ИСТИНА

All nuclear encoded eukaryotic mRNAs bear the m7G cap structure at their 5’-ends. 30 years ago, the scanning model was proposed to describe a mechanism of AUG selection in mRNAs with rather short and simple leaders. However, growing evidence suggests that many mammalian mRNAs possess a 5’UTR of more than 200 nt, with high GC content and complex secondary structure – features which are thought to impede their translation initiation by the mechanism of cap-dependent scanning. For such mRNAs, the presence of IRESes was suggested to explain their efficient translation, in parallel to IRES elements that had been discovered earlier within naturally uncapped mRNAs of some cytoplasmically propagating viruses. The IRES hypothesis is also appealing as an explanation of the relative resistance of some mRNA translation to special conditions of cell stress, mitosis or apoptosis, when cap-dependent translation is inhibited. For verification of the presence of an IRES in the given 5’UTR, the bicistronic test has been proposed as a major criterion. According to this approach, the sequence is declared as possessing IRES activity, if only stimulates, to any extent, the translation of the second gene in the bicistronic construct, as compared to the “empty” vector. The two most popular variants of the bicistronic test are usually applied are transfection of cultured cells with a bicistronic DNA plasmid (for in vivo tests) and translation of the corresponding bicistronic transcript in vitro in rabbit reticulocyte lysate, the widely used mammalian cell-free system. However, both of these methods have been shown recently to produce a collection of artifacts. The RNA transfection procedure has been suggested as an alternative approach, but poor IRES activity revealed by this method has been explained by hypothetical “nuclear events” required for IRESes to operate. In this work, we abandoned the comparison of different bicistronic mRNAs as a major criterion for testing the putative IRESes, since all eukaryotic mRNAs are both monocistronic and capped, and bicistronic constructs therefore represent an artificial situation. Instead, we aimed to identify the actual rather than potential mechanism of translation initiation directed by particular 5’UTRs (derived from beta-actin, beta-globin, LINE-1, Apaf-1, c-myc, hsp70 and several viral mRNAs) in various situations. Indeed, all the sequences we have tested, except viral IRESes, demonstrate extremely low translation level when being placed between two cistrons, both in Krebs-2 cells S30 extract and in living cells transfected with mRNAs. In contrast, they were unexpectedly efficient in the monocistronic context, irrespective of their length and GC content, contradicting the scanning model prediction. The m7G cap was indispensable for efficient translation of all monocistronic transcripts directed by putative cellular IRESes. Importantly, the m7G cap requirement was maintained also under conditions of cell stress or when the recombinant 4E-BP1 was added to the cell-free system, in contrast to the situation with viral IRES elements. The ability to direct translation under these stress conditions varies between different cap-dependent 5’UTRs. Our data show clearly that there is no need to invoke the IRES hypothesis for a long and structured 5’UTR of mRNA to explain their efficient translation under normal conditions and relative resistance of some mRNAs to partial eIF4F inactivation under conditions of stress.