Multiple C-terminal Motifs of the 46-kDa Mannose 6-Phosphate Receptor Tail Contribute to Efficient Binding of Medium Chains of AP-2 and AP-3*

The interaction of adaptor protein (AP) complexes with signal structures in the cytoplasmic domains of membrane proteins is required for intracellular sorting. Tyrosine- or dileucine-based motifs have been reported to bind to medium chain subunits ( m ) of AP-1, AP-2, or AP-3. In the present study, we have examined the interaction of the entire 67-amino acid cytoplasmic domain of the 46-kDa mannose 6-phosphate receptor (MPR46-CT) containing tyrosine- as well as dileucine-based motifs with m 2 and m 3A chains using the yeast two-hybrid system. Both m 2 and m 3A bind specifically to the MPR46-CT. In contrast, m 3A fails to bind to the cytoplasmic domain of the 300-kDa mannose 6-phosphate receptor. Mutational analysis of the MPR46-CT revealed that the ty-rosine-based motif and distal sequences rich in acidic amino acid residues are sufficient for effective binding to m 2. However, the dileucine motif was found to be one part of a consecutive complex C-terminal structure comprising tyrosine and dileucine motifs as well as clusters of acidic residues necessary for efficient binding of m 3A. Alanine substitution of 2 or 4 acidic amino acid residues of this cluster reduces the binding to m 3A much more than to m 2. The data suggest that the MPR46 is capable of interacting with different AP complexes using multiple partially overlapping sorting signals, which might depend on posttranslational modifications or subcellular localization of the receptor. The with primers: MPR46-CT- E55A/E56A/E58A/E59A, E58-F (5 9 -GGGGAGGAGTCAGCAGCAAGG-GATGACCAT-3 9 ) and E58-R (5 9 -ATGGTCATCCCTTGCTGCTGACTC- CTCCCC-3 9 ); MPR46-CT-E55A/E56A/E58A/E59A/D61A/D62A, D61-F (5 9 -TCAGCAGCAAGGGCTGCCCATTTATTACCA-3 9 ) and D61-R (5 9 -T- GGTAATAAATGGGCAGCCCTTGCTGCTGA-3 9 ). The mutation MPR46-CT-Y45A/V48A was introduced using the wild type MPR46-CT cDNA as template with primers Y45-F (5 9 -GTGCCTGCAGCAGCTCG-TGGTGCGGGGGATGACCAG-3 9 ) and Y45-R (5 9 -CTGGTCATCCCCC- GCACCACGAGCTGCTGCAGGCAC-3 9 ). For the MPR46-CT-StopH63-Y45A/V48A mutation, the MPR46-CT-StopH63 cDNA was used as template with primers Y45-F and Y45-R. All mutations were verified by sequencing the final products. The full coding regions of m 2 and m 3A were amplified by PCR using the following oligonucleotide primers: m and m m and m 3A-R (5 92 CGGGATC- CCGTCATGTCCTCACTTGGAA-3 The m 2 and m 3A PCR products were digested with Bgl II and Eco RI/ Bam HI, respectively, and cloned in frame

The 46-kDa mannose 6-phosphate receptor (MPR46) 1 mediates the transport of newly synthesized soluble lysosomal enzymes from the trans-Golgi network (TGN) to the endosomalprelysosomal compartment. Following the pH-induced dissociation of the complexes, the MPR either return to the TGN or undergo cycling via the plasma membrane (1). Targeting information contained within the 67-residue cytoplasmic domain of this type I integral membrane glycoprotein is respon-sible for the directed intracellular transport of MPR46 between TGN, endosomes, and the plasma membrane as well as for retention from lysosomal delivery. Mutational analyses have identified several signals including a tyrosine (Tyr 45 -Arg-Gly-Val 48 ) and dileucine-based motif (Leu 64 -Leu 65 ), and two aromatic residues (Phe 13 and Phe 18 ) required for efficient sorting in the TGN or rapid internalization at the plasma membrane (2)(3)(4). Additionally, the di-aromatic Phe 18 -Trp 19 motif and the phosphorylation of Ser 57 have been reported to be important for endosomal sorting (5,6). Finally, Cys 34 has been shown to be a reversible palmitoylated residue that might anchor the tail to the lipid layer and form a cytoplasmic loop (7). Both the intact palmitoylation site and the correct length of the loop are critical to prevent targeting of MPR46 to degrading compartments (7,8).
The signal structures in the cytoplasmic tail of the MPR46 form selective recognition sites for cytosolic proteins facilitating the incorporation of the receptors into transport vesicles. The best studied class of these cytosolic proteins comprises the family of adaptor proteins (AP; Ref. 9). AP-1 and AP-2 are heterotetramers comprising two large ϳ100-kDa chains (␤1 and ␥-adaptin for AP-1; ␤2 and ␣-adaptin for AP-2), a medium subunit of ϳ50 kDa (1 and 2) and a small polypeptide of ϳ20 kDa (1 and 2). AP-1 is involved in protein sorting at the TGN, whereas AP-2 mediates endocytosis from the plasma membrane. Recently, heterotetrameric AP-3 and AP-4 complexes have been identified (10,11), which are proposed to be involved in sorting at the TGN and endosomal membranes. The AP-3 complex containing a 3A isoform is ubiquitously expressed, whereas 3B is a component of the neuronal-specific AP-3 variant. Biosensor analysis has demonstrated the presence of several distinct binding sites for AP-1 and AP-2 in the cytoplasmic tail of the MPR46, which do not depend on dileucine signals (12). Using the yeast two-hybrid system, 1, 2, and 3 chains have been demonstrated to interact with triple repeat sequences of tyrosine-containing sorting signals of several integral membrane proteins (13). However, the composition of residues surrounding the tyrosine residue and the position of the tyrosine motif within the tail relative to the membrane appear to be important in determining chain specificity (13,14). The subunit of the adaptor complexes responsible for recognition of the dileucine motifs is a matter of debate (13,15,16).
In the present study we have examined the interaction of the entire cytoplasmic tail of the MPR46 (MPR46-CT) and various tail mutants with 2 and 3A using the yeast two-hybrid system. The results have revealed that both 2 and 3A bind to the MPR46-CT. The interactions are specific and depend on the presence of a complex configuration of different signals comprising not only the tyrosine and dileucine motif but also acidic amino acid residues localized between these motifs.

MATERIALS AND METHODS
[ 35 S]Methionine and the prestained Rainbow protein marker were from Amersham Pharmacia Biotech Europe (Freiburg, Germany). Oligonucleotide primers for PCR were synthesized by NAPS (Göttingen, Germany). The following reagents were obtained commercially as indicated: restriction enzymes and T4 DNA ligase (New England Biolabs, Schwalbach, Germany); glutathione-agarose (Sigma); Pfu and Taq DNA polymerase and the Quick ® Change site-directed mutagenesis kit (Stratagene Cloning System, La Jolla, CA); TNT ® coupled reticulocyte lysate system (Promega Corp., Madison, WI).
Vectors-The Gal4-DNA binding domain vector pAS2 and the Gal4activation domain vector pGAD424 were purchased from CLONTECH Laboratories (Heidelberg, Germany). The bacterial expression vector pGEX-4T-1 came from Amersham Pharmacia Biotech Europe and pSPUTK was obtained from Stratagene.
Yeast and Bacterial Strains-Yeast strains for two-hybrid system (MATCHMAKER) were purchased from CLONTECH. The yeast strain HF7c was used for cotransformation of the Gal4 constructs and SFY526 was used for ␤-galactosidase assays. The Escherichia coli strains of DH5␣ and BL-21 were obtained from CLONTECH and Amersham Pharmacia Biotech Europe, respectively.
Two-hybrid Analysis-The yeast reporter strain HF7c was cotransformed with wild type and mutant pAS2-MPR46-CT or pAS2-MPR300-CT-constructs and the pGAD424-2 and -3A constructs, respectively, using a lithium acetate-base method. To examine specificity of the interaction, HF7c cells were cotransformed with the plasmids pAS2 (Gal4DBD alone) and pAS2 lamin and the pGAD424-2 and -3A constructs. The double transformants were grown on SD agar medium lacking Trp, Leu, and His (ϩ 5 mM 3-amino-1,2,4-triazole, Sigma) for 5 days at 30°C before positive colonies were picked, restreaked onto triple minus plates, and assayed for the lacZ phenotype.
Production and Purification of GST Fusion Proteins-GST fusion constructs were made by ligation of wild type MPR46-CT-cDNA into the EcoRI-and SalI-restriction sites of the vector pGEX-4T-1 (Amersham Pharmacia Biotech Europe). The MPR300-CT-cDNA (amino acids 1-164) was amplified by polymerase chain reaction using PfuTurbo DNA polymerase (Stratagene) with primers MPR300CT-F (5Ј-GGAAT-TCCGCAAGAAGAAGAGGAGGGAAACAGTG-3Ј) and MPR300CT-R (5Ј-GGAATTCCTCAGATGTGTAAGAGGTC-3Ј). The polymerase chain reaction product was ligated into the EcoRI site of pGEX-4T-1. For GST-MPR46-CT fusion protein expression, E. coli BL-21 cells were transformed with the pGEX-4T-1-MPR46-CT construct and grown to an A 600 of 0.6 followed by induction of GST fusion protein expression with isopropyl-1-thio-␤-D-galactoside at a final concentration of 0.1 mM. After 4 h of additional growth at 37°C, E. coli BL-21 cells were pelleted and lysed by sonification in 10 mM phosphate-buffered saline, pH 7.4 (PBS), containing 1 mg/ml lysozyme, 25% (w/v) sucrose, 1 mM EDTA, 1% Triton X-100, and 1 mM phenylmethylsulfonyl fluoride. After centrifugation for 10 min at 10,000 ϫ g, the supernatant was mixed with 1 ml (50% v/v) glutathione-agarose (Sigma), equilibrated in PBS, incubated for 1 h at 4°C, and washed three times with 20 bead volumes of PBS containing 1% Triton X-100 and 1 mM phenylmethylsulfonyl fluoride. The fusion protein was eluted with 50 mM Tris-HCl, pH 8.0, containing 15 mM reduced glutathione. Eluted proteins were dialyzed overnight against PBS and stored in aliquots at Ϫ70°C. For MPR300-CT fusion protein expression, E. coli BL21 cells were grown to an A 600 of 0.1 and induced with 0.2 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 30°C (19). Lysis of E. coli BL21 cells, affinity purification, and elution of MPR300-CT fusion proteins were carried out as described above.
In Vitro Translation of 2 and 3A-The 2 and 3A cDNA was subcloned into Bg/II restriction site of the pSPUTK vector. The pSP64luciferase construct was obtained from Promega. In vitro translation was carried out by using the TNT ® coupled reticulocyte lysate system in the presence of 40 Ci of [ 35 S]methionine in a final volume of 50 l according to the manufacturer's instructions. 35 S-Labeled, in vitro translated products were centrifuged at 100,000 ϫ g for 5 min to remove insoluble materials.

Analysis of the Interaction of Wild Type MPR46 Cytoplasmic
Tail with 2 and 3A Subunits-Reporter yeast cells were first transformed with a plasmid encoding a fusion protein between the wild type MPR46-CT or MPR300-CT and the Gal4 DNA binding domain (pAS2-MPR46/300-CT). The strain was subsequently transformed with a plasmid encoding the 2 or 3A adaptor subunit fused to the transcriptional activation domain of Gal4 (pGAD424-2 or -3A). After an incubation period of 5 days in the presence of 3-amino-1,2,4-triazole (5 mM), the cells expressing the MPR46-CT fusion protein with either 2 or 3A exhibited a strong growth in a medium lacking histidine, and positive galactosidase activity (Fig. 1). The MPR300-CT failed to interact both with 2 and 3A.
The MPR46-CT Binds Directly to the 2 and 3A Subunit in Vitro-To confirm the results obtained with the two-hybrid system, the interaction of in vitro translated [ 35 S]methioninelabeled 2 and 3A with MPR46-CT and MPR300-CT expressed and purified as GST fusion protein was examined. Densitometric evaluation of the coprecipitated 2 subunit and correction by unspecific binding to GST alone revealed that about 4.3% of total 2 was specifically recovered on glutathione-agarose beads ( Fig. 2A). About 17% of the total 3A was coprecipitated with the MPR46-CT fusion protein, whereas the in vitro translated [ 35 S]methionine labeled luciferase used as a control was not. In coprecipitation experiments with the MPR300-CT fusion protein, 4.2% of total 2 but almost no 3A (1%) was recovered (Fig. 2B).
MPR46-CT Signal Structures Required for 2 Interaction-To define the specificity and the site responsible for the interaction with 2 in more detail, several plasmids encoding mutant forms of MPR46-CT (Fig. 3) and control constructs were cotransformed with the pGAD424-2-plasmid. The Nterminal deleted MPR46-CT ⌬1-16 and the C-terminal truncated MPR46-CT-StopH63, which lacks the dileucine motif, caused growth on histidine-free medium and activation of the lacZ reporter gene similar to the wild type MPR46-CT ( Fig. 3; Table I). Further C-terminal truncation resulting in cytoplasmic domains still containing the tyrosine-based signal (MPR46-CT-StopV48) or lacking both the dileucine and tyrosine motives (MPR46-CT-StopA44) reduced the interaction to 25 and 4%, respectively, of wild type MPR46-CT. The replacement of Tyr 45 and Val 48 by alanine in the entire cytoplasmic tail (MPR46-CT-Y45A/V48A) decreased ␤-galactosidase activity to 75% of the wild type tail. When these two residues were replaced in addition to the deletion of the dileucine motif (MPR46-CT-StopH63-Y45A/V48A), the ␤-galactosidase activity was reduced to 60% of the wild type tail. The inability of cells cotransformed with Gal4-binding domain alone, lamin C, and 2, to grow in the absence of histidine (Fig. 3) or to activate ␤-galactosidase (Table I) confirmed the specificity in the interaction between MPR46-CT and 2. The lack of ␤-galactosidase activity in cells coexpressing the wild type MPR300-CT fusion protein and 2 corresponded to the failure to grow on histidine-free medium (Fig. 1).
Interaction of Mutant MPR46-CT with 3A-When the var-ious MPR46-CT mutants were coexpressed with 3A in SFY 526 cells, quantification of the interaction measured by lacZ reporter gene activity revealed that the deletion of the dileucine motif (MPR46-CT StopH63) reduced ␤-galactosidase activity by 42% (Fig. 4). The MPR46-CT StopV48 and StopA44 mutants as well as the alanine substitution (Y45A/V48A) in the entire or in the truncated receptor tail (StopH63-Y45A/V48A) interacted weakly with 3A (10 -27% of ␤-galactosidase activity of the wild type MPR46-CT). The N-terminal deleted mutant MPR46-CT ⌬1-16 activated ␤-galactosidase similarly to the wild type tail. In control cells coexpressing lamin C, no interaction with 3A was observed (Fig. 4).   8) for 2 h at 4°C following absorption to glutathione-agarose beads for 1 h. After washing the beads, the bound radioactive proteins were detected by SDS-polyacrylamide gel electrophoresis and autoradiography. One fifth of the total amount of the in vitro-translated 2 (lane 1), luciferase (lane 4), and 3A (lane 6) were applied on the gel for comparison. B, GST or wild type GST-MPR300-CT fusion protein (25 g) were incubated with 35 S-labeled 2 (lanes 10 and 11) or 3A (lanes 13 and 14) as described above. Ten percent of the total amount of in vitro translated 2 (lane 9) and 3A (lane 12) was applied on the gel for comparison. Bound 2 (lanes 3 and 11) or 3A (lanes 8 and 14) was quantified using a phosphorimager and expressed as percentage of total 2 or 3A, respectively. Binding was corrected by the unspecific binding of 35 S-labeled 2 (lanes 2 and 10) or 3A (lanes 7 and 13) to GST, respectively. Autoradiograms of one representative experiment out of three are shown.

FIG. 3. Interaction of various mutant MPR46-CT with 2.
A, schematic illustration of wild type MPR46 cytoplasmic tail. The Gal4-DBD (black) is N-terminally fused to the MPR46-CT construct shown as single-letter code amino acid sequence. The numbering starts with Q following the transmembrane domain according to Johnson and Kornfeld (3). The tyrosine-based and the dileucine motifs (gray box) as well as the cluster of acidic amino acid residues (underlined) are indicated. B, yeast cells were cotransformed with the indicated pAS2 plasmids and plasmids expressing the pGAD424 fused to 2. The ability to grow on plates lacking histidine (ϪHis) for 5 days was tested. the cluster of acidic amino acids (residues 50 -62) is responsible for the reduced binding of mutant MPR46-CT StopV48 to 2 and 3A in comparison with the StopH63 mutant as well as for residual adaptor subunit binding in the StopH63-Y45A/V48A mutant, substitutions of acidic residues in the context of the entire tail were tested (Fig. 5). The coexpression of the mutant MPR46-CT containing double substitution of Glu 55 , Glu 56 , and the quadruple substitution of Glu 55 , Glu 56 , Glu 58 , and Glu 59 by alanine with 2 reduced ␤-galactosidase activity by 22 and 31%, respectively (Fig. 5C). These mutations impaired the interaction with 3A more strongly (20 and 18% of ␤-galactosidase activity of the wild type MPR46-CT). The substitution of Glu 55 , Glu 56 , Glu 58 , Glu 59 , Asp 61 , and Asp 62 by alanine reduced the interaction both with 2 and 3A to 26 and 5%, respectively, of wild type MPR46-CT. DISCUSSION Studies using the yeast two-hybrid system have established that tyrosine-based sorting signals in cytoplasmic domains of membrane receptors are important for the interaction with 1, 2, and 3 subunits of adaptor complexes AP-1, -2, and-3, respectively (10,13,14). In the present study, we have examined the full-length cytoplasmic tail of the MPR46 containing several independent signal structures including tyrosine-and dileucine-based motifs (residues 45-48 and residues 64 and 65, respectively), aromatic residues (residues 13, 18, and 19), and clusters of acidic amino acids (residues 50 -62) for its capability to interact with medium chains of AP-2 and AP-3 in the yeast two-hybrid system. Here we report that the tyrosine motif and a distal cluster of acidic amino acid residues but not the dileucine signal within the MPR46 tail are sufficient for effective binding to 2. In contrast, the interaction of the receptor tail with 3A depends on a complex C-terminal structure comprising tyrosine and dileucine motifs as well as a strong hydrophilic sequence.
The phenylalanine residues 13 and 18 as well as the tyrosine-based motif 45 YRGV 48 have been shown to mediate the internalization of the MPR46 from the cell surface (3). Whereas the former signal overlaps with the specific endosomal sorting signal 18 FW 19 , the tyrosine motif has the characteristics of the consensus motif YXX (where Y is tyrosine, X is any amino acid, and is a bulky hydrophobic amino acid) which can be found within the cytoplasmic domains of many endocytic transmembrane proteins recognizing 2 (13,20). Recently, crystallization studies of 2 complexed with tyrosine-containing internalization signal peptides have revealed that the peptides are bound to an extended conformation with separate pockets for both Y and residues (21). Consistent with the prediction that the dileucine signals, which are characteristically surrounded by polar and/or charged residues, will not be able to bind to the hydrophobic pocket structure of 2 (21), the dileucine signals of CD 3␥ have been proposed to interact with ␤2 adaptor subunits (16). However, surface plasmon resonance data have demonstrated the binding of dileucine-based sorting signals within the cytoplasmic tail of the major histocompatibility complex class II invariant chain to the 2 chain of AP-2 (18). Here we show that the dileucine motif of the MPR46 tail is not critical for the interaction with 2 in agreement with studies testing the binding of AP-2 to MPR46 tail peptides (15). Furthermore, the present data with C-terminal truncated MPR46 tails or with substitution of Tyr 45 and Val 48 by alanine in the context of the entire tail showed that, in addition to the tyrosine motif, a second signal in the stretch of amino acids 49 -63 is required for efficient 2 binding. Indeed, this study demonstrates that the substitution of six acidic residues within the compact cluster of acidic amino acids in this region ( 50 DDQLGEESEERDD 62 ) prevents the interaction with 2. The positive electrostatic surface potential near the YXX binding site of 2 (21) may support this interaction. Additionally, the acidic residues may also be important in vivo for the binding of the MPR46-CT to 2 because these amino acids are part of a casein kinase-2 phosphorylation site at Ser 57 (23). Since the nonphosphorylated MPR46 are not cycled via the plasma membrane (6), these receptors should fail to interact with AP2. Both a tyrosine-based motif and a cluster of acidic  residues have also been reported to function as independent signals within the cytoplasmic tail of furin to mediate TGN localization and internalization from the cell surface (24,25). Residues 2-16 of the MPR46 tail, which have been reported to affect AP-2 binding (12), are not important for MPR46 tail binding to the 2 chain, suggesting that these juxtamembrane residues may be involved in the binding of other AP-2 subunits. In contrast to the coprecipitation of 2 with the MPR300-CT, this interaction could not be demonstrated in the two-hybrid assay. The reason for these discrepancies is unclear. AP-3 has been shown to be involved in a clathrin-independent transport of membrane proteins to lysosomes in mammalian cells or to the vacuole in yeast, and to lysosome-related storage granules arising from the endocytic pathway like melanosomes in melanocytes, synaptic vesicles in neurons, as well as synaptic-like microvesicles in neuroendocrine cells (26 -30). Using the yeast two-hybrid system, tyrosine-based sorting motifs YXX of LAMP I and CD63 have been shown to interact with 3A and 3B chains (10,27) whereas in the cytoplasmic tails of LIMP II, tyrosinase, and vacuolar alkaline phosphatase or Vam3p, dileucine residues and acidic clusters were found to be important for AP-3 binding (26,28,31). Interestingly, we show for the first time the requirement for multiple C-terminal motifs within the cytoplasmic tail of the MPR46 comprising tyrosine-and dileucine-based signals, and clustered acidic amino acid residues to interact efficiently with 3A. Quantification of 3A interactions with various MPR46 tails by meas-urement of lacZ reporter gene activities revealed that the different signals were not additive but appeared to function in a cooperative manner. In addition, the data showed that the acidic residues Glu 55 , Glu 56 , Glu 58 , and Glu 59 are more critical for efficient binding to 3A than to 2. At least two signals located in the cytoplasmic tail of LIMP II, tyrosinase and synaptotagmin, have also been reported to be critical for correct intracellular sorting (28,30,32). It is likely that the structural distance between the tyrosine-and dileucine-based motifs and the acidic cluster in the cytoplasmic tail of MPR300 as well as their position with respect to the membrane (33, 34) compared with the MPR46 tail may be responsible for the failure of the MPR300 tail to bind to 3A.
Our present data are in contradiction to studies in which the MPR46 tail peptide immobilized on a sensor surface fails to bind AP-3 complexes (28). However, similar discrepancies have been reported for the 3A/AP-3 binding to the LAMP I tail (27,28). The different results obtained with our two-hybrid approach and with the plasmon resonance technique using the MPR46 tail (28) might be explained by the usage of 3A compared with purified AP-3 complexes from brain cytosol containing 3B. Another possibility might be the formation of receptor tail dimers in the two-hybrid system resembling prevalent dimeric MPR46 forms present in membranes (35), which might facilitate the binding of 3A dimers. The physiological significance of MPR46 binding to 3A, which preferentially interacts with lysosomal membrane proteins, is unclear. However, sub- stitution of Cys 34 or deletion of amino acid residues 20 -23 or 24 -29 as well as of the C-terminal end of the cytoplasmic tail interfere with endosomal sorting processes (7,8,22) and might direct the MPR46 to organelles for degradation in an AP-3-dependent manner.