As many of you may know, I have been examining how mRNAs are transported and localized within the cell and how the regulation of mRNA metabolism contributes to gene expression. From data accumulated recently within the "RNA Field", we know that transcription in eukaryotic cells is very sloppy - that is, a plethora of different RNA transcripts are generated from seemingly random pieces of DNA. As I explained in a recent post, some of this background transcription seems to play a role in regulating how the DNA is packed and thus allows for a tighter control of RNA production from protein coding genes. Moreover many eukaryotic genes have to be extensively processed before they are ready for export - it is likely that processing is not perfect and that at the end of the day many aberrant transcripts are produced. All this background transcription and all these potentially aberrant mRNAs pose a challenge to the cell - how does the cell "know" which strands of RNA to translate into proteins? The cell must have some "quality control" mechanism to filter out the good RNA from the bad (or the random). We know this because when we analyze the proteins produced in a typical cell by mass spec, the vast majority of proteins are synthesized from bonafide protein coding genes and not random pieces of genome or miss-processed transcripts.
So how could this work?
As in most biological systems, there is a network of different biological process that helps to sort out the wheat from the chaff. Here are some of the major "quality control" steps:
- Surveillance of mRNA transcription. As mRNAs are being synthesized, "good" and "random" transcripts might be identified. For example it is now known that certain silent parts of the genome are tagged with siRNAs that recognize transcripts that are made from these sequences. The siRNAs feedback onto the DNA and modify how that part of the genome is packaged.
- mRNA processing. After synthesis, "good" but not "random" RNAs should be processed.
- mRNA nuclear export. "Good" mRNAs should be exported from the nucleus into the cytoplasm. In contrast "random" or badly processed transcripts should be retained.
- mRNA degredation. "Good" transcripts should be stable while "random" or "bad" ones should be degraded.
- mRNA translation. "Good" but not "random" RNAs might be preferentially translated by the ribosome. In addition it is likely that the "good" transcripts encode proteins that are relatively stable while "random" or "bad" transcripts encode junk. This could imply a feedback mechanism from newly formed protein and transcript (here's an example of this potential mechanism).
One key idea is that these different quality control steps should all be coupled. For example I've blogged about how DNA packaging itself may influence whether locally produced transcripts that are stable or degraded and vice versa. Factors that recognize how DNA is packaged bind to RNA degradation enzymes and in turn RNA recognition complexes (such as the siRNA containing RITS complex) bind to factors that modify DNA packing. Looking at other processes, for example mRNA processing, RNA export and RNA decay, we would guess that these too should be linked together. One would guess that only "good" transcripts should be exported and that "bad" transcripts are not only retained within the nucleus but also degraded. So are they?
The data would indicate that this is so. As a nice example I would point to a recent paper from the Tollervey and Hurt labs that appeared in the January edition of PLoS Biology, where the function of the yeast gene Swt1 is examined. Previously it had been reported that this gene was required for the survival of yeast strains that were partially impaired in RNA nuclear export due to the inactivation of the TREX1 complex (TREX = transcription export, see this post). Swt1 in fact stands for Synthetic Lethal with TREX. In the PLoS paper the authors present good evidence that this gene encodes an enzyme that degrades RNA. When this gene is knocked out, the level of aberrant transcripts (in this case transcripts that have an intron but remain unspliced) rises and there is an increase in unexported, presumably malformed, nuclear mRNA. Out in the cytoplasm there is an increase in the number of unspliced mRNAs that are being translated into protein. The authors also demonstrate that Swt1 is required for the survival of strains that lack another RNA export complex (this one is called TREX2) or that lack nuclear pore proteins that inhibit the export of unprocessed transcripts. These genetic interactions between the Swt1 and mRNA export genes indicate that the two processes (RNA degradation and RNA export) talk to each other and coordinately regulate RNA quality control. Thus here we have a convergence of three quality control processes, mRNA processing, mRNA degradation and mRNA export. All of this makes a lot of sense. In my own experiments I have found that mRNAs that are not exported from mammalian cells are also rapidly degraded indicating a deep connection between export and decay. One mystery is that this RNA degradation enzyme seems to be localized to both the nucleus and the cytoplasm, although in certain circumstances in accumulates at the nuclear pores.
As a follow up (and some of this work might have been done, but I'll have to check) I would like to see how these various proteins act on a truly random piece of RNA, say one produced by an intergeneic region (that is a region between two bonafide genes). My guess is that these intergeneic RNAs might be trickier to handle. In addition it seems unclear how the cell would identify random pieces of sloppy transcription. In any case some big surprises are sure to come up in future studies.
Michal Skruzny, Claudia Schneider, Attila Racz, Julan Weng, David Tollervey, Ed Hurt
An Endoribonuclease Functionally Linked to Perinuclear mRNP Quality Control Associates with the Nuclear Pore Complexes
PLoS Biol (09) 7(1): e1000008. doi:10.1371/journal.pbio.1000008
Which of those mechanisms you mentioned would be subsumed under "non-sense mediated decay"?
NMD is a "measure" of whether an mRNA was properly processed. There are two ways in which NMD is activated:
The most known NMD mechanism is as follows. During splicing the splicesome deposits exon junction complex (EJC) at each splice site. After the mRNA reaches the cytoplasm (although this is still debated by some) the mRNA undergoes a pioneer round of translation. During this process the ribosome strips these EJCs off of the mRNA. If there was an error in splicing, the ribosome will translate intronic sequence, hit a premature termination codon, and fall off the transcript before having cleared off any EJCs deposited downstream. These intact EJCs, in combination with translation termination factors present on the same transcript, target the mRNA for decay.
Recently it has become clear that mRNAs with extremely long 3'UTRs, regardless of whether they have any unspliced introns, are also substrates for NMD mediated RNA decay. It is considered "NMD-mediated" because quite a few of the EJC components are required for this type of decay, but it is presently unclear what the exact mechanism is. In this case NMD is degrading any transcript with an abnormally long 3'UTR due to some mistake in 3'end processing.
Interestingly, both types of NMD-mediated decay require actively translating ribosomes. Thus the NMD mechanism represent a coupling between the processes of translation and mRNA decay.