The v-sis Protein Retains Biological Activity as a Type II Membrane Protein When Anchored by Various Signal-Anchor Domains, Including the Hydrophobic Domain of the Bovine Papilloma Virus E50ncoprotein

Membrane-anchored forms of the v-sis oncoprotein have been previously described which are oriented as type I transmembrane proteins and which efficiently induce autocrine transformation. Several examples of naturally occurring membrane-anchored growth factors have been identified, but all exhibit a type I orientation. In this work, we wished to construct and characterize membrane-anchored growth factors with a type II orientation. These experiments were designed to determine whether type II membrane-anchored growth factors would in fact exhibit biological activity. Additionally, we wished to determine whether the hydrophobic domain of the E5 oncoprotein of bovine papilloma virus (BPV) can function as a signal-anchor domain to direct type II membrane insertion. Type II derivatives of the v-sis oncoprotein were constructed, with the NH2 terminus intracellular and the COOH terminus extracellular, by substituting the NH2 terminal signal sequence with the signal-anchor domain of a known type II membrane protein. The signal-anchor domains of neuraminidase (NA), asialoglycoprotein receptor (ASGPR) and transferrin receptor (TR) all yielded biologically active type II derivatives of the v-sis oncoprotein. Although transforming all of the type II signal~anchor-sis proteins exhibited a very short half-life. The short half-life exhibited by the signal~anchor-sis constructs suggests that, in some cases, cellular transformation may result from the synthesis of growth factors so labile that they activate undeteetable autocrine loops. The E5 oncoprotein encoded by BPV exhibits amino acid sequence similarity with PDGF, activates the PDGF B-receptor, and thus resembles a miniature membrane-anchored growth factor with a putative type II orientation. The hydrophobic domain of the E5 oncoprotein, when substituted in place of the signal sequence of v-sis, was indistinguishable compared with the signal-anchor domains of NA, TR, and ASGPR, demonstrating its ability to function as a signal-anchor domain. NIH 31"3 cells transformed by the signal/anchor-sis constructs exhibited morphological reversion upon treatment with suramin, indicating a requirement for ligand/receptor interactions in a suramin-sensitive compartment, most likely the cell surface. In contrast, NIH 31"3 cells transformed by the E5 oncoprotein did not exhibit morphological reversion in response to suramin. T HE v-sis oncoprotein, closely related to the B chain of PDGE is synthesized with an NH2-terminal signal sequence to direct its translocation across the membrane of the RER and into the secretory pathway (Hannink and Donoghue, 1984). The v-sis oncoprotein and PDGF are usually viewed as examples of secreted growth factors which are released from cellular membranes prior to binding and activation of receptors. This view remains fundamentally unchanged by the recent observation that some forms of PDGF contain a basic amino acid sequence leading to association with the extracellular matrix (LaRochelle et al., 1991; Raines and Ross, 1992), which may restrict their ability to diffuse to other cells. PDGF belongs to a larger group of growth factors defined, in part, by homology among the receptors which they activate. The PDGF receptors, including the o~ receptor (Matsui et al., 1989) and the B receptor (Yarden et al., 1986), exhibit a "split-kinase " domain and define a family of receptors which includes: the stem cell factor (SCF) ~ receptor or c-k# (Qiu et al., 1988); the colony stimulating factor-1 (CSF1. Abbreviations used in this paper: ASGPR, asialoglycoprotein receptor; BPV, bovine papilloma virus; CSILI, colony stimulating factor-l; MLV, Moloney murine leukemia virus; NA, neuraminidase; RSV, l~us sarcoma virus; SCF, stem cell factor; TGF*a, transforming growth fi~ctor-~; TR, transferfin receptor; VSV-G, vesicular stomatitis virus glycoprotein. © The Rockefeller University Press, 0021-9525/93/11/549/12 $2.00 The Journal of Cell Biology, Volume 123, Number 3, November 1993 549-560 549 on O cber 9, 2017 jcb.rress.org D ow nladed fom 1) receptor or c-f ms (Coussens et al., 1986); and the vascular endothelial growth factor (VEGF) receptor or c-fit (De "Cries et al., 1992; Shibuya et al., 1990). Several of the growth factors for the "split-kinase" receptors are synthesized as membrane-anchored precursors. For instance, the growth factor CSF-1 is synthesized as two different membraneanchored precursors which are released by proteolysis (Kawasaki et al., 1985; Rettenmier and Roussel, 1988; Wong et al., 1987). The ligand of the c-kit receptor, referred to variously as SCE mast cell growth factor, or steel factor, is also synthesized as a membrane-anchored precursor which undergoes rapid proteolysis to release mature SCF (Anderson et al., 1990; Huang et al., 1990; Martin et al., 1990). In these cases, the membrane-anchored precursors are type I proteins in which the NH2 terminus is topologically extracellnlar and the COOH terminus remains within the cytoplasm ("N-out, C-in"). This orientation requires the presence of a conventional signal sequence located near the NH2 terminus of the nascent polypeptide, which is removed during translocation across the membrane, coupled with a stop-transfer domain or membrane anchor located near the COOH terminus. Although naturally occurring type I membrane-anchored forms of PDGF have not been identified, our laboratory has designed and described such constructs previously (Hannink and Donoghue, 1986a; Lee and Donoghue, 1991, 1992), using the membrane anchor of the vesicular stomatitis virus glycoprotein (VSV-G) to provide for membrane anchoring (Rose and Gallione, 1981; Adams and Rose, 1985; Guan et al., 1985). These prior studies demonstrated that membrane-anchored v-sis-G can still induce autocrine transformation (Hannink and Donoghne, 1980a), although its ability to induce PDGF receptor autophosphorylation is significantly reduced (Lee and Donoghue, 1991). Petti and DiMaio (1992) demonstrated that the E5 oncoprotein, encoded by bovine papiUomavirus (BPV), can be recovered in a complex with activated PDGF/3 receptors. The E5 oncoprotein is unusual due to its small size, only 44 amino acids, with a very hydrophobic NH2-terminal region and a hydrophilic COOH-terminal region (Horwitz et al., 1988, 1989). Petti et al. (1991) first noted slight amino acid similarity between E5 and PDGF-B. The region of similarity includes the last two Cys residues of the minimal transforming region of v-sis, previously identified by deletion analysis (Giese et al., 1987; Sauer and Donoghue, 1988). In addition, the tripeptide F~4SL~49V tS° in a putative receptor activating domain of v-sis (LaRochelle et al., 1989), also occurs in the E5 oncoprotein (Maher et al., 1993). Previous studies localized the E5 oncoprotein predominantly to Golgi membranes and/or the cell surface, with the COOH terminus topologically extracellular (Burkhardt et al., 1989). The presence of an NH2-terminal hydrophobic domain, together with a COOH-terminal hydrophilic domain exhibiting amino acid similarity to PDGF, suggests that E5 may function as a "miniature" membrane-anchored version of PDGE By this model, E5 would exhibit a type II orientation, i.e., "N-in, C-out S allowing the COOH-terminal hydrophilic domain to be extracellular and available for PDGF receptor activation. Several type 1I membrane-anchored proteins have been extensively characterized, including neuraminidase (NA) (Fields et al., 1982; Sivasubramanian and Nayak, 1987; Brown et al., 1988; Nayak and Jabbar, 1989; Kundu et al., 1991), asialoglycoprotein receptor (ASGPR) (Spiess et al., 1985; Spiess and Lodish, 1985, 1986), and transferrin receptor (TR) (Schneider et al., 1984; Zerial et al., 1986; Kundu et al., 1991). These proteins possess a %ignal-anchor" sequence located near the NH2 terminus. The signal-anchor sequence provides the dual function of a signal sequence, directing translocation across the membrane of the rough ER, as well as a membrane anchor, resulting in the topology "N-in, C-out" (Hartmann et al., 1989; High et al., 1991). In this study, we sought to determine whether the v-sis oncoprotein retains its biological activity when membrane anchored as a type II protein. To accomplish this, the DNA sequence encoding the signal sequence of v-sis was replaced with a sequence encoding the signal-anchor domain of a known type II membrane protein. We also wished to examine whether the hydrophobic domain of the E5 oncoprotein, when substituted in place of the normal signal sequence of the v-sis oncoprotein, could in fact function as a signalanchor domain. Our results indicate that the signal-anchor domains of NA, ASGPR and TR, as well as the hydrophobic domain of E5, all yield biologically active type II membrane-anchored derivatives of the v-sis oncoprotein. Although transforming, all of the type II signal~anchor-sis derivatives exhibited rapid turnover. These results indicate that there is, in principle, no reason to preclude the existence of naturally occurring membrane-anchored growth factors exhibiting a type II orientation. Materials and Methods Construction of Plasmids Encoding Signal~Anchor-Sis Fusions The signal~anchor-sis constructs were derived by substitution of the v-sis signal sequence by a heterologous signal-anchor domain. The parental plasmid contained a mutant v-sis gene in which the dibasic proteolytic processing site Lysll°-Arg m was previously mutagenized to Asnll°-Ser TM (Haw nink and Donoghue, 1986b). Cleavage at this dibasic processing site occurs as a late event in the secretory pathway, probably between the trans-Golgi compartment and the plasma membrane (Robbins et al., 1985; Lokeshwar et al., 1990; Lee and Donoghue, 1992). This cleavage removes the propeptide sequence and generates the NH2 terminus of the mature PDGE To prevent proteolytic separation of the growth factor domain from the signalanchor domain in the constructs reported here, it was essential to include the LysH°--*Asn, ArglH-'~Ser mutations in all clones. The parental plasmid, designated pRSV-sisNn°S TM, contains the mutant v-sis gene as a HindIII-ClaI restriction fragment in a standard plasmid vector under control of the Rous sarcoma virus (RSV) promoter. The original pRSV-A~ plasmid (obtained from S. Gould and S. Subramani, University of California, San Diego, La Jolla, CA) contains the RSV long terminal repeat to drive transcription of inserted genes followed by the SV-40 poly A addition site. The DNA sequence encoding the signal sequence of v-sis is easily removed from the parental plasmid pRSV-sisNHOSm as a HindIII-SstI restriction fragment, where the SstI site corresponds to nucleotide 3828 in the sequence of simian sarcoma virus (Devare et al., 1983). Removal of this restriction fragment removes the codoas for amino acids 1-59 of wild type v-sis protein. Synthetic restriction fragments encoding heterologous signal-anchor domains were prepared using two long complementary oligonucleotides, designed to produce HindIII and SstI overhangs when annealed. The oligonucleotides ranged in length from 83-124 bases, and were prepared using a DNA synthesizer (381A; Applied Biosystems, Foster City, CA) with customized coupling times, reagent delivery times, and column configurations to reduce cost and synthesis time. Approximately 25 #g of each crude oligonucleotide was applied to a 6% denaturing polyacrylamide gel, separated electrophoretically, and the band corresponding to each fulllength oligonucleotide was excised. Oligonucleotides were recovered by The Journal of Cell Biology, Volume 123, 1993 550 on O cber 9, 2017 jcb.rress.org D ow nladed fom overnight elution into 0.75 ml of elution buffer (containing 0.1% SDS, 0.5 M NHtAc, 10 mM MgAc2), and ethanol precipitation. Resuspended oligonucleotides were then ligated with 0.15 #g of the vector DNA, pRSVsis NH°sHI, which was previously cleaved with HindIH and SstI and purified by agarose gel electrophoresis. Recombinant clones were recovered by standard techniques and the sequence of the signal-anchor domains encoded by the HindHI-SstI fragments were confirmed by nucleotide sequencing. All synthetic oligonucleotides also encoded a XhoI restriction site adjacent to the HindIII site, so that the entirety of the coding region for each signal~anchor-sis construct could be swapped into other vectors as either XhoI-ClaI or as HindIII-ClaI restriction fragments. As an example, the NA-sis construct required the synthesis of two long oligonucleotides, designated D319 and D320, representing the sense strand and antisense strand, respectively. The sequence of the sense strand D319 oligonucleotide is: 5' AGCTTCTCGAGACC. AT(3. AAT. CCA. AAT. CAG. AAA. ATA. ATA. ACC. ATT. GGA. TCA. ATC. TGT. CTG. GTA. GTC. GGA. CTA. ATT. AGC. CTA. ATA. CTG. CAG. ATA. GGG. AAT. ATA. ATC. TCA. ATA. TGG. ATT. AGC. GAG. CT y. This oligonucleotide encodes the amino acid sequence MNPNQKIITIGSICLVVGLISLILQIGNIISIWISEL. The amino acids encoded by oligonucleotides D319/D320 represent amino acids 1-35 of the NA protein of human influenza virus (Fields et al., 1982), including the signal-anchor domain defined as amino acids 7-35 (Fields et al., 1982; Sivasubramanian and Nayak, 1987; Brown et al., 1988; Nayak and Jabbar, 1989; Kundu et al., 1991). It should be noted that the last two amino acids encoded by the D319/D320, as well as for the other constructs described below, correspond to E 5s and L 59 of wild type v-sis protein encoded at the unique SstI restriction site. In all constructs, the sequence surrounding the initiation codon was designed to provide for optimal initiation of translation (Kozak, 1986). The E5-sis construct was similarly designed using a pair of oligonucleotides designated D341/D342, which encode the amino acid sequence MPNLWFLLFLGLVAAMQLLLLLFLLLFFLVEL. The first 30 residues correspond to amino acids 1-30 of the BPV E5 oncoprotein (DiMaio et al., 1986). The TR-sis construct was designed using a pair of oligonucleotides designated D383/D384. These oligonucleotides encode the amino acid sequence MKRCSGSICYGTIAVIVFFLIGFMIGYLGYCKEL. Amino acids 2-32 in this sequence correspond to amino acids 60-90 of the human TR protein (Schneider et al., 1984), which includes the signal-anchor domain located at amino acids 65-88. The first Met residue encoded by D383/D384 was added to provide for initiation of translation. The ASGPR-sis construct was designed using a pair of oligonucleotides designated D385/D386. These oligonucleotides encode the amino acid sequence MPRLLLLSLGLSLLLLVVVC'VIGSEL. Amino acids 2-24 in this sequence correspond to amino acids 39-61 of the human ASGPR HI protein (Spiess et al., 1985), which includes the signal-anchor domain located at amino acids 41-59. AS for the preceding construct, the first Met residue encoded by D385/D386 was added to provide for translational initiation. Preparatory to DNA transfections into NIH 31"3 cells, DNA fragments encoding NA-sis, E5-sis, TR-sis, and ASGPR-sis were subcloned as XhoI-ClaI restriction fragments into the murine leukemia virus (MLV) expression vector pDM85, which was derived from the previously described retroviral vector pDD102 (Bold and Donoghue, 1985). In Vitro Transcription and Translation of Signal~Anchor-Sis Constructs DNA fragments encoding the signal~anchor-sis were also subcloned from the constructs described above into a vector derived from pSP64(polyA) (Promega Biotec, Madison, WI), designated pDIM31, which contains an SP6 promoter for in vitro transcription. As controls, other sis-related genes were subcloned into the SP6-promoter vector, including wild type v-sis and v-sis239-G (Hannink and Donoghue, 1986a) which will be designated simply as v-sis-G throughout this work. The 5'-capped and polyadenylated RNAs were transcribed in vitro as described (Melton, 1987). RNAs were analyzed by gel electrophoresis and subjected to in vitro translation in rabbit reticulocyte lysates containing 50 #Ci [35S]Cys (1,000 Ci/mmol). Translation products were resolved by SDS-PAGE using 15 % polyacrylamide in the separating gel, and detected by autoradiography of the dried gel. Cell Culture, Focus Assays, and Transient Expression Assays NIH 3T3 cells and CV-1 cells were cultured at 37°C in DME containing 10% calf serum. For focus assays, NIH 3T3 cells were transfected with the signal~anchor-sis constructs described above, under MLV retroviral promoter control, together with a replication competent helper provirus, pZAP (Hoffman et al., 1982). DNA transfections were carried out using a modified calcium phosphate transfection protocol (Chen and Okayama, 1987), as described previously (Maher et al., 1993). For some experiments, plates of transfected cells were allowed to overgrow to select for transformed cells by splitting cells weekly at a density of 1:6. For transient expression assays of protein expression, transfections into monkey CV-1 cells were carried out as described above using 12 #g DNA for each signal~anchor-sis construct under RSV promoter control. Metabolic Labeling and Immunoprecipitations NIH 3T3 or CV-I cells expressing the signal~anchor-sis proteins were used as described in the text for immunoprecipitation of radiolabeled protein. Before labeling, cells were rinsed and incubated for 15 rain in MEM lacking Cys and Met. For pulse-chase analyses, cells were labeled with 100 #Ci/ml each [35S]Cys and [3SS]Met for 30 min, rinsed twice with TBS, and chased for 10 min, 30 min, or 2 h in fresh DME. Labeled cells were subsequently lysed in 1.0 ml RLPA buffer (10 mM sodium phosphate [pH 7.0], 150 mM NaCI, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS, 1% Aprotinin) and clarified lysates were prepared. Immune complexes were formed as previously described (Hannink and Donoghue, 1986a), using a rabbit antiserum directed against bacterially synthesized v-sis protein generously provided by Ray Sweet and Keith Deen (Smith, Kline and French, King of Prussia, PA). Immune complexes were collected using protein A Sepharose (Sigma Immunochemicals, St. Louis, MO). Samples were run on a 12.5% SDS-POlyacrylamide gel and processed for fluorography to visualize proteins. Indirect Immunofluorescence To detect intracellular v-sis fusion proteins, cells were fixed in 3 % paraformaldehyde/PBS for 10 rain, followed by permeabllization in 1% Triton/PBS for 5 rain. Cells were then incubated with a rabbit antiserum directed against the v-sis protein, followed by a rhndamine-conjugated goat antirabbit antibody. To detect cell surface v-sis fusion proteins, cells were fixed with paraformaldehyde and incubated with antibodies without permeabilization, as described previously (Lee and Donoghue, 1992; Hannink and