The exosome ribonuclease complex functions in both the limited trimming of the 3′-ends of nuclear substrates during RNA processing events and the complete destruction of nuclear and cytoplasmic RNAs. The two RNases of the eukaryotic exosome, Rrp44 (rRNA-processing protein 44) and Rrp6, are bound at either end of a catalytically inert cylindrical core. RNA substrates are threaded through the internal channel of the core to Rrp44 by RNA helicase components of the nuclear TRAMP complex (Trf4–Air2–Mtr4 polyadenylation complex) or the cytoplasmic Ski (superkiller) complex. Recent studies reveal that Rrp44 can also associate directly with substrates via channel-independent routes. Although the substrates of the exosome are known, it is not clear whether specific substrates are restricted to one or other pathway. Data currently available support the model that processed substrates are targeted directly to the catalytic subunits, whereas at least some substrates that are directed towards discard pathways must be threaded through the exosome core.
Functions of the exosome
The exosome ribonuclease complex acts on all types of RNA at some point during their lifetime, functioning in: (i) the 3′-end maturation of stable RNAs (and some mRNAs); (ii) the complete degradation of fragments such as the 5′-ETS (5′-external transcribed spacer) that are released as a result of RNA processing reactions; (iii) quality control of all classes of nuclear and cytoplasmic RNAs; (iv) the constitutive degradation of unstable nuclear transcripts such as CUTs (cryptic unstable transcripts); and (v) the ultimate degradation of cytoplasmic mRNAs [1,2]. The exosome complex is conserved throughout eukarya and most archaea, and genetic depletion of any subunit except Rrp6 (rRNA-processing protein 6) causes a block in mitotic growth in yeast [3,4]. The importance of exosome function in human biology is reflected by the observations that active site mutations in Rrp44 correlate with the incidence of multiple myeloma and relapsed acute myeloid leukaemia [5,6], whereas mutations in the related enzyme Dis3L2 (DIS3-like exonuclease 2) are linked to Perlman syndrome and Wilm's tumour predisposition . Genetic complementation analyses underscore the functional conservation of exosome subunits through fungi, plants and metazoa [3,8–10].
Structure of the exosome
The eukaryotic exosome has a highly stable ‘core’, consisting of an asymmetric double-layered ring. The lower tier comprises six subunits (Rrp41, Rrp42, Rrp43, Rrp45, Rrp46 and Mtr3) that are structurally related to RNase PH. This is overlaid with an upper tier of three ‘cap’ subunits (Rrp4, Rrp40 and Csl4) that have S1 or KH (K homology) RNA-binding domains [11,12] (see Figure 1). A structurally homologous complex is found in archaea , which is itself similar in architecture to the bacterial PNPase (polynucleotide phosphorylase) enzyme . Catalytically, however, there are clear differences between these complexes. Although the archaeal complex contains multiple copies of fewer subunits, the subunits of the eukaryotic exosome have diverged through evolution. The eukaryotic exosome core has lost catalytic activity, but the substrate RNAs traverse the same pathway taken through the central channel of the archaeal complex [12,13]. The eukaryotic complex is directly associated with the eukaryotic-specific ribonucleases Rrp44 (also known as Dis3) and Rrp6 [3,4]. Rrp44 is bound to the core at the ‘bottom’ of the channel near the RNA exit site [15,16], whereas Rrp6 is bound to the ‘top’ of the core [12,17] (Figure 1). Rrp44 and Rrp6 are hydrolytic ribonucleases [18,19], whereas the archaeal complex and PNPase are phosphorolytic enzymes.
Structure of the exosome
The exosome RNase Rrp44
Rrp44 is related to the RNase II/RNase R family of 3′→5′ exoribonucleases . The catalytic RNB exonuclease domain is flanked at the N-terminal end by two tandem cold-shock domains (CSD1 and CSD2) and at the C-terminal end by an S1 domain (Figure 1). The cold shock and S1 domains contribute to RNA binding. In addition to its exonuclease activity, Rrp44 also has a distinct endonuclease activity that is associated with its PIN (PilT N-terminus) domain [21–23]. Full-length Rrp44 protein nevertheless degrades RNA in a 3′→5′ direction [3,24], indicating that the endo- and exo-nuclease activities of Rrp44 are functionally coupled. The PIN domain also mediates the interaction between Rrp44 and the exosome core [23,25]. Both RNase activities of Rrp44 are largely suppressed upon interaction with the exosome through allosteric effects that diminish its ability to bind RNA . In yeast, there is a single gene encoding Rrp44 and the protein functions in the nucleolus, the nucleoplasm and the cytoplasm [3,26]. In contrast, three Rrp44 homologues are found in human cells: Dis3 is nucleoplasmic and excluded from the nucleolus, whereas Dis3L and Dis3L2 are cytoplasmic [27,28]. Dis3L lacks endonuclease activity, but is associated with the exosome core, whereas Dis3L2 lacks the PIN domain and is not associated with the exosome. Dis3L2 plays a role in cytoplasmic mRNA degradation, in both human cells and in the fission yeast Schizosaccharomyces pombe [28,29]. Thus at least some eukaryotic Rrp44 homologues function independently of the exosome core complex. Indeed, a recent study reported that Rrp44 in budding yeast functions in mitochondria independently of the exosome core . Whether yeast Rrp44 can fully function when separated from the exosome remains to be determined. Notably, expression of the PIN domain alone is able to support slow growth .
The exosome RNase Rrp6
Rrp6 belongs to the RNase D family of 3′→5′ exoribonucleases that contain a DEDD exonuclease domain juxtaposed to one or more HRDC domains (helicase and RNase D C-terminal domains) . Rrp6 members of this family also contain an N-terminal PMC2NT domain  and a C-terminal exosome-interacting domain  (see Figure 1). Interaction of Rrp6 with the exosome core has little effect on its exonuclease activity, but allosterically stimulates the activity of Rrp44 . Yeast Rrp6 is a nuclear protein , whereas hRRP6 (human Rrp6 protein; also known as PM-Scl100) is restricted to the nucleolus . Three different Rrp6 proteins are expressed in Arabidopsis thaliana: Rrp6L1 and Rrp6L2 proteins are found in the nucleus, whereas Rrp6L3 is localized to the cytoplasm . Thus distinct forms of both Rrp44 and Rrp6 have evolved in different systems and presumably carry out specialized functions.
Rrp6 and the small nuclear RNA-binding protein Rrp47 (also known as Lrp1 in yeast, or C1D in human cells) form a heterodimer through an interaction between their N-terminal PMC2NT and Sas10/C1D domains [34,35]. This interaction provides mutual protein stabilization [36,37] and limits expression of the Rrp6–Rrp47 complex to the nucleus . The C-terminal region of Rrp47 mediates interactions with snoRNP (small nucleolar ribonucleoprotein) components that are critical for the normal 3′-end maturation of box C/D snoRNAs (small nucleolar RNAs) . Thus the Rrp6–Rrp47 interaction also allows RNA degradation to be coupled with substrate recognition or binding.
Threading substrates through the exosome
Rrp44 alone shields a stretch of nine to twelve nucleotides in RNase protection assays, and 31–33 nucleotides when associated with the exosome core . Substrate binding to Rrp44 requires a single-stranded 3′-terminus, but, once engaged with the RNA, Rrp44 is a highly processive exoribonuclease that efficiently degrades structured RNA . The energy released upon hydrolytic activity of Rrp44 is released in bursts, pulling substrates through the complex by a steric occlusion mechanism [39,40].
The function of the exosome complex in vivo is dependent on additional factors, notably the structurally closely related RNA helicases Mtr4 (also known as Dob1) and Ski2 (superkiller 2) [41,42]. Yeast Mtr4, together with the non-canonical poly(A) polymerases Trf4 or Trf5 and the RNA-binding proteins Air1 or Air2, constitute the TRAMP complexes (Trf4–Air2–Mtr4 polyadenylation complex) that stimulate exosome-mediated RNA degradation in vitro and in vivo [43–45]. The human protein hMTR4 is found in a homologous TRAMP complex together with hRRP6 in the nucleolus, as well as being a component of the distinct NEXT complex (nuclear exosome targeting complex) . This suggests that hMTR4, like yeast Mtr4, has a major role in targeting nuclear RNAs to the exosome and that it promotes the degradation of some substrates independently of their oligoadenylation. Ski2 is a component of the cytoplasmic Ski complex, which contains the scaffold protein Ski3 and two copies of the protein Ski8 [47,48], and is critical for exosome-mediated cytoplasmic mRNA degradation  and mRNA surveillance pathways [49–51].
RNase protection assays suggest that the Ski complex is positioned on top of the exosome core , as has been proposed for Mtr4 [53,54]. The RNA helicases Mtr4 and Ski2 are thought to unfold the 3′-end of substrates and thread the resultant ssRNA 3′-end into the exosome core for subsequent degradation by Rrp44. A number of conserved residues are found at the base of the central DExH core within Mtr4 and Ski2 and on the upper surface of the exosome core, suggesting that they might constitute a mutual-binding interface. Nevertheless, additional factors are required for a stable interaction between the RNA helicases and the exosome core. Mtr4 directly interacts with the exosome-associated protein Mpp6 , which may promote binding with the exosome core. Furthermore, hRRP6 is required for stable interaction between hMTR4 and the exosome core . The interaction between the yeast Ski complex and the exosome is dependent on another factor, Ski7 .
A conformational switch in Rrp44
Recent structural analyses resolve two distinct structural conformations of Rrp44 bound to the exosome core complex  (see Figure 2). In the presence of short structured RNAs, Rrp44 that is associated with the exosome core has an extended structure and the substrate can directly access either the exonuclease domain or the PIN domain. In this state, the CSD2 domain of Rrp44 is positioned at the RNA exit site at the bottom of the core, blocking the pathway of the substrate through the core. In the presence of longer RNAs, Rrp44 undergoes a substantial conformational change and becomes more compact. The exonuclease domain is rotated ~120°, bringing it close to the RNA exit site and displacing the CSD2 domain. In this conformation, Rrp44 is able to digest RNA that is threaded through the channel. These data suggest that Rrp44 bound to the exosome core can switch between an extended conformation that is accessible to direct substrate binding, and a compact conformation that is aligned for interaction with threaded substrates.
Exosome substrate targeting pathways
The path(s) taken by RNA to reach Rrp6 is less clear. An Rrp6 mutant lacking the C-terminal exosome-interaction domain is functional in vivo, indicating that at least some substrates can bind to Rrp6 independently of the exosome complex . In the context of the exosome complex, the RNA-binding activities of Rrp6 and Rrp44 are interconnected. The interaction between Rrp6 and the exosome core causes an increased RNA binding activity of Rrp44. Conversely, the presence of a substrate that is locked into the threaded pathway, by mutation of the Rrp44 exonuclease active site, decreased the RNA-binding activity of Rrp6 . It was suggested that Rrp6 substrates are threaded into the central channel of the exosome core and then diverted laterally to Rrp6. However, recent structures of the yeast exosome complex do not support such a pathway . Alternatively, the interrelated RNA-binding activities of exosome-associated Rrp6 and Rrp44 could reflect an allosteric mechanism that is mediated through the exosome core. Thus commitment of substrates to the threaded pathway to Rrp44 might suppress access of substrates to the catalytic domain of Rrp6, thereby ensuring that a single exosome complex can only be engaged with a substrate targeted to one or the other enzyme (see Figure 2).
Long and short pathways for distinct RNAs
Deletion of the exosome-interacting domain of Rrp6 does not impair its function in box C/D snoRNA 3′-end maturation or the trimming of the ‘5.8S+30’ processing intermediate, but does cause a defect in discard pathways that degrade the 5′-ETS fragment and truncated degradation intermediates of 5S rRNA . Furthermore, in vivo cross-linking studies have shown that Rrp44 and Rrp6 bind directly to 7S pre-rRNA and snoRNA processing intermediates, but very few contacts were observed with exosome core components . These data suggest that substrates that undergo limited trimming reactions are normally targeted directly to Rrp44 or Rrp6, whereas at least some substrates destined for complete degradation are dependent on threading through the exosome core. It is unclear whether Rrp6 can efficiently degrade its other substrates, such as pre-mRNAs, CUTs/SUTs (stable unannotated transcripts), tRNAs and snRNAs (small nuclear RNAs), when uncoupled from the exosome core.
Strikingly, the subset of Rrp6 substrates that accumulate in rrp6/mpp6 double mutants comprise those that are destined for complete degradation (CUTs, pre-mRNAs, 5′-ETS, truncated stable RNAs and snoRNA transcription read-through products) and exclude those that undergo limited 3′-end processing reactions (5.8S rRNA, box C/D snoRNAs) [61,62]. Given that Rrp6 substrates targeted to degradation are dependent on the interaction between Rrp6 and the exosome core, this suggests that the synthetic lethal relationship between loss-of-function rrp6 and mpp6 alleles  might reflect redundant functions of Rrp6 and Mpp6 in facilitating substrate threading into the exosome core. This model is supported by the direct physical interaction reported between hMPP6 and hMTR4  and the observation that the stable interaction between hMTR4 and the exosome core is dependent on hRRP6 .
Although the exosome was initially characterized through its role in the 3′-end processing of 5.8S rRNA , the overwhelming majority of exosome substrates are targeted to RNA discard pathways [60,63]. The exosome is able to degrade extensively structured pre-ribosomal RNP particles in a highly processive manner. It is not clear how substrates destined for limited processing could be effectively differentiated from those that are targeted to discard pathways if they were both engaged in a highly processive mode of degradation. One function proposed for the endonuclease activity of Rrp44 is to release substrates that are stalled in the exosome channel, thereby salvaging blocked exosome complexes . The endonuclease activity of Rrp44 is not, however, required for exosome-mediated RNA processing events in wild-type cells. Thus such a role presumably represents a fail-safe recycling function rather than an obligatory step in RNA processing.
Which RNAs might be degraded by Rrp44 using direct core-independent pathways, and how would these RNAs be targeted without threading through the exosome core? One major class of such substrates might be incorrectly folded tRNAs, which can be targeted directly to Rrp44 [65,66]. Notably, tRNA represents a significant proportion of the RNA subjected to cellular RNA discard pathways . Other small, structured RNAs targeted to discard pathways may well share this route. Such substrates are likely to require 3′-end tailing by TRAMP  or another nucleotidyltransferase  to facilitate efficient docking with Rrp44. A major current objective is to define the set of RNAs that are threaded through the exosome core and those that are targeted directly to either Rrp44 or Rrp6.
RNA UK 2014: An Independent Meeting held at Low Wood Hotel, Windermere, U.K., 24–26 January 2014. Organized and Edited by Niki Gray, Gracjan Michlewski and Steve West (University of Edinburgh, U.K.).
cryptic unstable transcript
5′-external transcribed spacer
helicase and RNase D C-terminal
human Rrp6 protein
small nucleolar RNA
small nucleolar ribonucleoprotein
complex, Trf4–Air2–Mtr4 polyadenylation complex