mRNA turnover plays a key role in the control of gene expression. Recent work has shown that proteins involved in mRNA turnover are located in multicomponent complexes which are tightly regulated. The control of mRNA stability is also intimately linked with translational processes. This article reviews the pathways and enzymes that control mRNA turnover in eukaryotic cells and discusses their mechanisms of control.
The development of a single cell into a living organism depends upon an array of proteins being present at the correct time and in the correct location. These proteins are instrumental in controlling the growth, division and movement of cells to form a three-dimensional embryo. The first step in the control of gene expression is the initiation of transcription which has been well studied. However, it is now clear that each step of the post-transcriptional pathway is also vitally important in determining the final levels of gene product. For example, in the fruitfly, Drosophila melanogaster, posttranscriptional steps such as RNA localization and suppression of translation are known to be crucial in embryonic development [1,2]. Until recently, it has generally been assumed that mRNA stability plays a passive, rather than active, role in the post-transcriptional pathway. However, recent work by my group and others has shown that mRNA stability is highly regulated, varies during development or in response to environmental factors and that specific RNAs can be targeted for degradation [1,3,4,5,6–7].
The importance of mRNA stability in developmental processes is revealed by the phenotypes of animals or plants where the expression of genes involved in controlling RNA stability has been disrupted. For example, mice deficient for TTP (tristetraprolin), an RNA-binding protein that regulates the stability of mRNAs such as GM-CSF (granulocyte/macrophage colony-stimulating factor) and TNF (tumour necrosis factor), develop a systemic inflammatory syndrome with autoimmunity and myeloid hyperplasia (bone marrow overgrowth) . In Caenorhabditis elegans, adults carrying mutations in dicer (dcr-1), a double-stranded endoribonuclease which plays a key role in the first steps of the RNAi (RNA interference) pathway, are sterile and have defects in developmental timing [9,10]. Finally, Arabidopsis mutant for DST1, which regulates the stability of RNAs such as CCL and SEN1, have defects in their circadian rhythm . These results show that ribonucleases and associated factors can be specifically regulated and can target key RNAs that are critical in these developmental and cellular processes.
The control of mRNA stability is intimately linked with the control of translation. This is best illustrated in the fruitfly Drosophila, where crucial protein gradients are often dependent on translational suppression of particular target RNAs, such as nanos and hunchback, followed by their degradation . The mRNAs being translated are now known to be held in a ‘closed loop’ conformation, with the 5′ cap being held in close proximity to the polyadenylated tail through a series of connecting proteins [eIF4E (eukaryotic initiation factor 4E), eIF4G and PABP (polyadenylated-binding protein)] [12,13–14]. Therefore mRNAs involved in translation would appear to be protected from exonucleolytic degradation. The proteins that specifically repress translation often appear to act by disrupting the RNP complex holding the 5′ and 3′ ends of the RNA together . Subsequent degradation of these mRNAs is likely to require interaction of the translation apparatus with mRNA-degradation machinery. Therefore investigating ribonucleases and ribonuclease-associated factors is not only essential for our understanding of the control of mRNA degradation, but is also likely to provide novel insights into the mechanisms of translational repression.
Enzymes and proteins involved in mRNA degradation
In the yeast Saccharomyces cerevisiae, the critical exoribonuclease in the 5′→3′ degradation pathway is Xrn1p [16,17–18]. This enzyme is extremely well conserved in all eukaryotes, with homologues in all eukaryotes, including Drosophila (pacman) and C. elegans (xrn-1) [5,7]. Xrn1p is a processive exonuclease hydrolysing RNA from the 5′ end releasing mononucleotides. The enzyme is not only involved in the normal decay of mRNA, but is also the major pathway in nonsense-mediated decay  and in RNAi . Eukaryotic cells also contain a related nuclease, Rat1p, which is localized in the nucleus and nucleolus, and is involved in the degradation of nuclear RNAs and in rRNA processing .
The critical cytoplasmic exoribonuclease in the 3′→5′ pathway is the exosome. This is a large protein complex containing multiple 3′→5′ exonucleases that also functions in a variety of nuclear RNA-processing reactions [22,23–24]. Structural analysis shows that the nine core exosome subunits form a ring structure which is likely to encircle the RNA. Most of the individual exosome subunits are highly conserved throughout eukaryotes and are also highly similar to bacterial 3′→5′ exoribonucleases. The reason for multiple exonucleases being arranged around a central cavity may be that they can be co-ordinately regulated and/or that particular exonucleases may have specificity for particular RNA sequences/structures, so that this ‘Swiss Army knife’ enzyme complex can rapidly degrade even the most difficult RNAs [21,23,24]. The activity of the exosome is dependent on the helicase Ski2 and associated proteins Ski3p and Ski8p . It is thought that Ski2p uses ATP hydrolysis to unwind RNA structures before digestion or to promote entry of the RNA into the central cavity of the exosome. The Ski2p–Ski3p–Ski8p complex seems to be recruited to the exosome through interactions with the cytoplasmic specific subunit, Ski7p [21,23,24].
For the processive exoribonuclease Xrn1p or the exosome to access the body of the mRNA, it must first be de-adenylated, decapped or endonucleolytically cleaved (Figure 1). De-adenylation occurs via a variety of de-adenylases, including the adenylase complex containing the nucleases Ccr4p and Pop2p/Caf1p along with the associated proteins Not1p–Not5p, Caf4, Caf16, Caf40 and Caf130p. Ccr4 and Pop2p/Caf1 are conserved among eukaryotes and mutations in their catalytic residues abolish de-adenylase activity [4,26]. Another adenylase complex consists of Pan2p and Pan3p, which in yeast, and presumably in other eukaryotes, is involved in shortening initially long polyadenylated tails to 55–75 nucleotides [4,27]. Finally, PARN (polyadenylated ribonuclease) has the major de-adenylase activity in mammalian cells, but is absent in D. melanogaster and S. cerevisiae [4,28].
Pathways of mRNA degradation in eukaryotes
Another way to provide access for exonucleases is to decap the RNA. This is achieved by the decapping proteins Dcp1p and Dcp2p which exist as a complex. The mechanism of action of the decapping complex is not fully understood, but it appears that Dcp2p cleaves the RNA and that this cleavage is stimulated by Dcp1p. Dcp2p contains a Nudix motif, which is found in a class of pyrophosphatases to yield the products m7GDP and a 5′-monophosphate mRNA. The decapping complex prefers substrates ≥25 nucleotides, which may prevent it from binding to mRNAs on which the translation initiation complex has already assembled .
Finally, a third way of accessing the mRNA is by endonucleolytic cleavage. Two examples of endonuclease cleavage providing access to exoribonucleases are nonsense-mediated decay and RNAi. In nonsense-mediated decay in Drosophila, mRNAs containing premature stop codons are endonucleolytically cleaved in the vicinity of the stop codon, by a ribosome-associated endonuclease or by the ribosome itself. The resulting 5′ fragment is degraded from the 3′ end by the exosome, whereas the 3′ fragment is degraded from the 5′ end by Xrn1p/Pacman . Using RNAi in Drosophila, long double-stranded RNAs are first processed by the double-stranded endonuclease Dicer into 21–22 nt siRNAs (small interfering RNAs). The siRNAs are incorporated into a multicomponent RISC (RNA-induced silencing complex) that cleaves target mRNAs at a site complementary to the siRNAs. Following cleavage, the 5′ fragment is degraded by the exosome, and the 3′ fragment is degraded by XRN1/Pacman .
Localization of ribonucleases in cytoplasmic particles
Recent work has shown that Xrn1p forms a multicomponent complex with the decapping proteins Dcp1 and Dcp2, the Lsm proteins (which are likely to form a heptameric ring encircling the RNA), the mRNA-degradation factor Pat1p and the DEAD (Asp-Glu-Ala-Asp)-box-containing helicase protein Dhh1p [30,31]. These proteins are located in cytoplasmic processing bodies known as P-bodies . P-bodies have also been observed in human, mouse cells and Drosophila ([32,33,34–35], and M.V. Zabolotskaya and S.F. Newbury, unpublished work]. In both yeast and human cells, it has been shown that these P-bodies are the sites where mRNAs are specifically decapped and degraded (Figure 2).
Transport of mRNAs from polysomes to P-bodies
P-bodies are dynamic structures which can change in size and number under different conditions. In yeast and human cells, the P-bodies have been shown to be affected dynamically by the availability of untranslated mRNA. For example, if the cells are treated with translation elongation inhibitors, which trap mRNAs in polysomes, P-bodies disappear within 5 min. In contrast, when mRNAs are driven off polysomes by conditions that decrease translation initiation, such as glucose deprivation, P-bodies rapidly increase in number and size [21,29]. In Drosophila and human cells, the de-adenylase Ccr4 also accumulates in P-bodies, suggesting a link between de-adenylation and decapping in these particles [36,37]. These results are consistent with the polysome pool and the P-body pool of RNAs being spatially distinct.
Relationship between mRNA degradation and translational repression of mRNAs
A fundamental aspect of control of mRNA degradation is the interface between mRNA translation in polysomes and mRNA degradation in P-bodies (Figure 2). What controls the transit of mRNAs between these particles? Recent exciting results have shown that reporter miRNAs (micro-RNAs) that are targeted for translational repression by mRNAs become concentrated in P-bodies in a miRNA-dependent manner . Argonaute proteins which are a core component of the RISC are found in P-bodies, suggesting that Argonaute binds to translationally repressed RNAs and delivers them to P-bodies. Alternatively, Argonaute, bound to the translationally repressed RNA, may nucleate the assembly of a P-body, which would be consistent with the dynamic nature of these structures . Localization of the translationally repressed RNAs to P-bodies would prevent protein synthesis, as the translation machinery is excluded from this structure. Since Argonaute binds to the Dcp1–Dcp2 complex, it might then promote decapping and subsequent degradation of the mRNA targets . These data therefore suggest that the sequestering of targeted mRNAs in P-bodies is crucial for translational repression.
The concept of de-adenylated mRNAs being stored in a repressed state is common in many biological contexts including early development and mRNA transport . For example, during Drosophila oogenesis, oskar mRNA is translationally repressed and is transported in particles to the posterior end of the oocyte before it is translationally activated. In axons and dendrites of human neurons, repressed RNAs are transported in large particles from the nucleus to the tip of the axon or dendrite where they are then translated . In some cases, similar proteins are involved in both translational repression and mRNA degradation. For example, the Drosophila homologue of the helicase Dhh1p, Me31b, which, in yeast, is required for efficient decapping , is necessary for the translational repression of bicoid and oskar mRNAs and is located in cytoplasmic particles in egg chambers . The Xenopus homologue of Dhh1p, Xp54, appears to repress mRNA translation directly and also is a major constituent of maternal storage particles . For C. elegans, the Dhh1p homologue cgh-1 is expressed specifically in the germline and early embryo, and is localized to P-granules and other putative mRNA–protein particles . Therefore transport of RNAs to P-bodies may not be one-way; as with other RNA granules, the mRNAs may re-emerge to be activated and translated.
Specificity of ribonucleases for RNA targets
A number of lines of evidence indicate that specific mRNAs can not only be targeted for degradation, but that the timing of this degradation can be regulated. This is particularly evident during early development, when maternally provided mRNAs are specifically degraded at a particular point in development . Further evidence that mRNA degradation can be specific comes from phenotypic analysis of yeast, C. elegans or Drosophila, where the expression of ribonucleases has been down-regulated. Mutations in XRN1 in S. cerevisiae led to a number of specific phenotypes, including larger cell size, increased doubling times, defective sporulation and sensitivity to the microtubule-depolymerizing drug benomyl [17,18]. In C. elegans, we have used RNAi to down-regulate the XRN1 homologue, xrn-1. The xrn-1(RNAi) embryos develop normally until the ventral enclosure stage, but then die because of failure of the epithelium to close along the ventral mid-line (Figure 3) . In Drosophila, hypomorphic mutations in the XRN1 homologue pacman, result in gastrulation defects and defects in thorax closure, where the epithelial cells fail to join along the dorsal side of the adult thorax (D.P. Grima, K.C. Wan, C. Roberts, J. Seago, C. Browne, Y. Okada and S.F. Newbury, unpublished work). These data show that only a subset of mRNAs must be degraded in order to produce these specific phenotypes. Therefore a mechanism to control the access of mRNA-degradation factors to particular mRNAs must be in place.
Failure of ventral enclosure in C. elegans embryos where xrn-1 has been down-regulated by RNAi
Another way that mRNA degradation might be controlled would be to modulate the expression levels of particular ribonucleases. Although the mechanisms of this type of regulation are not understood, it is clear that many ribonucleases and associated factors are differentially regulated during development. All of the RNA-degradation factors we have studied in Drosophila, including pacman/XRN1, twister/SKI2 and tazman/DIS3/RRP44, all vary in expression throughout development, both at the RNA and protein level [3,6,7]. Therefore, in metazoans, modulation of the expression of ribonucleases and/or associated factors may play an important role in the control of mRNA stability during development.
How can particular mRNAs be targeted? There are very few examples where the pathway that promotes targeted degradation of a metazoan mRNA has been worked out. However, some examples where progress has been made are given here. In the case of AREs (AU-rich elements) found in mRNAs such as TNF-α and interleukin-2, the specific ARE-binding protein TTP binds to the 3′-UTR (untranslated region) and promotes de-adenylation, probably by the deadenylase PARN. TTP also appears to interact physically with the 3′→5′ exosome and recruit it to the de-adenylated mRNA . As well as AREs, other sequences/structures within the 3′-UTR can also promote de-adenylation and repression of translation of target RNAs. For example, sequences within the 3′-UTR of hunchback RNA in Drosophila bind Pumilio and Nanos as well as another protein (brain tumour) and promote de-adenylation [43,44]. Further steps in the degradation/translation repression pathway are not known, but it would be interesting to find out whether the repressed hunchback RNA is degraded in P-bodies. RNAi- or miRNA-induced degradation pathways are also likely to be involved in destroying specific target RNAs in development as mutations in dcr-1/dicer affect oogenesis in C. elegans and Drosophila. One of the few well-characterized examples is the bantam miRNA which binds to the 3′-UTR of the pro-apoptotic mRNA hid (head involution defective) target RNA and promotes cell proliferation . Although it is thought that bantam acts by repressing translation of its target RNA, it is possible that the target RNA is also degraded in P-bodies. Understanding the mechanisms whereby specific mRNAs can be targeted is likely to lead to significant advances in the field and will also shed light on the control of translational repression.
The mechanisms of mRNA stability have proved to be highly regulated, involving multiprotein complexes and a plethora of specific RNA-binding proteins. The mRNA-degradation network is also intimately linked with the control of translation and other cellular processes such as nonsense-mediated decay and RNAi. Future work is likely to elucidate the links between these processes and mRNA stability, and will also shed light on the mechanisms of degradation of specific mRNAs. As highlighted by the widespread use of RNAi, this ancient and conserved method of gene regulation may also provide us with further tools to manipulate gene expression in eukaryotes.
Translation UK: Focused Meeting and Satellite to BioScience2005, held at Western Infirmary, Glasgow, U.K., 21–23 July 2005. Organized and Edited by M. Bushell (Nottingham, U.K.), S. Newbury (Newcastle upon Tyne, U.K.), G. Pavitt (Manchester, U.K.) and A. Willis (Nottingham, U.K.).
This work was supported by the BBSRC (Biotechnology and Biological Sciences Research Council).