Portal de Periódicos da CAPES

Quality Control of mRNA Function

2001; Cell Press; Volume: 104; Issue: 2 Linguagem: Inglês

10.1016/s0092-8674(01)00202-1

1097-4172

Lynne E. Maquat, Gordon Carmichael,

RNA regulation and disease

Most eukaryotic cellular proteins are generated by translation of mRNAs that have successfully passed a number of steps to insure their quality. Transcripts that fail to pass selection are degraded or otherwise prevented from engaging in protein synthesis. Recent studies have begun to elucidate the molecular mechanisms responsible for the pre- and cotranslational quality control (QC) of mRNA and the consequence to organisms when control fails. mRNA QC reflects physical interactions between components of the gene transcription, pre-mRNA processing, mRNA transport, and mRNA translation machineries. These interactions insure against transcript degradation when interactions are timely and proper, but elicit nuclear retention of the transcript, or degradation in the nucleus or cytoplasm, when interactions are delayed or abnormal. mRNA QC begins at the level of gene transcription. The cotranscriptional nature of pre-mRNA processing offers a number of opportunities for transcription to affect the quality and quantity of mRNA. Evidence that the transcriptional machinery can regulate pre-mRNA processing derives in part from the finding that sequence changes in the human fibronectin promoter can alter splice site selection, possibly through the differential recruitment of splicing factors, in this case SR proteins SF2/ASF and 9G8, to the CTD, the carboxyl terminal domain of the largest subunit of RNA polymerase II (5Cramer P. Cáceres J.F. Cazalla D. Kadener S. Muro A.F. Baralle F.E. Kornblihtt A.R. Mol. Cell. 1999; 4: 251-258Abstract Full Text Full Text PDF PubMed Scopus (250) Google Scholar). As additional evidence, the transcription factor TFIID recruits the 3′ end formation factor CPSF to promoters, where it is subsequently transferred to the CTD so as to direct endonucleolytic cleavage of nascent transcripts followed by polyadenylation of the upstream cleavage product (see, e.g., 8Hirose Y. Manley J.L. Genes Dev. 2000; 14: 1415-1429PubMed Google Scholar, and references therein). In fact, an abundance of data indicate that the CTD not only functions to regulate aspects of gene transcription, especially initiation, but also functions in association with factors required for pre-mRNA capping, splicing, and 3′ end formation to coordinate pre-mRNA synthesis and processing throughout transcription initiation, elongation, and termination (Figure 1) (8Hirose Y. Manley J.L. Genes Dev. 2000; 14: 1415-1429PubMed Google Scholar, 17Proudfoot N. Trends Biochem. Sci. 2000; 25: 290-293Abstract Full Text Full Text PDF PubMed Scopus (139) Google Scholar, and references therein). Providing perhaps the most striking visual evidence for the QC of mRNA formation throughout transcription, antibodies against splicing or polyadenylation factors react uniformly along the entire length of Xenopus oocyte lampbrush chromosomes, coincident with RNA polymerase II staining (7Gall J.G. Bellini M. Wu Z. Murphy C. Mol. Biol. Cell. 1999; 10: 4385-4402Crossref PubMed Scopus (234) Google Scholar). Somewhat unexpectedly, RNA polymerase II also functions in mRNA QC by playing a direct and active role in splicing and 3′ end formation in the absence of transcription (8Hirose Y. Manley J.L. Genes Dev. 2000; 14: 1415-1429PubMed Google Scholar, and references therein). The major steps of pre-mRNA processing appear to be coupled to each other as well, a feat also thought to be orchestrated by the CTD as an additional measure of QC. Efficient capping enhances both splicing and 3′ end formation, and efficient 3′ end formation enhances splicing (8Hirose Y. Manley J.L. Genes Dev. 2000; 14: 1415-1429PubMed Google Scholar, 17Proudfoot N. Trends Biochem. Sci. 2000; 25: 290-293Abstract Full Text Full Text PDF PubMed Scopus (139) Google Scholar, and references therein). Recent studies of Saccharomyces cerevisiae have uncovered a nuclear discard pathway for inefficiently spliced pre-mRNAs that appears to function in competition with the splicing pathway (3Bousquet-Antonelli C. Presutti C. Tollervey D. Cell. 2000; 102: 765-775Abstract Full Text Full Text PDF PubMed Scopus (305) Google Scholar). Transcripts are degraded primarily in a 3′ to 5′ direction by the exosome, but can also be degraded 5′ to 3′ by Rat1p. Both the exosome and Rat1p have orthologs in mammals, suggesting that similar modes of degradation exist in higher eukaryotes. In light of recent data demonstrating that transcription termination requires a poly(A) signal but not transcript cleavage (8Hirose Y. Manley J.L. Genes Dev. 2000; 14: 1415-1429PubMed Google Scholar, 17Proudfoot N. Trends Biochem. Sci. 2000; 25: 290-293Abstract Full Text Full Text PDF PubMed Scopus (139) Google Scholar, and references therein), it is conceivable that this discard pathway also eliminates transcripts that undergo inefficient 3′ end formation. After pre-mRNA processing, product mRNAs are exported to the cytoplasm. Export offers yet another form of mRNA QC because it is a selective and active process that discriminates between immature and mature mRNP. mRNA export is attributable to bound proteins that promote export, many of which are acquired as a consequence of pre-mRNA processing, and can be inhibited by bound proteins that block export, some of which are the consequence of incomplete pre-mRNA processing (see, e.g., 16Nakielny S. Dreyfuss G. Cell. 1999; 99: 677-690Abstract Full Text Full Text PDF PubMed Scopus (626) Google Scholar, and references therein). Defects in capping, splicing, and 3′ end formation inhibit export in both S. cerevisiae and mammalian cells (16Nakielny S. Dreyfuss G. Cell. 1999; 99: 677-690Abstract Full Text Full Text PDF PubMed Scopus (626) Google Scholar, and references therein). Roles for mature 5′ and 3′ ends, coupled with electron microscopic analysis of Balbiani ring mRNPs in the process of export (6Daneholt B. Cell. 1997; 88: 585-588Abstract Full Text Full Text PDF PubMed Scopus (124) Google Scholar), have led to a model where export is promoted by recognition of both ends of an mRNP at the nuclear pore. According to this model, failure to recognize either end would signal the presence of an aberrant particle, which would retard export (Figure 2). But proper capping and 3′ end formation are not sufficient to insure mRNA export, as would be expected since they are also characteristic of a number of incompletely spliced pre-mRNAs. Factors that block export, such as some components of spliceosomes and other non-shuttling proteins, must be removed. For example, COL1A1 transcripts defective in splicing initiate transport away from their site of synthesis but subsequently accumulate to abnormal levels within a nuclear domain rich in the SC-35 splicing factor (10Johnson C. Primorac D. McKinstry M. McNeil J. Rowe D. Lawrence J.B. J. Cell Biol. 2000; 150: 417-432Crossref PubMed Scopus (124) Google Scholar, and references therein). Additionally, the transport of many if not most mRNAs appears to reflect the binding of shuttling hnRNP and perhaps SR proteins (16Nakielny S. Dreyfuss G. Cell. 1999; 99: 677-690Abstract Full Text Full Text PDF PubMed Scopus (626) Google Scholar, and references therein) as well as a direct coupling of pre-mRNA splicing and mRNA export via splicing-dependent changes in mRNP structure. Recent studies have revealed that the general mRNA export factor Aly is recruited to pre-mRNA during spliceosome assembly and subsequently becomes tightly associated with spliced mRNPs in a way that promotes mRNP export (20Zhou Z. Lou M.-j. Straesser K. Katahira J. Hurt E. Reed R. Nature. 2000; 407: 401-405Crossref PubMed Scopus (387) Google Scholar). Aly (also called REF) and Y14, another protein implicated indirectly in mRNA export, have been shown to be components of an ≅335 kDa splicing-dependent complex centered 20–24 nucleotides upstream of the exon–exon junctions of mRNPs generated by splicing in HeLa cell nuclear extract (12Le Hir H. Izaurralde E. Maquat L.E. Moore M.J. EMBO J. 2000; 19: 6860-6869Crossref PubMed Scopus (677) Google Scholar, and references therein). A similar mRNP complex is generated by splicing in Xenopus oocytes (12Le Hir H. Izaurralde E. Maquat L.E. Moore M.J. EMBO J. 2000; 19: 6860-6869Crossref PubMed Scopus (677) Google Scholar). mRNAs that derive from genes that naturally lack introns might co-opt some of the same factors used for the export of spliced mRNAs, or they may use a different pathway. QC is also exerted during mRNA translation, another selective and active process that discriminates between improper and proper mRNP. Generally, the cap and poly(A) tail of an mRNA contribute to cytoplasmic stability, presumably by stimulating translation and, thereby, physically protecting the mRNA from accessibility to nucleases (see, e.g., 15Mitchell P. Tollervey D. Curr. Opin. Genet. Dev. 2000; 10: 193-198Crossref PubMed Scopus (233) Google Scholar, and references therein). A different type of translational QC is thought to survey all translated mRNAs in order to prevent the synthesis of proteins from those that prematurely terminate translation as a consequence of errors in gene expression. Errors include aberrant transcription initiation, inaccurate or inefficient splicing, the failure to incorporate selenocysteine at specific UGA codons of certain selenoprotein mRNAs, and frameshift or nonsense mutations within germ-line or somatic DNA. This QC mechanism has been called nonsense-mediated mRNA decay (NMD) or mRNA surveillance. NMD is, in essence, the downregulation of mRNA translation and activation of mRNA decapping and degradation as a consequence of the premature termination of translation (see, e.g., 9Jacobson, A., and Peltz, S.W. (2000). In Translational Control of Gene Expression, N. Sonenberg, J.W.B. Hershey, and M.B. Mathews, eds. (Cold Spring Harbor Press: Cold Spring Harbor, NY), pp. 827–847.Google Scholar, 14Maquat, L.E. (2000). In Translational Control of Gene Expression, N. Sonenberg, J.W.B. Hershey, and M.B. Mathews, eds. (Cold Spring Harbor Press: Cold Spring Habor, NY), pp. 849–868.Google Scholar, and references therein). NMD in S. cerevisiae is thought to take place in the cytoplasm. The trigger has been proposed to be an abnormal 3′-untranslated region, possibly reflecting an abnormal distance between a termination codon and the site of polyadenylation or, according to an alternative view, the presence of a loosely defined destabilizing element (9Jacobson, A., and Peltz, S.W. (2000). In Translational Control of Gene Expression, N. Sonenberg, J.W.B. Hershey, and M.B. Mathews, eds. (Cold Spring Harbor Press: Cold Spring Harbor, NY), pp. 827–847.Google Scholar, and references therein). The destabilizing element is hypothesized to recruit one of a number of shuttling proteins, e.g., Hrp1p, in the nucleus that, if located downstream of a termination codon, interacts with components of the translation termination complex in the cytoplasm in a way that elicits NMD. In contrast, NMD in mammalian cells can take place either in the cytoplasm or in association with nuclei, and is generally triggered by a larger than 50 to 55 nucleotide distance between a termination codon and a downstream exon–exon junction (14Maquat, L.E. (2000). In Translational Control of Gene Expression, N. Sonenberg, J.W.B. Hershey, and M.B. Mathews, eds. (Cold Spring Harbor Press: Cold Spring Habor, NY), pp. 849–868.Google Scholar, and references therein). The sensitivity of NMD to inhibitors of translation, regardless of the cellular site, has led to the proposal that nucleus-associated NMD reflects assessment of the translational reading frame either during the export of nuclear mRNA to the cytoplasm by cytoplasmic ribosomes or prior to export by a mechanism that must somehow be related to cytoplasmic translation. The dependence of NMD on the distance between the termination codon and a downstream exon–exon junction has been explained by a splicing-dependent alteration to mRNP that elicits QC by interacting with components of the translation termination complex. It is reasonable to think that such an alteration could be or could depend on one or more of the splicing-dependent proteins that bind 20–24 nt upstream of exon–exon junctions, as discussed above. Currently, the most likely alteration to mRNP that functions directly in NMD in human cells is the association of one of the recently described human Upf3 proteins (Figure 3). These proteins shuttle between nuclei and the cytoplasm, are far more abundant in the nucleus than the cytoplasm, and are orthologous to a protein required for NMD in S. cerevisiae (13Lykke-Andersen J. Shu M.D. Steitz J.A. Cell. 2000; 103: 1121-1131Abstract Full Text Full Text PDF PubMed Scopus (430) Google Scholar, 18Serin G. Gersappe A. Black J.D. Aronoff R. Maquat L.E. Mol. Cell. Biol. 2001; 21: 209-223Crossref PubMed Scopus (192) Google Scholar). In support of this idea, each human Upf3 protein associates selectively with spliced mRNA in vivo, and the tethering of human Upf3 protein (or either human Upf1 or Upf2 protein, which interact with human Upf3 protein) more than 50 to 55 nucleotides downstream of a normal termination codon can functionally replace the splicing-dependent alteration required for NMD (13Lykke-Andersen J. Shu M.D. Steitz J.A. Cell. 2000; 103: 1121-1131Abstract Full Text Full Text PDF PubMed Scopus (430) Google Scholar). When both the coding and noncoding strands of an RNA are produced simultaneously in the cell, the potential exists for the formation of double-stranded (ds) RNA. It is possible that many complementary RNAs are expressed within the nucleus, either by design (as exemplified by regulatory antisense transcripts) or by transcriptional read-through (as exemplified by the extension of transcription from one gene into the antisense strand of an adjacent gene). The most likely fate of such dsRNA is QC by RNA editing. Nuclear dsRNA is a substrate for members of the class of enzymes known as adenosine deaminases that act on RNA (ADARs; see e.g., 1Bass B.L. Trends Biochem. Sci. 1997; 22: 157-162Abstract Full Text PDF PubMed Scopus (167) Google Scholar, and references therein), first discovered in Xenopus. ADAR1 is ubiquitous in the animal kingdom. Under most conditions, it is almost exclusively confined to the nucleus. ADARs catalyze the conversion of adenosine (A) to inosine (I) within dsRNA. The resulting RNA contains I-U (uridine) base pairs that make the RNA duplex unstable, and may lead to partial or complete unwinding. ADAR editing of dsRNA is sensitive to duplex length. Duplexes of less than 15 base pairs are not edited in vitro, and maximal editing activity is observed with dsRNAs of about 100 base pairs. During polyoma virus infection, nuclear dsRNA is extensively modified, presumably by ADAR. Roughly 50% of the A residues in duplex regions are converted to I residues. Hyperedited molecules can be polyadenylated and are relatively stable, but they fail to be exported to the cytoplasm (Figure 4; see, e.g., 4Carmichael G.G. Kumar M. Zhang Z. Blood Cells. Mol. Dis. 2000; 26: 57-58Google Scholar, and references therein). The nuclear retention and eventual nuclear degradation of hyperedited RNAs provide a critical means for QC, since these RNAs, if exported and translated, would produce highly mutated proteins. It is tempting to speculate that retention may be mediated by the ≈60 kDa nuclear protein recently shown to bind extensively edited RNA in a sequence-independent, but highly cooperative manner (4Carmichael G.G. Kumar M. Zhang Z. Blood Cells. Mol. Dis. 2000; 26: 57-58Google Scholar, and references therein). In contrast to long duplexes, short RNA duplexes appear to be substrates for specific editing events that do not result in nuclear retention and, considering that I is recognized as guanosine (G) by the cellular translation machinery, produce specifically altered proteins (Figure 4; see, e.g., 1Bass B.L. Trends Biochem. Sci. 1997; 22: 157-162Abstract Full Text PDF PubMed Scopus (167) Google Scholar, and references therein). In higher eukaryotes, the QC pathway for cytoplasmic dsRNA is primarily an antiviral defense pathway that is quite distinct from the QC pathway for nuclear dsRNA (Figure 5). Uninfected mammalian cells rarely express dsRNA within the cytoplasm, most likely because of the ensuing dramatic effects on RNA levels, inhibition of protein synthesis, and, if prolonged, cell death (see, e.g., 11Kumar M. Carmichael G.G. Microbiol. Mol. Biol. Rev. 1998; 62: 1415-1434Crossref PubMed Google Scholar, and references therein). Most virus infections or other means of generating cytoplasmic dsRNA induce type I interferons (IFNs), including IFN-α and IFN-β. IFNs are multifunctional cytokines that modulate host immunological functions and can inhibit tumor cell growth and virus multiplication. A central player in cytoplasmic dsRNA activity is the dsRNA-activated protein kinase, PKR. Cells normally contain basal levels of PKR, but in an unphosphorylated and inactive form. Cytoplasmic dsRNA generates autophosphorylated PKR that can phosphorylate a number of substrates, including eIF2α, which inhibits protein synthesis. Cytoplasmic dsRNA and IFNs also activate the 2′, 5′-adenylate synthase (AS)/RNase L pathway so that 2′,5′-AS polymerizes ATP and other nucleotides in 2′,5′ linkages (Figure 5; 11Kumar M. Carmichael G.G. Microbiol. Mol. Biol. Rev. 1998; 62: 1415-1434Crossref PubMed Google Scholar, and references therein). These 2′,5′ oligoadenylates then activate the ribonuclease RNase L. RNase L can cleave both cellular and viral RNAs, although its primary function in vivo remains to be determined. As far as is known, dsRNA within the nucleus does not trigger the IFN, PKR or 2′-5′-AS pathways. In cells where the above pathways are inactive, or in lower eukaryotes, a separate pathway, RNA interference (RNAi), might provide the primary QC mechanism to eliminate cytoplasmic dsRNAs. In RNAi, dsRNA silences gene expression through the specific degradation of its cognate mRNA (see, e.g., 2Bass B.L. Cell. 2000; 101: 235-238Abstract Full Text Full Text PDF PubMed Scopus (331) Google Scholar, and references therein). Although RNAi has been reported in mouse embryos (see, e.g., 19Svoboda P. Stein P. Hayashi H. Schultz R.M. Development. 2000; 127: 4147-4156PubMed Google Scholar), these cells have not been reported to contain active PKR, and there is no evidence that RNAi ever works in mammalian cells where the more common PKR pathway is active. We speculate that in most mammalian cells the PKR pathway might be dominant over the RNAi pathway, or the RNAi pathway might be lacking altogether. While individual nuclear and cytoplasmic reactions required for the formation of functional mRNA can be carried out in isolation in vitro, it has become increasingly clear that many of the steps along the path from gene to protein are, in vivo, interdependent in a way that provides important mechanisms for the QC of mRNA function. In fact, nothing less would be expected of an efficiently operating assembly line, which should discard defective products rather than proceed to process them. Owing to space and reference constraints, this minireview describes neither the bulk of data demonstrating the interdependence of reactions required for mRNA biosynthesis, nor the finer details of these reactions. Future work will no doubt reveal the complete network of integrated events that ensures mRNA QC and yet-to-be-defined molecular constituents of this network.