Recognition of the Tryptophan-based Endocytosis Signal in the Neonatal Fc Receptor by the μ Subunit of Adaptor Protein-2*

Endocytosis of membrane proteins is typically mediated by signals present in their cytoplasmic domains. These signals usually contain an essential tyrosine or pair of leucine residues. Both tyrosine- and dileucine-based endocytosis signals are recognized by the adaptor complex AP-2. The best understood of these interactions occurs between the tyrosine-based motif, YXXΦ, and the μ2 subunit of AP-2. We recently reported a tryptophan-based endocytosis signal, WLSL, contained within the cytoplasmic domain of the neonatal Fc receptor. This signal resembles YXXΦ. We have investigated the mechanism by which the tryptophan-based signal is recognized. Both interaction assays in vitro and endocytosis assays in vivo show that μ2 binds the tryptophan-based signal. Furthermore, the WLSL sequence binds the same site as YXXΦ. Unlike the WXXF motif, contained in stonin 2 and other endocytic proteins, WLSL does not bind the α subunit of AP-2. These observations reveal a functional similarity between the tryptophan-based endocytosis signal and the YXXΦ motif, and an unexpected versatility of μ2 function.

Membrane proteins that enter cells at coated pits typically contain endocytosis signals in their cytoplasmic domains. Almost all of the endocytosis signals that have been described contain an essential tyrosine or pair of leucine residues (for review, see Ref. 1). Most tyrosine-based signals are of the form YXX⌽ (where ⌽ is an amino acid with a relatively large hydrophobic side chain and X can be any amino acid, although polar or positively charged residues are preferred (2)). The epidermal growth factor receptor, for example, contains the signal YRAL (3). Other tyrosine-based signals resemble the NPXY motif of the low density lipoprotein receptor (4). Dileucine signals contain adjacent leucine residues, one of which is sometimes replaced by isoleucine, valine, methionine, or alanine. A few endocytosis signals that differ from these patterns have been described, including an acidic patch in furin (5) and adjacent lysine residues in ERGIC-53 (6). Recently, we reported a tryptophan-based endocytosis signal in the neonatal Fc receptor, FcRn 1 (7).
FcRn internalized at coated pits at the apical plasma membrane of intestinal epithelial cells carries IgG from milk toward the blood circulation of suckling rats (8,9). The binding of IgG at the cell surface is allowed by the acidic pH of the intestinal lumen of the neonate (10 -12). This is atypical of FcRn. An example of its more widespread mechanism is that of IgG transport across the human placenta (13,14). IgG enters from maternal blood at near neutral pH and is thought to bind FcRn in acidified early endosomes (15)(16)(17)(18). IgG probably binds in early endosomes to FcRn expressed in other epithelia exposed to external fluid with a physiologically neutral bulk pH, including those of the adult intestine (19,20), liver (21), kidney (22), mammary gland (23)(24)(25)(26)(27), and lung (28). The functions of FcRn in these tissues remain to be established definitively. The function in capillary endothelium (29) is better understood. Here, FcRn protects from degradation IgG taken up from blood into early endosomes (30 -34). FcRn binds IgG at the surfaces of capillary endothelial cells, renal proximal tubule epithelial cells, glomerular epithelial cells, and trophoblast cells incubated in acidic media (22,(35)(36)(37). The presence of FcRn at the plasma membrane suggests that the receptor is delivered to early endosomes by endocytosis even in cells in which it binds IgG intracellularly (although an additional direct route from the trans Golgi network cannot be excluded). Thus, endocytosis signals are likely important for FcRn function in all cells in which it is expressed.
Both dileucine-and tyrosine-based endocytosis signals are recognized by AP-2 adaptor proteins at the plasma membrane (43,44). These tetrameric complexes link the membrane proteins containing the signals to the clathrin coats of pits (for review, see Ref. 45). AP-2 binds dileucine and tyrosine signals in different ways. Dileucine signals have been reported to bind both large subunits (␤ (46), ␥ or ␦ (47)), (48 -50), and subunits (47) of AP complexes. The YXX⌽ motif interacts with the medium subunit, 2 (51), and this is the best understood interaction of AP-2. Crystal structures of 2 with peptides containing YXX⌽ motifs show the two critical side chains accommodated by pockets in the adaptor protein (52). The tyrosine binding pocket includes Asp-176 and Trp-421 of 2 (52), and alanine substitution in these positions prevents YXX⌽ binding (53). YXX⌽ binding is also prevented by competition with the tyrosine structural analog tyrphostin A23 (54,55).
In this paper, we address the question of how the tryptophan-based endocytosis signal of rat FcRn, WLSL, is recognized. First we asked whether it interacted with 2. Then we asked whether it was recognized by 2 in a manner similar to YXX⌽. Additionally, in light of recent publications illustrating an interaction between tryptophan-containing signals and the ␣-adaptin ear domain (56 -58), we also explored the possibility of an interaction between FcRn and ␣-adaptin. Finally, we investigated the requirement for the tyrosine binding pocket of 2 in vivo by analyzing the effect of tyrphostin A23 on endocytosis. Our results showed that 2 recognized the WLSL and YXX⌽ endocytosis signals in analogous fashions. This revealed an unexpected versatility in the recognition of signals by 2.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture-Rat inner medullary collecting duct (IMCD) cells and their transfected derivatives were cultured as described previously (59). The lines expressing wild-type rat FcRn and mutants L322A/L323A and W311A/L322A/L323A have been described before (7).
Expression of GST Fusion Proteins-GST fusion proteins were expressed in Escherichia coli BL21 (DE3) RIL. Half-liter cultures were grown at 30°C to an A 600 of 0.8. Protein expression was induced with 1 mM isopropyl-␤-D-thiogalactopyranoside for 3 h at 30°C. Cells were harvested and frozen in liquid nitrogen. Thawed cells were resuspended in ice-cold phosphate-buffered saline (PBS), pH 7.4, containing lysozyme, 1 mM PMSF, and both mammalian and bacterial protease inhibitor cocktails (Sigma). Suspensions were sonicated on ice twice for 30 s at 50% duty cycle using a Fisher Scientific Sonic Dismembrator 550 with a microtip probe. Triton X-100 was added to a final concentration of 1%, and the lysates were cleared by centrifugation for 10 min at 12,000 rpm and 4°C in a Sorvall SS34 rotor (Newtown, CT). Each supernatant was incubated for 2 h at 4°C with 20 l of a 75% slurry of glutathione-Sepharose 4B (Amersham Biosciences) that had been washed with PBS, pH 7.4. The beads were washed three times with ice-cold PBS containing 1 mM PMSF. Fusion proteins were eluted by two incubations at room temperature with 100 l of 10 mM reduced glutathione, 50 mM Tris⅐HCl, pH 8. Pooled eluates were dialyzed at 4°C against PBS, pH 7.4. The protein concentrations were determined by A 280 using their calculated extinction coefficients. Glycerol was added to each to a concentration of 10%. Protein aliquots were frozen in liquid nitrogen and stored at Ϫ80°C.
GST-stonin was purified as described above, but with the following modifications. Protein expression was induced with 0.5 mM isopropyl-␤-D-thiogalactopyranoside for 4 h at 30°C. The thawed cells were re-suspended in ice-cold 3 M urea containing lysozyme, 1 mM PMSF, 2.5 g/ml leupeptin, and 0.7 g/ml pepstatin. After sonication, no Triton X-100 was added before the solution was centrifuged. The supernatant was dialyzed overnight at 4°C against PBS containing 0.5 mM PMSF, 2 mM EDTA, and 1 mM dithiothreitol. This solution was cleared by centrifugation for 20 min at 12,000 rpm at 4°C in a Sorvall SS34 rotor. The supernatant was incubated for 2 h at 4°C with 500 l of a 75% slurry of glutathione-Sepharose 4B that had been washed with PBS, pH 7.4. The beads were washed three times with ice-cold PBS containing 1 mM PMSF, 2 mM EDTA, and 1 mM dithiothreitol. Fusion proteins were eluted at room temperature with 1 ml of 20 mM reduced glutathione, 50 mM Hepes, pH 7.4, 150 mM NaCl, and 20 mM NaOH. Pooled eluates were dialyzed at 4°C against PBS, pH 7.4, containing 1 mM dithiothreitol.
Expression of His-tagged Fusion Proteins-His-tagged 2 was expressed in E. coli BL21 (DE3) RIL. Half-liter cultures were grown at 30°C to an A 600 of 0.8. Protein expression was induced by 0.5 mM isopropyl-␤-D-thiogalactopyranoside for 3 h at 30°C. Cell pellets were frozen in liquid nitrogen. Thawed cells were resuspended in ice-cold phosphate buffer, containing 1.14 M NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 ⅐7H 2 O, 1.4 mM KH 2 PO 4 , lysozyme, 1 mM PMSF, 2.5 g/ml leupeptin, and 0.7 g/ml pepstatin, pH 7.5. Suspensions were sonicated as above. CHAPS was added to a concentration of 2%, and the lysates were cleared as above. PBS was added to the supernatant to dilute the NaCl to 300 mM. Imidazole was added to a concentration of 10 mM, and each supernatant was incubated for 2 h at 4°C with 500 l of a 50% slurry of TALON Metal Affinity Resin (BD Biosciences Clontech, Palo Alto, CA) and 500 l of a 50% slurry of Sepharose CL-4B beads (Amersham Biosciences), prewashed in PBS, pH 7.4. The beads were washed three times with ice-cold phosphate buffer, containing 440 mM NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 ⅐H 2 O, 1.4 mM KH 2 PO 4 , 10 mM imidazole, 1 mM PMSF, 2.5 g/ml leupeptin, and 0.7 g/ml pepstatin, pH 7.4. Purified His-tagged fusion proteins were eluted in ice-cold phosphate buffer, containing 440 mM NaCl, 2.7 mM KCl, 4.3 mM Na 2 HPO 4 ⅐7H 2 O, 1.4 mM KH 2 PO 4 , 300 mM imidazole, 1 mM PMSF, 2.5 g/ml leupeptin, and 0.7 g/ml pepstatin, pH 7.4, for 1 h at 4°C. The eluate was dialyzed at 4°C against ice-cold Hepes buffer, containing 10 mM Hepes, 500 mM NaCl, and 10 mM ␤-mercaptoethanol, pH 7.4. Protein concentrations were determined, and proteins were frozen with glycerol and stored as described above.
Peptides-Peptides representing amino acids 302-315 of rat FcRn and of FcRn mutant W311A, and amino acids 989 -1002 of human epidermal growth factor receptor, each with an additional N-terminal cysteine (respectively, CRMRSGLPAPWLSLS, CRMRSGLPAPALSLS, and CLPSPTDSNFYRALM), were made by SynPep (Dublin, CA). Tyrphostins (Calbiochem) were stored as 350 mM (A23) or 8 mM (A51) stock solutions in Me 2 SO at Ϫ20°C and added to solutions just before use.
GST Pull-down Experiments-For each pull-down, 20 g of GST fusion protein was precleared by incubating in 500 l of PBS, 1 mM PMSF, pH 7.4, for 10 min at 4°C, then microcentrifuging for 5 min at 2,500 rpm. The supernatant was added to 20 l of glutathione-Sepharose (prewashed twice with PBS, pH 7.4) and incubated for 2 h at 4°C. The suspension was then microcentrifuged for 5 min at 2,500 rpm. The supernatant was removed, and the beads and bound proteins were kept on ice. For each pull-down, 10 g of His-tagged 2 or ␣-adaptin was precleared by incubating for 10 min at 4°C in 500 l of binding buffer (20 mM Hepes, 200 mM NaCl, 2 mM MgCl 2 , 0.5% Triton X-100, 1 mM PMSF, pH 8), then microcentrifuging for 5 min at 14,000 rpm. The supernatant was collected. For peptide competition experiments, 0.2 mM peptide was then incubated with the 2 supernatant for 15 min at room temperature. For tyrphostin competition experiments, 1:100 Me 2 SO, 3.5 mM A23, or 80 M A51 was then incubated with the 2 supernatant for 15 min at room temperature. The His-tagged fusion protein, precleared (and preincubated with peptide or tyrphostin as appropriate), was added to each tube containing GST proteins bound to glutathione-Sepharose. The beads were resuspended and then incubated for 3 h at 4°C. (All competition experiments were only incubated for 45 min.) They were washed three times for 5 min with binding buffer. After the final wash, the last of the buffer was removed with a fine, flat-ended, pipette tip. Proteins were eluted from the beads in 20 l of 2ϫ Laemmli sample buffer containing 2-mercaptoethanol (62) by heating for 5 min at 100°C. Proteins were resolved on 4 -20% acrylamide Novex Tris-glycine gels (Invitrogen). The gels were either stained with Coomassie Blue or probed by Western blotting with anti-polyhistidine tag antibodies.
Protein Iodination-The Fc fragment of human IgG (Jackson Immunoresearch, West Grove, PA) was labeled to a specific activity of ϳ0.5 Ci mol Ϫ1 with Na 125 I (PerkinElmer Life Sciences) using IODO-GEN (Pierce).
Fc Endocytosis Assay-Cells were seeded into 6-well plates and assayed when about 70% confluent. Cells were incubated with prewarmed DMEM, 1 mM KI, 1.5% fish gelatin, 25 mM Hepes (DMEM-KIGH), pH 8, for 1 h at 37°C. The cells were washed once with 1 ml of prewarmed DMEM-KIGH, pH 6 (except for the 0 time point, which was washed with cold DMEM-KIGH). One ml of prewarmed DMEM-KIGH, pH 6, containing 125 I-Fc (1 ϫ 10 6 cpm/per well) and either 1:1,000 Me 2 SO (as a vehicle control), 350 M A23 (from a 350 mM stock in Me 2 SO), or 8 M A51 (from an 8 mM stock in Me 2 SO) was added to the cells (except for the 0 time point to which cold DMEM-KIGH was added). In addition, 600 g/ml cold competitor IgG was added to half of the cells. The cells were incubated at 37°C for 0, 2, 4, or 8 min. Endocytosis was stopped by cooling the cells on ice and washing them four times with ice-cold, DMEM-KIGH, pH 6, containing tyrphostin or Me 2 SO. The cells were then incubated on ice for 45 min with 1 ml of DMEM-KIGH, pH 8. This medium was collected and counted (cpm surface1 ) in a CliniGamma 1272 ␥ counter (LKB Wallac, Piscataway, NJ). Another ml of DMEM-KIGH, pH 8, was added, collected, and counted (cpm surface2 ). The cells were lysed with 1 ml of 0.1 M NaOH, and the lysates were counted (cpm internal ). The percent of Fc internalized (internal/surface) was calculated with the Equation 1.
Specific binding and endocytosis were calculated by subtracting the percent of Fc internalized in the presence of competing IgG from the percent internalized in its absence. As a background correction, the percent of Fc internalized at time 0 was subtracted from the percent internalized at each time point. Fluid Phase Endocytosis Assay-Cells were seeded into 6-well plates and assayed when about 70% confluent. Cells were incubated with prewarmed DMEM-KIGH, pH 8, for 1 h at 37°C. The cells were washed once with 1 ml of prewarmed DMEM-KIGH, pH 6. Then, 1 ml of prewarmed DMEM-KIGH, pH 6, containing 30 M dextran-Alexa 488 (Molecular Probes, Eugene, OR) and either 350 M A23 (from a 350 mM stock in Me 2 SO) or 1 l of Me 2 SO was added to the cells. The cells were incubated at 37°C for 8 min. Endocytosis was stopped by cooling the cells on ice and washing them four times with ice-cold DMEM-KIGH, pH 6. The cells were then incubated on ice for 45 min with 1 ml of DMEM-KIGH, pH 8. This medium was removed, and the cells were washed a second time with 1 ml of DMEM-KIGH, pH 8. The cells were washed once with 1 ml of PBS and were subsequently lysed in 0.5 ml of PBS containing 1% Triton X-100. The amount of dextran-Alexa Fluor 488 taken up by the cells was measured with a fluorometer (Photon Technology International, Lawrenceville, NJ). These amounts were converted into apparent volumes internalized using a standard curve of the fluorescence of known amounts of dextran-Alexa 488 in PBS and Triton X-100.
CD4 Endocytosis Assay-IMCD cells were grown on coverslips and transfected with the plasmid encoding CD4 when about 60% confluent, using Lipofectamine and Plus Reagent in Opti-MEM (Invitrogen). The transfection reagents were left on the cells ϳ24 h. Anti-human CD4 (clone Q4120, Sigma), diluted 1:100 in Opti-MEM, was incubated with the cells at 37°C for 30 min with either 1:1,000 Me 2 SO or 350 M A23 (from a 350 mM stock in Me 2 SO). As a negative control for endocytosis, cells were incubated at 4°C without Me 2 SO or A23. After incubation, the cells were cooled on ice, washed with ice-cold PBS, and fixed for 30 min with 4% paraformaldehyde. Fixed cells were washed with high salt phosphate buffer (20 mM NaP i , pH 7.4, 500 mM NaCl) containing 10% goat serum. The surface CD4 was decorated with Alexa 488 goat antimouse IgG (Molecular Probes) diluted 1:200 in high salt phosphate buffer containing 10% goat serum. The cells were washed again with high salt phosphate buffer containing 10% goat serum and then blocked for 1 h at room temperature with unconjugated goat anti-mouse IgG (Jackson Immunoresearch) diluted 1:5 in high salt phosphate buffer containing 10% goat serum. After another wash with high salt phosphate buffer, the cells were permeabilized and blocked for 30 min at room temperature with high salt phosphate buffer containing 0.3% Triton X-100 and 30% goat serum. Internalized CD4 was decorated with Alexa 594 goat anti-mouse IgG (Molecular Probes) diluted 1:150 in permeabilization/blocking solution. Cells were washed with high salt phosphate buffer containing 0.3% Triton X-100, followed by high salt phosphate buffer alone, followed by 20 mM NaP i , pH 7.4. Coverslips were mounted using Vectashield (Vector Laboratories, Burlingame, CA). All images were visualized with a Leica TCS SP2 Confocal Scanning Microscope and analyzed with Leica Confocal Imaging Software (Leica MicroSystems, Exton, PA).

RESULTS
The Cytoplasmic Domain of FcRn Interacts with the 2 Subunit of AP-2-We expressed the cytoplasmic domain of FcRn as an N-terminal GST fusion protein and asked whether it could interact with purified His-tagged 2 in a pull-down assay. In Western blots of the proteins pulled down by GST-FcRn cyt , we detected substantially more His-tagged 2 than in controls using GST alone (Fig. 1, upper panel, second and third lanes). This indicates that the cytoplasmic domain of FcRn does indeed bind the 2 subunit of AP-2.
The cytoplasmic tail of FcRn contains an endocytosis signal dependent on Trp-311 in the sequence WLSL (7). We compared the binding of the cytoplasmic domain of FcRn in which Trp-311 was replaced by alanine with the wild-type tail. The W311A substitution reduced to background levels the amount of 2 pulled down, as detected by Western blot and Coomassie Blue staining (Fig. 1, upper and lower panels, fourth lane). Coomassie staining also confirmed that approximately equal amounts of GST or GST fusion were pulled down in all of the assays. (This control was done in all of the pull-down experiments but is shown only for Fig. 1.) The Tryptophan-based Sequence from FcRn, APWLSL, Is Sufficient to Bind 2-To study further the importance of the sequence WLSL in the interaction between FcRn and AP-2, we tested the ability of the trimer (APWLSL) 3 , with GST fused to the N terminus, to bind 2. GST-(APWLSL) 3 pulled down 2, as detected by Western blot (Fig. 2, third lane). In contrast, substitution of alanine for tryptophan within the trimer reduced the interaction with 2 to background (Fig. 2, fourth  lane). These results show that the APWLSL sequence from FcRn is sufficient to bind 2 and confirm the requirement of tryptophan for the interaction.
FcRn and the Tyrosine-based Motif Compete for Their Interaction to 2-To ask whether the tryptophan-based signal from FcRn bound the same site in 2 as a YXX⌽ endocytosis signal, we tested the ability of peptides representing each sequence to compete with FcRn for interaction with 2. The presence of a peptide containing the mutated FcRn signal, ALSL, did not interfere with the ability of GST-FcRn cyt to pull down 2, as detected by Western blot (Fig. 3, fourth lane). A peptide containing the intact WLSL signal inhibited the binding of GST-FcRn cyt to 2 (Fig. 3, fifth lane). Significantly, a peptide containing YRAL, a YXX⌽ signal from the epidermal growth factor receptor, also inhibited the interaction of GST-FcRn cyt with 2 ( Fig. 3, sixth lane). These results suggest that YXX⌽ and the tryptophan-based sequence in the cytoplasmic domain of FcRn bind the same site within 2.
FcRn Binds 2 at the YXX⌽ Binding Pocket-To ask directly whether FcRn interacts with the YXX⌽ binding pocket of 2, we tested the ability of GST-FcRn cyt and GST-(APWLSL) 3 to bind the mutant 2 W421A. The W421A substitution has been shown to prevent 2 from binding tyrosine-based motifs (53). Neither GST-FcRn cyt nor GST-(APWLSL) 3 was able to pull down His-tagged 2 W421A, as detected by Western blot (Fig.  4, fourth and sixth lanes). These results imply that the YXX⌽ binding pocket of 2 binds the WLSL signal in FcRn.
FcRn Does Not Interact with the ␣-Adaptin Ear Domain-WXXF motifs in stonin 2 mediate its interaction with both 2 and the ␣-adaptin ear domain (58). We compared the abilities of GST-FcRn cyt and GST-stonin 2 to interact with His-tagged ␣-adaptin ear domain and 2. GST-stonin 2 pulled down both ␣-adaptin and 2 (Fig. 5, fourth and eighth lanes). GST-FcRn cyt pulled down 2 but was unable to interact with ␣-adaptin (Fig.  5, third and seventh lanes). These results indicate that the WLSL sequence from FcRn is functionally distinct from the WXXF motif.
Tyrphostin A23 Inhibits the Interaction between FcRn and 2 and the Endocytosis of FcRn-Tyrphostins A23 and A51 are tyrosine analogs. We asked whether A23 or A51 affected the interaction between FcRn and 2. GST-FcRn cyt pulled down 2 in the presence of Me 2 SO, the solvent used for both tyrphostins.
It also pulled down 2 in the presence of tyrphostin A51 (Fig.  6, fifth and seventh lanes). In contrast, tyrphostin A23 markedly reduced the interaction between FcRn and 2 (Fig. 6, sixth  lane).
This inhibition enabled us to investigate the importance of the interaction between WLSL and 2 for the endocytosis of FcRn. Me 2 SO and tyrphostin A51 did not affect the uptake of 125 I-Fc by cell lines expressing FcRn or FcRn mutants. Tyrphostin A23 strongly inhibited endocytosis of FcRn L322A/ L323A, which depends solely on the WLSL signal (7) (Fig. 7A). A23 also inhibited endocytosis of wild-type FcRn (Fig. 7B) and FcRn W311A, which has only a dileucine endocytosis motif (Fig. 7C).
Tyrphostin A23 Does Not Inhibit Fluid Phase Uptake or Endocytosis of CD4 -To determine whether A23 reduced endocytosis generally, we tested its effect on fluid phase uptake, using Alexa 488-conjugated dextran as a tracer. After 8 min under the same conditions as the Fc endocytosis assay, there was no significant difference between fluid taken up in the presence of A23 (30 Ϯ 5 nl) or a Me 2 SO control (32 Ϯ 3 nl; mean Ϯ S.E., n ϭ 3, representative of three experiments).
Additionally, we tested the effect of A23 on the endocytosis of CD4, which depends upon a dileucine motif (63). We expressed CD4 in IMCD cells and let it bind and take up a monoclonal antibody at 37°C. We visualized surface CD4 using an Alexa 488-conjugated secondary antibody (green staining in Fig. 8, A-C), and internal CD4 with an Alexa 594 secondary after permeabilization (red staining). Fig. 8D shows that more than 80% of cells with surface CD4 internalized anti-CD4 at 37°C in the presence of A23 or a Me 2 SO control. When endocytosis was FIG. 2. The sequence APWLSL, which contains Trp-311 in FcRn, is sufficient to bind 2. His-tagged 2 was incubated with glutathione-Sepharose coupled to GST alone or to GST fused with a trimer of either the peptide APWLSL or APALSL. The 2 pulled down was detected with anti-His 6 . The data shown are from one experiment representative of at least five.

FIG. 3. A peptide containing the YXX⌽ motif blocks the interaction between FcRn and 2.
His-tagged 2 was incubated with or without peptides containing ALSL, WLSL, or YRAL. It was then put into a pull-down assay with glutathione-Sepharose coupled to GST or GST fused with the cytoplasmic domain of FcRn. The 2 pulled down was detected with anti-His 6 antibodies. The data shown are from one experiment representative of at least three.
FIG. 4. The interaction between FcRn and 2 depends on the YXX⌽ binding pocket. His-tagged wild-type (WT) 2 or its binding pocket mutant W421A was put into a pull-down assay with glutathione-Sepharose coupled to GST or to a GST fusion protein as indicated. The 2 pulled down was detected with anti-His 6 antibodies. The data shown are from one experiment representative of at least three.

FIG. 5. The cytoplasmic domain of FcRn does not interact with the ␣-adaptin ear domain.
His-tagged ␣-adaptin, or His-tagged 2, was put into a pull-down assay with glutathione-Sepharose coupled to GST alone, GST fused with FcRn cyt , or GST fused with the N-terminal domain of stonin 2. The first and fifth lanes contain, respectively, 1/10 of the amounts of ␣-adaptin and 2 that were put into each pull-down. One quarter of the protein from each pull-down was run on each of two gels. Anti-His 6 antibodies were used to detect ␣-adaptin and 2 on a Western blot. The data shown are from one experiment representative of at least three.
inhibited by keeping the cells on ice, fewer than 13% internalized the monoclonal antibody.

DISCUSSION
The tryptophan-based endocytosis signal in the sequence WLSL in FcRn is similar to the well characterized endocytosis signal YXX⌽ (7). Substitution of tryptophan for tyrosine in the YXX⌽ motifs of different receptors is reported either to maintain (64) or impair function as an endocytosis signal (65). This left it uncertain whether WLSL was likely to function in the same way as YXX⌽. The YXX⌽ motif is known to interact directly with the subunit of AP-2 (1). In the present study, we asked whether the tryptophan-based endocytosis signal functions like the tyrosine-based motif by analyzing its interaction with 2.
Using GST pull-down experiments, we found that the tryptophan-based endocytosis signal in FcRn is recognized by the subunit of AP-2. These results reveal a functional similarity between the tryptophan-based signal and YXX⌽ motifs. Considering this congruence, we predicted that our tryptophanbased sequence interacted within the YXX⌽ binding pocket of 2. We tested this hypothesis in three different experiments. First, we showed that the tyrosine-based motif competed with GST-FcRn cyt for its interaction with 2, implying that the tryptophan interacts at the same site on 2. We then showed that 2 W421A, mutated in its YXX⌽ binding pocket, could not interact with either GST-FcRn cyt or GST-(APWLSL) 3 . Third, we examined the effect of two tyrphostins on the interaction of the tryptophan signal with 2. Tyrphostins are used as tyrosine kinase inhibitors (66). Tyrphostin A23 also inhibits the interaction between 2 and a YXX⌽ motif, but A51 does not (54). In our assay, A23, but not A51, reduced the interaction between GST-FcRn cyt and His-tagged 2. These data showed that the tryptophan-based signal interacts with the YXX⌽ binding region of 2.
The crystal structure of a YXX⌽ peptide complexed with the binding domain of 2 illustrates the perfect peg-in-socket fit of both the tyrosine and the bulky hydrophobic ⌽ residue, each in its own pocket (52). Additionally, Owen and Evans (52) described a network of hydrogen bonds around the tyrosine residue and concluded that this explained the inability of phenylalanine to replace tyrosine successfully in most endocytosis signals. However, they did not comment on tryptophan. When we modeled the interaction of the tryptophanbased signal within the YXX⌽ binding pocket, only one hydrogen bond was lost compared with the tyrosine motif (Fig. 9). Significantly, the hydrogen bond with Asp-176 of 2 remained. In addition, the hydrophobic interaction between Trp-421 of 2 and tyrosine also existed between Trp-421 and tryptophan. This interaction is critical because mutating Trp-421 disrupts the interaction with YXX⌽ (53) and with APWLSL (above).
Leucine is the preferred ⌽ amino acid for interaction between YXX⌽ and 2 (2). The presence of leucine in the Wϩ3 position within WLSL might compensate for a lower binding energy of tryptophan compared with tyrosine, although we have found that isoleucine replacement maintains function (7), and phenylalanine is found in this position in bovine (24), pig (27) (GenBank accession no. AY204219), and sheep (40) FcRn. We note that the YTRF endocytosis signal in the transferrin receptor contains the preferred residue, arginine, in the Yϩ2 position (2) and tolerates replacement of tyrosine with tryptophan (64). In contrast, none of the residues Yϩ1, 2, or 3 is optimal in the cationic independent mannose 6-phosphate receptor YSKV signal, which is inactivated by tryptophan sub-FIG. 6. The tyrosine analog tyrphostin A23 inhibits the interaction between FcRn and 2. His-tagged 2 was incubated with or without tyrphostin A23 or tyrphostin A51. It was then put into a pull-down assay with glutathione-Sepharose coupled to GST or GST fused with FcRn cyt . Anti-His 6 antibodies were used to detect 2 on a Western blot. The data shown are from one experiment representative of at least three. DMSO, Me 2 SO. stitution (65). The presence of tryptophan in an endocytosis signal may thus place more constraints on the other three residues than does tyrosine.
Owen and Evans (52) also described the dimerization of 2 in the crystal structure, in which the individual YXX⌽ binding pockets and tyrosine motifs are placed in close proximity to each other. This structure suggests an enhanced strength and selectivity of binding to dimeric 2 by dimerized receptors. FcRn dimers are found in cells (67,68) and are required for high affinity IgG binding (69). The proximity of the C termini in a crystal structure of FcRn extracellular region dimers suggests that the cytoplasmic domains may be able to contact one another (70,71). Pairing of the tryptophan signals of two receptors might allow a strong, selective interaction with 2, as proposed by Owen and Evans.
Recent publications report an interaction between endocytic proteins containing the sequence WXXF and the ␣-adaptin ear domain (56 -58). The WXXF motif itself mediates this interaction. Because of the similarity between this motif and WLSL in FcRn, we explored the ␣-adaptin ear domain as a possible binding partner for FcRn. Whereas GST-stonin pulled down both ␣-adaptin and 2, GST-FcRn cyt was unable to interact with ␣-adaptin. In addition to binding ␣-adaptin, the WXXF sequence in stonin 2 interacts with the YXX⌽ binding pocket of 2 (58). Stonin 2 is a cytoplasmic protein. It has not been shown to mediate endocytosis but is suggested to regulate the interaction of 2 with tyrosine-based endocytosis signals (58). Together with the contrasting abilities to bind ␣-adaptin ear domain, this suggests that the WLSL sequence is functionally distinct from WXXF.
To determine whether the interaction between WLSL and 2 revealed by our pull-down assays was relevant to FcRn function in vivo, we studied the effects of tyrphostins on the endocytosis of Fc by cell lines expressing FcRn. Tyrphostin A23 is reported to inhibit the endocytosis of transferrin receptor (55). We found that A23 profoundly inhibited endocytosis of an FcRn mutant that lacks the dileucine motif and depends only on the tryptophan-based signal for uptake. Treatment with tyrphostin A51 reduced endocytosis only slightly. A51 does not inhibit endocytosis of transferrin receptor (55). Because we used both tyrphostins at concentrations 10 times higher than their IC 50 values for a tyrosine kinase (55), the greater reduction of endocytosis by A23 is unlikely to reflect inhibition of protein phosphorylation. However, the slight inhibition seen with A51 (possibly corresponding to a fraction of the A23 inhibition) might result from an effect on the cells caused by tyrosine kinase inhibition.
Surprisingly, tyrphostin A23 also inhibited endocytosis of wild-type FcRn and FcRn W311A. A23 did not inhibit fluid phase endocytosis or the dileucine-dependent uptake of CD4. We had previously found the dileucine motif sufficient for endocytosis (7). This therefore implies that dileucine-dependent endocytosis of FcRn depends upon the YXX⌽ binding pocket of  (52)). The dotted lines represent hydrogen bonds made by the tryptophan or tyrosine residue of the bound peptide.

2.
Although much is known about the interactions between tyrosine motifs and adaptor complexes, much of what is known about the interactions of the dileucine motif is contradictory. It has been reported that the dileucine motif interacts with the ␤ subunit of AP-1 (72). However, it has also been reported that it interacts with the subunits of AP-1, AP-2 (48,49), and AP-3 (50). Most recently, dileucine motifs were shown to bind complexes of AP-1 and ␥ subunits and of AP-3 and ␦ subunits (47). The relationship between tyrosine-and dileucine-based signals is also unclear. Overexpression assays, performed to determine whether tyrosine and dileucine motifs compete for saturable endocytic components, have provided inconsistent results. Some experiments have shown that although tyrosine and dileucine motifs can compete among themselves, they cannot compete with each other (73). However, it has also been documented that the overexpression of proteins containing either a tyrosine or dileucine motif can inhibit the internalization of the YXX⌽-containing transferrin receptor (74). In showing an interdependence of the tryptophan-and dileucine-based signals, our result is more consistent with the latter observation. As we describe below, this interdependence may arise from the particular contexts of these signals in FcRn.
The tryptophan sequence and dileucine motif lie close together in FcRn. Only three amino acids separate WLSL from DXXXLL. Therefore, it is possible that these two signals interact as a rigid unit with their binding sites in AP-2. For this model, the dileucine motif might be assumed to bind either 2 or ␤2 because these subunits are closely associated (75). Indeed, part of the ␤ subunit obstructs the YXX⌽ binding site of 2, and a conformational change must occur for the YXX⌽ motif to enter (75). However, our observation that FcRn cyt W311A, which contains the dileucine motif, did not pull down His-tagged 2 suggests that the interaction is either with ␤2 or with the N-terminal domain of 2, the portion that was not included in our His-tagged 2 construct. We speculate that the occupancy of the tyrosine/tryptophan binding pocket of 2 by tyrphostin A23 prevents FcRn (or FcRn W311A) approaching closely enough for the dileucine motif to bind AP-2.
Signals of the YXX⌽ type are anchored to 2 by the binding of the Y and ⌽ residues within pockets (52). The range of sequences that can bind in this way is clearly broad. Our understanding of this range comes largely from observations of the variation among endocytosis signals in proteins, and upon yeast two-hybrid studies (2,51). Such work has shown that the residues the ⌽ binding pocket can accommodate include leucine, phenylalanine, methionine, and isoleucine. Because tryptophan-based signals are rare and because the yeast twohybrid libraries used to identify binding sequences were built around tyrosine residues, the versatility of the Y binding pocket has not been emphasized previously. The finding that a tryptophan-based sequence in FcRn can bind the same site as YXX⌽ motifs reveals a still broader range of endocytosis signals capable of interacting with 2.