Nonsense-mediated RNA decay--a switch and dial for regulating gene expression

Abstract

Nonsense-mediated RNA decay (NMD) represents an established quality control checkpoint for gene expression that protects cells from consequences of gene mutations and errors during RNA biogenesis that lead to premature termination during translation. Characterization of NMD-sensitive transcriptomes has revealed, however, that NMD targets not only aberrant transcripts but also a broad array of mRNA isoforms expressed from many endogenous genes. NMD is thus emerging as a master regulator that drives both fine and coarse adjustments in steady-state RNA levels in the cell. Importantly, while NMD activity is subject to autoregulation as a means to maintain homeostasis, modulation of the pathway by external cues provides a means to reprogram gene expression and drive important biological processes. Finally, the unanticipated observation that transcripts predicted to lack protein-coding capacity are also sensitive to this translation-dependent surveillance mechanism implicates NMD in regulating RNA function in new and diverse ways.

Keywords: gene regulation; noncoding RNA; nonsense-mediated RNA decay; quality control; translation termination.

Figure 1

mRNP composition influences NMD substrate recognition. A: Translation of the entire protein-coding region of a normal mRNA by ribosomes serves to remodel the mRNP and clear RNA-binding proteins from this region of the RNA. Termination proximal to the poly(A) tail and poly(A) binding protein PAB1 is postulated to provide positive signals indicating termination is occurring within an appropriate mRNP context. B: mRNAs harboring an early stop codon undergo premature translation termination and, as a consequence, have an increased length of RNA downstream that fails to be remodeled and serves as a platform for a multitude of RNA binding proteins including several known to influence NMD substrate recognition, including the multi-protein exon junction complex deposited at exon junctions (fusion point of green and orange boxes) during pre-mRNA splicing (1) and the NMD factor UPF1 (2). Additionally, ribosomes terminating prematurely and distal to the poly(A) tail and PAB1 would fail to receive termination promoting signals thus indicating that termination is aberrant (3).

Figure 2

Nonsense codons can be introduced by genetic mutation or by errors or variation in transcription by RNA polymerase II. A: Genomic mutation (red bar) resulting in the introduction of a nonsense codon or frameshift in the open reading frame (ORF; green box) leads to 100% of the mRNA expressed from that locus containing a PTC at a fixed position (left). In contrast, nucleotide misincorporation during transcription by RNA polymerase II generates mRNA harboring PTCs at a frequency of <0.5% at variable positions within the mRNA ORF. B: Variations in transcription initiation site (left) create heterogeneity in the length of the mRNA 5′ UTR, impacting translation initiation and reading frame choice (discussed in detail in the text and see also Fig. 4). Similarly, the use of alternative 3′ end formation sites (right), due to mutations in cleavage and polyadenylation signals or regulated alternative polyadenylation (APA), generates heterogeneity in 3′ UTR length and differential susceptibility of the mRNA to NMD.

Figure 3

Alternative pre-mRNA splicing generates substrates for NMD by a variety of means. Exon skipping during pre-mRNA splicing generates an mRNA void of early nonsense codons and whose ORF is fully translated (top). Intron retention commonly results in inclusion of an in-frame PTC into the mature mRNA (1). Similarly, alternative 5′ or 3′ splice site usage leads to partial intron retention (2) and/or shifts in the translational reading frame that result in premature translation termination. Skipping (3) or inclusion of an exon of a nucleotide length that is not divisible by 3 results in frameshifts and premature termination. Finally, inclusion of a PTC-containing exon (i.e., a poison cassette exon) introduces a genomically-encoded premature termination codon and targets the splice variant for NMD (4).

Figure 4

Translational recoding can impact the position of translation termination. The position of translation initiation by 80 S ribosomes can be influenced by a variety of means and impacts the frame read by the translational machinery (left). The use of upstream, non-AUG codons for translation initiation (1) or downstream AUG codons due to leaky scanning (2), can result in translation of alternative reading frames and termination upstream of the natural stop codon. Additionally, translation of upstream open reading frames (uORFs) within the 5′ UTR (3) also leads to premature termination and targeting of the mRNA to NMD. Translational recoding during translation elongation (right) can occur as a consequence of programmed ribosomal frameshifting (1) or incorporation of selenocysteine at upstream UGA in selenoprotein mRNAs (2). The former event often directs translating ribosomes to nonsense codons in the −1 reading frame, while the latter facilitates bypass of the PTC and averts targeting of the mRNA to the NMD pathway. The extent by which translational recoding influences the susceptibility of an mRNA to NMD will depend on the frequency of the recoding event.

Translation of lncRNAs can have different outcomes depending on the biological function of the transcript

Translation of lncRNAs can produce biologically important peptides (1), suggesting that the genes expressing these transcripts warrant reannotation as protein-coding. lncRNA resulting from spurious transcription are predominantly cleared from the cell through degradation in the nucleus but can be targeted to surveillance by NMD if they escape to the cytoplasm and engage the translation machinery (2). For lncRNAs with regulatory functions in the cytoplasm, translation and targeting to NMD may act to modulate steady-state levels and activity of the RNA; in the absence of targeting by NMD, RNA levels are elevated, favoring enhanced function of that lncRNA (3). For lncRNAs whose function is to regulate gene expression in the nucleus, translation and rapid degradation by NMD in the cytoplasm would establish and maintain a concentration gradient that ensures the lncRNA population is principally nuclear and predominantly noncoding (4). Finally, NMD can assist in clearing the cell of lncRNAs that are sampled by the translation machinery in the process of de novo evolution of protein-coding genes (5). While lncRNA translation is necessary for the sampling of new peptides, rapid elimination of transcripts expressing nonfunctional products is important to protect the cell against deleterious effects.

Figure 5

NMD acts as a switch and a dial to modulate gene expression. A: One-hundred percent of transcripts expressed from genes harboring nonsense or frameshift mutations harbor PTCs and are potential substrates for NMD; efficient targeting of these RNAs to this surveillance pathway serves as an effective means to eliminate them from the cell (left; targeted transcripts shaded in gray). In contrast, when only a fraction of expressed transcripts harbor PTCs or are targeted to NMD, the pathway serves to eliminate only a subset of the RNA. Variation in the proportion of an mRNA population targeted to NMD may result from heterogeneity in 5′ or 3′ UTR length or the nature of the spliced isoform (middle), or due to the incomplete penetrance of an NMD feature (right). B: The overall impact of NMD on gene expression (bottom) is dependent upon the fraction of mRNA targeted for degradation (top). When a large percentage of expressed transcripts are targeted, such as for genomic mutations, NMD serves as a switch to ensure minimal expression from the aberrant gene. For stochastic errors in transcription or processing, the overall impact of NMD on gene expression is small and perhaps even experimentally undetectable. In contrast, targeting of mRNA between these extremes provides an important mechanism for fine-tuning gene expression that will have variable phenotypic consequence depending on the gene and overall levels of mRNA (and expressed protein) required by the cell.

Figure 5

NMD acts as a switch and a dial to modulate gene expression. A: One-hundred percent of transcripts expressed from genes harboring nonsense or frameshift mutations harbor PTCs and are potential substrates for NMD; efficient targeting of these RNAs to this surveillance pathway serves as an effective means to eliminate them from the cell (left; targeted transcripts shaded in gray). In contrast, when only a fraction of expressed transcripts harbor PTCs or are targeted to NMD, the pathway serves to eliminate only a subset of the RNA. Variation in the proportion of an mRNA population targeted to NMD may result from heterogeneity in 5′ or 3′ UTR length or the nature of the spliced isoform (middle), or due to the incomplete penetrance of an NMD feature (right). B: The overall impact of NMD on gene expression (bottom) is dependent upon the fraction of mRNA targeted for degradation (top). When a large percentage of expressed transcripts are targeted, such as for genomic mutations, NMD serves as a switch to ensure minimal expression from the aberrant gene. For stochastic errors in transcription or processing, the overall impact of NMD on gene expression is small and perhaps even experimentally undetectable. In contrast, targeting of mRNA between these extremes provides an important mechanism for fine-tuning gene expression that will have variable phenotypic consequence depending on the gene and overall levels of mRNA (and expressed protein) required by the cell.