Polyadenylation and nuclear export of mRNAs

Abstract

In eukaryotes, the separation of translation from transcription by the nuclear envelope enables mRNA modifications such as capping, splicing, and polyadenylation. These modifications are mediated by a spectrum of ribonuclear proteins that associate with preRNA transcripts, coordinating the different steps and coupling them to nuclear export, ensuring that only mature transcripts reach the cytoplasmic translation machinery. Although the components of this machinery have been identified and considerable functional insight has been achieved, a number of questions remain outstanding about mRNA nuclear export and how it is integrated into the nuclear phase of the gene expression pathway. Nuclear export factors mediate mRNA transit through nuclear pores to the cytoplasm, after which these factors are removed from the mRNA, preventing transcripts from returning to the nucleus. However, as outlined in this review, several aspects of the mechanism by which transport factor binding and release are mediated remain unclear, as are the roles of accessory nuclear components in these processes. Moreover, the mechanisms by which completion of mRNA splicing and polyadenylation are recognized, together with how they are coordinated with nuclear export, also remain only partially characterized. One attractive hypothesis is that dissociating poly(A) polymerase from the cleavage and polyadenylation machinery could signal completion of mRNA maturation and thereby provide a mechanism for initiating nuclear export. The impressive array of genetic, molecular, cellular, and structural data that has been generated about these systems now provides many of the tools needed to define the precise mechanisms involved in these processes and how they are integrated.

Keywords: RNA binding protein; RNA helicase; RNA splicing; gene expression pathway; nuclear pore; nuclear posttranscriptional modification; nuclear transport; polyadenylation; ribonuclear protein (RNP); spliceosome.

Figure 1.

Overview of nuclear transport pathways for macromolecules. These pathways are all based on a thermal ratchet mechanism in which energy is used to rectify Brownian motion by mediating the assembly of cargo:carrier complexes in the donor compartment and their disassembly in the acceptor compartment. Proteins and small RNAs are transported using karyopherin-based carriers to mediate movement through nuclear pores. In nuclear protein import (A) karyopherins (Kap; yellow) such as importin-β bind their cargoes (green) in the cytoplasm (often employing an adapter such as importin-α), and then, when the cargo:carrier complex reaches the nucleus, RanGTP (red) binding to the karyopherin dissociates the cargo, after which the karyopherin:RanGTP complex returns to the cytoplasm where the RanGTPase is activated by RanGAP, generating RanGDP that dissociates from the karyopherin, freeing it for a further import cycle. The RanGDP is then recycled to the nucleus where RanGEF (RCC1, Prp20) recharges it with GTP. The export of proteins and small RNAs (B) is mediated by an analogous pathway, except that here the cargo binds to the karyopherin (such as Crm1) complexed with RanGTP in the nucleus and is released in the cytoplasm following GTP hydrolysis. The export of mRNAs (C) employs a different pathway that uses the Mex67:Mtr2 (NXF1:NXT1) complex (Fig. 3) as a carrier to which binding in the nucleus and release in the cytoplasm are mediated by DEAD-box helicases (Sub2 and Dbp5, respectively) that hydrolyze ATP to remodel the mRNP. An additional feature of mRNA export is that it is necessary for the nuclear steps of the gene expression pathway to have been completed so that only fully matured transcripts are exported for translation in the cytoplasm. The machinery involved in these steps is summarized in Fig. 2.

Figure 2.

Outline of the gene expression pathway from transcription to translation for budding yeast. Transcripts are modified by the addition of 5′ caps and 3′ poly(A) tails and the splicing out of introns if present before a structural rearrangement, mediated by Sub and the TREX complex, removes Yra1 and attaches Mex67:Mtr2 to generate an export-competent mRNP to which Nab2 is also bound. The precise nature of the structural rearrangement remains to be established but may involve the generation of RNA hairpins that bind more strongly to the transport factor. The mRNP can then diffuse back and forth through nuclear pores as a result of interactions between Mex67:Mtr2 and FG-nucleoporins overcoming the barrier function of the pore. At the cytoplasmic face of the pore, the DEAD-box helicase Dbp5, working in conjunction with Nup42 and Gle1, remodels the RNA to release Mex67:Mtr2 and Nab2, thereby preventing the mRNA from returning to the nucleus. Pab1 also replaces Nab2 on the poly(A) tail. Although this sequence of processing steps tends to resemble a production line, the steps may not necessarily occur in a defined sequence, and nuclear export, the culmination of the nuclear phase of the gene expression pathway, appears to only require that all steps have been completed successfully. Metazoans have an analogous pathway but differ in some details, primarily in the addition of the EJC immediately after the 3′ splice site when exons are joined. In metazoans, the first EJC together with the 5′ cap facilitates the binding of the TREX complex and subsequent attachment of the Mex67:Mtr2 homologue NXF1:NXT1. IP6, inositol hexaphosphate; Pol II CTD, polymerase II C-terminal domain.

Figure 3.

Mex67:Mtr2/NXF1:NXT1 structure and interaction with CTE-RNA. A, domain structure of the S. cerevisiae Mex67:Mtr2 complex. The metazoan homologue, NXF1:NXT1 (also called TAP:P15), has a similar structure but has an additional unstructured arginine-rich domain at its N terminus. Mex67 contains four structural modules (RRM domain, LRR domain, NTF2-like domain, and UBA domain) that are connected by flexible linkers. Mtr2 also has an NTF2-like fold and binds to the Mex67 NTF2-like domain to form a heterodimer. The RRM, LRR, and NTF2-like domains bind to RNA, whereas the NTF2-like and UBA domains interact with FG-nucleoporins (based on Protein Data Bank (PDB) codes 1OAI and 4WWU). B, complex formed between viral CTE-RNA and the RRM and LRR domains of TAP (the human homologue of Mex67) showing the secondary structure of the RNA (based on PDB code 3RW6) that forms a bent hairpin (blue). Similar RNA secondary structures may be generated in the nucleus by helicases to facilitate the binding of Mex67:Mtr2 to generate export-competent mRNPs.

Figure 4.

Nab2 structure and its binding to poly(A) RNA. A, schematic illustration of the domains present in Nab2. The N-terminal Nab2N domain is essential and interacts with Mlp1 and Gfd1. The RGG domain also contains the nuclear localization sequence that is recognized by transportin to return Nab2 to the nucleus after an export cycle. There are seven Zn fingers arranged in three groups (ZnF1+2, ZnF3+4, and ZnF5–7). The fingers within each group interact so that they have a defined orientation to one another. B, arrangement of Nab2 Zn fingers 5–7 that impairs binding of a single poly(A) RNA chain to all three simultaneously (based on PDB code 5L2L). C, dimerization of Nab2 Zn fingers 5–7 brought about by binding A₁₁G RNA (54). The RNA binds to both Nab2 chains to generate the dimer, and it is likely that similar dimers can be formed with full-length Nab2 in vivo (based on PDB code 5L2L).

Figure 5.

Model for the regulation of poly(A) tail length by Nab2 in yeast. Poly(A) tail length in S. cerevisiae appears to be regulated by a mechanism analogous to that proposed for the way in which PABPN1 functions in metazoans (66, 67) to terminate polyadenylation by dissociating poly(A) polymerase from the cleavage and polyadenylation machinery (CPSF) so that it ceases to be processive. Each PABPN1 binds to ∼12–15 adenines, and when a sufficient number have been added, they form an approximately spherical aggregate that is proposed to stiffen the poly(A) tail, forcing poly(A) polymerase to dissociate from the CPSF, after which it ceases to be processive and dissociates from the poly(A) tail, terminating polyadenylation. An analogous model for polyadenylation termination in S. cerevisiae envisages that following cleavage, poly(A) polymerase (Pap1) synthesizes the poly(A) tail processively but only while it remains attached to the polyadenylation factor (CPF) through binding to Fip1 and Cft1 (for simplicity, the cleavage module has been omitted from the figure). Pap1 holds the growing poly(A) tail at one end while Rna15 and Cft1 hold its other end so that it forms a loop that can be accommodated by the flexibility of both the poly(A) chain and Fip1 (A). However, when a sufficient number of adenosines (probably ∼60) have been added to facilitate Nab2 binding, the reduction in flexibility of the poly(A) tail produced by Nab2 dimerization introduces a stress that detaches Pap1 from Cft1 and Fip1 (B). Once detached from Fip1, the processive activity of Pap1 is impaired and it dissociates from the tail, thereby terminating polyadenylation. The termination of polyadenylation then appears to be signaled to the TREX complex via CF1A component Pcf11 to initiate the formation of an export-competent mRNP. The highly schematic representation of the CF1A and polymerase modules of CPF is based on the recent cryo-EM structure of this complex (65).