ␣ 1a AR band only in the densest non-raft fractions (50) is probably a misinterpretation, as the antibody reportedly used (sc-1475 from Santa Cruz Biotechnology) is insensitive and directed against the ␣ 1d AR (early on the ␣ 1d AR was called the ␣ 1a/d AR).
A considerable number of reports suggests lipid rafts play a central role in the predominant ␣ 1a AR signaling cascade, namely the G q /PLC␤-mediated cleavage of PIP 2 into diacylglycerol and inositol triphosphate. This evidence includes the lipid raft localization of G q (51), PLC␤ (24,32), and G q -coupled GPCR family members (6), as well as data suggesting that localized pools of PIP 2 in lipid rafts are targeted by PLC␤ within minutes of activation (25,52,53), although the methodology of the earliest report (53) has been challenged (47). Although these experiments have been interpreted to suggest G q /PLC␤ mediated depletion of caveolae-associated pools of PIP 2 , evidence suggesting a single homogeneously distributed pool of PIP 2 (47) combined with the absence of a mechanism to prevent rapid PIP 2 redistribution (54) may suggest a decrease in caveolar PIP 2 -binding proteins (22,54,55) is the more likely explanation. Unfortunately, our ability to determine whether the ␣ 1a AR is present in caveolae is compromised by the localization of both caveolin 1 and ␣ 1a ARs in a moderate density fraction (fraction 4) also occupied by other types of membranes.
The fact that ␣ 1a ARs shift out of light rafts following agonist stimulation is strong evidence that these receptors have been activated. This behavior is not limited to rat-1 cells, as a similar shift of ␣ 1a AR was observed following stimulation of transiently expressed receptor introduced into COS-7 cells with an adenovirus vector. 4 The time frame of receptor exit from the light rafts is much slower than the acute signaling response of the ␣ 1a AR. Indeed, evidence suggests G q -coupled receptors induce PIP 2 cleavage (47) and calcium release (56,57) within seconds, whereas maximal rates of IP signaling from activated ␣ 1a ARs conclude in less than 1 min (38). Similar short time frames are observed for agonist-mediated ␣ 1a AR desensitization and phosphorylation (38,57). Consequently, the 3-10-min time frame of ␣ 1a AR exit from light rafts implies acute signaling events, including desensitization, have occurred within the raft environment. Delayed exit from rafts is not unprecedented as the tyrosine kinase, epidermal growth factor receptor (EGFR), has been shown to exit rafts at least 2 min after stimulation (58). A role for lipid rafts in enabling ␣ 1a AR signaling is also implied by evidence that lipid raft disruption inhibits maximal receptor response (52). Furthermore, the presence of G q and the other G proteins in light rafts provides the essential colocalization of effectors and receptor that is required for signaling. Although our data showing minimal raft exit by G q following ␣ 1a AR stimulation agree with published suggestions that PIP 2 cleavage is enhanced by colocalization of G q and PLC␤ in rafts, we consider it possible that the small fraction of activated GTP-bound G q leaves the raft environment, whereas most G q remains associated with rafts. Given that we did not observe PLC␤ in membrane rafts, our data provide little guidance on the location of activated PLC␤.
Indirect evidence is available suggesting both constitutive and agonist-mediated ␣ 1a AR internalization occur via the common GPCR internalization pathway involving GRK2-mediated receptor phosphorylation, ␤arrestin binding of the phosphorylated receptor, and internalization of the complex through clathrin-coated pits (28,34,38). Thus it was initially surprising that we were unable to detect clear colocalization of the ␣ 1a AR and clathrin in living cells, as this indicates only a small percentage of receptors could possibly be in clathrin pits at any one time. However, low clathrin pit occupancy is in fact consistent with slow constitutive and agonist-mediated internalization. In contrast to the ␣ 1b AR, which moves from the surface to predominantly internal distribution with a half-time of ϳ5 min (28,59,60), agonist treatment of the ␣ 1a AR causes little obvious redistribution by confocal analysis and only a 25-50% loss in surface receptor with a half-time of 15-30 min (28,38). Hypothetically, even if only 1% of ␣ 1a ARs reside in clathrin pits and the cycle time of pits is Ͻ1 min, then more than half of the surface receptor could internalize in 1 h, consistent with the observed rates of internalization.
We have provided clear evidence that unstimulated, surfaceassociated ␣ 1a AR is in membrane rafts. Nevertheless, during constitutive internalization basal ␣ 1a AR colocalizes with clathrin-coated pits. Under these conditions, internalizing ␣ 1a ARs may be drawn from a small proportion of receptors that are in bulk plasma membranes. Alternatively, clathrin-mediated internalization of lipid raft membranes could involve cointernalization with the ␣ 1a AR (61). In this regard, the ability of MCD by itself to decrease surface receptor density strongly suggests cholesterol depletion and raft disruption can increase internalization. Similarly, agonist-induced exit of the ␣ 1a AR from the light rafts precedes and appears to enable increased receptor internalization. Both of these observations suggest ␣ 1a AR internalization may be initiated by raft exit. However, agonist-induced exit of the ␣ 1a AR from light rafts occurs within a time frame (3-10 min) during which surface receptor assays and confocal analysis suggested modest receptor internalization. At this time we cannot directly identify the moderate density membrane structures that stimulated ␣ 1a AR enters when it leaves light rafts, as the receptor is the only protein identified in this environment. However, the discontinuity between raft exit and internalization strongly suggests most receptors have 5 T. Fujita, personal correspondence. FEBRUARY 1, 2008 • VOLUME 283 • NUMBER 5 moved into non-raft plasma membrane where it resides for some time before internalizing or recycling back into rafts. Movement of the ␣ 1a AR into the bulk plasma membrane is also consistent with the apparent shift of stimulated receptor to the denser fractions (4 and 5), which are relatively bereft of G proteins. Conjecturally, it is reasonable to suggest that retaining desensitized receptor at the surface may enable recycling of the ␣ 1a AR to an activable configuration, potentially explaining the modest desensitizations (Ͻ50%) observed for the ␣ 1a AR (38). Taken together, our observations suggest that exit of the ␣ 1a AR from lipid rafts enables but does not immediately initiate receptor internalization.

The ␣ 1a -Adrenergic Receptor Is a Membrane Raft Protein
Despite profound structural and functional differences, the EGFR displays remarkable similarity to the ␣ 1a AR in terms of membrane localization and trafficking (62). Like ␣ 1a ARs, the EGFR has been shown to occupy membrane rafts, both by the raft procedure adapted here (36) and by the more common high pH bicarbonate extraction (48,63,64). In addition the EGFR appears to associate with CT-B-labeled lipid rafts (64). Furthermore, the EGFR exits the raft environment several minutes after stimulation (58) and internalizes via clathrin-coated pits (65). Although internalization of the EGFR is more rapid than for the ␣ 1a AR, raft exit is still an independent event, as inhibition of clathrin-mediated internalization decouples the two processes (58). Clear evidence indicates the EGFR occupies non-caveolae lipid rafts (63), a conclusion consistent with the relative lack of colocalization of the ␣ 1a AR with caveolin 1. Given these similarities it is probably more than coincidence that G q -coupled receptors, including the ␣ 1 AR (66) and endothelin receptor (58), transactivate the EGFR resulting in crosstalk with profound medical implications (67).
Much additional work remains to characterize the membrane environments occupied by the ␣ 1a AR, the function of each environment, and the mechanisms of trafficking that shuttle this receptor and G q -coupled GPCRs in general. We believe the reproducible lipid raft procedure used in this study will provide a necessary vehicle for investigation of the role of rafts in ␣ 1a AR signaling not just through G q -mediated pathways but also G␤/␥-mediated and other pathways.