Portal de Periódicos da CAPES

Proteolytic Processing and Translation Initiation

2004; Elsevier BV; Volume: 279; Issue: 16 Linguagem: Inglês

10.1074/jbc.m312391200

1083-351X

Sylvain de Breyne, Romaine Stalder Monney, Joseph Curran,

Animal Virus Infections Studies

The four Sendai virus C-proteins (C′, C, Y1, and Y2) represent an N-terminal nested set of non-structural proteins whose expression modulates both the readout of the viral genome and the host cell response. In particular, they modulate the innate immune response by perturbing the signaling of type 1 interferons. The initiation codons for the four C-proteins have been mapped in vitro, and it has been proposed that the Y proteins are initiated by ribosomal shunting. A number of mutations were reported that significantly enhanced Y expression, and this was attributed to increased shunt-mediated initiation. However, we demonstrate that this arises due to enhanced proteolytic processing of C′, an event that requires its very N terminus. Curiously, although Y expression in vitro is mediated almost exclusively by initiation, Y proteins in vivo can arise both by translation initiation and processing of the C′ protein. To our knowledge this is the first example of two apparently independent pathways leading to the expression of the same polypeptide chain. This dual pathway explains several features of Y expression. The four Sendai virus C-proteins (C′, C, Y1, and Y2) represent an N-terminal nested set of non-structural proteins whose expression modulates both the readout of the viral genome and the host cell response. In particular, they modulate the innate immune response by perturbing the signaling of type 1 interferons. The initiation codons for the four C-proteins have been mapped in vitro, and it has been proposed that the Y proteins are initiated by ribosomal shunting. A number of mutations were reported that significantly enhanced Y expression, and this was attributed to increased shunt-mediated initiation. However, we demonstrate that this arises due to enhanced proteolytic processing of C′, an event that requires its very N terminus. Curiously, although Y expression in vitro is mediated almost exclusively by initiation, Y proteins in vivo can arise both by translation initiation and processing of the C′ protein. To our knowledge this is the first example of two apparently independent pathways leading to the expression of the same polypeptide chain. This dual pathway explains several features of Y expression. The Sendai virus P/C mRNA has become a paradigm for expressional elasticity, expressing six different polypeptide chains by ribosomal choice (namely, C′, P, C, Y1, Y2, and X) (1Curran J. Latorre P. Kolakofsky D. Semin. Virol. 1998; 8: 351-357Crossref Scopus (18) Google Scholar). The P protein initiates at the second start site (AUG104). It is an essential cofactor of the viral polymerase (2Curran J. Kolakofsky D. Adv. Virus Res. 1999; 54: 403-422Crossref PubMed Google Scholar). The first start site is an unusual ACG initiation codon at position 81 (ACG81) that gives rise to the C′ protein (3Curran J. Kolakofsky D. EMBO J. 1988; 7: 245-251Crossref PubMed Scopus (130) Google Scholar, 4Gupta K.C. Patwardhan S. J. Biol. Chem. 1988; 263: 8553-8556Abstract Full Text PDF PubMed Google Scholar). This is a member of an N-terminally nested set of four proteins referred to as the “C-proteins” whose open reading frame (ORF) 1The abbreviations used are: ORF, open reading frame; RRL, rabbit reticulocyte lysate; SeV, Sendai virus; HA, hemagglutinin; HCV, hepatitis C virus; IRES, internal ribosome entry site; CrPV, cricket paralysis virus; GFP, green fluorescent protein; PBS, phosphate-buffered saline; eIF, eukaryotic initiation factor; GMP-PNP, guanosine 5′-(β,γ-imino)triphosphate; wt, wild type; Stat, signal transducers and activators of transcription. overlaps that of P. The second member of this group, the C protein, initiates at AUG114 and is accessed by leaky ribosomal scanning (5Kozak M. Cell. 1986; 44: 283-292Abstract Full Text PDF PubMed Scopus (3589) Google Scholar). Its good Kozak consensus signal is consistent with its high expression levels in virally infected cells. It is also the major translation product when the P/C mRNA is expressed transiently in mammalian cells or in rabbit reticulocyte lysates (RRLs). The remaining members of this group are the Y proteins (Y1 and Y2). Their AUG start codons were mapped in vitro to positions 183 and 201 (6Patwardhan S. Gupta K.C. J. Biol. Chem. 1988; 263: 4907Abstract Full Text PDF PubMed Google Scholar). The C-proteins impact on both the readout of the viral genome (7Cadd T. Garcin D. Tapparel C. Itoh M. Homma M. Roux L. Curran J. Kolakofsky D. J. Virol. 1996; 70: 5067-5074Crossref PubMed Google Scholar, 8Curran J. Marq J.B. Kolakofsky D. Virology. 1992; 189: 647-656Crossref PubMed Scopus (151) Google Scholar) and the host cell response to viral infection. In particular, they function to disrupt type 1 interferon signaling pathways (9Garcin D. Curran J. Kolakofsky D. J. Virol. 2000; 74: 8823-8830Crossref PubMed Scopus (78) Google Scholar, 10Garcin D. Marq J.B. Goodbourn S. Kolakofsky D. J. Virol. 2003; 77: 2321-2329Crossref PubMed Scopus (47) Google Scholar). Although initiation from the first three start sites (ACG81/C′, AUG104/P, and AUG114/C) is readily explained by leaky scanning, results suggest that ribosomes can access the Y start codons by discontinuous scanning or shunting (11Curran J. Kolakofsky D. EMBO J. 1989; 8: 521-526Crossref PubMed Scopus (62) Google Scholar, 12Latorre P. Kolakofsky D. Curran J. Mol. Cell. Biol. 1998; 18: 5021-5031Crossref PubMed Scopus (93) Google Scholar). The seminal observation in this work was that changing the C′ ACG codon to AUG (referred to as AUG81) ablated expression of the P and C proteins but not Y1/Y2. In discontinuous scanning, ribosomes are loaded via the 5′ cap and then at a defined donor site translocate to an internal acceptor site located close to the start codon (13Gale Jr., M. Tan S.L. Katze M.G. Microbiol. Mol. Biol. Rev. 2000; 64: 239-280Crossref PubMed Google Scholar). The first description of a eukaryotic shunt was the SeV X protein (14Curran J. Kolakofsky D. EMBO J. 1988; 7: 2869-2874Crossref PubMed Scopus (68) Google Scholar). This is initiated from an AUG codon more than 1,500 nucleotides from the 5′ end of the P/C mRNA in a manner that is cap-dependent. However, the most studied shunts are those on the cauliflower mosaic virus 35 S RNA (15Futterer J. Kiss-Laszlo Z. Hohn T. Cell. 1993; 73: 789-802Abstract Full Text PDF PubMed Scopus (179) Google Scholar, 16Park H.S. Himmelbach A. Browning K.S. Hohn T. Ryabova L.A. Cell. 2001; 106: 723-733Abstract Full Text Full Text PDF PubMed Scopus (127) Google Scholar) and the adenovirus late mRNAs (17Yueh A. Schneider R.J. Genes Dev. 1996; 10: 1557-1567Crossref PubMed Scopus (168) Google Scholar, 18Yueh A. Schneider R.J. Genes Dev. 2000; 14: 414-421PubMed Google Scholar). We have been studying the expression of the Y proteins. A number of mutations were previously characterized that significantly enhanced Y levels in vivo, a result we attributed to increased initiation (19de Breyne S. Simonet V. Pelet T. Curran J. Nucleic Acids Res. 2003; 31: 608-618Crossref PubMed Scopus (15) Google Scholar). However, in this report we demonstrate that this “up-phenotype” is actually the consequence of a product-precursor relationship between the Y and C′ proteins. Curiously, although Y expression in vitro is mediated almost exclusively by initiation at the Y1183 and Y2203 AUG codons, Y proteins in vivo can arise both by translation initiation and processing of the C′ protein. DNA Constructions—All the constructs were made in the AUG81 background starting from the pBS SK AUG81, pBS SK Δ9, or pBS SK Δ9Y1Y2UGU clones (19de Breyne S. Simonet V. Pelet T. Curran J. Nucleic Acids Res. 2003; 31: 608-618Crossref PubMed Scopus (15) Google Scholar). Unless otherwise stated constructs carried the triple HA epitope tag fused to the C terminus of the C ORF. All changes were performed by PCR. The pEBS-PL episomal Epstein-Barr virus-based mammalian expression vector carries an SRα promoter and a hygromycin selection cassette (20Takebe Y. Seiki M. Fujisawa J. Hoy P. Yokota K. Arai K. Yoshida M. Arai N. Mol. Cell. Biol. 1988; 8: 466-472Crossref PubMed Google Scholar). Clones were transferred from pBS KS to pEBS-PL as SacI/XbaI fragments. The HCV IRES2b was removed as an NcoI/BspHI fragment from the pTZ18R plasmid clone (21Collier A.J. Tang S. Elliott R.M. J. Gen. Virol. 1998; 79: 2359-2366Crossref PubMed Scopus (69) Google Scholar) and introduced into the NcoI site of pBS SK AUG81, Δ9, and Δ9Y1Y2UGU. The triple HA tag was then fused to the C terminus by PCR. The cricket paralysis virus (CrPV) IRES was excised as an EcoRI/NcoI fragment from the clone pRdEMCV-CrPV-F and introduced into the NcoI site of pBS SK AUG81, Δ9Y1Y2UGU, and Δ9. This positioned the A site of the ribosome 18 nucleotides upstream of AUG81. To position the AUG81 in the ribosomal A site, we modified the IRES sequence, introducing an NcoI site at the translational start position and a compensating second site change that restored the critical pseudoknot of the IRES (22Wilson J.E. Pestova T.V. Hellen C.U. Sarnow P. Cell. 2000; 102: 511-520Abstract Full Text Full Text PDF PubMed Scopus (344) Google Scholar). This permitted fusion of the C′ AUG81 via its NcoI site. The bicistronic constructs were built starting from a pGEM4 plasmid clone containing the SeV M gene (23Mottet G. Muller V. Roux L. J. Gen. Virol. 1999; 80: 2977-2986Crossref PubMed Scopus (16) Google Scholar). This clone contained a unique EcoRI site downstream of the M gene into which was inserted the HCV-AUG81 as an EcoRI fragment. The GFP tag was amplified by PCR from pEBS-PL GFP (24Bontron S. Lin-Marq N. Strubin M. J. Biol. Chem. 2002; 277: 38847-38854Abstract Full Text Full Text PDF PubMed Scopus (60) Google Scholar) using oligonucleotides containing XbaI (5′) and NotI (3′) restriction sites. This was then used to replace the triple HA tag (which is fused to the end of the C ORF via an XbaI site) in cDNA clones expressing C′ (AUG81) and C. Cell Culture, Transient Transfection, and Metabolic Labeling—Cell culture, infections, transfections, and metabolic labeling were performed as previously described in de Breyne et al. (19de Breyne S. Simonet V. Pelet T. Curran J. Nucleic Acids Res. 2003; 31: 608-618Crossref PubMed Scopus (15) Google Scholar). Cells were transfected with the pEBS-PL clones using FuGENE (Roche Applied Science) according to the manufacturer's instructions. Cells were infected with SeV (strain Z) at a multiplicity of infection of 4. Antibodies and Indirect Immunofluorescence Assay—HeLa cells seeded on coverslips were transfected with pEBS-PL C′-GFP, pEBS-PL C-GFP, or both pEBS-PL C′-GFP and pEBS-PL C-HA as described in Ref. 25Donze O. Picard D. Nucleic Acids Res. 2002; 30: e46Crossref PubMed Scopus (242) Google Scholar. After 24 h the coverslips were washed in 150 mm NaCl, 100 mm Tris, pH 7.5, and the cells were fixed and permeabilized in 3% formal-dehyde, PBS during 20 min and 0.05% saponin, PBS during 5 min. They were then washed three times in PBS and blocked for 30 min at room temperature with PBS containing 1% bovine serum albumin. The anti-HA and anti-galactosyltransferase mouse antibodies were used at dilutions of 1:100 and 1:1000 in PBS, respectively. Coverslips were incubated with the antibodies for 20 min at room temperature. After three PBS washes, the secondary antibody conjugated with Alexa-568 was added at a dilution of 1:200 for 20 min at room temperature. Microscopic analyses were performed on a confocal laser scan fluorescence inverted microscope (LSM 410, Zeiss). Each time, the two channels were recorded either together or independently to ensure the absence of interference between the channels. In Vitro Transcription and Translation—Run-off capped transcripts were synthesized with T7 RNA polymerase in 800 μm ATP/CTP/UTP, 400 μm GTP, and 2 mm m7GpppG cap analogue (New England Biolabs). Capped mRNAs (50 μg/ml) were translated in a RRL (Promega) in the presence of 0.5 mm MgOAc, 75 mm KCl, a 20 μm concentration of each amino acid (except methionine), and 0.5 mCi/ml 35S Translabel. In the experiments using eIF4A, 2 μg of eIF4Awt, eIF4AR362Q, or elution buffer (150 mm imidazole, 110 mm KCl, 20 mm HEPES, pH 7.4) was added (26Pause A. Methot N. Svitkin Y. Merrick W.C. Sonenberg N. EMBO J. 1994; 13: 1205-1215Crossref PubMed Scopus (327) Google Scholar). Purification of eIF4Awt and eIF4AR362Q—The eIF4Awt and eIF4AR362Q clones were amplified by PCR and cloned downstream of the His6 tag in the bacterial expression vector pT7-7 His6. Transformed Escherichia coli strain BL21 were induced for 2 h with 0.4 mm isopropyl-1-thio-β-d-galactopyranoside when the culture reached an A600 of 0.5. Rifampicin (400 μg/ml) was added, and the incubation continued for a further 2 h. Cells were lysed in 300 mm NaCl, 20 mm HEPES, pH 7.4, 1% Nonidet P-40 by sonication, and the proteins were isolated on a Talon affinity column (Clontech) following the manufacturer's protocols. Analysis of Preinitiation Complexes (Toeprinting)—Primer extension analysis to position the 48 S ribosomal complex was performed as outlined in Ref. 27Kozak M. Nucleic Acids Res. 1998; 26: 4853-4859Crossref PubMed Scopus (60) Google Scholar. Briefly capped RNA (250 ng) was incubated with a 32P-labeled oligonucleotide corresponding to nucleotides 261–282 (2 × 106 cpm) in a final volume of 5 μl, heated to 50 °C, and then allowed to cool. The mixture was added to 8 μl of RRL containing either 1 mm GMP-PNP or 5 mm EDTA. This was incubated at 30 °C for 5 min and then transferred onto ice. The lysate containing the EDTA was adjusted to 5 mm MgCl2. Twenty microliters of toeprint buffer (50 mm Tris, pH 8.3, 75 mm KCl, 3 mm MgCl2,10mm dithiothreitol, 250 μm dNTPs, 1,000 units/ml Rnasin) and 100 units of Superscript II reverse transcriptase (Invitrogen) were then added, and the lysates were incubated at 37 °C for 30 min. Reactions were terminated by adding 30 μl of phenol-Tris/EDTA. Primer extension products were analyzed on an 8% polyacrylamide-urea gel alongside a sequence ladder. Enhanced Expression of the Y Proteins in the Δ9 Background Requires Expression of C′—In our previous report we described mutants within the AUG81 background that significantly increased (∼10-fold) the expression of the Y proteins, a pheno-type initially attributed to an increase in shunt-mediated translational initiation (19de Breyne S. Simonet V. Pelet T. Curran J. Nucleic Acids Res. 2003; 31: 608-618Crossref PubMed Scopus (15) Google Scholar). One of these, referred to as Δ9, removed nucleotides 189–197 (three codons) between the Y1 and Y2 AUGs. This increased expression was essentially independent of the nature of the Y1/Y2 codons (e.g. Δ9Y1Y2UGU also gave high levels of Y expression). The earlier studies had used a vaccinia-T7 transient expression system, but these phenotypes were conserved with the mammalian expression vector pEBS-PL, precluding the possibility that the viral infection had impacted on the expression patterns observed (Fig. 1, A and B). It should be noted that in the Δ9 background the Y1/Y2 proteins generally co-migrate on SDS-polyacrylamide gels (19de Breyne S. Simonet V. Pelet T. Curran J. Nucleic Acids Res. 2003; 31: 608-618Crossref PubMed Scopus (15) Google Scholar). With the objective of ensuring that the only products of the C ORF would be the Y proteins, we introduced a nucleotide down-stream of the initiation codon for C′ (at position +5 relative to the AUG81, i.e. nucleotide 85) within the Δ9 background, thereby creating a frameshift (Fig. 1A). We anticipated that such a construct would only express Y1/Y2 and would retain the Δ9 up-phenotype with regard to Y levels. Indeed, when expressed in mammalian cells, C′ expression was totally lost (Fig. 1C, lanes 4 and 5). However, Y protein levels dropped dramatically, approaching those observed in the AUG81 background. In addition, a second band was observed that co-migrated with the C protein (initiated from AUG114), a result that was somewhat surprising since the good Kozak context of the AUG81 start site had not been changed. Confirmation that this was the C protein came from mutating the AUG114 to GCG (Fig. 1C, lanes 6 and 7). The addition of 2 nucleotides also destroyed the phenotype characteristic of Δ9 (data not shown). However, addition of 3 nucleotides restored the C ORF and generated a slightly slower migrating C′ protein (Fig. 1C, lanes 8 and 9). There was no obvious expression of C, and Y protein levels were once again enhanced, i.e. the Δ9 phenotype had been restored. A Product-Precursor Relationship between Y and C′—Pulsechase experiments performed on virally infected cells had previously failed to provide unambiguous evidence for a product-precursor relationship between C′/C and the Y proteins. 2J. Curran, unpublished. However, the results outlined above demonstrated that the enhanced expression of the Y proteins observed in the Δ9 background was coupled to the expression of C′. We therefore performed a series of pulse-chase experiments in cells expressing the AUG81 and Δ9 clones (Fig. 2) and quantitated the proteins using a PhosphorImager. In the AUG81-transfected cells (Fig. 2A), the C′ levels decayed during the 5-h chase (t½ = 180 min), and there was a small but reproducible rise in the Y proteins. This effect was more pronounced in the Δ9 transfection in which the Δ9C′ protein turned over more rapidly (t½ = 69 min), whereas the Y proteins accumulated during the 1st h of the chase and then decayed with kinetics very similar to Δ9C′ (Fig. 2B). The presence of a product-precursor relationship between Y and Δ9C′ is most clearly demonstrated during a shorter chase period (Fig. 2C). In this latter experiment the Y1 and Y2 proteins were clearly discernable, and both accumulated at similar rates during the 60-min chase. The quantification of the gels also indicated that not all the Δ9C′ protein turned over to generate Y proteins. The results above indicate a product-precursor relationship between C′ and Y. However, is this the case in SeV-infected cells? Pulse-chase experiments were reperformed in virally infected cells, and the products of the C ORF were quantitated on a PhosphorImager (Fig. 2D). The C′ protein decayed with a half-life of 90 min, whereas the Y proteins accumulated slowly throughout the 5-h chase. Y Protein Expression in Vitro Is Exclusively via Initiation— The Y1/Y2 initiation codons were initially mapped on the P/C gene by mutating AUG186 and AUG201 and expressing the mRNAs in RRLs (6Patwardhan S. Gupta K.C. J. Biol. Chem. 1988; 263: 4907Abstract Full Text PDF PubMed Google Scholar). Such changes ablated all Y expression. Furthermore toeprinting analysis, which maps the position of the 48 S preinitiation complex on the mRNA, demonstrated the presence of ribosomes at the Y2 AUG (generally the major Y protein expressed in RRLs (11Curran J. Kolakofsky D. EMBO J. 1989; 8: 521-526Crossref PubMed Scopus (62) Google Scholar) (Fig. 3A)). In vitro translation of the P/C and AUG81 mRNAs yielded protein profiles similar to those observed in mammalian cells (11Curran J. Kolakofsky D. EMBO J. 1989; 8: 521-526Crossref PubMed Scopus (62) Google Scholar). However, in vitro translation of the Δ9 mRNA did not reproduce the enhanced expression of the Y proteins observed in vivo (compare Fig. 1B, lanes 3–6, with Fig. 3B, lanes 2 and 3), and Y expression remained AUG codon-dependent even when S10 HeLa extracts were added to the translation mixture (data not shown), suggesting that the RRL did not simply lack soluble cellular factors. Since Y expression was driven by translational initiation we decided to examine to what extent ribosome scanning was required. For this we obtained clones expressing eIF4A and a dominant negative form of eIF4A carrying the mutation R362Q (26Pause A. Methot N. Svitkin Y. Merrick W.C. Sonenberg N. EMBO J. 1994; 13: 1205-1215Crossref PubMed Scopus (327) Google Scholar). eIF4A is an RNA helicase that forms part of the eIF4F complex (28Imataka H. Sonenberg N. Mol. Cell. Biol. 1997; 17: 6940-6947Crossref PubMed Scopus (240) Google Scholar, 29Lamphear B.J. Kirchweger R. Skern T. Rhoads R.E. J. Biol. Chem. 1995; 270: 21975-21983Abstract Full Text Full Text PDF PubMed Scopus (471) Google Scholar, 30Pause A. Sonenberg N. EMBO J. 1992; 11: 2643-2654Crossref PubMed Scopus (530) Google Scholar). It functions to unwind the RNA during scanning. The eIF4AR362Q mutant effects in a dominant negative fashion initiation mediated via the 5′ cap but not that mediated via the HCV IRES (31Pestova T.V. Shatsky I.N. Fletcher S.P. Jackson R.J. Hellen C.U. Genes Dev. 1998; 12: 67-83Crossref PubMed Scopus (629) Google Scholar). Both wt and mutant proteins were His-tagged and purified from E. coli. Capped transcripts from the bicistronic clone M-IRESHCV-AUG81 (expression of the M protein is 5′ cap-dependent, whereas C′ is driven by the HCV IRES; Fig. 3C) were added to a RRL supplemented with 2 μg of either HiseIF4Awt or HiseIF4AR362Q (26Pause A. Methot N. Svitkin Y. Merrick W.C. Sonenberg N. EMBO J. 1994; 13: 1205-1215Crossref PubMed Scopus (327) Google Scholar). As a control, an equivalent volume of buffer was added. Addition of HiseIF4Awt to a RRL programmed with M-IRESHCV-AUG81 mRNA had virtually no effect on the expression of M, C′, Y1, and Y2 (Fig. 3C). However, addition of HiseIF4AR362Q dramatically reduced expression from the first cistron (M) while only marginally effecting expression from the second (C′, Y1, and Y2), confirming that ribosome recruitment by the HCV IRES does not require functional eIF4A (31Pestova T.V. Shatsky I.N. Fletcher S.P. Jackson R.J. Hellen C.U. Genes Dev. 1998; 12: 67-83Crossref PubMed Scopus (629) Google Scholar). To facilitate interpretation the image was quantitated on a PhosphorImager (Fig. 3C). The addition of HiseIF4AR362Q reduced both the M/C′ and M/Y protein ratios to a similar extent relative to the control (between 5- and 6-fold), whereas the C′/Y ratio was unchanged. We conclude that both ribosomal loading and subsequent initiation at the Y start codons are largely independent of eIF4A. The Y Proteins in Vivo: Protein Processing Versus Translation Initiation—The Y proteins in vivo can arise by processing of the C′ protein, but can they also be expressed via translation initiation as observed in vitro? To address this question we exploited the frameshift approach outlined earlier in combination with ribosome recruitment via the HCV IRES. The former assures that no C′ protein is expressed (i.e. no cleavage precursor), whereas the IRES eliminates the possibility that the Y start sites are accessed by leaky scanning (31Pestova T.V. Shatsky I.N. Fletcher S.P. Jackson R.J. Hellen C.U. Genes Dev. 1998; 12: 67-83Crossref PubMed Scopus (629) Google Scholar, 32Hellen C.U. Sarnow P. Genes Dev. 2001; 15: 1593-1612Crossref PubMed Scopus (805) Google Scholar). The IRES was positioned at the C′ AUG81 codon in a Δ9 background. A single nucleotide was inserted at position 103 (referred to as +1*, Fig. 4A) altering the ORF of the IRES-mediated translation product. This insertion was downstream of that presented earlier because the +5 position relative to the AUG has been reported to influence initiation at least that which is 5′ cap-dependent (33Boeck R. Kolakofsky D. EMBO J. 1994; 13: 3608-3617Crossref PubMed Scopus (81) Google Scholar, 34Grunert S. Jackson R.J. EMBO J. 1994; 13: 3618-3630Crossref PubMed Scopus (99) Google Scholar). The protein products expressed from this construct (HCV-C′+1*Δ9) were identical to those observed when the translation was 5′ cap-mediated (compare Fig. 4B, lanes 5 and 6 with lane 4). We were particularly surprised to observe a C protein given that the HCV IRES does not permit leaky scanning (31Pestova T.V. Shatsky I.N. Fletcher S.P. Jackson R.J. Hellen C.U. Genes Dev. 1998; 12: 67-83Crossref PubMed Scopus (629) Google Scholar). This suggests that either leaky scanning did occur in this particular context or that on this mRNA C was initiated by shunting. The Y proteins expressed in the HCV-C′+1*Δ9 background were lost when the Y1/Y2 codons were changed (Fig. 4B, lanes 7 and 8). Similar results were obtained when the experiment was performed in the AUG81 background (Fig. 4B, lanes 9–14), demonstrating that Y expression in vivo can be mediated via translation initiation in an AUG-dependent manner. This, in turn, indicates that the Y proteins observed in vivo from constructs in which the AUG183 and AUG201 have been mutated are generated exclusively via the processing of C′. Curiously the shorter C protein does not give rise to Y proteins when these AUG codons are changed (even in the Δ9 background; Fig. 4B, lanes 7 and 8 and lanes 11 and 12), indicating that although C turns over rapidly (t½ = 140 min, Fig. 2D), it is not processed to generate Y proteins. The in vivo and in vitro studies outlined above leave us with an apparent paradox since both propose different models to explain the expression of the same proteins. As an initial step at trying to resolve this we examined to what extent the proteins generated by processing of the C′ in vivo actually corresponded to the Y proteins initiated at AUG183 and AUG201, i.e. did they share a common N terminus. The approach we used took advantage of the fact that there is normally only one cysteine codon in the C ORF, and it is just downstream of Y2. Δ9Nsi is a Δ9 background in which the CG nucleotide pair just upstream of the Y2 AUG has been inverted thereby generating an NsiI restriction site (19de Breyne S. Simonet V. Pelet T. Curran J. Nucleic Acids Res. 2003; 31: 608-618Crossref PubMed Scopus (15) Google Scholar). This change also introduced a new cysteine codon just upstream of Y2 (Fig. 5A; the reason for its selection is indicated below). The cysteine codon downstream of Y2 was changed to tyrosine in a background in which the Y1/Y2 start codons were either AUG (Δ9*NsiMet) or UGU (Δ9*NsiCys). Transfected cells were metabolically labeled with either 35S Translabel or [35S]cysteine. Upon labeling with the former, both Δ9*NsiMet and Δ9*NsiCys gave the high Y expression characteristic of Δ9 (Fig. 5B, lanes 2–5). In the Δ9*NsiMet transfection, the Y proteins co-migrated (lanes 2 and 3), whereas with Δ9*NsiCys the two proteins were clearly separated (lanes 4 and 5). This change in Y protein mobility (in this particular case Y1) is probably the result of alterations in the primary sequence (arising from the methionine/cysteine mutations), changes that have also affected the mobility of the Δ9C′ proteins (compare lanes 2 and 3 with lanes 4 and 5). Although the Δ9Nsi background introduced a second cysteine codon upstream of Y2 it was selected for this experiment because in the Δ9*NsiCys clone the Y1 and Y2 proteins were clearly resolved (this is unfortunately not the case in the normal Δ9 construct). This was critical for the interpretation of the labeling patterns (see below). When labeled with [35S]cysteine, bands corresponding to C′ and Y1 were observed in the Δ9*NsiMet-transfected cells (Fig. 5B, lanes 7 and 8). The Y2 protein would not be detected since it contains no cysteine. In cells transfected with Δ9*NsiCys, the C′, Y1, and Y2 proteins were clearly visible (Fig. 5B, lanes 9 and 10). The radiolabeling of the faster migrating Y2 protein with [35S]cysteine positions its N terminus at either the precise site of the Y2 initiation codon or just 1 amino acid upstream (since the amino acid immediately N-terminal is now also a cysteine; Fig. 5A). The N Terminus of C′ Is Required for Processing to Generate Y1/Y2—We introduced the CrPV IRES just upstream of C′ in the Δ9 background. Remarkably this IRES directs protein synthesis on non-AUG codons in a manner that is methionyl-tRNA -independent, i.e. they initiate using the cognate charged tRNA (22Wilson J.E. Pestova T.V. Hellen C.U. Sarnow P. Cell. 2000; 102: 511-520Abstract Full Text Full Text PDF PubMed Scopus (344) Google Scholar, 35Sasaki J. Nakashima N. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 1512-1515Crossref PubMed Scopus (134) Google Scholar). Ribosomes recruited via this IRES would start translation on a GCU codon 6 amino acids before the start site of C′ (Fig. 6A). In cells this construct produced a slightly slower migrating form of Δ9C′ (Fig. 6B, lanes 3 and 4, Δ9C′*) consistent with IRES-driven expression. However, no Y proteins could be detected. The 6-amino acid addition did not change the stability of the C′ (Fig. 6C), suggesting that it had altered the processing pathway that generated the Y proteins. To rule out any effect of the IRES itself, we first modified it so that the critical pseudoknot that occupies the ribosomal A site is formed with AUG81 instead of GCU (22Wilson J.E. Pestova T.V. Hellen C.U. Sarnow P. Cell. 2000; 102: 511-520Abstract Full Text Full Text PDF PubMed Scopus (344) Google Scholar). This removes the N-terminal extension and ensures that the IRES-mediated product is identical to that expressed from the 5′ cap-driven Δ9. This modification restored the Δ9 expression phenotype (Fig. 6B, lanes 5 and 6). We next introduced the 6-amino acid extension immediately downstream of the C′ AUG such that C′* would be expressed in a 5′ cap-dependent manner. This destroyed the Δ9 phenotype (Fig. 6B, lanes 7 and 8). Mutation of Arg34 of the C′ Protein Severely Perturbs Processing—While performing these studies we changed the arginine codon immediately upstream of the Y1 AUG (Arg34) to either alanine or cysteine in the background Δ9 (Fig. 7A). These changes severely reduced Y expression in cells (Fig. 7B). Therefore, a single change in the primary sequence of the C′ protein just upstream from Y1 can ablate the Δ9 up-phenotype. The N-terminal Region of C′ May Serve as a Targeting Signal—The failure to process either the C′* or C protein suggests that the very N-terminal region of C′ itself plays a pivotal role in this event. The C′ and C protein differ by 11 amino acids (Fig. 6A), and assays