A position-specific 3'UTR sequence that accelerates mRNA decay.

The 3' untranslated regions (3'UTRs) of mammalian mRNAs direct an extensive range of alternative post-transcriptional outcomes, including regulation of mRNA decay and translation, contributing significantly to overall gene regulation. However, our knowledge of the underlying sequences and mechanisms is incomplete. We identified a novel 3'UTR sequence motif in mammals that targets mRNAs for transcript degradation. The motif is found in hundreds of mRNAs and is enriched in transcripts encoding regulatory proteins, such as transcription and signaling factors. Degradation of mRNAs containing the motif is mediated by the CCR4-NOT deadenylation complex. We identified hnRNPs A1 and A2/B1 as trans factors that directly bind to the motif, indicating a novel role for these proteins in deadenylation. Interestingly, a genome-wide analysis of the impact of this new regulatory pathway showed that the most active motifs are located within the 5' and 3'-terminal portions of 3'UTRs, whereas elements in the center tend to be inactive. The highly position-specific function of the motif adds a new layer of regulation to gene expression mediated by 3'UTRs.

The control of gene expression is fundamental to cell function. Although much of this control occurs at the level of transcription, post-transcriptional control is both prevalent and consequential. In mammals, the 3′ untranslated region of an mRNA (3′UTR) plays a major role in determining the mRNA's post-transcriptional fate. Mammalian 3′UTRs contain almost as much conserved sequence as that ascribed to transcriptional regulatory elements, yet much of this conservation cannot be attributed to known regulatory mechanisms. Furthermore, the degree of sequence conservation is greater for mammalian 3′UTRs than for those of other eukaryotes. Notably, transcriptome-wide mapping of protein binding sites demonstrates that 3′UTRs are bound by a diverse set of several hundred different proteins, most of which are uncharacterized. Therefore, it is clear that 3′UTR regulation involves myriad cis sequences and trans factors, many of which remain to be discovered and characterized. Thus, an essential component of understanding 3′UTR-mediated gene regulation is to decipher the mechanistic details of specific regulatory pathways triggered by cis-regulatory sequences within them.

We set out to understand the cellular function and underlying molecular mechanism of a novel 3′UTR motif (UAAC/GUUAU), which drew our attention for its striking prevalence (∼7% of mammalian 3′UTRs contain one or more copy) and its strong conservation. We demonstrated that the motif is repressive and established that the core 8 nucleotide sequence is both necessary to mediate regulation and sufficient when tested in many different sequence contexts. Moreover, the motif is active only within 3′UTRs and is ineffective in either introns or 5′UTRs. We found that regulation triggered by the motif acts by accelerated transcript degradation, via mRNA deadenylation mediated by the CCR4-NOT complex, the major mRNA deadenylase complex (Fig. 1). We purified trans factors that recognize the motif and identified the hnRNPs A1 and A2/B1, which we confirmed bind directly to the motif with high specificity. To examine the role of these trans factors genome-wide, we profiled the transcriptomes of cells depleted for hnRNP A1 and A2/B1 using RNAi. Our results confirm that motif-dependent repression requires these trans factors and indicate that hundreds of transcripts harboring the motif are subject to regulation by hnRNP A1 and A2/B1.

Figure 1.

Model of UAASUUAU-mediated mRNA deadenylation. hnRNP A2/B1 and A1 bind to UAASUUAU (black rectangles, S = G or C) preferentially at the 5′ and 3′ edges of 3′UTRs (first and last ∼300 nt) and recruit the CCR4-NOT complex to mRNAs to promote deadenylation.

Like many RNA binding proteins, both hnRNP A1 and A2/B1 possess a variety of different cellular roles, including mRNA processing, export, stability and translation. Our work has established a new role for these proteins. The two proteins are paralogs and, together with hnRNP A0 and A3, belong to the A/B family of heterogeneous nuclear ribonucleoproteins. The A/B family represents the most abundant cellular hnRNP proteins, with ∼3×107 copies of hnRNP A2/B1 in HeLa cells. Furthermore, mutations in hnRNP A1 and A2/B1 are strongly associated with human diseases, including cancers and neurological disorders, and abnormal regulation of these proteins is observed in several cancers, including lung cancer.

The impact of regulatory pathways can be gauged in several ways; informative metrics include the quantitative outcome of regulation and the number and types of targets impacted. The UAAC/GUUAU motif mediates approximately 2-fold repression, which, although subtle, is comparable to that typically mediated by individual miRNA target sites. The motif is found in the 3′UTRs of ∼7% of human genes and is frequently deeply conserved, with conserved instances of the motif occurring more than twice as frequently as expected by chance. Thus, in terms of prevalence and conservation, this motif is comparable to binding sites for some of the most deeply conserved miRNAs, sites that are themselves among the most prevalent and strongly conserved sequences within 3′UTRs. It is worth noting that transcripts encoding regulatory and signaling proteins are enriched in the motif, which implies that the pathway we have found contributes to large and complex regulatory networks. We also found effective instances of the motif in 3′UTRs of multiple core RNA decay enzymes, including the 5′ exonuclease XRN1, the decapping complex (DCP1a) and the CCR4-NOT deadenylase complex itself (CNOT1/4/6). Together, these results imply a broad regulatory impact for the pathway we have discovered, including auto-regulation of mRNA decay.

The sensitivity of the functional impact of the UAAC/GUUAU motif to its location within the 3′UTR is striking. Indeed, analysis of transcriptome profiling to discover functional sites suggests that only motifs located within the terminal portions of 3′UTRs (first and last ∼300 nucleotides) are competent to attenuate transcript abundance. This finding is reminiscent of miRNA target sites, which tend to be more effective when located toward the ends of 3′UTRs rather than in the center, although the degree of position-specificity appears significantly stronger for the UAAC/GUUAU than for miRNA target sites. For both miRNA target sites and the UAAC/GUUAU motif, the physical or biochemical mechanisms underlying their position-dependent effects remain unclear – investigating and understanding these mechanisms represent important opportunities for insights into the biology of 3′UTRs.

Several possible models could explain the positional sensitivity of 3′UTR regulatory elements. First, the central regions of 3′UTRs might be more likely to form occlusive secondary structures, thereby limiting access of trans factors to regulatory elements. Second, proximity of the motifs to the poly(A) tail and proteins that interact with the 3′ terminal mRNP might potentiate certain regulatory pathways. For example, it is easy to envision that trans factors found close to the 3′ terminus of a 3′UTR could have more effective interactions with PABP, deadenylases, or even the translation initiation machinery, when compared with such interactions with same trans factor bound elsewhere in a 3′UTR. These models are not mutually exclusive, nor may a single explanation apply to all sequence motifs. Nevertheless, it is likely that mechanistic insights into the phenomenon will be valuable in building toward a more comprehensive understanding of 3′UTR mediated gene regulation.

The enhanced efficacy of the UAAC/GUUAU motif when located toward the 5′ terminus of a 3′UTR is more difficult to rationalize. The most prevalent internal mRNA modification of higher eukaryotes is m6A (N6-methyladenosine). This modification has been implicated in affecting diverse aspects of post-transcriptional gene regulation, including mRNA translation, decay, splicing, alternative polyadenylation, and heat shock response. Transcriptome-wide maps indicate that m6A is found in at least ∼50% of mammalian mRNAs. More importantly, m6A is preferentially found within the 5′ end of 3′UTRs, suggesting a regulatory role restricted to a certain region within 3′UTRs. Importantly, analyzing CLIP-seq data, we found that hnRNP A2/B1 and A1 preferentially bind elements at the 5′ end of 3′UTRs, indicating a potential connection to m6A sites (Fig. 2). Consistent with this model, hnRNP A2/B1 was recently identified as an m6A-binding protein. We are currently exploring whether the abundance of m6A modifications close to the stop codon results in more effective recruitment (or activity) of hnRNP A2/B1 to UAAC/GUUAU motifs located near the 5′ end of 3′UTRs.

Figure 2.

Abundance of genome-wide binding sites for hnRNP A2/B1 and A1 (region −600 to 600 nt from the stop codon) within mRNAs analyzed using public CLIP-seq datasets, m6A abundance according to Ke et al.

Irrespective of the mechanistic basis for the position effect of the UAAC/GUUAU motif and other regulatory elements in 3′UTRs, there are clear implications when the prevalence of alternative 3′UTR isoforms are considered. Alternative splicing and, even more so, alternative polyadenylation can alter the length and composition of 3′UTRs for different transcripts of the same gene. Thus, a single gene may gain or lose cis-regulatory sites and consequential post-transcriptional gene regulation due to alternative mRNA processing. In addition, because the pathway we have found is so sensitive to the position of the motif within the 3′UTR, we suspect that alternative 3′UTR processing may have even greater impacts upon transcripts harboring the motif than for other cis-regulatory sites in 3′UTRs. That is, depending on the location of a motif, and the specific 3′UTR isoform(s) produced, alternative transcripts of the same gene will gain or lose sensitivity to the pathway we describe even in conditions where different isoforms all contain the motif. Moreover, even though hnRNP A1 and A2/B1 are expressed ubiquitously, dynamic regulation dependent on the motif will still occur as a result of alternative 3′UTR processing.

The 5′ and 3′ untranslated regions of human mRNAs contain hundreds of different cis-regulatory elements, creating an extraordinarily complex collection of post-transcriptional events with the potential to significantly impact gene regulation. Despite the remarkable progress in understanding post-transcriptional biology, it seems likely, in addition to position-specific effects of sequence motifs, that many novel layers affecting 3′UTR regulation remain to be discovered. Although the last decade has seen remarkable progress toward understanding the roles of 3′UTRs in gene regulation, the abundance of new approaches and techniques suited to the study of gene regulation will likely propel many additional insights into 3′UTR biology in the near future.