Saumitra Das¹, Michael Ott², Akemi Yamane¹, Arun Venkatesan¹, Sanjeev Gupta² and Asim Dasgupta¹

Department of Microbiology, Molecular Genetics and Immunolog, UCLA School of Medicine, 10833 Le Conte Avenue, Los Angeles, CA 90095-1747¹ , ²Department of Medicine, Albert Einstein, College of Medicine of Yeshiva University, Bronx, New York 10461-1602

Received 9/18/98 Accepted 9/23/98

2. INTRODUCTION

Mammalian plus strand RNA viruses cause a multitude of infectious diseases in humans and animals, and currently, no effective vaccination is available against the majority of these viruses. The RNA genomes of some of these viruses (picorna and flaviviruses) are translated by a common mechanism known as internal initiation of translation where the ribosomes bind to an internal sequence within the 5´ -untranslated region (UTR) near the initiator AUG of the viral RNA This sequence is termed internal ribosome binding site (IRES) (48). This method of eukaryotic protein synthesis is different from the cap-dependent translation of cellular mRNAs. This fundamental difference could serve as a target for antiviral agents that would preferentially stop viral proliferation without affecting the host. Studies from several laboratories suggest the importance of specific interaction of cellular proteins with the IRES elements of the viral RNAs. Apparently there is no resemblance in the RNA sequence amongst different IRES elements, yet they exhibit some common features both at the level of cis-acting RNA elements and transacting factors. Identification and characterization of specific protein factors that are involved in this process are necessary for a better understanding of this relatively new phenomenon of eukaryotic protein synthesis. We believe one or more common factors may be required exclusively for IRES-mediated translation and once identified, these factors may be targeted to inhibit viral translation.

Recently, we have isolated a small RNA (I-RNA) from the yeast Saccharomyces cerevisiae, which specifically inhibits translation programmed by the IRES sequences of different viral RNAs, including those inducing common cold, poliomyelitis and infectious and chronic hepatitis (HAV and HCV). The inhibitor RNA does not act as an antisense RNA. Rather, it specifically binds certain cellular proteins required for internal initiation of translation thereby preventing them from interacting with the 5´ -untranslated region (5´ -UTR) of viral RNA.

Since the naturally occurring small yeast RNA (I-RNA) appears to block translation of the viral RNA but does not significantly affect cellular capped mRNA translation, it is clearly important to determine whether the I-RNA or it’s derivatives can be used as antivirals against viruses that use IRES mediated translation. We have recently provided evidence that the I-RNA can fold into a stable secondary structure, which mimics part of the viral 5´ - UTR, and thus competes for protein binding (Venkatesan, Das and Dasgupta, unpublished). Thus, determination of its three dimensional structure might lead to the design of small molecules which would be biologically more effective in inhibition of viral RNA translation. The normal function of IRNA in yeast is not known. Some of the yeast genes are reported to be internally initiated (27, 28), and thus it is possible that IRNA may regulate the expression of these genes by inhibiting internal initiation of translation.

2.1. Internal initiation of translation in eukaryotes

To date, three different modes of initiation of eukaryotic mRNA translation have been discovered. The majority of the eukaryotic mRNAs are translated by the scanning mechanism whereby 40S ribosomal subunits bind to the 5´ -cap structure and then scan the mRNA in a 5´ to 3´ direction until an appropriate initiator AUG is encountered (34). The second method is the reinitiation mechanism where after translating an open reading frame, the 40S subunits resume scanning and start initiation at a further downstream AUG codon; the best example is the yeast GCN4 gene (21, 26). The third mechanism is known as the internal initiation. Certain viral and cellular mRNAs contain sequences in their 5´ -NCR that can direct cap-independent translation. The sequences mediating the internal recruitment of ribosomes have been termed the internal ribosome entry site (IRES) and can be located hundreds of nucleotides downstream from the 5´ -end of mRNA. The first example of initiation of translation by internal binding of ribosomes was found in picornaviral mRNAs (5,6,9,19,20,32,36,37,49,56). In sharp contrast to cellular mRNA translation, the naturally uncapped picornaviral RNAs translate inside infected cells in an environment where cap dependent translation of the majority of cellular capped mRNAs is impaired due to cleavage of the cap binding protein complex (39). It is through IRES mediated translation that picornaviral proteins are synthesized in infected cells, while cellular protein synthesis is inactivated.

Since internal initiation of translation is carried out by the cellular transitional apparatus in uninfected cells, the question arises whether cellular mRNAs can be translated by a similar cap-independent manner. It is possible that some capped mRNAs can be translated by both cap dependent scanning and internal ribosome binding mechanisms. Such RNAs may contain IRES sequences in their 5´ -non-coding regions that allow them to be translated at times when eIF-4F is nonfunctional and cannot bind to the 5´ -cap structure. Alternatively, an IRES located within the coding region of mRNA could render the mRNA functionally polycistronic and translation could result in the production of several protein products. In fact, there are reports that the cellular mRNAs encoding immunoglobulin heavy chain binding protein (38), Drosophila antennapedia gene (45), the mouse androgen receptor (24), and the mammalian fibroblast growth factor 2 gene (59) utilize IRES mediated initiation of protein synthesis. Recently, Chen and Sarnow have demonstrated that the mammalian ribosomes can translate a covalently closed circular mRNA if it includes a functional IRES element (11).

2.2. The structure of IRES elements in different groups of viruses

The 5´ -UTR plays a major role in translation of picornaviral RNAs. Usually the 5´ -UTR is relatively long (varies from 600 - 800) and can fold to generate structures consisting of a number of stems and loops. Interestingly though, the primary sequences are not conserved but the RNA secondary structures are very similar, suggesting a possible role of RNA folding in virus translation. On the basis of the sequence and structural similarities of the IRES elements, the picornaviruses can be classified into three groups, 1) entero and rhino virus, 2) cardio and apthovirus, and 3) hepatitis A virus (30 ). A pyrimidine rich tract (25 nt) upstream of the initiator AUG is found to be conserved in the 5´ -UTR of almost all picornaviral RNAs. Whether these conserved residues in loops and bulges are essential for the internal ribosome entry is not clear (1,29,33). A number of cellular protein factors have been shown to interact with the IRES elements but how they influence internal initiation is yet unclear. It is possible that the interaction helps in proper RNA folding and/or maintaining the higher order secondary structures of the 5-´ UTR RNA during translation. There are reports (but no direct evidence) that the IRES/protein interaction might help in assuming correct RNA folding to facilitate IRES/rRNA interaction. In fact, results from different laboratories have shown the existence of significant pseudoknot structures common among EMCV, TMEV, FVDV and HAV which might assume an efficient binding structure for the ribosomes during translation initiation. The base pairing between human 18S rRNA and the IRES elements of picornaviruses is another plausible model (30,31,53,).

Hepatitis C virus (HCV), a member of the flavivirus group, is the causative agent of hepatitis and is transmitted through blood. HCV infection often leads to chronic hepatitis, cirrhosis of liver and hepatocellular carcinoma. HCV 5´ -UTR is 341 nt long and is highly structured amongst different strains and isolates (10). The exact boundaries of the IRES is somewhat controversial; requirement of a short length of 5´ -proximal HCV polyprotein coding sequences has been reported for efficient IRES function (35, 58).

2.3. Cellular proteins that interact with the 5´ -UTR of the viral RNAs

Studies from several laboratories suggest the importance of specific interactions of cellular proteins with various elements containing secondary structures within the 5´ -UTR of viral RNAs. Two nuclear proteins have been identified that are apparently required for the efficient usage of the IRES by the host cell translation apparatus. One of them is a 52 Kd polypeptide (p52) which has been identified as the La autoantigen (41). The other cellular protein p57 appears to be identical to the polypyrimidine tract binding protein (PTB) (25). La is involved in RNA polymerase III transcription termination and PTB is a part of the nuclear spliceosome complex involved in RNA polymerase II transcript splicing. It is believed both La and PTB have dual roles in cytoplasmic mRNA translation and nuclear RNA biogenesis (11). Both La and PTB could be immunodepleted from in vitro translation extracts resulting in loss of picornaviral mRNA translation. Interestingly, translation in these depleted extracts could not be restored with the addition of purified La or PTB (25,57). These findings suggest that La and PTB may stimulate the IRES element in concert with additional factors to which they are tightly bound. The stimulatory effect of La and PTB on picornavirus IRES could be due to the recruitment of La and PTB associated factors to the IRES. Such factors could then directly mediate the binding of ribosomal subunits to the IRES. In contrast, recent studies using mutants having deletions of different stem loop domains of poliovirus 5´ -UTR demonstrated that interactions of neither p57 nor p52 is absolutely required for internal ribosome binding (22,23,50). Thus, functional significance of La and PTB in IRES-mediated translation is not clearly established and possible involvement of other RNA-protein and/or protein-protein interactions are being explored. In fact, several additional RNA binding proteins has been shown to be crosslinked to type 1 and type 2 IRES elements of poliovirus including p37, p39, p48, p70, p80, p100 and p110 (7,16,18,44). The 37 kD HeLa cell protein that binds to the IRES of HAV has been identified as cellular Glyceraldehyde-3-phosphate dehydrogenase(GAPDH) (55). Another protein of molecular mass 39 kD (binds to stem-loop IV of polio virus 5´ -UTR) has been identified as poly(rC) binding protein (PCBP2) (8, 17, 51), One or more of these cellular protein factors also bind to Hepatitis C 5´ -UTR and to the regulatory elements of human T-cell leukemia virus type 2 RNA (45,52,60,61,62). Thus, it will be interesting to identify the common cellular protein factors that specifically bind to all IRES elements with some sort of functional significance. There are few reports on the requirement of tissue specific transacting factors in IRES-mediated translation. It has been observed that liver-specific factors can stimulate the IRES function of the hepatitis A virus (20).

2.4. Requirement of canonical eIFs in IRES-mediated translation

Initiation of eukaryotic protein synthesis involves the assembly of 80S initiation complex containing initiator tRNA_i^MET, 40S and 60S ribosomal subunits at the initiation codon. In cap-dependent translation a 43S complex is formed that consists of eIF3 and a ternary complex of Met-tRNA_i^Met: eIF2: GTP bound to the 40S subunit. In the next step the 43S complex binds to the 5´ -end of the mRNA at the 5´ m7GpppX cap structure. The cap-dependent binding of ribosomes requires at least eukaryotic initiation factor eIF4F and ATP. eIF4F consists of three subunits, eIF4A, eIF4E, and eIF4G. eIF4A exhibits RNA-dependent ATPase activity, and together with eIF4B, RNA helicase activity. The eIF4G subunit coordinates the binding of eIF4F to the m7GpppCap by interacting with eIF4A , eIF4E, eIF3 and the RNA. The eIF4B-induced helicase activity of eIF4F melts the RNA secondary structure to generate a single stranded(ss) region near the cap. eIF4B then interacts with 18S rRNA to guide the 40S ribosomal subunit to the ss region of the mRNA (3,47,51,54,).

In case of cap-independent translation, the IRES elements promote binding of the 43S complex to a position far from the 5´ cap, and closer to the initiator AUG. Stimulation of EMCV- IRES mediated translation by one or more eIFs has been shown. eIF2/2B and eIF4B binding to the EMCV and FMDV IRES have been reported from different laboratories (4,43,54). In a recent report on in vitro reconstituted IRES-mediated translation, it was demonstrated that eIF2, eIF3 and eIF4F were required for initiation whereas eIF4B and to a lesser extent PTBP stimulated the process (51). Interestingly, translation of capped mRNA by the scanning mechanism absolutely requires the intact eIF4F holoenzyme complex. Infection with enteroviruses, rhinoviruses, or FMDV results in cleavage of the eIF4G component of the eIF4F rendering it inactive in promoting initiation on capped mRNA. However, the cleaved factor (incomplete eIF4F complex) stimulates IRES-dependent translation more efficiently than the intact eIF4F holoenzyme. It is believed that internal initiation is driven by the C-terminal cleavage product of eIF4G with its associated eIF4A. Despite the subtle differences, the overall similarity of the initiation factor requirements suggests that the two mechanisms can not be fundamentally very different (30). It is probably the first step i.e. the binding of the 43S complex to the internal sequence, that constitutes the primary difference between cap-dependent and IRES-mediated translation.