Interaction of Eukaryotic Initiation Factor eIF4B with the Internal Ribosome Entry Site of Foot-and-Mouth Disease Virus Is Independent of the Polypyrimidine Tract-Binding Protein

Abstract

Eukaryotic translation initiation factor 4B (eIF4B) binds directly to the internal ribosome entry site (IRES) of foot-and-mouth disease virus (FMDV). Mutations in all three subdomains of the IRES stem-loop 4 reduce binding of eIF4B and translation efficiency in parallel, indicating that eIF4B is functionally involved in FMDV translation initiation. In reticulocyte lysate devoid of polypyrimidine tract-binding protein (PTB), eIF4B still bound well to the wild-type IRES, even after removal of the major PTB-binding site. In conclusion, the interaction of eIF4B with the FMDV IRES is essential for IRES function but independent of PTB.

Picornaviral translation is initiated cap independently from an internal ribosome entry site (IRES) far downstream from the RNA 5′ end. IRES elements contain several conserved stem-loops, a core element in their 3′ region (7), and a pyrimidine tract followed by an AUG codon at their 3′ borders (4) (Fig. 1). Eukaryotic initiation factor 2 (eIF2), eIF3, eIF4A, eIF4B, and the central domain of eIF4G are required for formation of 48S initiation complexes with encephalomyocarditis virus (EMCV) IRES RNA and 40S ribosomal subunits (15). In addition, picornaviruses recruit unconventional cellular proteins, e.g., the 57-kDa polypyrimidine tract-binding protein (PTB). PTB binds to several IRES elements and enhances translation of foot-and-mouth disease virus (FMDV) (13, 14), and EMCV (2, 5).

FIG. 1.

(A) The FMDV O₁K IRES with its five stem-loops, the pyrimidine tract (py), and the authentic AUG. The blowup on the right shows the IRES 3′ region with nucleotide numbers. Either the entire stem-loop 4 (mutant pSP449-Δ4), its subdomain 4-1 (Δ4-1) or 4-2 (Δ4-2), or stretches of four [bulge(−4), positions 743 to 746] or five [bulge(−5), positions 743 to 747] bases in bulge 4-3 were deleted (boxes). Mutant bulge(+) has four additional bases (UAGC) inserted (arrowhead). In mutant bulge(UU), A residues 741 and 742 were changed to UU. All these single-site mutations were introduced into the otherwise complete IRES, whereas in the double mutants, these mutations were combined with the precise deletion of stem-loop 2. The box within subdomain 4-1 marks the conserved element discussed in the text. (B) UV cross-link assays with standard reticulocyte lysate (RRL) with [α-³²P]CTP-labeled RNAs representing the complete IRES or single-site IRES mutants as indicated. Relative translation efficiencies of the mutants are given at the bottom. wt, wild type; nt, nucleotides.

eIF4B stimulates eIF4A, which unwinds mRNA secondary structures in cap-dependent translation initiation. The functional domains of eIF4B are an RNA recognition motif domain that binds to 18S rRNA (9, 10), a dimerization domain (11), and a basic RNA-binding domain (9). eIF4B binds directly to the 3′ part of the FMDV IRES (12), and early studies reported a function of eIF4B in picornavirus translation (3). However, we do not know how eIF4B exerts its function at the IRES and whether factors like PTB are directly supporting its function. In this study, we show the effects of mutations in the FMDV IRES 3′ part both on eIF4B binding and on translation efficiency. A possible interaction between eIF4B and PTB was tested with a reticulocyte lysate devoid of endogenous PTB and various IRES mutants.

eIF4B binding to IRES core subdomains.

Several mutations were introduced in stem-loop 4 in the context of the complete IRES (Fig. 1A). IRES RNA was synthesized with [α-³²P]CTP (12) from plasmid pSP449 (8) and incubated with rabbit reticulocyte lysate (RRL; Promega) in the presence of 100 mM K⁺, and bound proteins were analyzed by UV cross-linking (14) (Fig. 1B). With the complete IRES (lane 1), eIF4B migrating with an apparent molecular mass of 80 kDa (12) was strongly labeled, and PTB appeared as a 57-kDa band (14). When the complete stem-loop 4 (FMDV positions 649 to 755) was removed, binding of eIF4B was strongly reduced (lane 2) (12). Also, PTB binding was largely affected. Simultaneously, proteins of about 48 and 65 kDa were labeled more intensely, most probably due to the loss of competition by the strongly RNA-binding proteins eIF4B and PTB.

Deletion of subdomains 4-1 and 4-2 also resulted in largely reduced binding of eIF4B (lanes 3 and 4). The whole subdomain 4-1 shows a conserved secondary structure and includes an element of absolute primary and secondary structure conservation (4), two single-stranded dinucleotide stretches separated by 4 stacked bp (marked in Fig. 1A). These unpaired residues are perhaps displayed as contact sites for a protein like eIF4B. Also, larger alterations in the bulge 4-3 in mutants bulge(+), bulge(−4), and bulge(−5) (lanes 5 to 7) resulted in an almost complete loss of eIF4B binding. Thus, these changes either affect sequences directly involved in eIF4B binding or cause changes in the secondary or tertiary structure of the IRES that interfere with eIF4B binding. In contrast, when only two A residues in the bulge 4-3 were mutated to UU, only a slight decrease in eIF4B binding was observed (lane 8), indicating that this mutation has an intermediate effect on the contact of eIF4B with the IRES. The presence of the first authentic initiator AUG of FMDV was not required for eIF4B binding (data not shown), consistent with the finding that no AUG is found in the consensus sequence for eIF4B binding (10).

Binding of PTB was affected more differentially. Deletion of subdomain 4-1 caused only a slight reduction of the PTB binding. However, deletion of subdomain 4-2 largely reduced binding of PTB, confirming that this subdomain is one of the determinants required for binding of PTB to the IRES 3′ area (6). In contrast, the alterations in the bulge 4-3 region do not affect PTB binding essentially (lanes 5 to 8), although the binding of eIF4B was affected remarkably. Thus, PTB binding appears to be independent of eIF4B.

Translation efficiencies parallel eIF4B binding.

The described IRES mutants were inserted into a dicistronic mRNA system with the luciferase gene monitoring IRES activity (plasmid pD12, derived from pD128 [14]). Translation was highly efficient under the control of the wild-type IRES. Correlating with the largely reduced level of binding of eIF4B, translation efficiency was reduced to about 4% (Fig. 1B, bottom) with the stem-loop 4 deletion mutant. Correspondingly, the other alterations reduced the efficiency of the IRES-directed translation to levels of no more than 17%, except the bulge(UU) mutation, where the translation efficiency was reduced only slightly. Thus, the translation efficiencies of all mutants correlate with the reduced efficiency of eIF4B binding. In contrast, PTB binding to these IRES mutants does not correlate with translation efficiencies. These results indicate that (i) the IRES core including stem-loop 4 and its subdomains is essential for IRES function, (ii) eIF4B physically contacts the IRES core, and (iii) eIF4B is directly involved in the process of FMDV translation initiation.

eIF4B binds to the IRES in the absence of PTB.

With some mutations in the IRES core, binding of PTB was impaired in parallel with that of eIF4B (Fig. 1B, Δ4 and Δ4-2), raising the question whether binding of eIF4B could be supported by PTB. In order to investigate a possible interaction of PTB and eIF4B, we used a modified RRL from which PTB was depleted with poly(U)-Sepharose (Fig. 2A, lane 3) (14). The amount of eIF4B in this lysate was only slightly reduced. The PTB-depleted RRL was still competent for IRES-dependent translation, although at a reduced rate due to removal of PTB (14). In this PTB-depleted RRL, eIF4B still bound well to the complete IRES (Fig. 2B, lane 1). The intensity of the eIF4B band was slightly lower than that obtained with untreated RRL (compare with Fig. 1), due to the slight reduction during the depletion procedure (compare with Fig. 2A). In contrast, a PTB band was almost not detectable, suggesting that eIF4B binding is independent of PTB. With all mutants, eIF4B binding to the IRES was strongly reduced, even with the bulge(UU) mutation (lanes 2 to 8). The translation efficiencies of these mutants in the PTB-depleted lysate (Fig. 2B, bottom) were mainly lower than those in normal RRL (Fig. 1B).

FIG. 2.

Interaction of eIF4B with the IRES in PTB-depleted RRL. (A) Removal of PTB from RRL, checked by UV cross-link assays with complete IRES RNA with 5 μl of normal RRL (lane 1) or RRL after one (lane 2) or two (lane 3) cycles of poly(U)-Sepharose treatment. (B) UV cross-link assays with the IRES RNAs as in Fig. 1B but with PTB-depleted RRL. Relative translation efficiencies of the mutants in PTB-depleted RRL are given at the bottom. Retic., reticulocyte; norm., normal; dep., depletion; wt, wild type.

Even after PTB depletion, it had to be considered that this lysate perhaps still contained trace amounts of PTB that could enhance eIF4B binding. To rule out this possibility, we deleted the stem-loop 2, the major PTB-binding site, from the IRES (compare with Fig. 1A). This Δ2 mutation was also introduced into the single-site mutants described above, creating double mutants lacking the major PTB-binding site in combination with the mutations affecting eIF4B binding. Since the Δ2 mutation reduced the efficiency of translation to about 20% (14) and the effects of this deletion on the IRES tertiary structure are not known, only the qualitative aspects of eIF4B binding were analyzed. Only eIF4B, not PTB, bound to the Δ2 mutant in normal RRL (Fig. 3A), although large amounts of PTB are present in this lysate (Fig. 1B, lane 1). This confirms that binding of eIF4B to the FMDV IRES is independent of PTB. In analysis of the described double mutants (Fig. 3A, lanes 2 to 8), eIF4B binding was impaired as with the single-site mutants (Fig. 2A). Binding of eIF4B was clearly detectable even when the IRESΔ2 mutant was used in the PTB-depleted RRL (Fig. 3B, lane 1), whereas the IRES core mutations reduced binding of eIF4B below the detection limit.

FIG. 3.

Interaction of eIF4B with IRES double mutants lacking the major PTB-binding site (stem-loop 2) in addition to the single-site mutations shown in Fig. 1. (A) UV cross-link assays with standard RRL; (B) UV cross-link assays with PTB-depleted RRL.

In conclusion, we have demonstrated that binding of eIF4B to the IRES is independent of the binding and even the physical presence of PTB. This implies that the stimulatory effect of PTB on FMDV translation (13, 14) is not caused by supporting the interaction of eIF4B with the IRES. In contrast, other mechanisms must account for the stimulatory action of PTB, like other protein-protein interactions or stabilization of the RNA tertiary structure by an RNA chaperone function. In turn, the binding of PTB is not supported by eIF4B. However, this was not to be expected since PTB is considered an unconventional factor that plays a role only in enhancing translation activity on top of a basic level (14).

A key role of eIF4B in FMDV translation initiation.

One can only speculate about the actual function of eIF4B. It may represent the key factor that links the small ribosomal subunit to the IRES. On one hand, eIF4B binds directly to the IRES core and may provide a basic platform for subsequent protein-protein interactions which guide the 40S subunit to the IRES 3′ border, e.g., an interaction of eIF4B with the ribosome-bound eIF3, perhaps cooperatively enhanced by formation of eIF4B dimers (11). Another link between IRES and the 40S subunit may be provided by interaction of eIF4B with eIF4G, since formation of ribosomal 48S initiation complexes with the EMCV IRES is stimulated threefold by eIF4B (15). On the other hand, eIF4B may have an important catalytic function in positioning the IRES on the ribosome. Binding of eIF4B to the FMDV IRES is ATP dependent, suggesting the involvement of the RNA helicase eIF4A (12). Moreover, eIF4B can catalyze RNA-RNA hybridizations (1). An eIF4B-eIF4A complex with a combined annealase-helicase activity may adjust the initiator AUG to the right position on the 18S rRNA by alternating annealing and melting events, resulting in correct hybridization of sequences in the IRES pyrimidine tract to complementary sequences in the 18S rRNA (1, 10, 16).