Portal de Periódicos da CAPES

In Vivo Analysis of Argos Structure-Function

1998; Elsevier BV; Volume: 273; Issue: 7 Linguagem: Inglês

10.1074/jbc.273.7.4275

1083-351X

Robert I. Howes, Jonathan D. Wasserman, Matthew Freeman,

HER2/EGFR in Cancer Research

The Drosophila Argos protein is the only known extracellular inhibitor of the epidermal growth factor receptor (EGFR). It is structurally related to the activating ligands, in that it is a secreted protein with a single epidermal growth factor (EGF) domain. To understand the mechanism of Argos inhibition, we have investigated which regions of the protein are essential. A series of deletions were made and tested in vivo; furthermore, by analyzing chimeric proteins between Argos and the activating ligand, Spitz (a transforming growth factor-α-like factor), we have examined what makes one inhibitory and the other activating. Our results reveal that Argos has structural requirements that differ from all known EGFR activating ligands; domains flanking the EGF domain are essential for its function. We have also defined the important regions of the atypical Argos EGF domain. The extended B-loop is necessary, whereas the C-loop can be replaced with the equivalent Spitz region without substantially affecting Argos function. Comparison of theargos genes from Drosophila melanogaster and the housefly, Musca domestica, supports our structure-function analysis. These studies are a prerequisite for understanding how Argos inhibits the Drosophila EGFR and provide a basis for designing mammalian EGFR inhibitors. The Drosophila Argos protein is the only known extracellular inhibitor of the epidermal growth factor receptor (EGFR). It is structurally related to the activating ligands, in that it is a secreted protein with a single epidermal growth factor (EGF) domain. To understand the mechanism of Argos inhibition, we have investigated which regions of the protein are essential. A series of deletions were made and tested in vivo; furthermore, by analyzing chimeric proteins between Argos and the activating ligand, Spitz (a transforming growth factor-α-like factor), we have examined what makes one inhibitory and the other activating. Our results reveal that Argos has structural requirements that differ from all known EGFR activating ligands; domains flanking the EGF domain are essential for its function. We have also defined the important regions of the atypical Argos EGF domain. The extended B-loop is necessary, whereas the C-loop can be replaced with the equivalent Spitz region without substantially affecting Argos function. Comparison of theargos genes from Drosophila melanogaster and the housefly, Musca domestica, supports our structure-function analysis. These studies are a prerequisite for understanding how Argos inhibits the Drosophila EGFR and provide a basis for designing mammalian EGFR inhibitors. The epidermal growth factor receptor (EGFR) 1The abbreviations used are: EGFR, epidermal growth factor receptor; EGF, epidermal growth factor; TGFα, transforming growth factor-α; S2, Schneider's line 2; PCR, polymerase chain reaction.1The abbreviations used are: EGFR, epidermal growth factor receptor; EGF, epidermal growth factor; TGFα, transforming growth factor-α; S2, Schneider's line 2; PCR, polymerase chain reaction. controls many aspects of animal growth, development, and cell proliferation (1Wiley L.M. Adamson E.D. Tsark E.C. BioEssays. 1995; 17: 839-846Crossref PubMed Scopus (50) Google Scholar). How these different roles are coordinated and their relative importance in normal development is not yet well understood, but it is clear that overactivation of the receptor in mammals is implicated in many forms of cancer (2Macias A. Perez R. Hagerstrom T. Skoog L. Anticancer Res. 1989; 9: 177-179PubMed Google Scholar, 3LeJeune S. Leek R. Horak E. Plowman G. Greenall M. Harris A.L. Cancer Res. 1993; 53: 3597-3602PubMed Google Scholar, 4Lee D.C. Fenton S.E. Berkowitz E.A. Hissong M.A. Pharmacol. Rev. 1995; 47: 51-85PubMed Google Scholar, 5Chia C.M. Winston R.M.L. Handyside A.H. Development. 1995; 121: 299-307PubMed Google Scholar). This overactivation can be caused by amplification of the receptor gene, by activating mutations in the receptor, or frequently by overexpression of activating ligands, which causes inappropriate autocrine signaling. The EGFR is a tyrosine kinase; it has an extracellular ligand binding domain and an intracellular kinase domain. As with other receptor tyrosine kinases, it forms dimers upon activation, and these then transphosphorylate each other, causing the activation of downstream signal transduction pathways (6Heldin C.H. Cell. 1995; 80: 213-223Abstract Full Text PDF PubMed Scopus (1427) Google Scholar). There are four members of the EGFR family in mammals (ErbB1–ErbB4), and they act in a complex set of overlapping functions, both as homo- and heterodimers (7Carraway K.L. Cantley L.C. Cell. 1994; 78: 5-8Abstract Full Text PDF PubMed Scopus (583) Google Scholar). They are activated by several classes of ligand, including EGF itself, transforming growth factor-α (TGFα), neuregulins, and others. The motif responsible for receptor activation in all known ligands is an EGF domain, comprising six characteristically spaced cysteines and a few other essential amino acids (8Groenen L.C. Nice E.C. Burgess A.W. Growth Factors. 1994; 11: 235-257Crossref PubMed Scopus (216) Google Scholar). Antagonists of the EGFR might have therapeutic potential, and much effort has been made to understand ligand binding and activation, to help design such inhibitors (9van Zoelen E.J.J. Lenferink A.E.G. Kramer R.H. van de Poll M.L.M. Pathol. Res. Practice. 1996; 192: 761-767Crossref PubMed Scopus (8) Google Scholar, 10Campion S.R. Niyogi S.K. Prog. Nucleic Acid Res. Mol. Biol. 1994; 49: 353-383Crossref PubMed Scopus (28) Google Scholar, 11Komoriya A. Hortsch M. Meyers C. Smith M. Kanety H. Schlessinger J. Proc. Natl. Acad. Sci. U. S. A. 1984; 81: 1351-1355Crossref PubMed Google Scholar, 12Hoeprich Jr., P.D. Langton B.C. Zhang J.W. Tam J.P. J. Biol. Chem. 1989; 264: 19086-19091Abstract Full Text PDF PubMed Google Scholar, 13van de Poll M.L.M. Lenferink A.E.G. van Vugt M.J.H. Jacobs J.J.L. Janssen J.W.H. Joldersma M. van Zoelen E.J.J. J. Biol. Chem. 1995; 270: 22337-22343Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar, 14Matsunami R.K. Campion S.R. Niyogi S.K. Stevens A. FEBS Lett. 1990; 264: 105-108Crossref PubMed Scopus (31) Google Scholar). Although several important regions have been identified, effective antagonists have remained elusive because of the apparent inability to uncouple receptor binding and activation; all mutants that still bind efficiently are activating.Like its mammalian counterparts, the Drosophila EGFR regulates many aspects of development (15Price J.V. Clifford R.J. Schupbach T. Cell. 1989; 56: 1085-1092Abstract Full Text PDF PubMed Scopus (214) Google Scholar, 16Schejter E.D. Shilo B.-Z. Cell. 1989; 56: 1093-1104Abstract Full Text PDF PubMed Scopus (174) Google Scholar, 17Raz E. Shilo B.-Z. Development. 1992; 114: 113-123PubMed Google Scholar, 18Raz E. Shilo B.-Z. Genes Dev. 1993; 7: 1937-1948Crossref PubMed Scopus (103) Google Scholar, 19Clifford R. Schupbach T. Development. 1992; 115: 853-872PubMed Google Scholar, 20Schweitzer R. Shilo B.-Z. Trends Genet. 1997; 13: 191-196Abstract Full Text PDF PubMed Scopus (224) Google Scholar, 21Freeman M. Cell. 1996; 87: 651-660Abstract Full Text Full Text PDF PubMed Scopus (689) Google Scholar). In flies, there is only one such receptor and it is equally similar to all four of theerbB genes of mammals, suggesting that it is the prototype receptor of the mammalian family (22Schejter E.D. Segal D. Glazer L. Shilo B.-Z. Cell. 1986; 46: 1091-1101Abstract Full Text PDF PubMed Scopus (65) Google Scholar, 23Wadsworth S.C. Vincent W.S. Bilodaeu-Wentworth D. Nature. 1985; 314: 178-180Crossref PubMed Scopus (35) Google Scholar). Its activating ligands fall into the recognized classes of mammalian ligands; Spitz and Gurken are TGFα-like molecules, whereas Vein resembles the neuregulins (24Rutledge B.J. Zhang K. Bier E. Jan Y.N. Perrimon N. Genes Dev. 1992; 6: 1503-1517Crossref PubMed Scopus (240) Google Scholar, 25Neuman-Silberberg F.S. Schupbach T. Cell. 1993; 75: 165-174Abstract Full Text PDF PubMed Scopus (506) Google Scholar, 26Schnepp B. Grumbling G. Donaldson T. Simcox A. Genes Dev. 1996; 10: 2302-2313Crossref PubMed Scopus (191) Google Scholar). There is also a unique extracellular inhibitor of the fly EGFR called Argos. Argos has the hallmarks of an EGFR ligand in that it is a secreted protein with a single EGF domain (albeit atypical) (27Freeman M. Klämbt C. Goodman C.S. Rubin G.M. Cell. 1992; 69: 963-975Abstract Full Text PDF PubMed Scopus (213) Google Scholar, 28Kretzschmar D. Brunner A. Wiersdorff V. Pflugfelder G.O. Heisenberg M. Schneuwly S. EMBO J. 1992; 11: 2531-2539Crossref PubMed Scopus (48) Google Scholar, 29Okano H. Hayashi S. Tanimura T. Sawamoto K. Differentiation. 1992; 52: 1-11Crossref PubMed Scopus (43) Google Scholar), and it has been shown to inhibit EGFR activation both in tissue culture and in the developing fly (30Schweitzer R. Howes R. Smith R. Shilo B.-Z. Freeman M. Nature. 1995; 376: 699-702Crossref PubMed Scopus (223) Google Scholar). EGFR activity is required for many aspects of fly development (20Schweitzer R. Shilo B.-Z. Trends Genet. 1997; 13: 191-196Abstract Full Text PDF PubMed Scopus (224) Google Scholar). Of relevance to this work, EGFR triggers the differentiation of cells in the eye and the wing vein (21Freeman M. Cell. 1996; 87: 651-660Abstract Full Text Full Text PDF PubMed Scopus (689) Google Scholar,31Sturtevant M.A. Roark M. Bier E. Genes Dev. 1993; 7: 961-973Crossref PubMed Scopus (285) Google Scholar). Loss of activity in the pathway (e.g. by loss-of-function mutations in EGFR or its activating ligands, or hyperactivity of Argos) causes loss of these structures; gain of activity (e.g. by loss-of-function mutations in argos) causes extra cells in the eye and wing veins.We would like to understand the mechanism of EGFR inhibition by Argos for theoretical and practical reasons. Our goal is to identify which parts of the molecule are necessary for its function to help design mammalian EGFR antagonists. To these ends, we have explored which parts of the Argos protein are essential for its inhibitory function. We have also compared it with the activating ligand, Spitz, to understand the structural elements that make one an activator and the other an inhibitor of the EGF receptor. With a similar goal, two groups have recently reported attempts to make chimeras between human EGF and the EGF domain of Argos, but neither succeeded in producing an inhibitor of the human EGFR (32van de Poll M.L.M. van Vugt M.J.H. Lenferink A.E.G. van Zoelen E.J.J. Biochemistry. 1997; 36: 7425-7431Crossref PubMed Scopus (22) Google Scholar, 33Lohmeyer M. Mason S. Gullick W.J. Int. J. Oncol. 1997; 10: 677-682PubMed Google Scholar). Our results suggest several reasons for this; although the atypical EGF domain is indeed a key element of Argos's inhibitory functions, other regions of the protein are also essential. These findings distinguish Argos from all other EGF-like factors, which appear to interact with the EGFR solely through the EGF domain; they also suggest that rational design of EGFR inhibitors should eventually be possible.DISCUSSIONThe wide range of functions of the members of the EGF receptor family is reflected in their large number of different activating ligands (54Carpenter G. Cohen S. J. Biol. Chem. 1990; 265: 7709-7712Abstract Full Text PDF PubMed Google Scholar, 55Derynck R. Cell. 1988; 54: 593-595Abstract Full Text PDF PubMed Scopus (538) Google Scholar, 56Riese D.J. Bermingham Y. Vanraaij T.M. Buckley S. Plowman G.D. Stern D.F. Oncogene. 1996; 12: 345-353PubMed Google Scholar, 57Riese D.J., II Kim E.D. Elenius K. Buckley S. Klagsbrun M. Plowman G.D. Stern D.F. J. Biol. Chem. 1996; 271: 20047-20052Abstract Full Text Full Text PDF PubMed Scopus (155) Google Scholar, 58Aviezer D. Yayon A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 12173-12177Crossref PubMed Scopus (0) Google Scholar, 59Carraway K.L. Burden S.J. Curr. Opin. Neurobiol. 1995; 5: 606-612Crossref PubMed Scopus (153) Google Scholar). The mature forms of EGF and TGFα are polypeptides that comprise a single EGF domain and little else (8Groenen L.C. Nice E.C. Burgess A.W. Growth Factors. 1994; 11: 235-257Crossref PubMed Scopus (216) Google Scholar). The neuregulins are much larger; they include a C-terminal EGF domain and a long N-terminal region, which may modulate their action (59Carraway K.L. Burden S.J. Curr. Opin. Neurobiol. 1995; 5: 606-612Crossref PubMed Scopus (153) Google Scholar). A minimal EGF domain from the neuregulins is sufficient to bind and activate their receptor (60Holmes W.E. Sliwkowski M.Y. Akita R.W. Henzel W.J. Lee J. Park J.W. Yansura D. Abadi N. Raab H. Lewis G.D. Shepard H.M. Kuang W.J. Wood W.I. Goeddel D.V. Vandlen R.L. Science. 1992; 256: 1205-1210Crossref PubMed Scopus (923) Google Scholar, 61Carraway III, K.L. Soltoff S.P. Diamonti A.J. Cantley L.C. J. Biol. Chem. 1995; 270: 7111-7116Abstract Full Text Full Text PDF PubMed Scopus (131) Google Scholar). Amphiregulin and heparin-binding EGF have very basic domains immediately N-terminal to their EGF domains, which are necessary for heparin binding and full biological activity (62Johnson G.R. Wong L. J. Biol. Chem. 1994; 269: 27149-27154Abstract Full Text PDF PubMed Google Scholar, 63Aviezer D. Yayon A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 12173-12177Crossref PubMed Scopus (125) Google Scholar). In this paper, we have shown that Argos, which is extended N- and C-terminally compared with EGF, and which is the only known EGF receptor inhibitor, has different structural requirements to any of the known activating ligands. For Argos to be an effective inhibitor, it needs its own EGF domain, the whole adjacent C terminus of the protein, and some of the region immediately N-terminal to the EGF domain (note that this region does not resemble the lysine-arginine rich heparin-binding regions of amphiregulin and heparin-binding EGF).Recently, the only other known inhibitor of an receptor tyrosine kinase has been identified. This factor, angiopoietin-2, is structurally very similar to the activating ligand for the Tie-2 receptor, which it blocks (64Maisonpierre P.C. Suri C. Jones P.F. Bartunkova S. Wiegand S. Radziejewski C. Compton D. McClain J. Aldrich T.H. Papadopoulos N. Daly T.J. Davis S. Sato T.N. Yancopoulos G.D. Science. 1997; 277: 55-60Crossref PubMed Scopus (2921) Google Scholar). Although Ang-2 binds the receptor, it fails to activate it, providing a simple explanation for its antagonistic function. The mechanism of Argos function must be more complex since it will inhibit ligand-independent activation of the receptor, which can be induced in tissue culture cells (30Schweitzer R. Howes R. Smith R. Shilo B.-Z. Freeman M. Nature. 1995; 376: 699-702Crossref PubMed Scopus (223) Google Scholar). This implies that Argos does not act by binding to and sequestering ligand, nor does it simply block the ligand binding site on the receptor. Receptor tyrosine kinases are activated by ligand-induced dimerization or oligomerization (65Ullrich A. Schlessinger J. Cell. 1990; 61: 203-212Abstract Full Text PDF PubMed Scopus (4583) Google Scholar), suggesting a simple mechanism for blocking receptor activation; we hypothesize that Argos inhibits EGFR dimerization. Our discovery that sequences from outside the EGF domain are essential for its function suggests that these extra domains participate in the hypothesized block to receptor dimerization. Note that, although there is strong circumstantial evidence that Argos acts directly on the receptor, the biochemical demonstration has yet to be reported. It is, however, clear that the inhibition occurs at the receptor itself (30Schweitzer R. Howes R. Smith R. Shilo B.-Z. Freeman M. Nature. 1995; 376: 699-702Crossref PubMed Scopus (223) Google Scholar), so the general proposal that Argos acts by causing a block to EGFR dimerization could hold even if it works indirectly.As well as identifying new functional domains of Argos, our results shed light on which parts of its EGF domain are essential for Argos to inhibit the receptor. The extended B-loop, which includes the C3-C4 stretch of amino acids, is an obvious candidate for being involved in Argos's unusual inhibitory action. We have shown that if this loop is replaced by the corresponding region of the Spitz EGF domain, which resembles human EGF, Argos function is abolished. This supports the idea that this region is necessary and participates in Argos inhibition. The B-loop in Musca Argos is of identical length and near-identical sequence to Drosophila Argos, consistent with its functional importance. Indeed, this spacing is also identical in Argos from the butterfly, Precis coenia, which diverged from Drosophila more than 200 million years ago, 4M. Kengaku and C. Tabin, personal communication. further supporting this conclusion. Interestingly, myxoma virus growth factor, which is related to EGF, has a slightly extended B-loop (13 amino acid residues) and activates the EGFR 200-fold less effectively than EGF (66Lin Y.Z. Ke X.H. Tam J.P. Biochemistry. 1991; 30: 3310-3314Crossref PubMed Scopus (30) Google Scholar). Despite intensive study, the role of this domain, which forms a β-sheet in human EGF, is still not clear, but there is substantial evidence implicating it in receptor binding and/or activation (11Komoriya A. Hortsch M. Meyers C. Smith M. Kanety H. Schlessinger J. Proc. Natl. Acad. Sci. U. S. A. 1984; 81: 1351-1355Crossref PubMed Google Scholar, 49Richter A. Conlan J.W. Ward M.E. Chamberlin S.G. Alexander P. Richards N.G.J. Davies D.E. Biochemistry. 1992; 31: 9546-9554Crossref PubMed Scopus (16) Google Scholar, 50Richter A. Drummond D.R. MacGarvie J. Puddicombe S.M. Chamberlin S.G. Davies D.E. J. Biol. Chem. 1995; 270: 1612-1616Abstract Full Text Full Text PDF PubMed Scopus (31) Google Scholar, 51Campion S.R. Matsunami R.K. Engler D.A. Niyogi S.K. Biochemistry. 1990; 29: 9988-9993Crossref PubMed Scopus (41) Google Scholar).The C-loop of EGF domains also contains critical residues for EGFR activation (9van Zoelen E.J.J. Lenferink A.E.G. Kramer R.H. van de Poll M.L.M. Pathol. Res. Practice. 1996; 192: 761-767Crossref PubMed Scopus (8) Google Scholar, 13van de Poll M.L.M. Lenferink A.E.G. van Vugt M.J.H. Jacobs J.J.L. Janssen J.W.H. Joldersma M. van Zoelen E.J.J. J. Biol. Chem. 1995; 270: 22337-22343Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar, 67Kramer R.H. Lenferink A.E.G. van Bueren-Koornneef I.L. van der Meer A. van de Poll M.L.M. van Zoelen E.J.J. J. Biol. Chem. 1994; 269: 8708-8711Abstract Full Text PDF PubMed Google Scholar). Both Spitz and Argos contain the two key amino acids (Gly-39 and Arg-41 from human EGF; Fig. 3 A), but in other ways the Argos C-loop looks very different from all known activating ligands, including Spitz. It is shorter and has increased positive charge; it is also identical in Musca, so it is surprising that replacing it with the more typical Spitz C-loop has little effect on Argos function. Perhaps the C-loop is necessary for receptor binding, but not for conferring inhibitory properties, making it is replaceable by a similar loop from another DrosophilaEGFR-binding protein, Spitz.A naturally occurring EGFR inhibitor like Argos might have therapeutic value in humans. We have failed to isolate mammalian homologues of Argos by various techniques, and it is possible that an EGFR inhibitor has evolved only in insects. We think it more likely that the conservation of EGF-like domains is not high enough over long enough stretches (their main conserved features are the spacing of cysteines and a few other critical amino acids) to make them easy to clone over long evolutionary distances. (Drosophila diverged from vertebrates 600–1000 million years ago). We are more hopeful of engineering an antagonist based on known mammalian ligands, as has been tried by several groups (9van Zoelen E.J.J. Lenferink A.E.G. Kramer R.H. van de Poll M.L.M. Pathol. Res. Practice. 1996; 192: 761-767Crossref PubMed Scopus (8) Google Scholar, 14Matsunami R.K. Campion S.R. Niyogi S.K. Stevens A. FEBS Lett. 1990; 264: 105-108Crossref PubMed Scopus (31) Google Scholar, 32van de Poll M.L.M. van Vugt M.J.H. Lenferink A.E.G. van Zoelen E.J.J. Biochemistry. 1997; 36: 7425-7431Crossref PubMed Scopus (22) Google Scholar, 33Lohmeyer M. Mason S. Gullick W.J. Int. J. Oncol. 1997; 10: 677-682PubMed Google Scholar), and our goal is to exploit Argos and Spitz to gain the necessary insight into the mechanisms of EGFR activation and inhibition. The epidermal growth factor receptor (EGFR) 1The abbreviations used are: EGFR, epidermal growth factor receptor; EGF, epidermal growth factor; TGFα, transforming growth factor-α; S2, Schneider's line 2; PCR, polymerase chain reaction.1The abbreviations used are: EGFR, epidermal growth factor receptor; EGF, epidermal growth factor; TGFα, transforming growth factor-α; S2, Schneider's line 2; PCR, polymerase chain reaction. controls many aspects of animal growth, development, and cell proliferation (1Wiley L.M. Adamson E.D. Tsark E.C. BioEssays. 1995; 17: 839-846Crossref PubMed Scopus (50) Google Scholar). How these different roles are coordinated and their relative importance in normal development is not yet well understood, but it is clear that overactivation of the receptor in mammals is implicated in many forms of cancer (2Macias A. Perez R. Hagerstrom T. Skoog L. Anticancer Res. 1989; 9: 177-179PubMed Google Scholar, 3LeJeune S. Leek R. Horak E. Plowman G. Greenall M. Harris A.L. Cancer Res. 1993; 53: 3597-3602PubMed Google Scholar, 4Lee D.C. Fenton S.E. Berkowitz E.A. Hissong M.A. Pharmacol. Rev. 1995; 47: 51-85PubMed Google Scholar, 5Chia C.M. Winston R.M.L. Handyside A.H. Development. 1995; 121: 299-307PubMed Google Scholar). This overactivation can be caused by amplification of the receptor gene, by activating mutations in the receptor, or frequently by overexpression of activating ligands, which causes inappropriate autocrine signaling. The EGFR is a tyrosine kinase; it has an extracellular ligand binding domain and an intracellular kinase domain. As with other receptor tyrosine kinases, it forms dimers upon activation, and these then transphosphorylate each other, causing the activation of downstream signal transduction pathways (6Heldin C.H. Cell. 1995; 80: 213-223Abstract Full Text PDF PubMed Scopus (1427) Google Scholar). There are four members of the EGFR family in mammals (ErbB1–ErbB4), and they act in a complex set of overlapping functions, both as homo- and heterodimers (7Carraway K.L. Cantley L.C. Cell. 1994; 78: 5-8Abstract Full Text PDF PubMed Scopus (583) Google Scholar). They are activated by several classes of ligand, including EGF itself, transforming growth factor-α (TGFα), neuregulins, and others. The motif responsible for receptor activation in all known ligands is an EGF domain, comprising six characteristically spaced cysteines and a few other essential amino acids (8Groenen L.C. Nice E.C. Burgess A.W. Growth Factors. 1994; 11: 235-257Crossref PubMed Scopus (216) Google Scholar). Antagonists of the EGFR might have therapeutic potential, and much effort has been made to understand ligand binding and activation, to help design such inhibitors (9van Zoelen E.J.J. Lenferink A.E.G. Kramer R.H. van de Poll M.L.M. Pathol. Res. Practice. 1996; 192: 761-767Crossref PubMed Scopus (8) Google Scholar, 10Campion S.R. Niyogi S.K. Prog. Nucleic Acid Res. Mol. Biol. 1994; 49: 353-383Crossref PubMed Scopus (28) Google Scholar, 11Komoriya A. Hortsch M. Meyers C. Smith M. Kanety H. Schlessinger J. Proc. Natl. Acad. Sci. U. S. A. 1984; 81: 1351-1355Crossref PubMed Google Scholar, 12Hoeprich Jr., P.D. Langton B.C. Zhang J.W. Tam J.P. J. Biol. Chem. 1989; 264: 19086-19091Abstract Full Text PDF PubMed Google Scholar, 13van de Poll M.L.M. Lenferink A.E.G. van Vugt M.J.H. Jacobs J.J.L. Janssen J.W.H. Joldersma M. van Zoelen E.J.J. J. Biol. Chem. 1995; 270: 22337-22343Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar, 14Matsunami R.K. Campion S.R. Niyogi S.K. Stevens A. FEBS Lett. 1990; 264: 105-108Crossref PubMed Scopus (31) Google Scholar). Although several important regions have been identified, effective antagonists have remained elusive because of the apparent inability to uncouple receptor binding and activation; all mutants that still bind efficiently are activating. Like its mammalian counterparts, the Drosophila EGFR regulates many aspects of development (15Price J.V. Clifford R.J. Schupbach T. Cell. 1989; 56: 1085-1092Abstract Full Text PDF PubMed Scopus (214) Google Scholar, 16Schejter E.D. Shilo B.-Z. Cell. 1989; 56: 1093-1104Abstract Full Text PDF PubMed Scopus (174) Google Scholar, 17Raz E. Shilo B.-Z. Development. 1992; 114: 113-123PubMed Google Scholar, 18Raz E. Shilo B.-Z. Genes Dev. 1993; 7: 1937-1948Crossref PubMed Scopus (103) Google Scholar, 19Clifford R. Schupbach T. Development. 1992; 115: 853-872PubMed Google Scholar, 20Schweitzer R. Shilo B.-Z. Trends Genet. 1997; 13: 191-196Abstract Full Text PDF PubMed Scopus (224) Google Scholar, 21Freeman M. Cell. 1996; 87: 651-660Abstract Full Text Full Text PDF PubMed Scopus (689) Google Scholar). In flies, there is only one such receptor and it is equally similar to all four of theerbB genes of mammals, suggesting that it is the prototype receptor of the mammalian family (22Schejter E.D. Segal D. Glazer L. Shilo B.-Z. Cell. 1986; 46: 1091-1101Abstract Full Text PDF PubMed Scopus (65) Google Scholar, 23Wadsworth S.C. Vincent W.S. Bilodaeu-Wentworth D. Nature. 1985; 314: 178-180Crossref PubMed Scopus (35) Google Scholar). Its activating ligands fall into the recognized classes of mammalian ligands; Spitz and Gurken are TGFα-like molecules, whereas Vein resembles the neuregulins (24Rutledge B.J. Zhang K. Bier E. Jan Y.N. Perrimon N. Genes Dev. 1992; 6: 1503-1517Crossref PubMed Scopus (240) Google Scholar, 25Neuman-Silberberg F.S. Schupbach T. Cell. 1993; 75: 165-174Abstract Full Text PDF PubMed Scopus (506) Google Scholar, 26Schnepp B. Grumbling G. Donaldson T. Simcox A. Genes Dev. 1996; 10: 2302-2313Crossref PubMed Scopus (191) Google Scholar). There is also a unique extracellular inhibitor of the fly EGFR called Argos. Argos has the hallmarks of an EGFR ligand in that it is a secreted protein with a single EGF domain (albeit atypical) (27Freeman M. Klämbt C. Goodman C.S. Rubin G.M. Cell. 1992; 69: 963-975Abstract Full Text PDF PubMed Scopus (213) Google Scholar, 28Kretzschmar D. Brunner A. Wiersdorff V. Pflugfelder G.O. Heisenberg M. Schneuwly S. EMBO J. 1992; 11: 2531-2539Crossref PubMed Scopus (48) Google Scholar, 29Okano H. Hayashi S. Tanimura T. Sawamoto K. Differentiation. 1992; 52: 1-11Crossref PubMed Scopus (43) Google Scholar), and it has been shown to inhibit EGFR activation both in tissue culture and in the developing fly (30Schweitzer R. Howes R. Smith R. Shilo B.-Z. Freeman M. Nature. 1995; 376: 699-702Crossref PubMed Scopus (223) Google Scholar). EGFR activity is required for many aspects of fly development (20Schweitzer R. Shilo B.-Z. Trends Genet. 1997; 13: 191-196Abstract Full Text PDF PubMed Scopus (224) Google Scholar). Of relevance to this work, EGFR triggers the differentiation of cells in the eye and the wing vein (21Freeman M. Cell. 1996; 87: 651-660Abstract Full Text Full Text PDF PubMed Scopus (689) Google Scholar,31Sturtevant M.A. Roark M. Bier E. Genes Dev. 1993; 7: 961-973Crossref PubMed Scopus (285) Google Scholar). Loss of activity in the pathway (e.g. by loss-of-function mutations in EGFR or its activating ligands, or hyperactivity of Argos) causes loss of these structures; gain of activity (e.g. by loss-of-function mutations in argos) causes extra cells in the eye and wing veins. We would like to understand the mechanism of EGFR inhibition by Argos for theoretical and practical reasons. Our goal is to identify which parts of the molecule are necessary for its function to help design mammalian EGFR antagonists. To these ends, we have explored which parts of the Argos protein are essential for its inhibitory function. We have also compared it with the activating ligand, Spitz, to understand the structural elements that make one an activator and the other an inhibitor of the EGF receptor. With a similar goal, two groups have recently reported attempts to make chimeras between human EGF and the EGF domain of Argos, but neither succeeded in producing an inhibitor of the human EGFR (32van de Poll M.L.M. van Vugt M.J.H. Lenferink A.E.G. van Zoelen E.J.J. Biochemistry. 1997; 36: 7425-7431Crossref PubMed Scopus (22) Google Scholar, 33Lohmeyer M. Mason S. Gullick W.J. Int. J. Oncol. 1997; 10: 677-682PubMed Google Scholar). Our results suggest several reasons for this; although the atypical EGF domain is indeed a key element of Argos's inhibitory functions, other regions of the protein are also essential. These findings distinguish Argos from all other EGF-like factors, which appear to interact with the EGFR solely through the EGF domain; they also suggest that rational design of EGFR inhibitors should eventually be possible. DISCUSSIONThe wide range of functions of the members of the EGF receptor family is reflected in their large number of different activating ligands (54Carpenter G. Cohen S. J. Biol. Chem. 1990; 265: 7709-7712Abstract Full Text PDF PubMed Google Scholar, 55Derynck R. Cell. 1988; 54: 593-595Abstract Full Text PDF PubMed Scopus (538) Google Scholar, 56Riese D.J. Bermingham Y. Vanraaij T.M. Buckley S. Plowman G.D. Stern D.F. Oncogene. 1996; 12: 345-353PubMed Google Scholar, 57Riese D.J., II Kim E.D. Elenius K. Buckley S. Klagsbrun M. Plowman G.D. Stern D.F. J. Biol. Chem. 1996; 271: 20047-20052Abstract Full Text Full Text PDF PubMed Scopus (155) Google Scholar, 58Aviezer D. Yayon A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 12173-12177Crossref PubMed Scopus (0) Google Scholar, 59Carraway K.L. Burden S.J. Curr. Opin. Neurobiol. 1995; 5: 606-612Crossref PubMed Scopus (153) Google Scholar). The mature forms of EGF and TGFα are polypeptides that comprise a single EGF domain and little else (8Groenen L.C. Nice E.C. Burgess A.W. Growth Factors. 1994; 11: 235-257Crossref PubMed Scopus (216) Google Scholar). The neuregulins are much larger; they include a C-terminal EGF domain and a long N-terminal region, which may modulate their action (59Carraway K.L. Burden S.J. Curr. Opin. Neurobiol. 1995; 5: 606-612Crossref PubMed Scopus (153) Google Scholar). A minimal EGF domain from the neuregulins is sufficient to bind and activate their receptor (60Holmes W.E. Sliwkowski M.Y. Akita R.W. Henzel W.J. Lee J. Park J.W. Yansura D. Abadi N. Raab H. Lewis G.D. Shepard H.M. Kuang W.J. Wood W.I. Goeddel D.V. Vandlen R.L. Science. 1992; 256: 1205-1210Crossref PubMed Scopus (923) Google Scholar, 61Carraway III, K.L. Soltoff S.P. Diamonti A.J. Cantley L.C. J. Biol. Chem. 1995; 270: 7111-7116Abstract Full Text Full Text PDF PubMed Scopus (131) Google Scholar). Amphiregulin and heparin-binding EGF have very basic domains immediately N-terminal to their EGF domains, which are necessary for heparin binding and full biological activity (62Johnson G.R. Wong L. J. Biol. Chem. 1994; 269: 27149-27154Abstract Full Text PDF PubMed Google Scholar, 63Aviezer D. Yayon A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 12173-12177Crossref PubMed Scopus (125) Google Scholar). In this paper, we have shown that Argos, which is extended N- and C-terminally compared with EGF, and which is the only known EGF receptor inhibitor, has different structural requirements to any of the known activating ligands. For Argos to be an effective inhibitor, it needs its own EGF domain, the whole adjacent C terminus of the protein, and some of the region immediately N-terminal to the EGF domain (note that this region does not resemble the lysine-arginine rich heparin-binding regions of amphiregulin and heparin-binding EGF).Recently, the only other known inhibitor of an receptor tyrosine kinase has been identified. This factor, angiopoietin-2, is structurally very similar to the activating ligand for the Tie-2 receptor, which it blocks (64Maisonpierre P.C. Suri C. Jones P.F. Bartunkova S. Wiegand S. Radziejewski C. Compton D. McClain J. Aldrich T.H. Papadopoulos N. Daly T.J. Davis S. Sato T.N. Yancopoulos G.D. Science. 1997; 277: 55-60Crossref PubMed Scopus (2921) Google Scholar). Although Ang-2 binds the receptor, it fails to activate it, providing a simple explanation for its antagonistic function. The mechanism of Argos function must be more complex since it will inhibit ligand-independent activation of the receptor, which can be induced in tissue culture cells (30Schweitzer R. Howes R. Smith R. Shilo B.-Z. Freeman M. Nature. 1995; 376: 699-702Crossref PubMed Scopus (223) Google Scholar). This implies that Argos does not act by binding to and sequestering ligand, nor does it simply block the ligand binding site on the receptor. Receptor tyrosine kinases are activated by ligand-induced dimerization or oligomerization (65Ullrich A. Schlessinger J. Cell. 1990; 61: 203-212Abstract Full Text PDF PubMed Scopus (4583) Google Scholar), suggesting a simple mechanism for blocking receptor activation; we hypothesize that Argos inhibits EGFR dimerization. Our discovery that sequences from outside the EGF domain are essential for its function suggests that these extra domains participate in the hypothesized block to receptor dimerization. Note that, although there is strong circumstantial evidence that Argos acts directly on the receptor, the biochemical demonstration has yet to be reported. It is, however, clear that the inhibition occurs at the receptor itself (30Schweitzer R. Howes R. Smith R. Shilo B.-Z. Freeman M. Nature. 1995; 376: 699-702Crossref PubMed Scopus (223) Google Scholar), so the general proposal that Argos acts by causing a block to EGFR dimerization could hold even if it works indirectly.As well as identifying new functional domains of Argos, our results shed light on which parts of its EGF domain are essential for Argos to inhibit the receptor. The extended B-loop, which includes the C3-C4 stretch of amino acids, is an obvious candidate for being involved in Argos's unusual inhibitory action. We have shown that if this loop is replaced by the corresponding region of the Spitz EGF domain, which resembles human EGF, Argos function is abolished. This supports the idea that this region is necessary and participates in Argos inhibition. The B-loop in Musca Argos is of identical length and near-identical sequence to Drosophila Argos, consistent with its functional importance. Indeed, this spacing is also identical in Argos from the butterfly, Precis coenia, which diverged from Drosophila more than 200 million years ago, 4M. Kengaku and C. Tabin, personal communication. further supporting this conclusion. Interestingly, myxoma virus growth factor, which is related to EGF, has a slightly extended B-loop (13 amino acid residues) and activates the EGFR 200-fold less effectively than EGF (66Lin Y.Z. Ke X.H. Tam J.P. Biochemistry. 1991; 30: 3310-3314Crossref PubMed Scopus (30) Google Scholar). Despite intensive study, the role of this domain, which forms a β-sheet in human EGF, is still not clear, but there is substantial evidence implicating it in receptor binding and/or activation (11Komoriya A. Hortsch M. Meyers C. Smith M. Kanety H. Schlessinger J. Proc. Natl. Acad. Sci. U. S. A. 1984; 81: 1351-1355Crossref PubMed Google Scholar, 49Richter A. Conlan J.W. Ward M.E. Chamberlin S.G. Alexander P. Richards N.G.J. Davies D.E. Biochemistry. 1992; 31: 9546-9554Crossref PubMed Scopus (16) Google Scholar, 50Richter A. Drummond D.R. MacGarvie J. Puddicombe S.M. Chamberlin S.G. Davies D.E. J. Biol. Chem. 1995; 270: 1612-1616Abstract Full Text Full Text PDF PubMed Scopus (31) Google Scholar, 51Campion S.R. Matsunami R.K. Engler D.A. Niyogi S.K. Biochemistry. 1990; 29: 9988-9993Crossref PubMed Scopus (41) Google Scholar).The C-loop of EGF domains also contains critical residues for EGFR activation (9van Zoelen E.J.J. Lenferink A.E.G. Kramer R.H. van de Poll M.L.M. Pathol. Res. Practice. 1996; 192: 761-767Crossref PubMed Scopus (8) Google Scholar, 13van de Poll M.L.M. Lenferink A.E.G. van Vugt M.J.H. Jacobs J.J.L. Janssen J.W.H. Joldersma M. van Zoelen E.J.J. J. Biol. Chem. 1995; 270: 22337-22343Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar, 67Kramer R.H. Lenferink A.E.G. van Bueren-Koornneef I.L. van der Meer A. van de Poll M.L.M. van Zoelen E.J.J. J. Biol. Chem. 1994; 269: 8708-8711Abstract Full Text PDF PubMed Google Scholar). Both Spitz and Argos contain the two key amino acids (Gly-39 and Arg-41 from human EGF; Fig. 3 A), but in other ways the Argos C-loop looks very different from all known activating ligands, including Spitz. It is shorter and has increased positive charge; it is also identical in Musca, so it is surprising that replacing it with the more typical Spitz C-loop has little effect on Argos function. Perhaps the C-loop is necessary for receptor binding, but not for conferring inhibitory properties, making it is replaceable by a similar loop from another DrosophilaEGFR-binding protein, Spitz.A naturally occurring EGFR inhibitor like Argos might have therapeutic value in humans. We have failed to isolate mammalian homologues of Argos by various techniques, and it is possible that an EGFR inhibitor has evolved only in insects. We think it more likely that the conservation of EGF-like domains is not high enough over long enough stretches (their main conserved features are the spacing of cysteines and a few other critical amino acids) to make them easy to clone over long evolutionary distances. (Drosophila diverged from vertebrates 600–1000 million years ago). We are more hopeful of engineering an antagonist based on known mammalian ligands, as has been tried by several groups (9van Zoelen E.J.J. Lenferink A.E.G. Kramer R.H. van de Poll M.L.M. Pathol. Res. Practice. 1996; 192: 761-767Crossref PubMed Scopus (8) Google Scholar, 14Matsunami R.K. Campion S.R. Niyogi S.K. Stevens A. FEBS Lett. 1990; 264: 105-108Crossref PubMed Scopus (31) Google Scholar, 32van de Poll M.L.M. van Vugt M.J.H. Lenferink A.E.G. van Zoelen E.J.J. Biochemistry. 1997; 36: 7425-7431Crossref PubMed Scopus (22) Google Scholar, 33Lohmeyer M. Mason S. Gullick W.J. Int. J. Oncol. 1997; 10: 677-682PubMed Google Scholar), and our goal is to exploit Argos and Spitz to gain the necessary insight into the mechanisms of EGFR activation and inhibition. The wide range of functions of the members of the EGF receptor family is reflected in their large number of different activating ligands (54Carpenter G. Cohen S. J. Biol. Chem. 1990; 265: 7709-7712Abstract Full Text PDF PubMed Google Scholar, 55Derynck R. Cell. 1988; 54: 593-595Abstract Full Text PDF PubMed Scopus (538) Google Scholar, 56Riese D.J. Bermingham Y. Vanraaij T.M. Buckley S. Plowman G.D. Stern D.F. Oncogene. 1996; 12: 345-353PubMed Google Scholar, 57Riese D.J., II Kim E.D. Elenius K. Buckley S. Klagsbrun M. Plowman G.D. Stern D.F. J. Biol. Chem. 1996; 271: 20047-20052Abstract Full Text Full Text PDF PubMed Scopus (155) Google Scholar, 58Aviezer D. Yayon A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 12173-12177Crossref PubMed Scopus (0) Google Scholar, 59Carraway K.L. Burden S.J. Curr. Opin. Neurobiol. 1995; 5: 606-612Crossref PubMed Scopus (153) Google Scholar). The mature forms of EGF and TGFα are polypeptides that comprise a single EGF domain and little else (8Groenen L.C. Nice E.C. Burgess A.W. Growth Factors. 1994; 11: 235-257Crossref PubMed Scopus (216) Google Scholar). The neuregulins are much larger; they include a C-terminal EGF domain and a long N-terminal region, which may modulate their action (59Carraway K.L. Burden S.J. Curr. Opin. Neurobiol. 1995; 5: 606-612Crossref PubMed Scopus (153) Google Scholar). A minimal EGF domain from the neuregulins is sufficient to bind and activate their receptor (60Holmes W.E. Sliwkowski M.Y. Akita R.W. Henzel W.J. Lee J. Park J.W. Yansura D. Abadi N. Raab H. Lewis G.D. Shepard H.M. Kuang W.J. Wood W.I. Goeddel D.V. Vandlen R.L. Science. 1992; 256: 1205-1210Crossref PubMed Scopus (923) Google Scholar, 61Carraway III, K.L. Soltoff S.P. Diamonti A.J. Cantley L.C. J. Biol. Chem. 1995; 270: 7111-7116Abstract Full Text Full Text PDF PubMed Scopus (131) Google Scholar). Amphiregulin and heparin-binding EGF have very basic domains immediately N-terminal to their EGF domains, which are necessary for heparin binding and full biological activity (62Johnson G.R. Wong L. J. Biol. Chem. 1994; 269: 27149-27154Abstract Full Text PDF PubMed Google Scholar, 63Aviezer D. Yayon A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 12173-12177Crossref PubMed Scopus (125) Google Scholar). In this paper, we have shown that Argos, which is extended N- and C-terminally compared with EGF, and which is the only known EGF receptor inhibitor, has different structural requirements to any of the known activating ligands. For Argos to be an effective inhibitor, it needs its own EGF domain, the whole adjacent C terminus of the protein, and some of the region immediately N-terminal to the EGF domain (note that this region does not resemble the lysine-arginine rich heparin-binding regions of amphiregulin and heparin-binding EGF). Recently, the only other known inhibitor of an receptor tyrosine kinase has been identified. This factor, angiopoietin-2, is structurally very similar to the activating ligand for the Tie-2 receptor, which it blocks (64Maisonpierre P.C. Suri C. Jones P.F. Bartunkova S. Wiegand S. Radziejewski C. Compton D. McClain J. Aldrich T.H. Papadopoulos N. Daly T.J. Davis S. Sato T.N. Yancopoulos G.D. Science. 1997; 277: 55-60Crossref PubMed Scopus (2921) Google Scholar). Although Ang-2 binds the receptor, it fails to activate it, providing a simple explanation for its antagonistic function. The mechanism of Argos function must be more complex since it will inhibit ligand-independent activation of the receptor, which can be induced in tissue culture cells (30Schweitzer R. Howes R. Smith R. Shilo B.-Z. Freeman M. Nature. 1995; 376: 699-702Crossref PubMed Scopus (223) Google Scholar). This implies that Argos does not act by binding to and sequestering ligand, nor does it simply block the ligand binding site on the receptor. Receptor tyrosine kinases are activated by ligand-induced dimerization or oligomerization (65Ullrich A. Schlessinger J. Cell. 1990; 61: 203-212Abstract Full Text PDF PubMed Scopus (4583) Google Scholar), suggesting a simple mechanism for blocking receptor activation; we hypothesize that Argos inhibits EGFR dimerization. Our discovery that sequences from outside the EGF domain are essential for its function suggests that these extra domains participate in the hypothesized block to receptor dimerization. Note that, although there is strong circumstantial evidence that Argos acts directly on the receptor, the biochemical demonstration has yet to be reported. It is, however, clear that the inhibition occurs at the receptor itself (30Schweitzer R. Howes R. Smith R. Shilo B.-Z. Freeman M. Nature. 1995; 376: 699-702Crossref PubMed Scopus (223) Google Scholar), so the general proposal that Argos acts by causing a block to EGFR dimerization could hold even if it works indirectly. As well as identifying new functional domains of Argos, our results shed light on which parts of its EGF domain are essential for Argos to inhibit the receptor. The extended B-loop, which includes the C3-C4 stretch of amino acids, is an obvious candidate for being involved in Argos's unusual inhibitory action. We have shown that if this loop is replaced by the corresponding region of the Spitz EGF domain, which resembles human EGF, Argos function is abolished. This supports the idea that this region is necessary and participates in Argos inhibition. The B-loop in Musca Argos is of identical length and near-identical sequence to Drosophila Argos, consistent with its functional importance. Indeed, this spacing is also identical in Argos from the butterfly, Precis coenia, which diverged from Drosophila more than 200 million years ago, 4M. Kengaku and C. Tabin, personal communication. further supporting this conclusion. Interestingly, myxoma virus growth factor, which is related to EGF, has a slightly extended B-loop (13 amino acid residues) and activates the EGFR 200-fold less effectively than EGF (66Lin Y.Z. Ke X.H. Tam J.P. Biochemistry. 1991; 30: 3310-3314Crossref PubMed Scopus (30) Google Scholar). Despite intensive study, the role of this domain, which forms a β-sheet in human EGF, is still not clear, but there is substantial evidence implicating it in receptor binding and/or activation (11Komoriya A. Hortsch M. Meyers C. Smith M. Kanety H. Schlessinger J. Proc. Natl. Acad. Sci. U. S. A. 1984; 81: 1351-1355Crossref PubMed Google Scholar, 49Richter A. Conlan J.W. Ward M.E. Chamberlin S.G. Alexander P. Richards N.G.J. Davies D.E. Biochemistry. 1992; 31: 9546-9554Crossref PubMed Scopus (16) Google Scholar, 50Richter A. Drummond D.R. MacGarvie J. Puddicombe S.M. Chamberlin S.G. Davies D.E. J. Biol. Chem. 1995; 270: 1612-1616Abstract Full Text Full Text PDF PubMed Scopus (31) Google Scholar, 51Campion S.R. Matsunami R.K. Engler D.A. Niyogi S.K. Biochemistry. 1990; 29: 9988-9993Crossref PubMed Scopus (41) Google Scholar). The C-loop of EGF domains also contains critical residues for EGFR activation (9van Zoelen E.J.J. Lenferink A.E.G. Kramer R.H. van de Poll M.L.M. Pathol. Res. Practice. 1996; 192: 761-767Crossref PubMed Scopus (8) Google Scholar, 13van de Poll M.L.M. Lenferink A.E.G. van Vugt M.J.H. Jacobs J.J.L. Janssen J.W.H. Joldersma M. van Zoelen E.J.J. J. Biol. Chem. 1995; 270: 22337-22343Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar, 67Kramer R.H. Lenferink A.E.G. van Bueren-Koornneef I.L. van der Meer A. van de Poll M.L.M. van Zoelen E.J.J. J. Biol. Chem. 1994; 269: 8708-8711Abstract Full Text PDF PubMed Google Scholar). Both Spitz and Argos contain the two key amino acids (Gly-39 and Arg-41 from human EGF; Fig. 3 A), but in other ways the Argos C-loop looks very different from all known activating ligands, including Spitz. It is shorter and has increased positive charge; it is also identical in Musca, so it is surprising that replacing it with the more typical Spitz C-loop has little effect on Argos function. Perhaps the C-loop is necessary for receptor binding, but not for conferring inhibitory properties, making it is replaceable by a similar loop from another DrosophilaEGFR-binding protein, Spitz. A naturally occurring EGFR inhibitor like Argos might have therapeutic value in humans. We have failed to isolate mammalian homologues of Argos by various techniques, and it is possible that an EGFR inhibitor has evolved only in insects. We think it more likely that the conservation of EGF-like domains is not high enough over long enough stretches (their main conserved features are the spacing of cysteines and a few other critical amino acids) to make them easy to clone over long evolutionary distances. (Drosophila diverged from vertebrates 600–1000 million years ago). We are more hopeful of engineering an antagonist based on known mammalian ligands, as has been tried by several groups (9van Zoelen E.J.J. Lenferink A.E.G. Kramer R.H. van de Poll M.L.M. Pathol. Res. Practice. 1996; 192: 761-767Crossref PubMed Scopus (8) Google Scholar, 14Matsunami R.K. Campion S.R. Niyogi S.K. Stevens A. FEBS Lett. 1990; 264: 105-108Crossref PubMed Scopus (31) Google Scholar, 32van de Poll M.L.M. van Vugt M.J.H. Lenferink A.E.G. van Zoelen E.J.J. Biochemistry. 1997; 36: 7425-7431Crossref PubMed Scopus (22) Google Scholar, 33Lohmeyer M. Mason S. Gullick W.J. Int. J. Oncol. 1997; 10: 677-682PubMed Google Scholar), and our goal is to exploit Argos and Spitz to gain the necessary insight into the mechanisms of EGFR activation and inhibition. We are grateful to Myriam Golembo and Ben-Zion Shilo for help with sequencing some of the constructs and Richard Smith for excellent technical assistance. Daniel Bopp and René Feyereisen generously sent us M. domestica cDNA libraries, and David Micklem and Michael Akam helped greatly with our attempts to clone Argos from various species. Tony Burgess at the Cambridge University Anatomy Department performed the scanning electron microscopy. We thank Sean Munro and Rob Kay for their comments on the manuscript.