Cell, Vol. 70, 93-104,
July 10, 1992, Copyright
0 1992 by Cell Press
A Novel Transforming Protein (SHC) with an SH2 Domain Is Implicated in Mitogenic Signal Transduction Giuliana Pelicci,’ Luisa Lanfrancone,” Francesco Grignani, l Jane McGlade,t Federica Cavallo,* Guido Forni,S lldo Nicoletti,’ Faust0 Grignani,’ Tony Paw&t and Pier Giuseppe Pelicci’ l lstituto Clinica Medica I Policlinico Monteluce University of Perugia 06100 Perugia Italy tDivision of Molecular and Developmental Biology Samuel Lunenfeld Research Institute Mount Sinai Hospital Toronto, Ontario M5G 1X5 Canada Slstituto di Microbiologia University of Turin 10126 Turin Italy
Summary A new SH2-containing sequence, WC, was isolated by screening cDNA libraries with SH2 representative DNA probes. The WC cDNA is predicted to encode overlapping proteins of 46.6 and 51.7 kd that contain a single C-terminal SH2 domain, and an adjacent glycinelproline-rich motif with regions of homology with the al chain of collagen, but no identifiable catalytic domain. Anti-SHC antibodies recognized three proteins of 46,52, and 66 kd in a wide range of mammalian cell lines. These SHC proteins complexed with and were phospholylated by activated epidermal growth factor receptor. The physical association of SHC proteins with activated receptbrs was recreated in vitro by using a bacterially expressed SHC SH2 domain. NIH 3T3 mouse fibroblasts that constitutively overexpressed SHC acquired a transformed phenotype in culture and formed tumors in nude mice. These results suggest that the SHC gene products couple activated growth factor receptors to a signallng pathway that regulates the proliferation of mammalian ceils. Introduction Src homology 2 (SH2) domains are noncatalytic regions that are conserved among a group of cytoplasmic signaling proteins. These signaling proteins are directly regulated by receptor tyrosine kinases and control the activation of mitogenic signal transduction pathways by tyrosine kinases (reviewed by Ullrich and Schlessinger, 1990; Cantley et al., 1991; Koch et al., 1991). One function of SH2 domains was revealed during investigation of the cytoplasmic processes stimulated by the binding of growth factors, such as epidermal growth factor (EGF) and platelet-derived growth factor (PDGF), to their
receptors (EGFR and PDGFR). Ligand activation of these receptors induces dimerization, triggers receptor tyrosine kinase activity, and leads to autophosphorylation (reviewed by Ullrich and Schlessinger, 1990). In addition, the activated receptors associate with and phophorylate a number of cytoplasmic proteins that contain SH2 domains, including phospholipase C (PLC)+ (Margoliset al., 1989; Meisenhelder et al., 1989; Wahl et al., 1989; Morrison et al., 1990), ~21” GTPase activating protein (GAP) (Ellis et al., 1990; Kaplan et al., 1990; Kazlauskas et al., 1990), phosphatidylinositol (PI) 3’-kinase (Pl3K) (Kazlauskas and Cooper, 1989; Coughlin et al., 1989; Bjorge et al., 1990), and Src and Src-like tyrosine kinases (Kypta et al., 1990). These cytoplasmic proteins are very probably among the targets through which growth factor receptors transmit the message to trigger mitogenesis. In some cases, there is evidence that SHe-containing proteins do, indeed, regulate signal transduction. PLC+ cleaves phosphatidylinositol4,5-biphosphate (PIP2) into the second messengers, diacylglycerol and inositol trisphosphate, which in turn stimulate protein kinase C and increase intracellular calcium (reviewed by Whitman and Cantley, 1988). GAP, by converting Ras from the active GTP-bound form to the inactivated GDP-bound form (Trahey and McCormick, 1987), exerts negative regulation on Ras, which is located on the tyrosine kinase mitogenic signaling pathway (Mulcahy et al., 1985; Smith et al., 1986). Pl3K phosphorylates PI to form PI3-phosphate, and the reaction is stimulated by PDGF (Whitman et al., 1988; Auger et al., 1989). Evidence is accumulating that Src and Src-like kinases are also involved in growth factor-signaling pathways. Indeed, it has been shown that PDGF stimulates Src kinase activity in fibroblasts (Gould and Hunter 1988; Kypta et al., 1990). The interactions between cytoplasmic signaling proteins and growth factor receptors are mediated by specific binding of SH2 domains to tyrosine-phosphorylated regions of the receptors (Koch et al., 1991). In vitro binding studies have demonstrated that the GAP, Src, PLCql, and Pl3K SH2 domains are sufficient for binding to activated growth factor receptors (Anderson et al., 1990; Moran et al. 1990; McGlade et al., 1992). Moreover, SH2 domains apparently bind directly to specific tyrosinephosphorylated sites (Reedijk et al., 1992). As SH2 domains are important for directing proteinprotein interactions during mitogenic signal transduction, it is not surprising that some of the proteins that contain these domains can be activated as oncogenes. Notably, the v-Crk-transforming protein contains an SH2 domain and an SH3 domain (a distinct sequence frequently found alongside SH2 elements), but has no identifiable catalytic domain (Mayer et al., 1988). Despite this fact, v-Crk induces the tyrosine phosphotylation of a group of proteins that become associated with the Crk SH2 domain in v-Crktransformed cells (Mayer and Hanafusa, 1990; Matsuda et al., 1990). v-Crk, therefore, promotes the formation of cytoplasmic signaling complexes, suggesting that the SH2 domains of signaling proteins bind not only to activated
Cell 94
.
growth factor receptors, but also to specific cytoplasmic phosphotyrosine-containing proteins. The Src and GAP SH2 domains are also involved in a network of interactions with cytoplasmic phosphoproteins, which appear to be important for their biological activity (Moran et al., 1990, 1991; Koch et al., 1991, 1992). As these findings suggest that SH2 domains couple tyrosine kinases to their targets, the isolation of new SH2containing genes should help to define the mechanisms through which tyrosine kinases regulate normal cell growth and induce neoplastic transformation. In searching for such new genes, we isolated a previously unidentified gene that encodes a novel SHPcontaining protein, which has been named SHC. Evidence is presented that suggests that SHC has a part in fibroblast growth regulation and that its constitutive expression leads to a transformed cell phenotype.
Isolation of a cDNA Clone Encoding a Novel SHP-Containing Protein We searched for novel SH2 sequences by screening human cDNA libraries with a DNA probe representative of the c-fes SH2 domain, the Kpnl-Taql fragment from nucleotide 1454-1655 of the reported human c-fes cDNA sequence (Alcalay et al., 1990). As no fes-specific transcripts were detected in B-lymphoid cell lines using the Kpnl-Taql DNA probe at high stringency conditions (data not shown), a cDNA library prepared from the Burkitt lymphoma P3HRI mRNA was screened at low stringency, and a singe clone, IGFl , isolated from approximately 5 x lo5 plaques. The hGF1 insert was characterized by nucleotide sequence analysis. Translation of the GFl sequence revealed a single long open reading frame (ORF), which, upon GenBank analysis, was found to contain a single stretch of about 100 aa with sequence homology with SH2 domains (Figure 1 and Figure 2). Since lhe XGFl insert did not contain either the 3’ or the S’end of the gene, two additional human cDNA libraries (fetal brain and KG1 myeloid leukemia cells) were screened with the GFl insert as DNA probe and 13 positive clones isolated and studied by restriction enzyme analysis: A limited map of the hGF1 is shown together with the four longest cDNAs in Figure 1A. All phage inserts were subcloned into the EcoRl site of the plasmid vector pGEM-3 (Promega, Biotec) and sequenced. The entire nucleotide sequence of the two longest clones, LGFl 1 and hGF13, and the 5’ and 3’ extremities (approximately 300 bp) of all other cDNA inserts were determined. The complete 3031 nt sequence of the human SHC mRNA is shown in Figure 1 B. It corresponds to clone hGFl1 from nucleotide 1-164, to clone LGF13 from nucleotide 165-2676, and to clone LGF16 from nucleotide 2677 to the 3’ end. It contains a single 1609 bp ORF, beginning at nucleotide position 14 and extending to an in-frame TGA termintion codon at position 1502 (Figure 1 B). The termination codon is followed by a 3’ untranslated region of 1530 bp, and a poly(A) tail of 19 nt is found at the 3’ end of the sequence (Figure 1B). Two in-frame ATGs, each surrounded by a
sequence (CTGGACATGA and AAAGTCATGG), which is in good agreement with the consensus sequence for initiation of translation in eukaryotes (Kozak, 1969) are found at positions 63 (ATG63) and 216 (ATG216), respectively. The ORFs beginning with these ATGs encode putative proteins of 473 and 426 aa, respectively, with molecular masses of 51.7 and 46.6 kd. The protein(s) encoded by the SHC cDNA was demonstrated using an in vitro transcription-translation system. The SHC cDNA sequence +l-+1647 was inserted into the pGEM-3 plasmid vector and transcribed in vitro under the control of the SP6 and T7 promoters to produce sense and antisense SHC mRNA. The in vitro synthesized SHC mRNAs were then translated in a rabbit reticulocyte lysate in the presence of [%S]methionine. Translation of the SHC sense mRNA revealed two major proteins. Tl&ir apparent molecularweights were in good agreement with the molecular weights of the two predicted SHC proteins. No translation products were seen using the antisense SHC mRNA (Figure lC).ThisfindingsuggeststhattheSHCATG63and ATG216 are both used as initiation sites for translation. SHC Contains a New SH2 Domain and a Collagen-Homologous Region Analysis. of the putative SHC protein sequence by GenBank disclosed a stretch of 93 aa, from 376-471, with sequence homology with the SH2 domains of other proteins (Figure 2A). Although the degree of homology varied from 41% to 60% and was highest with the Src (60%) and Crk (57%) SH2 domains, the SH2 sequence encoded by SHC had significant DNA homology (63.3% only over a 96 bp stretch, from nucleotide position 1194-1291) with the corresponding human fes sequence (data not shown). Analysis of the putative SHC protein sequence also revealed a stretch of 145 aa, from 233-377, which was 50% homologous to human al collagen (Figure 28) (Bernard et al., 1963; Chu et al., 1964). This collagen-homologous SHC region was extremely rich in glycine (10.3%) and proline (20%), both of which are also abundant in collagen a chains (33.6% and 23.6% in the collagen al chain SHChomologous region). Fifty percent of the glycine and 65% of the proline al collagen residues are conserved in the SHC collagen homologous region. Due to its homology with SH2 and collagen, we named the new gene SHC (Src homologous and collagen). The relative position of the SH2 and collagen-homologous regions of SHC is shown in Figure 1A. Identification of SHC Proteins in Mammalian Cells To investigate the expression of SHC gene products in vivo, antibodies were raised against a bacterial glutathione S-transferase (GST)-SHC fusion protein (see Experimental Procedures). These anti-SHC antibodies specifically immunoprecipitated the 46 and 52 kd proteins translated from transcribed SHC cDNA in vitro (Figure 3A). In order to compare the in vitro translation products with cellular SHC proteins, COS-1 monkey cells, COS-1 cells transfected with the SHC GFll cDNA (pECE-GFll expression vector, see below), Rat-2 fibroblasts, and human A431 cells were metabolically labeled with [YS]methionine,
A.
Figure 1. DNA and Primary Amino Acid Sequence and In Vitro Translation of the SHC cDNA (A) Schematic representation of the SHC sequence and limited restriction enzyme map of independent SHC cDNAs from the PBHRI (SGFI), fetal brain (IFBl), and KG1 ()IGFll , 1IGF13. XGFl6) human cDNA libraries. The box represents the ORF starting from the ATG83 with the SH2 and collagen-homologous (CH) regions indicated. The ATG218 (see text) is also indicated. Numbers above the maps indicate the relative position of the 5’ and 3’ extremities of each clone with respect to the SHC cDNA sequence shown in (6). Identification and length of poly(A) tails are also shown. B, BamHI; A, Accl; E, EcoRI. (6) SHC nucleotide and deduced amino acid sequences. The translation of the SHC sequence from ATG83 to the termination codon (’ l ‘) at position 1592 is shown below the nucleotide sequence. The ATG218, the putative poly(A) addition signal AlTAAA at position 2992, and the 5’ in-frame termination codon at position 11 are underlined. Nucleotide and amino acid positions are indicated on the left of each lane. The collagen-homologous region is underlined, and the SH2 domain is double underlined. Circles indicate five putative Ser/ Thr phosphorylation sites (Woodgett et al., 1986; Hanks et al., 1988) and small vertical bars a potential tyrosine phosphorylation site (Cooper et al., 1984). (C) In vitro translation of the SHC cDNA. The RNA corresponding to the SHC sequence from +I-+1647 (sense transcript; lane ST) or +1647-+l (antisense transcript; lane AT) was synthesized in vitro and translated in the presence of [%]methionine. The labeled proteins were separated by SDS-PAGE and visualized by fluorography. The size of the protein markers (MWM) is given on the right.
A.
Figure 2. The SHC SH2 and Collagen-Homologous Regions
UK (378)rKGKL*..rrE....rr..LLq.,.... ~L~ltTt.POmvLT.~LO...C..q
(A) Alignment of the SH2 domain of the SHC sequence with other SH2 sequences. Sequences were aligned by eye after preliminary GenBank screening. Gaps, indicated by dots, have been introduced to improve alignment of the sequences. Capitals indicate residues invariant or conserved among six or more SH2 sequences. Amino acid substitutions within each group: A, G, P, S, and T; L, I, V, and M; D, E, N. and Q; K, R, and H; F, Y, and W; and C are defined as conservative (Schwartz and Dayoff, 1979). Thenumbersin parentheses are the first (left) and last (right) amino acid residues in each sequence. The amino acid sequence of the following proteins is presented: human Src (Anderson et al., 1985); human abl (Shtivelman et al., 1986); human fes (Alcalay et al., 1990); bovine PLC-~1 (Stahl et al., 1988); bovine GAP (Vogel et al., 1988); avian v-Crk (Mayer et al., 1988); human Pl3K (Skolnik et al., 1991); avian tensin (Davis et al., 1991). Asterisks indiate that the 18 and 15 amino acids of crk and tensin, respectively, are omitted. (8) Comparison of the SHC collagen-related sequence with al collagen. Vertical lines, identical residues; colons, conserved residues.
WC UL FE6 PLC Y PLC C G&PY G4P c
(151) UffWll..RrE . . . . &rlLlnn .~...QFL~~lt.kmVcLN~6l.n c127, [email protected] . . . . PEy.puqi.u . . . ..aFLIIILLL.m.r.slnLrYa..6..p w606Q)wUIP..naE...VAE..LLvflS . . . . . . . . SbLIIILI qp.kQMW.Lnd...O..l WOO) Yls(L~.DprhiYr.LLtyciet~L~Tf vWtLSfVm G. .k WB).WWLT..M . . . .M.llbwprO . . . . . WLIlltm l .RWalSfrr . . . . G..k c1m VwKLd..nti....AE.rLrqn.*s . . . ..aYLInEwr.CYwLsf.L~.q..Tn.v (348) WnwlS..KqE . . . . lyn.LLatvg.al . . ..cYLmt PfnY8LYf.J.tUnlqr 0.48) rrruQu..RQ . . . . Av.sLLqlgh . . . . ..EY~L~I.PmFvLw..~...*8w zk Y (333, HlrpdlS..Rall . . ..H*.kLPdt~.D.....6YFLIllOnb.kPmtLTLr.k...~.nl Gel c wz4z1) Hvlun..w . . . .1DI.LLrgkr.D . . . ..MfLIIILmk q.acvYw.w...dE..e Tmuin (573) WU~IB..LO....AII LLkdrap . . . . . . wIImmhsfr.wgLMlVu.~w
2 4u IfI PLE I PLC c $5: CRll 0111 Y am1 c Trnh
B.
pLI.IL.lV..q*Ov..H..~Dkr.F,~LI~.VIiuPu.Le..Lqrpkr (471) WYXlrkl..DwC~.lT..8rtq..F~W,,Y.akh..Ad...~...LchRLtt.w.P (249) wYnl.ntd..e..LyJ.maKu.FnTLN!LulHI.~t..v...dE...LlttLhYv.aP (218) pF*ril~l..k..Lyr~..E.~..~~lPlLl~~l~tq.~.lt.k......K.~.. (344) VQ*llkv.~.Otp..kfflMl~~~ltll.qq..WCLrcr*f~.l..RLIH.P ~36) IklcN...~..oqMI1~.~...~~~l~.~~..y..k..n..U~l.. (756) Wf~l..ir..~I~....rr.F~l~.&h.Wcllk~......KLL’IMC (271, . ..fKl.cpcphqh..4....ry.Ml~lldn.rkql n..S.~y...dkePV.P (439) YIIIIIY*I~.2.rr~...E(“)l~~fY.klh.Y Ldt.lt......Li.PV.. (334) ..LiKl..fhrD..~fM PIt...IMln#Y .rmLlr(*pL...LdvulyH.. 01) V.KhNlnktat..a ~fr.PymI..raLkcLvllfY.*t~L~...bnltL~A (720) ,rll9tsr...outlk~.~.TCL&L~.~(rpUlp . . . . . . . . KL...VIP..(Yl)
Cdl 9s
Figure 3. Identification Anti-SHC Antibodies
of SHC Proteins
Using
(A) Proteins translated in vitro from the SHC cDNA (in vitro transl.) and cell lysates from [Yfjmethionine-labeled COS-1 (COS-1) or COS-1 cells transfected with pECE-GFll (COS-1 + SHC), Rat-2 fibroblasts (R2), or human epithelial cells (A431) were immunoprecipitated with either a nonspecific antibody (RaMlg) or 2 pg of affinity-purified anti-SHC antibody (aSHC). lmmunoprecipitates were resolved by SDS-PAGE, and gels exposed to X-ray film for 3 hr at -7OOC. The SHC proteins (pQQ, ~52. p4Q) are arrowed. Molecular weight markers (MWM) were run in the same gel, and 91corresponding molecular weights (x itY) are indicated to their left. m- (8) Untransfected COSI cells (-), or COS-1 tz5i.A.. --; 4% cells transfected with pECEGFl1 (+GFil). pECEGFl3 (ffiFl3), or pECEG$l ITTG (ffiF1 ITTG) were analyzed by immunoblotting with anti-SHC antibodies. (C) lmmunoblots of different rodent and human cell lines. Whole cell lysates were prepared from rat fibroblast (RI hER), dog epithelial cells (TRMP), and the following human cells: the AML-1, KGI, and HL-60 myeloid cell lines; the U937 monocytic cell line; the Jurkat, PEER, and CEM T cell lines; the CB33, PBHRI, and Namalwa (NAM.) B cell lines; the HeLa cervical cancer cell line; melanoma cells; the CALUI lung cancer cell line; the At72 glioblastoma cell line, and cells from primary cultures of normal fibroblasts (FIBROBL.) and mesothelium (MESOTH.). All cell lines were obtained from American Type Cell Culture. Fifty micrograms of total protein was separated by 10% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with affinity-purified anti-SHC antibodies.
lysed, and immunoprecipitated with anti-SHC antibodies. All cell types contain three proteins of 46, 52, and 66 kd (~46, ~52, and ~66) which were specifically precipitated by the anti-SHC antibodies (Figure 3A). Of these, p46 and ~52 comigrated with the SHC cDNA translation products. Tryptic peptide maps of the [%S]methionine-labeled human p46 and ~52 in vitro translation products were indistinguishable from maps of the comigrating rat cellular proteins precipitated with anti-SHC antibodies (data not shown), consistent with the notion that the immunoreactive p46 and ~52 cellular proteins are indeed encoded by the SHC gene. To confirm its coding potential, the SHC GFl 1 cDNA was inserted into a mammalian expression vector containing an SV40 origin (pECE-GFl 1) and transfected into COS-1 monkey cells. Although these cells express endogenous SHC proteins, astriking increase in the levelsof p46 and ~52 polypeptides recognized by anti-SHC antibodies was‘evident in the transfected cells (Figures 3A and 38). No elevation in the level of p66 was observed. The possibility that p46 and ~52 result from the alternative use of two putative translation initiation sites (ATG63 and ATG216, see Figure 1), was tested by overexpressing cDNAs that contain both or one of the two ATGs: the GFl 1 cDNA (Figure 1A) contains both ATG63 and ATG216, GFl 1lTG only ATG83, and GF13 only ATG218 (Figure 1A). The GFl 1lTG cDNA was obtained by mutagenizing the ATG218 of the GFl 1 cDNA (ATG-lTG) (see Experimental Procedures). Whereas levels of both p46 and p52 were increased in GFl-transfected COS cells, single polypeptides that apparently comigrated with the p46 or the ~52 were detected in GF13 or GFl llTG-transfected COS cells (Figure 38). These results suggest that the ATG83 and ATG218 can be used in vivo as initiation sites for
translation and that the p46 and ~52 polypeptides are very likely produced by alternative usage of translation initiation sites. The extent of SHC’protein expression in different cells was assessed by using affinity-purified anti-SHC antibodies for immunoblotting of lysates from a variety of cell types. Human tumor cell lines from melanoma, cervical cancer, lung cancer, and glioblastoma and normal human fibroblasts and mesothelial cells all contained three polypeptides of 46 kd, 52 kd, and 66 kd 61148, ~52, and ~66) that were recognized by anti-SHC antibodies (Figure 3C). As immunoreactive p48, ~52, and p86 SHC polypeptdes were also detected in a dog epithelial cell line and in a rat fibroblast line (Figure 3C), SHC proteins appear to be well consenred between mammalian species and between cell types of different embryological origin. Indeed, the p46 and the ~52 SHC products were present in every cell line examined. The larger p66 SHC protein was not detected in any human hematopoietic cell lines tested (Figure SC). The p46, ~52, and p66 polypeptides were not detected when an irrelevant antibody or pre-immune serum were used in immunoblots (Figure 3A and data not shown). SHC Proteins Are Sound by Activated EGF Receptors and Phosphorylated in Response to Growth Factor Stimulation Since SHC proteins have an SH2 domain, we considered the possibility that they interact with and are substrates for receptortyrosine kinases. To test this notion, we examined the fate of SHC proteins in cells stimulated with EGF. Rat-l cells engineered to overexpress the human EGFR (Rl hER) were stimulated with EGF for 5 min, lysed, and immunoprecipitated with anti-SHC antibodies. The antiSHC immunoprecipitates were then immunoblotted with
SHC: a Novel 97
SHP-Transforming
A.
Sequence
B. EGF -+
C. EGF
-+
EGF -++
+
-II
a P-Y
L
Figure 4. SHC Proteins Bind to Activated Receptor and Are Phosphorylated upon Stimulation
(I EGFR
I
a SHC
+ EGF
(A) Rl hER cells were serum starved for 48 hr, and unstimulated cells (-) or cells stimulated with 80 nM EGF for 5 min (+) were lysed and immunoprecipitated using 2 pg of affinity-purified anti-SHC antibody (aSHC). Immunoprecipitates were immunoblotted with either anti-phosphotyrosine antibodies (a P-Y) or anti-EGFR (a EGFR). (B) %rUt’n-Starved RI hER cells were labeled with 1 mCi/ml [“Pjorthophosphate for 3 hr and unstimulated (-) or cells stimulated with 80 nM EGF for 15 min (+) were lysed and immunoprecipitated with affinity-purified anti-SHC antibodies (upper panel). lmmunoprecipitates Phosphoamino acid analysis (lower panel) was performed (+) cells. Positionsof phosphoserine (S), phosphothreonine
were resolved by SDS-PAGE and transferred to nitrocellulose and autoradiographed. on the combined SHC proteins (~88. ~52, p48) from unstimulated (-) and EGF-stimulated (T), and phosphotyrosine (Y) markers are indicated. (C) Immobilized GST (GST) or fusion protein containing the SHC SH2 domain (GST-SH2) EGF-stimulated (+) A431 cells. Protein complexes were resolved by SDS-PAGE, transferred antibody.
antibodies to phosphotyrosine. In the absence of EGF stimulation, SHC proteins isolated from serum-starved cells contained low but detectable levels of phosphotyrosine. Stimulation with EGF induced a marked increase in the tyrosine phosphorylation of ~46, ~52, and p66 (Figure 4A). In addition, anti-SHC immunoprecipitates from EGFstimulated cells contained a novel phosphotyrosine-containing protein of 175 kd, corresponding to the apparent size of the EGFR. lmmunoblotting of the same samples with anti-EGFR antibodies confirmed that the EGFR coprecipitated with SHC proteins following EGF stimulation (Figure 4A). The same result was obtained when EGFstimulated ceils were first immunoprecipitated with antiEGFR antibodies and then blotted with anti-SHC antibodies, in which case all three SHC proteins were observed to coprecipitate with the EGFR (data not shown). The phosphorylation of SHC proteins was analyzed directly by metabolic labeling of Rl hER cells with [32P]orthophosphate (Pi). The SHC proteins were phosphorylated in serumstarved cells, and this phosphorylation was increased upon EGF stimulation (Figure 48). Phosphoamino acid analysis revealed that the SHC proteins from serumstarved cells were phosphorylated primarily on serine and to a low level on tyrosine. EGF stimulation induced a marked increase in the level of phosphotyrosine (Figure 46). Taken together, these results indicate that SHC proteins bind to activated EGF receptors and are phosphorylated in response to EGF stimulation. The SHC SH2 Domain Binds to EGF Receptors In Vitro SH2-containing proteins associate with ligand-activated growth factor receptors and other tyrosine phosphorylated intracellular proteins in vivo. In vitro binding experiments suggest that these associations are mediated by the direct interaction of SH2 domains with tyrosine phosphorylated sites on SHP-binding proteins (reviewed in Koch et al., 1991). To determine whether the SHC SH2 domain can
EGF EGF
was incubated to nitrocellulose,
with lysates from serum-starved (-) or and probed with anti-phosphotyrosine
stably associate with phosphotyrosine-containing proteins in vitro, the SHC SH2 region was expressd in bacteria as a GST fusion protein (GST-SH2). The GST-SH2 polypeptide was isolated from bacterial lysates by chromatography on glutathione sepharose. Immobilized bacterial proteins (GST-SHP and control GST) were then incubated with lysates from A431 cells, which overexpress the human EGFR, stimulated with EGF. SH2 complexes were recovered, washed extensively, and analyzed by immunoblotting with antibodies to phosphotyrosine. The GSTSH2 protein complexed specifically with a 175 kd tyrosine posphorylated protein in lysates of EGF-stimulated A431 cells (Figure 4.C). This phosphotyrosine-containing protein was subsequently shown to be the EGFR by immunoblotting with specific anti-receptor antibodies (data not shown). The SHC SH2 domain only bound to activated, autophosphorylated receptors. Taken together, these data demonstrate that the SHC SH2 domain binds to the autophosphorylated EGFR. The association between SHC proteins and activated growth factor receptors observed in vivo can therefore be recreated in vitro using the SHC SH2 domain. SHC Is Highly Conserved and Widely Expressed The degree to which SHC has been conserved in evolution was investigated by carrying out Southern blots of DNAs from different species hybridized to two SHC cDNA probes; one, the 11 RS2 probe, encodes a unique portion of the protein that contains neither the SH2 nor the collagen homologous regions; the other is the GFl cDNA probe, which encodes for a region of the SHC protein that spans the entire SH2 domain (see Experimental Procedures). As both probes hybridize to DNA fragments ranging from chicken to human, the SHC gene seems to have been conserved in vertebrate evolution (Figure 5A). Northern blot analysis of mRNAs from various human cell lines of different histologic derivation with DNA probes representative of different portions of the SHC cDNA (see
Cell 98
llRs2
Figure Gene
5. Evolutionary
GFl
Conservation
and RNA Expression
of the SHC
(A) Evolutionary conservation of the SHC gene. DNAs derived from the indicated species were hybridized with the 11 RS2 and the GFl DNA probes. Southern blots were exposed overnight except for the GFlhybridized chicken DNA, which was exposed for 4 days. (B) Northern blot analysis of SHC mRNA expression in human cells. Ten micrograms of total RNAs from the indicated cell types was hybridized to different SHC cDNA probes (see Results). This blot was hybridized with the GFl probe. The SHC 3.8, 3.4, and 2.8 kb transcripts are arrowed. The cell samples used for these RNA blots are described in the legend to Figure 3. 1
below) identified three different transcripts of approximately 3.8,3.4, and 2.8 kb, which were variably expressed in different mRNAs (Figure 5B). Hematopoietic cells only displayed the 3.4 kb transcript, whereas all three were variably expressed by the other cell types examined (normal mesothelium and fibroblasts; neoplastic lung, brain, and cervix). The three transcripts hybridized equally to three S/-/C DNA probes representative of the unique portion of the SHC protein (11 RS2 probe), the SH2 domain (GFl probe), and the 3’ untranslated region (GF13-0.9 probe) (data not shown). The 3.4 kb message is likely to represent the transcript that codes the SHC sequence we isolated, since the KG1 and P3HRI mRNAs used to prepare the libraries only contained this transcript (Figure 58). Since there was a strict correlation between expression of the 3.8 and 2.8 kb SHC-hybridizing transcripts and p66 in a variety of cell samples (compare Figure 3C and Figure 5B), either the 3.8 or the 2.8 kb transcript probably encodes ~66. Although the 3.8 and 2.8 kb transcripts could be encoded by SHC-related genes, this seems unlikely since all SHC-hybridizing transcripts had the same sensitivity to increasingly stringent hybridization conditions (data not shown).
Figure 6. Northern and Western Blot Analysis of NIH 3T3 Fibroblasts and NIH 3T3 Cell Lines Stably Transfected with the ZIP-SHC Construct (A) Ten micrograms of total RNA from the indicated samples were hybridized to the GFl probe. The endogenous SHC transcript and the genome-size viral transcript containing SHC sequences are arrowed. Ribosomal28Sand 18s markers are also indicated at the left of each panel. Northern blots in the lefthand panel were exposed for 12 hr, whereas the blot of the ZIPdHC9 and theZIP-SHCS(M) in the righthand panel was exposed for 2 hr to show that the smear on the 12 hr exposure of the ZIP-SHC9 blot is only due to overexpression of the SHC exogenous transcript. (B) Whole cell lysates from ZIP- or ZIP-SHC-expressing cell lines were prepared, and 50 ug of total protein was analyzed by 10% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with affinity-purified anti-SHC antibodies. The blot was then incubated with ‘?-protein A, and autoradiography was for 4 hr. Two additional ZIP-SHC clones (ZIP-SHC3 and ZIP-SHC13) are shown. Position of molecular weight markers (x 1O3) is indicated to the left, and SHC proteins (~66, ~52, and p46) are indicated at right.
NIH 3T3 Fibroblasts Overexpressing SHC Are Morphologically Transformed, Form Colonies in Agar, and Are Tumorigenic In Nude Mice The demonstration that the SHC SH2 domain is capable of binding activated growth factor receptors suggests that, like other SHP-containing proteins, SHC is involved in the cytoplasmic signaling of mitogenic stimuli. The capacity of the SHC protein to regulate cell growth was therefore investigated by analyzing the effects of constitutive SHC expression on the growth properties of NIH 3T3 fibroblasts. The EcoRI-Pvull 1.6 kb fragment, which contains the entire SHC coding region, was cloned into the ZIP expression vector (Cepko et al., 1984) under the control of the MLV promoter (ZIP-SHC). Cells transfected with ZIP or ZIP-SHC plasmids express the selectable 6418 antibiotic resistance phenotype. When ZIP-SHC was transfected into NIH 3T3 cells, it did not induce foci typical of ras transformation (data not shown). Polyclonal and clonal cell lines
$iC:
a Novel
SH2-Transforming
Sequence
that expressed SHC were therefore generated, so that the effects of SHC-constitutive expression on cell morphology andgrowth could be studied directly. One polyclonal (ZIPSHC bulk) and four clonal (ZIP-SHCl, ZIP-SHC2, ZIPSHCX, ZIP-SHCS) cell lines were selected for further analysis on the basis of their level of transfected SHC mRNA expression, which was highest in the ZIP-SHCS, lowest in the ZIP-SHC5, and intermediate in the other two (Figure 6A). Polyclonal (ZIP bulk) and clonal (ZIPl, ZIP4, ZIP5, ZIP6,ZIP8,ZIP9) cell lines that contained the parental ZIP expression vector were chosen as negative controls. To ensure that ZIP-SHC clones expressed normal-sized SHC proteins, they were analyzed by Western blotting with anti-SHC antibodies. All clones expressed much higher levels of SHC p46 and p52 than the normal NIH 3T3 control cell line. The levels of SHC proteins correlated roughly with exogenous SHC RNA levels in all ZIP-SHC cell lines (Figure 66). The morphology of the ZIP-SHC cell lines differed little from the parental NIH 3T3 or the control ZIP cell lines (NIH 3T3, ZIPl, ZIP-SHCl, and ZIP-SHC9 are compared in Figures 7a, 7b, 7c, and 78). ZIP-SHC cells were more refractile, more elongated, and piled up when they reached confluence. The number of ZIP-SHC cells on a given culture surface at confluence was approximately three times that of the ZIP cells, providing indirect evidence that SHC induces a loss of contact inhibition (data not shown).
We investigated the extent to which SHC could interfere with cell growth by testing the ability of ZIP and ZIP-SHC cell lines to form colonies in soft agar and induce tumors when injected into nude mice. ZIP-SHC or ZIP cells were plated in triplicate at various concentrations in 0.3% agar medium supplemented with 20% serum, and colonies were scored 14 days after plating. As reported in Table 1, the four ZIP-SHC clones and the polyclonal ZIP-SHC cell line formed colonies in soft agar at low frequencies, while the four control ZIP clones and the polyclonal ZIP cell line did not (representative micrographs of soft agar colonies from ZIP-SHCl and ZIPSHC9 cell lines are shown in Figures 8a and 8~). All mice injected with concentrations of between 1 and 20 x 1O6polyclonal ZIP-SHC cells developed tumors in an inverse dose-time dependent manner within lo-30 days (Table 1). Irrespective of the cell concentration inoculated, all tumors reached adiameter of 10 mm within 24-37days. Although all concentrations of ZIP cells provoked small tumors of I 0 mmy
1 5 10
2/3 (39 2 2) 213 (42 r 4) 3/3 (39.6 f 2.3)
-
20
3/3 (41.6 f 6.6)
-
1 5 10 20
414 414 4/4 414
414 414 414 4/4
ZIP-1 ZIP-8
July 10, 1992, Copyright
0 1992 by Cell Press
A Novel Transforming Protein (SHC) with an SH2 Domain Is Implicated in Mitogenic Signal Transduction Giuliana Pelicci,’ Luisa Lanfrancone,” Francesco Grignani, l Jane McGlade,t Federica Cavallo,* Guido Forni,S lldo Nicoletti,’ Faust0 Grignani,’ Tony Paw&t and Pier Giuseppe Pelicci’ l lstituto Clinica Medica I Policlinico Monteluce University of Perugia 06100 Perugia Italy tDivision of Molecular and Developmental Biology Samuel Lunenfeld Research Institute Mount Sinai Hospital Toronto, Ontario M5G 1X5 Canada Slstituto di Microbiologia University of Turin 10126 Turin Italy
Summary A new SH2-containing sequence, WC, was isolated by screening cDNA libraries with SH2 representative DNA probes. The WC cDNA is predicted to encode overlapping proteins of 46.6 and 51.7 kd that contain a single C-terminal SH2 domain, and an adjacent glycinelproline-rich motif with regions of homology with the al chain of collagen, but no identifiable catalytic domain. Anti-SHC antibodies recognized three proteins of 46,52, and 66 kd in a wide range of mammalian cell lines. These SHC proteins complexed with and were phospholylated by activated epidermal growth factor receptor. The physical association of SHC proteins with activated receptbrs was recreated in vitro by using a bacterially expressed SHC SH2 domain. NIH 3T3 mouse fibroblasts that constitutively overexpressed SHC acquired a transformed phenotype in culture and formed tumors in nude mice. These results suggest that the SHC gene products couple activated growth factor receptors to a signallng pathway that regulates the proliferation of mammalian ceils. Introduction Src homology 2 (SH2) domains are noncatalytic regions that are conserved among a group of cytoplasmic signaling proteins. These signaling proteins are directly regulated by receptor tyrosine kinases and control the activation of mitogenic signal transduction pathways by tyrosine kinases (reviewed by Ullrich and Schlessinger, 1990; Cantley et al., 1991; Koch et al., 1991). One function of SH2 domains was revealed during investigation of the cytoplasmic processes stimulated by the binding of growth factors, such as epidermal growth factor (EGF) and platelet-derived growth factor (PDGF), to their
receptors (EGFR and PDGFR). Ligand activation of these receptors induces dimerization, triggers receptor tyrosine kinase activity, and leads to autophosphorylation (reviewed by Ullrich and Schlessinger, 1990). In addition, the activated receptors associate with and phophorylate a number of cytoplasmic proteins that contain SH2 domains, including phospholipase C (PLC)+ (Margoliset al., 1989; Meisenhelder et al., 1989; Wahl et al., 1989; Morrison et al., 1990), ~21” GTPase activating protein (GAP) (Ellis et al., 1990; Kaplan et al., 1990; Kazlauskas et al., 1990), phosphatidylinositol (PI) 3’-kinase (Pl3K) (Kazlauskas and Cooper, 1989; Coughlin et al., 1989; Bjorge et al., 1990), and Src and Src-like tyrosine kinases (Kypta et al., 1990). These cytoplasmic proteins are very probably among the targets through which growth factor receptors transmit the message to trigger mitogenesis. In some cases, there is evidence that SHe-containing proteins do, indeed, regulate signal transduction. PLC+ cleaves phosphatidylinositol4,5-biphosphate (PIP2) into the second messengers, diacylglycerol and inositol trisphosphate, which in turn stimulate protein kinase C and increase intracellular calcium (reviewed by Whitman and Cantley, 1988). GAP, by converting Ras from the active GTP-bound form to the inactivated GDP-bound form (Trahey and McCormick, 1987), exerts negative regulation on Ras, which is located on the tyrosine kinase mitogenic signaling pathway (Mulcahy et al., 1985; Smith et al., 1986). Pl3K phosphorylates PI to form PI3-phosphate, and the reaction is stimulated by PDGF (Whitman et al., 1988; Auger et al., 1989). Evidence is accumulating that Src and Src-like kinases are also involved in growth factor-signaling pathways. Indeed, it has been shown that PDGF stimulates Src kinase activity in fibroblasts (Gould and Hunter 1988; Kypta et al., 1990). The interactions between cytoplasmic signaling proteins and growth factor receptors are mediated by specific binding of SH2 domains to tyrosine-phosphorylated regions of the receptors (Koch et al., 1991). In vitro binding studies have demonstrated that the GAP, Src, PLCql, and Pl3K SH2 domains are sufficient for binding to activated growth factor receptors (Anderson et al., 1990; Moran et al. 1990; McGlade et al., 1992). Moreover, SH2 domains apparently bind directly to specific tyrosinephosphorylated sites (Reedijk et al., 1992). As SH2 domains are important for directing proteinprotein interactions during mitogenic signal transduction, it is not surprising that some of the proteins that contain these domains can be activated as oncogenes. Notably, the v-Crk-transforming protein contains an SH2 domain and an SH3 domain (a distinct sequence frequently found alongside SH2 elements), but has no identifiable catalytic domain (Mayer et al., 1988). Despite this fact, v-Crk induces the tyrosine phosphotylation of a group of proteins that become associated with the Crk SH2 domain in v-Crktransformed cells (Mayer and Hanafusa, 1990; Matsuda et al., 1990). v-Crk, therefore, promotes the formation of cytoplasmic signaling complexes, suggesting that the SH2 domains of signaling proteins bind not only to activated
Cell 94
.
growth factor receptors, but also to specific cytoplasmic phosphotyrosine-containing proteins. The Src and GAP SH2 domains are also involved in a network of interactions with cytoplasmic phosphoproteins, which appear to be important for their biological activity (Moran et al., 1990, 1991; Koch et al., 1991, 1992). As these findings suggest that SH2 domains couple tyrosine kinases to their targets, the isolation of new SH2containing genes should help to define the mechanisms through which tyrosine kinases regulate normal cell growth and induce neoplastic transformation. In searching for such new genes, we isolated a previously unidentified gene that encodes a novel SHPcontaining protein, which has been named SHC. Evidence is presented that suggests that SHC has a part in fibroblast growth regulation and that its constitutive expression leads to a transformed cell phenotype.
Isolation of a cDNA Clone Encoding a Novel SHP-Containing Protein We searched for novel SH2 sequences by screening human cDNA libraries with a DNA probe representative of the c-fes SH2 domain, the Kpnl-Taql fragment from nucleotide 1454-1655 of the reported human c-fes cDNA sequence (Alcalay et al., 1990). As no fes-specific transcripts were detected in B-lymphoid cell lines using the Kpnl-Taql DNA probe at high stringency conditions (data not shown), a cDNA library prepared from the Burkitt lymphoma P3HRI mRNA was screened at low stringency, and a singe clone, IGFl , isolated from approximately 5 x lo5 plaques. The hGF1 insert was characterized by nucleotide sequence analysis. Translation of the GFl sequence revealed a single long open reading frame (ORF), which, upon GenBank analysis, was found to contain a single stretch of about 100 aa with sequence homology with SH2 domains (Figure 1 and Figure 2). Since lhe XGFl insert did not contain either the 3’ or the S’end of the gene, two additional human cDNA libraries (fetal brain and KG1 myeloid leukemia cells) were screened with the GFl insert as DNA probe and 13 positive clones isolated and studied by restriction enzyme analysis: A limited map of the hGF1 is shown together with the four longest cDNAs in Figure 1A. All phage inserts were subcloned into the EcoRl site of the plasmid vector pGEM-3 (Promega, Biotec) and sequenced. The entire nucleotide sequence of the two longest clones, LGFl 1 and hGF13, and the 5’ and 3’ extremities (approximately 300 bp) of all other cDNA inserts were determined. The complete 3031 nt sequence of the human SHC mRNA is shown in Figure 1 B. It corresponds to clone hGFl1 from nucleotide 1-164, to clone LGF13 from nucleotide 165-2676, and to clone LGF16 from nucleotide 2677 to the 3’ end. It contains a single 1609 bp ORF, beginning at nucleotide position 14 and extending to an in-frame TGA termintion codon at position 1502 (Figure 1 B). The termination codon is followed by a 3’ untranslated region of 1530 bp, and a poly(A) tail of 19 nt is found at the 3’ end of the sequence (Figure 1B). Two in-frame ATGs, each surrounded by a
sequence (CTGGACATGA and AAAGTCATGG), which is in good agreement with the consensus sequence for initiation of translation in eukaryotes (Kozak, 1969) are found at positions 63 (ATG63) and 216 (ATG216), respectively. The ORFs beginning with these ATGs encode putative proteins of 473 and 426 aa, respectively, with molecular masses of 51.7 and 46.6 kd. The protein(s) encoded by the SHC cDNA was demonstrated using an in vitro transcription-translation system. The SHC cDNA sequence +l-+1647 was inserted into the pGEM-3 plasmid vector and transcribed in vitro under the control of the SP6 and T7 promoters to produce sense and antisense SHC mRNA. The in vitro synthesized SHC mRNAs were then translated in a rabbit reticulocyte lysate in the presence of [%S]methionine. Translation of the SHC sense mRNA revealed two major proteins. Tl&ir apparent molecularweights were in good agreement with the molecular weights of the two predicted SHC proteins. No translation products were seen using the antisense SHC mRNA (Figure lC).ThisfindingsuggeststhattheSHCATG63and ATG216 are both used as initiation sites for translation. SHC Contains a New SH2 Domain and a Collagen-Homologous Region Analysis. of the putative SHC protein sequence by GenBank disclosed a stretch of 93 aa, from 376-471, with sequence homology with the SH2 domains of other proteins (Figure 2A). Although the degree of homology varied from 41% to 60% and was highest with the Src (60%) and Crk (57%) SH2 domains, the SH2 sequence encoded by SHC had significant DNA homology (63.3% only over a 96 bp stretch, from nucleotide position 1194-1291) with the corresponding human fes sequence (data not shown). Analysis of the putative SHC protein sequence also revealed a stretch of 145 aa, from 233-377, which was 50% homologous to human al collagen (Figure 28) (Bernard et al., 1963; Chu et al., 1964). This collagen-homologous SHC region was extremely rich in glycine (10.3%) and proline (20%), both of which are also abundant in collagen a chains (33.6% and 23.6% in the collagen al chain SHChomologous region). Fifty percent of the glycine and 65% of the proline al collagen residues are conserved in the SHC collagen homologous region. Due to its homology with SH2 and collagen, we named the new gene SHC (Src homologous and collagen). The relative position of the SH2 and collagen-homologous regions of SHC is shown in Figure 1A. Identification of SHC Proteins in Mammalian Cells To investigate the expression of SHC gene products in vivo, antibodies were raised against a bacterial glutathione S-transferase (GST)-SHC fusion protein (see Experimental Procedures). These anti-SHC antibodies specifically immunoprecipitated the 46 and 52 kd proteins translated from transcribed SHC cDNA in vitro (Figure 3A). In order to compare the in vitro translation products with cellular SHC proteins, COS-1 monkey cells, COS-1 cells transfected with the SHC GFll cDNA (pECE-GFll expression vector, see below), Rat-2 fibroblasts, and human A431 cells were metabolically labeled with [YS]methionine,
A.
Figure 1. DNA and Primary Amino Acid Sequence and In Vitro Translation of the SHC cDNA (A) Schematic representation of the SHC sequence and limited restriction enzyme map of independent SHC cDNAs from the PBHRI (SGFI), fetal brain (IFBl), and KG1 ()IGFll , 1IGF13. XGFl6) human cDNA libraries. The box represents the ORF starting from the ATG83 with the SH2 and collagen-homologous (CH) regions indicated. The ATG218 (see text) is also indicated. Numbers above the maps indicate the relative position of the 5’ and 3’ extremities of each clone with respect to the SHC cDNA sequence shown in (6). Identification and length of poly(A) tails are also shown. B, BamHI; A, Accl; E, EcoRI. (6) SHC nucleotide and deduced amino acid sequences. The translation of the SHC sequence from ATG83 to the termination codon (’ l ‘) at position 1592 is shown below the nucleotide sequence. The ATG218, the putative poly(A) addition signal AlTAAA at position 2992, and the 5’ in-frame termination codon at position 11 are underlined. Nucleotide and amino acid positions are indicated on the left of each lane. The collagen-homologous region is underlined, and the SH2 domain is double underlined. Circles indicate five putative Ser/ Thr phosphorylation sites (Woodgett et al., 1986; Hanks et al., 1988) and small vertical bars a potential tyrosine phosphorylation site (Cooper et al., 1984). (C) In vitro translation of the SHC cDNA. The RNA corresponding to the SHC sequence from +I-+1647 (sense transcript; lane ST) or +1647-+l (antisense transcript; lane AT) was synthesized in vitro and translated in the presence of [%]methionine. The labeled proteins were separated by SDS-PAGE and visualized by fluorography. The size of the protein markers (MWM) is given on the right.
A.
Figure 2. The SHC SH2 and Collagen-Homologous Regions
UK (378)rKGKL*..rrE....rr..LLq.,.... ~L~ltTt.POmvLT.~LO...C..q
(A) Alignment of the SH2 domain of the SHC sequence with other SH2 sequences. Sequences were aligned by eye after preliminary GenBank screening. Gaps, indicated by dots, have been introduced to improve alignment of the sequences. Capitals indicate residues invariant or conserved among six or more SH2 sequences. Amino acid substitutions within each group: A, G, P, S, and T; L, I, V, and M; D, E, N. and Q; K, R, and H; F, Y, and W; and C are defined as conservative (Schwartz and Dayoff, 1979). Thenumbersin parentheses are the first (left) and last (right) amino acid residues in each sequence. The amino acid sequence of the following proteins is presented: human Src (Anderson et al., 1985); human abl (Shtivelman et al., 1986); human fes (Alcalay et al., 1990); bovine PLC-~1 (Stahl et al., 1988); bovine GAP (Vogel et al., 1988); avian v-Crk (Mayer et al., 1988); human Pl3K (Skolnik et al., 1991); avian tensin (Davis et al., 1991). Asterisks indiate that the 18 and 15 amino acids of crk and tensin, respectively, are omitted. (8) Comparison of the SHC collagen-related sequence with al collagen. Vertical lines, identical residues; colons, conserved residues.
WC UL FE6 PLC Y PLC C G&PY G4P c
(151) UffWll..RrE . . . . &rlLlnn .~...QFL~~lt.kmVcLN~6l.n c127, [email protected] . . . . PEy.puqi.u . . . ..aFLIIILLL.m.r.slnLrYa..6..p w606Q)wUIP..naE...VAE..LLvflS . . . . . . . . SbLIIILI qp.kQMW.Lnd...O..l WOO) Yls(L~.DprhiYr.LLtyciet~L~Tf vWtLSfVm G. .k WB).WWLT..M . . . .M.llbwprO . . . . . WLIlltm l .RWalSfrr . . . . G..k c1m VwKLd..nti....AE.rLrqn.*s . . . ..aYLInEwr.CYwLsf.L~.q..Tn.v (348) WnwlS..KqE . . . . lyn.LLatvg.al . . ..cYLmt PfnY8LYf.J.tUnlqr 0.48) rrruQu..RQ . . . . Av.sLLqlgh . . . . ..EY~L~I.PmFvLw..~...*8w zk Y (333, HlrpdlS..Rall . . ..H*.kLPdt~.D.....6YFLIllOnb.kPmtLTLr.k...~.nl Gel c wz4z1) Hvlun..w . . . .1DI.LLrgkr.D . . . ..MfLIIILmk q.acvYw.w...dE..e Tmuin (573) WU~IB..LO....AII LLkdrap . . . . . . wIImmhsfr.wgLMlVu.~w
2 4u IfI PLE I PLC c $5: CRll 0111 Y am1 c Trnh
B.
pLI.IL.lV..q*Ov..H..~Dkr.F,~LI~.VIiuPu.Le..Lqrpkr (471) WYXlrkl..DwC~.lT..8rtq..F~W,,Y.akh..Ad...~...LchRLtt.w.P (249) wYnl.ntd..e..LyJ.maKu.FnTLN!LulHI.~t..v...dE...LlttLhYv.aP (218) pF*ril~l..k..Lyr~..E.~..~~lPlLl~~l~tq.~.lt.k......K.~.. (344) VQ*llkv.~.Otp..kfflMl~~~ltll.qq..WCLrcr*f~.l..RLIH.P ~36) IklcN...~..oqMI1~.~...~~~l~.~~..y..k..n..U~l.. (756) Wf~l..ir..~I~....rr.F~l~.&h.Wcllk~......KLL’IMC (271, . ..fKl.cpcphqh..4....ry.Ml~lldn.rkql n..S.~y...dkePV.P (439) YIIIIIY*I~.2.rr~...E(“)l~~fY.klh.Y Ldt.lt......Li.PV.. (334) ..LiKl..fhrD..~fM PIt...IMln#Y .rmLlr(*pL...LdvulyH.. 01) V.KhNlnktat..a ~fr.PymI..raLkcLvllfY.*t~L~...bnltL~A (720) ,rll9tsr...outlk~.~.TCL&L~.~(rpUlp . . . . . . . . KL...VIP..(Yl)
Cdl 9s
Figure 3. Identification Anti-SHC Antibodies
of SHC Proteins
Using
(A) Proteins translated in vitro from the SHC cDNA (in vitro transl.) and cell lysates from [Yfjmethionine-labeled COS-1 (COS-1) or COS-1 cells transfected with pECE-GFll (COS-1 + SHC), Rat-2 fibroblasts (R2), or human epithelial cells (A431) were immunoprecipitated with either a nonspecific antibody (RaMlg) or 2 pg of affinity-purified anti-SHC antibody (aSHC). lmmunoprecipitates were resolved by SDS-PAGE, and gels exposed to X-ray film for 3 hr at -7OOC. The SHC proteins (pQQ, ~52. p4Q) are arrowed. Molecular weight markers (MWM) were run in the same gel, and 91corresponding molecular weights (x itY) are indicated to their left. m- (8) Untransfected COSI cells (-), or COS-1 tz5i.A.. --; 4% cells transfected with pECEGFl1 (+GFil). pECEGFl3 (ffiFl3), or pECEG$l ITTG (ffiF1 ITTG) were analyzed by immunoblotting with anti-SHC antibodies. (C) lmmunoblots of different rodent and human cell lines. Whole cell lysates were prepared from rat fibroblast (RI hER), dog epithelial cells (TRMP), and the following human cells: the AML-1, KGI, and HL-60 myeloid cell lines; the U937 monocytic cell line; the Jurkat, PEER, and CEM T cell lines; the CB33, PBHRI, and Namalwa (NAM.) B cell lines; the HeLa cervical cancer cell line; melanoma cells; the CALUI lung cancer cell line; the At72 glioblastoma cell line, and cells from primary cultures of normal fibroblasts (FIBROBL.) and mesothelium (MESOTH.). All cell lines were obtained from American Type Cell Culture. Fifty micrograms of total protein was separated by 10% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with affinity-purified anti-SHC antibodies.
lysed, and immunoprecipitated with anti-SHC antibodies. All cell types contain three proteins of 46, 52, and 66 kd (~46, ~52, and ~66) which were specifically precipitated by the anti-SHC antibodies (Figure 3A). Of these, p46 and ~52 comigrated with the SHC cDNA translation products. Tryptic peptide maps of the [%S]methionine-labeled human p46 and ~52 in vitro translation products were indistinguishable from maps of the comigrating rat cellular proteins precipitated with anti-SHC antibodies (data not shown), consistent with the notion that the immunoreactive p46 and ~52 cellular proteins are indeed encoded by the SHC gene. To confirm its coding potential, the SHC GFl 1 cDNA was inserted into a mammalian expression vector containing an SV40 origin (pECE-GFl 1) and transfected into COS-1 monkey cells. Although these cells express endogenous SHC proteins, astriking increase in the levelsof p46 and ~52 polypeptides recognized by anti-SHC antibodies was‘evident in the transfected cells (Figures 3A and 38). No elevation in the level of p66 was observed. The possibility that p46 and ~52 result from the alternative use of two putative translation initiation sites (ATG63 and ATG216, see Figure 1), was tested by overexpressing cDNAs that contain both or one of the two ATGs: the GFl 1 cDNA (Figure 1A) contains both ATG63 and ATG216, GFl 1lTG only ATG83, and GF13 only ATG218 (Figure 1A). The GFl 1lTG cDNA was obtained by mutagenizing the ATG218 of the GFl 1 cDNA (ATG-lTG) (see Experimental Procedures). Whereas levels of both p46 and p52 were increased in GFl-transfected COS cells, single polypeptides that apparently comigrated with the p46 or the ~52 were detected in GF13 or GFl llTG-transfected COS cells (Figure 38). These results suggest that the ATG83 and ATG218 can be used in vivo as initiation sites for
translation and that the p46 and ~52 polypeptides are very likely produced by alternative usage of translation initiation sites. The extent of SHC’protein expression in different cells was assessed by using affinity-purified anti-SHC antibodies for immunoblotting of lysates from a variety of cell types. Human tumor cell lines from melanoma, cervical cancer, lung cancer, and glioblastoma and normal human fibroblasts and mesothelial cells all contained three polypeptides of 46 kd, 52 kd, and 66 kd 61148, ~52, and ~66) that were recognized by anti-SHC antibodies (Figure 3C). As immunoreactive p48, ~52, and p86 SHC polypeptdes were also detected in a dog epithelial cell line and in a rat fibroblast line (Figure 3C), SHC proteins appear to be well consenred between mammalian species and between cell types of different embryological origin. Indeed, the p46 and the ~52 SHC products were present in every cell line examined. The larger p66 SHC protein was not detected in any human hematopoietic cell lines tested (Figure SC). The p46, ~52, and p66 polypeptides were not detected when an irrelevant antibody or pre-immune serum were used in immunoblots (Figure 3A and data not shown). SHC Proteins Are Sound by Activated EGF Receptors and Phosphorylated in Response to Growth Factor Stimulation Since SHC proteins have an SH2 domain, we considered the possibility that they interact with and are substrates for receptortyrosine kinases. To test this notion, we examined the fate of SHC proteins in cells stimulated with EGF. Rat-l cells engineered to overexpress the human EGFR (Rl hER) were stimulated with EGF for 5 min, lysed, and immunoprecipitated with anti-SHC antibodies. The antiSHC immunoprecipitates were then immunoblotted with
SHC: a Novel 97
SHP-Transforming
A.
Sequence
B. EGF -+
C. EGF
-+
EGF -++
+
-II
a P-Y
L
Figure 4. SHC Proteins Bind to Activated Receptor and Are Phosphorylated upon Stimulation
(I EGFR
I
a SHC
+ EGF
(A) Rl hER cells were serum starved for 48 hr, and unstimulated cells (-) or cells stimulated with 80 nM EGF for 5 min (+) were lysed and immunoprecipitated using 2 pg of affinity-purified anti-SHC antibody (aSHC). Immunoprecipitates were immunoblotted with either anti-phosphotyrosine antibodies (a P-Y) or anti-EGFR (a EGFR). (B) %rUt’n-Starved RI hER cells were labeled with 1 mCi/ml [“Pjorthophosphate for 3 hr and unstimulated (-) or cells stimulated with 80 nM EGF for 15 min (+) were lysed and immunoprecipitated with affinity-purified anti-SHC antibodies (upper panel). lmmunoprecipitates Phosphoamino acid analysis (lower panel) was performed (+) cells. Positionsof phosphoserine (S), phosphothreonine
were resolved by SDS-PAGE and transferred to nitrocellulose and autoradiographed. on the combined SHC proteins (~88. ~52, p48) from unstimulated (-) and EGF-stimulated (T), and phosphotyrosine (Y) markers are indicated. (C) Immobilized GST (GST) or fusion protein containing the SHC SH2 domain (GST-SH2) EGF-stimulated (+) A431 cells. Protein complexes were resolved by SDS-PAGE, transferred antibody.
antibodies to phosphotyrosine. In the absence of EGF stimulation, SHC proteins isolated from serum-starved cells contained low but detectable levels of phosphotyrosine. Stimulation with EGF induced a marked increase in the tyrosine phosphorylation of ~46, ~52, and p66 (Figure 4A). In addition, anti-SHC immunoprecipitates from EGFstimulated cells contained a novel phosphotyrosine-containing protein of 175 kd, corresponding to the apparent size of the EGFR. lmmunoblotting of the same samples with anti-EGFR antibodies confirmed that the EGFR coprecipitated with SHC proteins following EGF stimulation (Figure 4A). The same result was obtained when EGFstimulated ceils were first immunoprecipitated with antiEGFR antibodies and then blotted with anti-SHC antibodies, in which case all three SHC proteins were observed to coprecipitate with the EGFR (data not shown). The phosphorylation of SHC proteins was analyzed directly by metabolic labeling of Rl hER cells with [32P]orthophosphate (Pi). The SHC proteins were phosphorylated in serumstarved cells, and this phosphorylation was increased upon EGF stimulation (Figure 48). Phosphoamino acid analysis revealed that the SHC proteins from serumstarved cells were phosphorylated primarily on serine and to a low level on tyrosine. EGF stimulation induced a marked increase in the level of phosphotyrosine (Figure 46). Taken together, these results indicate that SHC proteins bind to activated EGF receptors and are phosphorylated in response to EGF stimulation. The SHC SH2 Domain Binds to EGF Receptors In Vitro SH2-containing proteins associate with ligand-activated growth factor receptors and other tyrosine phosphorylated intracellular proteins in vivo. In vitro binding experiments suggest that these associations are mediated by the direct interaction of SH2 domains with tyrosine phosphorylated sites on SHP-binding proteins (reviewed in Koch et al., 1991). To determine whether the SHC SH2 domain can
EGF EGF
was incubated to nitrocellulose,
with lysates from serum-starved (-) or and probed with anti-phosphotyrosine
stably associate with phosphotyrosine-containing proteins in vitro, the SHC SH2 region was expressd in bacteria as a GST fusion protein (GST-SH2). The GST-SH2 polypeptide was isolated from bacterial lysates by chromatography on glutathione sepharose. Immobilized bacterial proteins (GST-SHP and control GST) were then incubated with lysates from A431 cells, which overexpress the human EGFR, stimulated with EGF. SH2 complexes were recovered, washed extensively, and analyzed by immunoblotting with antibodies to phosphotyrosine. The GSTSH2 protein complexed specifically with a 175 kd tyrosine posphorylated protein in lysates of EGF-stimulated A431 cells (Figure 4.C). This phosphotyrosine-containing protein was subsequently shown to be the EGFR by immunoblotting with specific anti-receptor antibodies (data not shown). The SHC SH2 domain only bound to activated, autophosphorylated receptors. Taken together, these data demonstrate that the SHC SH2 domain binds to the autophosphorylated EGFR. The association between SHC proteins and activated growth factor receptors observed in vivo can therefore be recreated in vitro using the SHC SH2 domain. SHC Is Highly Conserved and Widely Expressed The degree to which SHC has been conserved in evolution was investigated by carrying out Southern blots of DNAs from different species hybridized to two SHC cDNA probes; one, the 11 RS2 probe, encodes a unique portion of the protein that contains neither the SH2 nor the collagen homologous regions; the other is the GFl cDNA probe, which encodes for a region of the SHC protein that spans the entire SH2 domain (see Experimental Procedures). As both probes hybridize to DNA fragments ranging from chicken to human, the SHC gene seems to have been conserved in vertebrate evolution (Figure 5A). Northern blot analysis of mRNAs from various human cell lines of different histologic derivation with DNA probes representative of different portions of the SHC cDNA (see
Cell 98
llRs2
Figure Gene
5. Evolutionary
GFl
Conservation
and RNA Expression
of the SHC
(A) Evolutionary conservation of the SHC gene. DNAs derived from the indicated species were hybridized with the 11 RS2 and the GFl DNA probes. Southern blots were exposed overnight except for the GFlhybridized chicken DNA, which was exposed for 4 days. (B) Northern blot analysis of SHC mRNA expression in human cells. Ten micrograms of total RNAs from the indicated cell types was hybridized to different SHC cDNA probes (see Results). This blot was hybridized with the GFl probe. The SHC 3.8, 3.4, and 2.8 kb transcripts are arrowed. The cell samples used for these RNA blots are described in the legend to Figure 3. 1
below) identified three different transcripts of approximately 3.8,3.4, and 2.8 kb, which were variably expressed in different mRNAs (Figure 5B). Hematopoietic cells only displayed the 3.4 kb transcript, whereas all three were variably expressed by the other cell types examined (normal mesothelium and fibroblasts; neoplastic lung, brain, and cervix). The three transcripts hybridized equally to three S/-/C DNA probes representative of the unique portion of the SHC protein (11 RS2 probe), the SH2 domain (GFl probe), and the 3’ untranslated region (GF13-0.9 probe) (data not shown). The 3.4 kb message is likely to represent the transcript that codes the SHC sequence we isolated, since the KG1 and P3HRI mRNAs used to prepare the libraries only contained this transcript (Figure 58). Since there was a strict correlation between expression of the 3.8 and 2.8 kb SHC-hybridizing transcripts and p66 in a variety of cell samples (compare Figure 3C and Figure 5B), either the 3.8 or the 2.8 kb transcript probably encodes ~66. Although the 3.8 and 2.8 kb transcripts could be encoded by SHC-related genes, this seems unlikely since all SHC-hybridizing transcripts had the same sensitivity to increasingly stringent hybridization conditions (data not shown).
Figure 6. Northern and Western Blot Analysis of NIH 3T3 Fibroblasts and NIH 3T3 Cell Lines Stably Transfected with the ZIP-SHC Construct (A) Ten micrograms of total RNA from the indicated samples were hybridized to the GFl probe. The endogenous SHC transcript and the genome-size viral transcript containing SHC sequences are arrowed. Ribosomal28Sand 18s markers are also indicated at the left of each panel. Northern blots in the lefthand panel were exposed for 12 hr, whereas the blot of the ZIPdHC9 and theZIP-SHCS(M) in the righthand panel was exposed for 2 hr to show that the smear on the 12 hr exposure of the ZIP-SHC9 blot is only due to overexpression of the SHC exogenous transcript. (B) Whole cell lysates from ZIP- or ZIP-SHC-expressing cell lines were prepared, and 50 ug of total protein was analyzed by 10% SDS-PAGE, transferred to nitrocellulose, and immunoblotted with affinity-purified anti-SHC antibodies. The blot was then incubated with ‘?-protein A, and autoradiography was for 4 hr. Two additional ZIP-SHC clones (ZIP-SHC3 and ZIP-SHC13) are shown. Position of molecular weight markers (x 1O3) is indicated to the left, and SHC proteins (~66, ~52, and p46) are indicated at right.
NIH 3T3 Fibroblasts Overexpressing SHC Are Morphologically Transformed, Form Colonies in Agar, and Are Tumorigenic In Nude Mice The demonstration that the SHC SH2 domain is capable of binding activated growth factor receptors suggests that, like other SHP-containing proteins, SHC is involved in the cytoplasmic signaling of mitogenic stimuli. The capacity of the SHC protein to regulate cell growth was therefore investigated by analyzing the effects of constitutive SHC expression on the growth properties of NIH 3T3 fibroblasts. The EcoRI-Pvull 1.6 kb fragment, which contains the entire SHC coding region, was cloned into the ZIP expression vector (Cepko et al., 1984) under the control of the MLV promoter (ZIP-SHC). Cells transfected with ZIP or ZIP-SHC plasmids express the selectable 6418 antibiotic resistance phenotype. When ZIP-SHC was transfected into NIH 3T3 cells, it did not induce foci typical of ras transformation (data not shown). Polyclonal and clonal cell lines
$iC:
a Novel
SH2-Transforming
Sequence
that expressed SHC were therefore generated, so that the effects of SHC-constitutive expression on cell morphology andgrowth could be studied directly. One polyclonal (ZIPSHC bulk) and four clonal (ZIP-SHCl, ZIP-SHC2, ZIPSHCX, ZIP-SHCS) cell lines were selected for further analysis on the basis of their level of transfected SHC mRNA expression, which was highest in the ZIP-SHCS, lowest in the ZIP-SHC5, and intermediate in the other two (Figure 6A). Polyclonal (ZIP bulk) and clonal (ZIPl, ZIP4, ZIP5, ZIP6,ZIP8,ZIP9) cell lines that contained the parental ZIP expression vector were chosen as negative controls. To ensure that ZIP-SHC clones expressed normal-sized SHC proteins, they were analyzed by Western blotting with anti-SHC antibodies. All clones expressed much higher levels of SHC p46 and p52 than the normal NIH 3T3 control cell line. The levels of SHC proteins correlated roughly with exogenous SHC RNA levels in all ZIP-SHC cell lines (Figure 66). The morphology of the ZIP-SHC cell lines differed little from the parental NIH 3T3 or the control ZIP cell lines (NIH 3T3, ZIPl, ZIP-SHCl, and ZIP-SHC9 are compared in Figures 7a, 7b, 7c, and 78). ZIP-SHC cells were more refractile, more elongated, and piled up when they reached confluence. The number of ZIP-SHC cells on a given culture surface at confluence was approximately three times that of the ZIP cells, providing indirect evidence that SHC induces a loss of contact inhibition (data not shown).
We investigated the extent to which SHC could interfere with cell growth by testing the ability of ZIP and ZIP-SHC cell lines to form colonies in soft agar and induce tumors when injected into nude mice. ZIP-SHC or ZIP cells were plated in triplicate at various concentrations in 0.3% agar medium supplemented with 20% serum, and colonies were scored 14 days after plating. As reported in Table 1, the four ZIP-SHC clones and the polyclonal ZIP-SHC cell line formed colonies in soft agar at low frequencies, while the four control ZIP clones and the polyclonal ZIP cell line did not (representative micrographs of soft agar colonies from ZIP-SHCl and ZIPSHC9 cell lines are shown in Figures 8a and 8~). All mice injected with concentrations of between 1 and 20 x 1O6polyclonal ZIP-SHC cells developed tumors in an inverse dose-time dependent manner within lo-30 days (Table 1). Irrespective of the cell concentration inoculated, all tumors reached adiameter of 10 mm within 24-37days. Although all concentrations of ZIP cells provoked small tumors of I 0 mmy
1 5 10
2/3 (39 2 2) 213 (42 r 4) 3/3 (39.6 f 2.3)
-
20
3/3 (41.6 f 6.6)
-
1 5 10 20
414 414 4/4 414
414 414 414 4/4
ZIP-1 ZIP-8
