Signal-dependent trafficking of beta-amyloid precursor protein-transferrin receptor chimeras in madin-darby canine kidney cells.

We have investigated the intracellular trafficking of a chimeric molecule consisting of the cytoplasmic domain of the beta-amyloid precursor protein (APP) and the transmembrane region and external domain of the human transferrin receptor (TR) in Madin-Darby canine kidney cells. Newly synthesized APP-TR chimeras are selectively targeted to the basolateral surface by a tyrosine-dependent sorting signal in the APP cytoplasmic tail. APP-TR chimeras are then rapidly internalized from the basolateral surface and a significant fraction ( approximately 20-30%) are degraded. Morphological studies show that APP-TR chimeras internalized from the basolateral surface are found in tubulo-vesicular endosomal elements, internal membranes of multivesicular bodies, and lysosomes. APP-TR chimeras are also found in 60-nm diameter vesicles previously shown to selectively deliver wild-type TR to the basolateral surface; this result is consistent with the fact that 90% of internalized chimeras that are not degraded are selectively recycled back to the basolateral surface. APP-TR chimeras internalized from the apical surface are selectively transcytosed to the basolateral surface underscoring the importance of basolateral sorting in the endocytic pathway for maintaining the polarized phenotype. Tyr-653, an important element of the YTSI internalization signal in the APP cytoplasmic domain, is required for basolateral sorting in the biosynthetic and endocytic pathways. However, the structural features for basolateral sorting differ from those required for internalization.

Alzheimer's disease (AD) 1 is a progressive neurodegenerative disorder affecting ϳ1-6% of people over the age of 65. A characteristic neuropathological feature of AD is the senile plaque, which contains ␤-amyloid (A␤), a 39 -43 amino acid peptide derived from the ␤-amyloid precursor protein (APP), a type I integral transmembrane protein whose cellular function is unknown (1,2). Located on chromosome 21, the gene encod-ing APP is alternatively-spliced, and several isoforms of differing amino acid lengths are expressed. Genetic studies have revealed mutations in the coding region of APP genes isolated from a small minority of affected individuals with early onset familial AD, and it has been shown that one of these mutations, the Swedish double mutation, leads to secretion of elevated levels of A␤ (3,4). These observations led to the hypothesis that increased or abnormal A␤ production plays a central role in the pathogenesis of AD. While it is clear that generation of A␤ necessarily involves two proteolytic cleavages of APP: one in the extracellular domain and another in the transmembrane region, how APP is converted to A␤ in terms of proteolytic enzymes involved, cellular location of these cleavage events, and factors regulating its production remain to be determined (2). An alternative proteolytic cleavage by another unidentified enzyme (referred to as ␣-secretase) thought to reside on or near the plasma membrane leads to production of a large soluble fragment (APP s ) comprised of most of the APP extracellular domain (5,6). Significantly, this proteolysis occurs within the A␤ peptide, thus precluding its formation.
More recently, genetic studies have shown that mutations in two other related integral membrane proteins, presenilin 1 (PS-1) and presenilin-2 (PS-2), are associated with a much larger fraction of familial AD cases (7,8). These multitransmembrane-spanning proteins have a broad tissue distribution but little is known about their normal biological roles. A significant fraction of presenilin molecules are found in the endoplasmic reticulum; and at steady-state, the most abundant forms of the molecules appear to be ϳ30-kDa (NH 2 -terminal) and ϳ20-kDa (COOH-terminal) proteolytic fragments. A clue to the role of PS-1 and PS-2 in the pathogenesis of AD is that expression of presenilins containing naturally-occurring familial AD mutations seems to increase the amount of larger 42-43-amino acid forms of A␤ generated (8). These data support the idea that presenilins influence the metabolism of APP. However, whether PS-1 or PS-2 directly regulate the proteolytic processing of APP analogous to the regulation of the proteolytic processing of sterol regulatory element-binding proteins by a cleavage activating protein, SCAP (9), or by indirect effects on APP trafficking or proteolysis is unclear.
Analysis of APP trafficking in nonpolarized cells indicates that APP is transported to the cell surface via the constitutive biosynthetic pathway where it can be either cleaved by ␣-secretase to generate APP s or internalized and eventually degraded in lysosomes to yield amyloidogenic peptide fragments (10 -12). Studies of APP-TR chimeras suggest that the APP cytoplasmic tail contains two active internalization signals, GYENPTY and YTSI, and that both are involved in lysosomal targeting (13). Several independent studies of the trafficking of APP in MDCK cells, a well characterized polarized epithelial cell line, have shown that APP s and A␤ are preferentially released into the basolateral medium (14 -16). One group also demonstrated that at steady-state Ͼ80% of plasma membrane-associated APP resides on the basolateral surface (14) and subsequently showed that Tyr-653, the tyrosine residue of the YTSI internalization signal was important for basolateral expression (17). However, based on the observation that APPs derived from a tail-less mutant of APP was still secreted into the basolateral medium, De Strooper et al. (15) concluded that the most important basolateral sorting signal was located in the extracellular domain. Later, however, the same group reported that the cytoplasmic tail of APP was sufficient to target a heterologous protein to the basolateral surface (18). None of these studies addressed the question of whether basolateral sorting of APP occurred in the biosynthetic or endocytic pathways or both.
In this study, we have employed quantitative assays to analyze polarized trafficking of APP-TR chimeric molecules in MDCK cells in the absence of competing processing events such as APP s release. We show that polarized sorting of APP-TR in both the biosynthetic and endocytic pathways is important for its steady-state surface distribution. The basolateral sorting signal in the APP cytoplasmic tail is dependent on the tyrosine residue of the YTSI internalization signal but is not identical to this signal.

MATERIALS AND METHODS
APP-TR Chimeric Constructs-Generation of chimeric APP-TR by polymerase chain reaction was described previously (13). Mutant APP-TR chimeras were created by oligonucleotide-directed mutagenesis by the method of Kunkel (20). Briefly, mutants were screened and selected by restriction mapping and cloned into BH-RCAS(A), a retroviral expression vector derived from the Rous sarcoma virus, subtype A (RSV(A)) (21). All mutations were verified by dideoxynucleotide sequencing of the entire cytoplasmic domain of constructs in BH-RCAS(A) (22,23) using the Sequenase kit (U. S. Biochemical Corp., Cleveland, OH) according to the manufacturer's directions.
Expression of APP-TR Chimeras in MDCK Cells-To isolate recombinant virus for infection of MDCK cells, tissue culture medium from a confluent 10-cm dish of CEF was collected, passed through a 0.45-m filter, and centrifuged at 23,000 rpm for 2.5 h at 4°C in a Beckman SW40 Ti rotor. Viral pellet was resuspended in 1 ml of serum-free DME and passed through a 0.45-m filter. A clone of MDCK (strain II) cells which stably express the RSV(A) receptor (19) was plated the day before infection at a density of ϳ2 ϫ 10 4 cells/well in a Costar 24-well tissue culture plate (Costar Corp., Cambridge, MA). MDCK cells were maintained in DME supplemented with 10% (v/v) defined bovine calf serum. Infection of cells was achieved by incubation with 0.25-0.50 ml of concentrated recombinant virus for 12 h at 37°C, followed by addition of 1 ml of growth medium. Cells were then allowed to grow to confluency (roughly 3 days) without replacing media, and expression of APP-TR chimeras in MDCK cells was analyzed by indirect immunofluorescence using B3/25, a mouse monoclonal antibody against the external domain of the human TR (25) and a fluorescein isothiocyanate-conjugated goat anti-mouse secondary antibody (Cooper Biomedical, Malvern, PA). For subsequent trafficking analyses, several individual clones of MDCK cells expressing each of the various chimeras were isolated by limiting dilution.
Determination of the Distributions of APP-TR Chimeras on the Surface of Filter-grown MDCK Cells-Individual clones of MDCK cells expressing each chimera were seeded at high density (1.0 -1.5 ϫ 10 6 cells/filter) onto duplicate 24-mm Costar Transwell polycarbonate filters having a pore size of 0.4 m (Costar Corp., Cambridge, MA) and cultured for 3-5 days with media replaced on alternate days. Diferric human Tf (ICN Biomedicals, Costa Mesa, CA) was labeled with 125 I to a specific activity of 2-4 Ci/g using Enzymobeads (Bio-Rad) according to the manufacturer's directions. The day before the assay, 10 mM sodium butyrate (MCB, Inc., Cincinnati, OH) was included in the growth media to increase surface expression (26). The next day, cells were first incubated in serum-free DME containing 10 mM sodium butyrate for 1 h at 37°C, then shifted to 4°C, and washed once with ice-cold 0.5% BSA-PBS ϩ (PBS with 1 mM CaCl 2 and 1 mM MgCl 2 ). Cells were then incubated for 1 h at 4°C with BSA-PBS ϩ containing 4 g/ml 125 I-labeled Tf added to either the apical or the basolateral side of the monolayer and BSA-PBS ϩ added to the opposite side. Integrity of each monolayer was confirmed by measuring the amount of radioactivity in the opposing nonlabeling medium after 1 h at 4°C. Less than 1% of the 125 I-labeled Tf diffused through the intact monolayer. Unbound Tf was removed by washing monolayers 3 times at 4°C with ice-cold BSA-PBS ϩ , and filters were excised and counted in a ␥-counter to determine the amount of radioactivity bound to the cell surface. Nonspecificallybound Tf, determined by measuring the amount of radioactivity bound to either surface of uninfected MDCK cells, was subtracted as background. Two independent clones were analyzed for each APP-TR chimera.
Analysis of Tf Internalization of APP-TR Chimeras in MDCK Cells-The steady-state distribution of chimeras at 37°C was performed as described previously (27). Individual clones of MDCK cells expressing various APP-TR chimeras were plated in triplicate wells of Costar 24-well tissue culture plates at a density of 1.5 ϫ 10 4 cells/well. MDCK cells were grown overnight with the inclusion of 10 mM sodium butyrate in the media. The cells were preincubated in serum-free DME containing 10 mM sodium butyrate for 1 h and then incubated with 8 g/ml 125 I-labeled Tf in BSA-PBS for 1 h at 37°C. After aspiration of labeling medium, the cells were washed three times with 1 ml of ice-cold BSA-PBS, incubated twice for 3 min with 0.5 ml of 0.2 M acetic acid, 0.5 M NaCl (pH 2.4) to remove surface-bound 125 I-labeled Tf (28), and stripped from the wells with 1 M NaOH. More prolonged incubation with the acid wash did not change the amount of 125 I released. Radioactivity in the acid wash (surface-bound Tf) and radioactivity in the cell lysate (internalized Tf) were determined. Apparent internalization efficiencies were calculated relative to wild-type APP-TR chimera as described previously (13).
Tryptic Analysis of Surface Delivery of Newly Synthesized Chimeras in Filter-grown MDCK Cells-MDCK cells expressing APP-TR chimeras were seeded onto filters and grown as described for binding assays, with inclusion of 10 mM sodium butyrate in media 1 day prior to the experiment. Monolayers were preincubated for 20 min in methioninefree DME containing 10 mM sodium butyrate. Labeling of cells was performed from the basolateral side by placing filters over a 0.15-ml drop of methionine-free DME containing 0.4 mCi/ml Trans 35 S-label and 1% dialyzed fetal calf serum. 0.4 ml of methionine-free DME with 1% dialyzed fetal calf serum was added to the apical side to prevent cells from drying out. After incubation of cells at 37°C for 20 min, both surfaces of monolayers were washed 3 times with serum-free DME containing a 10-fold excess of methionine and cysteine, which was used in all subsequent steps requiring serum-free DME. 1 ml of serum-free DME was added to either surface, and monolayers were reincubated at 37°C for 30 min to allow a portion of newly synthesized chimeras to reach the plasma membrane.
To monitor cumulative levels of chimeras arriving at either surface during the 30-min chase period, 10 g/ml 1-tosylamide-2-phenylethyl chloromethyl ketone-treated trypsin (Worthington Biochemical Corp., Freehold, NJ) was included in either the apical or basolateral chase media, and 20 g/ml trypsin inhibitor isolated from chicken egg white (Sigma) was added to the opposite side. At the end of the chase period, cells were placed on ice, media from each surface, including an additional 0.3 ml of wash, was collected, and 20 g/ml trypsin inhibitor was added to samples containing trypsin. Media samples were spun to remove cellular debris which may contain uncleaved chimeras then the 70-kDa TR external domain tryptic fragment was immunoprecipitated, using the B3/25 monoclonal antibody directed against the human TR external domain (29). Immunoprecipitates were analyzed on 7.5% polyacrylamide gels, and gels were treated with Fluoro-Hance (Research Products International Corp., Mount Prospect, IL) according to the manufacturer's directions, dried, and exposed to X-AR film (Eastman Kodak, Rochester, NY) overnight. Quantitation of radioactivity was performed on a model 425 PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Alternatively, to determine the expression of newly synthesized protein at the cell surface after the 30-min chase, cells were cooled to 4°C and washed once with cold serum-free DME. Cells were incubated at 4°C for 30 min with serum-free DME containing 100 g/ml trypsin added to either surface with 100 g/ml trypsin inhibitor on the opposite surface. Media from each surface was collected and prepared for immu-noprecipitation as described. Monolayer integrity after either trypsinization condition was independently confirmed by the lack of detectable levels of the tryptic fragment in the media from the side opposite the surface exposed to trypsin, indicating that the tryptic fragment was not able to pass through the monolayer, consistent with previous use of this procedure (30).

Measurement of Recycling and Transcytosis of APP-TR Chimeras in Filter-grown MDCK Cells-MDCK cells expressing APP-TR and ATSA
chimeras were seeded onto filters and grown as described for binding assays, with inclusion of 10 mM sodium butyrate in media 1 day prior to the experiment. Monolayers were first incubated for 1 h at 37°C in serum-free DME containing 10 mM sodium butyrate, then incubated for 30 min at 37°C with 8 g/ml 125 I-labeled Tf in 0.5% BSA-DME added to either the apical (0.4 ml) or basolateral (0.15 ml drop) sides of the monolayers with BSA-DME added to the opposite side. Cells were then shifted to 4°C, and unbound Tf was removed by 3 washes with ice-cold BSA-DME. Surface-bound Tf was removed by incubating monolayers for 15 min at 4°C in 150 mM NaCl, 2 mM CaCl 2 , 20 mM sodium acetate buffer (pH 5.0) containing 50 M deferoxamine mesylate, followed by one wash with PBS ϩ and further incubation for 20 min at 4°C with PBS ϩ containing 50 M deferoxamine mesylate and 125 nM apo-Tf (27). Cells were then washed 3 times with ice-cold BSA-DME, and either the cell-associated radioactivity was determined immediately by excising filters and counting, or the monolayers were reincubated at 37°C for 90 min with prewarmed BSA-DME containing 100 g/ml unlabeled Tf. After the reincubation period at 37°C, the apical and basolateral media were collected, and the radioactivity which was released into each medium, as well as the radioactivity which remained cell-associated, was determined. Trichloroacetic acid soluble radioactivity in the media was determined by precipitation with 10% trichloroacetic acid for 30 min at 4°C. The total amount of radioactivity collected from monolayers reincubated at 37°C was similar to the amount bound to filters which had been counted immediately after surface stripping. Uninfected MDCK cells were treated in parallel, and corresponding values were subtracted as background from those of experimental cells. The efficiency with which surface-bound Tf was removed with deferoxamine mesylate was determined by incubating two sets of monolayers with 125 I-labeled Tf for 1 h at 4°C. For one set of monolayers, the radioactivity specifically bound to each surface was determined as described for the binding studies. The other set of monolayers were treated with deferoxamine mesylate, and the amount of radioactivity remaining bound was Ͻ10% of the total amount of radioactivity bound to untreated monolayers, indicating efficient surface-stripping by this procedure.
MDCK cells growing on the wall of the apical chamber of the Transwell filter unit have been reported to account for up to 40% of the total surface area accessible to the apical medium under standard growing conditions (31). To determine the contribution of these cells to the radioactivity released into the apical medium, monolayers were incubated apically with 125 I-labeled Tf, and filters were excised directly after washing (without 37°C reincubation period). Cells remaining associated with the unit after excision of the filter were lysed with 1 M NaOH and counted. Radioactivity associated with the filter unit after the 37°C reincubation period was subtracted, and the resulting value represented radioactivity released from cells growing on the chamber wall. This value, which was found to be ϳ25% of the total radioactivity collected in the apical media after reculture at 37°C, was subtracted from the apical media total.
Pulse-Chase Analysis of Degradation of APP-TR Chimeras in MDCK Cells-Approximately 2 ϫ 10 6 cells expressing each of the various APP-TR chimeras were plated on 6-cm tissue culture dishes and grown overnight. The following day, cells were washed twice with methioninefree DME, preincubated in methionine-free DME for ϳ20 min, and incubated for 30 min in 1.5 ml of methionine-free DME containing 0.12 mCi/ml Trans 35 S-label (ICN Biomedicals, Irvine, CA) and 1% dialyzed fetal calf serum. Pulse-labeled cells were chased for 0, 2, 4, or 8 h in growth medium made with DME containing a 10-fold excess concentration of methionine and cysteine. At each time point, labeled cells were placed on ice, washed three times with cold PBS, and solubilized with 1% Nonidet P-40/PBS. APP-TR chimeras were immunoprecipitated from postnuclear detergent extracts using B3/25. Immunoprecipitates were analyzed as described above.
Electron Microscopy-Tf-horseradish peroxidase and B3/25-horseradish peroxidase were prepared by conjugating 10 mg of Tf or B3/25 monoclonal antibody to 10 mg of horseradish peroxidase (type II, Sigma) using SPDP as described previously (32). Free horseradish peroxidase was removed from the conjugate using the FreeZyme conjugate purification kit (Pierce, Rockford, IL). 5 to 8-nm colloidal gold sols were made as described by Slot and Geuze (33). B3/25 monoclonal antibody was complexed to colloidal gold as described previously (32). Before incubation, gold complexes were washed by centrifugation in a Beckman airfuge at 150,000 ϫ g for 5 min.
For electron microscopy, cells were fixed in dilute Karnovsky fixative (2% paraformaldehyde, 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.5) (34), post-fixed in reduced osmium tetroxide, and embedded in Epon as described previously (28). Sections were either cut at 70 nm and stained with uranyl acetate and lead citrate or cut up to 1-m thick, stabilized with a thin film of evaporated carbon, and viewed unstained, in a Philips CM12 electron microscope. To increase the detection of APP-TR chimeras morphologically, monolayers were incubated in 10 mM sodium butyrate 12-16 h at 37°C prior to experimentation (35). Identical results were obtained in experiments performed without the inclusion of sodium butyrate.

APP Cytoplasmic Tail Mediates Basolateral Targeting of APP-TR Chimeras in MDCK Cells-It has been previously
shown that Ͼ80% of surface APP can be found on the basolateral surface of polarized MDCK cells (14). To investigate whether the APP cytoplasmic tail was sufficient to mediate basolateral targeting of a heterologous protein, we expressed chimeric APP-TR molecules (13) consisting of the APP cytoplasmic domain and the transmembrane region and extracellular domain of TR in MDCK cells and measured their surface distributions. Schematic diagrams of APP, TR, and chimeric APP-TR molecules are shown in Fig. 1a, with cytoplasmic tail amino acid sequences of APP, APP-TR, and mutant APP-TR chimeras provided in Fig. 1b. Stable expression of APP-TR chimeras was achieved using RCAS-BP(A), a replication-competent retroviral vector derived from RSV(A). Briefly, recombinant virus isolated from culture media of CEF transfected with RCAS-BP(A) constructs containing various APP-TR chimeras (13) was used to infect MDCK cells rendered susceptible to RSV(A) infection by stable introduction of the RSV(A) receptor as described previously (19).
Analysis of the surface distributions of APP-TR chimeras was performed on monolayers of cloned MDCK cell populations grown on 24-mm Costar Transwell filters (pore size 0.4 m) by binding 125 I-labeled Tf at 4°C to either the apical or basolateral surface as described under "Materials and Methods." Surface expression of APP-TR chimeras was increased by addition of 10 mM sodium butyrate to culture media 1 day prior to the experiment (16,26,36). As shown in Fig. 2, Ͼ80% of surface APP-TR chimeras were found on the basolateral domain, in contrast to tail-less TR molecules which were distributed almost equally on both surfaces as described previously (19). These results show that the APP cytoplasmic domain is sufficient to target APP-TR chimeras to the basolateral surface of MDCK cells.
Basolateral Targeting of APP-TR Chimeras Is Dependent on the Tyrosine of the YTSI Internalization Signal-Tyrosine residues have been shown to be important for the activity of several previously identified basolateral sorting signals (37). The APP cytoplasmic tail contains two tyrosine residues demonstrated to be important for the activity of the previously identified internalization signals, YTSI and GYENPTY (13). Therefore, to determine whether these tyrosine residues were required for basolateral targeting of APP-TR chimeras, we measured the surface distributions of mutant APP-TR chimeras in which these tyrosines were changed to alanines. As shown in Table I, alteration of YTSI to ATSI resulted in a sharp decrease in basolateral expression of APP-TR chimeras from ϳ80% basolateral to ϳ30% basolateral. However, alteration of both the tyrosine residues along with the glycine in the GY-ENPTY signal to AAENPTA resulted in only a slight decrease in basolateral expression. These results clearly demonstrate that selective expression of APP-TR chimeras on the basolateral surface is specifically dependent on the tyrosine residue of the YTSI internalization signal.

Trafficking of APP-TR Chimeras in MDCK Cells
Sequence Requirements for Basolateral Sorting Are Distinct from Those for Internalization-To compare the sequence requirements for basolateral sorting and rapid internalization, we performed alanine scanning mutations within the 4-residue internalization signal, YTSI, and measured effects on both polarized surface distribution and internalization. As shown in Table I, analysis of surface distributions of filter-grown MDCK cells expressing mutant APP-TR chimeras demonstrated that only the tyrosine appears critical for basolateral sorting, since independent alteration of each of the other 3 residues to alanine did not have significant effects on the surface distribution of APP-TR molecules.
Internalization efficiencies of mutant APP-TR chimeras expressed in MDCK cells grown on 24-well tissue culture plates were determined by measuring the steady-state distribution of 125 I-labeled Tf internalized for 60 min at 37°C as described under "Materials and Methods." Results of these assays are shown in Table II and show that mutation of either the tyrosine (ATSI) or isoleucine (YTSA) to alanine reduced the internalization efficiency to half that of wild-type APP-TR. Either mutation resulted in almost complete inactivation of this signal since mutation of both residues simultaneously (ATSA) did not lead to a significant additional decrease in internalization. In contrast, independent alteration of either the threonine (YASI) or serine (YTAI) residues in the YTSI signal to alanine did not decrease the internalization efficiency. Virtually identical internalization efficiencies were obtained for these mutant chimeras expressed in CEF (data not shown). The internalization data are consistent with the consensus sequence of 4-residue tyrosine-based internalization signals exemplified by the TR signal, YTRF, which require an aromatic residue at the first position and a bulky hydrophobic residue at the fourth position (38).   It is noteworthy that the ATSI and ATSA mutant APP-TR chimeras are not randomly distributed in the same manner as the tail-less TR mutant but are selectively expressed on the apical surface. Because of the complexity of APP-TR trafficking in MDCK cells, it is difficult to interpret this result. However, newly synthesized ATSA mutant molecules are delivered to the apical and basolateral surfaces in an essentially random fashion suggesting that the selective apical expression of the ATSI and ATSA mutant chimera only occurs in the endocytic pathway (see below).
Sorting of APP-TR Molecules to the Basolateral Surface Occurs at an Intracellular Site Along the Biosynthetic Pathway-Most current models of polarized sorting in MDCK cells propose that the predominant mechanism for selective expression of membrane proteins on the basolateral surface involves signaldependent sorting of newly synthesized molecules in the TGN (37). To investigate whether newly synthesized APP-TR chimeras are selectively delivered to the basolateral surface of MDCK cells, we pulse-labeled filter-grown cells expressing APP-TR chimeras with Trans 35 S-label and monitored the appearance of the chimeras on the apical and basolateral surfaces by applying trypsin to either surface and recovering the released 70-kDa TR external domain tryptic fragment by immunoprecipitation as described under "Materials and Methods." Trypsin was added either during the 30-min chase period at 37°C to allow the relative determination of cumulative amounts of chimeras appearing on the surface, or after the chase for an additional 30 min at 4°C to determine the surface distribution of newly synthesized chimeras at that time. The monolayer integrity of each filter was independently confirmed for each cleavage protocol by immunoprecipitating the tryptic fragment from the media opposite the side containing trypsin to assess its leakage. As shown in Fig. 3, ϳ90% of newly synthesized APP-TR chimeras delivered to the cell surface are sorted to the basolateral surface as determined by either trypsin cleavage protocol, indicating that newly synthesized APP-TR chimeras are initially sorted at an intracellular site along the biosynthetic pathway. The lack of detectable levels of the 70-kDa tryptic fragment in the opposite chamber to which trypsin was added (Fig. 3, see Trypsin (Ϫ) lanes) confirmed that the monolayer remained intact through the duration of either trypsin cleavage protocol (30).
To determine whether the tyrosine-based signal identified earlier was required for sorting of newly synthesized chimeras, we performed similar studies on cells expressing the ATSA mutant chimera and found that only ϳ40% of the newly synthesized mutants were directly delivered to the basolateral surface. This result demonstrates that sorting along the biosynthetic pathway requires either the tyrosine and/or isoleucine of the YTSI sorting signal and that inactivation of the basolateral sorting signal leads to an essentially random delivery of APP-TR chimeras to the apical and basolateral surfaces.
Basolateral Targeting of APP-TR Also Occurs Along the Endocytic Pathway-Since a majority of wild-type APP-TR chimeras are rapidly internalized and recycled, a sorting mechanism in the endocytic pathway must operate to return them to the basolateral surface to maintain the selective basolateral expression. To examine the fate of internalized chimeras the endocytic pathway of filter-grown MDCK cells expressing APP-TR chimeras was loaded from either the apical or basolateral surface with radiolabeled Tf. After stripping off surfacebound 125 I-labeled Tf to follow the trafficking of internalized Tf-chimeric complexes only, we measured trichloroacetic acid soluble and insoluble radioactivity released into the media bathing either surface during a 90-min chase period to evaluate recycling and transcytosis of the Tf-chimeric complexes. Under these conditions, Ͼ90% of the cell-associated radioactivity was released into the medium. These experiments demonstrate that rapid internalization of chimeras occurs from both surfaces and that ϳ20 -30% of internalized chimeras are degraded as determined by measuring the trichloroacetic acid soluble radioactivity in the media (data not shown). The degradation of APP-TR chimeras internalized from the cell surface is consistent with our previous report showing that a fraction of these chimeras is also degraded by this pathway in CEF (13). As shown in Fig. 4a, undegraded Tf-chimeric complexes internalized from the basolateral surface are recycled back to the same surface with high fidelity as ϳ90% of the intact 125 I-labeled Tf is released into the basolateral chamber. As shown in Fig. 4b, APP-TR chimeras internalized from the apical surface are selectively sorted by a transcytotic pathway to the basolateral surface as ϳ70% of the intact 125 I-labeled Tf loaded from the apical surface is released into the basolateral medium.
To evaluate the role of the tyrosine-based YTSI signal in polarized sorting within the endocytic pathway, we performed similar studies on cells expressing the ATSA mutant chimera. We found that, in contrast to APP-TR, only ϳ50% of basolaterally loaded ATSA mutant chimeras (Fig. 4c) are recycled back to the basolateral surface whereas apically loaded mutant APP-TR chimeras were selectively returned to the apical surface (Fig. 4d).
These results demonstrate that the polarized steady-state

FIG. 3. Surface delivery of newly synthesized APP-TR chimeras in filter-grown MDCK cells.
Filter-grown MDCK cells expressing APP-TR chimeras or ATSA mutant chimeras were pulse-labeled from the basolateral side with Trans 35 S-label for 20 min. Filters were washed and reincubated at 37°C for 30 min allowing newly synthesized chimeras to reach the cell surface. Surface delivery was monitored by two trypsin cleavage protocols: 10 g/ml trypsin was included in apical (A) or basolateral (B) media during the 30-min chase at 37°C or 100 g/ml trypsin was added after the chase for 30 min following shift of cells to 4°C. The 70-kDa TR external domain tryptic fragment was then immunoprecipitated from media and analyzed on SDS-polyacrylamide gels as described under "Materials and Methods." Dried gels were exposed to X-AR film overnight (Eastman Kodak, Rochester, NY). Immunoprecipitates were quantitated on a model 425 PhosphorImager (Molecular Dynamics). An autoradiograph from a representative experiment is shown, and the quantitative data shown is the mean Ϯ S.D. of results from three independent experiments in which two different clones of MDCK cells were analyzed for each chimera. distribution of APP-TR chimeras on the basolateral surface is maintained by an efficient sorting mechanism within the endocytic compartment and that the signal recognized by this sorting machinery is colinear with the YTSI internalization signal.
Both YTSI and GYENPTY Are Required for Rapid Degradation of APP-TR Chimeras in a Post-Golgi Endocytic Compartment-The recovery of significant levels of trichloroacetic acid soluble radioactivity in the experiments described above demonstrated that a significant fraction of APP-TR chimeras expressed on the cell surface were targeted to a prelysosomal/ lysosomal compartment in MDCK cells. To confirm that APP-TR chimeras were rapidly degraded in MDCK cells, we performed metabolic pulse-chase experiments to determine the rate of degradation of APP-TR chimeras. MDCK cells expressing APP-TR chimeras were pulse-labeled with Trans 35 S-label for 30 min and chased for various lengths of time; APP-TR chimeras were then isolated by immunoprecipitation and analyzed by SDS-polyacrylamide gel electrophoresis. As shown in Fig. 5, the APP-TR chimera was rapidly degraded (t1 ⁄2 ϭ ϳ2 h) and Ͻ10% of pulse-labeled molecules remained intact after an 8-h chase, clearly demonstrating, as in the case of CEF (13), that the APP cytoplasmic tail targets TR for rapid degradation in MDCK cells. The increased M r of APP-TR seen after 2 h results from passage through the Golgi compartment where oligosaccharide processing and synthesis are completed (29) and indicates that degradation occurs in a post-Golgi prelysosomal/lysosomal compartment.
To determine whether YTSI or GYENPTY sequences were required for targeting to a prelysosomal/lysosomal compartment in MDCK cells, we determined the degradation rates of mutant APP-TR chimeras in which important residues in each of the signals were altered to alanine. As shown in Fig. 5, inactivation of each sequence independently (ATSI or AAENPTA) resulted in only modest decreases in degradation rate, whereas alteration of both sequences in tandem (ATSI/ AAENPTA) resulted in a much more dramatic increase in half-life. These results clearly indicate that both tyrosine-based signals are required for efficient lysosomal targeting of APP-TR chimeras in MDCK cells.
Morphological Studies of APP-TR Chimera Trafficking in the Endocytic Pathway of MDCK Cells-For morphological studies of APP-TR trafficking in polarized MDCK cells, Tf-horseradish peroxidase conjugates and B3/25-gold complexes were used as tracers to delineate the endocytic pathways followed by internalized chimeras as described previously for wild-type TR (19). With Tf-horseradish peroxidase or B3/25-gold complexes applied from the basolateral surface, a broad variety of morphologically distinct intracellular membrane structures are labeled (Fig. 6). As shown in the low power view presented in Fig.  6a, Tf-horseradish peroxidase is found in tubular and vesicular endosomal elements throughout the cell. Occasionally, labeled coated 100-nm vesicles can be seen which probably represent vesicles derived from clathrin-coated pits pinching off the basolateral surface (inset, Fig. 6c). In addition to labeled tubulovesicular endosomal elements, typical multivesicular endosomes are also labeled with both Tf-horseradish peroxidase (Fig. 6c) and B3/25-gold (Fig. 6b). Tracers bound to APP-TR chimeras are associated with the internal membrane structures of the multivesicular vesicles as seen most clearly in the high power view of the multivesicular vesicle labeled with B3/25-gold tracer shown in Fig. 6b. In contrast, when these tracers are taken up by MDCK cells expressing wild-type TR, few multivesicular vesicles are labeled with labeling restricted to only the perimeter membrane (19). The labeling of the internal membranes of multivesicular bodies almost certainly represents APP-TR chimeras en route to lysosomes where they are degraded. Consistent with this view, classical dense lysosomes can be seen to be heavily labeled with Tf-horseradish peroxidase as shown in Fig. 6d. In addition, 60-nm diameter vesicles that are often found lying adjacent to the basolateral border (Fig. 6c) are also labeled with Tf-horseradish peroxidase tracer. Although uptake of tracers by MDCK cells expressing APP-TR chimeras from the apical surface was too low to delineate intracellular pathways from the apical surface, it has been shown in MDCK cells expressing wild-type TR that such 60-nm diameter structures function as exocytotic transport vesicles and contain both transcytosing Tf-horseradish peroxidase tracer (apically internalized) and recycling Tf-horseradish peroxidase tracer (basolaterally internalized). It is probable, therefore, that the 60-nm diameter vesicles labeled with tracer in cells expressing APP-TR chimeras perform the same function and selectively recycle undegraded chimeras internalized from the basolateral surface back to the same surface. DISCUSSION In this study, we have characterized the trafficking of APP-TR chimeras in MDCK cells with the objective of defining the role of the APP cytoplasmic tail in signal-dependent polarized trafficking. This approach only gives information about sorting signals in the cytoplasmic domain of APP, but has the major advantage that quantitative biochemical assays can be employed to study trafficking in both the biosynthetic and endocytic pathways of MDCK cells. Earlier studies showed that APP is selectively expressed on the basolateral surface of MDCK cells (14); however, whether newly synthesized APP were directly targeted to the basolateral surface was not determined. Although several groups showed that APP s generated from newly synthesized APP was selectively secreted into the basolateral medium (14 -16), subsequent work has shown that the polarized secretion of APP s occurs independently of the selective basolateral targeting of intact membrane-bound APP (17). Consequently, the basolateral secretion of APP s cannot be used as evidence that newly synthesized APP is directly targeted from the biosynthetic pathway to the basolateral surface. Thus, the results we report here showing that the polarized phenotype of APP-TR chimeras is generated by selective sorting of newly synthesized molecules from the biosynthetic pathway to the basolateral surface provide the first direct evidence that the cytoplasmic domain of APP contains a basolateral sorting signal that is recognized in the biosynthetic pathway of MDCK cells.
Our results also provide strong evidence that polarized trafficking of APP along the endocytic pathway is important for maintenance of the steady-state basolateral expression of APP observed in earlier studies (14), and represents a potentially significant issue not previously addressed in earlier reports. After delivery to the basolateral surface, APP-TR chimeras are rapidly internalized with ϳ50% of surface receptors residing in the endocytic pathway at steady state. The small fraction of APP-TR chimeras displayed on the apical surface are also internalized and enter the endocytic pathway. Regardless of whether they originated from the basolateral or apical surface, internalized APP-TR chimeras share a common fate. Although a significant fraction of the internalized APP-TR chimeras (20 -30%) is degraded as was previously shown for APP-TR chimeras expressed in CEF (13), the majority are selectively sorted to the basolateral surface. The efficiency with which undegraded APP-TR chimeras internalized from the basolateral surface were recycled back to the same surface was ϳ90%, a value similar to that reported for recycling of wild-type TR back to the basolateral surface of MDCK cells (19). The efficiency of transcytosis of APP-TR from the apical to basolateral surface was somewhat lower (ϳ70%) and may reflect signal-independent recycling of apically internalized chimeras from endosomal elements proximal to the apical surface. Evidence for such a recycling pathway was obtained from studies of the apical to basolateral transcytosis of human TR in MDCK cells that occurs with approximately the same efficiency (19).
Morphological studies employing Tf-horseradish peroxidase and B3/25-gold as tracers confirmed that APP-TR chimeras are internalized from the basolateral surface as these tracers were found in a variety of intracellular membrane structures including tubulo-vesicular endosomal elements extending into the apical cytoplasm, internal membranes of multivesicular bodies, lysosomes, as well as 60-nm diameter vesicles. These observations can be interpreted in light of recent electron microscopic studies of wild-type TR trafficking in polarized MDCK cells using Tf-horseradish peroxidase and B3/25-gold as tracers. In these studies, which additionally took advantage of the strong signal of a TR-horseradish peroxidase chimera in which the extracellular domain of the receptor was replaced with horseradish peroxidase, it was shown that the apical and basolateral endosomes of MDCK cells are extensively interconnected and that these endosomal elements are readily accessible to receptors internalized from either the basolateral or apical surface (19). It was also established that recycling TRs are delivered from endosomes to the basolateral border by a distinctive subset of 60-nm diameter exocytotic vesicles.
In this study, APP-TR chimeras were found in similar tubulo-vesicular endosomal elements and 60-nm diameter vesicles suggesting that the chimeras internalized from either the apical or basolateral surface enter the same interconnected endosomal system as TR and are selectively delivered to the basolateral surface by 60-nm diameter transport vesicles. However, experiments to directly establish this point were not feasible because of the relatively weak labeling obtained with the Tf-horseradish peroxidase and B3/25-gold tracers bound to internalized APP-TR chimeras compared with wild-type TR. In contrast to TR, APP-TR chimeras were also found on the internal membranes of multivesicular bodies and in mature lysosomes consistent with the biochemical evidence derived from MDCK cells and CEF (13) demonstrating that a significant fraction of internalized chimeras is degraded. Previous work has shown that internalized epidermal growth factor receptors destined for degradation in lysosomes are sorted from recycling receptors through spatial segregation in multivesicular bodies (39), and it appears that APP-TR chimeras are targeted to lysosomes by a similar process. Similarly, earlier studies by Yamazaki et al. (40) showed that a significant fraction of internalized APP was found in multivesicular bodies and lysosomes in nonpolarized cells. Although our morphological studies together with the finding that 20 -30% of Tf bound to internalized APP-TR chimeras is degraded establish that APP-TR chimeras traffic from the plasma membrane to lysosomes, it is also possible that a fraction of newly synthesized chimeras are targeted to lysosomes by a direct intracellular route as described for some lysosomal membrane glycoproteins and the major histocompatibility complex class II invariant chain (41)(42)(43)(44).
We have also analyzed the APP cytoplasmic sorting signals that determine the trafficking of APP-TR chimeras in MDCK cells. Previous studies of the steady state distribution of wildtype and mutant APP molecules on MDCK cells had established that the basolateral expression of APP was dependent on Tyr-653 (17), the tyrosine residue that is important for the activity of the YTSI internalization signal located in the APP cytoplasmic domain (13). However, whereas Haass et al. (17) found that mutant APP molecules lacking most the cytoplasmic domain were randomly distributed on the apical and basolateral surfaces of polarized MDCK cells, De Strooper et al. (15) concluded that the major basolateral sorting signal of APP was located in the external domain of the molecule based on the polarized secretion of APP s . Although De Strooper et al. (18) subsequently reported that the cytoplasmic domain of APP was sufficient to target a heterologous protein to the basolateral surface of MDCK cells, the sorting signals present in the cytoplasmic domain of APP that determine trafficking of the molecule in MDCK cells remain poorly defined. We have shown here that the cytoplasmic domain of APP contains a basolateral sorting signal that is dependent on Tyr-653, confirming the work of Haass et al. (17). We have also shown that the APP basolateral sorting signal is distinct from the YTSI internalization signal because its activity is not dependent on the isoleucine residue. Finally we have shown that efficient targeting of APP-TR chimeras to lysosomes requires that both tyrosine-based internalization signals, YTSI and GYENPTY, in the APP cytoplasmic domain be active in accordance with re-sults of degradation of APP-TR chimeras in CEF (13).
In summary, we have shown that the cytoplasmic domain of APP contains a tyrosine-dependent basolateral sorting signal that is active in the basolateral and endocytic pathways of MDCK cells. APP-TR chimeras are targeted directly from the biosynthetic pathway to the basolateral surface of MDCK cells but basolateral sorting of chimeras in the endocytic pathway is important for maintenance of the polarized phenotype. The discovery that mutations in presenilins are associated with a majority of familial Alzheimer's cases and that these molecules are expressed predominantly in the endoplasmic reticulum has currently led to efforts to determine whether A␤ production can be generated in the endoplasmic reticulum under direct regulation by presenilins. However, the relationship between presenilins and A␤ production is still unclear as is the role of APP trafficking. Studies of the basic cell biology of both molecules may lead to a better understanding of the relationship between the two molecules and their physiological roles.