Targeting the nuclear RNA exosome: Poly(A) binding proteins enter the stage

ABSTRACT

Centrally positioned in nuclear RNA metabolism, the exosome deals with virtually all transcript types. This 3′–5′ exo- and endo-nucleolytic degradation machine is guided to its RNA targets by adaptor proteins that enable substrate recognition. Recently, the discovery of the ‘Poly(A) tail exosome targeting (PAXT)’ connection as an exosome adaptor to human nuclear polyadenylated transcripts has relighted the interest of poly(A) binding proteins (PABPs) in both RNA productive and destructive processes.

Introduction

To manage the constant and pervasive production of RNA, eukaryotic nuclei rely heavily on RNA degradation systems. These aid in the processing of mature RNA from precursors and in the removal of transcriptional and RNA processing by-products, transcripts produced in excess and otherwise nuclear retained RNA. A chief player here is the 3′–5′ exo- and endo-nucleolytic RNA exosome, conserved in all studied eukaryotes.^1-5 This multisubunit protein complex resides in both the nucleus and the cytoplasm, where it deals with most known RNA biotypes.⁶ The nuclear exosome, relevant for this article, is composed of an inert doughnut-shaped core of 9 proteins (EXO-9) and achieves its catalytic activity from associated ribonucleases; the exonuclease hRRP6/Rrp6p and the exo- and endo-nuclease hRRP44 (DIS3)/Rrp44p, positioned on opposite sides of the central core channel (Fig. 1).

Figure 1.

RNA exosome adaptors in S. cerevisiae and human nuclei. Principle overview of the main factors interacting with, and providing target specificity to, the exosome in S. cerevisiae (left) and human (right) nuclei (see text for details). hMTR4-Mtr4p (green) offers a main anchor linking the exosome to its different adaptors (indicated by black arrows). Question marks indicate possible interactions of Nsa1p and PICT1, UTP18 and DGCR8 with Mtr4p and hMTR4, respectively, motivated by protein homologies. Localization and function of C1D and MPP6 in the human nucleoplasm remain an open question (as question marks indicate). Note that proteins with the symbol ‘p’ derive from S. cerevisiae.

Although, the exosome might target some of its substrates directly, available data suggest that it uses various RNA-binding adaptor complexes. Indeed, a functional scheme with a ‘ready-to-use’ RNA decay machine connected to a suitable adaptor, providing substrate specificity, is emerging as a general theme. In line with this, we recently described a new human exosome adaptor, which targets polyadenylated nuclear RNAs. Consequently, we coined this adaptor the ‘poly(A) tail exosome targeting (PAXT) connection’.⁷ Here we discuss organizational strategies for exosomal targeting of nuclear RNA, possible biologic implications of the PAXT connection and its relation to fellow nuclear RNA decay pathways.

Organization of exosome co-factors and adaptors

In all species studied, proper function of both nuclear and cytoplasmic exosomes critically depends on RNA helicase activities.⁸ Human examples of such ATP-dependent exosome-associated helicases are the cytoplasmic and nuclear proteins SKI2 (SKIV2L) and hMTR4 (SKIV2L2), respectively. hMTR4/Mtr4p is a DExH superfamily 2 (SF2) helicase conserved from yeast to man and its proposed role is to unwind secondary structure and facilitate the injection of RNA substrate into the channels of the nuclear exosome.^9-13 However, besides this key activity, hMTR4/Mtr4p also provides a universal anchor for nuclear exosome adaptors (Fig. 1). The N-terminal domain of hMTR4/Mtr4p is responsible for exosome interaction, in particular via hRRP6/Rrp6p and C1D/Rrp47p, an hRRP6/Rrp6p co-factor.^14,15 This is for example relevant within nucleoli of human cells, where hRRP6 and C1D participate in pre-rRNA processing.¹⁶ Like C1D/Rrp47p, the M-phase phosphoprotein 6 protein, MPP6/Mpp6p, appears to serve as a general exosome co-factor,^17-19 but although C1D, MPP6/Mpp6p and Mtr4p display affinities for RNAs harbouring poly(G), poly(C/U) and poly(A) tracts, respectively,^9,16-18 it is unclear whether this directly guides the exosome to some of its RNA targets or whether it reflects supportive roles for the exosome in, for example, substrate organization or unwinding.

Recently, other proteins have emerged as clear cases of direct adaptors between substrate RNA and the nuclear exosome through their interactions with hMTR4/Mtr4p. In an apparent simple arrangement, the S. cerevisiae RNA-binding proteins, Nop53p and Utp18p, bind the so-called arch domain of Mtr4p, facilitating 5.8S rRNA processing and the degradation of 5′external transcribed spacer (5′ETS) RNAs, respectively (Fig. 1).²⁰ In particular, Nop53p cross-links close to the 3′end of mature 5.8S rRNA on its 35S pre-rRNA precursor and adjacent to an Mtr4p cross-linking site, which is lost upon Nop53p depletion. Whether the human counterparts of Nop53p and Utp18p, PICT1 and UTP18, harbour similar hMTR4 interaction motifs and functions remains to be investigated. Another nucleolar example is the hMTR4-interacting protein WDR74, which is somewhat similar to the S. cerevisiae ribosome biogenesis factor Nsa1p and suggested to recruit hMTR4 and the exosome to 12S rRNA for its processing into 5.8S rRNA.²¹ Finally, the human RNA binding protein DGCR8 is suggested to target the nucleolar exosome to the degradation of mature snoRNA and telomerase RNA (hTR).²² However, an involvement here of hMTR4 for exosomal anchoring has not been directly analyzed.

In the nucleoplasm of human cells, a common organization of 2 exosome adaptors has been described in our laboratory. In both instances, an RNA-binding protein, believed to provide target specificity to the exosome, connects to hMTR4 via a large bridging protein (Fig. 1). In case of the trimeric nuclear exosome targeting (NEXT) complex, the zinc-finger protein ZCCHC8 bridges hMTR4 to RBM7, which binds RNA rather promiscuously, reflecting the broad substrate range of NEXT toward early unprocessed transcripts.^7,23-26 In case of the PAXT connection, the zinc-finger protein ZFC3H1 bridges hMTR4 to the nuclear poly(A) binding protein PABPN1, which primarily binds polyadenylated RNA species,^7,27,28 Noteworthy, however, the tight MTR4-ZFC3H1 dimer interacts with PABPN1 in a more transient and partly RNA-dependent manner, suggesting a role for yet unknown factors in yielding an active PAXT connection. In contrast, the proteins of the hMTR4-ZCCHC8-RBM7 trimer can be co-precipitated in virtually equal stoichiometries.²⁶ As a possible way of engaging NEXT and PAXT with certain capped RNAs, both assemblies physically associate with components of the cap binding complex (CBC) and its associated factors ARS2 and ZC3H18.^7,23

A particularly complex architecture of an exosome linkage to its RNA substrate is reflected by the 3-way organization of the Trf4p-Air1p/2p-Mtr4p polyadenylation (TRAMP4),^29-31 Nrd1p-Nab3p-Sen1p (NNS)³² and exosome complexes in the S. cerevisiae nucleoplasm (Fig. 1). Here, the NNS complex is recruited to short sequence motifs (GUAA/G and UCUUG recognized by Nrd1p and Nab3p, respectively) in the nascent RNA.³² This elicits transcription termination, through the helicase activity of Sen1p, and mediates a direct interaction between Nrd1p and the RNA exosome, mediated by TRAMP4. An interaction which establishes a ‘hand-over’ of substrate to the exosome as exemplified by its complete degradation of cryptic unstable transcript (CUTs) or its 3′end processing of stable sn(o)RNA.^33-38 A functional homolog of the NNS complex appears to be lacking in mammalian cells, but the above mentioned ARS2 protein might serve a similar role in linking transcription termination to exosome activity.^23,39

A distinct Trf5p-Air1p-Mtr4p polyadenylation (TRAMP5) complex also exists in S. cerevisiae nuclei (Fig. 1).^40-42 In addition to the substrates mentioned above, TRAMP4/5 complexes aid the exosome in the degradation of 23S pre-rRNA, aberrant 5S rRNA forms and hypomethylated initiator tRNAMet as well as in the processing of diverse pre-rRNA species.^43-45 While Trf4p and Trf5p endow TRAMP4 and TRAMP5 with poly(A) polymerases activity, the Air1p and Air2p zinc-knuckle proteins bind RNA. Together with the helicase activity of Mtr4p, these capacities of TRAMP4/5 are suggested to aid in the exosomal degradation of structured substrates, which may require subsequent rounds of 3′adenylation by Trf4/5p to reach completion.⁴⁶ In this sense, TRAMP complexes may be best described as exosome co-factors, still requiring adaptors for proper RNA targeting. Except for the NNS complex connection of TRAMP4, it is not clear how TRAMP4/5 complexes are recruited to transcript substrates.

TRAMP structural organization and activity is conserved in S. pombe and in human cells through the Mtr4-Air1-Cid14 and hMTR4-ZCCHC7-PAPD5/7 complexes, respectively.^26,47-49 In these species, TRAMP complexes appear to largely reside in nucleoli, serving the exosome in the processing and degradation of pre-rRNA species. However, in S. pombe, TRAMP components also participate in the degradation of heterochromatic transcripts via the recruitment of the exosome and the RNA interference (RNAi) machinery.⁴⁷ Moreover, some hMTR4-dependent exosome adaptor schemes are conserved in S. pombe, where, the hMTR4-related protein, Mtr4-like protein 1 (Mtl1), interacts with the ZFC3H1-homologous Red1 protein to form a scaffold coined the Mtl1-Red1 core (MTREC) complex.^50-52 MTREC further associates with other proteins, including the S. pombe PABPN1 homolog Pab2, suggesting the existence of a PAXT-like activity. ZCCHC8 and RBM7 do not appear to be conserved in S. pombe, whereas the A. thaliana hMTR4-like helicase, HEN2, localizes to the nucleoplasm and co-purifies with 2 proteins related to human ZCCHC8 and one protein related to RBM7.^53,54 Continued exploration of exosome-targeting components and strategies across eukaryotic species will reveal how nuclear exosomes achieve their substrate specificities and how changing environmental conditions may balance the use of different adaptor complexes according to substrate abundances and cellular needs.

Poly(A) binding proteins (PABPs) in RNA metabolism

Eukaryotic PABPs are both nuclear and cytoplasmic at steady-state, but often shuttle between these compartments.⁵⁵ The recent description of PAXT highlights the versatility of nuclear PABPs across species. In eukaryotic nuclei, multiple PABPs cover the poly(A) tail, which reaches a length of ∼80–90 nt in S. cerevisiae and ∼250 nt in human cells. In addition, PABPN1, and possibly also the S. cerevisiae nuclear PABP Nab2p, bind the nascent tail during its synthesis.^56-61 Notably, mechanistic differences may exist, as Nab2p is not a homolog of PABPN1, which seems to be absent in S. cerevisiae. Regardless, these PABP-scaffolds on nuclear poly(A) tails were proposed to serve a protective function, shielding the newly produced RNA against nucleolytic attack while directing RNAs with any business in the cytoplasm for nuclear export.^62,63 It is now clear, however, that PABPs may also invoke RNA degradation by directly recruiting ribonucleolytic activities.

In S. cerevisiae, Nab2p protects newly synthesized mRNA against exosomal degradation, likely during the distributive phase of the polyadenylation reaction where the growing tail is most vulnerable.⁵⁹ Significantly, Nab2p also connects to mRNA export factors,⁶³ perhaps further aiding in moving the transcript along its productive path and escaping nuclear decay. However, for transcripts with longer nuclear residence times, like certain pools of pre-mRNA, Nab2p appears to attract the exosome machinery, causing transcript turnover.⁶⁴ Such double-faced activity of a PABP, pending the maturation level of the RNA, might provide an in-built nuclear timer ensuring that transcripts failing to exit the nucleus in a timely manner will be eliminated (Fig. 2).⁶⁵ PABPN1 holds similar properties as Nab2p; it binds the nascent poly(A) tail,^56,57,66 it is implicated in RNA export⁶² and it participates in nuclear RNA decay.⁵⁵ The latter activity was suggested before the discovery of the PAXT connection, when it was described that PABPN1 and the poly(A) polymerases α and γ contribute to nuclear exosome degradation of RNA via the so-coined ‘PABPN1 and PAP-mediated RNA decay (PPD)’ pathway.^27,28,67 We suggest that physical interactions demonstrated by the PAXT connection explain, at least part of, the biology of the PPD pathway. However, it should be kept in mind that nuclear RNA decay pathways are still insufficiently characterized and that central factors might engage with more than one type of nucleolytic activity (see below). Regardless, it seems plausible that the concept of one individual PABP affecting nuclear levels of polyadenylated RNA both positively and negatively is conserved between S. cerevisiae and human cells.

Figure 2.

PABPs provide timing of RNA nuclear residence. Proceeding their poly(A) tail recognition by nuclear PABPs (red balloons), processed RNAs are exported to the cytoplasm (indicated by green light). In contrast, export-impaired or -delayed RNAs are turned over (red light). Thus, PABPs ‘sense’ the nuclear exposure time (represented by a the timer) of polyadenylated RNAs and provide the platform for nuclear RNA decay.

Surprisingly, the S. pombe homolog of PABPN1, Pab2, is not involved in the RNA polyadenylation reaction, perhaps being obsolete due to the short size of S. pombe poly(A) tails (∼40 nt) and/or the high processivity of the Pla1 poly(A) polymerase in fission yeast.^56,68,69 Yet, Pab2 does engage in nuclear exosome-related activities like the degradation of meiosis-specific transcripts and unspliced pre-mRNAs via MTREC^51,70,71 as well as the 3′end trimming of extended snoRNA precursors.⁷² Instead, an RNA protective function was recently suggested for the fission yeast homolog of S. cerevisiae Nab2, spNab2, which competes with Pab2 for specific polyadenylated transcripts to prevent exosome-mediated decay.⁷³ This highlights the fact that genomes of studied model organisms encode more than just a single PABP and that these may have redundant or even opposing functions; both Pab2 and spNab2 deleted S.pombe cells are viable and produce mRNAs with normal poly(A) tail lengths, motivating the finding that the fission yeast homolog of S. cerevisiae Pab1, the cytoplasmic spPabp, complements essential functions of Pab2- and spNab2-null cells.⁷⁴ Similarly, overexpression of the normally cytoplasmic Pab1p rescues the viability of genetic or conditional Nab2p depletion in S. cerevisiae and Pab1p likely also binds to nascent poly(A) tails.^59,63,75 Finally, the zinc-finger protein ZC3H14, a human homolog of Nab2p, was recently ascribed an RNA protective function,⁷⁶ and like Nab2p, ZC3H14 was suggested to control poly(A) tail length and modulate processing of pre-mRNAs.⁷⁷

An additional degradative role in RNA maturation has also recently been ascribed to PABPN1. Interestingly this occurs in a PAXT-independent manner and instead of the nuclear exosome involves the poly(A)-specific ribonuclease PARN.^78-80 A particularly well-characterized substrate for this pathway is the hTR, which uses the canonical poly(A) polymerases α and γ, PABPN1 and PARN for its maturation - in their absence mature hTR forms are significantly decreased.⁷⁹ Consistent with PABPN1 activity outside of the PAXT connection, depletion of the PAXT subunit ZFC3H1 has no effect on mature hTR levels (Fig. 3A). The RNA exosome does nevertheless take part in hTR biology in that it aids in the degradation of 3′extended hTRs derived from imperfect 3′end processing events.⁸⁰ However, in this case the NEXT complex is engaged (Fig. 3B), consistent with targeting of these RNA species not being dependent on prior polyadenylation.

Figure 3.

Processing and degradation of hTR. Screen shots of the hTR gene and RNAseq data from.⁷ (A) hTR maturation depends on the PAXT-independent poly(A) tail recognition by PABPN1 and the ensuing 3′end trimming by PARN (see text for details). Accordingly, depletion of PABPN1 (‘PABPN1 Kd’), but not ZFC3H1 (‘ZFC3H1 k_d’), decreases levels of mature hTR. Orange dashed lines indicate the maximum RNA signal levels in PABPN1 depletion conditions. (B) Extended hTR 3′ends (highlighted by red box) of impaired hTR precursors are targets for NEXT-dependent exosome degradation in the nucleoplasm.

The ability of PABPN1 to engage in RNA productive and destructive processes raises several questions. While nuclear residence time may well determine whether poly(A) tail-bound PABPN1 gets a chance to associate with decay activity or not, it is less clear what controls whether PABPN1 associates with the exosome or with PARN. Does this decision govern whether the transcript is targeted for processing or complete decay? Also, how do functional RNAs requiring nuclear stability counteract nucleolytic targeting? Perhaps they avoid polyadenylation altogether, as exemplified by sn(o)RNAs or by creating structures incompatible with PABPN1 binding or exosomal activity. Clearly more research is required to detail aspects of PABP function in controlling cellular poly(A) RNA levels. Still, our realization that these proteins function dually in production and decay provides a biologic framework in which this can be done.

Disclosure of potential conflicts of interest

No potential conflicts of interests were disclosed.

Acknowledgments

We thank Agnieszka Tudek, Manfred Schmid, Michal Lubas and Domenico Libri for fruitful comments on this article.

Funding

This work was supported by the ERC (grant 339953), the Danish National Research Council, the Lundbeck- and the Novo Nordisk-Foundations.