Cells respond to stress stimuli through coordinated changes in gene expression. The regulation of translation is often used under these circumstances because it allows immediate and selective changes in protein levels. There are many examples of translational control in response to stress. Here we examine two representative models, the regulation of eukaryotic initiation factor-2alpha by phosphorylation and internal ribosome initiation through the internal ribosome-entry site, which illustrate the importance of translational control in the cellular stress response and apoptosis.

SUMMARYProtein synthesis involves the translation of ribonucleic acid information into proteins, the building blocks of life. The initial step of protein synthesis consists of the eukaryotic translation initiation factor 4E (eIF4E) binding to the 7-methylguanosine (m7-GpppG) 5′cap of mRNAs1,2. Low oxygen tension (hypoxia) represses cap-mediated translation by sequestering eIF4E through mammalian target of rapamycin (mTOR)-dependent mechanisms3–6. While the internal ribosome entry site is an alternative translation initiation mechanism, this pathway alone cannot account for the translational capacity of hypoxic cells7,8. This raises a fundamental question in biology as to how proteins are synthesized in periods of oxygen scarcity and eIF4E inhibition9. Here, we uncover an oxygen-regulated translation initiation complex that mediates selective cap-dependent protein synthesis. Hypoxia stimulates the formation of a complex that includes the oxygen-regulated hypoxia-inducible factor 2α (HIF-2α), the RNA binding protein RBM4 and the cap-binding eIF4E2, an eIF4E homologue. PAR-CLIP10 analysis identified an RNA hypoxia response element (rHRE) that recruits this complex to a wide array mRNAs, including the epidermal growth factor receptor (EGFR). Once assembled at the rHRE, HIF-2α/RBM4/eIF4E2 captures the 5′cap and targets mRNAs to polysomes for active translation thereby evading hypoxia-induced repression of protein synthesis. These findings demonstrate that cells have evolved a program whereby oxygen tension switches the basic translation initiation machinery.

rogrammed cell death (apoptosis) plays a critical part in regulating cell turnover during embryogenesis, metamorphosis, tissue homeostasis and viral infection 1 . Dysregulation of apoptosis occurs in such pathologies as cancer, autoimmunity, immunodeficiency and neurodegeneration. Proteins of the inhibitor-ofapoptosis (IAP) family are intrinsic cellular suppressors of apoptosis and are represented by highly conserved members found from insect viruses to mammals 2-4 . The most potent mammalian IAP is the X-linked IAP, or XIAP 5 , whose mechanism of action involves direct inhibition of caspases 3 and 7, key proteases of the apoptotic cascade 6 . Cellular control of XIAP expression should be fundamental to a cell's ability to modulate its responses to apoptotic stimuli. However, XIAP messenger RNA is expressed in most tissues and cells at fairly constant levels 5 , indicating that translational control of XIAP levels may be an important regulatory mechanism. Here we characterize the primary genomic structure and function of XIAP, and show that XIAP expression is controlled at the translational level, specifically through an internal ribosome-entry site (IRES).Several features of XIAP mRNA indicate that it may be translationally regulated, including an unusually long 5′ untranslated region (UTR) (>5.5 kilobases (kb) for murine and >1.6 kb for human XIAP transcripts) with predicted complex secondary structure and numerous potential translation start sites upstream of the authentic initiation codon. This UTR would be expected to present a significant obstacle to efficient translation by conventional ribosome scanning 7 . An alternative mechanism of translation initiation, mediated through the IRES, has been identified in picornaviruses and in a few cellular mRNAs 8 . Thus we tested whether the 5′ UTR of XIAP mRNA could be involved in translation initiation from reporter-based bicistronic mRNA transcripts encoding β-galactosidase and chloramphenicol aceytyltransferase (CAT) (for example, see ref. 9). (Translation of β-galactosidase is driven by the 5′ mRNA methylguanosine cap.) Both human and mouse XIAP 5′ UTRs directed translation of the second cistron (encoding CAT) at 150-fold higher levels than those produced without the 5′ UTR or with the 5′ UTR in reverse orientation, suggesting the presence of an IRES (Fig. 1a). No activity was detected when using the identical DNA segments cloned into a promoterless construct, confirming P Figure 1 The 5′ UTR of mouse and human XIAP mRNA contains functional IRES elements. a, DNA segments corresponding to the indicated regions of the 5′ UTR of human (h) or mouse (m) XIAP transcripts were inserted (in the indicated directions) into the XhoI site of the linker region (LR) of the bicistronic plasmid pβgal/CAT (where βgal is β-galactosidase); HeLa cells were transfected with these plasmids (2 µg each). The promoterless CAT reporter plasmid pCATbasic/UTR was constructed by inserting the indicated 5′ UTR region into the pCATbasic vector, and in this case HeLa cells were co-transfected ...

The cell has many ways to regulate the production of proteins. One mechanism is through the changes to the machinery of translation initiation. These alterations favor the translation of one subset of mRNAs over another. It was first shown that internal ribosome entry sites (IRESes) within viral RNA genomes allowed the production of viral proteins more efficiently than most of the host proteins. The RNA secondary structure of viral IRESes has sometimes been conserved between viral species even though the primary sequences differ. These structures are important for IRES function, but no similar structure conservation has yet to be shown in cellular IRES. With the advances in mathematical modeling and computational approaches to complex biological problems, is there a way to predict an IRES in a data set of unknown sequences? This review examines what is known about cellular IRES structures, as well as the data sets and tools available to examine this question. We find that the lengths, number of upstream AUGs, and %GC content of 59-UTRs of the human transcriptome have a similar distribution to those of published IRES-containing UTRs. Although the UTRs containing IRESes are on the average longer, almost half of all 59-UTRs are long enough to contain an IRES. Examination of the available RNA structure prediction software and RNA motif searching programs indicates that while these programs are useful tools to fine tune the empirically determined RNA secondary structure, the accuracy of de novo secondary structure prediction of large RNA molecules and subsequent identification of new IRES elements by computational approaches, is still not possible.

No abstract