lished results). Briefly, a heterozygote was grown in 50 pg/mI of hygromycin for 10 passages, and LPG+ cells were removed by agglutination with monoclonal antibody CA7AE. Several clonal lines containing homozygous HYG/HYG replacements were recovered (termed fpg2- knockouts). 20. In vitro LPG synthesis was assayed with uridine diphosphate-Gal, guanosine diphosphate-1 4C-Ia- beled Man, and Leishmania microsomal membranes as described (6). In the presence of 0.1% Triton X-1 00,1580 + 220 cpm was incorporated with the wild type and 950 + 350 cpm with C3P0 (65%). Incorporation (1.3%) was observed previously with R2D2 (6). Without detergent, 4760 + 660 cpm was incorporated with the wild type and 500 + 80 cpm (11 %) with C3PO. 21. S. J. Turco, M. A. Wilkerson, D. R. Clawson, J. Biol. Chem. 259, 3883 (1984). 22. J. K. Lovelace and M. Gottlieb, Mol. Biochem. Para- sitol. 22, 19 (1987). 23. Protein immunoblot analyses with the mouse monoclonal antibody to gp63, 235, showed iden- tical mobilities for gp63 from the wild type, C3PO, and the lpg2- knockout, all of which which were clearly distinct from nonglycosylated forms ob- tained after tunicamycin treatment. 24. A. Descoteaux, N. Dean, S. M. Beverley, unpub- lished data. 25. A. Jardim, D. L. Tolson, S. J. Turco, T. W. Pearson, R. W. Olafson, J. Immunol. 147, 3538 (1991); A. Jardim et al., Biochem. J. 305, 305 (1995); H. M. Flinn, D. Rangarajan, D. F. Smith, Mol. Biochem. Parasitol. 65, 259 (1994). 26. D. F. Smith and D. Rangarajan, Glycobiology 5, 161 (1995). 27. E. Handman, L. F. Schnur, T. W. Spithill, G. F. Mitch- ell, J. Immunol. 137,3608 (1986); T. B. McNeely and S. J. Turco, ibid. 144, 2475 (1990); M. Elhay et al., Mol. Biochem. Parasitol. 40, 255 (1990); S. H. Reiner, S. Zheng, Z. F. Wang, L. Stowring, R. M. Locksley, J. Exp. Med. 179,447 (1994); R. Cappai et al., Parasitology 108, 397 (1994). 28. A. Descoteaux, S. J. Turco, S. M. Beverley, unpub- lished data. 29. C. Whitfield and M. A. Valvano, Adv. Microb. Physiol. 35, 135 (1993); P. Reeves, in Bacterial Cell Wall, J.-M. Ghuysen and R. Hakenbeck, Eds. (Elsevier, New York, 1994), pp. 281-317. 30. H. L. Callahan and S. M. Beverley, J. Biol. Chem. 266,18427 (1991). 31. A. Cruz, C. M. Coburn, S. M. Beverley, Proc. Natl. Acad. Sci. U.S.A. 88, 7170 (1991). 32. D. L. Tolson, S. J. Turco, R. P. Beecroft, T. W. Pear- son, Mol. Biochem. Parasitol. 35,109 (1989). 33. R. M. Mortensen, D. A. Conner, S. Chao, A. A. T. Geisterfer-Lowrance, J. G. Seidman, Mol. Cell. Biol. 12, 2391 (1992). 34. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, lie; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gin; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. 35. J. Kyte and R. F. Doolittle, J. Mol. Biol. 157, 105 (1982). 36. A. Bello, B. Nare, D. Freedman, L. Hardy, S. M. Beverley, Proc. Natl. Acad. Sci. U.S.A. 91, 11442 (1994). 37. U. K. Laemmli, Nature 227, 680 (1970); K. Katakura and A. Kobayashi, Infect. Immun. 56, 2856 (1988). 38. T. lig, D. Harbecke, M. Wiese, P. Overath, Eur. J. Biochem. 217, 603 (1993). 39. We thank H. Flinn for assistance, D. Russell for advice on cellular localization, N. Dean for exten- sive advice and communication of submitted manuscripts, D. Dwyer for discussions, and S. Cilmi, D. Dobson, F. Gueiros-Filho, and S. Singer for comments. The mouse monoclonal antibody to gp63, 235, was provided by D. Russell and W. R. McMaster. This work was supported by NIH grant A131078. A.D. is the recipient of a Centennial Fel- lowship from the Medical Research Council of Can- ada, and SM.B. and S.J.T. are Burroughs-Well- come Scholars in Molecular Parasitology. 23 May 1995; accepted 8 August 1995 Interaction of Tyrosine-Based Sorting Signals with Clathrin-Associated Proteins Hiroshi Ohno, Jay Stewart, Marie-Christine Fournier, Herbert Bosshart, Ina Rhee, Shoichiro Miyatake, Takashi Saito, Andreas Gallusser, Tomas Kirchhausen, Juan S. Bonifacino* Tyrosine-based signals within the cytoplasmic domain of integral membrane proteins mediate clathrin-dependent protein sorting in the endocytic and secretory pathways. A yeast two-hybrid system was used to identify proteins that bind to tyrosine-based signals. The medium chains (Ku and p2) of two clathrin-associated protein complexes (AP-1 and AP-2, respectively) specifically interacted with tyrosine-based signals of several integral membrane proteins. The interaction was confirmed by in vitro binding assays. Thus, it is likely that the medium chains serve as signal-binding components of the clathrin-de- pendent sorting machinery. Targeting of integral membrane proteins to endosomes, lysosomes, the basolateral plasma membrane, and the trans-Golgi net- work (TGN) is largely mediated by sorting signals contained within the cytoplasmic domain of the proteins [reviewed in (1)1. Many of these sorting signals consist of continuous sequences of four to six amino acids containing a critical tyrosine residue. A subset of tyrosine-based signals conforms to the canonical motif YXX0, where Y is tyrosine, X is any amino acid, and 0 is an amino acid with a bulky hydrophobic side chain (1). Although much has been learned in recent years about the structure and function of tyrosine-based signals, the mo- lecular mechanisms involved in their rec- ognition are still poorly understood. Previ- ous studies have provided evidence for an association of cytoplasmic domains bearing tyrosine-based signals with clathrin-associ- ated protein complexes (2). However, the exact identity of the signal-binding proteins and the molecular details of the recognition event remain to be established. We decided to search for proteins that interact with tyrosine-based sorting signals, using a yeast two-hybrid approach (3). As a "bait" in the two-hybrid system, we used a triple repeat of the tyrosine-containing se- quence SDYQRL (4, 5) from the cytoplas- mic tail of the integral membrane protein TGN38 (6). This sequence has the charac- teristics of a YXXQ motif and mediates both internalization from the cell surface and localization to the TGN (7). Screening H. Ohno, M.-C. Fournier, H. Bosshart, I. Rhee, J. S. Bonifacino, Cell Biology and Metabolism Branch, Nation- al Institute of Child Health and Human Development, Na- tional Institutes of Health, Bethesda, MD 20892, USA. J. Stewart, A. Gallusser, T. Kirchhausen, Department of Cell Biology and Center for Blood Research, Harvard Medical School, Boston, MA 02115, USA. S. Miyatake and T. Saito, Division of Molecular Genetics, Center for Biomedical Science, Chiba University School of Medicine, Chiba 260, Japan. *To whom correspondence should be addressed. of a mouse spleen complementary DNA (cDNA) library (-2.5 X 106 clones) result- ed in the isolation of two clones that inter- acted specifically with the (SDYQRL)3 bait sequence (8). The two clones (termed 3M2 and 3M9) corresponded to the medium chain (R2) of the plasma membrane, clath- rin-associated protein complex AP-2 (9). In addition to 112, the AP-2 complex contains two large chains (ox- and P-adaptin) and one small chain (u2) (10). Using growth on histidine-deficient (-His) plates as an assay (1 1), we found that proteins encoded by both 3M2 and 3M9 interacted not only with the (SDYQRL)3 repeat but also with a single SDYQRL se- quence and with the full-length TGN38 cytoplasmic tail (Fig. 1A). Mutation of the tyrosine (Y) residues in all three contexts abolished interaction with the [2 clones (Fig. lA). The binding specificity of p, was further characterized by mutation of each residue of the SDYQRL sequence individu- ally to alanine. Only the Y and L residues were absolutely required for interaction with 3M9, whereas mutation of the S, D, and Q residues had no detectable effect, and mutation of the R residue decreased but did not completely abolish the ability to grow on -His plates (Fig. 1 B). Thus, 112 was capable of interacting with the sequence SDYQRL in various contexts and under sequence requirements that were consistent with those defined in studies in vivo (7). To corroborate the results obtained with the two-hybrid system, we tested whether in vitr(-translated, 35S-methionine-labeled R2 was capable of interacting with various se- quences appended to glutathione-S-trans- ferase (GST) (Fig. 2). We observed that both the 3M2 and 3M9 forms of R2 bound to GST-(SDYQRL)3 but not to GST- (SDGQRL) or to GST (Fig. 2A). In vitro- translated luciferase, used as a negative con- trol, did not interact with any of the GST fusion proteins tested (Fig. 2A). Binding of SCIENCE * VOL. 269 * 29 SEPTEMBER 1995 1 872 on September 2, 2008 www.sciencemag.org Downloaded from