Functional Dissection of an AP-2 β2 Appendage-binding Sequence within the Autosomal Recessive Hypercholesterolemia Protein*

The autosomal recessive hypercholesterolemia (ARH) protein plays a critical role in regulating plasma low density lipoprotein (LDL) levels. Inherited defects in ARH lead to a hypercholesterolemia that closely phenocopies that caused by a defective LDL receptor. The elevated serum LDL-cholesterol levels typical of ARH patients and the pronounced accumulation of the LDL receptor at the cell surface of hepatocytes in ARH-null mice argue that ARH operates by promoting the internalization of the LDL receptor within clathrin-coated vesicles. ARH contains an amino-terminal phosphotyrosine-binding domain that associates physically with the LDL receptor internalization sequence and with phosphoinositides. The carboxyl-terminal half of ARH contains a clathrin-binding sequence and a separate AP-2 adaptor binding region providing a plausible mechanism for how ARH can act as an endocytic adaptor or CLASP (clathrin-associated sorting protein) to couple LDL receptors with the clathrin machinery. Because the interaction with AP-2 is highly selective for the independently folded appendage domain of the β2 subunit, we have characterized the ARH β2 appendage-binding sequence in detail. Unlike the known α appendage-binding motifs, ARH requires an extensive sequence tract to bind the β appendage with comparably high affinity. A minimal 16-residue sequence functions autonomously and depends upon ARH residues Asp253, Phe259, Leu262, and Arg266. We suggested that biased β subunit engagement by ARH and the only other β2 appendage selective adaptor, β-arrestin, promotes efficient incorporation of this mechanistically distinct subset of CLASPs into clathrin-coated buds.

Clathrin-coated vesicles are short lived transport intermediates fabricated solely for the preferential sorting and intracel-lular trafficking of select transmembrane proteins, bound ligands, and lipids (1)(2)(3). Clathrin coats assemble at several discrete intracellular sites where, typically, different cargo molecules are sorted. As the two major clathrin-associated heterotetrameric adaptor proteins (AP-1 and AP-2) are generally restricted to particular sorting stations within the cell, a long standing model has been that adaptors direct the process of cargo selection (4). Different subunits of the adaptor complex directly engage different classes of sorting sequences. The 2 subunit binds to YXXØ-type (where Ø is a bulky hydrophobic amino acid) sorting signals in a phosphorylation-dependent manner (3,5,6). In the basal state, the orientation of the 2 subunit within AP-2 makes the YXXØ interaction surface inaccessible. Phosphorylation on 2 residue Thr 156 is believed to alter the conformation of 2 to allow productive engagement of the YXXØ sequence (7)(8)(9). Another sorting sequence termed the (DE)XXXL(LI) motif is recognized by AP-1, AP-2, and AP-3 (6). Recent evidence suggests that the (DE)XXXL(LI) sequence is bound by a hemicomplex of the ␥ and 1 subunits in AP-1 or the ␦ and 3 subunits in AP-3 (10), although the precise location of the interaction site remains to be determined.
Not all cargo molecules sorted into clathrin-coated vesicles appear to utilize heterotetrameric adaptors directly, because not all sorting sequences conform to the YXXØ or (DE)XXX-L(LI) types. At the plasma membrane, there are several other internalization signals/tags known (6), and the growing consensus is that alternate adaptor proteins govern the endocytosis of proteins with these signals (3,11,12). The internalization of low density lipoprotein (LDL) 1 particles is a good example of the activity of alternate adaptors in clathrin-mediated endocytosis. In humans, the LDL receptor, a type I transmembrane glycoprotein, plays a critical role in regulating plasma LDLcholesterol levels. Inherited defects in the receptor cause hypercholesterolemia and, if untreated, early onset coronary heart disease and myocardial infarction. The 50-amino acid cytosolic domain of the LDL receptor contains an 802 FDNPVY sequence that is necessary and sufficient for rapid clathrinmediated endocytosis of LDL (13). Expression of a receptor with a single nucleotide point mutation generating a Tyr 807 3 Cys substitution, which inactivates the FXNPXY internaliza-* This work was supported in part by National Institutes of Health Grant R01 DK53249 and in part by the Senior Vice Chancellor for the Health Sciences, University of Pittsburgh, in support of the Competitive Medical Research Fund. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Supported in part by the Renal and Epithelial Biology Training Grant 5T32 DK061296-02 from the National Institutes of Health and in part by American Heart Association Predoctoral Fellowship Award 0415428U.
ʈ Supported by a Wellcome Senior Research fellowship in basic biomedical Science (to D. J. O.).
** To whom correspondence should be addressed: Dept. of Cell Biology and Physiology, 3500 Terrace St., S325BST, University of Pittsburgh School of Medicine, Pittsburgh, PA 15261. Tel.: 412-648-9711; Fax: 412-648-9095; E-mail: traub@pitt.edu. 1 The abbreviations used are: LDL, low density lipoprotein; ARH, autosomal recessive hypercholesterolemia; CLASP, clathrin-associated sorting protein; DiI, 3,3Ј-dioctadecylindocarbocyanin; GFP, green fluorescent protein; GPCR, G protein-coupled receptor; GST, glutathione S-transferase; His 6 , hexahistidine; mAb, monoclonal antibody; PBS, phosphate-buffered saline; PTB, phosphotyrosine binding; PtdIns(4,5)P 2 , phosphatidylinositol 4,5-bisphosphate; BSA, bovine serum albumin; DTT, dithiothreitol; Ni-NTA, nickelnitrilotriacetic acid; PTB, phosphotyrosine binding; ITC, isothermal titration calorimetry. tion sequence, results in a clinical phenotype that is similar to individuals who express no detectable LDL receptor (14). There is in vitro evidence for the vital FXNPXY sequence binding the AP-2 2 subunit (K d ϳ50 M) but as an atypical variant of the YXXØ-type interaction utilizing the Tyr and distal amino acids (FDNPVYQKT). 2 (The boldface identifies the distal amino acids being referred to). Other biochemical work confirms a weak association with 2 but suggests a binding site separate from the YXXØ interaction surface (15). Yet overexpression of the LDL receptor in HeLa cells does not impede the efficient clathrin-mediated endocytosis of either transferrin or epidermal growth factor receptors (16). In fact, RNA interference silencing of AP-2 protein levels has a negligible effect on the sorting of the FXNPXY sequence, although transferrin uptake is almost completely blocked (17). ␤-Arrestin1 and ␤-arrestin2 can bind physically to the cytoplasmic segment of the LDL receptor in a manner that depends on the FDNPVY Tyr 807 residue and a distal Ser 833 when phosphorylated (18). However, ␤-arrestinnull cells show that whereas these proteins can moderately enhance LDL internalization, they are not necessary for the process (18). Current evidence suggests that LDL internalization is, instead, managed primarily by two phosphotyrosinebinding (PTB) domain proteins, termed Disabled-2 (Dab2) and the autosomal recessive hypercholesterolemia (ARH) protein (19 -23).
The amino-terminal PTB domain of both Dab2 and ARH binds physically to the FXNPXY sequence and, synchronously and noncompetitively, to the plasma membrane phospholipid phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P 2 ) (21,22,24,25). These properties allow specific cargo recognition at the cell surface, and overexpression of a tandem Dab2 PTB domain construct in COS-7 cells prevents LDL uptake without interfering with transferrin internalization (20). Although ARH and Dab2 might be functionally redundant to some degree, ARH appears particularly important for LDL uptake by hepatocytes, the major cell type involved in clearance of circulating LDL (26). Individuals homozygous for defects in the ARH gene have a clinical hypercholesterolemia that is essentially a phenocopy of familial hypercholesterolemia, but with normal LDL receptor production (27,28). Like Dab2 (19,20), the carboxyl-terminal region of ARH binds to both clathrin and the AP-2 adaptor (21)(22)(23). These properties, together with the capacity to bind PtdIns(4,5)P 2 and cargo (the FXNPXY sequence), designate Dab2 and ARH as CLASPs (clathrin-associated sorting proteins). Importantly, ARH is the first human disease directly attributable to a defect in a component of the endocytic clathrin machinery. These CLASPs are presumed to expand the sorting repertoire of the endocytic clathrin coat and to act analogously to two other alternate adaptors, ␤-arrestin1 and -2. Following ligand stimulation, the internalization of many G protein-coupled receptors (GPCRs) is driven by the ␤-arrestins, which rapidly translocate to the phosphorylated GPCR on the cell surface and mesh with existing clathrin-coated structures at the surface by binding to both clathrin and AP-2 (29,30).
To bind to AP-2, ARH, Dab2, the ␤-arrestins, and several other CLASPs engage the independently folded appendage domains that project off the large ␣ and ␤2 subunits of AP-2 (11,31,32). ␤-Arrestin and ARH are notable because they alone have extreme selectivity for ␤2 appendage over the ␣ appendage (21,22). Here we present a detailed analysis of the side chain requirements for specific binding to the AP-2 ␤2 appendage in human ARH. Our studies provide a comprehensive description of the endocytic interaction determinants located within vertebrate ARH proteins.

EXPERIMENTAL PROCEDURES
DNA Constructs-The recombinant proteins consisting of glutathione S-transferase (GST) fused to ARHC1 (human residues 180 -308), AP-2 ␣ C appendage (murine residues 701-938), ␣ C appendage (Q782A) point mutant, AP-2 ␤2 appendage (rat residues 714 -951), and the epsin 1 DPW domain (rat residues 229 -407) fusion proteins have been described previously (22,(33)(34)(35). GST-ARHM1 (human ARH residues 254 -269) was generated by ligation of complementary oligonucleotides, after annealing and digestion, into EcoRI/XhoI cleaved pGEX-4T-1. The GST-ARHM2 was produced in two steps using the GST-ARHM1 plasmid as a template. First, an XbaI site was introduced into the center of the ARHM1 sequence without changing the encoded protein sequence. Then a second annealed oligonucleotide was ligated into EcoRI/XbaIcut GST-ARHM1 to generate residues 248 -269 of human ARH fused in-frame with GST. The green fluorescent protein (GFP)-ARHM2 was constructed similarly but using a single annealed oligonucleotide pair ligated into pEGFP-C1 between the EcoRI and BamHI restriction sites. ARH (full length) and the PTB domains (residues 1-179) were generated by insertion of appropriate PCR products into NheI/BglII digested pEGFP-N1, whereas ARHC1 was cloned into pEGFP-C1 using the EcoRI and BamHI sites. His 6 -tagged ARH was created by cloning PCR amplified full-length ARH between the NdeI and NotI sites of pET28. The GST-␤-arrestin1 (rat residues 331-418) construct was kindly provided by Marc Caron. Directed mutations were introduced into the appropriate vectors using the QuikChange system (Stratagene) with the required mutagenic primer pairs (the sequences of which are all available upon request). All the clones and mutations were verified by automated dideoxynucleotide sequencing.
Protein and Tissue Extract Preparation-GST and the various GST fusion proteins were produced in Escherichia coli BL21 cells. The standard induction protocol entails shifting log-phase cultures (A 600 ϳ0.6) from 37°C to room temperature and then adding isopropyl-1-thio-␤-Dgalactopyranoside to a final concentration of 100 M. After 3-5 h at room temperature with constant shaking, the bacteria were recovered by centrifugation at 15,000 ϫ g max at 4°C for 15 min and stored at Ϫ80°C until used. Bacteria were lysed on ice in 50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 0.2% (w/v) Triton X-100, 10 mM ␤-mercaptoethanol with sonication or in B-PER reagent (Pierce). Insoluble material was removed by centrifugation at 23,700 ϫ g max at 4°C for 15 min, and then the GST fusions were collected on GSH-Sepharose. After extensive washing in PBS, GST fusions were eluted with 10 mM Tris-HCl, pH 8.0, 10 mM GSH, 5 mM DTT on ice and dialyzed into PBS, 1 mM DTT before use in binding assays. Several of the purified fusion proteins were cleaved from the GST with thrombin (Amersham Biosciences) while still immobilized upon GSH-Sepharose. Digestion was as recommended by the manufacturer, followed by addition of the irreversible thrombin inhibitor D-Phe-Pro-Arg chloromethyl ketone (Calbiochem) to a final concentration of 25 M. His 6 -ARH was produced in E. coli BL21 (DE3) by induction of log-phase cultures with 1 mM isopropyl 1-thio-␤-Dgalactopyranoside at room temperature for 3 h. Cleared lysates were incubated with Ni-NTA-agarose, and bound protein was eluted in 50 mM Tris-HCl, pH 7.5, 300 mM sodium chloride, 100 mM imidazole.
Cytosol was prepared from frozen rat brain (PelFreez) by sequential differential centrifugation after homogenization in 25 mM HEPES-KOH, pH 7.2, 250 mM sucrose, 2 mM EDTA, and 2 mM EGTA supplemented with 1 mM phenylmethylsulfonyl fluoride and Complete (Roche Applied Science) protease inhibitor mixture. The 105,000 ϫ g max supernatant is defined as cytosol and was stored in small aliquots at Ϫ80°C. Before use in binding assays, thawed samples of rat brain cytosol were adjusted to 25 mM HEPES-KOH, pH 7.2, 125 mM potassium acetate, 5 mM magnesium acetate, 2 mM EDTA, 2 mM EGTA, and 1 mM DTT (assay buffer) by addition of a 10ϫ stock and then centrifuged at 245,000 ϫ g max (TLA-100.4 rotor) at 4°C for 20 min to remove insoluble particulate material.
Binding Assays-Pull-down-type assays, in a total volume of 300 l, were as described previously (34,36). Typically, 50 -400 g of GST and 2 D. J. Owen, unpublished observations. ; lanes e and f) immobilized on GSH-Sepharose was incubated with rat brain cytosol. After centrifugation, aliquots corresponding to 1/75 of each supernatant (S) and 1/8 of each washed pellet (P) were resolved by SDS-PAGE and either stained with Coomassie Blue or transferred to nitrocellulose. Portions of the blots were probed with the anti-clathrin HC mAb TD.1 and anti-AP-1 ␤1 and AP-2 ␤2 subunit mAb 100/1, or anti-2 subunit serum or affinity-purified AE/1 anti-AP-1 ␥ subunit, or RY/1 anti-1 subunit antibodies. B, ϳ50 g of immobilized GST (lanes a and b) or GST-␤2 appendage (lanes c-j) were incubated with rat brain cytosol alone (lanes a-d) or with cytosol supplemented with 20 M epsin 1 DPW domain (residues 229 -407; lanes e and f), 20 M human ARH (residues 180 -308; lanes g and h), or 20 M ␤-arrestin1 (residues 331-418; lanes i and j). After centrifugation, aliquots of ϳ1/40 of each supernatant (S) and 1/6 of each washed pellet (P) were resolved by SDS-PAGE and either stained with Coomassie Blue or transferred to nitrocellulose. Portions of the blots were probed anti-epsin 1 antibodies or anti-AP180 or anti-amphiphysin mAbs, or an anti-eps15 polyclonal serum. The arrowhead indicates a nonspecific band detected in all the supernatant fractions by the anti-eps15 serum used. Note the complete inhibition of binding by the ARH polypeptide, save the eps15 interaction. C, approximately 1 (lanes a and b), 2 (lanes c and d), or 4 g of His 6 -ARH (lanes e-p), immobilized on Ni-NTA-agarose, was incubated with 1 g of GST-␤2 appendage (lanes a-f) or GST-␤2(W841A) (lanes g and h), GST-␤2(E849A) (lanes i and j), GST-␤2(Y888V) (lanes k and l), GST-␤2(E902A) (lanes m and n), or GST (lanes o and p) in the presence of 0.25 mg/ml carrier acetylated BSA (AcBSA). After centrifugation, aliquots corresponding to 1/60 of each supernatant (S) and 1/8 of each washed pellet (P) were resolved by SDS-PAGE and either stained with Coomassie Blue or transferred to nitrocellulose. The blot was probed with an anti-GST mAb. D, ribbon representation of the structure of the bilobed ␤2 subunit appendage. The location of the four side chains mutated is shown on the platform subdomain. The extent of binding of wild type GST-␤2 appendage or the W841A, E849A, Y888V, or E902A mutants to immobilized His 6 -ARH is shown semi-quantitatively in the inset table.
the GST fusion proteins were first each immobilized upon ϳ25 l of packed GSH-Sepharose by incubation at 4°C for 2 h with continuous mixing. The Sepharose beads containing the required immobilized proteins were then washed and resuspended to 50 l in assay buffer. Clarified rat brain cytosol or purified, thrombin-cleaved ␣ C or ␤2 appendage (in the presence of 0.1 mg/ml carrier BSA) was added, and the tubes were incubated at 4°C for 60 min with continuous gentle mixing. For the competition assays, thrombin-cleaved proteins were added directly into the assay mixture to a final concentration of 20 M in the presence of 25 M D-Phe-Pro-Arg chloromethyl ketone. The GSH-Sepharose beads were then recovered by centrifugation at 10,000 ϫ g max at 4°C for 1 min, and an aliquot of each supernatant was removed and adjusted to 100 l with SDS-sample buffer. After washing the GSH-Sepharose pellets four times each with ϳ1.5 ml of ice-cold PBS by centrifugation, the supernatants were aspirated, and each pellet was resuspended in SDS-sample buffer.
Isothermal Titration Calorimetry (ITC) Measurements-ARH-derived peptides were synthesized by the Peptide Synthesis Facility of the University of Pittsburgh. The GST-␤2 appendage fusion protein, produced in E. coli (DE3) pLysS (Novagen), was purified by affinity chromatography using GSH-Sepharose resin. The column matrix was washed extensively with 10 mM HEPES, pH 7.5, 200 mM sodium chloride, 4 mM ␤-mercaptoethanol and incubated overnight with thrombin (Serva Electrophoresis). ␤2 appendage was eluted, concentrated, and further purified by Superdex 200 gel filtration chromatography in 50 mM HEPES, pH 7.5, 100 mM sodium chloride, 0.5 mM DTT. Mutant ␤2 appendage was expressed and purified as per wild type protein. ITC experiments were performed on a VP-ITC instrument (Microcal) at 10°C in ITC buffer (50 mM HEPES, pH 7.5, 100 mM sodium chloride). 4 -8 l of 1 mM ARH-derived peptide was injected into 1.4 ml of 90 M ␤2 appendage in 31-33 aliquots. The titration data were integrated and normalized in Origin (Microcal) to determine the reaction stoichiometry, n, and equilibrium constant, K a (ϭ K d Ϫ1 ). Yeast Two-hybrid Interaction Screens-The Matchmaker GAL4 twohybrid system (Clontech) was used. ARHC1 (residues 180 -308) was cloned into EcoRI-and BamHI-cut pGBKT7, whereas the ␤2 appendage sequence was placed into pGADT7 digested with EcoRI and XhoI. After confirmation of the insert sequences, mutations were generated using the QuikChange system (Stratagene). Saccharomyces cerevisiae strain AH190 was transformed with the appropriate plasmid combinations according to the manufacturer's protocols and selected first on SD minimal medium plates lacking Leu and Trp. Individual clones were selected and then streaked/spotted onto plates of SD medium lacking Leu and Trp, on plates without His, Leu, and Trp, or onto plates lacking Ade, His, Leu, and Trp.
Electrophoresis and Immunoblotting-Samples were resolved on polyacrylamide gels prepared with an altered acrylamide/bisacrylamide (30: 0.4) ratio stock solution. The decreased cross-linking generally improves resolution but also affects the relative mobility of several proteins, most noticeably AP180 and epsin 1. After SDS-PAGE, proteins were either stained with Coomassie Blue or transferred to nitrocellulose in ice-cold 15.6 mM Tris, 120 mM glycine. Blots were usually blocked overnight in 5% skim milk in 10 mM Tris-HCl, pH 7.8, 150 mM NaCl, 0.1% Tween 20, and then portions were incubated with primary antibodies as indicated in the  DDGLDEAFSRLAQSRT c ␤2 appendage platform subdomain individual figure legends. After incubation with horseradish peroxidaseconjugated anti-mouse or anti-rabbit IgG, immunoreactive bands were visualized with enhanced chemiluminescence. Cell Culture, Transfections, and Immunofluorescence-CV-1 and HeLa SS6 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum at 37°C in a humidified atmosphere with 10% CO 2 and plated onto 12-mm round glass coverslips 1 day prior to transfection. For transient transfections in a 35-mm dish, 6 l of Lipofectamine 2000 (Invitrogen) was mixed with 375 l of Opti-MEM (Invitrogen) and 500 ng of DNA and incubated at room temperature for 45 min before adding the mixture to the dish. The following day the cells were permeabilized for 1 min on ice in 0.3% saponin in assay buffer, fixed for 20 min in 2% paraformaldehyde, blocked with 10% goat serum, 0.2% saponin in PBS for 30 min, and finally incubated sequentially with primary and secondary antibodies diluted with 10% goat serum and 0.05% saponin in PBS each for 1 h. Epifluorescent images were acquired using a 63ϫ objective lens on a Nikon Microphot-FXL using appropriate excitation and emission filter cubes, and confocal microscopy was performed using an Olympus Fluoview BX61 confocal with argon 488 and helium-neon 543 laser lines and appropriate filters.
3,3Ј-Dioctadecylindocarbocyanin (DiI)-LDL Internalization-HeLa cells were grown and transfected as described above, except that the cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% lipoprotein-deficient fetal calf serum following transfection to up-regulate LDL receptors. Cells were incubated with 10 g/ml DiI-LDL (Biomedical Technologies) for 40 min at 37°C and then fixed with 2% paraformaldehyde for 20 min for examination by epifluorescence microscopy. For quantitation, fields of cells were imaged using a 20ϫ objective and analyzed using Metamorph (Universal Imaging). A binary mask corresponding to all of the transfected cells was generated and applied to the red channel (LDL). These transfected cells were then manually scored for no internalization (8-bit DiI-LDL intensity Ͻ80), partial internalization (structures present with 8-bit intensity 80 -180), or normal internalization (structures present with 8-bit intensity Ͼ180). For transferrin internalization, cells were transfected and cultured in lipoprotein-deficient medium as above but were incubated in serum-free medium for 1 h prior to the addition of 10 g/ml DiI-LDL for 25 min at 37°C, followed by addition of 25 g/ml biotin-conjugated human transferrin and further incubation at 37°C for an additional 15 min. Cells were then fixed in 2% paraformaldehyde for 20 min, blocked, and permeabilized with 10% goat serum, 0.05% saponin in PBS for 30 min, incubated with Cy5-conjugated streptavidin with 10% goat serum, 0.05% saponin in PBS for 1 h, washed, mounted, and then imaged.

RESULTS AND DISCUSSION
ARH and ␤-Arrestin1 Association with the AP-2 ␤2-Subunit Appendage-We compared the ability of GST-ARHC1 (residues 180 -308) and a GST fusion of the carboxyl-terminal segment of ␤-arrestin1 (residues 331-418) to associate with cytosolic clathrin and adaptors in vitro. Both fusion proteins affinityisolate clathrin and AP-2 (Fig. 1A, lanes d and f) as judged by the ␤2 and 2 subunits of the heterotetrameric adaptor complex. Specific antibodies directed against the ␥ and 1 subunits of the Golgi/endosome-localized AP-1 adaptor complex show that, in vitro, AP-1 is also able to bind to both the immobilized ARH and ␤-arrestin1 fusions as well (Fig. 1A, lanes d and f, left  panels). GST alone, however, fails to interact with these clathrin-associated components (Fig. 1A, lane b), and the proteins remain completely within the unbound supernatant fraction (Fig. 1A, lane a).
Experiments utilizing clathrin-depleted rat brain cytosol show that both AP-1 and AP-2 still interact efficiently with immobilized GST-ARHC1, excluding the possibility that the presence of AP-1 in these pull-down experiments is indirectly mediated by clathrin (data not shown). Rather, the association of the expressed ARH and ␤-arrestin1 segments with both AP-1 and AP-2 in vitro reflects the strong selectivity for the ␤ appendage that projects off the heterotetrameric adaptor core (21,22,(37)(38)(39). The primary sequence of the ␤1 and ␤2 appendage is 85% identical, and indeed, the carboxyl-terminal regions of either ARH or ␤-arrestin1 can inhibit productive association of soluble rat brain accessory proteins with the immobilized  g and h) immobilized on GSH-Sepharose was incubated with rat brain cytosol. After centrifugation, aliquots corresponding to 1/40 of each supernatant (S) and 1/8 of each washed pellet (P) were resolved by SDS-PAGE and either stained with Coomassie Blue or transferred to nitrocellulose. Portions of the blots were probed with the anti-clathrin HC mAb TD.1 and anti-AP-1 ␤1 and AP-2 ␤2 subunit mAb 100/1, or anti-2 subunit serum. B, ITC binding studies of ARH-derived peptide binding to wild type or mutant ␤2 appendage. Raw ITC data (top) and concentration-normalized integrated data (bottom) from calorimetric titrations of 1 mM ARH-derived peptide into 0.09 mM ␤2 appendage are shown. C, equilibrium dissociation constants (K d ) and reaction stoichiometries (n).
GST-␤2 appendage. Alone, the GST-␤2 appendage binds to a characteristic cohort of partner proteins (40) (Fig. 1B, lane d), including AP180, epsin 1, and traces of amphiphysin I and II. These proteins do not pellet with GST alone (Fig. 1B, lane b) and are instead present within the supernatant fraction (Fig.  1B, lane a). In the presence of a 5-fold molar excess of monomeric ARHC1 protein over the ␤2 appendage, AP180, epsin, and amphiphysin also remain entirely soluble (Fig. 1B, lane g). This indicates that ARH engages a common surface upon the ␤2 appendage platform subdomain, occluding other partner interactions. By contrast, in a similar type of competition assay, the ARHC1 segment has negligible inhibition on ␣ C appendage binding (22).
In other experiments designed to validate that ARH engages the same interaction surface on the ␤2 appendage bound by the ␤-arrestins (37-39), we observed that alteration of certain platform subdomain residues strongly decrease ARH associations. In particular, W841A and Y888V substitutions essentially eliminate the binding of His 6 -ARH in vitro (Fig. 1C). These two aromatic residues are situated on the perimeter of the putative hydrophobic binding pocket of the ␤2 platform subdomain (Fig.  1D) (40). Two other side chains in this region, Glu 849 and Glu 902 , when changed to Ala, reduce but do not abolish His 6 ARH binding (Fig. 1, C and D). The involvement of these two acidic residues agrees with an earlier study (21). Most interestingly, none of the added protein segments that bind to the ␤2 appendage platform subdomain or the platform mutations (40) interfere significantly with the association of eps15 with the ␤2 appendage (Fig. 1B). This suggests eps15 might use an alternative binding surface to engage the ␤2 appendage.
An Autonomous and Transplantable ␤2 Appendage Interaction Sequence-Previous truncation studies (21,22) reveal that side chains between ARH residues 263 and 283 are required for productive ␤ appendage engagement. With the objective of attempting to decipher recognizable interaction motifs for the ␤ appendage, we performed alanine-scanning mutagenesis in the context of the GST-ARHC1 model protein to pinpoint more precisely the AP-2 binding region. Conversion of 256 DEAF to AAAA (Ala 256 -259 ) or 260 SRLA to AAAA (Ala 260 -263 ) almost completely eliminated AP-2 binding (Fig. 2, lanes j and l) without altering clathrin binding detectably in pull-down-type experiments. A 264 QSRT to AAAA substitution (Ala 264 -267 ) also had a significant inhibitory effect (Fig. 2, lane n), in agreement with the reported importance of Arg 266 in AP-2 binding (21). A 252 DDGL to AAAA switch (Ala 252-255 ) was also inhibitory, albeit less severe (lane h), whereas more proximal or distal tetraalanine substitutions had no notable effect on adaptor associations. Clathrin engagement by GST-ARHC1 was not affected by any of the mutations tested. This reiterates that the clathrin and AP-2 binding determinants within ARH are clearly distinct (21,22).
Our mutagenesis experiments further delimited the AP-2 binding region within ARH to a 16-residue stretch 252 DDG-LDEAFSRLAQSRT. This tract is considerably longer than the DP(FW) (41)(42)(43), FXDXF (43), and WXX(FW)X(DE) (44 -46) sequence motifs that engage the alternative AP-2 appendage, the ␣ subunit appendage (Table I). To determine whether this ARH sequence can interact with AP-2 in isolation, we fused either GLDEAFSRLAQSRTNP (GST-ARHM1) or VWELDDG-LDEAFSRLAQSRTNP (GST-ARHM2) in-frame with GST. In pull-down assays, the GST-ARHM1 fusion protein binds to AP-2 extremely weakly (Fig. 3A, lane d) but is clearly above the signal observed with the GST control protein (lane b). The GST-ARHM2 fusion protein encompassing the entire 16-residue binding stretch binds to AP-2 indistinguishably from the entire carboxyl-terminal half of ARH (GST-ARHC1, Fig. 3A,  lane h). This authenticates the mapped sequence as an independent and transplantable AP-2 binding region. Some clath-rin is also recovered with the GST-ARHM2 pellet (Fig. 3A, lane f) despite the absence of the proximal 212 LLNLD clathrin box sequence. Notably however, the amount of clathrin bound is considerably less than that seen with the GST-ARHC1 (Fig. 3A, lane h) and may relate to our (36) and others (39,47) observations of overlapping AP-2 and clathrin binding properties of some interaction sequences.
ITC experiments confirm both that the whole 22-mer ARHM2 segment and the minimal 16-residue DDGLDEAFSR-LAQSRT sequence bind to the platform subdomain of the ␤2 appendage with moderately high affinity and that a smaller ARH-derived peptide is incompetent to engage the ␤2 appendage. Titrating a 16-residue ARH-derived DDGLDEAFSR-LAQSRT peptide into a solution of purified monomeric ␤2 appendage indicates that the K d for this interaction is ϳ2 M (Fig. 3, B and C), whereas the longer 22-residue ARHM2 peptide displays a slightly higher affinity (K d ϭ 1.3 M). The extended sequence is necessary for productive binding to the ␤2 appendage as a shorter peptide, which corresponds to the internal core of the sequence (DEAFSRLA), exhibits no detectable interaction (data not shown). These results, along with the pull-down data (Fig. 3A), clearly indicate that the internal EAF tripeptide sequence is not able mimic the DP(FW) motif to promote optimal ␤2 appendage engagement. The ITC data also reveal that there is a single binding site for the ARH peptide upon the ␤2 appendage. With a K d of 1-2 M, the affinity of ARH for the ␤2 appendage is ϳ5-fold higher than the best interaction motif for the AP-2 ␣ subunit appendage; the WXX(FW)X(DE) motif binds the ␣ appendage with a K d of ϳ10 M, albeit to a topologically distinct surface, located on the sandwich subdomain of the ␣ appendage (35). Mutation of key residues (Trp 841 and Tyr 888 ) that comprise the site of high hydrophobic potential, believed to be the major ligand binding surface on the ␤2 appendage platform subdomain (Fig. 1D), disrupt ARH peptide binding without significantly affecting appendage folding (confirmed by CD spectroscopy, not shown). A W841A platform mutation alters the K d for the ARH peptide Ͼ40-fold, whereas a Y888V substitution abolishes any detectable association with the ARH peptide (Fig. 3, B and C). Thus, this work confirms the presence of spatially analogous proteinprotein interaction sites on the ␣ and ␤2 platform subdomains but shows that although the ␣ site recognizes short DP(FW) and FXDXF peptides with relatively low affinity (K d ϳ50 -100 M), the ␤2 site can recognize a longer sequence with much higher affinity (K d ϳ1-2 M).
To validate the activity of the ␤2 appendage-binding sequence in vivo, we fused residues 180 -308 (ARHC1) or residues 248 -269 (ARHM2) to the carboxyl terminus of GFP. In both cases, transient transfection of the fusion protein into either CV-1 or HeLa cells targets the GFP to a profusion of punctate structures randomly scattered throughout the cell (Fig. 4, A and C, left panels), although a diffuse cytosolic pool is also evident. The vast majority of the punctate structures colocalize with AP-2 (see insets in A and C), positively identifying them as endocytic clathrin-coated regions. Thus, the ARHM2 sequence alone appears able to bind to the ␤2 appendage in vivo.
Yeast Two-hybrid Analysis-The interaction between the ARHC1 segment and the ␤2 appendage is sufficient to reconstitute GAL4-dependent transcriptional activity in S. cerevisiae (Fig. 5). Yeast transformed with ARHC1 fused to the GAL4 DNA binding domain and the entire ␤2 appendage fused to the GAL4 activation domain grow on plates lacking Ade, His, Leu, and Trp, whereas either plasmid alone cannot sustain growth under these conditions (Fig. 5). Importantly, four separate alterations (W841A, E849A, Y888V, and E902A) to ␤2 appendage platform residues abolish the interaction with ARH, in accord with all the biochemical data ( Figs. 1 and 3). These results validate the binary association of ARH with the AP-2 ␤2 appendage in an independent assay system.
Delineation of Key Residues within the ARH ␤2 Appendage Binding Tract-Comparison of the minimal ARH ␤2-binding segment with the ␤2 binding region from ␤-arrestin1 and -2 shows some common sequence features. In the ␤-arrestins, several functionally important side chains (Fig. 6A, orange asterisks) have been identified previously (38,39,48). Similar analysis of selected residues within the ARH ␤2-binding tract clearly reveals that Phe 259 and Leu 262 are required for append-age binding in vitro. Neither GST-ARHC1 (F259A) nor GST-ARHC1 (L262A) binds to cytosolic AP-2 (Fig. 6B, lanes l and p  compared with d), although the clathrin interaction remains intact. Analogous results are obtained in affinity interaction assays using purified components, including full-length His 6tagged ARH (Fig. 6C); thus these experiments define at least one key residue within the DEAF and SRLA sequence blocks which, when switched to tetra-alanine, essentially abolish AP-2 binding (Figs. 2 and 6C). Most significantly, Phe 259 is phylogenetically conserved among vertebrate ARH proteins and can be aligned with a phylogenetically conserved Phe in the ␤-arrestin binding region. Alteration of this corresponding Phe (F391A in FIG. 6. Identification of key side chains in the ARH AP-2-binding sequence. A, alignment of the ␤2 appendage binding regions of human (Hs) ARH and rat (Rn) ␤-arrestin1 and rat ␤-arrestin2 with phylogenetically conserved residues boxed in magenta, chemically conserved residues in yellow, and conserved residues between ARH and the ␤-arrestins are outlined in purple. Amino acids that on the basis of interaction assays are identified to play a critical role in either ␤-arrestin (orange asterisks) or ARH (green asterisks) are shown, as is the relative location of the ARHM1 and ARHM2 segments. B, approximately 50 g of either GST (lanes a and b), GST-ARHC1 (lanes c and d), GST-ARHC1 (D253A) (lanes e and f), GST-ARHC1 (E257A) (lanes g and h), GST-ARHC1 (E257K) (lanes i and j), GST-ARHC1 (F259A) (lanes k and l), GST-ARHC1 (R261A) (lanes m and n), GST-ARHC1 (L262A) (lanes o and p), or GST-ARHC1 (R266A) (lanes q and r) immobilized on GSH-Sepharose was incubated with rat brain cytosol. After centrifugation, aliquots corresponding to 1/40 of each supernatant (S) and 1/8 of each washed pellet (P) were resolved by SDS-PAGE and either stained with Coomassie Blue or transferred to nitrocellulose. Portions of the blots were probed with the anti-clathrin HC mAb TD.1 and anti-AP-1 ␤1 and AP-2 ␤2 subunit mAb 100/1, or anti-2 subunit serum. C, approximately 1 g of either His 6 -ARH (lanes a and b), His 6 -ARH (Ala 252-255 ) (lanes c and d), His 6 -ARHC1 (Ala 256 -259 ) (lanes e and f), His 6 -ARHC1 (F259A) (lanes g and h), His 6 -ARHC1 (R266A) (lanes i and j), or no His 6 -tagged protein (lanes k and l) immobilized on Ni-NTA-agarose was incubated with 1 g of GST-␤2 appendage in the presence of carrier BSA. After centrifugation, aliquots corresponding to 1/40 of each supernatant (S) and 1/8 of each washed pellet (P) were resolved by SDS-PAGE and either stained with Coomassie Blue or transferred to nitrocellulose. The blot was probed with an anti-GST mAb. D, approximately 50 g of either GST (lanes a and b), GST-ARHC1 (lanes c and d), GST-ARHC1 (D252A) (lanes e and f), GST-ARHC1 (D253A) (lanes g and h), or GST-ARHC1 (DD 3 AR) (lanes i and j) immobilized on GSH-Sepharose was incubated with rat brain cytosol. After centrifugation, aliquots corresponding to 1/75 of each supernatant (S) and 1/8 of each washed pellet (P) were resolved by SDS-PAGE and either stained with Coomassie Blue or transferred to nitrocellulose. Portions of the blots were probed with the anti-AP-1 ␣ subunit mAb 100/2, anti-2 subunit serum, or the anti-clathrin HC mAb TD.1.
␤-arrestin1) also completely abolishes AP-2 binding without similarly affecting the clathrin association (38,39,48). That Phe 259 is critical for ARH binding to AP-2 is further underscored by the effect of the F259A substitution on the intracellular distribution of transiently expressed GFP-ARHC1 and GFP-ARHM2 (Fig. 4, B and D). In both cases, colocalization with AP-2 is lost. We also find that Arg 266 is necessary for ␤2 appendage engagement (Fig. 6, A, green asterisks, and B), as reported by others (21). Once again, this underscores the extent of the linear sequence tract necessary for productive binding to AP-2. Importantly, introduction of the disruptive F259A mutation into ARH in the context of the yeast two-hybrid system similarly abolishes the association, and the transformed yeasts grow upon plates lacking Leu and Trp but not on plates without Ade, His, Leu, and Trp (Fig. 5).
In both ARH and the ␤-arrestins, a vicinal Asp doublet is positioned five amino acids before the necessary Phe residue (Fig. 6A). A D252A substitution does not change adaptor binding appreciably, whereas a D253A mutation has a moderate effect on the ARH⅐AP-2 interaction (Fig. 6, B, lane f compared with lane d, and D), in accord with the relatively weaker inhibitory effect of the DDGL tetra-alanine substitution (Ala 252-255 , Fig. 2). However, substitution of the Asp doublet to Ala-Arg (GST-ARHC1(DD 3 AR)) results in strong inhibition of AP-2 binding without interfering with the clathrin association (Fig. 6D). The involvement of these Asp residues (particularly Asp 253 ) in binding to AP-2 most probably explains the difference observed in the AP-2 ␤2 appendage binding properties of the core DEAFSRLA peptide compared with the longer ARH peptides encompassing the DDGL block (Fig. 3). This acidic residue pair has yet to be evaluated in the ␤-arrestin system.
A number of amino acids that have a strong negative effect on ␤-arrestin1/2-␤2 appendage interactions when altered (38,39,48) have only weak or negligible effects on ARH⅐␤2 binding and vice versa. For example, an ARH R261A substitution has minimal effect on AP-2 binding (Fig. 6B, lane n), although mutation of the analogous residue (Arg 394 ) in ␤-arrestin2 has severe inhibitory consequences (48). On the other hand, the ARH E257K change does interfere with AP-2 binding (Fig. 6B, lane j) but an analogous alteration in ␤-arrestin1 (E389R) has minimal effect (38). And despite the apparent requirement for basic residues at the distal end of the binding sequence, the number, chemical nature, and positioning of these residues differ between ARH and the ␤-arrestins (38,39). In summary, this analysis shows that although certain phylogenetically conserved residues are commonly positioned within the regions of ARH and the ␤-arrestins that bind to the ␤2 appendage, these do not necessarily correspond to key functional residues. It remains to be determined whether these shared amino acids play another role, perhaps in the regulation of the accessibility of the ␤2-binding sequence.
Functional Assessment of the Role of the ␤2 Appendage-and Clathrin-binding Motifs in ARH-Finally, we transiently expressed GFP-tagged ARH in HeLa cells to monitor the effect on LDL uptake relative to neighboring untransfected cells. This approach was selected because it has unfortunately become apparent that it is experimentally unfeasible to introduce ARH mutants into immortalized lymphoblasts derived from ARH patients, 3 and these are the only available cell types derived from ARH patients that exhibit a defect in LDL receptor uptake in vitro (49). Our transfection experiments show that overexpression of full-length ARH potently inhibits the internalization of fluorescent DiI-LDL, blocking DiI-LDL internal-ization completely in 57% of transfected cells and significantly impairing uptake in 39% of transfected cells (Fig. 7, A and E). Because the PTB domain of ARH binds synchronously to PtdIns(4,5)P 2 and the FXNPXY internalization sequence (22), and, alone, overexpression of the related Dab2 PTB domain blocks LDL internalization (20), we also tested the effect of overexpression of only the ARH PTB domain. We find that although this portion of ARH also inhibits LDL internalization, the inhibition is less severe, blocking DiI-LDL internalization in 31% of cells and impairing it in 42% of cells (Fig. 7, B and E). Notably, this inhibition is not due to a general inhibition of endocytosis by ARH but rather a specific impairment of LDL uptake. Internalization of transferrin is normal in cells overexpressing full-length ARH-GFP (Fig. 7C, compare DiI-LDL (red) to transferrin (blue)). Importantly, overexpression of ARH-GFP also does not alter the steady-state distribution of either clathrin or AP-2 (Fig. 7D). In cells expressing low levels of the fluorescent ARH, good colocalization of this protein and AP-2 is clearly evident (Fig. 7D, arrowheads), but higher levels of expression do not prevent clathrin and AP-2 from forming the characteristic punctate structures at the cell surface.
Because the inhibition of DiI-LDL uptake caused by the ARH PTB domain did not quantitatively mirror that of full-length ARH, we measured LDL uptake in cells expressing full-length ARH with mutations in either the ␤2 appendage binding region, the clathrin binding region, or both. Similar to expression of the ARH PTB domain alone, expression of full-length ARH bearing either F259A or R266A mutations inhibits DiI-LDL internalization on a par with that of the ARH PTB domain, with 38 and 43% of cells transfected with ARH F259A and R266A, respectively, demonstrating completely impaired DiI-LDL uptake (Fig. 7E). Interestingly, although disruption of the clathrin-binding site reduces ARH-mediated inhibition of DiI-LDL uptake (59% of cells inhibited), simultaneous mutation of both the clathrin-binding site and the ␤2-binding site does not significantly relieve the DiI-LDL inhibition beyond that observed with ARH F259A substitution alone (Fig. 7E). Our interpretation of all these uptake results is that the PTB domain and the ␤2-binding site are the most important functional regions of ARH with respect to LDL internalization and, together, act in parallel to produce maximal inhibition upon overexpression by preventing LDL receptor sorting into clathrin coats both at the level of cargo recognition and incorporation into existing clathrin-coated structures.
Conclusions-The minimal 16-residue sequence we have delineated within human ARH that appears both necessary and sufficient to bind to the platform subdomain of the ␤2 appendage is noticeably longer than any other known endocytic interaction motif (Table I). The affinity of this interaction is also significantly higher than other interactions between short binding motifs and the ␣ and ␤2 appendages or the terminal domain of the clathrin HC (Table I). In common with these motifs, however, is a critical aromatic side chain that, in all probability, packs into the cleft of high hydrophobic surface potential on the ␤2 platform (40). This supposition is consistent with the pronounced effect that either a W841A or Y888V substitution has on the binding of ARH to the ␤2 appendage in vitro. We have been unable to verify the reported importance of the DLF triplet at the carboxyl terminus of Xenopus ARH (23); neither truncation nor site-directed mutagenesis of this triplet in human ARH affected AP-2 binding (data not shown), and this DLF triplet is not conserved in the Zebrafish ARH sequence.
It is noteworthy that the ARH carboxyl-terminal segment can out-compete all of the ␤2 appendage binding partners present in brain cytosol, except for eps15. This validates the higher apparent affinity of the ARH interaction motif compared with the tandemly arrayed DP(FW) motifs found in proteins that also bind to the platform subdomain of the AP-2 ␣ appendage, like epsin, AP180, and amphiphysin (43). Yet, unlike these proteins, ARH does not bind to the ␣ appendage (22). This probably reflects the fact that the DP(FW) motif engages both appendage types in the context of a type I ␤-turn that inserts the vital aromatic side chain into a hydrophobic cleft in the platform without numerous other peripheral contacts (43).
Consequently, it appears that the ARH (and ␤-arrestin) sequence is tailored for favorable extended associations with the ␤2 appendage platform, whereas the other proteins, which also engage the ␣ appendage, display less optimal and/or less extensive modes of ␤2 platform engagement. We therefore suggest that, because of the higher affinity of dedicated interaction motifs, the ␤2 platform site represents a relatively privileged binding surface that allows for efficient association of both ARH and the ␤-arrestins with the assembling endocytic clathrin coat. Is there any experimental evidence for the ␤2 platform being a privileged site? We believe that our data showing that both GFP-ARHC1 and GFP-ARHM2 fusion proteins target, without disrupting, AP-2-containing structures in the cell and that full-length ARH overexpression potently inhibits LDL uptake without interfering with the endocytosis of transferrin, which is totally dependent on AP-2 (17,50), supports the notion that the ␤2 appendage platform can be accessed as a privileged interaction surface.
The biological importance of a relatively restricted binding site for association with the polymerizing clathrin coat is that it ensures effective sorting of designated cargo by CLASPs, like ARH and the ␤-arrestins, that contain only a single AP-2 interaction motif. In other words, a privileged site guards against the alternate adaptor being quickly dislodged by other proteins like epsin or AP180/CALM. The latter group of endocytic proteins appears to function in a mechanistically distinct manner, because multiple, tandemly arrayed AP-2 (and clathrin) interaction motifs ensure that these proteins are steady-state components of clathrin coats assembling at the cell surface. Indeed, whereas immunofluorescence and electron microscopy show proteins like CALM, Dab2, and epsin are routinely concentrated in clathrin structures (20,34,51), live cell imaging shows that ␤-arrestin is normally soluble in the cytosol and only translocates to the cell surface following agonist stimulation of GPCRs to promote desensitization and clathrin-dependent internalization (52,53). A similar molecular ushering activity, governed in part by the ␤2 platform interaction sequence, could account for the critical role ARH plays in the internalization of LDL particles by the liver.