Identification of a Novel Apical Sorting Motif and Mechanism of Targeting of the M2 Muscarinic Acetylcholine Receptor*

Previous studies have shown that the M2 receptor is localized at steady state to the apical domain in Madin-Darby canine kidney (MDCK) epithelial cells. In this study, we identify the molecular determinants governing the localization and the route of apical delivery of the M2 receptor. First, by confocal analysis of a transiently transfected glycosylation mutant in which the three putative glycosylation sites were mutated, we determined that N-glycans are not necessary for the apical targeting of the M2 receptor. Next, using a chimeric receptor strategy, we found that two independent sequences within the M2 third intracellular loop can confer apical targeting to the basolaterally targeted M4 receptor, Val270-Lys280 and Lys280-Ser350. Experiments using Triton X-100 extraction followed by OptiPrep™ density gradient centrifugation and cholera toxin β-subunit-induced patching demonstrate that apical targeting is not because of association with lipid rafts. 35S-Metabolic labeling experiments with domain-specific surface biotinylation as well as immunocytochemical analysis of the time course of surface appearance of newly transfected confluent MDCK cells expressing FLAG-M2-GFP demonstrate that the M2 receptor achieves its apical localization after first appearing on the basolateral domain. Domain-specific application of tannic acid of newly transfected cells indicates that initial basolateral plasma membrane expression is required for subsequent apical localization. This is the first demonstration that a G-protein-coupled receptor achieves its apical localization in MDCK cells via transcytosis.

The renal epithelial Madin-Darby canine kidney (MDCK) 2 cell line provides a model system for the study of protein targeting because of the ability of this cell line, when grown to confluency, to form a polarized monolayer (1). Polarized epithelial cell surface membranes are divided into apical and basolateral domains possessing distinct protein and lipid compositions that are separated by tight junctions. This spatial asymmetry is generated and maintained by one of three pathways. Proteins can be targeted directly to the apical or basolateral membrane from the trans-Golgi network (TGN) via the exocytotic pathway (2). They can also be targeted indirectly by being delivered to one domain, typically the basolateral domain, endocytosed, and then either recycled back to the original membrane or redirected to the opposite domain in a process termed transcytosis (3,4). Alternatively, proteins can be randomly targeted to both domains and achieve their asymmetric distribution by selective stabilization or retention at one plasma membrane (5)(6)(7). Although other epithelial cell types, including hepatocytes and intestinal cells, primarily utilize the indirect pathway for apical protein delivery (8), MDCK cells primarily use the direct pathway (3).
In MDCK cells, newly synthesized apical and basolateral membrane proteins are segregated into separate transport vesicles within the TGN by virtue of sorting signals within the protein (2,9,10). The routing of basolaterally targeted proteins expressed in this cell line is mediated by discrete cytosolic amino acid sequences that frequently resemble endocytosis signals. These sequences often contain a critical tyrosine residue within an NPXY or YXX⌽ motif (where ⌽ is a bulky hydrophobic amino acid), a dihydrophobic motif, a cluster of acidic residues, or a combination of these elements (11)(12)(13). Recent evidence suggests that in some cases these motifs are recognized by the clathrin adapter proteins AP-1, AP-2, or AP-3 that sequester them into basolateral transport vesicles (3, 14 -16). Molecular signals mediating apical transport in MDCK cells are much more diverse. The apical targeting of proteins within MDCK cells can be mediated by the lipid anchor of glycosylphosphatidylinositol (GPI)-anchored proteins, by protein Nor O-linked glycans, or by amino acid residues located in the extracellular, transmembrane, or cytoplasmic domains of proteins (3,(17)(18)(19). The prevalent model of apical membrane sort-* This work was supported by Grant R01NS26920 from the National Institutes of Health. 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. 1  ing proposes that apical proteins are sequestered into apical transport vesicles via association with sorting platforms rich in cholesterol and sphingolipid-rich lipid rafts at the level of the trans-Golgi network (20,21). The distribution of the mAChR subtypes when exogenously expressed in MDCK cells has been characterized. Although the M 1 , M 4 , and M 5 receptors appeared to be nontargeted, the M 2 and the M 3 receptors displayed an apical and a basolateral distribution, respectively (22). The M 3 basolateral sorting sequence has been defined in a previous study as a 7-amino acid motif residing in the N-terminal portion of the third intracellular loop (22,23). In this study, the molecular signals mediating the apical targeting of the M 2 mAChR and the mechanism of delivery to the apical membrane were investigated.
In this study we use M 2 /M 4 chimeric receptor constructs to show that there is apical targeting information within the third intracellular loop of the M 2 mAChR. We use sequences from the M 2 third intracellular loop appended to the C terminus of M 4 to define two adjacent sequences of amino acids that were able to confer apical targeting to the M 4 receptor in a positionindependent fashion as follows: an 11-amino acid sequence, Val 270 -Lys 280 and a 71-amino acid sequence, Lys 280 -Ser 350 . Furthermore, we demonstrate by using deletion analysis that either of these apical sorting sequences is sufficient to direct the apical sorting of the M 2 receptor. To address the mechanism by which the M 2 receptor cargo is sequestered into apically targeted vesicles, we determine whether or not the M 2 receptor is raft-associated by using both biochemical and confocal analyses. Finally, we use metabolic labeling followed by surface biotinylation, immunocytochemical imaging of the cell-surface staining pattern of newly transfected confluent MDCK cells expressing the M 2 -GFP, and domain-specific application of the cell-impermeable cross-linking agent tannic acid to demonstrate that the M 2 receptor is initially transported to the basolateral domain prior to its subsequent apical localization.

EXPERIMENTAL PROCEDURES
Cell Culture-MDCK (strain II) cells were obtained from Dr. Keith Mostov (University of California, San Francisco). Cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 100 units/ml penicillin G, and 0.1 mg/ml streptomycin sulfate at 37°C in a humidified 10% CO 2 environment.
Transfection of mAChRs and Chimeric Constructs-To analyze the targeting of mAChR constructs, MDCK cells were seeded at high density (1 ϫ 10 6 cells/well) on 24.5-mm polycarbonate Transwell filters (0.4-m pore size; Costar, Cambridge, MA) in DMEM supplemented with 10% fetal bovine serum (1 ml in the apical and 2.5 ml in the basolateral compartments) 24 h before transfection. For each construct to be transfected, 4 g of DNA was added to a final volume of 100 l of DMEM (without serum or antibiotics), and 8 l of Lipofectamine 2000 (1:2 ratio of DNA:Lipofectamine 2000 reagent; Invitrogen) was added to 92 l of DMEM (without serum or antibiotics). The two solutions were combined and incubated on a bench top for at least 30 min before being added to the medium in the apical compartment of the plated MDCK cells. Twenty four hours after the start of transfection, the wells of transfected cells were replaced with fresh media.
Fluorescent images were collected in both the x-y and x-z planes using a Leica TCS SP1/NT confocal microscope (Leica Microsystems, Inc., Exton, PA.) using ϫ100, 1.4 N.A. oil immersion lens in the W. M. Keck Imaging Center at the University of Washington. For each x-y image, a projected z-series image was generated by taking images at 0.6-m intervals from the apical to the basolateral regions of the cells and compressing to a single x-y image. Images were processed using Adobe Photoshop. Quantitation of the apical/basolateral distribution of mAChR deletion constructs was performed using the public domain NIH Image program (developed at the National Institutes of Health). The mean pixel intensity/unit area was determined by manually outlining the areas of interest in the raw (unprocessed) x-z images. Data are expressed as percent basolateral of total staining and are presented as means Ϯ S.E. for the number of experiments indicated. Between 2 and 16 images were quantitated for each experiment. Data were processed using Microsoft Excel.
Construction of Epitope-tagged Chimeric and C-terminal Fusion mAChRs-A modified FLAG epitope (DYKDDDDA) was added to the extracellular N termini of the porcine M 2 (clone Mc7 (25)) and human M 4 (26) mAChR coding sequences immediately after the initiator methionines using PCR as described (22). The M 2 and M 4 receptors were cloned into the KpnI/EcoRI and EcoRV sites of site of pcDNA3.1 (Invitrogen), respectively. M 2 /M 4 chimeric mAChRs were constructed by using sequential PCR using Pfu or Pfu Turbo polymerase (Stratagene) as described previously (27) to replace parts of the M 2 coding sequence with the homologous regions of M 4 coding sequence as described previously. pFM 2 , pFM 4 , or M 2 /M 4 chimeric constructs were used as PCR templates for subsequent chimeras. All PCR-amplified constructs were designed with KpnI and EcoRI sites at their 5Ј-and 3Ј-ends, respectively, and ligated into pcDNA3.1. The sequences comprising the M 2 /M 4 chimeras are as follows, with the numbers in parentheses representing the amino acid residues of M 2 that were substituted into M 4  C-terminal fusion proteins M 4 ϩM 2 (208 -252), M 4 ϩM 2 (208 -280), and M 4 ϩM 2 (208 -352), in which M 2 sequences were fused to the C terminus of M 4 , were generated by sequential PCR using pFM 2 and pFM 4 as a template. The above C-terminal fusion constructs were inserted into the BglII/EcoRI site of pCDPS. The sequences comprising the M 4 C-terminal M 2 appendages are as follows: M 2 (208 -327), coding nt 622-981; M 2 (208 -280), coding nt 622-840 of M 2 ; and M 2 (208 -252), coding nt 622-756.
A C-terminal M 2 -GFP fusion protein was constructed by PCR amplifying FLAG-M 2 with an N-terminal EcoRI site and a C-terminal XbaI site (which replaced the C-terminal ochre stop codon) and ligating the PCR product into pCS2 ϩ XLT vector (28) that was engineered to produce an in-frame N-terminal GFP fusion protein, pCS2 ϩ XLT-GFP (obtained from R. T. Moon).
Generation of MDCKII Cells Stably Expressing FLAG-tagged M 2 mAChR-MDCKII cells stably expressing FLAG-tagged M 2 mAChR were generated by transfecting MDCK II cells cultured in a 10-cm plate at 60 -80% confluency using the calcium phosphate precipitation method (30) with 10 g of the pFM 2 receptor in pcDNA3.1. Beginning at 48 h after transfection, stable transformants were selected by treatment with 1 mg/ml G418 (Geneticin; Invitrogen) for at least 4 weeks following transfection. Drug-resistant cells were selected and then screened by both sterile fluorescence-assisted cell sorting (FACS) and immunocytochemistry. For FACS analysis, cells were dissociated from plates using trypsin (as performed in cell culturing), pelleted by centrifugation, suspended in PBS with 3% BSA and 2 g/ml monoclonal anti-FLAG antibody, and incubated on ice for 30 min. Cells were again pelleted by centrifugation, rinsed 2ϫ in PBS, and resuspended in PBS with 3% BSA with 1:250 anti-mouse FITC-conjugated secondary antibody, and incubated for 30 min. Cells were rinsed 2ϫ in PBS, resuspended in PBS with 3% BSA, and subjected to FACS analysis (Coulter Elite; Beckman Coulter). Selected cells (75,000) were plated, grown, and split into a 96-well culture dish (Costar) at a dilution of 1 cell/ml. Clones were expanded and further evaluated for FLAG-M 2 receptor expression by immunocytochemistry as described under "Immunocytochemical Analysis." Detergent Extraction and Flotation in OptiPrep TM Density Gradients-MDCK cells stably expressing FLAG-tagged M 2 mAChR were grown 3-5 days post-confluency on three 10-cm culture dishes with daily media changes before use. Cells were washed twice in ice-cold PBS and lysed for 30 min in 3 ml of ice-cold TNE (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA), 5 mM dithiothreitol, with 0.5% Triton X-100 and protease inhibitors (Complete; Roche Applied Science) on ice. The lysate was scraped from the dishes, and the plates were rinsed with 1 ml of the ice-cold TNE; the lysates from the three plates were combined and centrifuged at 10,000 ϫ g for 5 min. Membranes were resuspended in 3 ml of the same buffer and were homogenized by passing the sample through a 25-gauge needle 20 times, on ice. The extract was brought to 40% OptiPrep TM in a final volume of 4 ml and sequentially overlaid with 6 ml of 30% and 2 ml of 5% OptiPrep TM . Gradients were centrifuged for 18 h at 39,000 rpm at 4°C in a Beckman SW41Ti rotor. The opalescent band at the 5-30% OptiPrep TM interface, containing lipid rafts, was collected as the Triton X-100-insoluble fraction, and the 40% OptiPrep TM layer was harvested as the Triton X-100soluble fraction. 25-l aliquots were taken from each 1-ml fraction; 60 -80 l of Laemmli sample buffer was added to each fraction, and the sample was heated for 4 min at 95°C and subjected to SDS-PAGE on a 10% polyacrylamide gel. Antibodies used were subtype-specific monoclonal anti-M 2 (1:500) as described (31), polyclonal anti-caveolin (1:5000; Transduction Laboratories), and polyclonal anti-human transferrin antibodies (1:200; Santa Cruz Biotechnology, H-300). Renaissance Western blot chemiluminescent reagent (PerkinElmer Life Sciences) was used to detect immunoreactivity.
Quantitation of the distribution of protein in fractions throughout the gradient was performed using the public domain NIH Image program (developed at the National Institutes of Health). The mean pixel intensity/unit area was determined by manually outlining the areas of interest. The percent of total was determined by dividing the pixel intensity of each band by the sum of the intensities of all the bands throughout the gradient. Data were processed using Microsoft Excel; graphing was performed on CA-Cricket Graph III.1.
Cholera Toxin and Antibody-induced Patching-MDCK cells stably expressing FLAG-tagged M 2 mAChR were plated on MDCK cells seeded at 3.5 ϫ 10 5 cells/well on 2-well glass chamber slides (4.2 cm 2 /well; Nalge Nunc International) and grown to confluency for 3-4 days. AlexaFluor 594 conjugate of cholera toxin ␤-subunit (ChTx␤; Molecular Probes) was used to label endogenous glycosphingolipids. Cells were incubated with ChTx␤ (10 g/ml) in DMEM-HEPES (25 mM HEPES), 0.2% BSA on ice for 30 min and then rinsed three times with ice-cold PBS. For the nonpatched (control) condition, cells were rinsed, fixed, and stained as described under "Immunocytochemical Analysis." Lipid raft aggregation, or patching of ChTx␤, was induced by incubating the cells with anti-ChTx␤ antibody (1:250 in DMEM-HEPES, 0.2% BSA; Calbiochem-Novabiochem) for 20 min at 37°C. Cells were removed from the incubator, rinsed, fixed, and stained as described previously. Co-patching of M 2 mAChR and ChTx␤ was performed by simultaneously incubating cells on ice with ChTx␤ and anti-FLAG antibody (2.5 g/ml) in DMEM-HEPES, 0.2% BSA. After rinsing three times with ice-cold PBS, cells were patched with both anti-ChTx␤ antibody (as described above) and FITC-conjugated goat anti-mouse secondary antibody (1:250; Cappel Research Products, Durham, NC) for 20 min at 37°C. Cells were rinsed three times with PBS, fixed, and visualized as described above. As a positive control, cells were labeled simultaneously with BODIPY FL-labeled C5 ganglioside GM1 (500 nM on PBS; Molecular Probes) and ChTx␤ for 30 min on ice and were either subsequently fixed or patched as described above with anti-ChTx␤.
Quantitation of colocalization was performed using the plug-in RG2B_colocalization to ImageJ (version http://rsb.info. nih.gov). The colocalization measurements were taken using a minimum threshold pixel intensity adjusted between 100 and 200 and set equivalently for both channels to correspond to the image intensity. The minimum ratio for pixel intensity between the two channels was set to 0.5. Colocalization pixels were displayed as an image with maximum pixel intensity (255). The RGB output of the plug-in was split into green, red, and blue images, and each image was quantitated by measuring the total pixel area above the threshold intensity used in colocalization determination (see above). Results are displayed as percent colocalization as determined by dividing the area of colocalization pixels by the total pixel area over the threshold of whichever channel was limiting.
Functional Confirmation of Intact Monolayers Prior to Metabolic Labeling-For polarity experiments, MDCKII cells were seeded at a density of 1 ϫ 10 6 cells/24.5-mm polycarbonate membrane filter (Transwell chambers, 0.4-m pore size; Costar, Cambridge, MA) and cultured for 5-8 days with medium changes every day. Prior to each functional or immunocytochemical experiment, the integrity of the monolayer was assessed by adding [ 3 H]methoxyinulin (PerkinElmer Life Sciences) to the apical medium and monitoring the leak of [ 3 H]methoxyinulin from the apical compartment to the basolateral compartment by sampling and counting the basolateral medium in a scintillation counter after a 1-h incubation at 37°C. Chambers with greater than 3% leak per h were discarded.
Metabolic Labeling/Biotinylation Strategy for Determining Surface Delivery of the M 2 mAChR-MDCK cells were metabolically labeled, surface-biotinylated, and evaluated as described (32). MDCK cells stably expressing the FLAG-M 2 receptor cells grown on Transwells were washed twice with PBS and grown for 45 min in methionine/cysteine-free medium (Invitrogen) supplemented with 4 mM glutamine and 5% fetal calf serum. Following removal of this medium, cells were inverted (cell side down), and 150 -200 l of methionine/cysteine-free medium with 5% fetal bovine serum containing 1 Ci/l 35 S-Express protein labeling mixture (NEG-072; PerkinElmer Life Sciences) was added to the bottom of inverted filters. The cells were incubated at 37°C for various lengths of time. Each time point reflects a pooled sample from three 24.5-mm wells. At the end of the labeling period, cells were washed three times with PBS/CM (PBS containing 0.5 mM CaCl 2 and 2 mM MgCl 2 ). Sulfo-NHS-LC-biotin (Pierce) at 1 mg/ml in PBS buffer was then applied to the apical or basolateral surface and incubated for 30 min on ice in order to covalently label surface proteins. The opposite compartment not being treated with biotin was incubated with PBS/CM containing 1% (w/v) BSA. After the incubation, cells were washed once in 50 mM Tris to quench and three times for 5 min with PBS/CM. Cells were harvested in 1 ml of ice-cold PBS using a rubber policeman and pelleted by centrifugation. Membranes were solubilized by resuspending the pellet in solubilization buffer as follows: 1% digitonin (Calbiochem) and 0.1% cholate (Sigma) in phosphate buffer (10 mM KH 2 PO 4 , pH 7.0, 1 mM EDTA, 50 mM NaCl) with protease inhibitors (Complete; Roche Applied Science) and rotating at 4°C for 6 h. Membranes were pelleted by centrifugation, and the supernatant was rotated overnight at 4°C with 9.0 g/ml anti-FLAG M 2 monoclonal antibody (Sigma). 50 l of a 1:1 slurry of protein G-agarose (Roche Applied Science) was added to the membrane/antibody mixture after preincubating the agarose beads for 1 h in 2.5% BSA in solubilization buffer followed by two 5-min washes with solubilization buffer alone and rotated overnight at 4°C. The next day the protein G-agarose was pelleted by centrifugation. Pellets were washed three times for 5 min with 0.5% digitonin and 0.1% cholate in phosphate buffer with protease inhibitors (Complete; Roche Applied Science). Following the last wash, proteins were eluted from the protein G-agarose beads by incubating in 100 l of SDS sample buffer at 90°C for 20 min and agitated every 5 min. A second elution was performed in 50 l of sample buffer. Samples were combined and subjected to SDS-PAGE on a 10% polyacrylamide gel and subsequent autoradiography using BioMax MS-1 film and a BioMax TransScreen-LE (PerkinElmer Life Sciences). The identity of the radioactive band as that of the M 2 receptor was confirmed by analyzing membranes of M 2 stably expressing MDCK cells by Western analysis using an affinity-purified subtype-specific polyclonal anti-M 2 (1:200) raised to the third intracellular loop of mouse M 2 mAChR and matching the electrophoretic mobility of the immunoreactive band with that of the radioactive band.
Time Course of M 2 -GFP Surface Expression and Tannic Acid Treatment-Cells were plated near confluency at 1 ϫ 10 6 cells/ well in Transwell filters (0.4 m pore size; Costar, Cambridge, MA). 36 h later, cells were transfected with 4 g of the M 2 C-terminal GFP fusion protein FLAG-M 2 -GFP in pCS2 ϩ XLT using Lipofectamine as described above. 12 h after the start of the transfection, when cell surface staining of the M 2 -GFP receptor construct was just detectable by immunocytochemical methods, 0.5% tannic acid (Sigma) in serum-free DMEM was applied to either the apical or basolateral compartments of Transwells for an additional 30, 60, or 120 min. The opposite compartment to the one being treated with tannic acid contained standard growth media. Control wells at each time point were not treated with tannic acid and are displayed in Fig. 10. The integrity of the monolayer and confirmation of polarization was determined by staining with ␤-catenin antibody after 2 h of tannic acid treatment of either the basolateral or apical compartments. Cells were rinsed three times with PBS at the end of the incubation, fixed, labeled, and stained using goat anti-mouse Alexa-568-conjugated secondary antibody (1:500; Molecular Probes) using nonpermeabilized conditions, or stained with ␤-catenin antibody (1:200; Transduction Laboratories) using permeabilized conditions, as described under "Immunocytochemical Analysis."

Localization of Wild Type M 2 mAChR and a Glycosylationdefective Mutant M 2 Receptor in MDCK Cells-Many studies
have indicated that N-or O-glycosylation is critical not only for the exit of newly synthesized proteins from the endoplasmic reticulum (33,34) and the Golgi (35) but also for targeting of newly synthesized proteins to the correct membrane domain of polarized epithelial cells (35)(36)(37). Despite mounting evidence indicating that apical sorting can occur independently of N-glycosylation (18, 38 -40), a role for N-glycans in apical sorting continues to be supported by recent evidence that N-glycans alone can mediate apical sorting of the membrane dipeptidase, independent of raft association (41). Furthermore, the addition of glycosylation sites to the basolaterally localized Na,K-ATPase ␤ 1 resulted in the redirection of this subunit to the apical domain of HGT-1 cells (a human gastric adenocarcinoma cell line) and the redistribution of the associated Na,K-ATPase ␣ 1 subunit to the apical domain as well (42).
The M 2 mAChR is a sialoglycosylated transmembrane protein (43) that has three potential sites for N-glycosylation. A previous study has shown that N-glycosylation of the M 2 receptor is not required for cell-surface localization or ligand binding and does not confer increased stability against receptor degradation in Chinese hamster ovary cells (29); however, the importance of N-glycans in mediating the membrane targeting of the M 2 receptor within a polarized cell type has not been investigated. To determine whether N-glycosylation plays a role in the steady-state apical localization of the M 2 mAChR in MDCK cells, a glycosylation-defective mutant in which the three asparagine codons were substituted with aspartate residues (29) was transfected into MDCK cells, and the steady-state distribution of the mutant within the polarized monolayer was analyzed by immunocytochemistry and confocal microscopy. Similar to the wild type receptor, the glycosylation-defective mutant was localized to the apical domain of MDCK cells (Fig. 1, A and B, respectively). This result demonstrates that the steady-state distribution of the M 2 mAChR to the apical domain of confluent MDCK cells is independent of N-glycosylation.
The Third Intracellular Loop of the M 2 mAChR Contains Apical Sorting Information-Whereas the M 2 receptor displays an apical distribution when expressed in MDCK cells, the M 4 receptor was shown previously to exhibit an apparently nonpolarized, cytoplasmic distribution under permeabilizing staining conditions when transiently expressed in MDCK cells (22). Because of the differential localization of these receptors and their high degree of amino acid identity (56% identity, 78% if the third intracellular loop is excluded), we tested the use of M 2 /M 4 receptor chimeric constructs in a gain-of-function approach to determine which domain of M 2 could confer apical targeting to the M 4 receptor. To visualize the plasma membrane distribution of the M 2 and M 4 receptor constructs, we used nonpermeabilizing conditions to stain for cell-surface expression of the extracellular N-terminal FLAG epitope of transiently transfected filter-grown MDCK cells. Under these conditions, the FLAG-M 4 construct was localized to the basolateral domain on the cell surface (Fig. 3B). Because of the differential distribution of the M 2 and M 4 receptors on the cell surface, we were able to pursue the approach of defining an M 2 apical sorting signal using M 2 /M 4 chimeras. A schematic representation of the chimeric constructs is presented in Fig. 2. Fig. 3 shows the steady-state localizations of the chimeric receptors. The first construct, M 4 /M 2 (208 -467), contains the N-terminal portion of M 4 through the predicted fifth transmembrane (TM5) region and M 2 sequence from the N terminus of the third intracellular loop to the C terminus of the receptor. This chimera displayed an apical dis-tribution in MDCK cells, similar to that of the wild type M 2 receptor (Fig. 3, A and C). This result suggested that a region C-terminal to TM5 of the M 2 receptor is sufficient to confer apical sorting to the M 4 receptor. To test this hypothesis, we examined the surface distribution of a receptor that contained the M 4 sequence through the M 4 third intracellular loop, and M 2 sequence from TM6 through the C terminus of the receptor, M 4 /M 2 (391-467). As shown in Fig. 3D, this receptor was basolaterally targeted, suggesting that the M 2 apical targeting information was contained in the third intracellular loop of M 2 . To further examine this possibility, a pair of constructs representing the inverse of the constructs tested above was generated. These constructs consisted of the N-terminal M 2 sequence either up to M 2 (1-207)/M 4 or through M 2 (1-390)/M 4 , the M 2 third intracellular loop within the M 4 context. In support of the hypothesis that the M 2 third intracellular loop contains apical sorting information, the M 2 (1-207)/M 4 , which lacked the M 2 third intracellular loop, displayed a basolateral steady-state distribution, whereas the M 2 (1-390)/M 4 construct, which encompassed the M 2 third intracellular loop, was apically directed (Fig. 3, E and F). To determine whether the third intracellular loop alone was sufficient to confer apical targeting to the M 4 receptor, the M 2 third intracellular loop sequence was substituted into the context of the M 4 receptor in the construct M 4 /M 2 (208 -390). The steady-state distribution of M 4 /M 2 (208 -390) was apical (Fig. 3G). These results suggest that the M 2 third intracellular loop contains sufficient sorting information to redirect the basolaterally targeted M 4 receptor to the apical domain in MDCK cells.
The M 2 Apical Sorting Sequence Is Not Dependent on Its Context within the Third Intracellular Loop-Results from the chimeric constructs discussed above indicate that the M 2 third intracellular loop, when substituted into the M 4 receptor context, can redirect the basolaterally targeted M 4 receptor to the apical domain of MDCK cells. The redirection of the resulting construct could be due to either the addition of an apical sorting  signal from the M 2 receptor or the removal of a primary basolateral sorting determinant from the M 4 receptor. Indeed, there is support for the existence of a hierarchy in sorting signals of proteins expressed in MDCK cells. Removal of a basolateral sorting sequence often (44,45), but not always (19,46), results in the unmasking of recessive apical sorting information and consequential apical targeting of the protein. To distinguish between these possibilities, and to identify a smaller region of the M 2 receptor that could confer apical targeting to the M 4 receptor, three constructs were made that consisted of M 2 third intracellular loop sequences appended to the C terminus of M 4 as follows: M 4 ϩM 2 (208 -327), M 4 ϩM 2 (208 -280), and M 4 ϩM 2 (208 -252) which are diagrammed in Fig. 4A. As shown in Fig. 4B, the constructs consisting of longer M 2 appendages to M 4 , M 4 ϩM 2 (208 -327), and M 4 ϩM 2 (208 -280), exhibited an apical distribution similar to that of wild type M 2 (Fig. 4B, panels 1, 3, and 4), whereas the shorter appendage M 2 (208 -252) displayed an apical distribution similar to that of the wild type M 4 receptor (Fig. 4B, panels 2 and 5). These results indicate that M 2 sorting information can be conferred to the full-length M 4 receptor, that the functionality of the sorting sequence is not dependent on its position within the third intracellular loop, and that M 2 amino acids Glu 253 -Lys 280 contain sufficient apical targeting information to redirect the basolaterally targeted M 4 receptor to the apical domain.
There Are Redundant M 2 Apical Targeting Determinants within the M 2 Third Intracellular Loop-The observation that M 2 amino acids Glu 253 -Lys 280 within the larger context of amino acids 208 -280 was sufficient to confer apical targeting to the basolaterally targeted M 4 receptor led us to ask if a minimal sequence resided within this larger context that was sufficient to confer apical targeting information. Smaller overlapping M 2 third intracellular loop sequences were appended to the full-length M 4 receptor and tested as described above. Additionally, because there is evidence for redundancy of targeting sequences within proteins such as the transforming growth factor-␣ precursor (47), the low density lipoprotein receptor (11), and the ␣ 2A -adrenergic receptor (48) (Fig. 5B). In contrast, M 2 amino acids 260 -280, when appended to the M 4 C terminus, did result in an apical distribution similar to wild type M 2 . In light of these results, an additional construct was made to address the question of whether the M 2 residues 270 -280 held sufficient apical targeting information to confer apical targeting to the M 4 receptor. As shown in Fig. 5B, panel 6, the M 4 ϩM 2 (270 -280) was apically targeted indicating that 11 amino acids from the third intracellular loop of M 2 can indeed confer apical targeting to the full-length M 4 receptor.
Attempts at further defining the apical sorting sequence within Lys 280 -Ser 350 by dividing it into smaller sequences and appending these to the C terminus of the M 4 receptor resulted in constructs that did not express well. 3 It was therefore not possible to determine whether a smaller region of this sequence was sufficient to confer apical targeting. Nonetheless, these results indicate that there are redundant apical targeting sequences within the third intracellular loop of M 2 and that M 2 amino acids 270 -280 and 280 -350 each contain sufficient sorting information to confer apical targeting to the M 4 receptor. Both apical sequences are indicated in Fig. 5C.  necessary for the apical targeting of the M 2 receptor. First, to determine whether deletion of the M 2 third intracellular loop was sufficient to cause mis-localization of the receptor, amino acids 220 -380 (comprising 89% of the third intracellular loop) were deleted. Although less than 10% (8.0 Ϯ 4.4%, n ϭ 6, where n is the number of independent transfections, between 2 and 14 cells were quantitated in each experiment; mean Ϯ S.E.) of the total staining for the wild type receptor was present in the basolateral domain, the deletion mutant receptor M 2 (del 220 -380) displayed a significant increase in basolateral staining (30.8 Ϯ 10.1%, n ϭ 5). A series of mutants containing smaller overlapping deletions was generated to determine the necessity of those sequences for the apical targeting of the M 2 receptor. Although deletion mutants M 2 (del 220 -280), M 2 (del 280 -325), and M 2 (del 303-380) displayed little or no increase in the amount of basolateral staining of the receptor when compared with wild type (10.4 Ϯ 4.6%, n ϭ 2; 10.5 Ϯ 4.1%, n ϭ 5; 23.0 Ϯ 5.7, n ϭ 6), deletion of amino acids 227-315, which disrupts both apical sorting sequences, resulted in the largest degree of basolateral accumulation (40.5 Ϯ 6.4%, n ϭ 5). These results support the finding that the M 2 receptor contains redundant sorting information because either sorting sequence was sufficient to direct the receptor apically, and only when both were deleted was there significant mis-localization of the receptor.
The M 2 Receptor Is Not Found in Detergent-resistant Membranes-Sphingolipid-and cholesterol-rich microdomains called rafts are postulated to form sorting platforms within the TGN that concentrate apically targeted proteins into transport vesicles (49,50). Association with rafts has been shown to be crucial for the correct targeting of a number of apically directed proteins in MDCK cells, as disruption of rafts by cholesterol depletion often results in the mistargeting of these proteins (51)(52)(53). To determine whether the M 2 mAChR receptor achieves its apical localization in MDCK cells through association with rafts, confluent MDCK cells stably expressing FLAG-M 2 were extracted with the nonionic detergent Triton X-100 on ice followed by density gradient centrifugation using a multistep gradient. As shown in Fig.  7A, the M 2 receptor was not found in appreciable amounts in the detergent-resistant buoyant fractions of the gradient that contained the well characterized raft marker caveolin-1. It was, however, found in highest concentration (94% of total) in the high density detergent-soluble fractions together with the nonraft marker transferrin at the bottom of the gradient. Quantitation of the immunoreactivity throughout the density gradient of each of the proteins is provided in Fig. 7B. These results indicate that by the method of detergent extraction, the M 2 mAChR is not significantly associated with detergent-resistant lipid rafts.
The M 2 Receptor Is Not Found in Cholera Toxin-induced Patches-Because resistance to solubilization in the nonionic detergent Triton X-100 as a method for determining raft association has been criticized as an indirect method open to a variety of interpretations that may not truly reflect raft size, structure, composition, or even existence (54, 55), we used cross-linking of ChTx␤ patches to obtain visual evidence of M 2 association with lipid rafts in intact cells. MDCK cells stably expressing the FLAG-M 2 mAChR were labeled with the AlexaFluor-conjugated ChTx␤-594. The pentavalent cholera toxin ␤-subunit binds glyco-sphingolipids and exhibits a particularly strong affinity for the GM1 glycosphingolipid, a widely used indicator for the distribution of lipid rafts (56). Cholera toxin binding to five cell surface GM1 molecules leads to co-clustering of various glycosylated raft-associated proteins that are further aggregated, or "patched," by exposure to anti-ChTx␤ antibody, which brings entire lipid microdomains together into patches (21,57).
To verify the ability of patched ChTx␤-594 to aggregate the cellsurface raft marker GM1, live MDCK cells were labeled with a fluorescent-conjugated GM1 lipid, GM1-BODIPY, treated with ChTx␤-594, and then either fixed or patched with anti-ChTx␤ antibody and then fixed. As shown in Fig. 8A, panel 1, and quantitated in Fig. 8C, unpatched GM1-BODIPY displayed a homogeneous distribution and minimally colocalized with ChTx␤-594 (23.5 Ϯ 1.7% colocalization, mean Ϯ S.D., average of three cells from a single experiment). After patching with anti-ChTx␤ antibody (Fig. 8A, panel 2), both the ChTx␤-594 and the GM1-BODIPY labeling appeared more punctate, and there was a significant increase in colocalization between anti-ChTx␤-induced patches and aggregated GM1-BODIPY (67.6 Ϯ 19.9% colocalization, three cells from a single experiment). We next investigated whether or not the M 2 receptor was associated with ChTx␤-induced patches. When live cells were treated with ChTx␤-594, fixed, and stained with anti-FLAG antibody, a homogeneous distribution of both FLAG-M 2 and ChTx␤-594 was detected, and the overlay of the two images (Fig.  8B, panel 3) as well as quantitation indicate a low level of colocalization (15.3 Ϯ 3.2% colocalization, three cells from two independent experiments; Fig. 8C). When ChTx␤-594labeled cells were patched using anti-ChTx␤ antibody, a more punctate distribution of ChTx␤-594 labeling was apparent (Fig. 8B,  panel 4); however, there was minimal effect on the distribution of either the M 2 receptor or on colocalization of the M 2 recep-  tor with ChTx␤-594 patches (14.6 Ϯ 8.0% colocalization, five cells from three independent experiments; Fig. 8C). To increase the resolution of the staining for the M 2 receptor, and to enhance the visualization of colocalization with lipid rafts, both the M 2 receptor and ChTx were co-patched by simultaneously treating cells with mouse anti-FLAG antibody and ChTx␤-594 on ice, and then patching with anti-mouse FITC-conjugated secondary antibody and anti-ChTx␤ antibody (Fig. 8B, panel 5). Under these conditions, staining for the M 2 receptor indeed appeared more punctate; however, there was little increase in colocalization between M 2 receptor and ChTx␤-594 labeling (24.2 Ϯ 3.5% colocalization, eight cells from two independent experiments; Fig. 8C), as the vast majority of receptor labeling was found outside the ChTx␤-induced patches. 35

S-Metabolic Labeling and Surface Biotinylation of MDCK Cells Stably Expressing Epitope-tagged M 2 mAChR-Although
in most epithelial cells, including MDCK cells, newly synthesized apical and basolateral proteins are targeted directly from the TGN to their final surface localization, the Na ϩ /K ϩ -ATPase (58) and the ␣ 2B -adrenergic receptor (6) achieve their final basolateral localization after random surface delivery and selective stabilization at the basolateral domain, whereas the polymeric immunoglobulin A receptor (pIgA-R) (59, 60) is targeted directly to the basolateral surface of MDCK cells and subsequently achieves its apical localization via transcytosis.
To determine the pathway by which the M 2 mAChR achieves its steady-state apical distribution when expressed in MDCK cells, FLAG-M 2 stably expressing cells were metabolic labeled and surface-biotinylated, and proteins were immunoprecipitated with anti-FLAG antibody, followed by streptavidin precipitation, and analyzed by SDS-PAGE, followed by autoradiography. As shown in Fig. 9, at 15 min, newly synthesized epitope-tagged M 2 mAChR was first detected on the basolateral surface, whereas the amount of receptor present on the apical domain was similar to that of control (representing nonbiotinylated samples). At 30 min, the majority of newly synthesized epitope-tagged M 2 mAChR remained localized at the basolateral surface (83.6 Ϯ 13.7%, n ϭ 2 independent experiments); however, significant apical accumulation was observed. By 60 min the receptor showed a greater accumulation on the apical domain when compared with the basolateral domain (60.0 Ϯ 3.8%, n ϭ 2). The identity of the band as the M 2 receptor was ensured by concurrently subjecting a membrane sample of the same M 2 stably expressing MDCK cells to SDS-PAGE and matching the migration of the immunoreactive M 2 band (using a monoclonal anti-M 2 antibody) with that of the radioactive signal. 3 These results suggest that the M 2 mAChR accumulates initially on the basolateral surface of MDCK cells, and within an hour the majority of the receptor is present on the apical domain.
Time Course of Expression of the M 2 mAChR in Newly Transfected MDCK Cells as Monitored by Confocal Microscopy-As discussed above, the metabolic labeling studies indicate that the M 2 mAChR accumulates initially within the basolateral domain of MDCK cells. As a further test of this result, the fusion protein FLAG-tagged M 2 -GFP was transiently transfected into MDCK cells. The surface expression of the newly synthesized M 2 receptor was analyzed by staining (red) for the FLAG epitope under nonpermeabilizing conditions, whereas GFP fluorescence allows visualization of total cellular M 2 expression. Because preliminary experiments indicated that the receptor started to appear on the cell surface 12 h after the start of transfection, the time course of receptor expression was determined using 12 h as the initial time point. Although the expression level varied among cells, only the highest expressing cells were analyzed at the stated time points. Twelve hours after transfection (t ϭ 0; Fig. 10), the surface staining of very faint M 2 -GFP expressing cells was undetectable above background staining. Within 30 min (t ϭ 30; Fig. 10), the surface distribution of the receptor in the most highly expressing cells appeared basolateral. By 60 min, the apical domain was enriched with M 2 receptor compared with expression within the basolateral domain (t ϭ 60; Fig. 10). At 120 min, the M 2 distribution appeared apical in highly expressing MDCK cells (t ϭ 120; Fig. 10). However, if the gain at which the confocal images were collected was unaltered from that which was used at earlier time points, the image corresponding to t ϭ 120 min did indeed display M 2 immunoreactivity basolaterally as well. 3 The gain was lowered for collection at these later time points in order to avoid over-  . Metabolic labeling/biotinylation strategy for determining surface delivery of the M 2 mAChR. The amount of newly synthesized receptor delivered to the apical versus basolateral surface was quantified by biotinylating one surface or the other of metabolically labeled MDCKII cells stably expressing the FLAG-M 2 mAChR in Transwell culture and subsequently isolating radiolabeled receptor on the biotinylated surface through sequential streptavidin-agarose and immunoprecipitation. Cells were surface-biotinylated on either the apical (Ap) or basolateral (Bl) surface after metabolically labeling with 35 S-Express protein labeling mixture for the times indicated as follows: 15, 30, and 60 min. The control lane (C) represents MDCK cells metabolically labeled for 2 h that were not surface-biotinylated and were otherwise treated similarly to other samples. The molecular mass marker at 75 kDa is indicated.
exposure. Images were collected at similar gains for both the green and red channels to reflect what was observed using the microscope alone. The lack of yellow in the overlay projection at earlier time points is because of the intensity differences between the two labels. To ensure that surface staining was not artifactual, an analogous experiment was performed by transiently transfecting FLAG-M 2 in MDCK cells and monitoring the time course of appearance of cell surface staining as described above using a fluorescein-conjugated secondary antibody, which yielded similar results. 3 These results corroborate the results obtained by metabolic labeling and surface biotinylation discussed above, and again indicate basolateral surface expression of the M 2 receptor prior to appearance in the apical membrane.
Inhibition of Surface Delivery of Vesicles Containing Newly Synthesized M 2 mAChR by Tannic Acid Treatment-To confirm that the M 2 mAChR was first delivered to the basolateral domain before its subsequent appearance at the apical membrane, MDCK cells were transfected with FLAG-M 2 -GFP, and 12 h after the start of transfection cells were treated for varying lengths of time with tannic acid, a cell-impermeable fixative that cross-links carbohydrate groups and does not diffuse across the tight junctions (61,62), in order to block the fusion of vesicles bearing newly synthesized proteins to either the apical or basolateral domain. As described above, 12 h after the start of transfection, the surface staining of cells expressing FLAG-M 2 -GFP was not detectable above background staining (t ϭ 0; Fig.  11A). Cells treated on the apical domain with tannic acid for 30 min displayed a basolateral distribution similar to that which was observed at the t ϭ 30 control sample in the previous section (t ϭ 30; Fig. 11A, comparable with Fig. 10, panel 2). After 60 and 120 min of apical tannic acid treatment, the lateral staining intensity increased compared with the previous time points, whereas the delivery of M 2 to the apical domain remained blocked (t ϭ 60 and 120; Fig. 11A). When the basolateral domain of filter-grown MDCK cells was treated with tannic acid for 30, 60, and 120 min, M 2 surface immunoreactivity was not apparent on either the basolateral or the apical domain but did accumulate intracellularly (t ϭ 30, 60, 120; Fig. 11B). These results show that the M 2 mAChR accumulates basolaterally when vesicle fusion to the apical domain is blocked by tannic acid treatment, and that apical surface expression is inhibited by blocking the initial delivery of the M 2 mAChR to the basolateral domain of MDCK cells. The latter finding confirms that the M 2 receptor must first be targeted to the basolateral domain before undergoing transcytosis to the apical domain of MDCK cells.
The physical integrity and polarization of the MDCK monolayer following treatment with tannic acid were confirmed by staining for basolaterally targeted ␤-catenin. The x-z view indicates that in MDCK cells transfected with M 2 -GFP (green), treated for 2 h with tannic acid on the apical domain and stained for ␤-catenin (red), the integrity of the monolayer is maintained and the basolateral distribution of ␤-catenin remains unperturbed (Fig. 11C). Similar results were obtained for cells treated with tannic acid on the basolateral domain. 3

DISCUSSION
The goal of this study was to characterize the molecular signals involved in the apical targeting of the M 2 mAChR in MDCK cells and address the mechanism by which it achieves its steady-state distribution. The apical targeting of a transmembrane protein in MDCK cells can be mediated by a variety of signals, including glycosylation motifs and amino acid sequences located within the extracellular, transmembrane, or intracellular domain (18,35,63,64). Both N-and O-glycans are postulated to mediate apical targeting in a number or proteins (35,36,65,66) via incorporation into lipid rafts via putative raft-associated carbohydrate-binding lectins localized to the TGN (20,67). However, there are examples where apical targeting may depend on glycosylation, yet be independent of raft association (68,69). We found that glycosylation of the M 2 receptor is not required for its targeting in MDCK cells.
While investigating the possibility of utilizing M 2 /M 4 chimeric receptors to determine the location of the M 2 receptor apical sorting signal, we determined that the cell surface distribution of the M 4 receptor is basolateral when transiently expressed in MDCK cells. The differential steady-state distribution of the M 2 and M 4 receptors allowed us to use a gain-offunction approach to identify apical targeting determinants within the M 2 mAChR. The high degree of sequence identity between these two receptors and their similar ability to functionally couple to the G i The apical targeting determinants of the M 2 receptor were able to override the basolateral sorting information contained in the M 4 receptor. Although a hierarchy of strength in targeting signals has been demonstrated for a number transmembrane proteins in which deletion of a basolateral targeting sequence reveals the existence of a recessive apical targeting determinant (45,70,71), there are studies demonstrating that an apical sorting sequence can override a basolateral targeting sequence as demonstrated here and by others (19,(72)(73)(74). Interestingly, in a previous study, . Cells were stained with mouse anti-FLAG antibody, labeled with Alexa-568-conjugated secondary antibody, and visualized using confocal microscopy as described under "Experimental Procedures." The 1st column indicates GFP fluorescence; the 2nd column indicates Alexa-568 fluorescence, and the 3rd column is an overlay of the first two columns. C, ␤-catenin staining of tannic acid-treated MDCK cells transfected with M 2 -GFP. Filter grown MDCK cells transfected with FLAG-tagged M 2 -GFP were treated for 2 h with 0.5% tannic acid on the apical or the basolateral domain. Cells were fixed, labeled with anti-␤-catenin antibody, stained with Alexa-568-conjugated secondary antibody, and visualized using confocal microscopy as described under "Experimental Procedures." Images shown are for cells treated apically; identical images were obtained for cells treated basolaterally. The upper panels are images taken in the x-y plane (projected z-series), and the lower panels are x-z images of a vertical section taken at or near the midpoint of the image. Bar ϭ 20 m. the M 2 targeting sequence was not able to override the M 3 basolateral targeting determinant when this M 3 determinant was appended to the C terminus of the M 2 receptor (22). These results suggest that there is no specific hierarchy in the strength of basolateral compared with apical sorting sequences in MDCK cells, but rather that the strength of the signal may be an inherent property of the signal itself.
A series of M 2 third intracellular loop deletion mutants suggested that there are multiple regions within the M 2 third intracellular loop that contribute to its apical targeting. Because deletion of nearly the entire third intracellular loop did not result in a fully unpolarized distribution of the receptor, other sequences besides those investigated here may contribute to the apical targeting of the M 2 receptor. The notion that a protein carries multiple autonomous and redundant targeting motifs is not novel. Examples include the transforming growth factor-␣ precursor (47), the low density lipoprotein receptor (11), and the kidney AE1 anion exchanger (75). Consistent with this idea, the M 3 mAChR receptor also displays similar redundancy of targeting signals as deletion of the primary basolateral targeting sequence identified in the M 3 receptor did not perturb its steady-state basolateral distribution (22).
It is particularly interesting that the sorting sequences of the two muscarinic receptors examined to date contain sorting determinants localized to their third intracellular loops. The majority of polytopic membrane proteins, including the G-protein-coupled receptors rhodopsin, the metabotropic glutamate receptor, and the serotonin 5-HT1B receptor, whose expression pattern has been characterized in MDCK cells, is targeted by motifs localized to their C-terminal domains (46, 72, 76 -84). A few multipass transmembrane proteins with cytosolic targeting sequences present exceptions to this paradigm. The calcium ATPase type 2 splice form contains a 45-amino acid insert within the first intracellular loop that confers apical targeting to an otherwise basolaterally targeted protein (85), the multidrug resistance protein 1 contains basolateral targeting information in its third cytoplasmic loop (86), and the liver Na ϩ -taurocholate cotransporting polypeptide and prominin retain their basolateral and apical distribution, respectively, despite C-terminal truncation (73,87). To date, none of the previously identified apical targeting sequences demonstrate homology or amino acid similarity to the targeting sequences identified within the M 2 mAChR. Whether structural similarity is conserved between apical sorting determinants has yet to be determined.
The apical sorting of many transmembrane and GPI-anchored proteins in MDCK cells is mediated by their incorporation into rafts (49,51). Raft incorporation can occur via the association of GPI anchors with the exoplasmic leaflet of the plasma membrane (88), via N-or O-glycans present on the extracellular domains of proteins (35,37), or via amino acids within the transmembrane domain or membrane-proximal regions of integral membrane proteins (89,90). The relevance of raft association for apical targeting depends on the protein being investigated. In the case of the p75 neurotrophin receptor, the presence and location of membrane-proximal O-glycosylation is critical for both raft association and apical delivery (91). In some proteins, such as hemagglutinin, neuraminidase, and megalin, sequences mediating apical sorting are distinct from amino acids involved in raft association, and raft association alone is not sufficient to mediate apical targeting of these proteins (89,90). Still other apically targeted membrane proteins such as CD3⑀ (63), the dipeptidyl peptidase IV (92), the glycine transporter GLYT2 (69), the glutamate transporter excitatory amino acid transporter (93), the metabotropic glutamate receptor 1b (78), and the guanylyl cyclase C (76) are not raft-associated as determined by Triton X-100 extraction. In this study, by both Triton X-100 extraction and antibody crosslinking of cholera toxin-induced patches, the M 2 mAChR was determined to be minimally associated with lipid rafts. This finding suggests that the mechanism by which the M 2 mAChR receptor is sequestered into apically targeted vesicles at the level of the TGN is not via direct association with rafts. For this protein and other nonraft associated polytopic membrane proteins possessing cytosolic apical targeting domains, the mechanism by which the protein is concentrated in apically targeted cargo vesicles has yet to be determined. Given the evidence presented here, it is indeed possible that this subset of proteins is packaged with basolaterally targeted proteins and redirected apically after achieving a basolateral membrane distribution.
Proteins can achieve their final membrane localization through one of three pathways in epithelial cells: direct delivery, selective stabilization after random delivery, and via transcytosis to one plasma membrane domain following delivery to the opposite domain (3). The data presented here demonstrate that by metabolic 35 S labeling and surface immunoprecipitation, as well as confocal analysis of newly transfected confluent MDCK cells, that the M 2 mAChR is first directed to the basolateral domain before it accumulates on the apical domain. This represents the first report of a seven-transmembrane receptor undergoing transcytosis in MDCK cells. For some epithelial cell types other than MDCK cells, transcytosis is a predominant mechanism by which proteins achieve their apical localization (94). Although in MDCK cells most newly synthesized apical and basolateral proteins are targeted directly from the TGN to their final surface localization, the Na ϩ /K ϩ -ATPase (58) and the ␣ 2B -adrenergic receptor (6) achieve their final basolateral localization after random surface delivery and selective stabilization at the basolateral domain, and the polymeric immunoglobulin A receptor (59,60) is targeted directly to the basolateral surface of MDCK cells and subsequently achieves its apical localization via transcytosis. Indeed, there are a number of other proteins utilizing the transcytotic pathway in MDCK cells (40,95). Whether the initial basolateral localization of these proteins is because of infidelity in apical targeting or is reflective of a physiological role for transcytosis is not known.
The data presented here show that the apical accumulation of the M 2 mAChR occurs subsequent to its basolateral localization. 35 S labeling and surface biotinylation assays of stably transfected MDCK cells demonstrate that the M 2 receptor accumulates at the basolateral surface starting 15 min after radioactive label is incorporated into newly synthesized receptor. The M 2 receptor only starts to appreciably accumulate over background levels at the apical surface ϳ30 min after receptors are synthesized, with the majority achieving apical localization 1 h after incorporation of the radiolabel. By immunocytochem-ical methods, the expression of newly transfected FLAG-tagged M 2 -GFP in MDCK cells is also first detected in the basolateral domain. Only at subsequent time points does apical surface expression of the M 2 receptor become detectable.
Polishchuk et al. (62) used domain-specific treatment with tannic acid to selectively block vesicle fusion and protein delivery to either the apical or basolateral domain of MDCK cells. Blocking fusion of transport vesicles to the apical domain resulted in the basolateral accumulation of the M 2 receptor, whereas blocking fusion of transport vesicles to the basolateral domain blocked the appearance of the receptor at both the basolateral and apical domains. These results are consistent with the results of both the metabolic labeling and time course analyses shown in Fig. 9 and provide further support for a requirement for basolateral transport and transcytosis prior to the ultimate transport of the M 2 mAChR to the apical membrane. Blocking basolateral fusion of cargo vesicles by tannic acid resulted in the lack of any surface expression of the M 2 receptor. These results suggest that basolateral surface expression is prerequisite to subsequent apical surface accumulation of the receptor and that there is a post-plasma membrane sorting mechanism for at least some apically targeted proteins in MDCK cells. We also attempted to determine whether the apical targeting determinant located within amino acids 270 -280 contains solely apical targeting information or whether this sequence was required for the basolateral transit of the receptor. Unfortunately, receptor expression of the M 2 C-terminal appendages to the M 4 mAChR was not sufficiently strong to observe pre-steady-state accumulation at either membrane domain at time points earlier than 24 h post-transfection. 3 The conclusions of the work of Polishchuk et al. (62) have recently been criticized by Paladino et al. (96), who suggested that membrane polarity is not fully achieved at 3 days in culture and that treatment with tannic acid can interfere with segregation of the apical and basolateral domains. However, our experiments reveal that polarization is achieved within the limits of our time course by the apical and basolateral localization of the M 2 receptor and ␤-catenin, respectively, as shown in Fig. 11C. Furthermore, the metabolic labeling and domain-specific biotinylation experiments were performed on stably transfected cells Ն5 days after initiation of polarization, and provide independent confirmation of the transcytosis of the M 2 receptor.
In conclusion, the results presented here demonstrate the M 2 mAChR is an apically targeted protein that achieves its localization independent of raft association and that basolateral localization is prerequisite to apical accumulation. This is the first reported example of a seven-transmembrane receptor that is predominantly localized to the apical surface by undergoing transcytosis from the basolateral domain in MDCK cells. This may represent a model mechanism utilized by other proteins for their localization in MDCK cells as well as other polarized cell types.