MOLECULAR AND CELLULAR BIOLOGY, Mar. 1990, p. 1OOO-1009 Copyright 0 1990, American Society for Microbiology 0270-7306/90/031000-10$02.00/0 Vol. 10, No. 3 The src Protein Contains Multiple Domains for Specific Attachment to Membranes J. M. KAPLAN,'.'t H. E. VARMUS,2.3 AND J. M. The G. W. Hooper Research Foundation,' Department of Microbiology and Immunology,3 and Department of Biochemistry and Biophysics,2 University of California Medical Center, San Francisco, California 94143 Received 30 August 1989/Accepted 20 November 1989 The proteins encoded by the oncogene v-src and its cellular counterpart e-src (designated generically here as pw? are tightly associated with both plasma membranes and intracellular membranes. This association is due in part to the amino-terminal myristylation of ppWm, but several lines of evidence suggest that amino-terminal portions of the protein itself are also involved. We now report that pw" contains at least three domains which, in conjunction with myristylation, are capable of mediating attachment to membranes and determining subceitular localization. We identified these domains by fusing various portions of p@"" to pyruvate kinase, which is normally a cytoplasmic protein. Amino acids 1 to 14 of ppaoVr are sufficient to mediate both myristylation and the attachment of pyruvate kinase to cytoplasmic granules. In contrast, amino acids 38 to 111 mediate association with the plasma membrane and perinuclear membranes, whereas amino acids 204 to 259 mediate association primarily with perinuclear membranes. We conclude that ppMY" contains independent domains that target the protein to distinctive subcellular locations and thus may facilitate diverse biological functions of the protein. The products of the retroviral oncogene v-src and its cellular counterpart c-src (designated generically here as ppW") are protein tyrosine kinases that are associated with membranes. Association of pp60"" with membranes is es- sential for transformation by v-src (19). In various cell types, the viral and cellular src proteins are associated with plasma membranes (7, 8, 21,44), with perinuclear membranes (36), with secretory organelles in both chromaffin cells and plate- lets (30, 33), and with growth cones in developing neurons (27, 41). Little is known about how pp6o"'c is specifically targeted to these subcellular locations. The amino terminus of pv is covalently coupled to a 14-carbon fatty acid, myristic acid (4, 39). The myristyl moiety promotes the association of pp60."' with membranes (2, 11). On the other hand, not all myristylated src proteins are associated with membranes (3, ll), and some nonmyri- stylated src proteins are membrane associated, albeit weakly (12, 20). These results suggest that pp60"" contains amino acid sequences which, in conjunction with amino-terminal myristylation, mediate the association of the protein with membranes. Previous work has implicated the amino-terminal 10 kilo- daltons (kDa) of ppW" in attachment to membranes (22, 35). Here we report work that dissects the amino-terminal domain into several regions that independently target and attach the protein to membranes in distinctive subcellular locations. MATERIALS AND METHODS Materials. The following were gifts: monoclonal anti- ppWc antibody 327 (MAb327). from J. Brugge (25); rabbit antibodies specific for the chicken M1 pyruvate kinase (PK), from B. Roberts; simian virus 40 expression vectors contain- ing the chicken MI PK cDNA (RL142PKlOX, RL18PK8X, and RL18PK12X, which differ by the reading frame at a * Corresponding author. t Present address: Massachusetts Institute of Technology, Biol- ogy 56-629, Cambridge, MA 02139. polylinker sequence), from B. Roberts (16); and a plasmid containing the chicken fibroblast c-src cDNA, p5H, from H. Hanafusa. The E3 src gene is a deletion mutant which lacks the sequence encoding amino acids 8 to 37 of pp6OSrc, as previously described (17). Reagents were obtained as fol- lows: protease inhibitors, paraformaldehyde, bovine serum albumin (fraction V), and detergents from Sigma Chemical Co. ; fluorescein-conjugated goat anti-mouse immunoglobu- lin G, goat anti-rabbit immunoglobulin G, and normal goat serum from the Jackson Immunochemicals Co.; protein A conjugated to Sepharose CL-4B from Pharmacia Fine Chem- icals; rhodamine-conjugated phalloidin from Molecular Probes Inc.; ~-[~'S]methionine from ICN Radiochemicals; [3H]myristic acid, 22.4 CVmmol (50 mCi/ml in dimethyl sulfoxide) from Du Pont, NEN Research Products; restric- tion enzymes, Escherichia coli DNA polymerase 1, and T4 DNA ligase from New England BioLabs, Inc.; human serum fibronectin from Collaborative Research Inc.; and Gold Seal glass cover slips (18 by 18 mm; thickness, 1.5 mm) from VWR Scientific. Metaboli labeling, immunoprecipitation, and subcellular fractionation. COS7 cells were labeled with ~-["S]methio- nine for 18 h in Dulbecco modified Eagle medium containing 10% the normal concentration of t-methionine, 10% dia- lyzed fetal calf serum, and 200 pCi of ~-[~~Slmethionine per ml. COS7 cells were labeled with [3H]myristic acid for 18 h in Dulbecco modified Eagle medium with 10% fetal calf serum, 1% dimethyl sulfoxide, and 500 pCi of [3Hlmyristic acid per ml (5). Labeled cell extracts were prepared with lysis buffer (20 mM Tris hydrochloride [pH 8.01, 150 mM NaCI, 1% Nonidet P-40, 50 kg of soybean trypsin inhibitor per ml, 50 pg of aprotinin per ml, 20 pg of leupeptin per ml, and 1 mg of fraction V bovine serum albumin per ml). Immunoprecipitations with polyclonal rabbit sera and mono- clonal antibodies were performed as described previously (21). Crude cell extracts prepared by Dounce homogeniza- tion of hypotonically swelled cells were fractionated into cytosol and membrane fractions by differential centrifuga- tion, as previously described (21). Subcellular fractionations VOL. 10,1990 uc25BPK 411 II MEMBRANE ATTACHMENT OF src PROTEIN 1001 + +FMPNQ srcl 1 1 PK - srQOyA8-111)PK 0 ucSllPK 0 rrclll(A15-54)PK 0 I] wcl4PK 0 rrc7PK 0 (+) + WPNP ++G + - CYT Q PM+PN PN FIG. 1. Summary of the membrane association of src-PK proteins. The sequences derived from pp60"' in all of the src-PK proteins are depicted. The restriction sites used to create internal deletions of src sequences and to fuse the src sequences to the coding region of PK are indicated. Residue numbers correspond to the sequence of ppWC encoded by the Prague C strain of Rous sarcoma virus (39). Myristylation (Myr) of hybrid proteins, when analyzed, is indicated by +. When src-PK proteins associated with membranes, myristylation was assumed [and denoted (+)I but not tested. Membrane association (MA) of hybrid proteins, as determined by biochemical fractionations, is indicated by f. The intracellular distribution of hybrid proteins was determined by immunofluorescence (IF). Association of hybrid proteins with plasma membranes (PM), with the cytoplasm (CYT), with perinuclear membranes (PN). or with cytoplasmic granules (G) is indicated. At the bottom of the figure the N-terminal259 amino acids of ppWC are divided into functional domains. The first 7 amino acids of pp6(yn' 1 act as a recognition sequence for myristylation. Amino acids 1 to 14 cause association with cytoplasmic granules (G). Amino acids 38 to 111 ( 1, together with the myristylation signal, cause association with both the plasma membrane and perinuclear membranes (PM+PN). Amino acids 204 to 259 ( OlEl), together with the first 14 amino acids, causes association solely with perinuclear membranes (PN). The myristyl moiety is indicated by the crooked line in the diagram of src259PK (top of figure). were quantitated by scanning autoradiograms with a Bio- Rad densitometer. Construction of recombinant sre-PK genes. The N-terminal amino acid sequence of the hybrid src-PK proteins is derived from pw (B77 strain of Rous sarcoma virus), whereas the C-terminal amino acids are derived from PK. The src-PK proteins are designated by the C-terminal boundary of the domain of ppW" that is fused to PK. and in cases in which this domain contains internal deletions, the amino acids that are lost are indicated in parentheses. Residue numbers reported correspond to the sequence of the Prague C strain of Rous sarcoma virus (40). The amino acids derived from the chicken M1 PK (residues 120 to 529) are the same in all cases (26). The nucleotide sequence encoding PK was derived from plasmid RL142PKlOX, RL18PK8X, or RL18PK12X as tequired to maintain an open reading frame. The src7PK plasmid (previously known as R7-PK) was described previ- ously (17). Sequences encoding various N-terminal amino acids of pp6(y" were fused to the coding region of PK by cutting both genes with restriction endonucleases, repairing the ends with E. coli DNA polymerase I, and fusing the resulting blunt ends. Internal deletions of src sequences were constructed by cutting the src gene with restriction endonucleases and fusing the repaired ends. The restriction . sites used in these constructions are indicated in Fig. 1. The SruI site at codon 7 is a restriction polymorphism which was created by converting lysine 7 to arginine (17). A simian virus 40 vector expressing the chicken fibroblast pp6(y", pSVSH, was constructed by replacing the PK cDNA of RL142PKlOX with the c-src cDNA derived from p5H. Further details about the construction of these genes will be provided upon request. Transfection of recombinant src-PK genes. COS7 cells were exposed for 18 h to calcium phosphate precipitates contain- ing 20 pg of recombinant src-PK DNA and then subjected to a 2-min shock with Tris-buffered saline (0.8% NaCI, 0.1% glucose, 0.038% KCI, 0.2% Tris hydrochloride, 0.06% Tris base, 4.5 mg of phenol red per liter) containing 25% dimethyl sulfoxide (15). Transfected COS7 cells were labeled for 18 h with either ~-["S]rnethionine or ['H]myristic acid starting 36 h after the transfection had begun. Immunofluorescence. Glass cover slips were coated with human serum fibronectin (100 pg/d in Dulbecco phosphate- buffered saline [PBS]) for 1 h at 37°C in a humidified chamber. Transfected COS7 cells were seeded onto glass cover slips 24 h after the transfection had begun. After 8 to 12 h, the cover slips were washed with PBS. cells were fixed with 3.5% paraformaldehyde in PBS for 20 minutes and 1002 KAPLAN ET AL. MOL. CELL. BIOL. FIG. 2. The first 111 amino acids of pp60sm contain a membrane-anchoring domain. COS7 cells were transfected with plasmids encoding native PK (lanes 1 to 31, src7PK (lanes 4 to 6). src259PK (lanes 7 to 9), srclllPK (lanes 10 to 12), or src204(A8-111)PK (lanes 13 to 15) as described in Materials and Methods. Transfected cells were labeled with ~-[~~S]methionine for 12 h. Crude extracts (T; lanes 1,4,7,10, and 13) were fractionated into cytosol (S; lanes 2,5,8,11. and 14) and a membrane pellet (P; lanes 3,6,9,12, and 15) by differential centrifugation. Equivalent amounts of each fraction were analyzed by immunoprecipitation with a polyclonal rabbit anti-PK antibody. The positions of 200-, 97-, 69-, and45-kDa molecular mass markers are indicated. The src-PK proteins are indicated by the arrowheads. permeabilized with 0.1% Triton X-100 in PBS, and cover slips were coated with 10% normal goat serum in PBS to block nonspecific binding sites. * Primary antibodies used were monoclonal anti-pp60.'" 327 (10 pg/ml in PBS) and polyclonal rabbit anti-PK (1/200 dilution in PBS). Fluores- cein-conjugated antibodies were used at a dilution of 1/200 in PBS. Rhodamine-conjugated phalloidin was used as recom- mended by the manufacturer. Stained cover slips were washed several times with PBS and then with absolute ethanol and were embedded in glycerol containing 2% propyl gallate, an antibleaching reagent. Microscopy was done with either a Zeiss Photomicroscope 111 or an inverted Olympus fluorescence microscope. Cells were photographed with Kodak Tn-X film at ASA 200. In general. 1 to 5% of the transfected COS7 cells gave a fluorescent signal following this procedure; this signal presumably corresponds to the fraction of cells that took up the DNA. The surrounding nontluorescent cells acted as an internal control for the specificity of the antibodies used. RESULTS To test the role of N-terminal amino acid sequences of ppW" in association with membranes and in subcellular localization, we constructed a set of genes encoding hybrid src-PK proteins (Fig. 1). Since myristylation is normally required for membrane association, we included the myri- stylation signal in all of these hybrid genes. The association of the hybrid proteins with membranes was analyzed first by a crude biochemical fractionation and subsequently by im- munofluorescence. Myristylation is not sufecient for membrane anchorage. We have previously shown that the first 7 amino acids of pvr act as a sufficient recognition sequence for myristylation (17). Fusing sequences encoding the first 7 amino acids of ppWC to the coding region of PK creates a hybrid gene, src7PK, which encodes a myristylated protein (see Fig. 6). If myristylation is a sufficient cause of membrane association, the src7PK protein should attach to membranes. Cells expressing either the native PK protein or the src7PK protein were fractionated into soluble and membrane fractions, and these fractions were assayed for PK proteins (Fig. 2). Virtually all of the native PK protein (Fig. 2, lanes 1 to 3) and the src7PK protein (Fig. 2, lanes 4 to 6) was recovered in the soluble fraction (Fig. 2, lane 2 and 5, respectively), indicating that a second domain of pp6os" is required in conjunction with myristylation to cause mem- brane association. The gels illustrated in Fig. 2 contained variable amounts of proteins that appeared ancillary to the specific products of the transfected DNAs (marked by arrowheads). We do not know the identities of these ancillary proteins, but they were present in all specimens, including cells that had not been transfected. We therefore attribute them to nonspecific immunoprecipitation. Anchoring PK to membranes with a domain from pWm. Since previous reports had implicated N-terminal sequences in pp6OV-"" as membrane-anchoring domains (9, 12, 19, 22, 34), we fused sequences encoding the first 259 amino acids of the protein to the coding sequences of PK, creating a hybrid gene called src259PK. Cells expressing the src259PK protein were fractionated into cytosol and membrane fractions, and these fractions were assayed for the src259PK protein (Fig. 2, lanes 7 to 9). Most (ca. 90%) of the src259PK protein was recovered in the membrane fraction (lane 9). This result demonstrates that sequences in the N-terminal half of pp6os", together with myristylation, are sufficient to mediate association with membranes. The srcZ59PK protein and pp60'= associate with mem- branes in a similar manner. Before further characterizing the membrane-anchoring domains made up of the N-terminal VOL. 10, 1990 MEMBRANE ATTACHMENT OF src PROTEIN 1003 FIG. 3. Influence of myristylation of subcellular localization. COS7 cells were transfected with plasmids encoding either the src259PK protein (lanes 1 to 3) or the N7-259PK protein (lanes 4 to 6). Transfected cells were labeled with ~-['~S]methionine for 12 h. Crude extracts (T, lanes 1 and 4) were fractionated into cytosol (S; lanes 2 and 5) and a membrane pellet (P; lanes 3 and 6) by differential centrifugation. Equivalent amounts of each fraction were analyzed by immunoprecipitation with a polyclonal rabbit anti-PK antibody. The src259PK protein is indicated by the arrow. The positions of 97- and 6PkDa molecular size markers are indicated. 259 amino acids of pp60"", we tested whether the src259PK and pv proteins associate with membranes in a similar manner. We first asked whether myristylation is required for the association of src259PK with membranes. Replacement of lysine 7 with an asparagine largely abolishes myristylation of ppW" (17). Therefore, we converted lysine 7 to aspara- gine in src259PK and evaluated the subcellular distribution of the mutant protein (Fig. 3, lanes 4 to 6). The protein (designated N7-259PK) was found predominantly (ca. 87%) in the soluble fraction of cells (Fig. 3, lane 5), implying that the membrane association of src259PK, like that of pp6W itself, requires N-terminal myristylation. A second hallmark of the association of pp60"" with membranes is that, like other peripheral membrane proteins, pv can be removed from membranes by alkaline extrac- tions (Fig. 4, lanes 4 and 5). Similarly, the src259PK protein is extracted from membranes following extraction with base (Fig. 4, lanes 9 and 10). On the other hand, neither pp60-'" (Fig. 4, lanes 2 and 3) nor src259PK (Fig. 4, lanes 6 to 8) is efficiently extracted from membranes by 0.3 M NaCl. We conclude that by several criteria, src259PK and pp60"" appear to associate with membranes in a similar manner. The first 111 amino acids of the src protein are sufficient for membrane association. Previous work indicated that the N-terminal 10 kDa of ppW" is required to anchor the protein to membranes (22, 34). To better define the respon- sible amino acid sequences, we fused the coding region of PK to sequences encoding each of the following domains: the first 204 amino acids of pp60"" (src204PK); the first 111 amino acids of ppW" (srclllPK); and the first 204 amino acids of ppW". from which residues 8 to 111 were deleted [src204(A&lll)PK]. Cells expressing these proteins were fractionated as described above (Fig. 2). Although 75% of the src204(AS-l11)PK protein was recov- ered in the cytosol (Fig. 2, lanes 13 to 151, most of the src204PK protein (not shown in Fig. 2, but see Fig. 4, lanes 11 to 15) and ca. 65% of the srclllPK protein (Fig. 2, lanes 10 to 12) were recovered with the membrane pellet (Fig. 2, lanes 10 to 12). The proteins encoded by both src204PK (Fig. 4. lanes 11 to 15) and srclllPK (Fig. 4, lanes 16 to 20) associated with membranes in a manner similar to pr, FIG. 4. The src-PK proteins and ppWw associate with membranes in a similar manner. B31 cells (lanes 1 to 5) or COS7 cells transfected with plasmids encoding src259PK (lanes 6 to 10). src204PK (lanes 11 to 15). or srclllPK (lanes 16 to 20) proteins were labeled with ~4~'SIrnethionine for 12 h. Crude extracts (T; lanes 1.6, 11, and 16) were adjusted to either 0.3 M NaCl (lanes 2, 3.7, 8, 12, 13. 17, and 18) or 0.1 M NaOH (lanes 4, 5.9,10,14.15,19, and 20) and then fractionated into cytosol (S; lanes 3,5,8,10,13.15.18, and 20) and a membrane pellet (E lanes 2. 4. 7, 9, 12, 14, 17. and 19) by differential centrifugation. Equivalent amounts of each fraction were analyzed by immunoprecipitation with either MAb327 antibody (lanes 1 to 5) or a polyclonal rabbit anti-PK antibody (lanes 6 to 20). The positions of pWm and the src-PK proteins are indicated by the arrowheads. 1004 KAPLAN ET AL. MOL. CELL. BIOL. FIG. 5. The first 111 amino acids of pw contain at least two membrane-anchoring domains. COS7 cells were transfected with plasmids encoding the srclllPK (lanes 1 to 3). the srcl4PK (lanes 4 to 6). the srcUPK (lanes 7 to 9), the srclll(A8-37)PK (lanes 10 to 12), the srclll(Al5-54)PK (lanes 13 to 15). or the src259(A1&203)PK (lanes 16 to 18) proteins. Transfected cells were labeled with ~-["S]rnethionine for 12 h. Crude extracts fl; lanes 1,4,7,10,13, and 16) were fractionated into cytosol (S: lanes 2,5,8,11,14, and 17) and a membrane pellet (P; lanes 3,6,9,12,15, and 18) by differential centrifugation. Equivalent amounts of each fraction were analyzed by immunoprecipitation with a polyclonal rabbit anti-PK antibody. The src-PK proteins are indicated by the arrowheads. Note that the srcl4PK protein migrates slightly farther than a background protein, which prevented quantitation of the membrane association of srcl4PK. because they were extracted only partially from membranes by 0.3 M NaCI, but were extracted quantitatively by base. We conclude that the first 111 amino acids of pp60-'" are sufficient to cause association with membranes and that the association is similar to that found with native pp60"". By contrast, amino acids 111 to 204, together with N-terminal myristylation, permit only limited association with mem- branes. The first 111 amino acids of the src protein contain at least two independent membrane-anchoring domains. The domain represented by the first 111 amino acids of pw was divided into several segments, and the sequences encoding these segments were fused to the coding region of PK. The association of these hybrid proteins with membranes was analyzed by biochemical fractionation as described above. (i) The first 14 amino acids of the src protein contain a membrane-anchoring domain. It was previously shown that fusing the first 14 amino acids of pp60sK to any one of several other proteins causes these proteins to associate with mem- branes in crude biochemical fractionations (1, 31, 32). This result was confirmed by fusing sequences encoding the first 14 amino acids of pwr to the PK-encoding region, creat- ing srcl4PK. An appreciable fraction of the srcl4PK protein was recovered with the membrane pellet after subcellular fractionation (Fig. 5, lanes 4 to 6), although precise quanti- tation was prohibited by the presence of a nearly comigrating protein. Other hybrid proteins that contained the first 14 amino acids of pp6os'c also associated with membranes: 72% of src54PK (Fig. 5, lanes 7 to 9), 72% of srclll(A15-54)PK (Fig. 5, lanes 13 to 15), and 67% of src259(A14-203)PK (Fig. 5, lanes 16 to 18) were recovered with the membrane pellet. These results leave open the possibility that amino acids 15 to 54, 55 to 111, and 204 to 259 also contain membrane- anchoring domains, a possibility that will be confirmed below by immunofluorescence (see Fig. 7). (i) Amino acids 38 to 111 of the src protein contain a membrane-anchoring domain. Two previous observations indicated that portions of p- outside the first 14 amino acids contribute to membrane attachment. First, removal of more than 8 kDa from the amino terminus of ppW" failed to detach the protein from membranes (22). Second, the protein encoded by the deletion mutant E3 src, which lacks amino acids 8 to 37, attaches to membranes (17). We sought and identified a second membrane-anchoring domain by fusing the sequence encoding the first 81 amino acids of the E3 src protein to the coding region of PK, creating the srclll(A8-37)PK gene. Most (66%) of the srclll(A137)PK protein (Fig. 5, lanes 10 to 12) was recov- ered with the membrane pellet (Fig. 5, lane 12) following subcellular fractionation. This implies that amino acids 38 to 111 of pp60"" also contain a membrane-anchoring domain. Failure of chimeric proteins to bind membranes cannot be attributed to poor myristylation. An alternative explanation for the behavior of the cytoplasmic src7PK protein and the membrane-associated src259PK and srcl4PK proteins is that they differ not by the presence of a membrane-anchoring domain, but rather in the stoichiometry of myristylation. If the association of pp60.'" with membranes were mediated solely by the myristyl moiety, only the myristylated mole- cules would associate with membranes. This possibility was tested by comparing the stoichiometry of myristylation of the src7PK (Fig. 6, lanes 2 and 5), the srcl4PK (Fig. 6, lanes 3 and 6). and the src259PK (Fig. 6, lanes 1 and 4) proteins. Cells expressing these proteins were labeled with either ~-[~'S]methionine (lanes 1 to 3) or ['Hlmynstic acid (lanes 4 to 6). and the PK proteins were analyzed by immunoprecip- itation. Since it had previously been shown that 84% of pvc molecules are myristylated (4), we arbitrarily used the same value for the stoichiometry of myristylation of src259PK. In comparison, approximately 60% of the srcl4PK protein and 17% of the src7PK protein were myristylated (estimates derived from Fig. 6). Thus, if myristylation were sufficient to cause membrane association, 17% of the src7PK protein VOL. 10, 1990 MEMBRANE ATI'ACHMENT OF src PROTEIN 1005 FIG. 6. The stoichiometries of myristylation of the src259PK. the src7PK. and the srcl4PK proteins are similar. COS7 cells were transfected with plasmids encoding the src259PK (lanes 1 and 4). the src7PK (lanes 2 and 5). or the srcl4PK (lanes 3 and 6) proteins. Transfected cells were labeled with either ~-[~?S]methionine (lanes 1 to 3) or [-'H]myristic acid (lanes 4 to 6) for 12 h. Labeled cells were solubilized with lysis buffer, and PK proteins were immunoprecip- itated with a polyclonal anti-PK antibody. The positions of 97-, 69-, and 45-kDa molecular mass markers are indicated. The PK proteins are indicated by the arrows. Stoichiometries mentioned in Re- sults were calculated as follows, by using data from densitometry: (myr-xlmyr-src259) O'S-~/~~S-src259) would be expected to be membrane associated, whereas only 3% was observed. We also analyzed the membrane association of myristy- lated proteins. If the only relevant difference between the cytoplasmic src7PK and src204(A1111)PK proteins, on the one hand, and the membrane-associated srr259PK and srcl4PK proteins, on the other, were the proportion of the molecules that are myristylated, all of the myristylated molecules would be membrane associated. Cells expressing these proteins were labeled with ['Hlrnyristic acid and subsequently divided into membrane and cytosol fractions. Virtually all of the myristate-labeled src7PK protein was recovered in the cytoplasmic fraction, whereas most of the myristate-labeled src259PK protein and the srcl4PK protein was recovered with the membrane pellet (data not shown). Similarly. much of the myristate-labeled src204(A8-lll)PK protein was cytoplasmic. We conclude that the inability of the src7PK and src204(A1111)PK proteins to associate with membranes does not reflect inefficient myristylation. Localization of src-PK proteins by immunofluorescence. The intracellular distribution of ppW. was previously ana- lyzed by immunofluorescence and electron microscopy (30, 36.37.41, 44). The results indicated that in different types of cells, pp60"' may be associated with a variety of subcellular compartments, including plasma membranes (typically at sites of focal adhesion), perinuclear membranes, and secre- tory granules. Therefore, we used immunofluorescence to characterize the intracellular distribution of src-PK proteins. (i) p@= associates with plasma membranes, perinuclear membranes, and cytoplasmic granules in transfected COS7 cells. The distribution of pp60"" in COS7 cells following transfection with a plasmid containing the c-src gene was determined by immunofluorescence. The intracellular distri- bution of ppW" in COS7 cells appeared to be a composite of three subcellular locations: plasma membranes, pennu- clear membranes, and cytoplasmic granules (Fig. 7A). (U) The PK protein, the src7PK protein, and the srcm(A8- 111)PK protein are cytoplasmic. The intracellular distribution of the PK and hybrid proteins was analyzed by staining fixed cells with a polyclonal anti-PK antibody. The PK and src7PK proteins had a distribution representing the cyto- plasm (Fig. 7B and C): the margins of cells expressing these proteins were poorly visualized after staining with anti-PK antibody, and the intensity of the staining was proportional to the thickness of the cell (Le., to the volume of underlying cytoplasm). The src204(A&lll)PK protein appeared to be equally distributed between the plasma membrane and the cytoplasm (Fig. 7G). These results confirm our interpreta- tion of the biochemical fractionations described above. (iii) Most of the membrane-associated src-PK proteins have an intracellular distribution similar to pwm. The src259PK protein (Fig. 7D), the src204PK protein (Fig. 7E), the srrlllPK protein (Fig. 7F), the srclll(A8-37)PK protein (Fig. 7H), the srrS4PK protein (Fig. 71), and the srelll(Al5- 54)PK protein (Fig. 7J) had a distribution similar to that of ppW" itself. Often cells expressing srr-PK proteins that associate with the plasma membrane appeared to have a filamentous component to the fluorescent signal (for exam- ple, see Fig. 7E). The filamentous structures also contained actin, as demonstrated by staining with rhodamine-conju- gated phalloidin, an actin-specific stain (data not shown). We do not know the nature of these filamentous structures, but they conceivably could correspond to sites of adhesion. (iv) The src259(A1&203)PK protein associates specifically with perinuclear membranes. Interestingly, the src259(Al& 203)PK protein associated primarily with perinuclear mem- branes (Fig. 7K). Since the distribution of the src259(A14 203)PK and the srcl4PK (see below) proteins were different, we conclude that amino acids 204 to 259 contain a domain that preferentially targets proteins to perinuclear mem- branes. (v) The srcl4PK protein associates with cytoplasmic gan- ules. Cells expressing the srcl4PK protein had a punctate pattern of cytoplasmic fluorescence after staining with anti- PK antibody (Fig. 7L). These fluorescent spots had an apparent diameter of 0.2 km and were uniformly distributed throughout the cytoplasm. We suspect that this distribution corresponds to a membranous organelle, rather than a large proteinaceous complex, because the srcl4PK protein does not sediment rapidly after dissolution of membranes with 1% Triton X-100 (data not shown). (The photographic reproduc- tion of Fig. 7L is not adequate to exclude a perinuclear location for some of the srcl4PK protein. However, other exposures of the same field and examination of numerous other cells sustain our conclusion that srcl4PK is located principally in cytoplasmic granules, rather than in either of the other two locations reported here.) The srcl4PK is the only hybrid protein that associates principally with cytoplasmic granules. Although most of the E FIG. 7. Membrane-anchoring domains target proteins to specific subcellular locations. COS7 cell were transfected with plasmids encoding ppXF (A), PK (B), src7PK (C), src259PK (D). m204PK (E), mlllPK 0, srr204(A&lll)PK (G), stclll(A8-37)PK (H), src54PK (I), st~lll(A15-54)PK (J), sm259(A14203)PK (K), or sml4PK (L) proteins. At 48 h after exposure to these DNAs, the cells were seeded onto glass cover slips. After an additional 24 h, the cells were bed with paraformaldehyde, permeabilized with Triton X-100, and stained with either anti-ppW MAb327 or a polyclonal anti-PK antibody. as described in Materials and Methods. Perinuclear membranes (PN) and cytoplasmic granules (G) are indicated. VOL. 10.1990 MEMBRANE ATTACHMENT OF src PROTEIN 1007 src-PK proteins that contain amino acids 1 to 14 were also found in cytoplasmic granules, proteins that contain addi- tional portions of ppW" associated with other membranes as well. For example, both the src54PK protein (Fig. 71) and the srclll(A15-54)PK protein (Fig. 75) associated with the plasma membrane and perinuclear membranes, in addition to cytoplasmic granules. Because the distribution of these proteins differs from srcl4PK. amino acids 15 to 54 and 55 to 111 must contain additional sorting information. These re- sults imply that multiple domains of pp60"" can affect the intracellular distribution of a protein. DISCUSSION Several lines of evidence suggest that the association of pv with membranes is mediated in part by interaction between pw and one or more membrane proteins. First, pp6os" is found in diverse subcellular locations that vary from one type of cell to another. These locations include the plasma membrane in fibroblasts (8,21,44), particularly focal contacts (37); perinuclear membranes in fibroblasts (36); cytoplasmic granules in platelets (33) and chromaan cells (30); and growth comes in neurons (27, 41). These specific- ities suggest that pp60"" interacts with other proteins in the various locations. Second, reconstitution of purified pp60"" into lipid vesicles in vitro requires the addition of membrane proteins (34), and the binding of pp60"" to membranes utilizes at least one protein receptor that is both specific and saturable (35). Third, the protein tyrosine kinase encoded by lck associates with two transmembrane glycoproteins, the CD4 and CD8 antigens of T lymphocytes (38, 42). and the catalytic activity of lck can be modulated by cross-linking the CD4 antigen (43). Since the lck protein is closely related to pp6os", it seems likely that the latter also interacts specifically with membrane proteins and that the interaction may help govern the activity of pp60"". If the attachment of pv to membranes resembles that of the Ick protein, the following predictions should be borne out: (i) pv should behave like a peripheral membrane protein in schemes for solubilization; (ii) myristylation should not be sufficient for the association of ppW" with membranes; (iii) specific domains of pp6o"'c should mediate its association with membranes; and (iv) these membrane- anchoring domains will target pp60"" to specific subcellular locations. The work reported here fulfills each of these predictions. We acknowledge one caveat. Our work was performed with chimeric proteins formed between portions of pw and the cytosolic protein PK. We cannot presently refute the possibility that the conformation (or other features) of these chimeras has influenced the outcome of our experiments in ways that are artifactual. It is reassuring, however, that the chimera src259PK (which contains all of the membrane- targeting domains that we have defined within pp60"") distributes within cells in a manner identical to that of ppW" itself. In addition, the conclusions reached here are in accord with previous observations on the roles of myri- stylation and other domains of pp60"" in the attachment to membranes (for examples, see references 22, 34, and 35). The src protein is a peripheral membrane protein. Previous reports have demonstrated that the association of pw with membranes cannot be disrupted by chelating agents or by extreme concentrations of salt, whereas nonionic deter- gents completely extract pwm from membrane vesicles (21.22). These results led to the provisional conclusion that pp60IK is an integral membrane protein. In contrast, we have shown that pp60"' is a peripheral membrane protein, because it can be extracted from membrane vesicles with alkali. Similarly, hybrid src-PK proteins that are associated with membranes are also extracted with alkali. Perhaps extraction from membranes with alkali will be a general feature of myristylated proteins associated with membranes. The role of myristylation in the membrane association of ~~60'~. Myristylation is neither necessary nor sufficient for the association of pp6os" with membranes. Several forms of pp60"' are myristylated yet cytoplasmic (3, 11, 29). Con- versely, nonmyristylated forms of ppWc have been de- scribed which still associate with membranes (12, 20, 22). Our results provide further support for the notion that the myristyl moiety is not sufficient for the association of pp60"' with membranes: the myristylated proteins src7PK and src204(A8-lll)PK are cytoplasmic when analyzed by either biochemical fractionation or immunofluorescence (see Fig. 1 for a summary). What, then, is the role of myristylation in the attachment of ppW" to membranes? First, the modification may in- crease the affinity of the protein for membranes: forms of pp60"" that lack myristylation but are nevertheless found on membranes can be solubilized by ionic strength alone. Second, myristylation may initiate the binding of pp60"' to membranes: typically, nonmyristylated forms of pp60"" never attach to membranes. The full association rf pp60"" with membranes requires one or more portions of the protein itself, however, and once the association is established, it may be independent of myristylation: cleavage of mem- brane-bound pr with protease removes 8 kDa from the amino terminus, yet fails to disrupt the association of the remainder of the protein with the membranes (22). Membrane-anchoring domains target molecules to distinct subcellular locations. Previous reports have demonstrated that ppW' is found in plasma membranes, perinuclear membranes, and cytoplasmic granules. The results reported here are unusual in that pp6oI'c is found in all of these locations in a single type of cell, a circumstance that we cannot presently explain but that might be due either to the use of COS7 cells or to the exceptional abundance of pp@F produced in the cells (10- to 20-fold more than found in cells transformed by Rous sarcoma virus [data not shown]). Whatever its cause, this circumstance allowed us to demon- strate that distinctive domains of ppscv" are required for targeting the protein to each subcellular location. Amino acids 1 to 14 of pptW" cause proteins to associate primarily with cytoplasmic granules in COS7 cells. Perhaps these amino acids specifically target pp60"" to granules in chromaffin cells and platelets as well. Amino acids 38 to 111 cause proteins to associate with both plasma membranes and perinuclear membranes. We suspect that these amino acids make up multiple membrane-anchoring domains, because in conjunction with amino acids 1 to 14, either amino acids 15 to 54 or amino acids 55 to 111 are sufficient to cause association with plasma membranes, perinuclear mem- branes, and cytoplasmic granules. Amino acids 204 to 259, together with amino acids 1 to 14, cause a protein to associate primarily with perinuclear membranes. Wc con- clude that the subcellular localization of pp6os" is mediated by multiple domains within the protein. Is the subcellular distribution of pp60'" controlled by phosphorylation? How is it that in most types of cells, ppWC is localized to one subcellular compartment and not to others? It appears possible that the activity of membrane- anchoring domains is regulated by phosphorylation. For example, we have shown that amino acids 1 to 14 of pp60"" 1008 KAPLAN ET AL. MOL. CELL. BIOL. target proteins specifically to the membranes of cytoplasmic granules. Within and close to this domain are serine residues that are known to be phosphorylated by cellular kinases: serine 12 by kinase C (13, 14) and serine 17 by the cyclic AMP-dependent protein kinase (6, 10, 18). Perhaps phos- phorylation of these sites helps regulate the subcellular location of pWc. There is precedent for this speculation in the finding that phosphorylation of the epidermal growth factor receptor by kinase C causes internalization of recep- tor molecules (24). The viral and cellular src proteins differ in their amino acid sequence, including regions used here in chimeras. Since the chimeras contained sequences derived only from viral src, the conclusions that we have reached regarding the targeting of src protein to membranous compartments apply rigor- ously only to the viral version of the protein. Alternative splicing in neurons results in the synthesis of a novel form of ppWC (pp60+) (23,28). Perhaps the localization of pp60+ in neurons to growth cones is in part due to these structural alterations. Where within the cell is the transforming activity of ppW" effected? Transformation by the viral version of pp60"" requires that the protein be associated with membranes (see reference 19 for a review), presumably because the crucial substrates of pp6os" are also membrane associated. Since ppW" is capable of associating with various membranes, it is not clear whether transformation results from the activity of ppWC at a single or at multiple subcellular locations. Surprisingly, many of the mutant src proteins with dele- tions that begin at codon 15 retain transforming activity. None of these proteins has been analyzed by immunofluo- rescence. Our results indicate that these transforming src proteins would associate primarily with perinuclear mem- branes and cytoplasmic granules. No mutant allele of src has been described in which all of the membrane-anchoring domains have been deleted. In most of the available mutants, the encoded proteins retain the first 14 amino acids of pp60"'; among these mutant proteins, only the product of the NYl8-3 allele is not associated with membranes (11). The NY18-3 src gene lacks only the sequences that normally encode amino acids 169 to 264, yet it encodes a myristylated protein that is located in the cytoplasm. Our results do not explain these properties. Perhaps the conformation of this protein precludes associa- tion with membranes. ACKNOWLEDGMENTS We thank J. Brugge, B. Roberts, and H. Hanafusa for generously providing reagents. J.M.K. thanks Kathleen Weston and David Morgan for their critical (and often helpful) comments and Lynn Vogel for assistance in preparing the manuscript. The work reported here was supported by Public Health Service grants CA 39832 (to H.E.V.) and CA 44338 (to J.M.B.) from the National Institutes of Health and by funds from the George Williams Hooper Research Foundation (to J.M.B.). J.M.K. was a Fellow of the University of California Medical Scientist Training Program. LITERATURE CITED 1. Brooks-WUson, A. R.. E. Ball, and T. Pawson. 1989. The myristylation signal of ptW'K functionally complements the N-terminal fps-specific region of P13Wagqp'. Mol. Cell. Biol. 9:2214-2219. 2. Buss, J. E., M. P. Kamps, K. Gould, and B. M. Senon. 1986. The absence of myristic acid decreases membrane binding of p6Wm but does not affect tyrosine protein kinase activity. J. Virol. 232468474. 3. Buss, J. E., M. P. Kamps, and B. M. Sefton. 1985. Myristic acid is attached to the transforming protein of Rous sarcoma virus during or immediately after synthesis and is present in both soluble and membrane-bound forms of the protein. Mol. Cell. Biol. 4:2697-2704. 4. Buss, J. E., and B. M. Senon. 1984. Myristic acid, a rare fatty acid, is the lipid attached to the transforming protein of Rous sarcoma virus and its cellular homolog. J. Virol. 537-12. 5. Chow, M., J. E. Newman, D. Filman, J. M. Hogle, D. J. Rowlans, and F. Brown. 1987. Myristylation of picornavirus capsid protein VP4 and its structural significance. Nature (Lon- don) 32248246. 6. Cdlett, M. S., E. Erikson, and R. L. Erikson. 1979. Structural analysis of the avian sarcoma virus transforming protein: sites of phophorylation. J. Virol. 29770-781. 7. Courtneidge, S. A., and J. M. Bishop. 1982. Transit of pp60'" to the plasma membrane. Proc. Natl. Acad. Sci. USA 797117- 7121. 8. Courtmidge, S. A., A. D. Levinson, and J. M. Bishop. 1980. The protein encoded by the transforming gene of avian sarcoma virus (pp60"') and a homologous protein in normal cells (pp60pm'o~''r) are associated with the plasma membrane. Proc. Natl. Acad. Sci. USA 71:3783-3787. 9. Cross, F. R., E. A. Garber, D. Pelh, and H. Hanafusa. 1984. A short sequence in the p64F N terminus is required for myristylation and membrane association and for cell transfor- mation. Mol. Cell. Biol. 4:1834-1842. 10. Cross, F. R., and H. Hanahwa. 1983. Local mutagenesis of Rous sarcoma virus: the major sites of tyrosine and serine phosphor- ylation for pp6Ww are dispensable for transformation. Cell 3459747. 11. Garber, E. A., F. R. Cross, and H. Hanafusa. 1985. Processing of ~60""" to its myristylated membrane-bound form. Mol. Cell. Biol. 52781-2788. 12. Garber, E. A., and H. Hanafusa. 1987. NH,-terminal sequences of two src proteins that cause aberrant transformation. Proc. Natl. Acad. Sci. USA M80-85. 13. Gentry, L. E., K. E. Chaffin, M. Shoyab, and A. F. Purchio. 1986. Novel serine phosphorylation of pp6Oc-''' in intact cells after tumor promoter treatment. Mol. Cell. Biol. 6:735-738. 14. Goukl, K. L., J. R. Woodgett, J. A. Cooper, J. E. Buss, D. Shnlloway, and T. Hunter. 1985. Protein kinase C phosphory- lates ~~60''~ at a novel site. Cell 4284-57. 15. Graham, F. L., and A. J. van der Eb. 1973. A new technique for the assay of infectivity of human adenovirus 5 DNA. Virology 52:456-467. 16. Kidderon, D., B. L. Roberts, W. D. Richardson, and A. E. Smith. 1984. A short amino sequence able to specify nuclear location. Cell 39499-509. 17. Kaplan, J. M., G. Mardon, J. M. Bishop, and H. E. Varmus. 1988. The first seven amino acids encoded by the v-src once- gene act as a myristylation signal: lysine-7 is a critical determi- nant. Mol. Cell. Biol. 82435-2441. 18. Karess, R. E., and H. Hanafusa. 1981. Viral and cellular sic genes contribute to the structure of recovered avian sarcoma virus transforming protein. Cell 24.155-164. 19. Krueger, J. G., E. A. Garber, and A. R. Goldberg. 1983. Subcellular localization of ppaolK in RSV transformed cells. Cum. Top. Microbiol. Immunol. 107:51-124. 20. Krueger, J. G., E. A. Garber, A. R. Goldberg, .od R. HanafWa. 1982. Changes in amino-terminal sequences of ppW" lead to decreased membrane association and decreased in vivo tumor- igenicity. Cell m889-896. 21. Krueger, J. G., E. Wang, and A. R. Goldberg. 1980. Evidence that the STC gene product of Rous sarcoma virus is membrane associated. Virology 101:2W. 22. Levinsoo, A. D., S. A. Courtncldge, and J. M. Bishop. 1981. Structural and functional domains of the Rous sarcoma virus transforming protein (pp6(y'c). Roc. Natl. Acad. Sci. USA 78: 1624-1628. 23. Levy, J. B., T, Dorai. LA. Wang, and J. S. Brugge. 1987. The structurally distinct form of pp60c"'c detected in neuronal cells is encoded by a unique c-src mRNA. Mol. Cell. Biol. 2 4142-4145. Vor. 10,1990 MEMBRANE ATTACHMENT OF src PROTEIN 1009 24. Lin, C. R., W. S. Chen, C. S. Latpr, C. D. Carpenter, G. N. GUI, R. M. Evans, and M. G. Rasenteld. 1986. Protein kinase C phosphorylation at Thr 654 of the unoccupied EGF receptor and EGF binding regulate functional receptor loss by independent mechanisms. Cell 44:839-845. 25. Lipsich, L. A., A. J. Lewis, and J. S. Brugge. 1983. Isolation of monoclonal antibodies that recognize the transforming proteins of avian sarcoma viruses. J. Virol. 48:352-360. 26. Lonberg, N., and W. Gilbert. 1983. Primary structure of chicken pyruvate kinase mRNA. Roc. Natl. Acad. Sci. USA 803661- 3665. 27. Mantas, P. F., M. Aubry, C. G. Shores, L. Frame, and K. H. Plenninger. 1988. c-src gene product in developing rat brain is enriched in nerve growth cone membranes. Proc. Natl. Acad. Sci. USA 85.5001-5006. 28. Martiaez, R., B. Mathey-Prevot, A. Bernards, and D. Baltimore. 1987. Neuronal pp605" contains a six-amino acid insertion relative to its non-neuronal counterpart. Science 232411-415. 29. Olson, E. N., D. A. Towkr, and L. Glnsor. 1985. Specificity of fatty acid acylation of cellular proteins. J. Biol. Chem. 260: 3784-3190. 30. Pprsons, S. J., and C. E. Cnutz. 1986. @-'"activity detected in the chromaffin granule membrane. Biochem. Biophys. Res. Commun. 134:736-743. 31. Pellman, D., E. A. Garber, F. R. Cross, and H. Hpnpluss. 1985. Fine structural mapping of a critical NHrterminal region of pW". Roc. Natl. Acad. Sci. USA 82:1623-1628. 32. Pellmpn, D., E. A. Garber, F. R. Cross, and H. Hanafusa. 1985. An N-terminal peptide from ~60'~ can direct myristylation and plasma membrane localization when fused to heterologous proteins. Nature (London) 314374377. 33. Rendu, F., M. Lebret, S. Danieh, R. Fagad, S. Levy- Tdedpao, and S. Fkher. 1989. High pp60-src level in human platelet dense bodies. Blood 731545-1551. 34. Re&, M. D. 1988. Reconstitution of the Rous sarcoma virus transforming protein ~p60"-'~ into phospholipid vesicles. Mol. Cell. Biol. 81896-1905. 35. Resh, M. D. 1989. Specific and saturable binding of pv to plasma membranes: evidence for a myristyl-src receptor. Cell 36. Resh, M. D., and R. L. Erikson. 1985. Highly specific antibody to Rous sarcoma virus src gene product recognizes a novel population of pp6Oy~"" and p@"" molecules. J. Cell Biol. 100:409-418. 37. Rohmhneider, L. R. 1980. Adhesion plaques of Rous sarcoma virus-transformed cells contain the src gene product. Proc. Natl. Acad. Sci. USA 723514-3518. 38. Rudd, C. E., J. M. TrevUlyan, J. D. Dasgupta, L. L. Won& and S. F. Schlossmpn. 1988. The CD4 receptor is complexed in detergent lysates to a protein-tyrosine kinase (pp58) from hu- man T lymphocytes. Roc. Natl. Acad. Sci. USA 85:519&5194. 39. schultz, A. M., L. E. Henderson, S. Oroszlnn, E. A. Garber, and H. Hanafusa. 1985. Amino terminal myristylation of the protein kinase p6Osrc, a retroviral transforming protein. Science 227: 427-429. 40. Schwartz, D. E., R. Tizard, and W. Gilbert. 1983. Nucleotide sequence of Rous sarcoma virus. Cell 329353469. 41. Sobuc, K., and K. Kanda. 1988. Localization of ~p60f-''~ in growth cone of PC12 cells. Biochem. Biophys. Res. Commun. 157:1383-1390. 42. VeUlette, A., M. A. Bookman, E. M. Hod, and J. B. Bokn. 1988. The CD4 and CD8 cell surface antigens are associated with the internal membrane tyrosine-protein kinase p56"'. Cell 43. Veillette, A., M. A. Bookman, E. M. Hod, L. E. Smdson, and J. B. Wen. 1989. Signal transduction through the CD4 receptor involves the activation of the internal membrane tyrosine- protein kinase ~56'''. Nature (London) 338257-259. 44. Willingham, M. C., G. Jay, and I. Pastan. 1979. Localization of the ASV src gene product to the plasma membrane of trans- formed cells by electron microscopic immunocytochemistry. Cell 18125-134. 58:281-286. 55:301-308.