Translation initiation in eukaryotes is generally dependent on a structure termed the cap (m
7GpppN, where m is a methyl group and N is any nucleotide), which is present at the 5′ends of all nuclear transcribed mRNAs. The cap recruits the 40S ribosomal subunit through its interaction with the cap binding protein complex, eIF4F, which forms a bridge between the mRNA and the 40S ribosomal subunit through eIF3 (
19,
27). Following its recruitment to the mRNA 5′end, the 40S ribosomal subunit in association with several initiation factors, is thought to migrate along the 5′ untranslated region (5′UTR) until an initiation codon is encountered (
27). However, in a sizeable fraction of eukaryotic cellular mRNAs (
9,
25,
70), the 40S ribosomal subunit is recruited to the mRNA via an alternative mechanism using an internal ribosome binding site (IRES). IRESs were first discovered in picornavirus mRNAs (poliovirus and encephalomyocarditis virus), which are naturally uncapped (
38,
54), and subsequently identified in many cellular mRNAs (reviewed in references
25 and
70; see the Internal Ribosome Entry Site database [
7] for an updated list). Strikingly, cellular IRESs are mainly found in mRNAs that are translated under stress conditions (
31). These conditions include hypoxia, apoptosis, serum starvation (
26,
29,
32,
43,
45), and mitosis (
10,
58). Thus, control of IRES-dependent translation plays a critical role in regulation of cell growth and survival.
To understand the mechanism of control of IRES function, it is imperative to study the mode by which the 40S ribosome subunit is recruited to the IRES. To this end, we and others undertook the identification and characterization of cellular proteins that interact with the poliovirus IRES. The studies of Meerovitch et al. led to the identification of the first IRES
trans-acting factor (ITAF), the La autoantigen (
47,
48). Subsequently, La was shown to bind to the hepatitis C virus (HCV) IRES near the initiation codon and to stimulate HCV RNA translation in a rabbit reticulocyte lysate at low concentrations (
1,
2). The HCV IRES-La interaction has been recapitulated in a
Saccharomyces cerevisiae three-hybrid system (
57). In addition, La binds to cellular IRESs, such as X-linked inhibitor of apoptosis (XIAP) (
30) and BiP (
42).
Initiation of poliovirus translation from the correct initiation site in a rabbit reticulocyte lysate is feeble, and aberrant products are readily detected (
8,
14), possibly due to limiting amounts of one or more translation factors. Interestingly, the amount of La protein present in reticulocyte lysate is very low compared to extracts from nucleated cells (
48), and the addition of recombinant La enhances poliovirus RNA translation and eliminates the production of aberrant proteins (
48,
65). It was thus concluded that La protein is a bona fide ITAF. However, the physiological significance of these studies was questioned (
35,
36), because of the large amounts of recombinant La that had been used (10-fold more recombinant La than the amount of La protein present in HeLa cell extract). Thus, the possibility that La is not a physiological ITAF but rather mimics a genuine poliovirus ITAF, owing to its RNA binding properties was raised (
35,
36,
72). To address this issue, we studied the requirement of La autoantigen for poliovirus translation in vivo. Using small interfering RNA (siRNA) and a dominant-negative mutant of La (LaDN), we demonstrate that La is required for optimal translation of poliovirus. In addition, we demonstrate that LaDN inhibits the initiation of formation of 48S ribosome complexes on HCV and poliovirus mRNAs.