Biochemical Society Transactions

RNA UK 2010

3′ uridylation and the regulation of RNA function in the cytoplasm

Chris J. Norbury


Degradation of cytoplasmic mRNAs is an important aspect of the regulation of gene function in eukaryotes. Much of what is currently known about the underlying pathways of mRNA decay is derived from studies of the budding yeast Saccharomyces cerevisiae, in which mRNA turnover is initiated by deadenylation, followed either by decapping and 5′→3′ degradation or by further 3′→5′ exonucleolysis. Our studies using RNA cRACE (circularization-based rapid amplification of cDNA ends) techniques indicate that mRNA decapping in the fission yeast Schizosaccharomyces pombe often does not require prior deadenylation. Furthermore, the poly(A) polymerase-related, cytoplasmic enzyme Cid1 catalyses uridylation of a variety of functionally diverse poly(A)+ mRNAs and hence stimulates decapping as part of a novel mRNA turnover pathway. The pathways initiated by uridylation and deadenylation stimulate decapping in a partially redundant fashion, but urg1 mRNA is stabilized in mutants lacking cid1. Accumulation of uridylated RNAs in an lsm1 mutant suggests an involvement of the Lsm1–7 complex in recognition of the 3′ uridylation tag and recruitment of the decapping machinery. Recent reports from other groups suggest that in metazoans, which unlike budding yeast contain Cid1 orthologues, 3′ uridylation by such enzymes is used to regulate miRNA (microRNA) and siRNA (small interfering RNA) biogenesis and activity. It has further been suggested that uridylation is an important regulatory modification of non-polyadenylated replication-dependent histone mRNAs. This modification may also form the basis of a widespread mechanism for the initiation of the decay of polyadenylated mRNAs in organisms other than fission yeast.

  • Cid1
  • fission yeast
  • mRNA turnover
  • terminal U-transferase


The steady-state level of any given mRNA, and hence its biological activity, is clearly a function both of its rate of synthesis and of its rate of decay, and in numerous instances mRNA turnover in the cytoplasm has been identified as an important level of regulation of gene expression [1,2]. Much of what is known about the mechanisms and regulation of cytoplasmic mRNA decay has come from studies of the budding yeast Saccharomyces cerevisiae, in which exonucleolytic removal of most of the poly(A) tail is the initiating event [3]. The residual oligo(A) ‘stump’ is no longer able to maintain interaction of the mRNA with the PABP [poly(A)-binding protein], an important activator of translation, but instead binds the seven-membered Sm-like complex Lsm1–7, which in turn recruits the decapping enzyme heterodimer Dcp1–Dcp2. The subsequent enzymatic removal of the 5′ cap structure serves to suppress translation further and to allow access of 5′→3′ exonucleases, and hence degradation of the mRNA from either or both ends. Deadenylation undoubtedly also plays an important role in other eukaryotes, but the extent to which this is generally rate-limiting for mRNA turnover is not yet apparent. Recent studies from my laboratory have focused on mRNA decay in the fission yeast Schizosaccharomyces pombe, which in evolutionary terms is approximately as distant from S. cerevisiae as it is from metazoans [4]. Surprisingly, removal of the poly(A) tail does not seem to be a precondition of mRNA decapping for some (and, by extension, many) fission yeast mRNAs. In addition, fission yeast possess an additional, parallel decay pathway that involves modification of the target mRNA by 3′ uridylation, rather than deadenylation, as a prerequisite for decapping.

Pathways of mRNA deadenylation

The Ccr4–Not complex, which has diverse roles in transcriptional regulation and RNA turnover, plays a major role in removal of poly(A) tails in S. cerevisiae, where initial mean tail lengths are of the order of 70 nucleotides long [5]. In metazoan cells the initial mean tail length is rather longer, approx. 200 nucleotides, and deadenylation has two distinct phases. First, the poly(A)-selective exonuclease Pan2-Pan3 [which in budding yeast instead acts mainly in the nuclear trimming of newly synthesized poly(A) tails] reduces tails to approx. 50 nucleotides, with the remaining deadenylation being performed either by the Ccr4–Not complex, as in the budding yeast model, or by PARN [poly(A)-specific ribonuclease] [6,7]. PABP is required for efficient recruitment of Pan2-Pan3 to mRNA 3′ ends [8], but conversely antagonizes the exonucleolytic activity of the Ccr4–Not complex. Once deadenylation has reduced the tail to 10–15 nucleotides, PABP is no longer able to protect the mRNA from recognition by Lsm1–7 and subsequent decapping. In metazoans, RNA-binding proteins that recognize specific regulatory elements in the 3′ UTR (untranslated region) have been shown to mediate recruitment of the Ccr4–Not deadenylase to mRNA 3′ ends [9], or to determine the balance between PARN-dependent deadenylation and cytoplasmic polyadenylation [10]. In budding yeast mRNA 3′ UTRs are less complex, and this mode of mRNA regulation appears to be correspondingly less prevalent.


As deadenylation proceeds, there is presumably competition between Lsm1–7 and PABP for shortening cytoplasmic mRNA 3′ ends; although the Lsm1–7 heptamer preferentially binds short tails in vitro, this aspect of the decay pathway is far from fully understood. In any event, Lsm1–7 recruitment serves to promote mRNA decapping, by inhibiting further 3′→5′ degradation by the cytoplasmic exosome and by direct recruitment of the Dcp1–Dcp2 heterodimer [11]. The subcellular location of the decapping process has been the subject of much debate. While the presence of decapping factors in cytoplasmic foci termed p-bodies (processing bodies) suggested that these represented dedicated decapping sites to which mRNAs might be transported [12], recent work in S. cerevisiae indicates instead that decapping takes place on polysomes [13]. Recruitment of Lsm1–7, decapping and subsequent 5′→3′ degradation of mRNAs that are still in the process of being translated would minimize the risk of ‘run-off’ translation of mRNAs that had suffered 3′→5′ degradation extending into the open reading frame.

mRNA decay in fission yeast

We have used a cRACE (circularization-based rapid amplification of cDNA ends) strategy to investigate the sequence of events in mRNA decay in fission yeast [4]. This approach allows the selective cloning and sequencing of the 5′ and 3′ ends of individual mRNA molecules, as well as discrimination between those mRNAs with an intact 5′ cap and those that have undergone decapping. The technique relies on the use of T4 RNA ligase to circularize RNA molecules that have 5′ phosphate and 3′ hydroxy termini, followed by nested RT (reverse transcription)–PCR using primers corresponding to sequences on either side of the ligated junction. Decapped mRNAs have the necessary 5′ phosphate, so will be selectively circularized in the absence of any further treatment. Conversely, capped mRNA sequences can be isolated by first removing free 5′ phosphates from decapped species using alkaline phosphatase, and then using tobacco acid pyrophosphatase to cleave the cap structure in such a way as to leave a 5′ phosphate suitable for ligation.

Analysis of many hundreds of cRACE products has allowed us to build up a detailed picture of mRNA turnover in fission yeast. Our efforts were initially focused on transcripts of the α-actin gene act1, a housekeeping gene with a level of expression sufficient to make the cRACE approach relatively straightforward. Surprisingly, the length distribution of poly(A) tails of decapped act1 mRNAs was indistinguishable from that of capped transcripts [4]. Extension of this analysis to several other genes showed a varying level of dependence of decapping on deadenylation; the mRNAs of some genes, including pof9 and urg1, exhibited the same lack of deadenylation seen with act1, whereas others, such as hcn1, showed a pattern of deadenylation prior to decapping resembling that previously described in budding yeast. We conclude that, for many transcripts in Schizosaccharomyces pombe, deadenylation is not a prerequisite for decapping. The co-existence of deadenylation-dependent and deadenylation-independent mRNA turnover pathways could confer on fission yeast the possibility of differential regulation of distinct mRNA subpopulations that are channelled preferentially into one or other of these pathways.

mRNA uridylation in fungi and human cells

The polyadenylated mRNAs of all six S. pombe genes analysed by the cRACE approach exhibited some level of 3′ uridylation, a modification we had previously observed on act1 transcripts and had attributed to the cytoplasmic nucleotidyl transferase Cid1 [4,14]. Although Cid1 produced in Escherichia coli or purified by tandem-affinity chromatography from S. pombe apparently possesses highly processive poly(U) polymerase activity, most mRNAs modified by uridylation in a Cid1-dependent manner in vivo had only one or two 3′ uridylyl residues [14]. This could reflect stringent regulation of Cid1 activity in vivo, or perhaps the co-existence of active and potentially processive Cid1 with a poly(U)-specific exonuclease. This situation would be somewhat analogous to the balanced poly(A) polymerase and PARN activities associated with deadenylated mRNAs in Xenopus oocytes [10], or to the co-operation between the poly(A) polymerase activity of the TRAMP (Trf4/5-Air1/2-Mtr4 polyadenylation) complex and the exonucleolytic activity of the nuclear exosome [1517].

Sequencing of cRACE products from S. pombe mutants defective in the decapping activator Dcp1, the Lsm1 component of the Lsm1–7 complex, Ccr4, Pan2, Pan3 or a PARN orthologue allowed us to assemble a detailed model of the relationship between Cid1-mediated uridylation and mRNA decay [4]. For example, deletion of lsm1 or growth of a temperature-sensitive dcp1 mutant at its semi-permissive temperature led to an increase in the abundance of uridylated mRNAs, clearly suggesting that uridylation precedes Lsm1-mediated decapping. These data further suggested that uridylation promotes decapping in a manner that is at least partially redundant with deadenylation, as deletion of cid1 caused decapped mRNAs to have shorter poly(A) tails, whereas deletion of ccr4 led not only to longer poly(A) tails, but also to a substantial increase in the proportion of uridylated mRNAs.

In line with the notion that Cid1-mediated uridylation can be rate-limiting for mRNA degradation, we found that the half-life of the urg1 mRNA was almost doubled on deletion of cid1; a similar degree of stabilization was seen on deletion of ccr4, but deletion of pan2, pan3 or the PARN orthologue had no such effect. The greatest increase in urg1 mRNA half-life was seen on deletion of lsm1, supporting a model in which the Lsm1–7 complex is required for both Ccr4–Not- and Cid1-mediated decapping pathways [4].

The 3′ uridylation pathway of mRNA decay would seem to be absent from S. cerevisiae as, aside from the canonical nuclear poly(A) polymerase, its genome encodes only two Cid1-related nucleotidyl transferases, Trf4 and Trf5, and these are both thought to act as A-transferases in the TRAMP-exosome pathway of nuclear RNA turnover. Furthermore, our cRACE investigation of budding yeast ACT1 mRNA revealed no uridylated sequences [4]. In contrast, a recent report describes the addition of CU tails to polyadenylated mRNAs in Aspergillus [18], suggesting that Cid1-related RNA turnover pathways exist in diverse fungi, and that their absence from S. cerevisiae may be an exception. The question of whether or not polyadenylated mRNAs are subject to cytoplasmic uridylation in metazoans remains open at the time of writing, but it is clear that untemplated uridine residues can be added to the non-polyadenylated replication-dependent histone mRNAs in human cells [19]. This modification has been implicated in the rapid decay of these mRNAs following inhibition of DNA replication, although the identity of the terminal U-transferase(s) responsible is currently contentious.

Uridylation of other cytoplasmic RNA substrates

Several recent reports have described the uridylation of cytoplasmic non-mRNA substrates in mammalian cells and nematode worms. The Cid1-related U-transferase termed ZCCHC11/TUT4 (in mammals) or PUP-2 (in Caenorhabditis elegans) can be recruited to cytoplasmic pre-miRNAs [miRNA (microRNA) precursors] by the RNA-binding protein Lin28, resulting in the 3′ uridylation of the pre-miRNA and its subsequent degradation by an as yet poorly characterized mechanism [2023]. In this way ZCCHC11-dependent uridylation serves to inhibit the generation of mature, biologically active miRNAs of the let-7 family. A related but distinct function of ZCCHC11 results in the 3′ uridylation of mature miRNAs of the miR26 family, which again serves to inhibit their biological function, and hence allows the expression of a variety of pro-inflammatory mediators such as IL-6 (interleukin-6) [24]. The recent demonstration of 3′ uridylation of siRNAs (small interfering RNAs), again in C. elegans, identified yet another class of small RNA substrates for a Cid1-like U-transferase [25]. In this case, uridylation appears to function as part of a delicately poised network that determines the flow of siRNA traffic to distinct argonaute proteins with roles in RNAi (RNA interference)-induced transcriptional silencing, post-transcriptional gene silencing, heterochromatin maintenance and/or chromosome segregation.

Open questions

Several aspects of the regulation and function of RNA 3′ uridylation are currently unknown. Cid1 was first identified in fission yeast through its involvement in cell-cycle checkpoint control; its overexpression confers selective resistance to the combination of hydroxyurea, which inhibits DNA replication, and caffeine, which inhibits the cell-cycle checkpoint signalling kinase Rad3 [26,27]. Conversely, cid1 deletion results in loss of checkpoint integrity on exposure of cds1-mutant cells to hydroxyurea. Despite the elucidation of the Cid1-dependent mRNA decapping pathway, the basis of the checkpoint phenotypes seen when cid1 expression is altered has yet to be determined. It is likely that genome-wide deep sequencing of mRNA 3′ ends will be required to resolve this issue. It seems reasonable to expect that there will be some level of selectivity in uridylation by Cid1, even though the available data indicate that its role in mRNA turnover is a global one. Although it is tempting to draw parallels between the observation of S-phase-specific histone mRNA uridylation in human cells and the S–M checkpoint function of Cid1 in S. pombe, there is so far no indication that homologous mRNA targets are involved in the two systems.

The identification of structurally diverse RNA targets for ZCCHC11, including pre-miRNAs, miRNAs and mRNAs, together with the identification of Lin28 as a ZCCHC11 partner protein with selective RNA-binding activity, suggests that RNA U-transferases in general might employ multiple RNA-binding protein partners to direct them to specific substrates. Affinity purification of native U-transferase complexes and MS and/or identification of specific genetic interactions are likely to prove valuable in the identification of these partners.

Finally, it is pertinent to ask whether the activities of Cid1 and related U-transferases are regulated, for example in response to replication stress. Its identification as a cell-cycle checkpoint component led us to search for regulatory post-translational modifications of Cid1, particularly phosphorylation, that might provide a link to known checkpoint signalling mechanisms. Although modifications have been identified, none has so far been found to influence Cid1 activity, or to be cell-cycle checkpoint arrest-specific. Alternative possibilities include checkpoint-dependent modification of selective RNA-binding partners, which could direct Cid1 and related U-transferases to alternative RNA substrates. It is equally possible that the 3′ uridylation tag is interpreted differently under different conditions, for example through modification of the Lsm1–7 heptamer. In either case, regulation of the downstream consequences could be achieved without direct modulation of the transferases, which could remain constitutively active, as assayed in vitro. Consistent with this type of model, 3′ uridylated act1 transcripts in S. pombe were found in the decapped fraction in exponentially growing cells, but in the capped fraction in S-phase-arrested cells [4,14]. In the latter cells, uridylated mRNAs may not only escape decapping, but also be subjected to other processes, such as translational repression. Clarification of all of these points awaits further investigation.


Research in my laboratory is supported by Cancer Research UK [grant number C6016/A11587]; and the Biotechnology and Biological Sciences Research Council [grant number BB/F013531/1].


  • RNA UK 2010: An Independent Meeting held at The Burnside Hotel, Cumbria, U.K., 22–24 January 2010. Organized and Edited by Jeremy Brown and Nick Watkins (Newcastle, U.K.).

Abbreviations: cRACE, circularization-based rapid amplification of cDNA ends; miRNA, microRNA; PABP, poly(A)-binding protein; PARN, poly(A)-specific ribonuclease; pre-miRNA, miRNA precursor; siRNA, small interfering RNA; TRAMP, Trf4/5-Air1/2-Mtr4 polyadenylation; UTR, untranslated region


View Abstract