The Vesicular Acetylcholine Transporter Interacts with Clathrin-associated Adaptor Complexes AP-1 and AP-2*

In neuronal cells the neurotransmitter acetylcholine is transferred from the cytoplasm into synaptic vesicles by the vesicular acetylcholine transporter (VAChT). The cytoplasmic tail of VAChT has been shown to contain signals that direct its sorting and trafficking. The role of clathrin-associated protein complexes in VAChT sorting to synaptic vesicles has been examined. A fusion protein between the VAChT cytoplasmic tail and glutathione S-transferase was used to identify VAChT-clathrin-associated protein adaptor protein 1, adaptor protein 2 and adaptor protein 180 complexes from a rat brain extract. In vivo coimmunoprecipitation confirmed adaptin α and adaptin γ complexes, but adaptor protein 180 complexes were not detected by this technique. Deletion and site directed mutagenesis show that the VAChT cytoplasmic tail contains multiple trafficking signals. These include a non-classical tyrosine motif that serves as the signal for adaptin α and a dileucine motif that serves as the signal for adaptin γ. A classical tyrosine motif is also involved in VAChT trafficking, but does not interact with any known adaptor proteins. There appear to be two endocytosis motifs, one involving the adaptor protein 1 binding site and the other involving the adaptor protein 2 binding site. These results suggest a complex trafficking pathway for VAChT.

In neuronal cells the neurotransmitter acetylcholine is transferred from the cytoplasm into synaptic vesicles by the vesicular acetylcholine transporter (VAChT). The cytoplasmic tail of VAChT has been shown to contain signals that direct its sorting and trafficking. The role of clathrin-associated protein complexes in VAChT sorting to synaptic vesicles has been examined. A fusion protein between the VAChT cytoplasmic tail and glutathione S-transferase was used to identify VAChT-clathrin-associated protein adaptor protein 1, adaptor protein 2 and adaptor protein 180 complexes from a rat brain extract. In vivo coimmunoprecipitation confirmed adaptin ␣ and adaptin ␥ complexes, but adaptor protein 180 complexes were not detected by this technique. Deletion and site directed mutagenesis show that the VAChT cytoplasmic tail contains multiple trafficking signals. These include a non-classical tyrosine motif that serves as the signal for adaptin ␣ and a dileucine motif that serves as the signal for adaptin ␥. A classical tyrosine motif is also involved in VAChT trafficking, but does not interact with any known adaptor proteins. There appear to be two endocytosis motifs, one involving the adaptor protein 1 binding site and the other involving the adaptor protein 2 binding site. These results suggest a complex trafficking pathway for VAChT.
The neurotransmitter acetylcholine (ACh) 1 is synthesized in the cytosol of cholinergic nerve terminals, transported into synaptic vesicles, and secreted upon calcium influx triggered by an extracellular signal. The vesicular acetylcholine transporter (VAChT) transports this ACh from the cytoplasm into synaptic vesicles in exchange for two protons (1,2). VAChT is localized to the membrane of the small synaptic vesicle in central cholinergic neurons (3) and synaptic-like microvesicles (SLMV) and endosomes in PC12 cells (4 -6).
The trafficking of VAChT has been most thoroughly studied in PC12 cells where it is believed to involve a multistep process. It has been proposed, although not proven, that VAChT, along with synaptophysin (7), is trafficked from the trans-Golgi network (TGN) to the plasma membrane via the constitutive secretory pathway, then from the plasma membrane to an endosomal compartment, and finally from the endosomal compartment to SLMVs (6).
Membrane trafficking between organelles of the endocytic and secretory pathways is mediated by transport vesicles that shuttle between different compartments. Both the generation of transport vesicles and the selection of protein cargo for inclusion in these vesicles are dependent on the function of coat proteins attached to the cytosolic face of the appropriate membrane (8 -10).
Clathrin-coated vesicles represent the best characterized system of membrane protein trafficking in eukaryotic cells. The major components of clathrin-coated vesicles are clathrin and adaptor protein (AP) complexes. Clathrin provides the structural component while the adaptor protein complexes select the vesicle cargo and promote clathrin-lattice formation onto the respective membrane (11) and recruit accessory proteins to the site of vesicle formation. There are five known adaptor protein complexes associated with clathrin; adaptor proteins 1 through 4 (AP-1, AP-2, AP-3, and AP-4) and adaptor protein 180 (AP-180). The AP-1 complexes are associated with the TGN and are involved in the transport of proteins to the endosomal/lysosomal system and to the cell surface. AP-2 complexes are found at the plasma membrane and participate in the internalization of cell surface proteins (12)(13)(14), while AP-180 is found in synaptic vesicles of neuronal cells (15)(16)(17)(18).
Adaptor protein 3 (AP-3) is a more recently identified adaptor protein associated with the transport of proteins from the TGN and/or endosomes to intracellular vacuoles, and seems to be important in neuronal protein trafficking (19 -22). AP-3 is believed to function in the formation of synaptic vesicles from endosomes (23). However, AP-3 does not associate with clathrin (19 -22). Yet another recently described adaptor protein, AP-4, has no known function (24,25).
Selection of cargo membrane proteins by adaptor protein complexes is dependent on a "sorting signal," usually located in a cytoplasmic domain of the cargo membrane protein. There are two known sorting signals for interaction with clathrincoated vesicles, a tyrosine-based sorting signal, YXXØ or NPXY (where X is any amino acid and Ø is a large hydrophobic amino acid (leucine, isoleucine, phenylalanine, methionine, valine) and the dileucine-based sorting signal. Both AP-1 and AP-2 have been found to recognize tyrosine-based and dileucinebased sorting signals (26,27). AP-3 can also recognize the dileucine-sorting motif (28 -30).
It is believed that at least two different sorting machineries are necessary for the trafficking of newly synthesized VAChT from the TGN to its final destination on the synaptic vesicle. Sorting from the TGN to the plasma membrane is suggested to involve the constitutive secretory pathway, and would not require a special sorting system. However, endocytosis from the plasma membrane to an endosomal compartment and intracellular trafficking from the endosomal compartment to synaptic vesicles likely require sorting machineries. Although there is evidence that clathrin-mediated endocytosis contributes to the sorting of VAChT in PC12 cells, the steps and pathway underlying the sorting and trafficking of newly synthesized transporter has yet to be determined. We now report on the identification of clathrin-coated vesicles that are involved in VAChT trafficking.

EXPERIMENTAL PROCEDURES
DNA Constructs and Mutagenesis-A cDNA corresponding to the 60 amino acid cytoplasmic carboxyl tail of VAChT (residues 471-530), VAChT CTD , was generated by PCR and subcloned into the EcoR1/NotI restriction sites of the bacterial expression vector pGEX5X-3 by adding appropriate restriction sites to the PCR primers.
Wild-type rat VAChT, subcloned into the EcoRI and XbaI sites of pBluescript KS ϩ (Invitrogen), was used as a template for mutagenesis. Site-directed mutagenesis was performed using the QuikChange Sitedirected Mutagenesis kit (Invitrogen). To prepare deletion mutants a stop codon was inserted at the appropriate site of the mutagenic primer. To verify that only the desired mutation was introduced, the mutated portion was sequenced by the ThermoSequenase Radiolabeled Terminator Cycle Sequencing Kit (Amersham Biosciences). Full-length mutant VAChTs were subcloned into the mammalian expression vector pcDNA3 or pcDNA3.1 (Invitrogen), and the 60 amino acid cytoplasmic carboxyl tail of mutant VAChT was subcloned into pGEX5X-3.
Cell Culture and Preparation of Stable Transfectants-PC12 cells were cultured in 5% CO 2 at 37°C in RPMI 1640 media containing L-glutamine (Invitrogen), 10 mM HEPES, 1 mM sodium pyruvate, 0.45% glucose, 10% horse serum, 5% bovine calf serum, and 1% penicillin/streptomycin. Electroporation was used for transfection of wild type and mutant VAChT cDNAs into PC12 cells. For electroporation, cells were detached from plates with trypsin/EDTA, washed with ice-cold phosphate-buffered saline (PBS), and resuspended in 800 l of cold PBS at a cell density of ϳ6 ϫ 10 7 cells/ml. The resuspended cells were mixed with 50 g of plasmid DNA. After a 10-min incubation on ice, the cell-DNA mixture was transferred to a 0.4-cm gap cuvette (Bio-Rad), electroporated (0.2 kV, 975 microfarads) using a Bio-Rad Gene Pulser II, then replated in culture media and cultured for 48 -72 h. Stable transformants were selected with 0.5 mg/ml of Geneticin (Invitrogen) or 0.5 mg/ml of Zeocin (Invitrogen).
GST Fusion Protein Production and in Vitro Binding Assays-Wild type and mutant VAChT cytoplasmic tail constructs in pGEX5X-3 were transformed into E. coli BL21. 5 ml of bacteria, grown overnight in LB media containing 100 g/ml of ampicillin was transferred to 500 ml of 2ϫ YT media (1.6% tryptone, 1% yeast extract, 0.5% NaCl) containing ampicillin and cultured at 37°C an additional 3-5 h until the OD 600 reached 0.5-0.7. Protein expression was then induced by the addition of 0.1 mM isopropyl ␤-D-thiogalactoside (IPTG) for 1-2 h at 27°C. After induction bacteria were pelleted, washed, and resuspended in cold PBS containing a protease inhibitor mixture (Roche Applied Science) and 0.2 mM phenylmethylsulfonyl fluoride (Sigma). Bacteria were disrupted with a French Pressure Cell at 500 -1000 p.s.i. Cell debris was removed by centrifugation at 14,000 ϫ g for 30 min, and the resulting bacterial lysate was either used immediately or stored at Ϫ80°C until use.
GST fusion proteins were purified by incubation with glutathioneagarose beads (Sigma) for 2 h at 4°C followed by washing with cold PBS. In most experiments the fusion proteins were not eluted from the beads. The amount of bound fusion protein was quantitated with an aliquot after elution from the beads with Coomassie Plus protein assay reagent (Pierce). Rat brain extracts were prepared by homogenization of frozen rat brain (Pel-Freez Biologicals) using a Potter-Elvejhem homogenizer by several up and down strokes in Cytosol Buffer (25 mM Hepes, pH 7, 125 mM potassium acetate, 2.5 mM magnesium acetate, 1 mM dithiothreitol, and 0.1% glucose) containing protease inhibitor mixture (Roche Applied Science), 0.2 mM phenylmethylsulfonyl fluoride (Sigma), 10 mM EDTA, and 0.1% of CHAPS. The homogenate was then centrifuged at 100,000 ϫ g for 30 min and the resulting supernatant used immediately.
Preparation of Postnuclear Supernatants-Postnuclear supernatants were prepared as described in Kim et al. (31). Briefly, transfected cells were trypsinized and rinsed with ice-cold PBS containing 10 mM EDTA. Cell pellets were resuspended in homogenization buffer containing 0.32 M sucrose, 10 mM HEPES-KOH (pH 7.4), 4% protease inhibitor mixture (Roche Applied Science), and 0.2 mM phenylmethylsulfonyl fluoride. The cell suspension was homogenized on ice with a Potter-Elvejhem homogenizer by several up and down strokes within a 5-min period. Postnuclear supernatants were then collected by centrifugation at 800 ϫ g for 10 min. The protein concentration was measured with the Coomassie Plus protein assay reagent based on the Bradford method (Pierce).
Immunoprecipitation-1 ml of postnuclear supernatant (ϳ5 mg of total protein) was incubated with the cross-linking reagent 3,3-dithiobis-(sulfosuccinimidyl)propionate (DTSSP) at 5 mM for 2 h at 4°C. After cross-linking, 100 l of 10% Nonidet P-40 was added and insoluble material removed by centrifugation at 20,000 ϫ g for 20 min. The resulting cell lysate was subjected to immunoprecipitation. The cell lysate was first precleared with 30 l of protein G-agarose for 1 h at 4°C, then incubated with anti-VAChT antibody for 1 h, and then for an additional hour after adding protein G-agarose. Immunoprecipitates were collected by centrifugation and washed four times with homogenization buffer supplemented with 100 mM NaCl and 1% Nonidet P-40. The immunoprecipitated proteins were eluted with SDS sample buffer and analyzed by SDS-polyacrylamide gel electrophoresis and Western blotting.
Organelle Immunoisolation-Post-nuclear supernatants (500 l containing ϳ250 g of total protein) derived from wild type or mutant VAChT transfected PC12 cells were precleared with 10 l of mouse preimmune serum and 10 l of Dynabeads® M-450 (Dynal ASA, Oslo, Norway) for 1 h at 4°C. The precleared post nuclear supernatant was incubated with 2 g of monoclonal antibody to synaptophysin (Chemicon) for 1 h at 4°C. Dynabeads® M-450 (25 l bed volume) conjugated to rabbit anti-mouse antibody (Dynal ASA) was added and incubated for 2 h at 4°C followed by four washes in homogenization buffer. The immunoisolated vesicles were then extracted in SDS sample buffer and subjected to SDS-polyacrylamide gel electrophoresis and Western analysis.
Cell Fractionation (Linear Sucrose Gradient Centrifugation)-Postnuclear supernatant (500 l; ϳ5 mg of total protein) was loaded onto a 10-ml linear sucrose gradient from 0.6 to 1.6 M sucrose in 10 mM HEPES buffer (pH 7.4) and centrifuged at 30,000 rpm for 6 h in an SW40 Ti rotor in a Beckman LE-80 centrifuge at 4°C. Fractions of 500 l were collected from the bottom of the tube, 10 l of each fraction was mixed with 10 l of 2ϫ SDS sample buffer and subjected to SDS-polyacrylamide gel electrophoresis and Western blotting. A mouse monoclonal antibody against synaptophysin (Chemicon) or a rabbit antiserum against secretogranin II was used to detect synaptic-like microvesicles and large dense core vesicles, respectively. The latter antiserum was a generous gift from Dr. Jonathan Scammell, University of Alabama. A goat polyclonal antibody against VAChT (Chemicon) was used.

SDS-Polyacrylamide Gel Electrophoresis and Western Blot
Analysis-For the detection of mutant protein expression, 5 l of postnuclear supernatant (50 g of total protein) containing the mutant VAChT was diluted to 500 l with homogenization buffer and pelleted by centrifugation at 50,000 rpm for 1 h in a TLA-100 -3 rotor using a Beckman TL-100 ultracentrifuge. The pellet was resuspended in 20 l of SDS sample buffer and separated by 10% SDS-PAGE. Separated proteins were electrophoretically transferred to an Immobilon P membrane and blocked with Tris-buffered saline containing 10% nonfat dry milk and 0.1% Tween 20 for 1 h at room temperature. The membrane was sequentially incubated with primary antibody (goat anti-rVAChT antibody), then with peroxidase-conjugated porcine anti-goat antibody for 1 h each at room temperature. Between each incubation, the blot was washed three times for 10 min with Tris-buffered saline and 0.1% Tween 20. Immune complexes were visualized by enhanced chemiluminescence (ECL) (Amersham Biosciences).
Immunofluorescence Microscopy-For immunostaining, cells were plated onto glass coverslips coated with 5 g/ml laminin (Sigma), fixed for 20 min with cold methanol and blocked in PBS containing 5% bovine calf serum and 5% horse serum. The cells were then incubated with primary antibody in the same buffer for 20 min at room temperature, washed three times, incubated an additional 20 min at room tempera-ture with the appropriate secondary antibody in the same buffer, and washed again three times with blocking buffer and three times with PBS. The coverslip were mounted on glass slides with Vectorshield mounting solution. Cells were viewed on a Nikon Eclipse E600 Microscope. To detect the VAChT primary antibody, the secondary antibody was a fluorescein-conjugated sheep anti-goat antibody (Chemicon). For the monoclonal antibody to synaptophysin (Chemicon), Texas red-conjugated bovine anti-mouse antibody (Vector Laboratories) was used.

RESULTS
The C-terminal Cytoplasmic Tail of VAChT Interacts with AP Complexes-The cytoplasmic tail of a number of membrane proteins has been shown to contain information for targeting to various intracellular destinations (32)(33)(34)(35). The VAChT cytoplasmic carboxyl tail fits into this group as it has been shown to be involved in the targeting of this transporter to synaptic vesicles (35). This segment of VAChT contains two putative sorting motifs; a dileucine motif preceded by a phosphorylation site and two possible tyrosine-based sorting motifs, Fig. 1A.
To analyze for an interaction of the VAChT cytoplasmic tail with adaptor protein (AP) complexes, we initially utilized a fusion protein consisting of the carboxyl cytoplasmic domain of rVAChT fused to GST (GST-VAChT CTD ) to pull-down complexes from a rat brain extract. Thus GST-VAChT CTD was mixed with a rat brain extract, precipitated with glutathione agarose, washed, and subjected to SDS-PAGE followed by Western blot analysis. The resolved proteins were probed with antisera to adaptin ␣ (a subunit of the AP-2 complex), adaptin ␥ (a subunit of the AP-1 complex), ␤-NAP (a subunit of the AP-3 complex), and AP-180. As shown in Fig. 2, adaptin ␣, adaptin ␥ and AP-180, but not ␤-NAP were co-precipitated with GST-VAChT CTD . GST alone did not form any detectable complexes.
To confirm the interaction of the VAChT cytoplasmic tail with specific AP complexes we sought to demonstrate the presence of these complexes in vivo. Although PC12 cells express endogenous VAChT mRNA as detected by PCR, VAChT protein expression is too low to be detected by immunoprecipitation and Western blot analysis. Therefore, as done in other studies (4,35) full-length VAChT was stably transfected into PC12 cells to produce PC12 VAChT . In general cargo protein-adaptor protein complex interactions are transient and unstable. A post-nuclear supernatant prepared from PC12 VAChT was therefore treated with the homobifunctional cross-linking reagent DTSSP for 2 h at 4°C to fix complexes. After lysis of the vesicles, samples were immunoprecipitated with anti-VAChT antisera and subjected to SDS-PAGE followed by Western blot analysis. This procedure led to the identification of adaptin ␣ and adaptin ␥ complexes, Fig. 3. No complexes or VAChT protein were detected with non-transfected PC12 cells, nor was an AP-180 complex detected by this procedure. Since PC12 cells contain far less AP-180 than rat brain (data not shown), this experiment cannot exclude VAChT interacting with AP-180 in neurons.
Analysis of VAChT Deletion Mutants-Based on the experiments described above, it was possible to demonstrate that the VAChT C-terminal cytoplasmic tail forms specific AP complexes. We next set out to identify the sequence(s) within the C-terminal tail of VAChT that interacts with each AP complex. As noted above and shown in Fig. 1, VAChT contains both putative dileucine and tyrosine motifs that could be involved in adaptor complex formation. A series of deletion mutants were constructed and are shown in Fig. 1B. The mutant VAChT ⌬C11 has the most C-terminal putative tyrosine-sorting motif deleted, the mutant VAChT ⌬C37 has both C-terminal putative tyrosine-sorting motifs deleted, while the mutant VAChT ⌬C53 has the above deletions plus the dileucine motif and the phosphorylation site deleted.
The interaction with adaptin ␣ was essentially eliminated in all of the mutants, even VAChT ⌬C11 , in which the C-terminal 11 amino acids that contain a putative tyrosine motif was deleted. This finding suggests that the C-terminal 11 amino acids contain the AP-2 interaction signal. The adaptin ␥ interaction was diminished in the VAChT ⌬C11 and VAChT ⌬C37 deletion constructs, and was absent in the VAChT ⌬C53 construct in which the phosphorylation site and the dileucine motif were deleted (Fig. 4). However, all of these mutants exhibited interaction with AP-180 and since together they represent deletion of all but the first 8 amino acids of the C-terminal tail we consider the interaction with AP-180 as being nonspecific. Supporting this suggestion is the finding that a construct in which the first 8 amino acids (residues 471-478) of VAChT were deleted still bound to AP-180.
We examined the consequence of eliminating the interaction with adaptin ␣ or adaptin ␥ by examining the trafficking of these deletion mutants in PC12 cells. Full-length VAChT containing each of the deletion mutants was constructed in the expression vector pcDNA3.1. Since the C-terminal truncations eliminate the epitope recognized by available VAChT antibodies, the hemagglutinin epitope (HA), YPYDVPDYA, was attached to the N-terminal of VAChT for detection purposes. Since the HA epitope contains tyrosine residues that might affect trafficking we examined trafficking of wild type and VAChT ⌬C11 with and without the HA epitope. The VAChT ⌬C11 mutant reacted with the anti-C-terminal VAChT antibody. We found no difference. This observation is in agreement with the data of Tan et al. (36) who also showed that the HA epitope did not affect VAChT subcellular localization in PC12 cells. Immunofluorescence was used to compare the trafficking of the deletion mutants. As shown in Fig. 5, all of the mutants exhibit a different subcellular localization from wild type VAChT, which is found in a perinuclear region colocalized with synaptophysin, the marker for synaptic-like microvesicles. PC12 cells expressing the VAChT ⌬C11 mutant, in which the interaction with AP-2 was eliminated and the interaction with AP-1 was diminished, showed staining mostly on the plasma membrane, but a small fraction still colocalized with synaptophysin. This expression pattern appeared very similar to what we observed with a dileucine mutant, Fig. 5. However, PC12 cells expressing the VAChT ⌬C37 or the VAChT ⌬C53 mutant exhibited staining throughout the cell. The mis-sorting of the VAChT ⌬C11 and VAChT ⌬C37 mutants was further confirmed by sucrose density gradient fractionation of transiently transfected PC12 cells. As shown in Fig. 6, transiently transfected wild type VAChT colocalized with synaptophysin the marker for synaptic-like microvesicles. In agreement with immunofluorescent staining the VAChT ⌬C11 mutant was partially colocalized with synaptophysin, but was reproducibly shifted one to two fractions in the direction of the more dense part of the sucrose gradient. In a similar analysis the VAChT ⌬C37 mutant was clearly mislocalized appearing as two distinct peaks in different parts of the sucrose gradient. Neither peak co-migrated with either secretogranin II or synaptophysin.
A Non-classical Tyrosine Motif Is Involved in AP-2 Binding-The deletion mutation experiments described above indicate that the phosphorylation/dileucine-containing motif is involved in AP-1 binding and that the last 11 amino acids of the C terminus of VAChT (residues 520 to 530) contain a binding site for AP-2 and contribute to AP-1 binding. In order to determine whether the VAChT dileucine motif is a recognition signal for binding to AP-1 or AP-2 complexes, we substituted alanine residues yielding the mutant VAChT L485A/L486A . As shown in Fig. 7A, this dileucine motif mutation abolished binding to adaptin ␥, the subunit of AP-1. In contrast, this mutation had no effect on binding to the AP-2 complex in agreement with the deletion mutagenesis studies described above.
The AP-2 binding region does not contain a typical tyrosinebased sorting motif. However, it contains three tyrosine residues, Tyr 524 , Tyr 526 , and Tyr 527 , Fig. 1. Interestingly, the VAChT cytoplasmic tail does contain a typical tyrosine-based sorting motif YDAV at residues 494 -497. When a Y494A mutant VAChT was expressed in PC12 cells, it did not exhibit normal trafficking as judged by its lack of co-localization with FIG. 4. Mapping the sites of interaction of rVAChT with AP complexes using deletion mutants. Rat brain extracts were incubated with GST fusion proteins containing deletion mutants of the rVAChT cytoplasmic tail. Complexes were isolated on GST-agarose, separated by SDS-PAGE, and then probed with specific antisera as described in Fig. 2. synaptophysin as judged by immunofluorescence (data not shown). To further examine the potential functionality of the four C-terminal tyrosine residues of VAChT, mutants Y494A, Y524A, and an Y526A/Y527A double mutant were constructed. As shown in Fig. 7B, the Y494A mutation did not affect the interaction with any of the adaptin molecules. In contrast Y524A showed reduced interaction with both adaptin ␣ and adaptin ␥, but not with AP-180, further supporting the nonspecific interaction with AP-180. The interaction with adaptin ␣ was virtually abolished in the double mutant Y526A/Y527A, while a reduced interaction with adaptin ␥ was seen with this mutant.
Since both the mutant Y524A and the double mutant Y526A/ Y527A exhibit reduced interaction with adaptin ␣, we made a triple mutant Y524A/Y526A/Y527A to determine if the interaction with adaptin ␣ could be completely eliminated. Using this mutant we found that the binding to adaptin ␣ is virtually abolished, however the interaction with adaptin ␥ is diminished but still detectable, Fig. 8A. These findings were confirmed by in vivo co-immunoprecipitation experiments using stably transfected PC12 cells, Fig. 8B.
To characterize the function of the tyrosine containing motif in trafficking, we isolated from post nuclear supernatants of wild type and mutant VAChT-transfected PC12 cells synaptophysin-containing vesicles. This was accomplished using a mouse anti-synaptophysin antibody in conjunction with a rabbit anti-mouse antisera conjugated to Dynabeads. Although wild type VAChT was nearly quantitatively precipitated by this procedure, the Y524/526/527A mutant VAChT was barely detectable, Fig. 8C.
To further confirm mislocalization of the Y524/526/527A mutant VAChT, we immunostained transfected PC12 cells with anti-VAChT antibody. The Y524/526/527A mutant VAChT showed staining on the plasma membrane and within a perinuclear region that colocalized with synaptophysin (Fig. 9). This staining pattern is distinct from wild type VAChT, but is similar to that seen with VAChT ⌬C11 and the dileucine mutant VAChTs in PC12 cells (Fig. 5) as well as the double mutant Y526/527A (data not shown) indicating that the tyrosine-con-  6. Sucrose linear gradient fractionation of wild type and deletion mutant VAChT. Post-nuclear supernatants from wild type or rVAChT mutant-transfected PC12 cells were fractionated on a 0.6 -1.6 M continuous sucrose gradient. Fractions were probed with antisecretogranin II antisera (SgII) to locate large dense core vesicles, anti-synaptophysin antisera (Syn) to locate small synaptic-like microvesicles, and anti-VAChT antisera (VAChT) to locate rVAChT.

FIG. 7. The dileucine motif (A) and the tyrosine motif (B) interact with AP-1 and AP-2 complexes, respectively.
A rat brain extract was incubated with GST fusion proteins containing the Cterminal tail of rVAChT with the indicated mutations, treated with GST-agarose beads, and then subjected to Western blot analysis using anti-adaptin ␣ antibody, anti-adaptin ␥ antibody, or anti-AP-180 antibody.
taining region of the VAChT cytoplasmic tail is involved in endocytosis.

DISCUSSION
Recognition of sorting signals within the cytoplasmic tail of membrane proteins by adaptor protein complexes is a crucial step in membrane protein sorting. The adaptor protein complexes, AP-1, AP-2, and AP-3, have all been shown to recognize tyrosine-and leucine-based sorting signals, the most common sorting signals within membrane protein cytoplasmic tails. Most intracellular membrane proteins contain a cytoplasmic tail bearing more than one of these sorting signals. Examples are the mannose 6-phosphate receptors (34,37), epidermal growth factor receptor (38,39), CD4 (40), and invariant chain (41,42). VAChT has now been shown to join this group having both tyrosine-and leucine-based sorting signals within its carboxyl cytoplasmic tail.
We have used GST pull-down assays to show that the VAChT cytoplasmic tail interacts with adaptor protein 1, adaptor protein 2, and possibly, but unlikely, adaptor protein 180, but not with adaptor protein 3. In vivo co-immunoprecipitation assays confirm the interaction with adaptor protein 1 and adaptor protein 2, but low endogenous levels of AP-180 precluded confirming this complex in vivo. Deletion and site-specific mutagenesis studies show that the phosphorylated serine/ dileucine containing region represents a signal for interaction with AP-1, while a tyrosine-containing signal within residues Tyr 524 to Tyr 527 interacts with AP-2. In addition tyrosine 494 is shown to be critical for VAChT trafficking, but does not form a detectable complex with AP-1, AP-2, or AP-3. We could not confirm the report of Barbosa et al. (43) who found that the dileucine motif interacts with AP-2 in a yeast two-hybrid system assay. One possibility is that the dileucine motif of VAChT forms a complex with AP-2, as proposed by Barbosa et al. (43), but that this complex is either too transient or too weak to be detected by the pull-down and in vivo co-immunoprecipitation assays used in this study.
Although the VAChT C-terminal Tyr 524 -527 tyrosine motif does not fit the consensus sequence of a classical tyrosine motif, deletion mutagenesis, and site-directed mutagenesis shows this region is involved in the interaction with the AP-2 complex. Mutation of this motif also leads to a reduction of the interaction of AP-1 with VAChT. Since AP-2 interactions are believed to be involved in internalization from the plasma membrane, we expected that eliminating the Tyr 524 -527 motif would trap VAChT on the plasma membrane. Indeed the deletion mutant VAChT ⌬C11 and the point mutant Y524A/Y526A/ Y527A did show a significant amount of plasma membrane VAChT indicating the Tyr 524 -527 tyrosine-containing motif is involved in VAChT internalization. The finding that a fraction of VAChT containing a disrupted or missing tyrosine sorting motif also is found on synaptic-like microvesicles suggests that this motif either affects the rate of internalization as previously seen with a dileucine mutant (36) or is part of a more complex signal, likely with the dileucine motif, which in its absence is inefficiently endocytosed. This indicates that the VAChT cytoplasmic tail contains at least two internalization signals that may function independently or interactively, the Tyr 524 -527 tyrosine motif and the dileucine motif. Perhaps both motifs are required for efficient internalization, and that the presence of just one permits slow internalization.
In addition to the Tyr 524 -527 tyrosine-sorting motif, the Cterminal tail of VAChT contains a classical tyrosine-based sorting motif (YXXØ) consisting of Tyr 494 -Asp 495 -Ala 496 -Val 497 , Fig. 1. Ohno et al. (44) showed the presence of a valine in the Ø position disfavors interaction with adaptor proteins. However, a valine at the Ø position, although unfavorable for interaction with known adaptor proteins, does lead to sorting activity, via unknown complexes (44 -47). Deletion of the 494 YDAV classical tyrosine motif in conjunction with deleting the Tyr 524 -527 tyrosine-sorting motif completely disrupts VAChT trafficking in PC12 cells. Thus the classical tyrosine-based sorting motif YDAV of VAChT is also involved in VAChT sorting, but not through interaction with the well characterized adaptor proteins.
The finding that deletion or site specific mutagenesis of the Tyr 524 -527 tyrosine motif in VAChT eliminated AP-2 binding as well as mislocalized the transporter to the plasma membrane is in keeping with the previously described participation of AP-2 in the internalization of cell surface proteins (10,(12)(13)(14). However, the observation that mutation of the dileucine/phosphorylation motif, which results in predominantly plasma membrane VAChT, is the site of interaction with AP-1 is an unexpected finding. AP-1 complexes have been shown to be associated with the TGN and be involved in the transport of proteins from the TGN to the endosome/lysosome or from the endosome to the TGN system. However, in polarized epithelial cells, AP-1 complexes are involved in trafficking from TGN to the cell surface (48,49).
Based on the proposed model for VAChT trafficking shown in Fig. 10A, VAChT, along with other synaptic proteins such as synaptophysin, move from the TGN to the plasma membrane via constitutive secretory vesicles. These proteins are then trafficked from the plasma membrane to synaptic-like microvesicles following endocytosis. To enter the constitutive secretory pathway does not require interaction with an adaptor protein. Thus it is difficult to explain the function of the AP-1 interaction with VAChT. Based on the known involvement of AP-1 in trafficking from the TGN we might have expected to see the dileucine/phosphorylation mutant in an intracellular compartment. However, when the AP-1 interaction is blocked, i.e. by mutation of the dileucine/phosphorylation motif, VAChT could reach the plasma membrane via a default pathway as seen with a VAChT mutant completely lacking a cytoplasmic C-terminal tail (36). Even if this occurs, the presence of an intact Tyr 524 -527 tyrosine motif that can interact with AP-2 would have been expected to lead to the internalization of plasma membrane VAChT. Thus the finding that the dileucine/ phosphorylation mutant is retained on the plasma membrane is inconsistent with the proposed role of AP-1 in a TGN-trafficking role. However, in a number of membrane protein trafficking systems, one sorting motif serves more than one sorting function (50 -52). For example the dileucine motif of the clusters of differentiation 3␥ (CD3 ␥), serves as a signal for both AP-1 and AP-2 binding (53). The dileucine/phosphorylation motif of VAChT may also have a dual function; one of containing an internalization signal, but not via an AP-2 complex, and the other a signal for AP-1-dependent trafficking.
An alternative pathway for VAChT trafficking, suggested by Tao-Cheng and Eiden (54) and by Varoque and Erickson (35), utilizes the regulated secretory sorting pathway, as do VMATs, instead of the constitutive secretory pathway, Fig. 10B. Electron microscopy studies showed endogenous VAChT, as well as transfected VAChT, appearing mainly in SLMVs, but also in LDCVs in PC12 cells (1,35,54). Similarly, in a study by Krantz et al. (55) analysis of individual PC12 cells transfected with recombinant rVAChT showed ϳ40% of the VAChT in LDCVs. Mutation of serine 480 to glutamate, mimicking phosphorylation, increased LDCV VAChT to 60%. If VAChT were to follow the constitutive secretory sorting pathway it is difficult to understand how it could appear in dense core vesicles. Therefore, these published findings coupled with the VAChT interaction with AP-1 seen here could be interpreted to suggest that VAChT may directly sort from the TGN into immature dense core vesicles and then move to endosomes before maturating in SLMVs (54), Fig. 10B. This possibility clearly warrants future investigation. AP-3 is believed to function in the formation of synaptic vesicles from endosomes (23). In neither our in vitro nor in vivo binding studies could an interaction between AP-3 and the VAChT cytoplasmic tail be detected. Based on these findings one is left to conclude that either the trafficking of VAChT from the endosome to the synaptic vesicles utilizes a novel sorting machinery or that the interaction between AP-3 and VAChT is relatively weak, perhaps because of the requirement for a linker protein. There is precedence for the involvement of linker proteins interacting with AP-1 and AP-2 as seen with the two mannose 6-phosphate receptors, TIP (MPR tail interaction protein of 47 kDa) and GGA (56,57).
In conclusion, we have shown that VAChT interacts with the TGN-associated AP-1 clathrin complex as well as the plasma membrane AP-2 complex. AP-1 is predominantly recognized by the phosphorylation/dileucine containing region of the VAChT C-terminal cytoplasmic tail. This phosphorylation/dileucine containing region not only recognized AP-1, but was shown to be involved in VAChT endocytosis both in this study and the study of Tan et al. (36). AP-2 interacts with a non-classical tyrosine-containing motif and is also involved in internalization. Although the interaction of VAChT with AP-2 seems to function similarly as seen with other transporters, the interaction with AP-1 does not. Thus the function of the VAChT-AP complexes identified in this study do not seem to exhibit the same trafficking functions as seen in other systems, suggesting the formation of novel complexes. An alternative explanation is that overexpression of VAChT leads to anomalous sorting events and mislocalization of VAChT. Although this possibility FIG. 10. Proposed pathways for VAChT trafficking in neuroendocrine cells. A, VAChT trafficking through the constitutive pathway to the plasma membrane and then to synaptic-like microvessicles. This pathway is adapted from that proposed for synaptophysin (7). Newly synthesized VAChT first moves from the TGN to the plasma membrane (PM) via constitutive secretory vesicles (CSV), step 1A. VAChT is then transported to early endosomes (EE), by endocytosis via endosomal vesicles (EV), step 2A, followed by transport to SLMV, step 3A. An external stimulus causes VAChT to fuse with the plasma membrane and release acetylcholine, step 4A. VAChT is then recycled to SLMV via early endosomes, step 5A. B, VAChT trafficking through the regulated secretory pathway (adapted from that proposed for the biogenesis of large dense core vesicles (58). Newly synthesized VAChT moves from the TGN to early endosomes (EE) through predense core vesicles (PDCV), step 1B, then directly to synaptic-like microvesicles, step 2B. After release of acetylcholine by an external stimulus (step 3B), VAChT is recycled to SLMV by endocytosis (steps 4B and 2B).
cannot be conclusively ruled out, the expressed recombinant VAChT migrates in a sucrose gradient at the buoyant density of synpatic-like microvesicles and is co-precipitated with synaptophysin a known synaptic-like microvesicle protein. Additionally, the observation that in PC12 cells recombinant VAChT is trafficked to synaptic-like microvesicles while, when transfected, the related transporter VMAT2 is trafficked to dense core vesicles (4) supports the notion that overexpression per se does not lead to artifactual transporter trafficking.