Recognition of Eukaryotic Initiation Factor 4G Isoforms by Picornaviral Proteinases
2002; Elsevier BV; Volume: 277; Issue: 46 Linguagem: Inglês
10.1074/jbc.m208006200
ISSN1083-351X
AutoresNicole Foeger, Walter Glaser, Tim Skern,
Tópico(s)Animal Disease Management and Epidemiology
ResumoThe leader proteinase (Lpro) of foot and mouth disease virus is a papain-like cysteine proteinase. After processing itself from the polyprotein, Lpro then cleaves the host protein eukaryotic initiation factor (eIf) 4GI, thus preventing protein synthesis from capped mRNA in the infected cell. We have investigated Lpro interaction with eIF4GI and its isoform, eIF4GII. Lpro, expressed as a catalytically inactive fusion protein with glutathione S-transferase, binds specifically to eIF4G isomers in rabbit reticulocyte lysates. Deletion and specific mutagenesis were used to map the binding domain on Lpro to residues 183–195 of the C-terminal extension and to residue Cys133. These residues of the C-terminal extension and Cys133 are adjacent in the crystal structure but lie about 25 Å from the active site. The region on eIF4GI recognized by the Lpro C-terminal extension was mapped to residues 640–669 using eIF4GI fragments generated by proteolysis or byin vitro translation. The Lpro cleavage site at Gly674↓Arg675 was not necessary for binding. Similar experiments with human rhinovirus 2A proteinase (2Apro), a chymotrypsin-like cysteine proteinase that also cleaves eIF4G isoforms, revealed that 2Apro can also bind to eIF4GI fragments lacking its cleavage site. These experiments strongly suggest a novel interaction between picornaviral proteinases and eIF4G isoforms. The leader proteinase (Lpro) of foot and mouth disease virus is a papain-like cysteine proteinase. After processing itself from the polyprotein, Lpro then cleaves the host protein eukaryotic initiation factor (eIf) 4GI, thus preventing protein synthesis from capped mRNA in the infected cell. We have investigated Lpro interaction with eIF4GI and its isoform, eIF4GII. Lpro, expressed as a catalytically inactive fusion protein with glutathione S-transferase, binds specifically to eIF4G isomers in rabbit reticulocyte lysates. Deletion and specific mutagenesis were used to map the binding domain on Lpro to residues 183–195 of the C-terminal extension and to residue Cys133. These residues of the C-terminal extension and Cys133 are adjacent in the crystal structure but lie about 25 Å from the active site. The region on eIF4GI recognized by the Lpro C-terminal extension was mapped to residues 640–669 using eIF4GI fragments generated by proteolysis or byin vitro translation. The Lpro cleavage site at Gly674↓Arg675 was not necessary for binding. Similar experiments with human rhinovirus 2A proteinase (2Apro), a chymotrypsin-like cysteine proteinase that also cleaves eIF4G isoforms, revealed that 2Apro can also bind to eIF4GI fragments lacking its cleavage site. These experiments strongly suggest a novel interaction between picornaviral proteinases and eIF4G isoforms. Members of the eukaryotic initiation factor 4 (eIF4) 1The abbreviations used are: eIF, eukaryotic initiation factor; CTE, C-terminal extension; FMDV, foot and mouth disease virus; Lpro, leader proteinase; PV, poliovirus; RRL, rabbit reticulocyte lysate; SVDV, swine vesicular disease virus; 2Apro, 2A proteinase; GST, glutathioneS-transferase; HRV, human rhinovirus group of initiation factors play a central role in the initiation of protein synthesis in eukaryotes (1Merrick W.C. Hershey J.W.B. Sonenberg N. Hershey J.W.B. Mathews M.B. Translational Control of Gene Expression. 39. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2000: 33-88Google Scholar). These factors comprise the cap-binding protein eIF4E, the RNA helicase eIF4A, the RNA-binding protein eIF4B and a scaffold protein that occurs in two isoforms, eIF4GI and eIF4GII (2Gingras A.C. Raught B. Sonenberg N. Annu. Rev. Biochem. 1999; 68: 913-963Crossref PubMed Scopus (1762) Google Scholar). In mammalian systems, eIF4E, eIF4A, and eIF4G can be isolated as a complex designated eIF4F. eIF4F together with eIF4B and eIF3 recognizes the cap structure at the 5′ end of the cellular mRNA, unwinds RNA secondary structure in an ATP-dependent manner, and enables binding of the 40 S ribosomal subunit to the mRNA (3Pain V.M. Eur. J. Biochem. 1996; 236: 747-771Crossref PubMed Scopus (637) Google Scholar, 4Morley S.J. Curtis P.S. Pain V.M. RNA (N. Y.). 1997; 3: 1085-1104PubMed Google Scholar, 5Hentze M.W. Science. 1997; 275: 500-501Crossref PubMed Scopus (193) Google Scholar). The eIF4F complex is an important control point for the overall rate of protein synthesis (6Rhoads R.E. J. Biol. Chem. 1993; 268: 3017-3020Abstract Full Text PDF PubMed Google Scholar, 7Sonenberg N. Gingras A.C. Curr. Opin. Cell Biol. 1998; 10: 268-275Crossref PubMed Scopus (506) Google Scholar). During the replication of certain picornaviruses such as FMDV, HRV, and PV, translation of capped cellular mRNA in the infected cell is shut down by specific cleavage of the eIF4G isoforms (8Kräusslich H.G. Nicklin M.J. Toyoda H. Etchison D. Wimmer E. J. Virol. 1987; 61: 2711-2718Crossref PubMed Google Scholar, 9Devaney M.A. Vakharia V.N. Lloyd R.E. Ehrenfeld E. Grubman M.J. J. Virol. 1988; 62: 4407-4409Crossref PubMed Google Scholar, 10Lloyd R.E. Grubman M.J. Ehrenfeld E. J. Virol. 1988; 62: 4216-4223Crossref PubMed Google Scholar, 11Gradi A. Imataka H. Svitkin Y.V. Rom E. Raught B. Morino S. Sonenberg N. Mol. Cell. Biol. 1998; 18: 334-342Crossref PubMed Scopus (247) Google Scholar) (Fig.1). Initiation of protein synthesis from viral mRNA is unaffected because it initiates internally and does not require a cap structure (12Jackson R.J. Howell M.T. Kaminski A. Trends Biochem. Sci. 1990; 15: 477-483Abstract Full Text PDF PubMed Scopus (281) Google Scholar, 13Belsham G.J. Sonenberg N. Microbiol. Rev. 1996; 60: 499-511Crossref PubMed Google Scholar, 14Belsham G.J. Jackson R.R. Sonenberg N. Hershey J.W.B. Mathews M.B. Translational Control of Gene Expression. 39. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2000: 869-900Google Scholar). Substantial genetic evidence shows that the papain-like Lpro of FMDV and the chymotrypsin-like 2Apro of HRVs and enteroviruses cleave eIF4G isoforms in vivo, thus preventing recruitment of capped mRNA to the 40 S ribosomal subunit (15Bernstein H.D. Sonenberg N. Baltimore D. Mol. Cell. Biol. 1985; 11: 2913-2923Crossref Scopus (76) Google Scholar, 16Brown C.C. Piccone M.E. Mason P.W. McKenna T.S. Grubman M.J. J. Virol. 1996; 70: 5638-5641Crossref PubMed Google Scholar, 17Sakoda Y. Ross-Smith N. Inoue T. Belsham G.J. J. Virol. 2001; 75: 10643-10650Crossref PubMed Scopus (22) Google Scholar, 18Kanno T. Mackay D. Wilsden G. Kitching P. Virus Res. 2001; 80: 101-107Crossref PubMed Scopus (10) Google Scholar). Using purified eIF4F or recombinant eIF4E and eIF4G and recombinant proteinases, several groups demonstrated that eIF4G can be cleaved directly in vitro (19Liebig H.-D. Ziegler E. Yan R. Hartmuth K. Klump H. Kowalski H. Blaas D., W., S. Frasel L. Lamphear B. Rhoads R. Kuechler E. Skern T. Biochemistry. 1993; 32: 7581-7588Crossref PubMed Scopus (164) Google Scholar, 20Kirchweger R. Ziegler E. Lamphear B.J. Waters D. Liebig H.D. Sommergruber W. Sobrino F. Hohenadl C. Blaas D. Rhoads R.E. Skern T. J. Virol. 1994; 68: 5677-5684Crossref PubMed Google Scholar, 21Ventoso I. MacMillan S.E. Hershey J.W. Carrasco L. FEBS Lett. 1998; 435: 79-83Crossref PubMed Scopus (48) Google Scholar, 22Bovee M.L. Lamphear B.J. Rhoads R.E. Lloyd R.E. Virology. 1998; 245: 241-249Crossref PubMed Scopus (42) Google Scholar). Furthermore, the eIF4G·eIF4E complex was shown to be a better substrate for HRV 2Apro than eIF4G alone (23Haghighat A. Svitkin Y. Novoa I. Kuechler E. Skern T. Sonenberg N. J. Virol. 1996; 70: 8444-8450Crossref PubMed Google Scholar). The availability of eIF4E is also important for Lpro cleavage of eIF4GI; the addition of the eIF4E-binding protein (eIF4E-BP1) to a cleavage reaction rendered the eIF4GI refractory to Lpro cleavage (24Ohlmann T. Pain V.M. Wood W. Rau M. Morley S.J. EMBO J. 1997; 16: 844-855Crossref PubMed Scopus (51) Google Scholar). Purification of in vitro cleavage products has allowed the determination of the Lpro cleavage site to be between Gly674 and Arg675; in contrast, rhino- and enterovirus 2Apro cleave between Arg681 and Gly682 (20Kirchweger R. Ziegler E. Lamphear B.J. Waters D. Liebig H.D. Sommergruber W. Sobrino F. Hohenadl C. Blaas D. Rhoads R.E. Skern T. J. Virol. 1994; 68: 5677-5684Crossref PubMed Google Scholar, 25Lamphear B.J. Yan R. Yang F. Waters D. Liebig H.D. Klump H. Kuechler E. Skern T. Rhoads R.E. J. Biol. Chem. 1993; 268: 19200-19203Abstract Full Text PDF PubMed Google Scholar, 26Zamora M. Marissen W.E. Lloyd R.E. J. Virol. 2002; 76: 165-177Crossref PubMed Scopus (44) Google Scholar). eIF4GI (1600 amino acids in total) numbering is according to Byrd et al. (27Byrd M.P. Zamora M. Lloyd R.E. Mol. Cell. Biol. 2002; 22: 4499-4511Crossref PubMed Scopus (72) Google Scholar). Generally, the proteinase concentration present in the infected cell at the time of eIF4G cleavage is much lower than the amount of Lpro or 2Apro required for eIF4G cleavagein vitro (20Kirchweger R. Ziegler E. Lamphear B.J. Waters D. Liebig H.D. Sommergruber W. Sobrino F. Hohenadl C. Blaas D. Rhoads R.E. Skern T. J. Virol. 1994; 68: 5677-5684Crossref PubMed Google Scholar, 22Bovee M.L. Lamphear B.J. Rhoads R.E. Lloyd R.E. Virology. 1998; 245: 241-249Crossref PubMed Scopus (42) Google Scholar, 28Belsham G.J. McInerney G.M. Ross-Smith N. J. Virol. 2000; 74: 272-280Crossref PubMed Scopus (149) Google Scholar). This has led to the suggestion that the picornaviral proteinases may activate latent cellular proteinases which then amplify the response (29Lloyd R.E. Toyoda H. Etchinson D. Wimmer E. Ehrenfeld E. Virology. 1986; 150: 299-303Crossref PubMed Scopus (42) Google Scholar, 30Wyckoff E.E. Lloyd R.E. Ehrenfeld E. J. Virol. 1992; 66: 2943-2951Crossref PubMed Google Scholar). This view has been strengthened at least for PV by the observation that the majority of the eIF4GI-cleaving activity isolated from infected cells does not appear to co-purify with 2Apro-containing fractions (31Bovee M.L. Marissen W.E. Zamora M. Lloyd R.E. Virology. 1998; 245: 229-240Crossref PubMed Scopus (44) Google Scholar). In contrast, recent work seems to indicate that the majority of eIF4GI cleavage in PV-infected cells is indeed performed by 2Apro(26Zamora M. Marissen W.E. Lloyd R.E. J. Virol. 2002; 76: 165-177Crossref PubMed Scopus (44) Google Scholar). A further level of complexity has recently been introduced into the eIF4G cleavage reaction. eIF4GI and eIF4GII are cleaved at different rates in cells infected by PV or HRV14 (32Gradi A. Svitkin Y.V. Imataka H. Sonenberg N. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 11089-11094Crossref PubMed Scopus (271) Google Scholar, 33Svitkin Y.V. Gradi A. Imataka H. Morino S. Sonenberg N. J. Virol. 1999; 73: 3467-3472Crossref PubMed Google Scholar); during infection with HRV2, however, the isoforms are cleaved simultaneously (34Seipelt J. Liebig H.D. Sommergruber W. Gerner C. Kuechler E. J. Biol. Chem. 2000; 275: 20084-20089Abstract Full Text Full Text PDF PubMed Scopus (52) Google Scholar). The reasons for the molecular basis behind this difference remain unclear. We have recently developed an RRL system in which FMDV Lproor HRV 2Apro cleave eIF4GI at concentrations at those found during infection (35Glaser W. Skern T. FEBS Lett. 2000; 480: 151-155Crossref PubMed Scopus (45) Google Scholar). We also showed that deletion of the 18-amino acid CTE of FMDV Lpro, which lies 25 Å or more from the active site, significantly reduced eIF4GI cleavage (36Glaser W. Cencic R. Skern T. J. Biol. Chem. 2001; 276: 35473-35481Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar). However, processing of the viral polyprotein cleavage site presented on a substrate in trans remained unaffected. This suggested that the CTE either stabilizes a conformation of Lpro necessary for eIF4GI cleavage or that it is involved in a direct recognition of eIF4GI. Here, we show that the CTE of FMDV Lpro is indeed part of a domain away from the active site that directly recognizes eIF4GI and define the region recognized by the CTE on eIF4GI to be distinct from that cleaved by Lpro. Furthermore, we demonstrate that HRV2 2Apro can also bind directly to eIF4GI, again at a site away from the amino acids between which it cleaves. Plasmid LbproVP4VP2, which encodes FMDV amino acids 29–364 corresponding to the Lbpro (37Medina M. Domingo E. Brangwyn J.K. Belsham G.J. Virology. 1993; 194: 355-359Crossref PubMed Scopus (126) Google Scholar) form of Lpro (29–201), VP4 (202–286), and part of VP2 (287–364) has been described (36Glaser W. Cencic R. Skern T. J. Biol. Chem. 2001; 276: 35473-35481Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar). Fragments encoding HRV2 2Apro were amplified from nucleotides 3173–3586 of the HRV2 cDNA (19Liebig H.-D. Ziegler E. Yan R. Hartmuth K. Klump H. Kowalski H. Blaas D., W., S. Frasel L. Lamphear B. Rhoads R. Kuechler E. Skern T. Biochemistry. 1993; 32: 7581-7588Crossref PubMed Scopus (164) Google Scholar). The deletions and mutations of Lbproand 2Apro were generated by standard PCR mutagenesis and introduced as EcoRI/XhoI fragments into the plasmid pGEX5X (Amersham Biosciences) as required. Fragments of eIF4GI for in vitro translation were amplified from plasmid pSKHC1 (38Yan R.Q. Rychlik W. Etchison D. Rhoads R.E. J. Biol. Chem. 1992; 267: 23226-23231Abstract Full Text PDF PubMed Google Scholar), which contains the entire eIF4GI cDNA, and cloned as EcoRI/HincII fragments into pBluescriptKS (Stratagene). Rabbit polyclonal antisera raised against either the N terminus (kindly provided by R. Rhoads) or the C terminus (kindly provided by A. Gradi) of eIF4GI were diluted 1:8000 and 1:1000, respectively; a rabbit polyclonal antiserum (also from A. Gradi) against the C terminus of eIF4GII was diluted 1:1000. The mouse monoclonal antibody against eIF4E (Transduction Laboratories) was diluted 1:500. Secondary horseradish peroxidase-conjugated antibodies were diluted 1:10,000 (Bio-Rad). Escherichia coli JM101 or BL21 cells were transformed with plasmids encoding the GST-Lbpro and GST-2Apro fusion proteins or GST alone. To express GST-Lbpro, an overnight culture was diluted 1:10 in 50 ml of medium, isopropyl-1-thio-β-d-galactopyranoside was added to a final concentration of 2 μm, and the cells were incubated at 30 °C for 3 h. For GST-2Apro the overnight culture was diluted 1:10 in 250 ml of medium, the cells were incubated to an A 600 of 0.8 and induced for 3 h at 18 °C by the addition of isopropyl-1-thio-β-d-galactopyranoside to a final concentration of 0.3 μm. The respective proteins were purified on glutathione-agarose resin (Amersham Biosciences) using standard techniques. Glutathione-Sepharose beads coated with GST fusion proteins were incubated in binding buffer (50 mm Tris-Cl pH 7.4, 10 mm EDTA, 150 mm NaCl) with either an aliquot (8 μl) of RRL or with radiolabeled in vitro translated proteins for 2 h at 4 °C. After three washes with binding buffer, bound proteins were eluted by boiling in SDS-PAGE-loading buffer, resolved by SDS-PAGE, and visualized by Western blotting (using enhanced chemiluminescence system of Pierce for detection) or fluorography. In vitro expression of radiolabeled proteins for GST pull-down assays was performed in RRLs (Quick Coupled Transcription/Translation system; Promega) in the presence of [35S]methionine (20 μCi/reaction; Hartmann Analytic, Germany). Labeled proteins were resolved by SDS-PAGE, and gels were dried and exposed to x-ray films. In vitrotranslations in RRLs (Promega) to examine Lbproself-processing and eIF4GI cleavage were performed using in vitro transcribed RNAs as described (35Glaser W. Skern T. FEBS Lett. 2000; 480: 151-155Crossref PubMed Scopus (45) Google Scholar, 36Glaser W. Cencic R. Skern T. J. Biol. Chem. 2001; 276: 35473-35481Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar). FMDV Lpro is the first protein encoded on the polyprotein. Because protein synthesis on the FMDV RNA can initiate at 1 of 2 AUG codons lying 84 nucleotides apart, two forms of Lpro, designated Labpro and Lbpro, are synthesized (39Sangar D.V. Newton S.E. Rowlands D.J. Clarke B.E. Nucleic Acids Res. 1987; 15: 3305-3315Crossref PubMed Scopus (87) Google Scholar); these forms have the same enzymatic properties (37Medina M. Domingo E. Brangwyn J.K. Belsham G.J. Virology. 1993; 194: 355-359Crossref PubMed Scopus (126) Google Scholar). All work described here was performed with the shorter Lbpro (FMDV amino acids 29–201). Cleavage of eIF4GI by Lbpro both in vivo andin vitro is highly efficient (20Kirchweger R. Ziegler E. Lamphear B.J. Waters D. Liebig H.D. Sommergruber W. Sobrino F. Hohenadl C. Blaas D. Rhoads R.E. Skern T. J. Virol. 1994; 68: 5677-5684Crossref PubMed Google Scholar, 28Belsham G.J. McInerney G.M. Ross-Smith N. J. Virol. 2000; 74: 272-280Crossref PubMed Scopus (149) Google Scholar, 35Glaser W. Skern T. FEBS Lett. 2000; 480: 151-155Crossref PubMed Scopus (45) Google Scholar, 40Thomas A.A.M. Scheper G.C. Kleijn M. Deboer M. Voorma H.O. Eur. J. Biochem. 1992; 207: 471-477Crossref PubMed Scopus (22) Google Scholar). To investigate the mechanism of this reaction further, we decided to examine whether a GST-Lbpro fusion protein could bind to the eIF4G isoforms present in an RRL. To construct this fusion protein (GST-LbproC51A), we used an inactive Lbpromutant in which the active site cysteine residue had been replaced by an alanine residue (41Ziegler E. Borman A.M. Kirchweger R. Skern T. Kean K.M. J. Virol. 1995; 69: 3465-3474Crossref PubMed Google Scholar), thus preventing cleavage of the eIF4G isoforms during the experiment. GST-LbproC51A was bound to glutathione-coupled Sepharose (the purity of the GST-LbproC51A fusion protein is shown in Fig.2 A, lane 2) and incubated with RRL as a source of eIF4G isomers. After washing, proteins bound to GST-LbproC51A were resolved by SDS-PAGE and detected by immunoblotting (Fig. 2, B–D). eIF4GI and eIF4GII, which migrate as a series of bands with a relative molecular mass of about 220 kDa (11Gradi A. Imataka H. Svitkin Y.V. Rom E. Raught B. Morino S. Sonenberg N. Mol. Cell. Biol. 1998; 18: 334-342Crossref PubMed Scopus (247) Google Scholar, 42Etchison D. Milburn S.C. Edery I. Sonenberg N. Hershy J.W.B. J. Biol. Chem. 1982; 257: 14806-15810Abstract Full Text PDF PubMed Google Scholar), are retained by the GST-LbproC51A fusion (Fig. 2, B andC, lanes 3). About 20–25% of the eIF4GI in the RRL aliquot was present in the bound fraction. Furthermore, the 25 kDa eIF4E protein, which forms a complex with the eIF4G isomers (43Sonenberg N. Hershey J.W.B. Mathews M.B. Sonenberg N. Translational Control. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N. Y.1996: 245-270Google Scholar), was found in this fraction as well (Fig. 2 D, lane 3). No binding of these proteins was observed to the GST protein alone (Fig. 2, B–D, lanes 2). As can be seen in Fig.3, Lbpro possesses the typical two domain α-helix/β-sheet fold of a papain proteinase. In addition, Lbpro is unique in having an 18-amino acid CTE (44Guarné A. Tormo J. Kirchweger K. Pfistermueller D. Fita I. Skern T. EMBO J. 1998; 17: 7469-7479Crossref PubMed Scopus (117) Google Scholar); removal of this region has been demonstrated to delay cleavage of eIF4GI without affecting cleavage of the viral polyprotein sequence presented in trans (36Glaser W. Cencic R. Skern T. J. Biol. Chem. 2001; 276: 35473-35481Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar). We therefore investigated whether the CTE was responsible for Lbpro binding of eIF4GI. Thus, we expressed GST-LbproC51A fusions carrying different CTE deletions and examined their ability to bind eIF4GI (Fig.4). eIF4GI is still retained by an Lbpro lacking residues 196–201 (LbproC51A-6; Fig. 4 B, lane 3); Lbpro-6 cleaves eIF4GI at wild-type levels (36Glaser W. Cencic R. Skern T. J. Biol. Chem. 2001; 276: 35473-35481Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar). Binding is, however, almost completely abolished by removal of the entire CTE (residues 183–201, LbproC51A-18; Fig. 4 B, lane 4); Lbpro-18 was significantly delayed in eIF4GI cleavage (36Glaser W. Cencic R. Skern T. J. Biol. Chem. 2001; 276: 35473-35481Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar). Furthermore, no binding whatsoever was observed using a mutant with an unrelated sequence in the CTE (GST-Lb pro C51A CTE*, Fig. 4 B, lane 5), which was severely delayed in eIF4GI cleavage (36Glaser W. Cencic R. Skern T. J. Biol. Chem. 2001; 276: 35473-35481Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar). Thus, amino acids 183–195 of Lbpro are necessary for the interaction with eIF4GI. However, they are not sufficient, because a GST fusion protein bearing just the residues 183–195 did not bind eIF4GI (data not shown). This implied that another part of the Lbpro was also required for interaction with eIF4GI. Examination of the Lbpro structure (Fig. 3) showed that the loop between β-sheets 4 and 5 was close to the residues at the beginning of the CTE. Moreover, this loop contains the residue Cys133, which was observed to form an intermolecular disulfide bridge with a neighboring molecule in the protein crystal of Lbpro (44Guarné A. Tormo J. Kirchweger K. Pfistermueller D. Fita I. Skern T. EMBO J. 1998; 17: 7469-7479Crossref PubMed Scopus (117) Google Scholar), suggesting that it was unusually reactive and, thus, perhaps important for Lbpro function. To test whether Cys133 was also involved in binding and cleaving eIF4GI, we substituted Cys133 with serine (this does not affect the structure of the enzyme (45Guarné A. Hampoelz B. Glaser W. Carpena X. Tormo J. Fita I. Skern T. J. Mol. Biol. 2000; 302: 1227-1240Crossref PubMed Scopus (55) Google Scholar)) and first examined the ability of this mutant to carry out self-processing at the Lbpro/VP4 junction and eIF4GI cleavage. This mutant performs self-processing (i.e. intramolecular cleavage) at wild-type levels, because no uncleaved material was observed with either wild-type or mutant enzyme (Fig. 5 A). In addition, this mutant also carries out intermolecular cleavage at the viral polyprotein cleavage site with essentially identical kinetics to those of the wild type (data not shown). However, the ability of the C133S mutant to cleave eIF4GI was clearly impaired (Fig. 5 B, compare the wild-type and mutant panels). 50% cleavage of eIF4GI was observed between 4 and 8 min with the wild-type enzyme; with the mutant enzyme, 50% cleavage was not seen until 30 min. Correspondingly, the ability of this mutant to bind eIF4GI was reduced to that of the mutant lacking 18 amino acids in the CTE (compare Fig. 5 D, lane 3, with Fig.4 B, lane 4), suggesting that replacement of Cys133 had severely impaired the ability of Lbpro to bind eIF4GI and had, thus, retarded cleavage. However, the ability of the mutant to self-process at wild-type levels indicates that its overall structure was intact and the active site was fully functional. Thus, the recognition of the two known substrates of Lbpro, namely the Lbpro/VP4 junction in the polyprotein and eIF4GI, involve different parts of the protein. To identify the region on eIF4GI recognized by the CTE/Cys133 domain of Lbpro, we first cleaved eIF4GI proteolytically into the cleavage products by translating active Lbpro in RRLs (Fig. 6,A–C). The ability of the GST-LbproC51A fusion protein to bind the cleavage products was then examined as described above. The N-terminal cleavage product (cpN, Fig.6 B, lane 2) was retained on the GST-LbproC51A-Sepharose (Fig. 6 B, lane 4). Once again, eIF4E was also bound (data not shown). In contrast, no binding of the C-terminal cleavage product (cpC) was observed (Fig. 6 C, lane 4). Because the affinity of the anti-cpC antiserum appeared low, we repeated the experiment using three times as much Lbpro-treated RRL; nevertheless, cpC was still not detectable (data not shown). Thus, these results imply a direct interaction between Lbpro and cpN (residues 1–674). To further define the binding site on eIF4GI recognized by the Lbpro CTE/Cys133 domain, we synthesized a deletion of cpN containing amino acids 260–674 (Fig.6 A), expressed it in an RRL using [35S]methionine as radioactive label, and incubated the RRL containing the labeled protein with GST-LbproC51A coupled to glutathione-Sepharose. Bound proteins were resolved by SDS-PAGE, and the gels were subjected to fluorography. This protein, encompassing amino acids 260–674 (Fig. 6 D), was retained by the GST-LbproC51A fusion protein. Next, we synthesized a 6-amino acid C-terminal deletion of cpN, 260–669, thus removing the amino acids that comprise the N-terminal part of the Lbpro cleavage site on eIF4GI. Nevertheless, this protein was retained by the GST-LbproC51A fusion protein (Fig. 6 E). In contrast, the fragment 260–640 (Fig.6 F), which terminates 34 amino acids from the cleavage site, was not bound. Thus, the Lbpro CTE/Cys133recognition domain lies between amino acids 640 and 669 of eIF4GI. The eIF4GI fragments shown in Fig. 6, D–F, were all capable of binding to eIF4E (data not shown), demonstrating that they were not incorrectly folded in any way. Ohlmann et al. (24Ohlmann T. Pain V.M. Wood W. Rau M. Morley S.J. EMBO J. 1997; 16: 844-855Crossref PubMed Scopus (51) Google Scholar) show that Lbpro can only cleave eIF4GI when eIF4E is available for binding. Therefore, we investigated whether eIF4E could modulate binding of the eIF4GI fragments. To this end, we synthesized a series of N-terminal deletions of cpN (Fig.7 A). All of the fragments were bound by the GST-LbproC51A fusion protein (Fig. 7,B–D) even though the eIF4GI fragment from amino acid 640 to 820 lacks the eIF4E binding site (609–623 (46Marcotrigiano J. Gingras A.C. Sonenberg N. Burley S.K. Mol. Cell. 1999; 3: 707-716Abstract Full Text Full Text PDF PubMed Scopus (411) Google Scholar, 47Mader S. Lee H. Pause A. Sonenberg N. Mol. Cell. Biol. 1995; 15: 4900-4997Crossref Google Scholar)). However, the retention of this fragment, which was extended relative to the others to enable it to be resolved by SDS-PAGE, was significantly reduced compared with the two fragments containing the eIF4E binding site. The 2Apro of entero- and rhinoviruses, chymotrypsin-like cysteine proteinases structurally unrelated to Lbpro, also cleave the eIF4G isoforms. Thus, we investigated whether the HRV2 2Aprocould bind these isoforms in a similar manner to Lbpro(Fig. 8). Accordingly, we generated a GST-2AproC106A fusion protein (Fig. 8 A) that was again inactivated through the replacement of the active site cysteine by an alanine. Fig. 8 B shows that the 2Aprofusion protein could also bind eIF4GI. Once again, eIF4E was found in the bound material (Fig. 8C). The eIF4GII isoform was also bound by GST-2AproC106A (data not shown). The region on eIF4GI recognized by HRV2 2Apro was examined by once again using an active Lbpro to generate the cleavage products of eIF4GI (Fig. 9,A and B). We again observed that GST-2AproC106A also retained cpN (Fig.9 B) but not cpC (data not shown). That Lbpro cleaves eIF4GI between Gly674 and Arg675, 7 amino acids N-terminal from the 2Aprocleavage site between Arg681 and Gly682, indicates the 2Apro is also not simply binding eIF4GI at its cleavage site. To investigate whether 2Apro recognized the same amino acids as Lbpro, we carried out pull-down experiments using deletions of eIF4GI expressed in RRLs. The 2Apro fusion protein retained the deletion 260–674 (Fig.9 C) but failed to retain the deletions 260–669 or 260–674 (Fig. 9, D and E). This is in contrast to the Lbpro fusion, which also retained the eIF4GI fragment 260–669 (Fig. 6 E). This indicates that the binding sites of the two proteinases are similar but not identical and suggests an important role for eIF4GI amino acids 669–674 for recognition by HRV2 2Apro. The efficient cleavage of eIF4GI during picornaviral infection is an important determinant of virulence. Deletion of the Lpro-coding sequence from FMDV leads to an attenuated phenotype in vivo (16Brown C.C. Piccone M.E. Mason P.W. McKenna T.S. Grubman M.J. J. Virol. 1996; 70: 5638-5641Crossref PubMed Google Scholar); an attenuated strain of SVDV, an enterovirus closely related to coxsackievirus B5, has been described in which the 2Apro can carry out self-processing but is hampered in its ability to process eIF4GI (17Sakoda Y. Ross-Smith N. Inoue T. Belsham G.J. J. Virol. 2001; 75: 10643-10650Crossref PubMed Scopus (22) Google Scholar, 18Kanno T. Mackay D. Wilsden G. Kitching P. Virus Res. 2001; 80: 101-107Crossref PubMed Scopus (10) Google Scholar). Thus, the understanding of the interaction between picornaviral proteinases and the eIF4G isoforms is crucial to an understanding of picornaviral virulence. The data presented here indicate that cleavage of eIF4GI by Lbpro or 2Apro is a multi-step reaction that is more complex than a simple recognition by the enzymes of their respective cleavage sites. Instead, the first step in the reaction appears to be proteinase binding to a distinct domain on eIF4GI separate from the amino acids at the respective cleavage sites on eIF4GI. Furthermore, for the Lbpro at least, we have shown that the active site of the enzyme is not involved in the initial recognition of eIF4GI. Nevertheless, this binding reaction appears to promote cleavage, presumably by bringing the catalytic site into close proximity of the cleavage site. After cleavage, Lbpro or 2Apro must be released from the N-terminal cleavage product to be able to process further eIF4GI molecules. This release of the enzyme may be driven, initially at least, by the low concentration of the eIF4GI cpN/Lbpro species. Furthermore, the fact that not all the eIF4GI was bound in the pull-down experiments indicates that binding is reversible. Thus, the existence of an eIF4GI binding domain on Lbpro, which is separate from the active site of the enzyme, would appear to be a major factor in promoting efficient cleavage of eIF4GI at low concentrations of enzyme. The experiments described here were all performed in RRLs. Thus, it might be argued that the interaction between the proteinases and the eIF4G isomers are not direct but may be mediated by RNA or a bridging protein. The RRLs were all previously treated with micrococcal nuclease, which would argue against the involvement of RNA. We have also repeated some of these experiments using a nuclease-treated S30E. coli extract; binding was observed in the presence of eIF4E between the GST-LbproC51A fusion protein and in vitro translated eIF4GI fragments. This argues against a bridging protein, because it seems unlikely that such a specific protein is conserved between mammals and bacteria. How do the CTE of Lbpro and residue Cys133 act together to form the eIF4GI binding domain? The side chains of Cys133 and Leu188 from the CTE are within 8.5 Å, indicating that they, together with other residues of the CTE and the β4-β5 loop (Fig. 3), could bind to a single domain on eIF4GI between residues 640 and 669. It is of interest that Cys133is conserved in all FMDV sequences so far available, except for a number of recently isolated FMDV serotypes from East and South Africa (48van Rensburg H. Haydon D. Joubert F. Bastos A. Heath L. Nel L. Gene. 2002; 289: 19-29Crossref PubMed Scopus (52) Google Scholar). In these strains, threonine is found at residue 133; serine has not been described. It will be of interest to examine the eIF4G cleavage behavior of these serotypes. The CTE represents the most divergent part of Lbpro in terms of sequence and structure. Nevertheless, it appears to be playing a major role in substrate binding and specificity; moreover, it is noteworthy that this region is completely absent in papain, the prototype enzyme of the family of cysteine proteinases. Papain does not produce defined cleavage fragments of eIF4GI. 2B. Hampoelz and T. Skern, unpublished data. Ohlmann et al. (24Ohlmann T. Pain V.M. Wood W. Rau M. Morley S.J. EMBO J. 1997; 16: 844-855Crossref PubMed Scopus (51) Google Scholar) have previously proposed that the substrate for Lbpro is the eIF4GI·eIF4E complex. They observed that the addition of the eIF4E-binding protein eIF4E-BP1 before the addition of Lbpro prevented cleavage of eIF4GI, presumably because the eIF4GI·eIF4E complex could not be formed. For HRV2 2Apro, Haghighat et al. (23Haghighat A. Svitkin Y. Novoa I. Kuechler E. Skern T. Sonenberg N. J. Virol. 1996; 70: 8444-8450Crossref PubMed Google Scholar) demonstrate using purified recombinant proteins that the eIF4G·4E complex was cleaved much more efficiently than eIF4GI alone. The addition of exogenous eIF4E to a molar ratio of 4:1 increased the cleavage efficiency by at least 50-fold. Moreover, biochemical and genetic data have been obtained showing that yeast eIF4GI undergoes an unfolded to folded transition on binding eIF4E (49Hershey P.E. McWhirter S.M. Gross J.D. Wagner G. Alber T. Sachs A.B. J. Biol. Chem. 1999; 274: 21297-21304Abstract Full Text Full Text PDF PubMed Scopus (76) Google Scholar) as suggested previously by Ohlmann et al. (24Ohlmann T. Pain V.M. Wood W. Rau M. Morley S.J. EMBO J. 1997; 16: 844-855Crossref PubMed Scopus (51) Google Scholar). This may be one of the reasons why the eIF4G·4E complex is the preferred substrate for picornaviral proteinases. It was thus surprising that Lbprowas able, albeit weakly, to bind the eIF4GI fragment 640–820, which does not contain the eIF4E binding site. This discrepancy can be explained by proposing that amino acids 640–669 in full-length eIF4GI must attain a particular conformation to be recognized by Lbpro. This conformation can apparently be achieved by the small fragment of amino acids 640–820 alone; the apparently weaker binding of this fragment indeed suggests that it is not all in the correct conformation for Lbpro recognition. However, in the 1600- amino acid full-length eIF4GI molecule, binding of eIF4E may substantially increase binding and cleavage by Lbpro. The binding interaction between HRV2 2Apro and eIF4GI illustrates that the recognition of a specific binding site distinct from the cleavage is not restricted to the FMDV Lbpro. Nevertheless, the binding sites are not identical; the HRV2 2Apro site extends to include eIF4GI amino acid 674, whereas the C-terminal boundary of the Lbpro site was defined as amino acid 669. Given that the two proteinases have completely different tertiary structures, it is perhaps not surprising that the binding sites are not identical. Furthermore, the respective binding sites may serve to dock the proteinase at a particular distance from the cleavage site. Because the cleavage sites lie seven amino acids apart, this may also apply constraints to the binding sites, which require them to be different. The binding domain on the HRV2 2Apro remains unclear despite the availability of its structure (50Petersen J.F. Cherney M.M. Liebig H.D. Skern T. Kuechler E. James M.N. EMBO J. 1999; 18: 5463-5475Crossref PubMed Scopus (98) Google Scholar). Mutants of PV 2Apro have been described in the literature that impair or abrogate eIF4GI cleavage but which do not affect self-processing (15Bernstein H.D. Sonenberg N. Baltimore D. Mol. Cell. Biol. 1985; 11: 2913-2923Crossref Scopus (76) Google Scholar, 51Yu S.F. Lloyd R.E. Virology. 1991; 182: 615-625Crossref PubMed Scopus (66) Google Scholar). These mutants do not, however, cluster on any particular part of the 2Apro molecule, and it remains to be seen whether direct binding to eIF4GI is affected. The eIF4G isoforms act as protein adaptors during the initiation of protein synthesis (4Morley S.J. Curtis P.S. Pain V.M. RNA (N. Y.). 1997; 3: 1085-1104PubMed Google Scholar, 5Hentze M.W. Science. 1997; 275: 500-501Crossref PubMed Scopus (193) Google Scholar). Binding sites on eIF4G isoforms have been described for several proteins; these include other proteins of the eIF4 group (eIF4E, the cap-binding protein, and eIF4A, the RNA helicase), eIF3, the poly(A)-binding protein, and mitogen-activated protein kinase-interacting kinase-1 (MNK1) (2Gingras A.C. Raught B. Sonenberg N. Annu. Rev. Biochem. 1999; 68: 913-963Crossref PubMed Scopus (1762) Google Scholar). In addition, a viral protein, the NSP3 protein of human rotavirus, has been shown to bind eIF4G at the same site as poly(A)-binding protein (PABP) and to be able to displace PABP from its binding site (52Piron M. Vende P. Cohen J. Poncet D. EMBO J. 1998; 17: 5811-5821Crossref PubMed Scopus (302) Google Scholar). However, none of these proteins have been reported to bind to the region on eIF4GI (or to the equivalent region of eIF4GII, which is about 65% identical to this region of eIF4GI) identified here as being recognized by the picornaviral proteinases. Because viral proteins often evolve to use cellular binding sites, it seems feasible that a cellular protein may exist that can recognize this region. However, data base searches did not reveal any cellular proteins with a sequence related to the CTE of Lbpro (data not shown). The Lbpro of FMDV is among the smallest papain-like enzymes. Nevertheless, Lbpro has evolved a novel site away from the substrate binding pocket to accelerate cleavage of eIF4G isoforms. Once again, the pressure to compress viral genomes has resulted in the evolution of a versatile and multifunctional protein. We thank D. Blaas and J. Seipelt for critical reading and A. Gradi, R. Rhoads, N. Sonenberg, and Y. Svitkin for reagents.
Referência(s)