PerspectiveRNA Events

No End to Nonsense

See allHide authors and affiliations

Science  11 Oct 2002:
Vol. 298, Issue 5592, pp. 370-371
DOI: 10.1126/science.1078096

Two of the hottest topics in eukaryotic gene expression research involve absolute nonsense: nonsense-mediated mRNA decay (NMD) and nonsense-associated altered splicing (NAS). “Nonsense” in this case refers to a type of mutation in mRNA transcripts that causes the protein synthesis machinery to terminate prematurely their translation into proteins. Nonsense mutations were originally thought to affect only the length, and therefore the function, of the encoded protein. However, it is now apparent that they can dramatically decrease the half-lives of mutant mRNAs as well as alter the pattern of precursor mRNA (pre-mRNA) splicing (see the figure). The molecular basis of the latter phenomenon (NAS) is particularly mysterious, because it is generally accepted that nonsense mutations cannot be recognized as nonsense until after the splicing process is complete. Two papers, one by Mendell et al. on page 419 of this issue (1) and another by Wilkinson and co-workers in a recent issue of Molecular Cell (2), now begin to unravel this mystery by showing that NMD and one type of NAS (reading frame-dependent NAS) are functionally distinct processes that rely on different, but overlapping, sets of proteins.

NMD is the quality control system by which mRNAs containing premature stop (nonsense) codons are selectively eliminated by eukaryotic cells. It is thought that by removing these defective mRNAs, NMD protects cells from potential damage due to inappropriately truncated proteins. To date, a number of proteins required for NMD have been identified in a variety of organisms, and analysis of how they regulate this process is well under way (3, 4). NAS, on the other hand, has proven much more controversial (57). Although numerous examples of NAS have been described, most of these can be readily explained by conventional mechanisms involving chance disruption of RNA sequences called exonic splicing enhancers (ESEs). ESEs are target sites for proteins that help to define pre-mRNA splice sites; their disruption by almost any type of mutation can cause altered splicing. Therefore, the promotion of alternate splicing by most nonsense mutations is simply due to the destruction of a key recognition element for the splicing machinery and has nothing to do with the ability of nonsense mutations to be recognized subsequently as stop signals during protein synthesis. Recently, however, Wilkinson and co-workers convincingly demonstrated that NAS of certain T cell receptor gene transcripts does require that the mutations act as protein synthesis stop signals (2, 8). They showed that certain nonsense mutations only mediated alternate splicing if they were “in frame” with a start signal, meaning that they had to be detected by the protein translation machinery. In addition to the T cell receptor case, available evidence suggests that nonsense mutations can alter splicing in a reading frame-dependent manner in other systems as well (912).

Circumventing stop signals.

Alternate fates of mRNA transcripts containing nonsense mutations. (Top) Shown is a pre-mRNA harboring a nonsense stop signal within an internal exon (colored box). In the majority of such molecules, introns (lines) are removed at the usual sites, resulting in retention of the nonsense mutation in the mature mRNA. Such aberrant mRNAs are then subject to degradation by NMD. (Bottom) In some cases, however, a nonsense mutation can activate an alternate splicing pathway (NAS), yielding a stable mRNA lacking the mutation. These two effects of nonsense mutations are dependent on different protein factors. Whereas NMD requires both Upf1 and Upf2, reading frame-dependent NAS is Upf2-independent.


One aspect of T cell receptor NAS that is particularly difficult to reconcile with our current understanding of eukaryotic gene expression is that this type of NAS depends on the mRNA reading frame, something that is not established until after splicing is complete. A pressing challenge, therefore, is to understand how the apparent downstream process of reading-frame recognition can feed back to alter the apparent upstream process of pre-mRNA splicing. Given that reading frame-dependent NAS is triggered by the same signals that trigger NMD, one possibility is that NAS is an indirect consequence of NMD. If so, then both phenomena should be dependent on the same subcellular machinery. What both groups now show is that this is not what actually occurs. Depletion of two factors requisite for NMD, Upf1 and Upf2, abrogated NMD of mutant T cell receptor mRNAs. In contrast, reading frame-dependent NAS proved resistant to Upf2 depletion (1, 2). Yet, because both processes do require Upf1, there is clearly some mechanistic overlap between them.

In their study, Mendell et al. used RNA interference (RNAi) to eliminate Upf1 and Upf2 from their test cells. RNAi can be initiated in mammalian cells by incubating them with small pieces of double-stranded RNA, one strand of which is complementary to a target mRNA. These small double-stranded RNAs cause the targeted mRNA to be degraded. RNAi allows the researcher to “knock down” individual mRNAs, and thereby remove the proteins they encode, without having to introduce mutations at the DNA level. After determining a knockdown phenotype, one generally wants to know what specific part of the target protein is responsible for the eliminated activity. Thus, having shown that Upf1 is necessary for both NMD and reading frame-dependent NAS, Mendell et al. (1) wondered whether Upf1 plays the same role in both processes. To address this question, they used a strategy known as “allele-specific RNAi.” By changing the sequence of the Upf1 gene encoded on a plasmid, they created a version of Upf1 that was not subject to RNAi by the double-stranded RNAs used to knock down the endogenous protein. They could then rescue both NMD and NAS in the RNAi-induced cells by introducing this altered plasmid. This strategy allowed them to determine the effects of two different substitution mutations in conserved regions of Upf1. One of these mutations supported NAS, but failed to rescue NMD. Thus, the functions of Upf1 in NAS and NMD are genetically separable, again demonstrating that these two processes are related yet mechanistically different.

Although the two new studies rule out one possible mechanism for reading frame-dependent NAS, we are left with many more unanswered questions. Of utmost interest is the unknown mechanism by which recognition of the mRNA reading frame can feed back to alter pre-mRNA splicing. One way to address this problem is to identify the proteins required. Identification of Upf1 as the first such factor thus represents a crucial step toward this goal.


View Abstract

Navigate This Article