Copyright © 2001, American Society for Microbiology. All Rights Reserved.
Detailed Analysis of the Requirements of Hepatitis A Virus
Internal Ribosome Entry Segment for the Eukaryotic
Initiation Factor Complex eIF4F
ANDREW M. BORMAN,* YANNE M. MICHEL,ANDKATHERINE M. KEAN
Unite´ Postulante de Re´gulation de la Traduction Eucaryote et Virale, CNRS URA 1966, Institut Pasteur, 75724 Paris Cedex 15, France
Received 16 January 2001/Accepted 29 May 2001
The hepatitis A virus (HAV) internal ribosome entry segment (IRES) is unique among the picornavirus IRESs in that it is inactive in the presence of either the entero- and rhinovirus 2A or aphthovirus Lb proteinases. Since these proteinases both cleave eukaryotic initiation factor 4G (eIF4G) and HAV IRES activity could be rescued in vitro by addition of eIF4F to proteinase-treated extracts, it was concluded that the HAV IRES requires eIF4F containing intact eIF4G. Here, we show that the inability of the HAV IRES to function with cleaved eIF4G cannot be attributed to inefficient binding of the cleaved form of eIF4G by the HAV IRES. Indeed, the binding of both intact eIF4F and the C-terminal cleavage product of eIF4G to the HAV IRES was virtually indistinguishable from their binding to the encephalomyocarditis virus IRES, as assessed by UV cross-linking and filter retention assays. Rather, we show that HAV IRES activity requires, either directly or indirectly, components of the eIF4F complex which interact with the N-terminal fragment of eIF4G. Effectively, HAV IRES activity, but not that of the human rhinovirus IRES, was sensitive to the rotavirus nonstructural protein NSP3 [which displaces poly(A)-binding protein from the eIF4F complex], to recombinant eIF4E-binding protein (which prevents the association of the cap eIF4E-binding protein eIF4E with eIF4G), and to cap analogue.
Translation initiation on the uncapped genomic RNA of hepatitis A virus (HAV), a picornavirus, occurs after internal entry of ribosomes toward the 3⬘ end of a long, structured 5⬘
untranslated region (UTR), rather than by ribosome scanning from the 5⬘end (9, 13). Such internal initiation of translation requires the presence of an IRES (internal ribosome entry segment), which, for all of the animal picornaviruses studied, comprises some 450 nucleotides (nt) of complex secondary and presumably tertiary RNA structure (for reviews, see references 20 and 33). While IRESs have been formally identified in the genomes of all picornaviruses studied, they exhibit consider-able structural and functional divergence. Indeed, on the basis of both sequence and structural features and the requirements for optimal activity, three types of picornavirus IRES have been distinguished: type I entero- and rhinovirus (e.g., polio-virus and human rhinopolio-virus [HRV]) IRESs, the type II cardio-virus (encephalomyocarditis cardio-virus [EMCV]) and aphthocardio-virus (foot-and-mouth disease virus [FMDV]) IRESs, and the type III HAV IRES (2, 7, 40, 41).
The classification of the HAV IRES separately from the other picornavirus IRESs is merited at several levels. As men-tioned above, there exists very little sequence similarity be-tween the HAV element and the other picornavirus IRESs, and in many respects the HAV IRES is unique in the condi-tions required for optimal activity in vitro. Most interestingly, it is the only picornavirus IRES whose activity is severely
in-hibited in vitro and in cell culture in the presence of the entero-and rhinovirus 2A proteinases entero-and the aphthovirus Lb protein-ase (2, 5, 48). Treatment of translation extracts with these proteinases is without effect for the type II IRESs (37, 50) and actually stimulates type I IRES activity (6, 50). This stimulation is mediated via cleavage of eukaryotic initiation factor 4G (eIF4G) (6), a component of the eIF4F holoenzyme complex involved in the translation of capped cellular mRNAs. The eIF4F complex comprises the central scaffold molecule eIF4G, which binds the cap-binding protein eIF4E and the ATP-de-pendent RNA helicase eIF4A toward its N and C termini, respectively (for a review, see reference 35). The 2A and Lb proteinases cleave eIF4G in the same region, at sites separated by only seven amino acids, resulting in the liberation of two primary cleavage products (25, 28). The N-terminal cleavage product still associates with eIF4E and is dispensable for type I and II IRES activity (37), whereas the C-terminal domain binds eIF4A and is sufficient to drive type I and II IRES-mediated translation (6, 37). Indeed, direct functional interac-tions between the C-terminal region of eIF4G and the type II EMCV and FMDV IRESs have been reported (32, 39; for a review, see reference 23).
Since HAV IRES activity was abolished by treatment of translation extracts with either the 2A or Lb proteinase and could be fully restored upon readdition of eIF4F, it was con-cluded that the HAV IRES requires eIF4F containing intact eIF4G (5). However, it is now known that mammalian eIF4G as part of the eIF4F complex also interacts with poly(A)-binding protein (PABP) (19, 43), the protein implicated in binding the poly(A) tails at to the 3⬘ end of most eukaryotic mRNAs and in cooperative enhancement of translation of * Corresponding author. Mailing address: Unite´ Postulante de
Re´g-ulation de la Traduction Eucarvote et Virale, CNRS URA 1966, In-stitut Pasteur, 25 rue du Dr. Roux, 75724 Paris Cedex 15, France. Phone: (33) 1 40 61 33 55. Fax: (33) 1 40 61 30 45. E-mail: ambomb @pasteur.fr.
7864
on November 9, 2019 by guest
http://jvi.asm.org/
capped, polyadenylated mRNAs (12, 45). PABP was recently shown to be cleaved by the enterovirus 2A proteinases in the virus-infected cell and in vitro (22, 24). Thus, it is possible that PABP cleavage explains the inhibitory effects, at least of the 2A proteinase, on HAV IRES activity. We wished to examine this hypothesis and also to evaluate the possibility that HAV IRES inactivation upon eIF4G cleavage derives from a re-quirement for components of the intact eIF4F holoenzyme complex other than eIF4G.
In the present study, we demonstrate that inhibition of HAV IRES-driven translation correlates with cleavage of eIF4G and not PABP and that the Lb proteinase shows no detectable proteolytic activity for endogenous PABP in reticulocyte ly-sates. In addition, we present data which indicate that the presence of both eIF4E and PABP in the eIF4F complex is required for optimal HAV IRES activity, even though the HAV IRES-containing RNAs used were neither capped nor polyadenylated.
MATERIALS AND METHODS
Plasmids. The monocistronic construct p0p24, which contains the human immunodeficiency virus type 1 (HIV-1LAI) gene encoding the p24 protein under the control of the bacteriophage T710 promoter, has been described previously (34). The sequences corresponding to the complete HAV IRES (nt 44 to 738) were excised from the previously described full-length infectious cDNA clone of HAV (p16HM175) (21) by digestion withNcoI andAflIII, the cut ends were filled in, and the fragment was inserted into the filled-inBamHI site of p0p24 to generate pHAVp24. The sequences corresponding to the full HRV IRES were excised from pXLJHRV10-611 (4) by digestion withSalI andNcoI and inserted into p0p24 digested with the same enzymes, to produce pHRVp24.
Proteins and antibodies.Recombinant wild-type HRV type 2 (HRV2) 2A proteinase and FMDV O1k Lb proteinase, which had been purified to homoge-neity as described previously (25, 30), were stored in H100 buffer (10 mM HEPES-KOH [pH 7.5], 100 mM KCl, 1 mM MgCl2, 0.1 mM EDTA, 7 mM
-mercaptoethanol). Rabbit eIF4F and the N- and C-terminal cleavage products of eIF4F liberated following treatment with 2A proteinase (CpNand CpC, re-spectively) were purified as described previously (6, 26, 27). These proteins were dialyzed extensively against H100 buffer prior to use. A recombinant fragment of rotavirus NSP3 protein encompassing amino acids 163 to 313, which had been overexpressed and purified exactly as described previously (42, 43), was dialyzed against H100 buffer prior to use. Histidine-tagged eIF4E-binding protein 1 (4E-BP1 [PHAS-1]), which had been overexpressed and purified as described previ-ously (36), was diluted in H100 buffer prior to use. Rabbit anti-eIF4G peptide 7 antiserum (raised against residues 327 to 342), rabbit anti-eIF4G peptide 6 serum (raised against amino acids 1230 to 1248), and monoclonal antibody 10E10 raised against human PABP have been described previously (14, 49).
In vitro transcription and translation.The synthesis of radiolabelled probes was performed as described previously (3). For probes corresponding to the HAV IRES, plasmid p16HM175 was digested to completion withAflIII to allow in vitro transcripts spanning nt 1 to 738 of the HAV genome to be generated. The EMCV IRES probe, which starts at the poly(C) tract and ends at nt 848, was synthesized from p-CITE (Novagen) which had been linearized by digestion with NcoI.
For RNAs destined for in vitro translation, in vitro transcription reactions and purification and quantification of synthesized RNAs were performed exactly as described by Michel et al. (34). Purified mRNAs were subsequently translated in nuclease-treated rabbit reticulocyte lysates (Flexi reticulocyte lysate; Promega) supplemented with 5% (by volume) HeLa cell S10 extracts as described else-where (5). Reactions (12l by volume) contained 50% (by volume) reticulocyte lysate, 90 and 0.75 mM added KCl and MgCl2, respectively, and 10g of in vitro-transcribed RNAs per ml. Reaction mixtures were preincubated with the indicated concentrations of the 2A and Lb proteinases for 10 min at 30°C prior to proteinase inactivation, using elastatinal for the 2A proteinase (final concen-tration, 500M) and E-64 (Sigma) for the Lb proteinase (final concentration, 400M) and addition of RNA. Reaction mixtures supplemented with rotavirus NSP3 protein, recombinant 4E-BP1, or cap analogue were incubated on ice for 10 min before addition of RNA. Translations were performed for 90 min at 30°C, before analysis of [35S]methionine-labeled translation products by sodium
dode-cyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) using gels contain-ing 20% polyacrylamide, as described previously (11). Dried gels were exposed to Bio-max film (Amersham) typically for 24 to 48 h. Quantification of translation efficiencies was performed densitometrically, exactly as described previously (5), using multiple exposures of each gel to ensure that the linear response range of the film was respected.
UV laser cross-linking and filter retention assays.Radiolabeled RNAs (ap-proximately 0.1 pmol; 100,000 cpm) were incubated with 0.1 to 0.2g of purified proteins in cross-linking buffer (10 mM HEPES-KOH [pH 7.2], 3 mM MgCl2, 5% [by volume] glycerol, 1 mM dithiothreitol, 80 mM KCl, 0.1 mg ofEscherichia coli23S rRNA per ml; 10-l final reaction volumes) for 10 min at room tem-perature before being irradiated at 266 nm using an NdYAG laser with a single high-intensity pulse (5-ns duration, 0.5⫻1011-W/m2intensity, approximate dose of 100 J/m2). Following irradiation, samples were digested with RNases A and T
1 (0.25 mg/ml and 8 units/l, respectively) for 30 min at 37°C and prepared for SDS-PAGE as described previously (3).
Filter retention assays were performed essentially as described by Clever et al. (10). Briefly, binding reaction mixtures (30-l final volume) containing binding buffer (20 mM HEPES-KOH [pH 7.3], 80 mM KCl, 0.5 mM MgCl2, 2 mM dithiothreitol 5% [by volume] glycerol, 0.1 mg of bovine serum albumin per ml, 10g ofE. colitRNA per ml) and the indicated concentrations of purified eIF4F, CPN, or CpCwere assembled on ice prior to the addition of 20 ng of radiolabeled probes (approximately 0.1 pmol). After 10 min at 30°C, reaction mixtures were transferred to prewetted nitrocellulose disks (0.45-m pore size; HAWP type; Millipore) and washed with 5 ml of binding buffer. The radioactivity retained on filters was quantified by scintillation counting. Data were corrected for the amount of probe retained in the absence of added target protein, which was typically less than 3% of total radioactivity. Experiments were performed in duplicate, and the average counts are given, expressed as the percentage of bound probe relative to the maximal binding achieved with a given RNA target in each experiment. The apparent dissociation constant (Kd) of each interaction
was taken as the concentration of protein required to attain 50% of the maximal binding.
Western blotting analysis.Western blotting analysis of eIF4G or PABP was performed exactly as described previously (6), using rabbit anti-eIF4G peptide 6 and 7 antisera (for detection of CpCand CpN, respectively) or monoclonal antibody 10E10 (for PABP) as primary antibodies. Membranes were then incu-bated with horseradish peroxidase-linked goat anti-mouse or anti-rabbit second-ary antibodies and were revealed by enhanced chemiluminescence (ECLplus; Amersham) or a commercial diaminobenzidine-peroxidase substrate kit (Vector Laboratories Inc.).
RESULTS
We found previously that the HAV IRES is inactive in translation extracts in which eIF4G is cleaved by the picorna-virus 2A or Lb proteinase. Rescue of HAV IRES-driven trans-lation was possible by back-addition of purified rabbit eIF4F containing intact eIF4G. Since eIF4G was the only component of the eIF4F complex known at that time to be cleaved by both the 2A and Lb proteinases, it was concluded that HAV IRES function requires intact eIF4G (5). However, it was recently demonstrated that the enterovirus 2A proteinases cleave an-other protein known to associate with mammalian eIF4F, namely, PABP (22, 24). Thus, the first aim of the present study was to determine whether proteinase-mediated inhibition of HAV IRES activity results from cleavage of eIF4G, PABP, or both. For the experiments described here, we used monocis-tronic mRNAs from which the synthesis of the HIV-1 p24 protein is driven by either the HAV or the HRV2 IRES (Fig. 1A). Translation reactions were performed in rabbit reticulo-cyte lysate under conditions essentially the same as those used in a previous study demonstrating that the HAV IRES was inactive upon eIF4G cleavage (5) (see Materials and Meth-ods). Thus, reactions contained near-optimal concentrations of added salt and trace quantities of HeLa cell S10 extract to enable HRV IRES-driven translation and were programmed
on November 9, 2019 by guest
http://jvi.asm.org/
with relatively low concentrations (10g/ml) of in vitro-tran-scribed, uncapped, monocistronic mRNAs.
Inhibition of HAV IRES-driven translation correlates with
cleavage of eIF4G and not PABP.As shown previously (5), the
treatment of translation extracts with relatively low concentra-tions (20g/ml) of active HRV2 2A or FMDV O1k Lb pro-teinase abrogates the translation of HAV IRES-driven trans-lation (Fig. 1B, Active 2A and Active Lb lanes). Such inhibition is dependent on proteolytic activity, since treatment of extracts with proteinases which had been first inactivated by incubation with their specific inhibitors did not affect HAV IRES activity (Fig. 1B, Inactive 2A and Inactive Lb lanes).
To investigate the possible mechanistic basis of inhibition, we examined the integrity of both eIF4G and PABP, the re-cently identified substrate of the enterovirus 2A proteinases, by Western blot analysis of translation reactions at various times throughout a 90-min incubation at 30°C with active 2A or Lb proteinase. While eIF4G was rapidly cleaved by the HRV2 2A and particularly the FMDV O1k Lb proteinase, there was no evidence of cleavage of PABP under the reaction conditions used (Fig. 1C). However, when translation extracts were incu-bated for long periods of time (over 3 h) with high concentra-tions of active 2A proteinase (130g/ml) and at 37°C rather than 30°C, a 40% reduction in the amount of intact PABP was observed (Fig. 1D). Interestingly, no change in the quantity of intact PABP was found in reactions treated with the Lb pro-teinase under equivalent conditions (Fig. 1D). The absence of detectable PABP cleavage in our standard translation assay conditions and the apparent inability of the Lb proteinase to cleave PABP even after extended incubation demonstrate that inactivation of the HAV IRES by proteinase treatment corre-lates with cleavage of eIF4G and not with that of PABP.
The HAV IRES can bind both intact eIF4G and the
proteo-lytic C-terminal fragment with similar affinities.The
C-termi-nal cleavage product of eIF4G with its associated eIF4A and eIF3 (Cpc-eIF4A-eIF3) is sufficient to drive type I and type II IRES activity. Furthermore, a recombinant fragment of eIF4G which overlaps CpCcan bind directly to the EMCV IRES and can functionally replace intact eIF4G in 48S initiation complex formation assays on this IRES (38, 39). Thus, we examined
FIG. 1. (A) Schematic representation of plasmids used to generate monocistronic IRES-carrying mRNAs. The HIV-1 p24 coding region is shown as an open box. The 5⬘ UTRs derived from picornavirus sequences are depicted as thick lines; the numbers below the HRV2 and HAV IRESs are based on the viral genome sequences and denote the first and last nucleotides of picornavirus sequence present. The junction between the 5⬘UTR and the p24 coding region is detailed, and the initiation codon for p24 synthesis is underlined. (B) In vitro translation of monocistronic uncapped HAV p24 mRNA in the pres-ence of the 2A or Lb proteinase. Standard rabbit reticulocyte lysate translation reactions (see Materials and Methods) were preincubated for 10 min at 30°C with H100 buffer (no RNA and buffer lanes), recombinant 2A and Lb proteinases (20g/ml; Active 2A and Active Lb lanes), or recombinant 2A and Lb proteinases which had been previously inactivated by incubation for 10 min at 4°C with elastatinal and E-64, respectively (20g/ml; Inactive 2A and Inactive Lb lanes). Reactions were then programmed with uncapped mRNA transcribed in vitro from pHAVp24 (10g/ml) (see Materials and Methods). A control reaction was programmed with water (no RNA lane).
Trans-lations were processed as described in Materials and Methods. The autoradiograph of the dried 20% polyacrylamide gel is shown. The position of the HIV-1 p24 translation product is marked. Translation efficiency derived from densitometric quantification is plotted below each lane. (C) Integrity of eIF4G and PABP in translation reactions treated with the 2A and Lb proteinases. Standard translation reactions were incubated at 30°C with the indicated concentrations of purified recombinant 2A and Lb proteinases for the times indicated above each lane prior to analysis by Western blotting as described in Materials and Methods, using antibodies raised against CpN(top) and PABP (bot-tom). The positions of intact eIF4G, CpN, and PABP are indicated. (D) PABP is cleaved upon extended incubation with high concentra-tions of 2A but not Lb proteinase. Standard translation reacconcentra-tions were incubated at 37°C with the indicated concentrations of purified recom-binant 2A and Lb proteinases for the times indicated above each lane prior to analysis by Western blotting as described in Materials and Methods, using antibodies raised against PABP. The position of intact PABP is indicated. The amount of PABP detected in each reaction was quantified by densitometry and is indicated below each lane as a percentage of the amount detected in the absence of proteinase treat-ment (0 min lane).
on November 9, 2019 by guest
http://jvi.asm.org/
[image:3.612.88.256.71.546.2]whether the failure of the HAV IRES to function upon cleav-age of eIF4G was due to its inability to bind the CpC proteo-lytic fragment. Toward this end, we digested purified rabbit eIF4F with HRV2 2A proteinase and repurified the digestion products (see Materials and Methods). Figure 2A shows a Western blot analysis of the resulting protein preparations, using antibodies which recognize CpNand CpC, respectively. While cleaved eIF4F contains large quantities of both CpNand CpC, the separated cleavage products show only low levels of cross-contamination. Thus, only trace concentrations of CpC are present in the CpNpreparation, and vice versa.
These protein preparations were first used in classical UV Laser cross-linking assays with radiolabeled probes corre-sponding to the full-length EMCV or HAV IRES (see Mate-rials and Methods). Figure 2B shows the results of a represen-tative cross-linking experiment. Both the HAV and EMCV IRESs could be cross-linked to intact eIF4G, as evidenced by the detection of a specific32P-labeled protein of approximately 220 kDa in reactions containing pure eIF4F. Treatment of eIF4F with 2A proteinase prior to cross-linking resulted in the
loss of this labeled 220-kDa protein and the presence of a new labeled protein of approximately 97 kDa. The size of this product corresponds well with the estimated size of CpC (97 kDa). Cross-linking experiments using the purified cleavage products of eIF4F confirmed that this 97-kDa32P-labeled pro-tein was indeed CpC. No specific protein was radiolabeled in reactions containing CpN. However, in reactions containing CpC, a predominant protein identical in size to that observed with cleaved eIF4F (97 kDa) was evidenced (Fig. 2B, compare eIF4F/2A and CpC lanes), independently of the IRES probe used. Thus, it seems that CpCcan be cross-linked specifically to both the EMCV and HAV IRESs, even though the latter requires intact eIF4G for activity. Interestingly, a smaller, less intensely labeled protein was also detected in cross-linking reactions containing cleaved eIF4F or purified CpCand either IRES. A protein of similar apparent size was also detectable in reactions containing intact eIF4F after prolonged exposures of the gels (data not shown). Given its approximate size (43 kDa), it seems likely that this species corresponds to eIF4A. It should be noted that the intensity of labeling of the different cross-FIG. 2. (A) Purified rabbit eIF4F was treated with 2A proteinase, and the CpNand CpCwere separated as described in Materials and Methods; 180, 180, 170, or 80 ng of eIF4F, 2A proteinase-treated eIF4F, or purified CpNor CpC, respectively, was submitted to Western blotting using antibodies raised against CpN(top) and CpC(bottom). The positions of intact eIF4G, CpN, and CpCare indicated. (B) UV Laser cross-linking of eIF4F to the HAV and EMCV 5⬘UTRs. UV cross-linking reactions containing 180 ng of intact eIF4F or 2A proteinase-treated eIF4F, 170 ng of purified CpN, or 80 ng of purified CpCwere assembled exactly as described in Materials and Methods and supplemented with radiolabeled probes corresponding to the full EMCV (left) or HAV IRES (right). After cross-linking and RNase digestion, reactions were analyzed by SDS-PAGE. The autoradiograph of the dried 15% gel is shown; positions of the approximately 220-, 97-, and 43-kDa proteins labeled after cross-linking are indicated. The right-hand panel was exposed approximately three times longer than the left-hand panel. (C) The EMCV and HAV IRESs bind eIF4G and CpCwith similar affinities. Filter binding assays were performed exactly as described in Materials and Methods, using radiolabeled probes corresponding to the HAV (circles) or EMCV (squares) IRES and the given concentrations of intact eIF4F (top) purified CpC(filled symbols, bottom), or purified CpN(open symbols, bottom). The amount of retained RNA (expressed as a percentage of the maximal retained RNA) is plotted against protein concentration.
on November 9, 2019 by guest
http://jvi.asm.org/
[image:4.612.71.533.70.370.2]linked proteins was reproducibly greater using the EMCV rather than the HAV radiolabeled RNAs. However, the HAV probe (738 nt) is significantly longer than the EMCV counter-part (450 nt) and is considerably richer in U residues (232, as opposed to 83). Thus, are used approximately three fold more EMCV than HAV probe in order to include equivalent quan-tities of radiolabel in cross-linking reactions.
Since the molar ratio of protein to RNA in UV Laser cross-linking experiments was of the order of 6 to 1, it remains possible that the affinity of the HAV IRES for CpC is much lower than that of EMCV. To assess the relative affinities of the two IRESs for intact eIF4G (as part of eIF4F) and for CpC, the different protein preparations were used in filter retention assays (see Materials and Methods). Intact eIF4F and purified CpC bound the HAV and EMCV IRESs to approximately equal degrees (Fig. 2C). The apparentKds of these interactions
(as judged by the concentration of protein required for half-maximal retention of RNA) were 0.2 nM (for the HAV and EMCV IRES interactions with eIF4F) and 0.4 nM and 0.7 nM for HAV-CpC and EMCV-CpC, respectively. In contrast, vir-tually no radiolabeled HAV or EMCV IRES RNAs were re-tained in reactions containing purified CpN. Thus, the inability of the HAV IRES to function in the presence of CpCcannot be explained by a reduced capacity to bind this protein fragment.
The HAV IRES requires eIF4E and PABP bound to eIF4G
for activity.The inability of CpC-eIF4A-eIF3 to drive HAV
IRES activity, even though CpCis apparently bound with high affinity by this element, led us to investigate whether the com-ponents of the full eIF4F complex which remain associated with CpNare required by the HAV IRES. Toward this end, we used two different recombinant proteins with the proven ca-pacity to displace either eIF4E or PABP from the full eIF4F holoenzyme complex. The rotavirus nonstructural protein NSP3 has recently been shown to interact with the N-terminal part of eIF4G and to evict PABP from eIF4F (42, 43). The recombinant fragment of NSP3 (amino acids 163 to 313) used for the present study is also capable of displacing PABP from eIF4G (8, 34) but retains no detectable affinity for RNA, as measured in gel retardation and UV cross-linking assays (42). 4E-BP1 (PHAS-1) has previously been shown to inhibit the interaction between eIF4E and eIF4G (15, 31) by competing for binding to a dorsal site on eIF4E (44).
Thus, translation reactions were treated with various con-centrations of either 4E-BP1 or NSP3 fragment before being programmed with in vitro-transcribed mRNAs carrying the HAV IRES (Fig. 3A and B). To control against nonspecific inhibitory effects of the two protein preparations, independent reactions were programmed with an mRNA under the control of the type I HRV2 IRES (HRV1 p24), which does not require CpN-eIF4E-PABP for activity (6). Low concentrations of ei-ther NSP3 or 4E-BP1 significantly inhibited translation initi-ated from the HAV IRES. The effects of 4E-BP1 were partic-ularly dramatic, with 50% translation inhibition induced when recombinant protein concentrations exceeded 100 nM. Con-versely, HRV IRES activity was slightly but reproducibly stim-ulated by the addition of either recombinant protein. We also verified that the concentrations of 4E-BP1 and NSP3 used were indeed displacing considerable proportions of their re-spective target proteins (Fig. 3D and E), either by testing the resistance of eIF4G to cleavage by the Lb proteinase when
eIF4E is displaced from the eIF4F complex by 4E-BP1 (36) (Fig. 3D) or by measuring the amount of PABP which can be coimmunoprecipitated in the presence of NSP3, using antibod-ies directed against eIF4G as described previously (8, 34, 43) (Fig. 3E). The degrees of displacement of eIF4E by 4E-BP1 and of PABP by NSP3 were significant and correlated well with the levels of inhibition of HAV IRES activity described above. In effect, the concentrations of NSP3 required to displace 80% of PABP from eIF4G inhibited HAV IRES activity by between 40 and 70% (compare Fig. 3A and E), and the concentrations of 4EBP1 which caused maximal displacement of eIF4E from eIF4G also inhibited HAV IRES-driven translation by greater than 50% (compare Fig. 3B and D). Thus, the removal of either eIF4E or PABP from the eIF4F complex inhibits HAV, but not HRV, IRES-driven translation.
Given that HAV IRES activity requires eIF4E as part of the eIF4F complex, we evaluated the effects of added cap analogue on cap-independent translation driven by either the HAV or HRV IRES (Fig. 3C). Similarly to 4E-BP1 and NSP3, cap analogue slightly stimulated HRV IRES activity but signifi-cantly inhibited translation driven by the HAV element. How-ever, the concentration of cap analogue required to inhibit HAV IRES-driven translation by 50% (66 M) was 1 to 2 orders of magnitude higher than the concentrations of either 4E-BP1 and NSP3 which induced the equivalent inhibition (100 or 990 nM, respectively, compared to an estimated con-centration of endogenous eIF4G of 10 to 20 nM [5]). A recent report has shown that even higher concentrations of analogue can almost abolish translation driven by the HAV, but not the EMCV or poliovirus, IRES (1a).
DISCUSSION
The poliovirus and coxsackievirus 2A proteinases were re-cently shown to cleave PABP, in addition to eIF4G, in vitro and in the virus-infected cell (22, 24). In the present study, we have shown that inactivation of the type III HAV IRES by the HRV 2A and FMDV Lb proteinases correlates with cleavage of the eIF4G component of eIF4F rather than with proteolysis of PABP. Effectively, no Lb proteinase-mediated cleavage of PABP could be evidenced, even upon prolonged incubation at very high concentrations of proteinase, yet this proteinase still inactivated HAV IRES activity when included in translation assays. Furthermore, there was no detectable cleavage of PABP by the rhinovirus 2A proteinase under the conditions of our translation assays, although proteolysis was observed upon extended incubation. Thus, the ability to degrade PABP ap-pears to be a conserved feature of all entero- and rhinovirus 2A proteinases. Interestingly, the kinetics of PABP cleavage seem to be similar for the entero- and rhinovirus 2A protein-ases, in the sense that the cleavage of PABP was in all cases significantly slower than that of eIF4G (compare results in references 22 and 24 and this study). Hence, it appears that if PABP degradation contributes to the shutoff of host cell trans-lation upon infection with entero- and rhinoviruses, it is likely to be late in the infectious cycle.
The type I and II IRESs function efficiently with eIF4F which has been cleaved by the picornavirus 2A or Lb protein-ase. In effect, CpCof eIF4G with its associated eIF4A and eIF3 is sufficient to support internal initiation of translation on these
on November 9, 2019 by guest
http://jvi.asm.org/
elements. Indeed, a direct interaction between the FMDV and EMCV IRESs and CpCor a central region of eIF4G overlap-ping CpChas been demonstrated previously (32, 39), and the central fragment of eIF4G together with eIF4A and eIF3 can functionally substitute for intact eIF4F in 48S initiation com-plex formation on the EMCV IRES (38). We thus investigated whether the inability of the HAV IRES to function with cleaved eIF4F resulted from failure of this IRES to bind CpC. The HAV and EMCV IRESs could both be cross-linked by UV irradiation to intact eIF4G (as part of purified eIF4F) or to purified CpC, and filter retention assays confirmed that the HAV and EMCV IRESs bind intact eIF4F and CpC with
similar, high affinities (apparent dissociation constants all less than 1 nM). It is possible that the actualKds for these
inter-actions are even lower than those measured here. In effect, our estimations are based on the concentration of purified protein required to retain the different RNAs and do not take into account the proportion of protein in the preparation capable of binding RNA, which is likely to be less than 100%. It should also be noted that Lomakin et al. recently measured the affinity of the interaction between the central domain of eIF4G and the region of the EMCV IRES to which it binds (31a). Even though those authors employed gel retardation using a trun-cated EMCV IRES and a recombinant internal fragment of FIG. 3. Effects of cap analogue, 4E-BP1, and rotavirus NSP3 on HAV and HRV IRES-driven translation. Standard translation reactions programmed with the indicated uncapped RNAs (10g/ml) were supplemented with the concentrations indicated above each lane of recombinant rotavirus NSP3 protein (A; concentrations in micrograms per milliliter), recombinant 4E-BP1 protein (B; reactions (a) through i were made 0, 25, 50, 100, 200, 400, 800, 1,600, and 3,200 nM, respectively, in recombinant 4E-BP1), and cap analogue (C; reactions a through g were made 0, 0.22, 1.1, 5.5, 11, 22 and 66M, respectively, in cap analogue). An asterisk denotes the position of a strongly labeled smaller p24-derived product in reactions programmed with HRV p24 RNA, whose synthesis results from leaky scanning and initiation of ribosomes at an internal AUG in the p24 coding region. The protein larger than the p24 polypeptide synthesized from HRV p24 mRNAs results from translation initiation at an upstream AUG at position 449 in the HRV2 5⬘UTR as previously described (4). Relative translation efficiency (as a percentage of that measured in the absence of each inhibitor) derived from densitometric quantification is plotted below each panel (filled circles, HAV IRES activity; open squares, HRV IRES activity). The mean values from two independent experiments (⫾standard deviation) are plotted in each case. (D) Displacement of eIF4E from eIF4G as measured by Lb proteinase-mediated cleavage of eIF4G. Translation reactions were incubated for 10 min on ice with H100 buffer or 4E-BP1 in H100 buffer (final concentrations, from left to right, of 30, 60, 120, and 250 nM) before being treated with H100 buffer or Lb proteinase in H100 buffer (Lb⫹lanes; final proteinase concentration of 15g/ml) for 15 min at 30°C. The integrity of eIF4G was then assessed by Western blotting as described in Materials and Methods. The positions of intact eIF4G and the primary and secondary N-terminal cleavage products of eIF4G (1° CpN and 2° CpN, respectively) are indicated. The quantity of intact eIF4G or 2° CpN in each reaction was evaluated by densitometry and is plotted below each lane. Since the anti-eIF4G antibody used recognizes the primary cleavage product of eIF4G much more efficiently than it recognizes either the intact moleule or the secondary cleavage product, the signal for 1° CpN was saturating in all reactions which had been treated with Lb proteinase. (E) Effect of NSP3 on coimmunoprecipitation of PABP and eIF4G. The degree of displacement of PABP from eIF4G was assessed by immunoprecipitation of complexes using antibodies raised against eIF4G followed by Western blot analysis of immunoprecipitates using antibodies directed against PABP as described previously (8, 34). The data presented are modified from reference 34.
on November 9, 2019 by guest
http://jvi.asm.org/
eIF4G to assess the binding affinity, the values obtained (af-finities of 5 to 6 nM) are similar to those reported here with either the intact EMCV or HAV IRES. Importantly, such data suggest that the IRES-eIF4G interaction is of sufficient affinity to drive internal initiation of translation and would also allow effective competition of these IRESs with capped, cellular mRNAs for eIF4G binding (31a). These mechanistic consid-erations aside, the data presented here also argue against the suggestion that the HAV IRES fails to function with CpC because it can not efficiently bind this cleavage product.
It remains possible that CpCbinding to the HAV IRES does not allow IRES activity, for example, because eIF4F and CpC bind to different sites on the HAV IRES and the CpCbinding site is incompetent for 48S complex formation. Indirect evi-dence in favor of such a hypothesis derives from the similar binding affinities of intact eIF4G and CpCfor the HAV IRES despite the failure of the apparently nonfunctional CpC to behave as a dominant negative inhibitor of HAV IRES activity (5). By toeprinting assays, we attempted to identify the exact binding sites of eIF4F and CpCon the HAV IRES. Unfortu-nately, we were unable to convincingly locate any such site, possibly because of the relatively low concentrations of purified proteins in our preparations (data not shown). However, in light of the effects of recombinant 4E-BP1 and rotavirus NSP3 proteins on HAV IRES activity, it seems unlikely that differ-ential binding sites for eIF4F and CpCalone could explain the failure of CpCto sustain HAV IRES activity. Either recombi-nant protein was capable of inhibiting HAV IRES activity to the same degree as the 2A or Lb proteinase, indicating that both PABP and eIF4E must remain associated with eIF4G for translation to be efficiently driven by the HAV IRES. At first sight, these results seem at odds with what is currently known concerning the mechanism(s) of IRES-mediated translation initiation, since they indicate that an mRNA carrying the HAV IRES resembles a capped, polyadenylated mRNA in terms of its requirements for the eIF4F complex. However, since HAV IRES-driven translation is naturally cap independent and the in vitro-transcribed RNAs used in the experiments described here were neither capped nor polyadenylated, it can be postu-lated that eIF4E and PABP are not needed to fulfill their classical roles of binding the 5⬘cap and 3⬘poly(A) tail. It has recently been shown that plant cap-binding protein can bind specific internal sequences of viral RNAs (46; K. Browning, personal communication). We cannot rule out that some such cap mimicry is operational in the picornavirus IRESs and serves to tether eIF4F to an internal site within the HAV IRES. However, it should be noted that we never detected any significant cross-linking of eIF4E to the HAV or EMCV IRES (Fig. 2), arguing against such a hypothesis. Instead, several lines of evidence indicate that in fact the conformation of eIF4G is altered upon binding of either PABP or eIF4E. For instance, the HRV 2A protease is virtually incapable of cleav-ing eIF4G in the absence of eIF4E (17). Similarly, an addi-tional effect of 4E-BP1-mediated displacement of eIF4E from eIF4G is a conformational change in eIF4G which renders it uncleavable by the Lb proteinase (36) (Fig. 3D). Indeed, yeast eIF4E was recently shown to induce folding of a peptide com-prising the region of eIF4G with which it interacts (18). In addition, binding of eIF4E to eIF4G increases the affinity of the eIF4E-cap interaction (16), and binding of wheat germ
PABP to eIF4G increases the affinities of both the PABP-poly(A) and eIF4E-cap interactions (29, 47). Thus, the confor-mation of eIF4G seems to depend on the proteins which it binds, and eIF4G is also apparently capable of relaying con-formational changes to its different binding partners. In light of those observations, a reasonable interpretation of the data presented here would be that eIF4E and PABP are required by the HAV IRES to induce the correct conformation of eIF4G not for binding per se but for essential signaling events and that eviction of either protein pushes eIF4G into a nonfunctional conformation. Indeed, a similar explanation has been ad-vanced to explain the surprising capacity of 4E-BP1 to inhibit the translation of uncapped cellular mRNAs (36), translation which has been shown to be stimulated by eIF4G cleavage (6, 37). From the data presented here, one then has to postulate that binding of cap analogue to eIF4E in the context of eIF4F can apparently also direct conformational changes in eIF4G. However, as Ali et al. point out in the accompanying paper (1), it is difficult to envisage the mechanism of such a conforma-tional change given that the cap binding site on eIF4E is not close to the region which interacts with eIF4G. While further studies will be required to fully explore these hypotheses, the HAV IRES is the only such element identified to date which apparently requires eIF4E as part of the eIF4F complex for internal initiation of translation. Thus, the type III HAV IRES differs significantly from the type I entero- and rhinovirus and type II cardio- and aphthovirus IRESes with respect to its requirement for components of the eIF4F holoenzyme com-plex.
ACKNOWLEDGMENTS
We are grateful to Ce´cile Malnou and Sylvie van der Werf for their interest in this work and to Pascal Roux for help with laser cross-linking assays. We thank Tim Skern and Ernst Kuechler for the gift of purified recombinant 2A and Lb proteinases, Bob Rhoads for purified rabbit eIF4F and antibodies raised against eIF4G, Simon Morley for purified eIF4E-BP1, Didier Poncet and Nathalie Castagne´ for purified rotavirus NSP3 protein, and M. Go¨rlach for antibodies against PABP. We also thank Matthias Hentze, Encarna Martinez-Salas, and Karen Browning for communicating unpublished results.
Work in the laboratory of K.M.K. is supported by the Programme de Recherche Clinique de l’Institut Pasteur, by a Contrat d’Incitation a` la Recherche en vue d’Applications (CCV no. 8) from the Pasteur Insti-tute, by grant 6495 from the Association Franc¸aise contre les Myopa-thies (AFM), and by funding from the Agence Nationale de Recher-ches sur le SIDA (ANRS). Y.M.M. is supported by a doctoral fellowship from the Association pour la Recherche sur le Cancer (ARC).
REFERENCES
1.Ali, I. K., L. McKendrick, S. J. Morley, and R. J. Jackson.2001. Activity of the hepatitis A virus IRES requires association between the cap-binding translation initiation factor (eIF4E) and eIF4G. J. Virol.75:7854–7863. 1a.Bergamini, G., T. Preiss, and M. W. Hentze.2000. Picornavirus IRESes and
the poly(A) tail jointly promote cap-independent translation in a mammalian cell-free system. RNA6:1773–1780.
2.Borman, A. M., J.-L. Bailly, M. Girard, and K. M. Kean.1995. Picornavirus internal ribosome entry segments: comparison of translation efficiency and the requirements for optimal internal initiationin vitro. Nucleic Acids Res. 23:3656–3663.
3.Borman, A., M. T. Howell, J. Patton, and R. J. Jackson.1993. The involve-ment of a spliceosome component in internal initiation of human rhinovirus RNA translation. J. Gen. Virol.74:1775–1788.
4.Borman, A., and R. J. Jackson.1992. Initiation of translation of human rhinovirus RNA: mapping the internal ribosome entry site. Virology188: 685–696.
5.Borman, A. M., and K. M. Kean.1997. Intact eukaryotic initiation factor 4G
on November 9, 2019 by guest
http://jvi.asm.org/
is required for Hepatitis A virus internal initiation of translation. Virology 237:129–136.
6.Borman, A. M., R. Kirchweger, E. Zeigler, R. E. Rhoads, T. Skern, and K. M. Kean.1997. eIF4G and its proteolytic cleavage products: effect on initiation of protein synthesis from capped, uncapped and IRES-containing mRNAs. RNA3:186–196.
7.Borman, A. M., P. Le Mercier, M. Girard, and K. M. Kean.1997. Compar-ison of picornaviral IRES-driven internal initiation of translation in cultured cells of different origins. Nucleic Acids Res.25:925–932.
8.Borman, A. M., Y. M. Michel, and K. M. Kean.2000. Biochemical charac-terisation of cap-poly(A) synergy in rabbit reticulocyte lysates: the eIF4G-PABP interaction increases the functional affinity of eIF4E for the capped mRNA 5⬘end. Nucleic Acids Res.28:4068–4075.
9.Brown, E. A., S. P. Day, R. W. Jansen, and S. M. Lemon.1991. The 5⬘
nontranslated region of hepatitis A virus: secondary structure and elements required for in vitro translation. J. Virol.65:5828–5838.
10. Clever, J., C. Sassetti, and T. G. Parslow.1995. RNA secondary structure and binding sites forgaggene products in the 5⬘packaging signal of human immunodeficiency virus type 1. J. Virol.69:2101–2109.
11. Dasso, M. C., and R. J. Jackson.1989. Efficient initiation of mammalian mRNA translation at a CUG codon. Nucleic Acids Res.17:6485–6497. 12. Gallie, D. R.1998. A tale of two termini: a functional interaction between the
termini of an mRNA is a prerequisite for efficient translation initiation. Gene216:1–11.
13. Glass, M. J., and D. F. Summers.1992. A cis-acting element within the hepatitis A virus 5⬘-non-coding region required forin vitrotranslation. Virus Res.26:15–31.
14. Go¨rlach, M., C. G. Burd, and G. Dreyfuss.1994. The mRNA poly(A)-binding protein: localization, abundance and RNA-poly(A)-binding specificity. Exp. Cell Res.211:400–407.
15. Haghighat, A., S. Mader, A. Pause, and N. Sonenberg.1995. Repression of cap-dependent translation by 4E-binding protein 1: competition with p220 for binding to eukaryotic initiation factor-4E. EMBO J.14:5701–5709. 16. Haghighat, A., and N. Sonenberg.1997. eIF4G dramatically enhances the
binding of eIF4E to the mRNA 5⬘-cap structure. J. Biol. Chem.272:21677– 21680.
17. Haghighat, A., Y. Svitkin, I. Novoa, E. Kuechler, T. Skern, and N. Sonen-berg.1996. The eIF4G-eIF4E complex is the target for direct cleavage by the rhinovirus 2A proteinase. J. Virol.70:8444–8450.
18. Hershey, P. E., S. M. McWhirter, J. D. Gross, G. Wagner, T. Alber, and A. B. Sachs.1999. The cap-binding protein eIF4E promotes folding of a functional domain of yeast translation initiation factor eIF4G1. J. Biol. Chem.274: 21297–21304.
19. Imataka, H., A. Gradi, and N. Sonenberg. 1998. A newly identified N-terminal amino acid sequence of human eIF4G binds poly(A)-binding pro-tein and functions in poly(A)-dependent translation. EMBO J.17:7480– 7489.
20. Jackson, R. J., and A. Kaminski.1995. Internal initiation of translation in eukaryotes: the picornavirus paradigm and beyond. RNA1:985–1000. 21. Jansen, R. W., J. E. Newbold, and S. M. Lemon.1988. Complete nucleotide
sequence of a cell culture-adapted variant of hepatitis A virus: comparison with wild-type virus with restricted capacity forin vitroreplication. Virology 163:299–307.
22. Joachims, M., P. C. Van Breugel, and R. E. Lloyd. 1999. Cleavage of poly(A)-binding protein by enterovirus proteases concurrent with inhibition of translation in vitro. J. Virol.73:718–727.
23. Kean, K. M., Y. M. Michel, and A. M. Borman.1999. Viral exceptions to 5⬘-end dependent initiation of translation: is there really a difference in the mechanism of ribosome recruitment to capped and IRES containing mRNAs. Curr. Top. Virol.1:191–201.
24. Kerekatte, V., B. D. Keiper, C. Badorff, A. Cai, K. U. Knowlton, and R. E. Rhoads.1999. Cleavage of poly(A)-binding protein by coxsackievirus 2A protease in vitro and in vivo: another mechanism for host protein synthesis shutoff? J. Virol.73:709–717.
25. Kirchweger, R., E. Ziegler, B. J. Lamphear, D. Waters, H. D. Liebig, W. Sommergruber, F. Sobrino, C. Hohenadl, D. Blaas, R. E. Rhoads, and T. Skern.1994. Foot-and-mouth disease virus leader proteinase: purification of the Lb form and determination of its cleavage site on eIF4␥. J. Virol. 68:5677–5684.
26. Lamphear, B. J., R. Kirchweger, T. Skern, and R. E. Rhoads.1995. Mapping of functional domains in eukaryotic protein synthesis initiation factor 4G (eIF4G) with picornaviral proteases. J. Biol. Chem.270:21975–21983. 27. Lamphear, B. J., and R. Panniers.1990. Cap binding protein complex that
restores protein synthesis in heat-shocked Ehrlich cell lysates contains highly phosphorylated eIF4E. J. Biol. Chem.265:5333–5338.
28. Lamphear, B. J., R. Yan, F. Yang, D. Waters, H. D. Liebig, H. Klump, E. Kuechler, T. Skern, and R. E. Rhoads.1993. Mapping the cleavage site in protein synthesis initiation factor eIF-4g of the 2A proteases from human coxsackievirus and rhinovirus. J. Biol. Chem.268:19200–19203.
29. Le, H., R. L. Tanguay, M. L. Balasta, C. C. Wei, K. S. Browning, A. M. Metz, D. J. Goss, and D. R. Gallie.1997. Translation initiation factors eIF-iso4G and eIF-4B interact with the poly(A)-binding protein and increase its RNA binding activity. J. Biol. Chem.272:16247–16255.
30. Liebig, H. D., E. Ziegler, R. Yan, K. Hartmuth, H. Klump, H. Kowalski, D. Blaas, W. Sommergruber, L. Frasel, B. Lamphear, R. E. Rhoads, E. Kuechler and T. Skern.1993. Purification of two picornaviral 2A protein-ases: interaction with eIF4 and influence onin vitrotranslation. Biochemistry 32:7581–7588.
31. Lin, T. A., X. Kong, T. A. Haystead, A. Pause, G. Belsham, N. Sonenberg, and J. C. Lawrence, Jr.1994. PHAS-I as a link between mitogen-activated protein kinase and translation initiation. Science266:653–656.
31a.Lomakin, I. B., C. U. T. Hellen, and T. V. Pestova.2000. Physical association of eukaryotic initiation factor 4G (eIF4G) with eIF4A strongly enhances binding of eIF4G to the internal ribosome entry site of encephalomyocarditis virus and is required for internal initiation of translation. Mol. Cell. Biol. 20:6019–6029.
32. Lopez de Quinto, S., and E. Martinez-Salas.2000. Interaction of the eIF4G initiation factor with the aphthovirus IRES is essential for internal transla-tion initiatransla-tion in vivo. RNA6:1380–1392.
33. Meerovitch, K., and N. Sonenberg.1993. Internal initiation of picornavirus RNA translation. Semin. Virol.4:217–227.
34. Michel, Y. M., D. Poncet, M. Piron, K. M. Kean, and A. M. Borman.2000. Cap-poly(A) synergy in mammalian cell-free extracts: investigation of the requirements for poly(A)-mediated stimulation of translation initiation. J. Biol. Chem.275:32268–32276.
35. Morley, S. J., P. S. Curtis, and V. M. Pain.1997. Translation’s mystery factor begins to yield its secrets. RNA3:1085–1104.
36. Ohlmann, T., V. M. Pain, W. Wood, M. Rau, and S. J. Morley.1997. The proteolytic cleavage of eukaryotic initiation factor (eIF) 4G is prevented by eIF4E binding protein (PHAS-I; 4E-BP1) in the reticulocyte lysate. EMBO J.16:844–855.
37. Ohlmann, T., M. Rau, V. M. Pain, and S. J. Morley.1996. The C-terminal domain of eukaryotic protein synthesis initiation factor (eIF) 4G is sufficient to support cap-independent translation in the absence of eIF4E. EMBO J. 15:1371–1382.
38. Pestova, T. V., C. U. T. Hellen, and I. N. Shatsky.1996. Canonical eukaryotic initiation factors determine initiation of translation by internal ribosomal entry. Mol. Cell. Biol.16:6859–6869.
39. Pestova, T. V., I. N. Shatsky, and C. U. T. Hellen.1996. Functional dissection of eukaryotic initiation factor 4F: the 4A subunit and the central domain are sufficient to mediate internal entry of 43S preinitiation complexes. Mol. Cell. Biol.16:6870–6878.
40. Pilipenko, E. V., B. M. Blinov, B. L. Chernov, T. M. Dimitrieva, and V. I. Agol.1989. Conservation of the secondary structure elements of the 5⬘ -untranslated region of cardio- and aphthovirus RNAs. Nucleic Acids Res. 17:5701–5711.
41. Pilipenko, E. V., B. M. Blinov, L. I. Romanova, A. N. Sinyakov, S. V. Maslova, and V. I. Agol.1989. Conserved structural domains in the 5⬘ -untranslated region of picornaviral genomes: an analysis of the segment controlling translation and neurovirulence. Virology168:201–209. 42. Piron, M., T. Delaunay, J. Grosclaude, and D. Poncet.1999. Identification of
the RNA-binding, dimerization, and eIF4GI-binding domains of rotavirus nonstructural protein NSP3. J. Virol.73:5411–5421.
43. Piron, M., P. Vende, J. Cohen, and D. Poncet.1998. Rotavirus RNA-binding protein NSP3 interacts with eIF4GI and evicts the poly(A) binding protein from eIF4F. EMBO J.17:5811–5821.
44. Ptushkina, M., T. von der Haar, M. M. Karim, J. M. Hughes, and J. E. McCarthy.1999. Repressor binding to a dorsal regulatory site traps human eIF4E in a high cap-affinity state. EMBO J.18:4068–4075.
45. Sachs, A. B., P. Sarnow, and M. W. Hentze.1997. Starting at the beginning, middle and end: translation initiation in eukaryotes. Cell89:831–838. 46. Wang, S., K. S. Browning, and W. A. Miller.1997. A viral sequence in the
3⬘-untranslated region mimics a 5⬘cap in facilitating translation of uncapped mRNA. EMBO J.16:4107–4116.
47. Wei, C.-C., M. L. Balasta, J. Ren, and D. J. Goss.1998. Wheat germ poly(A) binding protein enhances the binding affinities of eukaryotic initiation factor 4F and (iso)4F for cap analogues. Biochemistry37:1910–1916.
48. Whetter, L. E., S. P. Day, O. Elroy-Stein, E. A. Brown, and S. M. Lemon. 1994. Low efficiency of the 5⬘nontranslated region of hepatitis A virus RNA in directing cap-independent translation in permissive monkey kidney cells. J. Virol.68:5253–5263.
49. Yan, R., W. Rychlik, D. Etchison, and R. E. Rhoads.1992. Amino acid sequence of the human protein synthesis initiation factor eIF4␥. J. Biol. Chem.267:23226–23231.
50. Ziegler, E., A. M. Borman, R. Kirchweger, T. Skern, and K. M. Kean.1995. Foot-and-mouth disease virus Lb proteinase can stimulate rhinovirus and enterovirus IRES-driven translation and cleave several proteins of cellular and viral origin. J. Virol.69:3465–3474.