RNA-quality control by the exosome

Key Points

  • The processing and degradation of RNA sequences is ubiquitous, and these activities can be separated into different classes on the basis of their mechanism and function. The clearest functional differences exist between RNA maturation that will generate a usable RNA from its precursor and RNA degradation that will destroy the RNA completely.

  • There is mounting evidence that all RNA-maturation pathways in eukaryotes are continuously monitored by surveillance systems. A key component of the RNA-surveillance machinery is the exosome complex of 3′→5′ exonucleases. On different substrates this complex is responsible for either total RNA degradation or accurate RNA processing, implying a precise distinction between different classes of substrate.

  • Despite the presence of multiple catalytic sites, the purified yeast exosome is almost entirely inactive. However, several activating cofactors have recently been identified.

  • The role of polyadenylation in eukaryotic cells seems different on either side of the nuclear envelope. The long poly(A) tails on cytoplasmic mRNAs promote stability and translation. By contrast, defective nuclear RNAs are identified and 'tagged' with short oligo(A) tails by the TRAMP polyadenylation complex prior to degradation by the exosome. Such tagging of RNAs by polyadenylation could be conceptually similar to the role of polyubiquitylation in targeting proteins for degradation by the proteasome complex.

  • Key features of RNA degradation have been conserved throughout evolution. Recent structural analyses indicate that the eukaryotic exosome resembles a complex between bacterial PNPase and RNase II, the most active degradative exonucleases in Escherichia coli lysates. Moreover, the role of eukaryotic poly(A) polymerases in targeting nuclear RNAs for degradation is strikingly similar to the role of oligoadenylation in stimulating RNA degradation in bacteria.


The exosome complex of 3′→5′ exonucleases is an important component of the RNA-processing machinery in eukaryotes. This complex functions in the accurate processing of nuclear RNA precursors and in the degradation of RNAs in both the nucleus and the cytoplasm. However, it has been unclear how different classes of substrate are distinguished from one another. Recent studies now provide insights into the regulation and structure of the exosome, and they reveal striking similarities between the process of RNA degradation in bacteria and eukaryotes.


The quantity of RNA that is produced by a cell is quite remarkable. A fast-growing yeast cell produces some 2,000 molecules of ribosomal RNA precursor per minute1 ? comprising some 14 million nucleotides ? as well as mRNAs from thousands of genes and an abundance of small functional RNAs, including transfer RNAs, small nucleolar RNAs (snoRNAs) and small nuclear RNAs (snRNAs)1. Compounding the metabolic cost of this synthesis, almost all RNA species undergo several post-transcriptional processing reactions to produce functional RNA molecules. Moreover, most RNAs function as RNA?protein complexes that are usually generated by elaborate assembly and maturation pathways. The reason for the ubiquity of RNA processing is unclear, but it is plausible that it exists to allow quality-control systems to distinguish between the mature ribonucleoprotein particles (RNPs) and their precursors. There is mounting evidence that the processes that convert the RNA primary transcript to the mature RNA?protein complex are continuously monitored by surveillance systems.

A key component of the RNA-surveillance machinery is the exosome complex of 3′ →5′ exonucleases. This complex degrades many types of RNA that have been targeted by surveillance activities in both the nucleus and cytoplasm, and it is also responsible for the precise trimming, in a 3′→5′ direction, of the 3′ ends of nuclear precursors to several RNA species. This dual role of the exosome in total RNA degradation and accurate RNA processing implies a precise distinction between different classes of substrate. Many RNAs are also subject to degradation from the 5′ terminus by the 5′ exonucleases Rat1 (XRN2 in humans) and Xrn1. However, due to space limitations, these pathways will not be discussed in this review and the reader is referred to Ref. 2.

Here we will review recent advances in our understanding of RNA degradation and maturation (see also Box 1). In particular, new mechanistic insights have been gained from analyses of the structure of the exosome complex of 3′→5′ ribonucleases. Important advances have also been made in understanding the activation of the exosome during the surveillance of nuclear RNA maturation in eukaryotes. These studies have also uncovered a largely unanticipated role for RNA polyadenylation in stimulating nuclear RNA degradation, which is in direct contrast to the role of poly(A) tails in stabilizing cytoplasmic mRNAs.

Functions of the exosome

Nuclear and cytoplasmic forms of the exosome complex were initially identified in yeast3,4 (see Table 1 for the composition of the yeast exosome complexes). These complexes share ten common components, but they differ in the presence of the GTPase Ski7 in the cytoplasmic complex, and the RNase Rrp6 (ribosomal RNA-processing protein-6) and putative nucleic-acid-binding protein Rrp47 (also known as Lrp1) in the nuclear complex4,5,6,7,8,9,10.

Table 1 Components of the exosome and exosome-like complexes

The nuclear exosome functions in the 3′ processing of the precursors to stable RNAs, including the 5.8S rRNA component of mature ribosomes and many snoRNAs3,11,12,13,14. The nuclear exosome is responsible for the surveillance and degradation of aberrant nuclear precursors of many types of RNA including pre-mRNAs, pre-tRNAs and pre-rRNAs6,15,16,17,18,19,20,21,22,23. Regulated 3′ degradation by the nuclear exosome is also implicated in the control of expression levels of some mRNAs24,25,26,27 (see Table 2 for a list of substrate RNAs). These exosome substrates therefore include both RNA species that should be matured by the removal of nucleotides to a precisely defined end point, and defective RNAs that will undergo rapid and complete degradation. This dual role implies the existence of specific signals that are associated with different substrate RNAs. These must allow the complex to identify many structurally diverse substrates and to reliably distinguish between the different classes of RNA. The precise identity of these signals remains poorly understood, but they are very likely to be comprised of sequences both in the substrate RNA and in specific associated proteins.

Table 2 Characterized substrates and cofactors for the exosome

The only known substrates for the cytoplasmic exosome are mRNAs. The exosome participates in the 3′ turnover of normal mRNAs8,28, and therefore helps to determine mRNA abundance and, so, protein-synthesis rates. Also, the exosome rapidly degrades mRNAs with structural defects; the nonsense-mediated decay (NMD) pathway degrades mRNAs with premature translation-termination codons29,30,31, whereas the non-stop decay pathway degrades mRNAs that lack a termination codon altogether32,33. In human cells, the exosome is also recruited to, and rapidly degrades, mRNAs that contain specific A+U-rich sequence elements (AREs)34,35,36,37. These are present in many mRNAs that encode proteins for which transient expression is important, including growth factors and proto-oncogenes. Furthermore, the exosome degrades the 5′ fragments of mRNAs that are cleaved in the no-go decay pathway, which targets mRNAs on which translation has stalled38,39. The 5′ fragments of mRNAs that have been cleaved by the RNA interference (RNAi) pathway in Drosophila melanogaster cells are similarly degraded by the exosome40.

Structure and mechanism of the exosome

RNA exonuclease complexes exist in all three domains of life, and emerging data indicate that they share much in common at both a structural and functional level.

PNPase structure. Bacteria, along with many mitochondria and chloroplasts, contain polynucleotide phosphorylase (PNPase) (Fig. 1a,b). This is a homotrimer in which each monomer contains S1 and heterogeneous nuclear (hn)RNP K-homology (KH) RNA-binding domains41 that are important for substrate binding42. In addition, each monomer contains two domains that are related to RNase PH and that are therefore referred to as PH domains. RNase PH is a phosphorolytic exonuclease ? it uses inorganic phosphate to break the RNA backbone, thereby releasing mononucleotide 5′-diphosphate products. This reaction is reversible, and elevated nucleotide diphosphate concentrations lead to polymerization with the release of a phosphate product. It was this activity that led to the naming of PNPase.

Figure 1: Structure of the exosome complex.

The relative size and distribution of conserved domains between the PNPase from the bacterium Streptomyces antibioticus41 (a,b) and the exosome complex from the archaeon Archaeoglobus fulgidus45,46,47 (c,d) are compared. (a) PNPase ribbon model, viewed from the S1 and KH domain face. (b) PNPase ribbon model, side view. (c) Archaeal exosome ribbon model, viewed from the Rrp4 face. (d) Archaeal exosome ribbon model, side view, with the putative locations of Rrp6 and Rrp44 in the yeast exosome indicated. Each PNPase subunit contains a tandem repeat of the RNase PH domain (one PH domain is shown in green and the other in blue), together with S1 (orange) and KH (red) RNA-binding domains at the C terminus. In the archaeal exosome, the RNase PH domains are located on alternating Rrp41 (blue) and Rrp42 (green) subunits, whereas the S1 (orange) and KH (red) domains are present on the Rrp4 subunits. All other regions of both protein complexes are shown in yellow to convey where equivalent domains lie. As previously noted47, the relative positions of the S1 and KH domains seem to be reversed between the two structures; however, these domains were incompletely resolved in the PNPase crystal structure and could be mobile. (d) The diameters of the spheres that represent the eukaryotic-specific exosome components Rrp6 (114 kDa) and Rrp44 (84 kDa) are approximately proportional to their masses relative to the core exosome (285 kDa) of Saccharomyces cerevisiae. Blue arrows indicate the reported two-hybrid interactions of these proteins60, and the sizes of the arrow heads reflect the number and strength of the interactions that are detected. We therefore propose that Rrp6 and Rrp44 interact with each other on the face of the exosome that lies opposite the S1- and KH-domain proteins. This positioning of Rrp44 and Rrp6 might isolate these proteins from substrates that are targeted for degradation by the PH-domain protein homologues. The structures were generated using Pymol (see The PyMOL Molecular Graphics System in Further Information).

In the Escherichia coli PNPase enzyme, only one PH domain is catalytically active in each monomer41. In the PNPase crystal structure, the six PH domains of the trimer have a ring-like organization (Fig. 1a,b) and a central channel is formed at the interface between the three subunits41,43. Perhaps unexpectedly, the access channels to the active sites do not face towards the S1 and KH RNA-binding domains. Instead, they face the opposite way, and they probably require coordinated interactions with the RNA substrate to guide it into the catalytic centre. Studies on E. coli PNPase, however, do indicate that the RNA substrate travels from the RNA-binding domains to the active site of the PH domain via this central pore44. Consistent with this, a single-stranded 3′ RNA end is required for PNPase activity.

Archaeal exosome. Recent structural analyses of the archaeal exosome (Fig. 1c,d)45-47 show that it contains the same domains as PNPase, with a similar overall structure as previously proposed48,49. The difference is that in the archaeal complex some of these domains are present in separate polypeptides. Six proteins with RNase PH domains form a ring that is comprised of three identical Rrp41?Rrp42 heterodimers. In each heterodimer, only Rrp41 is catalytically active, although Rrp42 contributes to the substrate-binding site45. One face of this ring is associated with a protein trimer with RNA-binding domains that is comprised of either Rrp4 or Csl4 (Refs 45?47). The position of this trimer, on the face of the PH-domain ring opposite the active sites of the Rrp41 subunits, is similar to the location of the S1 and KH RNA-binding domains of PNPase. Rrp4 and Csl4 each contain an S1 RNA-binding domain. In addition, Rrp4 contains a KH RNA-binding domain, whereas the archaeal Csl4 protein has a zinc-ribbon RNA-binding domain47. The crystal structures of the archaeal exosome complexes contain homotrimers of either Rrp4 or Csl4, but heterogeneous complexes might form in vivo.

The structure of the archaeal exosome also has similarities to that of the proteasome, as previously indicated50, because the active sites of the Rrp41 components are reached from a barrel-like central cavity to which substrate access is limited. The interface between the three heterodimers of the PH-domain ring generates two main entrances to this cavity. The larger is located on the face of the complex that does not bind the Rrp4 and Csl4 trimers. However, this is not believed to be the normal route for access of RNA to the active sites because of the negative charge distribution around the entrance. Instead, substrate RNAs are likely to enter a channel through the Rrp4- or Csl4-trimer rings, the 'S1 pore', and this process is presumably promoted by the RNA-binding domains of Rrp4 and Csl4. This channel has a width of 15 Å with the Rrp4 trimer and 18 Å with the Csl4 trimer. The RNA can then access the central RNase PH cavity via a narrow channel, some 8?10 Å in width, which excludes regions of secondary structure. Positive charges around this pore could facilitate entry of the RNA45,46,47.

The single-stranded RNA reaches the active site within an Rrp41 subunit via an RNA-binding pocket that is four nucleotides in length46. This pocket coordinates the extreme 3′-terminal nucleotides of the RNA substrate and guides them into the catalytic centre, despite the opposing orientation of the access channels of the active sites relative to the S1 and KH RNA-binding domains. Liberated nucleotides are most probably released through the other side of the access channel, which also connects the active site with the solvent. The distance from the interface of the S1 pore with the RNase PH cavity to the PH-domain active site is 50?60 Å, which corresponds to 7?9 nucleotides. This indicates that the archaeal exosome should be able to degrade RNAs only if they have a single-stranded tail that is greater than this length, and this conclusion has been experimentally verified46.

Predicting the eukaryotic exosome structure. The core structure of the eukaryotic exosome is assumed to resemble closely that of the archaeal complex. However, the composition of the exosome is more complex in all of the eukaryotes that have been examined so far, including trypanosomes, flies, humans and yeast4,35,51,52,53,54 (Table 1). The six RNase PH domains are each present in distinct proteins: eukaryotic Rrp41, Rrp46 and mRNA transporter protein-3 (Mtr3) are more closely related to archaeal Rrp41, whereas eukaryotic Rrp42, Rrp43 and Rrp45 are more closely related to archaeal Rrp42. In the archaeal complex, Rrp42-like subunits are catalytically inactive, and this might also be the case in the eukaryotic exosome.

The eukaryotic exosome also contains a Csl4 homologue and two proteins, Rrp4 and Rrp40, that are homologous to archaeal Rrp4 (Refs 3, 4, 55). In yeast, both Rrp4 and Rrp40 are efficiently precipitated with tagged Csl4 (J.LC. and D.T., unpublished observations), making it likely that each yeast exosome contains all of these proteins. As in the archaeal complex, the RNA-binding domains of these proteins are presumed to be important for substrate binding and for insertion into the S1 pore, which leads to the central cavity and the catalytic sites of the PH-domain homologues. However, this has not been experimentally verified. Recombinant Rrp4 from yeast and Arabidopsis thaliana showed exonuclease activity, digesting substrates into mononucleotides3,55. However, a similar in vitro exonuclease activity has not been reported for Rrp40 or Csl4, and no in vitro activity was detected for recombinant archaeal Rrp4 (Ref. 47).

In addition, the yeast exosome includes Rrp44 (also known as Dis3), which is a member of the RNase II family of processive, hydrolytic exonucleases3. Another exonuclease (Rrp6 in yeast, PM-Scl100 in humans) is present only in the nuclear version of the yeast exosome4,56. This protein is homologous to the RNase T/D family of bacterial exonucleases, and it also shows exonuclease activity in vitro6,57.

The archaeal exosome does not include any homologues of Rrp44 or Rrp6, so neither the archaeal exosome structure nor the structure of PNPase provide any clues as to the location of the eukaryotic subunits. However, several two-hybrid interactions have been reported for yeast and human Rrp44 and Rrp6 (Refs 58?60). The PH-domain protein Rrp43 interacted with both the N-terminal, Pin-C domain of Rrp44 and the C-terminal fragment of Rrp6. The N-terminal fragment of Rrp6 interacted with each of the Rrp41-like subunits of the exosome (Rrp41, Rrp46 and Mtr3), whereas its exonuclease domain, which constitutes the N-terminal half of the protein, interacted with Rrp44. Based on these data, we propose that Rrp44 and Rrp6 interact with each other on the face of the PH-domain ring that lies opposite the Rrp4, Rrp40 and Csl4 subunits and the entry pore to the active sites of the Rrp41-like subunits (Fig. 1d).

Rrp44 can be stripped off the exosome in high salt concentrations, which leaves the remainder of the complex intact9. This is consistent with the model that Rrp44 does not form part of the structural core of the complex. However, the genetic depletion of Rrp44 results in RNA-processing defects that closely resemble mutations in the nine components of the 'core', archaeal-like complex. This indicates that Rrp44 might also have a role in promoting an active conformation in the core complex. Given the large size of the pore in the 'front' face of the central cavity, it is possible that RNAs that enter through the Rrp4?Rrp40?Csl4 ring could traverse the central cavity to reach an Rrp44?Rrp6 complex on the other side of the complex.

By contrast, the depletion or mutation of Rrp6 (including point mutations in the active site) results in phenotypes of activity on many nuclear RNA substrates that are distinct from those of other exosome mutants4,9,11,13,21,23,27,56,61. These findings make it unlikely that substrate binding by Rrp6 involves RNA transit through the core complex, and it also shows that the remainder of the exosome is fully functional in the absence of Rrp6.

Overall, the eukaryotic exosome is thought to resemble a complex between homologues of bacterial PNPase and RNase II, which make up the major RNA-degradation activity in E. coli lysates. Our predicted localization of Rrp44 and Rrp6 places them far from the site of RNA entry to the Rrp4?Rrp40?Csl4 ring and the active sites of the Rrp41-like subunits. This implies that substrates that are targeted for phosphorolytic digestion by the RNase PH-like subunits follow a quite different pathway from RNAs that are targeted to the hydrolytic Rrp44 and Rrp6 subunits (Fig. 1d).

Role of RNA helicases. To access the active sites of the PH-domain ring, RNA substrates must pass through a pore that excludes double-stranded RNA47, and RNA helicases are very likely to contribute to this translocation. RNA helicases have ATP-dependent translocase activity on single-stranded nucleic acids, which could unwind the structure of substrate RNAs and perhaps push the RNA through the pore into the central cavity of the exosome. Consistent with this idea, the DExH-box RNA helicase Mtr4 (also known as Dob1) is required for most nuclear activities of the exosome ? the exceptions being activities that specifically require Rrp6, such as the final trimming of stable RNA precursors11,14,61. A closely related cytoplasmic RNA helicase, Ski2, is similarly required for all known functions of the yeast cytoplasmic exosome28,30,33,40,62. In human cells, another DExH-box RNA helicase, RHAU, is implicated in activated mRNA turnover that is stimulated by AREs34.

These proposed roles of RNA helicases in substrate unwinding and translocation into the central cavity of the exosome are potentially similar to the functions of the ATPases in the regulatory 19S cap complex of the proteasome. These ATPases open up secondary structures in protein substrates and subsequently translocate the substrates into the lumen of the proteasomal core structure.

Activation of the exosome

Despite the abundance of in vivo substrates, initial in vitro analyses indicated that the yeast exosome has low intrinsic exonuclease activity on naked RNA substrates3. It was therefore believed that the eukaryotic exosome generally has little RNase activity. This might be important in protecting the cell against inappropriate RNA degradation ? the activated exosome would be a dangerous beast to have roaming unchecked through the transcriptome. Several different factors and complexes have now been identified that can activate the exosome on defined classes of transcript (Table 2; Fig. 2). Two broad classes of activating cofactors can be distinguished. Some factors bind specific RNA sequences in substrate RNAs, whereas others are likely to recognize and target RNA?protein complexes primarily on the basis of their specific structural features.

Figure 2: Activation of the exosome.

Despite the presence of several catalytic sites, the exosome is almost entirely inactive without the aid of cofactors. A diverse array of substrates and activating cofactors have been identified, and they fit broadly into four categories: surveillance, sequence-dependent degradation, sequence-independent degradation and maturation. 1 | The cytoplasmic exosome is activated by the Ski complex (Ski2?Ski3?Ski8), along with the helicase Ski7, for the degradation of improperly processed RNAs such as those lacking a stop codon (non-stop decay) or containing a premature stop codon (nonsense-mediated decay (NMD)). NMD also requires the Upf1?Upf2?Upf3 surveillance complex, the Upf1 component of which interacts with the exosome via Ski7. 2 | In the nucleus, the TRAMP complex (Trf4?Air1?Mtr4 in the diagram) targets defective precursors to ribosomal RNAs, transfer RNAs, small nuclear RNAs (snRNAs), small nucleolar RNAs (snoRNAs) and other transcripts. 3 | Sequence-dependent recruitment of the cytoplasmic exosome in human cells involves factors including TTP and KSRP (not shown), which directly recognize A+U-rich sequence elements (AREs) and cause mRNA destabilization. 4 | In the nucleus, Nrd1, perhaps acting together with its binding partner Nab3, can recruit the exosome to mRNAs, snoRNAs and probably other RNAs that contain defined sequence motifs. Other exosome substrates lack defined sequence motifs that could specifically function in exosome recruitment. 5 | All mRNAs undergo turnover, a process that is not generally dependent on defined RNA sequences but that involves the Ski complex and the exosome, as well as the decapping and 5′-degradation machinery. 6 | In the nucleus, exosome-dependent RNA-degradation pathways that are unlikely to involve the recognition of simple RNA sequences include the degradation of spacer elements that have been excised from rRNA precursors and the 3′ trimming of pre-rRNAs and snoRNAs. However, these activities require cofactors: the RNA helicase Mtr4 or the putative RNA-binding protein Rrp47. 7S is a 3′-extended precursor to the 5.8S rRNA, which is 3′ matured by the exosome. ETS, external transcribed spacer; M7G, 7-methylguanosine; TMG, trimethylguanosine.

Sequence-specific cofactors. Sequence-specific cofactors include the nuclear RNA-binding protein Nrd1, which is implicated in promoting RNA degradation by the exosome both in vitro and in vivo63,64. The precursors of many RNA species contain Nrd1-binding sites63,64,65,66, which potentially function as targets for exosome-mediated degradation. It might be that, by default, all RNAs with binding sites for Nrd1 are targeted for complete degradation by the exosome. Normal processing (that is, 3′ cleavage and polyadenylation of mRNAs or 3′ cleavage of the precursors to snoRNAs and snRNAs) would remove the Nrd1-binding site and protect the processed RNAs against degradation by the exosome. This would potentially constitute a quality-control system that degrades RNAs that fail to undergo efficient RNA processing.

In human cells, ARE-mediated degradation is an important mRNA-turnover pathway that involves the recruitment of the exosome, either by direct association with ARE-binding proteins (ARE-BPs) such as AUF1/hnRNPD, KSRP and TTP35,37 or by association of ARE-BPs with the exosome-associated RNA helicase RHAU34. ARE-mediated decay also occurs in Saccharomyces cerevisiae67. This activity requires a TTP homologue, but the role of the exosome has not yet been reported.

Non-sequence-specific cofactors. Other cofactors for the exosome act on many different substrates that lack clear similarities in RNA structure or sequence. This indicates that more complex features of the RNA?protein structure are being recognized and targeted. The first general exosome cofactor to be identified was the yeast DExH-box RNA helicase Mtr4, which is required for the activities of the nuclear exosome in both RNA maturation and degradation14.

Mtr4 is also present in the Trf?Air?Mtr4 polyadenylation (TRAMP) complexes, which polyadenylate target RNAs and activate the exosome both in vitro and in vivo17,18,19,68 (J.H. and D.T., unpublished observations). Two distinct forms of the TRAMP complexes, TRAMP4 and TRAMP5, have been identified in yeast. In addition to Mtr4, each complex contains a poly(A) polymerase, which is either Trf4 (in TRAMP4) or Trf5 (in TRAMP5). Both TRAMP complexes also contain a zinc-knuckle protein ? either Air1 or Air2, which seem to be functionally redundant17,18,19,68. The TRAMP complexes are thought to bind RNA substrates through the zinc-knuckle, putative RNA-binding domains that are present in Air1 and Air2. Mtr4 might then actively recruit the exosome to TRAMP substrates, as its depletion leads to RNA hyperadenylation in vivo, which indicates that polyadenylation has been uncoupled from degradation68. The TRAMP complexes are predicted to have a dual function: poly(A)-tail addition makes the RNA a better substrate for the exosome, whereas its interactions with the exosome make the exosome a better enzyme. Together, these activities probably have an important role in allowing the exosome to degrade through long, highly structured RNAs and RNA?protein complexes rapidly and processively.

The TRAMP4 and TRAMP5 complexes show clear substrate preferences in vivo68,69, but the basis of this selectivity is unclear. The TRAMP complexes seem to function specifically in nuclear RNA-surveillance pathways, as no roles in exosome-mediated RNA maturation have yet been shown. Nuclear surveillance of a defective pre-tRNA was reduced by mutations in the Mtr4 and Trf4 components of the TRAMP complex and by mutations in Rrp44 or Rrp6 (Refs 19, 20, 70). This indicates that the TRAMP complex specifically directs this substrate to the face of the exosome that associates with Rrp6 and Rrp44 (Fig. 1d).

Precursors to several small stable RNAs that contain binding sites for Nrd1 are also known to be stabilized by exosome and/or TRAMP mutations. In particular, extended forms of the U14 snoRNA are stabilized in strains that are defective in either Nrd1 or the TRAMP4 complex, as well as exosome mutants, which hints at functional interactions between these pathways18,63,64. Consistent with this idea, the TRAMP components Trf4 and Air2 could be co-precipitated with Nrd1, as could several exosome components64. Moreover, the Nrd1-associated, RNA-binding protein Nab3 co-precipitated with Mtr4 (Refs 71, 72), which indicates the existence of additional contacts.

Mtr4 is much more abundant than the TRAMP complex and it is required, together with the nuclear exosome, for RNA-processing and -degradation activities that seem to be largely TRAMP independent (for example, pre-rRNA processing)14,18. In these pathways, we predict that Mtr4 functions to recruit the exosome in association with other proteins, including components of the pre-ribosome. Candidate pre-ribosome-associated exosome-recruitment factors include human MPP6 (Ref. 73) and yeast Nop53 (Refs 74, 75).

A further activation pathway for the nuclear exosome involves the putative RNA-binding protein Rrp47, which seems to function specifically together with Rrp6 in the maturation of stable RNAs ? snoRNAs and rRNA. Overall, the data support the hypothesis that the TRAMP complex, Nrd1 and Rrp47 each target nuclear RNA substrates directly to Rrp6 and/or Rrp44, rather than to the PH-domain proteins.

Cytoplasmic cofactors. All known cytoplasmic activities of the exosome require the same set of cofactors: the cytoplasmic-exosome-specific component Ski7 (a GTPase with homology to translation factors) and the Ski2, Ski3 and Ski8 proteins, which form a complex7,8,62,76,77. Ski2 is a DExH-box putative RNA helicase that is homologous to Mtr4; however, Ski3 and Ski8 show no detectable homology to Trf4, Trf5, Air1 or Air2. As well as participating in mRNA 3′ turnover, the Ski2?Ski3?Ski8 complex and the Ski7?exosome complex function in the NMD and non-stop decay mRNA-surveillance pathways29,30,33,77,78. The exosome components that are involved in the degradation of mRNAs have not been established, but point mutations in Rrp41, Csl4 and Rrp4 were each shown to interfere with mRNA turnover8,28,76, which indicates that these RNAs might enter the central cavity from the Rrp4?Rrp40?Csl4 face. Other substrates for the cytoplasmic exosome probably include the transcripts from the M-satellite of the L-A virus, carried by most laboratory strains of yeast, which is translated to produce the so-called 'killer' protein79. These transcripts differ from normal mRNAs in that they lack both 5′ cap structures and poly(A) tails. Screens for the overexpression of killer protein generated the super-killer (Ski) mutants, which included many RNA-turnover factors such as the 5′ exonuclease Xrn1 (also known as Ski1) as well as the Ski proteins and the exosome components Rrp41 (also known as Ski6) and Csl4 (also known as Ski4)8,28,76,80,81.

Little is known about how any of these cofactors actually activate the exosome, but the data indicate that the physical recruitment of the exosome to specific RNA substrates has a key role in stimulating RNA degradation in both the nucleus and the cytoplasm.

Nuclear polyadenylation and degradation

In E. coli cell extracts, the 3′→5′ exonucleases PNPase and RNase II account for most of the RNA-turnover activity. These enzymes have low intrinsic activity on substrates with stem-loop structures at their 3′ ends, such as mRNAs that result from intrinsic transcription terminators or the repeated extragenic palindrome (REP) stabilizer elements, both of which introduce very stable secondary structures into the RNA82,83. However, the addition of a small number of A residues greatly enhances the degradation of these RNAs and the oligo(A) tail is believed to create a 'landing pad' from which the nucleases can initiate degradation.

This function of bacterial oligo(A) tails in promoting RNA degradation is in marked contrast to the role of the polyadenylation of eukaryotic mRNAs (Fig. 3a). The poly(A) tail of mRNAs stimulates export to the cytoplasm, where it promotes mRNA stability and translation. In yeast, the presence of an intact poly(A) tail is a mark of a successfully processed transcript. Therefore, poly(A) tails are closely monitored by the exosome, which can apparently distinguish normal transcripts from hypo- or hyperadenylated transcripts, as part of its role in the nuclear pre-mRNA-surveillance process6,22,61.

Figure 3: Roles of polyadenylation in eukaryotic cells.

In eukaryotes, there is an apparent dichotomy in the functions of RNA poly(A) tails on either side of the nuclear envelope. a | Normal pre-mRNA molecules, which are transcribed by RNA polymerase II (Pol II), undergo co-transcriptional cleavage at defined sites in the 3′ untranslated region (3′ UTR) and they are then polyadenylated by the canonical poly(A) polymerase Pap1. These poly(A) tails are usually covered with nuclear poly(A)-binding proteins before the poly(A) polymerase releases the RNA, which might protect it from degradation. The pre-mRNAs are spliced and exported to the cytoplasm where their poly(A) tails and associated proteins confer stability and signal its readiness for translation. The rate of cytoplasmic mRNA deadenylation varies greatly for different RNA species and is an important factor in determining mRNA lifetimes. In yeast, the major mRNA-deadenylation activity is provided by the Ccr4?Pop2 dimer. Following deadenylation, the mRNA can be decapped and 5′ degraded by Xrn1 or 3′ degraded by the exosome. During early metazoan development, deadenylated mRNAs are frequently stable but translationally inert; translation is only initiated after polyadenylation by cytoplasmic poly(A) polymerases. b,c | In the nucleus, RNA polymerases I (Pol I) and III (Pol III) generate functional RNAs, such as ribosomal RNA and transfer RNA, respectively, to which a poly(A) tail is added by the TRAMP complexes only after a defect in maturation is detected. In wild-type cells, the tails are probably normally very short and the adenylated RNA molecules are rapidly degraded by the exosome. In consequence, the population of these adenylated molecules in a wild-type cell decreases rapidly. RNA polymerase II also generates a class of cryptic unstable transcripts from intergenic regions, which are rapidly degraded by the TRAMP?exosome system (see part a). M7G, 7-methylguanosine.

However, analyses in yeast revealed that snRNAs, snoRNAs, pre-rRNAs and pre-tRNAs could all undergo polyadenylation, and this was apparently not linked to RNA stability but was associated with RNA degradation by the exosome11,13,20,84(Fig. 3b,c). Strains that lacked the nuclear exosome component Rrp6 accumulated polyadenylated precursors of many non-coding RNAs. It therefore seemed likely that these polyadenylated RNAs were identified and 'tagged' as aberrant by a surveillance system, and they would normally be rapidly degraded to prevent the formation of defective RNA and RNA?protein complexes. Such tagging of RNAs by polyadenylation would be conceptually similar to the role of polyubiquitylation in targeting proteins for degradation by the proteasome complex.

Recent analyses have shown that these poly(A) tails are added by the TRAMP complexes (see above). Known TRAMP substrates include defective pre-tRNAs, pre-snoRNAs, pre-snRNAs and pre-rRNAs, aberrant forms of the U6 snRNA and 5S rRNA as well as a class of normally cryptic, unstable RNA polymerase II transcripts (CUTs) of unknown function17,18,19,20,68,70. This list includes the products of RNA polymerases I, II and III, showing the widespread use of the TRAMP?exosome system in nuclear RNA surveillance in yeast. Notably, these substrates have no discernable common features in RNA structure or sequence, supporting the idea that the TRAMP complexes recognize specific features of the RNA?protein structure.

A crucial difference between the functions of polyadenylation in cytoplasmic mRNA stability and nuclear RNA degradation might lie in the processivity of the reactions. Canonical pre-mRNA cleavage and polyadenylation by the poly(A) polymerase Pap1 is closely coupled to transcription and is highly processive, adding 250 nt in humans and 60?90 nt in yeast. This might ensure that no free 3′ end is available until a very long tail has been synthesized and covered by poly(A)-binding proteins such as Pab1. By contrast, Trf4 and Trf5 seem to have low processivity18,19,68,85, so the 3′ ends of tails that are added by the TRAMP complexes might frequently be free and available to the exosome. Under normal circumstances, tails that have been added by TRAMP in vivo might never be long enough for Pab1 binding.

There is currently only limited evidence that the TRAMP pathway operates in human cells, but a recent report indicates that at least one human exosome target undergoes polyadenylation86. Nascent transcripts from the human β-globin gene are cut at the co-transcriptional cleavage (CoTC) site, which lies downstream of a conventional pre-mRNA cleavage and polyadenylation site87. Following CoTC site cleavage, the 3′ product is degraded from its free 5′ end by the 5′ exonuclease XRN2 ? an activity that is important for subsequent transcription termination88. The 5′ fragment is degraded by the exosome, and cloning of the pre-mRNA CoTC cleavage products reveals that its 3′ end frequently undergoes oligo(A) addition86. It could be that CoTC-site cleavage has a surveillance function, as failure of the pre-mRNA to undergo rapid cleavage at the canonical polyadenylation site will lead to complete degradation by the exosome.

The nuclease activity of the RNase PH and PNPase family is reversible, allowing these proteins to function as polymerases. The pools of free ATP and ADP are larger than those of other nucleotides and, probably as a consequence, this polymerase activity generates tails that are largely but not exclusively poly(A). In the archaea Sulfolobus solfataricus and Sulfolobus acidocaldarius, polynucleotidylation is performed by the exosome89, analogously to polynucleotide addition by PNPase, which occurs in at least some bacteria and chloroplasts90,91,92. In principle, the activity of the RNase PH homologues in the eukaryotic exosome should also be reversible, leading to polynucleotide-tail addition. It is, therefore, notable that the tails on the cleaved human globin mRNAs include a small number of nucleotides other than adenosines86. Whether this reflects a polynucleotide-polymerase activity of the human exosome remains to be determined.

The localization and composition of the exosome complex might be more diverse in human cells than in yeast: Rrp6, which is restricted to the yeast nucleus, and Ski7, which is cytoplasmic in the yeast exosome, are each reported to localize to both the nucleus and the cytoplasm in human cells31,93. Moreover, the subcellular distribution of different core exosome subunits is apparently not identical in D. melanogaster cells94, perhaps indicating the existence of distinct subcomplexes. It seems clear that many details remain to be discovered.

Budding yeast have only two members of the Trf family of poly(A) polymerases, both of which are nuclear and have been implicated in RNA degradation. However, in other eukaryotes, including humans and fission yeast, other homologues are present in both the nucleus and the cytoplasm. In fission yeast, Cid12 is required for the small interfering RNA (siRNA)-dependent formation of heterochromatin around the centromeres, although its precise role is unclear. The best characterized cytoplasmic poly(A) polymerases are the Gld2 proteins, which add poly(A) tails to specific mRNA species and have key roles in development by activating the translation of stored maternal mRNAs (see Refs 95, 96 and the references therein).

Localization of nuclear RNA surveillance

Many RNA-processing and RNA-degradation activities are concentrated in specific subcellular regions; for example, ribosome synthesis occurs in the nucleolus and translational repression and 5′ degradation of cytoplasmic mRNAs occurs in P-bodies. The surveillance of at least some defective pre-ribosomes might take place in a distinct subregion of the nucleolus ? termed the 'No-body'97. In a strain that is defective in the nuclear export of pre-60S ribosome subunits, the late pre-rRNAs and mature 25S rRNA are polyadenylated by the TRAMP4 complex and then rapidly degraded by the exosome. Under these conditions, the polyadenylated pre-ribosomes are strongly concentrated in the No-body together with the TRAMP and the exosome complexes. In strains with defects in either the TRAMP or the exosome, this No-body enrichment is lost and both pre-rRNA and rRNA become stabilized97.

An apparently similar subnucleolar focus of polyadenylated RNA is also seen in strains with mutations in the TRAMP component Mtr4 (Ref. 98) or the exosome subunit Rrp6 (T. Carneiro, C. Caralho, D.T. and M. Carmo-Fonseca, submitted), which is consistent with the No-body localization of other RNAs that are undergoing surveillance or that are targeted for degradation. There has been extensive discussion of 'transcription factories', in which RNA-synthesis and RNA-processing factors are proposed to associate to facilitate efficient RNA maturation. It seems that there might also be an RNA 'demolition site' in which the degradation machinery is concentrated.

Conclusions and perspectives

The data reviewed here indicate that key features of RNA degradation have been conserved throughout evolution. In E. coli lysates, the most active degradative exonucleases are PNPase and RNase II ? and recent structural analyses indicate that the eukaryotic exosome resembles a complex between these enzymes. This is coupled with the realization that eukaryotic poly(A) polymerases have important roles in nuclear RNA degradation, probably by providing unstructured, single-stranded 'landing pads' for degradative exonucleases. This is similar to the role of oligoadenylation in bacteria.

A key outstanding question in understanding RNA-surveillance and quality-control pathways is: how can 'defective' RNA and RNA?protein complexes be distinguished from normal processing intermediates? For example, ribosome synthesis and pre-mRNA splicing take place in very large and highly dynamic complexes, and it is not clear how maturation defects can be rapidly and specifically identified. We favour a model that is based on kinetic proofreading, in which the degradation machinery is recruited by factors that are normally components of the precursor RNPs. If maturation proceeds with rapid kinetics, it will normally be completed before the binding of the surveillance machinery can occur. However, any delay in maturation, whether due to defects in the RNA itself or in its association with the large number of RNA-binding proteins, will result in RNA degradation, regardless of the actual nature of the defect. This type of mechanism avoids the problems that are inherent in attempting to identify specific defects. For example, ribosome synthesis in yeast involves at least 250 proteins and 70 snoRNAs as well as 7 kb of pre-rRNA, and the number of possible defects is clearly immense.

Another problem in understanding the mechanism and activation of the exosome is its role in either the precise trimming or the total degradation of different RNAs (or indeed the same RNAs under different conditions). The answer will most likely require detailed structural information on the exosome, its three-dimensional conformations while it is in contact with its RNP substrates, and the mechanism by which specific substrates are directed to specific exonuclease active sites. A clearer understanding of the relative use of phosphorolytic and hydrolytic activities of the exosome on known in vivo substrates will be an important first step. However, the striking distinction between the roles of polyadenylation for promoting mRNA stability versus promoting efficient RNA degradation reminds us that the devil is, as ever, in the detail.


  1. 1

    Warner, J. R. The economics of ribosome biosynthesis in yeast. Trends Biochem. Sci. 24, 437?440 (1999).

    CAS  PubMed  Google Scholar 

  2. 2

    Parker, R. & Song, H. The enzymes and control of eukaryotic mRNA turnover. Nature Struct. Mol. Biol. 11, 121?127 (2004).

    CAS  Google Scholar 

  3. 3

    Mitchell, P., Petfalski, E., Shevchenko, A., Mann, M. & Tollervey, D. The exosome; a conserved eukaryotic RNA processing complex containing multiple 3′→5′ exoribonuclease activities. Cell 91, 457?466 (1997).

    CAS  PubMed  Google Scholar 

  4. 4

    Allmang, C. et al. The yeast exosome and human PM-Scl are related complexes of 3′→5′ exonucleases. Genes Dev. 13, 2148?2158 (1999).

    CAS  PubMed  PubMed Central  Google Scholar 

  5. 5

    Peng, W. T. et al. A panoramic view of yeast noncoding RNA processing. Cell 113, 919?933 (2003).

    CAS  PubMed  Google Scholar 

  6. 6

    Burkard, K. T. & Butler, J. S. A nuclear 3′?5′ exonuclease involved in mRNA degradation interacts with Poly(A) polymerase and the hnRNA protein Npl3p. Mol. Cell. Biol. 20, 604?616 (2000).

    CAS  PubMed  PubMed Central  Google Scholar 

  7. 7

    Araki, Y. et al. Ski7p G protein interacts with the exosome and the Ski complex for 3′-to-5′ mRNA decay in yeast. EMBO J. 20, 4684?4693 (2001).

    CAS  PubMed  PubMed Central  Google Scholar 

  8. 8

    van Hoof, A., Staples, R. R., Baker, R. E. & Parker, R. Function of the ski4p (Csl4p) and Ski7p proteins in 3′-to-5′ degradation of mRNA. Mol. Cell. Biol. 20, 8230?8243 (2000).

    CAS  PubMed  PubMed Central  Google Scholar 

  9. 9

    Mitchell, P. et al. Rrp47p is an exosome-associated protein required for the 3′ processing of stable RNAs. Mol. Cell. Biol. 23, 6982?6992 (2003).

    CAS  PubMed  PubMed Central  Google Scholar 

  10. 10

    Yavuzer, U., Smith, G. C., Bliss, T., Werner, D. & Jackson, S. P. DNA end-independent activation of DNA-PK mediated via association with the DNA-binding protein C1D. Genes Dev. 12, 2188?2199 (1998).

    CAS  PubMed  PubMed Central  Google Scholar 

  11. 11

    van Hoof, A., Lennertz, P. & Parker, R. Yeast exosome mutants accumulate 3′-extended polyadenylated forms of U4 small nuclear RNA and small nucleolar RNAs. Mol. Cell. Biol. 20, 441?452 (2000).

    CAS  PubMed  PubMed Central  Google Scholar 

  12. 12

    Zanchin, N. I. & Goldfarb, D. S. The exosome subunit Rrp43p is required for the efficient maturation of 5.8S, 18S and 25S rRNA. Nucleic Acids Res. 27, 1283?1288 (1999).

    CAS  PubMed  PubMed Central  Google Scholar 

  13. 13

    Allmang, C. et al. Functions of the exosome in rRNA, snoRNA and snRNA synthesis. EMBO J. 18, 5399?5410 (1999).

    CAS  PubMed  PubMed Central  Google Scholar 

  14. 14

    de la Cruz, J., Kressler, D., Tollervey, D. & Linder, P. Dob1p (Mtr4p) is a putative ATP-dependent RNA helicase required for the 3′ end formation of 5.8S rRNA in Saccharomyces cerevisiae. EMBO J. 17, 1128?1140 (1998).

    CAS  PubMed  PubMed Central  Google Scholar 

  15. 15

    Grosshans, H., Deinert, K., Hurt, E. & Simos, G. Biogenesis of the signal recognition particle (SRP) involves import of SRP proteins into the nucleolus, assembly with the SRP-RNA, and Xpo1p-mediated export. J. Cell Biol. 153, 745?762 (2001).

    CAS  PubMed  PubMed Central  Google Scholar 

  16. 16

    Allmang, C., Mitchell, P., Petfalski, E. & Tollervey, D. Degradation of ribosomal RNA precursors by the exosome. Nucleic Acids Res. 28, 1684?1691 (2000).

    CAS  PubMed  PubMed Central  Google Scholar 

  17. 17

    Wyers, F. et al. Cryptic Pol II transcripts are degraded by a nuclear quality control pathway involving a new poly(A) polymerase. Cell 121, 725?737 (2005). Shows that the TRAMP complex and the exosome rapidly degrade transcripts that are produced from intergenic regions. These transcripts are very common, being produced from up to 10% of intergenic regions under normal conditions.

    CAS  PubMed  Google Scholar 

  18. 18

    LaCava, J. et al. RNA degradation by the exosome is promoted by a nuclear polyadenylation complex. Cell 21, 713?724 (2005).

    Google Scholar 

  19. 19

    Vanacova, S. et al. A new yeast poly(A) polymerase complex involved in RNA quality control. PLoS Biol. 3, e189 (2005). A biochemical analysis of the activity of the TRAMP complex, showing that it can recruit the exosome to degrade an aberrant tRNA in vitro.

    PubMed  Google Scholar 

  20. 20

    Kadaba, S. et al. Nuclear surveillance and degradation of hypomodified initiator tRNAMet in S. cerevisiae. Genes Dev. 18, 1227?1240 (2004). First report that polyadenylation by Trf4 is linked to RNA degradation by the exosome.

    CAS  PubMed  PubMed Central  Google Scholar 

  21. 21

    Torchet, C. et al. Processing of 3′ extended read-through transcripts by the exosome can generate functional mRNAs. Mol. Cell 9, 1285?1296 (2002).

    CAS  PubMed  Google Scholar 

  22. 22

    Hilleren, P., McCarthy, T., Rosbash, M., Parker, R. & Jensen, T. H. Quality control of mRNA 3′-end processing is linked to the nuclear exosome. Nature 413, 538?542 (2001).

    CAS  PubMed  Google Scholar 

  23. 23

    Bousquet-Antonelli, C., Presutti, C. & Tollervey, D. Identification of a regulated pathway for nuclear pre-mRNA turnover. Cell 102, 765?775 (2000).

    CAS  PubMed  Google Scholar 

  24. 24

    Lee, A., Henras, A. K. & Chanfreau, G. Multiple RNA surveillance pathways limit aberrant expression of iron uptake mRNAs and prevent iron toxicity in S. cerevisiae. Mol. Cell 19, 39?51 (2005). Reports that mRNA-encoding proteins that are involved in iron metabolism are subject to various different nuclear RNA-surveillance pathways.

    CAS  PubMed  Google Scholar 

  25. 25

    Kuai, L., Das, B. & Sherman, F. A nuclear degradation pathway controls the abundance of normal mRNAs in Saccharomyces cerevisiae. Proc. Natl Acad. Sci. USA 102, 13962?13967 (2005).

    CAS  PubMed  Google Scholar 

  26. 26

    Roth, K. M., Wolf, M. K., Rossi, M. & Butler, J. S. The nuclear exosome contributes to autogenous control of NAB2 mRNA levels. Mol. Cell. Biol. 25, 1577?1585 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  27. 27

    Houalla, R. et al. Microarray detection of novel nuclear RNA substrates for the exosome. Yeast 23, 439?454 (2006).

    CAS  PubMed  Google Scholar 

  28. 28

    Anderson, J. S. J. & Parker, R. P. The 3′ to 5′ degradation of yeast mRNAs is a general mechanism for mRNA turnover that requires the SKI2 DEVH box protein and 3′ to 5′ exonucleases of the exosome complex. EMBO J. 17, 1497?1506 (1998).

    CAS  PubMed  PubMed Central  Google Scholar 

  29. 29

    Takahashi, S., Araki, Y., Sakuno, T. & Katada, T. Interaction between Ski7p and Upf1p is required for nonsense-mediated 3′-to-5′ mRNA decay in yeast. EMBO J. 22, 3951?3959 (2003).

    CAS  PubMed  PubMed Central  Google Scholar 

  30. 30

    Mitchell, P. & Tollervey, D. An NMD pathway in yeast involving accelerated deadenylation and exosome-mediated 3′→5′ degradation. Mol. Cell 11, 1405?1413 (2003).

    CAS  PubMed  Google Scholar 

  31. 31

    Lejeune, F., Li, X. & Maquat, L. E. Nonsense-mediated mRNA decay in mammalian cells involves decapping, deadenylating, and exonucleolytic activities. Mol. Cell 12, 675?687 (2003).

    CAS  PubMed  Google Scholar 

  32. 32

    Frischmeyer, P. A. et al. An mRNA surveillance mechanism that eliminates transcripts lacking termination codons. Science 295, 2258?2261 (2002).

    CAS  PubMed  Google Scholar 

  33. 33

    van Hoof, A., Frischmeyer, P. A., Dietz, H. C. & Parker, R. Exosome-mediated recognition and degradation of mRNAs lacking a termination codon. Science 295, 2262?2264 (2002).

    CAS  PubMed  PubMed Central  Google Scholar 

  34. 34

    Tran, H., Schilling, M., Wirbelauer, C., Hess, D. & Nagamine, Y. Facilitation of mRNA deadenylation and decay by the exosome-bound, DExH protein RHAU. Mol. Cell 13, 101?111 (2004).

    CAS  PubMed  Google Scholar 

  35. 35

    Chen, C. Y. et al. AU binding proteins recruit the exosome to degrade ARE-containing mRNAs. Cell 107, 451?464 (2001).

    CAS  PubMed  Google Scholar 

  36. 36

    Mukherjee, D. et al. The mammalian exosome mediates the efficient degradation of mRNAs that contain AU-rich elements. EMBO J. 21, 165?174 (2002).

    CAS  PubMed  PubMed Central  Google Scholar 

  37. 37

    Gherzi, R. et al. A KH domain RNA binding protein, KSRP, promotes ARE-directed mRNA turnover by recruiting the degradation machinery. Mol. Cell 14, 571?583 (2004).

    CAS  PubMed  Google Scholar 

  38. 38

    Doma, M. K. & Parker, R. Endonucleolytic cleavage of eukaryotic mRNAs with stalls in translation elongation. Nature 440, 561?564 (2006). First report of a yeast mRNA-surveillance pathway that is initiated by RNA cleavage.

    CAS  PubMed  PubMed Central  Google Scholar 

  39. 39

    Tollervey, D. RNA lost in translation. Nature 440, 425?426 (2006).

    CAS  PubMed  Google Scholar 

  40. 40

    Orban, T. I. & Izaurralde, E. Decay of mRNAs targeted by RISC requires XRN1, the Ski complex, and the exosome. RNA 11, 459?469 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  41. 41

    Symmons, M. F., Jones, G. H. & Luisi, B. F. A duplicated fold is the structural basis for polynucleotide phosphorylase catalytic activity, processivity, and regulation. Struct. Fold. Des. 8, 1215?1226 (2000).

    CAS  Google Scholar 

  42. 42

    Stickney, L. M., Hankins, J. S., Miao, X. & Mackie, G. A. Function of the conserved S1 and KH domains in polynucleotide phosphorylase. J. Bacteriol. 187, 7214?7121 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  43. 43

    Valentine, R. C., Thang, M. N. & Grunberg-Manago, M. Electron microscopy of Escherichia coli polynucleotide phosphorylase molecules and polyribonucleotide formation. J. Mol. Biol. 39, 389?391 (1969).

    CAS  PubMed  Google Scholar 

  44. 44

    Spickler, C. & Mackie, G. A. Action of RNase II and polynucleotide phosphorylase against RNAs containing stem-loops of defined structure. J. Bacteriol. 182, 2422?2427 (2000).

    CAS  PubMed  PubMed Central  Google Scholar 

  45. 45

    Lorentzen, E. et al. The archaeal exosome core is a hexameric ring structure with three catalytic subunits. Nature Struct. Mol. Biol. 12, 575?581 (2005). The first structure of the archaeal exosome, which showed that only one of the archaeal RNase PH-like exosome subunits is an active ribonuclease, although both are required for activity.

    CAS  Google Scholar 

  46. 46

    Lorentzen, E. & Conti, E. Structural basis of 3′ end RNA recognition and exoribonucleolytic cleavage by an exosome RNase PH core. Mol. Cell 20, 473?481 (2005). Following from the previous analysis, the authors show here how the exosome binds and degrades substrate RNAs.

    CAS  PubMed  Google Scholar 

  47. 47

    Buttner, K., Wenig, K. & Hopfner, K. P. Structural framework for the mechanism of archaeal exosomes in RNA processing. Mol. Cell 20, 461?471 (2005). This work characterizes the structure of the complete archaeal exosome.

    PubMed  Google Scholar 

  48. 48

    Symmons, M. F., Williams, M. G., Luisi, B. F., Jones, G. H. & Carpousis, A. J. Running rings around RNA: a superfamily of phosphate-dependent RNases. Trends Biochem. Sci. 27, 11?18 (2002).

    CAS  PubMed  Google Scholar 

  49. 49

    Aloy, P. et al. A complex prediction: three-dimensional model of the yeast exosome. EMBO Rep. 3, 628?635 (2002).

    CAS  PubMed  PubMed Central  Google Scholar 

  50. 50

    van Hoof, A. & Parker, R. The exosome: a proteasome for RNA? Cell 99, 347?350 (1999).

    CAS  PubMed  Google Scholar 

  51. 51

    Andrulis, E. D. et al. The RNA processing exosome is linked to elongating RNA polymerase II in Drosophila. Nature 420, 837?841 (2002).

    CAS  PubMed  Google Scholar 

  52. 52

    Brouwer, R., Pruijn, G. J. & van Venrooij, W. J. The human exosome: an autoantigenic complex of exoribonucleases in myositis and scleroderma. Arthritis Res. 3, 102?106 (2001).

    CAS  PubMed  Google Scholar 

  53. 53

    Brouwer, R. et al. Three novel components of the human exosome. J. Biol. Chem. 276, 6177?6184 (2001).

    CAS  PubMed  Google Scholar 

  54. 54

    Estevez, A. M., Kempf, T. & Clayton, C. The exosome of Trypanosoma brucei. EMBO J. 20, 3831?3839 (2001).

    CAS  PubMed  PubMed Central  Google Scholar 

  55. 55

    Chekanova, J. A., Dutko, J. A., Mian, I. S. & Belostotsky, D. A. Arabidopsis thaliana exosome subunit AtRrp4p is a hydrolytic 3′→5′ exonuclease containing S1 and KH RNA-binding domains. Nucleic Acids Res. 30, 695?700 (2002).

    CAS  PubMed  PubMed Central  Google Scholar 

  56. 56

    Briggs, M. W., Burkard, K. T. & Butler, J. S. Rrp6p, the yeast homologue of the human PM-Scl 100-kDa autoantigen, is essential for efficient 5.8 S rRNA 3′ end formation. J. Biol. Chem. 273, 13255?13263 (1998).

    CAS  PubMed  Google Scholar 

  57. 57

    Phillips, S. & Butler, J. S. Contribution of domain structure to the RNA 3′ end processing and degradation functions of the nuclear exosome subunit Rrp6p. RNA 9, 1098?1107 (2003).

    CAS  PubMed  PubMed Central  Google Scholar 

  58. 58

    Raijmakers, R., Egberts, W. V., van Venrooij, W. J. & Pruijn, G. J. Protein?protein interactions between human exosome components support the assembly of RNase PH-type subunits into a six-membered PNPase-like ring. J. Mol. Biol. 323, 653?663 (2002).

    CAS  PubMed  Google Scholar 

  59. 59

    Estevez, A. M., Lehner, B., Sanderson, C. M., Ruppert, T. & Clayton, C. The roles of intersubunit interactions in exosome stability. J. Biol. Chem. 278, 34943?34951 (2003).

    CAS  PubMed  Google Scholar 

  60. 60

    Lehner, B. & Sanderson, C. M. A protein interaction framework for human mRNA degradation. Genome Res. 14, 1315?1323 (2004).

    CAS  PubMed  PubMed Central  Google Scholar 

  61. 61

    Milligan, L., Torchet, C., Allmang, C., Shipman, T. & Tollervey, D. A nuclear surveillance pathway for mRNAs with defective polyadenylation. Mol. Cell. Biol. 25, 9996?10004 (2005). Describes how the exosome responds to mRNA molecules with defective poly(A) tails.

    CAS  PubMed  PubMed Central  Google Scholar 

  62. 62

    Wang, L., Lewis, M. S. & Johnson, A. W. Domain interactions within the Ski2?3?8 complex and between the Ski complex and Ski7p. RNA 11, 1291?1302 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  63. 63

    Arigo, J. T., Carroll, K. L., Ames, J. M. & Corden, J. L. Regulation of yeast NRD1 expression by premature transcription termination. Mol. Cell 21, 641?651 (2006). Reports that autoregulation of Nrd1 expression involves the RNA-binding proteins Nrd1 and Nab3 as well as the exosome.

    CAS  PubMed  Google Scholar 

  64. 64

    Vasiljeva, L. & Buratowski, S. Nrd1 Interacts with the nuclear exosome for 3′ processing of RNA polymerase II transcripts. Mol. Cell 21, 239?248 (2006). Reports a second exosome-stimulating complex. The Nrd1 protein binds to a specific target-RNA sequence and recruits the exosome to facilitate degradation.

    CAS  PubMed  Google Scholar 

  65. 65

    Morlando, M. et al. Coupling between snoRNP assembly and 3′ processing controls box C/D snoRNA biosynthesis in yeast. EMBO J. 23, 2392?2401 (2004).

    CAS  PubMed  PubMed Central  Google Scholar 

  66. 66

    Steinmetz, E. J., Conrad, N. K., Brow, D. A. & Corden, J. L. RNA-binding protein Nrd1 directs poly(A)-independent 3′-end formation of RNA polymerase II transcripts. Nature 413, 327?331 (2001).

    CAS  PubMed  Google Scholar 

  67. 67

    Puig, S., Askeland, E. & Thiele, D. J. Coordinated remodeling of cellular metabolism during iron deficiency through targeted mRNA degradation. Cell 120, 99?110 (2005). Reports that an ARE-mediated mRNA-degradation system functions in yeast and involves the yeast homologue of the well characterized human ARE-binding protein TTP.

    CAS  PubMed  Google Scholar 

  68. 68

    Houseley, J. & Tollervey, D. Yeast Trf5p is a nuclear poly(A) polymerase. EMBO Rep. 7, 205?211 (2005).

    PubMed Central  Google Scholar 

  69. 69

    Egecioglu, D. E., Henras, A. K. & Chanfreau, G. F. Contributions of Trf4p- and Trf5p-dependent polyadenylation to the processing and degradative functions of the yeast nuclear exosome. RNA 12, 26?32 (2006). References 68 and 69 describe a second TRAMP complex, and show that the two TRAMP complexes have different substrate affinities.

    CAS  PubMed  PubMed Central  Google Scholar 

  70. 70

    Kadaba, S., Wang, X. & Anderson, J. T. Nuclear RNA surveillance in Saccharomyces cerevisiae: Trf4p-dependent polyadenylation of nascent hypomethylated tRNA and an aberrant form of 5S rRNA. RNA 12, 508?521 (2006). Extends the range of TRAMP substrates and shows that the complex acts on newly transcribed pre-tRNA.

    CAS  PubMed  PubMed Central  Google Scholar 

  71. 71

    Gavin, A. C. et al. Functional organization of the yeast proteome by systematic analysis of protein complexes. Nature 415, 141?147 (2002).

    CAS  PubMed  PubMed Central  Google Scholar 

  72. 72

    Gavin, A. C. et al. Proteome survey reveals modularity of the yeast cell machinery. Nature 440, 631?636 (2006).

    CAS  PubMed  Google Scholar 

  73. 73

    Schilders, G., Raijmakers, R., Raats, J. M. & Pruijn, G. J. MPP6 is an exosome-associated RNA-binding protein involved in 5.8S rRNA maturation. Nucleic Acids Res. 33, 6795?6804 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  74. 74

    Thomson, E. & Tollervey, D. Nop53p is required for late 60S ribosome subunit maturation and nuclear export in yeast. RNA 11, 1215?1224 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  75. 75

    Granato, D. C. et al. Nop53p, an essential nucleolar protein that interacts with Nop17p and Nip7p, is required for pre-rRNA processing in Saccharomyces cerevisiae. FEBS J. 272, 4450?4463 (2005).

    CAS  PubMed  Google Scholar 

  76. 76

    Brown, J. T., Bai, X. & Johnson, A. W. The yeast antiviral proteins Ski2p, Ski3p, and Ski8p exist as a complex in vivo. RNA 6, 449?457 (2000).

    CAS  PubMed  PubMed Central  Google Scholar 

  77. 77

    Inada, T. & Aiba, H. Translation of aberrant mRNAs lacking a termination codon or with a shortened 3′-UTR is repressed after initiation in yeast. EMBO J. 24, 1584?1595 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  78. 78

    Cao, D. & Parker, R. Computational modeling and experimental analysis of nonsense-mediated decay in yeast. Cell 113, 533?45 (2003).

    CAS  PubMed  Google Scholar 

  79. 79

    Wickner, R. B. Double-stranded and single-stranded RNA viruses of Saccharomyces cerevisiae. Annu. Rev. Microbiol. 46, 347?375 (1992).

    CAS  PubMed  Google Scholar 

  80. 80

    Benard, L., Carroll, K., Valle, R. C., Masison, D. C. & Wickner, R. B. The ski7 antiviral protein is an EF1-α homolog that blocks expression of non-Poly(A) mRNA in Saccharomyces cerevisiae. J. Virol. 73, 2893?2900 (1999).

    CAS  PubMed  PubMed Central  Google Scholar 

  81. 81

    Ridley, S. P., Sommer, S. S. & Wickner, R. B. Superkiller mutations in Saccharomyces cerevisiae suppress exclusion of M2 double-stranded RNA by L-A-HN and confer cold sensitivity in the presence of M and L-A-HN. Mol. Cell. Biol. 4, 761?770 (1984).

    CAS  PubMed  PubMed Central  Google Scholar 

  82. 82

    Folichon, M. et al. Fate of mRNA extremities generated by intrinsic termination: detailed analysis of reactions catalyzed by ribonuclease II and poly(A) polymerase. Biochimie 87, 819?826 (2005).

    CAS  PubMed  Google Scholar 

  83. 83

    Khemici, V. & Carpousis, A. J. The RNA degradosome and poly(A) polymerase of Escherichia coli are required in vivo for the degradation of small mRNA decay intermediates containing REP-stabilizers. Mol. Microbiol. 51, 777?790 (2004).

    CAS  PubMed  Google Scholar 

  84. 84

    Kuai, L., Fang, F., Butler, J. S. & Sherman, F. Polyadenylation of rRNA in Saccharomyces cerevisiae. Proc. Natl Acad. Sci. USA 101, 8581?8586 (2004).

    CAS  PubMed  Google Scholar 

  85. 85

    Haracska, L., Johnson, R. E., Prakash, L. & Prakash, S. Trf4 and Trf5 proteins of Saccharomyces cerevisiae exhibit poly(A) RNA polymerase activity but no DNA polymerase activity. Mol. Cell. Biol. 25, 10183?10189 (2005). Reports that both Trf4 and Trf5 are poly(A) polymerases.

    CAS  PubMed  PubMed Central  Google Scholar 

  86. 86

    West, S., Gromak, N., Norbury, C. J. & Proudfoot, N. J. Adenylation and exosome-mediated degradation of co-transcriptionally cleaved pre-messenger RNA in human cells Mol. Cell 21, 437?443 (2006). Evidence is presented that human mRNA molecules that fail to be correctly cleaved and polyadenylated are rapidly targeted by the exosome for degradation.

    CAS  PubMed  Google Scholar 

  87. 87

    Teixeira, A. et al. Autocatalytic RNA cleavage in the human β-globin pre-mRNA promotes transcription termination. Nature 432, 526?530 (2004).

    CAS  PubMed  Google Scholar 

  88. 88

    West, S., Gromak, N. & Proudfoot, N. J. Human 5′→3′ exonuclease Xrn2 promotes transcription termination at co-transcriptional cleavage sites. Nature 432, 522?525 (2004). Resolving a long-standing argument, West et al . show that cleavage occurs prior to transcriptional termination, the continuing RNA polymerase is then chased down by a 5′→3′ exonuclease that triggers its release.

    CAS  PubMed  Google Scholar 

  89. 89

    Portnoy, V. et al. RNA polyadenylation in Archaea: not observed in Haloferax while the exosome polynucleotidylates RNA in Sulfolobus. EMBO Rep. 6, 1188?1193 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  90. 90

    Yehudai-Resheff, S., Hirsh, M. & Schuster, G. Polynucleotide phosphorylase functions as both an exonuclease and a poly(A) polymerase in spinach chloroplasts. Mol. Cell. Biol. 21, 5408?5416 (2001).

    CAS  PubMed  PubMed Central  Google Scholar 

  91. 91

    Mohanty, B. K. & Kushner, S. R. Polynucleotide phosphorylase functions both as a 3′→5′ exonuclease and a poly(A) polymerase in Escherichia coli. Proc. Natl Acad. Sci. USA 97, 11966?11971 (2000).

    CAS  PubMed  Google Scholar 

  92. 92

    Bollenbach, T. J., Schuster, G. & Stern, D. B. Cooperation of endo- and exoribonucleases in chloroplast mRNA turnover. Prog. Nucleic Acid. Res. Mol. Biol. 78, 305?337 (2004).

    CAS  PubMed  Google Scholar 

  93. 93

    Zhu, B. et al. The human PAF complex coordinates transcription with events downstream of RNA synthesis. Genes Dev. 19, 1668?1673 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  94. 94

    Graham, A. C., Kiss, D. L. & Andrulis, E. D. Differential distribution of exosome subunits at the nuclear lamina and in cytoplasmic foci. Mol. Biol. Cell 17, 1399?1409 (2006). Reports that some exosome components show differential enrichments at distinct locations, which indicates that all human exosome components might not obligatorily function as a complex.

    CAS  PubMed  PubMed Central  Google Scholar 

  95. 95

    Rouhana, L. et al. Vertebrate GLD2 poly(A) polymerases in the germline and the brain. RNA 11, 1117?1130 (2005).

    CAS  PubMed  PubMed Central  Google Scholar 

  96. 96

    Kwak, J. E., Wang, L., Ballantyne, S., Kimble, J. & Wickens, M. Mammalian GLD-2 homologs are poly(A) polymerases. Proc. Natl Acad. Sci. USA 101, 4407?4412 (2004).

    CAS  PubMed  Google Scholar 

  97. 97

    Dez, C., Houseley, J. & Tollervey, D. Surveillance of nuclear-restricted pre-ribosomes within a subnucleolar region of Saccharomyces cerevisiae. EMBO J. 25, 1534?1546 (2006). The authors show that defective pre-ribosomes are rapidly and completely degraded by the TRAMP and exosome in nucleolar 'No-bodies'.

    CAS  PubMed  PubMed Central  Google Scholar 

  98. 98

    Kadowaki, T. et al. Isolation and characterization of Saccharomyces cerevisiae mRNA transport-defective (mtr) mutants. J. Cell Biol. 126, 649?659 (1994).

    CAS  PubMed  Google Scholar 

Download references

Author information



Ethics declarations

Competing interests

The authors declare no competing financial interests.

Related links

Related links


David Tollervey's homepage

The PyMOL Molecular Graphics System


Small nucleolar RNA

(snoRNA). A small RNA molecule that functions in ribosome biogenesis in the nucleolus. Most snoRNAs direct site-specific base modification in the pre-ribosomal RNAs, whereas a small number are required for pre-ribosomal RNA cleavage.

Small nuclear RNAs

Five small RNA species (U1, U2, U4, U5 and U6) that form the core of the pre-mRNA-splicing system in the nucleus.

Ribonucleoprotein particle

A complex of proteins and RNA. In many cases, the proteins can recognize their cognate mRNA molecules (selective binding) and mediate their delivery to specific regions within the cell.

Nonsense-mediated decay

The process by which the cell destroys mRNAs in which translation has been prematurely terminated owing to the presence of a nonsense codon within the coding region.

Non-stop decay

An RNA-degradation pathway for mRNAs that do not contain a translation-termination codon and in which the translating ribosomes stall at the end of the transcript.

No-go decay

An RNA-degradation pathway that is induced by the stalling of a translating ribosome. The RNA is cleaved in the vicinity of the stalled ribosome, and this is followed by exonuclease degradation of the cleaved fragments. The 5′ fragment is degraded by the exosome, whereas the 3′ fragment is degraded by the 5′ exonuclease Xrn1.

S1 domain

A putative RNA-binding domain that was initially identified in ribosomal protein S1, and that is present in a large number of RNA-associated proteins.

KH domain

An evolutionarily conserved single-stranded-RNA-binding domain that was originally identified in the human heterogeneous nuclear (hn)RNP K protein.


A large multisubunit protease complex that selectively degrades multi-ubiquitylated proteins. It contains a 20S particle that has catalytic activity and two regulatory 19S particles.

DExH-box RNA helicases

RNA helicases that are related to DEAD-box proteins, and that contain an Asp-Glu-x-His (DExH) conserved motif.

ARE-mediated mRNA degradation

A process of rapid degradation of mRNAs that is mediated by the recruitment of the exosome by factors that bind to A+U-rich sequence elements (AREs) that are generally located in the 3′ untranslated region of the mRNAs.

Zinc-knuckle domain

This CX2CX4HX4C motif binds two zinc atoms and has been implicated in interactions between proteins and with single-stranded nucleic acids.

5′ cap structure

A structure that is located at the 5′ end of eukaryotic mRNAs and that consists of m7GpppN (where m7G represents 7-methylguanosine, p represents a phosphate group and N represents any base).


(Processing body). The cytoplasmic site of mRNA degradation. mRNAs on which translation has ceased are released from the polysome pool and recruited to P-bodies, where they accumulate together with the mRNA decapping and 5′-degradation machinery.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Houseley, J., LaCava, J. & Tollervey, D. RNA-quality control by the exosome. Nat Rev Mol Cell Biol 7, 529–539 (2006). https://doi.org/10.1038/nrm1964

Download citation

Further reading


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing