For Export and Decay, Splicing Helps Along the Way

RNA'S QUALITY CONTROL

First observed in the early 1980s in mammals, NMD ensures that only mRNAs with full coding potential are translated into protein. In mammals, if an mRNA has a splicing junction more than 50 nucleotides downstream of the first in-frame stop codon, it becomes targeted for NMD. But how the cell could possibly recognize the difference between the premature termination codons and the real termination codon was pure enigma, says Steitz, because this requires that the cytoplasmic translation machinery know the nuclear processing history of a given mRNA. Therefore, either translation components exist in the nucleus, or somehow presplicing history is communicated to the cytoplasm. And so, researchers speculated that mRNA must somehow remember where its introns were before being spliced.

"We called [it] a marker," says Miles Wilkinson, University of Texas, MD Anderson Cancer Center, Houston, "but we had no idea what it was or whether or not it even existed." He adds that the marker explanation reconciles the disparity: "It didn't require hypothesizing anything weird like nuclear translation."

Lynne Maquat, University of Rochester, NY, and others initially hypothesized that splicing-dependent changes in the proteins accompanying mRNA likely facilitate some sort of molecular memory. A protein would mark the exon-exon junctions in the nucleus, and the mark would be carried to the cytoplasm where it is read by translation machinery. Moore and others described that mark as the EJC. The spliceosome deposits the complex roughly 24 nucleotides upstream of exon-exon junctions in a sequence-independent manner.

In 2000, searching for the hypothesized mark, Lykke-Andersen and Steitz described Upf2 and Upf3, human homologs of yeast NMD proteins.² They then created an artificial mark by tethering the factors onto mRNA downstream of a premature termination codon, which resulted in NMD. "We were so lucky that it mimicked an intron and had that effect," says Lykke-Andersen. Izaurralde comments, "It was the first time to show that tethered proteins could recruit all the necessary machinery to degrade mRNA."

Back-to-back Science papers the following year from Steitz's group and University of Pennsylvania's HHMI investigator, Gideon Dreyfuss, confirmed a functional link between splicing and NMD.⁵⁶ "To have proteins that associate in the nucleus ... then remain on the mRNA in the same position after export, to me, is the most exciting aspect," says Dreyfuss. "It's really molecular memory. It's amazing."

Even though all organisms examined except bacteria manifest NMD, neither the mechanism nor the factors are conserved, says Maquat. "And, as far as I can tell, the role of the EJC in NMD is evident only in mammals."

THE EJC SPRINGBOARD

At the same time the Science papers came out, Moore's group published this issue's second Hot Paper, describing a connection between the exon-exon junction complex and factors involved in NMD and mRNA export.³ "We found the mark, the EJC, and then we had to prove that it was related to NMD," says Moore. They showed that the EJC interacted with Upf2 and Upf3, known NMD factors, says Steitz. "Then the whole story came together."

Courtesy of Elisa Izaurralde

In this assay, proteins of interest are expressed as fusions with the coat protein of bacteriophage MS2. The reporter gene harbors high affinity MS2-binding sites in the 3' UTR. When the stop codon is located at least 50 nucleotides upstream of these biniding sites, NMD factors (e.g., EJC proteins such as Y14) recruit additional components of the NMD machinery and the reporter mRNA is degraded.

This work also confirmed a connection between splicing and mRNA export, at least for very short pre-mRNAs containing no poly-A tails that are injected into Xenopus oocytes, so-called artificial mRNAs. "At the time, there was a controversy, and it's still going on, as to the extent to which splicing contributes to the export of mRNA," says Moore. In general, spliced mRNAs regardless of length are exported well. By studying spliced artificial mRNAs that carry or do not carry an EJC, Moore's group showed that the EJC permits efficient export of spliced mRNA by providing a binding platform for the export factors REF and TAP. "The first thought was, 'Wow, the EJC not only causes the mRNA to get degraded by NMD, but it's also important for mRNA [export],"' says Wilkinson. "It provided one explanation for a longstanding observation that if you add introns to your construct, you get much higher gene expression."

REF: EXIT ASSISTANT

That same year, Izaurralde's group published the third Hot Paper describing REF's role in mRNA export. Their findings suggest that both spliced and unspliced (intronless) mRNAs use common export factors.⁴ This notion went against the prevailing model and published work stating that spliced mRNA exited the nucleus more efficiently. The key to this research is that it tested intronless mRNAs. "We show that it doesn't matter if there is splicing or not. REF was able to bind TAP and promote export of even unspliced RNA," says Izaurralde. Moore's work shows that shorter RNAs are inefficiently exported unless they carry an EJC.³ Izaurralde determined that this effect becomes less pronounced with longer RNA sequences.⁴ The probability for REF and other mRNA export adaptors to bind to mRNAs independent of splicing is likely to increase with RNA length, providing an explanation for this length effect, says Moore.

The paper might be highly cited for many reasons, says Izaurralde. The REF proteins are part of the newly discovered EJC on which many groups are working. But the most likely reason, she says, is that the original model was extremely popular, and now most people cite the Hot Paper for its suggestion that splicing is not required for mRNA export, although it may enhance its efficiency.

The idea that splicing has so many effects on downstream mRNA metabolism has not generally been appreciated until now. Once studied in isolation, many disciplines now converge. "There were the splicers, the transcriptors, the translators," says Moore. "We didn't talk to each other as much before, but now we are. That's very exciting."

Linda B. Schultz LBSchultz@sciwriting.com is a freelance science writer in Atlanta, Ga.