AP180 and AP-2 Interact Directly in a Complex That Cooperatively Assembles Clathrin*

Clathrin-coated vesicles are involved in protein and lipid trafficking between intracellular compartments in eukaryotic cells. AP-2 and AP180 are the resident coat proteins of clathrin-coated vesicles in nerve terminals, and interactions between these proteins could be important in vesicle dynamics. AP180 and AP-2 each assemble clathrin efficiently under acidic conditions, but neither protein will assemble clathrin efficiently at physiological pH. We find that there is a direct, clathrin-independent interaction between AP180 and AP-2 and that the AP180-AP-2 complex is more efficient at assembling clathrin under physiological conditions than is either protein alone. AP180 is phosphorylated in vivo, and in crude vesicle extracts its phosphorylation is enhanced by stimulation of casein kinase II, which is known to be present in coated vesicles. We find that recombinant AP180 is a substrate for casein kinase II in vitro and that its phosphorylation weakens both the binding of AP-2 by AP180 and the cooperative clathrin assembly activity of these proteins. We have localized the binding site for AP-2 to amino acids 623–680 of AP180. The AP180/AP-2 interaction can be disrupted by a recombinant AP180 fragment containing the AP-2 binding site, and this fragment also disrupts the cooperative clathrin assembly activity of the AP180-AP-2 complex. These results indicate that AP180 and AP-2 interact directly to form a complex that assembles clathrin more efficiently than either protein alone. Phosphorylation of AP180, by modulating the affinity of AP180 for AP-2, may contribute to the regulation of clathrin assembly in vivo.

AP-2 and AP180 co-localize to budding coated vesicles in neuronal cells arrested with GTP␥S (28). Furthermore, AP-2 and AP180 co-immunoprecipitate from coated vesicle extracts (14) and co-purify from various cell extracts (36,37). However, since AP-2 and AP180 also both bind to clathrin, it is unclear if their co-immunoprecipitation and co-purification reflect an indirect interaction with clathrin in a ternary complex or a direct interaction between these two proteins. We show here that AP-2 and AP180 interact directly, and we map the binding site for AP-2 on AP180 to amino acids 623-680. Both the ␣ and ␤ subunits of AP-2 bind to this same region of AP180.
We also find that AP180 and AP-2 together are more efficient at assembling clathrin than is either AP180 or AP-2 alone. The assembly activity of the two proteins together is significantly greater than can be accounted for by simply adding the assembly activities of each protein alone. Further, we show that disruption of the AP180/AP-2 interaction by a recombinant fragment of AP180 containing the AP-2 binding site inhibits the enhanced assembly activity of the AP180 plus AP-2 combination. This indicates that formation of the AP180-AP-2 com-plex is required to obtain the supra-additive effects on clathrin assembly observed when both of these proteins are added to an assembly reaction.
Coated vesicles also contain an associated CKII, and clathrin light chain ␤ has been shown to be a good substrate for this enzyme (38). AP180 is known to be phosphorylated in vivo, from both labeling studies of cultured neurons (13) and phosphatase studies of mouse brain extracts (16). Furthermore, AP180 has been shown to be phosphorylated in an assembly protein fraction under conditions in which CKII is stimulated (39). To extend these studies, we carried out experiments with purified proteins to determine if AP180 is a substrate for CKII. We find that there are three CKII phosphorylation sites in AP180, all within the central 42-kDa domain of the protein.
Since this domain also contains residues important for clathrin assembly and AP-2 binding, we evaluated the effects of phosphorylation on these activities. While phosphorylation has no effect on the clathrin assembly activity of AP180 alone, it weakens the interaction between AP180 and AP-2. Apparently as a consequence of this, phosphorylation also reduces the assembly activity of the AP180/AP-2 combination by an amount consistent with its effect on AP180-AP-2 affinity.
We propose a model in which AP180 and AP-2 associate to form a clathrin assembling complex. Phosphorylation, by modulating the affinity of AP180 for AP-2, may influence formation of this complex and thereby contribute to the regulation of clathrin assembly.

Materials
Glutathione-Sepharose 4B was obtained from Amersham Pharmacia Biotech. Monoclonal antibodies against ␣ or ␤ subunits of AP-2, monoclonal antibodies against the ␥ subunit of AP-1, anti-IgG, ATP, bovine ␣-casein, polylysine, calf intestinal alkaline phosphatase agarose, trypsin, trypsin inhibitor, and the catalytic subunit of PKA were from Sigma. [␥-32 P]ATP was from DuPont. CaMKII and CKII were from New England BioLabs. PKC was from Roche Molecular Biochemicals.
For all the reactions, 0.6 M protein, 25 M ATP, and 1 Ci of [␥-32 P]ATP in buffer A were used in a reaction volume of 50 l. For phosphorylation by CKII, 100 g/ml polylysine, 130 mM KCl, and 30 units/l CKII in buffer A was added (except for the experiment shown in Fig. 10B, where the amount of enzyme is indicated on the displayed plot). For phosphorylation by CaMKII, 2.4 M calmodulin, 2 mM CaCl 2 , and 15 units/l CaMKII in buffer A were added. For phosphorylation by PKA, 10 units/l PKA in buffer A was added. For phosphorylation by PKC, 2 mM CaCl 2 , 100 g/ml phosphatidylserine, 20 g/ml diacylglycerol, and 0.25 milliunits/ml PKC in buffer A were added. In all cases, the reaction was carried out by adding kinases and incubating for 1 h at 30°C. The reaction was stopped by adding 5ϫ SDS sample buffer, followed by boiling for 5 min. The samples were electrophoresed on 10% SDS-polyacrylamide gels, followed by staining with Coomassie Blue. Protein amounts were determined by densitometric analysis relative to GST and bovine AP180 standards (Molecular Dynamics Personal Densitometer SI). The radioactivity in the labeled proteins was determined using a PhosphorImager (Molecular Dynamics, Inc., Sunnyvale, CA).
Characterization of the AP180 and AP-2/AP-1 Interaction-All of the procedures were performed at 4°C or on ice unless otherwise indicated. Bovine AP-2 was dialyzed into buffer B (50 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride) overnight and clarified by centrifugation at 400,000 ϫ g for 10 min. 100 l of glutathione-Sepharose beads coupled with a 10 M concentration of either GST-AP180 or other GST-AP180 deletion fragments was mixed with 0.63 M (final concentration) bovine AP-2 in buffer C (0.05% Tween 20, 0.1% gelatin, and 2 mM EGTA in buffer B) in a final volume of 200 l. The mixture was incubated for 1 h with gentle rocking before the beads were pelleted in a microcentrifuge by centrifugation at 1,300 ϫ g for 5 min. The supernatant was removed, and the beads were washed three times with 500 l of buffer C. Pelleted beads were resuspended in 100 l of 2ϫ SDS sample buffer, followed by boiling for 5 min. The samples were analyzed by SDS-PAGE, followed by Western blot analysis using a monoclonal antibody to the ␣ subunit of AP-2 and detected by ECL (NEN Life Science Products). Data were quantitated using the Personal Densitometer SI (Molecular Dynamics) to determine the percentage of AP-2 bound. GST served as the negative control. The interaction between GST-AP180 and AP-1 was measured utilizing the same assay. The interactions between the GST-AP180 fusion proteins and the ␤ 2 subunit of AP-2 or the ␣-ear domains of AP-2 were measured utilizing the same assay, except that 20 M fusion proteins were coupled to the beads, and the final concentration of ␤ 2 subunit or the ␣-ear domains in the reaction was 4 M. The inhibition of the AP180/AP-2 interaction by M11 was measured similarly, but the indicated amounts of M11 were preincubated with bovine AP-2 for 30 min before the mixture was added to the fusion protein-coupled beads.
Preparation of Phosphorylated and Dephosphorylated Fusion Proteins for either AP180-AP-2 Binding or Clathrin Assembly Assays-2.4 M protein, 160 g/ml polylysine, 100 M ATP, and 32 units/l CKII were mixed in buffer A in a final volume of 300 l, and the reaction was incubated for 80 min at 30°C. For proteins used in the AP180-AP-2 binding assays, the phosphorylated fusion proteins were incubated with 50 l of pelleted glutathione-Sepharose beads by rocking for 4 h at 4°C, followed by three washes each of 500 l of buffer B and used immediately. In each experiment, it was verified that the same amount of phosphorylated and unphosphorylated proteins had bound to the beads. For proteins used in the clathrin assembly assays, the phosphorylated fusion proteins were incubated with 200 l of pelleted glutathione-Sepharose beads by rocking for 2 h at 4°C, followed by three washes each of 500 l of buffer D (50 mM Tris-HCl, 1 mM MgCl 2 , pH 8.0) and elution with 400 l of buffer E (20 mM glutathione, 50 mM Tris-HCl, 1 mM MgCl 2 , pH 8.0). Half of the eluate (phosphorylated GST-AP180) was dialyzed into 10 mM Tris-HCl, pH 8.5. The other half of the eluate was dialyzed into 50 mM Tris-HCl, 1 mM MgCl 2 , pH 8.0, for 4 h and was added to insoluble calf intestinal alkaline phosphatase-agarose at a ratio of 0.35 g of protein/unit of phosphatase. The dephosphorylation was carried out at room temperature by rocking for 1 h, and the reaction was stopped by gentle spinning to remove the phosphatase agarose. The dephosphorylated GST-AP180 was subsequently dialyzed into 10 mM Tris-HCl, pH 8.5, for use in a clathrin assembly assay.
As a control, a parallel phosphorylation reaction with the addition of 4 Ci of [␥-32 P]ATP was carried out under the same conditions in a volume of 50 l to ensure that the stoichiometry of phosphorylation was similar to that described in Fig. 10. The mock phosphorylated control proteins were incubated under identical conditions but without the addition of CKII.
Preparation of Cytosol-All of the procedures were performed at 4°C. Fresh bovine brain tissue was mixed at a weight ratio of 1:2 with buffer B and blended for 3 ϫ 20 s, followed by centrifugation at 14,500 ϫ g for 20 min. The supernatant was further clarified by centrifugation at 125,000 ϫ g for 1 h and concentrated 5 times using 50% saturated ammonium sulfate in 0.5 M Tris-HCl, pH 7.0. It was then suspended in buffer B and used at 7 mg/ml (final concentration) under the same conditions as used for pure AP-2 in the characterization of the AP180/ AP-2 interaction.
Partial Proteolysis of AP-2-Partial digestion of AP-2 was carried out by mixing AP-2 and trypsin at a weight ratio of 1:500 in 10 mM Tris-HCl, pH 8.0, for 40 min at room temperature. The digestion was stopped by the addition of soybean trypsin inhibitor at a weight ratio of 5:1 relative to trypsin. 1 ⁄10 volume of 1 M MES, pH 6.2 was added to the reaction mixture, and the mixture was incubated on ice for 2 h. Under these conditions the "trunk" of AP-2 aggregated, leaving the "ear" or appendage domains in a soluble form that was easily separated by centrifugation at 400,000 ϫ g for 15 min at 4°C (45).
Co-immunoprecipitation of AP-2 and AP180 with AP180 Monoclonal Antibody-Clathrin-coated vesicles (4 -5 mg), prepared according to Nandi et al. (46) were stripped of clathrin by incubation in 10 mM Tris-HCl, pH 8.5, followed by sedimentation at 400,000 ϫ g for 15 min at 4°C. The partially decoated vesicles, which still contained APs were suspended in 5 ml of 4% Triton X-100 in 10 mM Tris-HCl, pH 8.5, and incubated on ice for 15 min. The suspension was centrifuged at 400,000 ϫ g at 4°C for 15 min, and the supernatant was retained. The supernatant was applied to either an affinity column at room temperature made with affinity-purified monoclonal antibodies against AP180 (5 mg of mAb F1-20/ml of CNBr-activated Sepharose 4B) or applied to a control Sepharose 4B column. Unbound protein was washed from the columns with 10 column volumes of phosphate-buffered saline, and bound protein was eluted from the columns with 50 mM glycine, 150 mM NaCl, pH 2.3. The samples were electrophoresed on 4 -20% SDS-polyacrylamide gels, followed by Western blot analysis with antibodies against AP180 and AP-2.
Clathrin Assembly Assay-For assembly reactions carried out at pH 6.7, mock-phosphorylated GST-AP180, phosphorylated GST-AP180 (P-GST-AP180), M11, and clathrin were dialyzed overnight into 10 mM Tris-HCl, pH 8, 2 mM dithiothreitol at 4°C, followed by clarification at 400,000 ϫ g and 4°C for 10 min. The indicated assembly protein was mixed with clathrin, and the assembly was initiated by adding 1 ⁄10 volume of 1 M MES-NaOH, pH 6.7. The final conditions in the reaction were 0.5 M clathrin, 0.1 M MES-NaOH, 9 mM Tris-HCl, pH 6.7, and the indicated amounts of fusion proteins in a final volume of 200 l. The mixture was incubated on ice for 45 min and then was centrifuged at 400,000 ϫ g and 4°C for 6 min. The upper 80% of the supernatant was removed and analyzed by SDS-PAGE and Coomassie Blue staining. The percentage of clathrin assembly was determined by the relative depletion of clathrin from the solution before and after centrifugation and quantitated by densitometry of Coomassie Blue-stained gels (Molecular Dynamics Personal Densitometer SI). In the absence of GST-AP180, 3% of the clathrin was depleted by centrifugation.

RESULTS
Protein Purification-Highly purified clathrin, AP-1, and AP-2 were prepared from bovine brain clathrin-coated vesicles (Fig. 1). Care was taken to avoid contamination of the clathrin by the assembly proteins and of the assembly proteins by clathrin. Western blot analysis revealed that the purified proteins had Ͻ0.2% cross-contamination. Recombinant mouse GST-AP180 was prepared from a bacterial expression vector and purified on glutathione-Sepharose. SDS-PAGE of the purified protein revealed a major band with an apparent M r of 180,000 and a number of minor lower molecular weight bands (Fig. 1). Because the staining pattern of the Coomassie Bluestained gel was identical to an immunoblot with an anti-AP180 monoclonal antibody, we conclude that the lower molecular weight bands are proteolytic fragments of AP180. AP180 is known to be protease-sensitive (17), and attempts at further purification did not yield a preparation that was significantly more enriched for the high molecular weight band.
Characterization of the AP180/AP-2 Interaction-Bovine brain cytosolic extracts were incubated with glutathione-Sepharose to which a series of different GST-fusion proteins had been bound. Western blot analysis of the retained proteins with an ␣-adaptin-specific antibody revealed that both ␣-adaptin a and ␣-adaptin c were retained on the GST-AP180 resin but not on the GST resin, indicating that AP-2 had bound specifically to AP180 (Fig. 2A). Moreover, we found that AP-2 was retained on columns made with fusion proteins that included either amino acids 305-901 (C58) or 305-744 (M42) of mouse AP180 but was not retained on columns made with fusion proteins that included amino acids 1-304 (N33) or 745-901 (C16), indicating that the binding site is located in the central 42-kDa domain between residues 305 and 744.
To evaluate the likelihood that this is a physiologically relevant protein/protein interaction, we carried out immunoprecipitation studies. Because AP180 and AP-2 both co-localize to CCVs, we isolated CCVs from bovine brain. Since both proteins are known to interact directly with clathrin, we extracted the CCVs in low ionic strength buffer to remove the clathrin, followed by detergent extraction of the clathrin-depleted vesicles. We found that antibodies against AP180 co-immunoprecipitated AP-2 from this extract (Fig. 2B). This demonstrates that the interaction between AP-2 and AP180 is not dependent on the presence of clathrin. To assess whether the interaction between AP180 and AP-2 is direct or indirect, we carried out binding assays utilizing highly purified bovine AP-2. The AP-2 was incubated with glutathione-Sepharose to which a series of GST fusion proteins had been bound. Analysis of the bound material revealed that AP-2 was retained only on the columns made with either fulllength GST-AP180, the 58-kDa C-terminal domain, or the middle 42-kDa domain (Fig. 3), consistent with the previous experiment. This experiment demonstrates that there is a direct interaction between AP-2 and AP180 and that the binding site is within the middle 42-kDa domain of AP180.
To further localize the binding site on AP180 to which AP-2 binds, we engineered a series of progressive deletion mutants of mouse AP180 and studied the binding of purified AP-2 to these mutants (Fig. 4). We found a sharp loss in AP-2 binding between fragments C27 and C22, suggesting that the AP-2 binding site is within amino acids 623-680. This was confirmed by the finding that AP-2 bound to glutathione-Sepharose resin to which a recombinant fragment consisting of amino acids 623-680 (M5) of AP180 had been attached. These data are summarized in Fig. 4.
AP-2 is associated with clathrin-coated vesicles budding from plasma membranes, while AP-1 is associated with clathrin-coated vesicles budding from Golgi membranes. Since AP180 has not been found in association with CCVs budding from Golgi membranes, we examined its binding to AP-1, expecting that it would not bind. However, when GST-AP180 attached to glutathione-Sepharose beads was incubated with purified AP-1 (with no detectable AP-2 contamination), the AP-1 bound efficiently (Fig. 5). Since the ␤ subunits of AP-2 and AP-1 are 84% identical (89% similar) (44,47), the above observation suggested that binding of AP180 to both AP-1 and AP-2 might be a consequence of the high degree of structural similarity between the AP-1 and AP-2 ␤ subunits. We therefore expressed and purified the ␤ 2 subunit of AP-2 and evaluated its FIG. 2. AP-2 is pulled out of bovine brain cytosol by AP180 affinity chromatography, and is co-immunoprecipitated with AP180 by an anti-AP180 monoclonal antibody. A, bovine brain cytosol was incubated with glutathione-Sepharose beads to which GST, GST-AP180, GST-N33, GST-M42, GST-C58, or GST-C16 was attached. The material bound to each resin was subjected to SDS-PAGE, followed by Western blot analysis using antibodies to the ␣ subunit of AP-2. The doublet of bands labeled as AP-2 are the ␣ a (upper) and ␣ c (lower) subunits. B, clathrin-coated vesicle extracts were incubated with either Sepharose 4B beads to which the anti-AP180 Mab F1-20 was covalently attached or with underivatized Sepharose 4B. The material bound to the beads was subjected to SDS-PAGE, followed by Western blot analysis using antibodies to either the ␣ subunit of AP-2 or AP180.
FIG. 3. Purified bovine AP-2 binds to the middle 42-kDa domain of AP180. Purified bovine AP-2 was incubated with glutathione-Sepharose beads to which GST, GST-AP180, GST-N33, GST-M42, GST-C58, or GST-C16 was attached. The material bound to each resin was subjected to SDS-PAGE, followed by Western blot analysis using antibodies to the ␣ subunit of AP-2. A, a blot from a typical experiment is displayed. B, a quantitative analysis of the average percentage of binding of AP-2 from six independent experiments, with error bars indicating one S.E., is plotted. ability to bind to GST-AP180. Consistent with this hypothesis, the ␤ 2 subunit of AP-2 did indeed bind to a site within amino acids 623-680 of AP180, similar to native tetrameric AP-2 (Fig.  6A). While the binding of recombinant ␤ 2 to AP180 is not as efficient as of native AP-2, we also found that the clathrin assembly properties of ␤ 2 are weaker than those of native AP-2 (consistent with previously published observations (48), and data not shown), suggesting that the ␤ 2 subunit may be only partially active following purification from inclusion bodies.
Because AP180 has been reported to have been pulled out of a cytosolic extract by passage over an affinity column made with the "ear" domain of the ␣ subunit of AP-2, it has been proposed that AP180 interacts with the ␣-ear domain of AP-2 (36). In order to evaluate this hypothesis, we used limited proteolysis by trypsin to generate hinge-ear domains of AP-2, and assayed them for binding to glutathione-Sepharose beads to which we attached different recombinant fragments of GST-AP180. We found that there is indeed a direct interaction between the ␣-hinge-ear domain of AP-2 and AP180 (Fig. 6B). Moreover, we found that the ␣-hinge-ear binds to a site within the same stretch of amino acids on AP180 to which either native AP-2 or recombinant ␤ 2 binds.
The AP180-AP-2 Complex Cooperatively Assembles Clathrin-Since we found that AP180 and AP-2 associate directly to form a complex, we set out to study the effects of AP180-AP-2 complex formation on clathrin assembly. We carried out a quantitative comparison of the relative efficiency of clathrin assembly by AP180 alone, AP-2 alone, and AP180 plus AP-2 in combination under physiologically relevant protein concentrations (45) and pH. In reactions that contained 0.2 M clathrin at pH 7.0, neither AP180 alone nor AP-2 alone assembled clathrin efficiently (Fig. 7). However, assembly by the AP180/AP-2 combination under these conditions was very efficient. Because the assembly activity of AP180 and AP-2 in combination was significantly greater than would be expected from simply adding the assembly activities of the individual proteins, we conclude, by definition, that AP180 and AP-2 in combination cooperatively assemble clathrin.
To test whether the enhancement in assembly activity is a direct consequence of the AP180/AP-2 interaction, we utilized the recombinant AP180 fragment M11 characterized in Fig. 4 to perturb the AP180/AP-2 interaction. M11 contains the AP-2 binding site, binds to AP-2 (Fig. 4), and would therefore be expected to competitively inhibit AP-2 binding to AP180. Indeed, M11 was an effective inhibitor of the AP180/AP-2 interaction (Fig. 8A). We found that M11 itself did not possess any intrinsic clathrin assembly activity (Fig. 8B). Then we used M11 to evaluate the effects of disruption of the AP180/AP-2 interaction on the enhanced assembly observed by the AP180/ AP-2 combination. We found that M11 was indeed an effective inhibitor of clathrin assembly by the AP180/AP-2 combination, suggesting that AP180-AP-2 complex formation is required for cooperative assembly (Fig. 8C).
AP180 Is a Substrate for Casein Kinase II-AP180 is phosphorylated in vivo (13,16). To identify which kinases might be involved in this phosphorylation, we incubated purified GST-AP180 with CKII, PKA, CaMKII, or PKC in the presence of [␥-32 P[ATP. Following SDS-PAGE and autoradiography, it was found that AP180 was labeled only in the presence of CKII and PKA, indicating that AP180 is a substrate for these two kinases (Fig. 9). No labeling of AP180 was seen by CaMKII or PKC, although the phosphorylation of histone H1 by PKC and the autophosphorylation of CaMKII indicate that the enzymes FIG. 4. The AP-2 interacting region of AP180 is within amino acids 623-680. A series of progressive deletion mutants of AP180 were prepared as described under "Experimental Procedures." Purified bovine AP-2 was incubated with Glutathione-Sepharose beads to which the indicated GST fusion protein had been attached. The material bound to each resin was subjected to SDS-PAGE, followed by Western blot analysis using antibodies to the ␣ subunit of AP-2.

FIG. 5. AP180 binds to AP-1 in vitro.
Purified bovine AP-1 was incubated with glutathione-Sepharose beads to which either GST-AP180 or GST had been attached. The material bound to each resin was subjected to SDS-PAGE, followed by Western blot analysis using antibodies to the ␥ subunit of AP-1. were active on these well characterized substrates. This indicates that AP180 is not a substrate for these enzymes. Since coated vesicles are known to contain CKII and AP180 had previously been observed to be phosphorylated in an assembly protein fraction under conditions in which CKII would have been active (39), we decided to further characterize the phosphorylation by CKII. We measured the phosphorylation of the various domains of bacterially expressed GST-AP180 by purified CKII (Fig. 10) and found that only fusion proteins containing amino acids 305-744 were phosphorylated, indicating that all of the phosphorylation sites are within the middle 42-kDa domain of AP180 (Fig. 10A). We also found that all of the fusion proteins that were phosphorylated incorporated 3 mol of phosphate/mol of protein, indicating that there are three CKII phosphorylation sites within the central 42-kDa domain of AP180 (Fig. 10B).
Effect of Phosphorylation on AP180 Function-Since the two known functions of AP180, AP-2 binding (Fig. 3) and clathrin assembly (39), involve residues within the central 42-kDa domain of AP180, we studied the effects of phosphorylation on these two activities. Either unphosphorylated or CKII-phosphorylated GST-AP180 resin was incubated with purified bovine AP-2, and the amount of AP-2 retained on the resin was measured. We found that 56% less AP-2 was bound to the phosphorylated GST-AP180 resin than to the unphosphorylated GST-AP180 resin, that 48% less AP-2 was bound to the phosphorylated GST-M42 resin than to the unphosphorylated GST-M42 resin, and that 57% less AP-2 was bound to the phosphorylated GST-C58 resin than to the unphosphorylated GST-C58 resin (Fig. 11). These experiments indicate that phosphorylation of AP180 by CKII decreases its interaction with AP-2. We also compared the effects of CKII phosphorylation of GST-AP180 on AP180-mediated clathrin assembly. We observed no difference in clathrin assembly mediated by phosphorylated GST-AP180 versus unphosphorylated GST-AP180, as measured in a quantitative, factor-dependent clathrin assembly assay (Fig. 12). This also served as a control to monitor the activity of the phosphorylated protein.
AP180 Phosphorylation Modulates Clathrin Assembly Promoted by the AP180-AP-2 Complex-We then set out to evalu- FIG. 8. An AP180 recombinant fragment that inhibits the AP180/AP-2 interaction also disrupts clathrin assembly by the AP180/AP-2 combination. A, purified bovine AP-2 was incubated with glutathione-Sepharose beads to which GST-AP180 was attached in the presence of the indicated amounts of recombinant fragment M11. The material bound to the resin was subjected to SDS-PAGE, followed by Western blot analysis using antibodies to the ␣ subunit of AP-2. A quantitative analysis of the average percentage of AP-2 bound from three independent experiments, with error bars indicating one S.E., is plotted. B, clathrin assembly by M11 was carried out at pH 6.7. A quantitative analysis of the average percentage of clathrin assembly from three independent experiments, with error bars indicating one S.E., is plotted. C, clathrin assembly reactions containing 0.2 M clathrin, 0.3 M GST-AP180, and 0.3 M AP-2 were carried out at pH 7.0 in the presence of the indicated amounts of M11. A quantitative analysis of the percentage of clathrin assembly is plotted. ate the effects of AP180 phosphorylation on the assembly activity of the AP180/AP-2 combination. We carried out assembly reactions at pH 7.0 utilizing clathrin and AP180/AP-2 combinations formed with AP-2 and either GST-AP180, mock-phosphorylated GST-AP180, phosphorylated GST-AP180, or phosphorylated GST-AP180 that was subsequently dephosphorylated with alkaline phosphatase (Fig. 13). The combinations made with AP-2 and either GST-AP180 or mock-phosphorylated GST-AP180 both showed more assembly activity than the expected additive effects of the individual proteins. However, the assembly activity of the combination made with CKII phosphorylated AP180 was reproducibly reduced by an amount consistent with its reduced affinity for AP-2. This reduction in activity can be specifically attributed to phosphorylation, since full clathrin assembly activity was restored to the protein by treatment with alkaline phosphatase.

DISCUSSION
The finding that AP180 and AP-2 co-localize to the same budding CCV (28) raises the question of why there needs to be two clathrin assembly proteins in the same CCV. It has long been argued that the cell uses APs to promote clathrin assembly to ensure uniformly sized cages and to ensure that the assembly reaction is efficient under physiological conditions (23,49). Clathrin, in the absence of an assembly protein, will form irregularly sized cages but only under conditions of low pH and high calcium (23,50,51). While in vitro assembly reactions promoted by either AP180 or AP-2 proceed at higher pH than reactions without an AP, the conditions used are still not physiological (5,25). AP-2 and AP180 assembly reactions are typically carried out at pH 6.4 -6.7 (5,25,44,49). Even under these conditions, abnormally high concentrations of proteins are required (25) relative to their available concentrations in cells (45,52). It may be significant that assembly reactions utilizing the crude coat protein fraction (which contains both AP180 and AP-2, among other factors) proceed more easily at lower protein concentrations and less acidic pH than reactions utilizing purified APs (23,46). It is therefore possible that AP180 and AP-2 work cooperatively, at relatively low concentrations, to promote efficient clathrin assembly under physiological pH and ionic conditions.
Our observation that AP180 and AP-2 together give greater clathrin assembly activity than expected from simply adding the assembly activity of each protein alone suggests that cooperative effects between AP180 and AP-2 may indeed be important in clathrin assembly. Such cooperative effects may be mediated through a direct interaction between AP180 and AP-2 (Fig. 3). Consistent with previously reported affinity chromatography experiments (36), we found that AP180 interacts directly with the hinge-ear domain of the ␣ subunit of AP-2 (Fig. 6B). In addition, we found that AP180 also interacts directly with the ␤ subunit of AP-2 (Fig. 6A). We determined that the region of AP180 that interacts with both of the large subunits of AP-2 is contained within amino acids 623-680 (Fig.  4). A recombinant AP180 fragment containing this binding site will disrupt both the AP180/AP-2 interaction, and the enhanced clathrin assembly activity obtained when AP180 and AP-2 are added together to an assembly reaction. The latter observation strongly suggests that the enhanced assembly activity of the AP180/AP-2 combination depends upon formation of an AP180-AP-2 complex.
While these studies have demonstrated that AP180 and AP-2 interact directly in vitro, we believe that it is very likely that they also interact in vivo for the following reasons. Both AP180 and AP-2, along with clathrin and dynamin, have been colocalized to clathrin-coated vesicles budding from nerve terminal plasma membranes by immunoelectronmicroscopy (28). Previously, it was found that AP180 and AP-2 could be coimmunoprecipitated from extracts of clathrin-coated vesicles with antibodies to AP180 (14). However, the extracts used in this study were prepared under conditions in which clathrin, a common binding partner for AP180 and AP-2, would have been present, making it impossible to distinguish between the existence of a direct AP180/AP-2 interaction versus formation of a ternary complex with clathrin. We therefore repeated these experiments utilizing extracts of vesicles from which the clathrin was first selectively removed and found that AP180 and AP-2 were still co-immunoprecipitated (Fig. 2B). It has also been shown that AP180 can be pulled out of a detergent extract FIG. 11. Phosphorylation of AP180 by CKII weakens its interaction with AP-2. Pure bovine AP-2 was incubated with glutathione-Sepharose beads to which either GST, or the indicated phosphorylated or unphosphorylated fusion protein was attached. The material bound to each resin was subjected to SDS-PAGE, followed by Western blot analysis using antibodies to the ␣ subunit of AP-2. of brain membranes on an affinity column made with the ear domain of the ␣ subunit of AP-2 (36). The affinity chromatography experiments reported here are complementary, since they show that AP-2 is pulled out of cytosolic extracts with immobilized AP180 (Fig. 2A). Together, these studies suggest that AP180 and AP-2 directly interact in vivo.
In addition to measuring a direct interaction between AP180 and AP-2, we also measured a direct interaction between AP180 and AP-1 in vitro. However, the latter interaction may not be biologically relevant, since AP180 has not been found in the Golgi, where AP-1-containing vesicles are found. The ␤ 2 subunit of AP-2 is highly homologous to the ␤ 1 subunit of AP-1, and it may be that AP180 is binding to both of these proteins through conserved features of the ␤ subunits. In vivo, the specificity of AP180 for AP-2 may be controlled by the spatial distribution of these molecules. These studies also raise the possibility that other members of the tetrameric AP gene family may interact with other members of the monomeric AP gene family in vivo. It may be that every cell membrane utilizes a member of the monomeric AP family, and a member of the tetrameric AP family for clathrin-coated pit formation. Our observation that CALM (22) can directly interact with both AP-1 and AP-2 in vitro 2 is consistent with this proposal.
The in vitro phosphorylation of AP180 by CKII characterized here is also likely to occur in vivo for the following reasons. First of all, AP180, AP-2, and CKII are all found in the same subcellular organelle, the clathrin-coated vesicle (5,11,28,53). Secondly, when phosphorylation studies were carried out on extracts of clathrin-coated vesicles, AP180 was phosphorylated under conditions in which CKII is stimulated (39). Last, AP180 is labeled by [␥-32 P]ATP in cultured neurons (13), and AP180 found in mouse brain synaptosomes can be desphosphorylated by alkaline phosphatase (16).
Our observation that phosphorylation of AP180 affects its binding to AP-2 provides support for the long standing idea that coated vesicle dynamics may be influenced by protein phosphorylation (13,38). Interestingly, AP180 phosphorylation has no detectable effect on AP180 assembly activity. Instead, the effect of AP180 phosphorylation on clathrin assembly appears to be mediated through its effects on AP180-AP-2 assembly. This provides further support for the idea that the enhanced assembly activity of the AP180/AP-2 combination depends upon formation of a complex between these two proteins. While the quantitative effect of AP180 phosphorylation on clathrin assembly by the AP180/AP-2 combination is not large, it is statistically significant and completely consistent with the effect of phosphorylation on AP180-AP-2 affinity.
It may be that the regulation of coat protein phosphorylation is tied to cycles of coated vesicle assembly and disassembly, since there have to be mechanisms that prevent the coat proteins from interacting both with each other and with clathrin in inappropriate situations. For example, within nerve terminals the membrane protein composition of a mature synaptic vesicle is nearly identical to the membrane composition of a clathrincoated vesicle (54), yet somehow the coat proteins are kept off of the synaptic vesicle. In support of this view is a report that phosphorylation of AP-2 modulates its clathrin binding properties (55) as well as a report that phosphorylation by unknown kinase(s) regulates the association and dissociation cycle of the clathrin-based endocytic machinery that consists of clathrin, AP-2, AP180, dynamin, amphiphysin, and Eps15 (56). Of course, other factors might also be involved. For example, inositol lipid phosphorylation might be coordinated with cycles of exo-and endocytosis. Both AP-2 and AP180 bind to inositol lipids (34,35), and the phosphorylation state of the inositol lipid has been shown to differentially affect its inhibition of AP180-mediated clathrin assembly (35). Thus, while the effects of AP180 phosphorylation on clathrin assembly that we measure are not large, they may be only one component of a mechanism that integrates multiple phosphorylation events involving both the protein and lipid components of the membrane trafficking machinery.