mRNA in dendrites: this message will self-destruct in 10 seconds

I just wanted to post a brief entry on the newest paper from the Moore lab. I've already hinted at some of these results in a post from last year.

As you all know, mRNA is made and processed in the nucleus. During mRNA synthesis, all the introns must be removed and all the exon sequences must be glued together to produce mature mRNA. At each exon-exon fusion site (called exon junctions) a protein complex, called the exon junction complex (EJC), is deposited. Once processing is over and done, the mature mRNA is exported from the nucleus into the cytoplasm. Then ribosomes engage the mRNA and translate the transcript into proteins. Translation begins at a start site and ends at a termination codon. During the first (or pioneer) round of translation, the ribosome bumps off any EJCs that may have been deposited along the length of the transcript. Now in the course of mRNA processing, sometimes mistakes occur, for example an intron may have not been spliced out. Since introns often have premature stop codons, any ribosome that translates an intron-containing message will fall of once it reaches the intron. Since the ribosome falls off of the transcript, any EJCs that were deposited downstream from that stop codon are no longer removed. These EJCs (in collaboration with translational-termination factors) targets these mis-processed mRNAs to processing-bodies (P-bodies) where the mRNAs are degraded. The whole process of destroying intron-containing transcripts is called Nonsense-mediated decay (NMD).

P-bodies are strange creatures, we know that RNA clumps together in these structures and that various conditions can help to target transcripts to these aggregates. mRNAs can be targeted to these structures by NMD and miRNAs. When cells undergo stress certain mRNAs are first targeted to larger aggregates called stress granules that then pass its transcripts to P-bodies. There are many RNA-aggregates in cells that may or may not be related to P-bodies. P-body like aggregates are seem in oocytes and these structures are where all the mRNAs that an egg inherits from the mother are stored. P-body like structures are also seen in neurons. These "RNA granules" travel up dendrites and are though to provide storage/delivery for certain transcripts. These dendritic transcripts may then be expressed right at the synapse upon neuronal stimulation. Local translation in neurons helps strengthen synaptic connections (i.e. LTP).

So what's the connection between all these RNA aggregates? Similar proteins are found in all of these structures. One puzzle is that the two functions that are often ascribed to the aggregates (destroying/storing mRNAs) are almost at the opposite end of the spectrum of RNA metabolism. On the one hand these RNA-bodies serve as a vault, where you would keep your most precious items, and on the other hand these granule is the trash compactor where you would dump anything you needed to get rid of.

In the new paper, Corinna Giorgi from the Moore lab demonstrated that eIF4AIII, a protein that anchors the EJC to the mRNA and is sometimes found in P-bodies, is present in neuronal mRNA granules. She isolated eIF4AIII derived from neuronal extracts, and discovered that this protein is bond to many particular mRNAs previously described as residing in these granules. Interestingly if she reduces the level of eIF4AIII, the level of these transcripts and the amount of translated protein increases.

In an interesting twist one of these mRNAs have introns at the very end of the transcript after the natural stop codon (in the 3' untranslated region or 3'UTR). Hence this transcript is a natural substrate for NMD! The idea would be that when these mRNAs are processed in the nucleus, an EJC is loaded onto their 3'UTR. Once they are exported and packaged into an RNA granule they are transported up dendrites to post-synaptic sites. When synaptic signalling occurs the mRNAs are briefly translated into protein. The ribosome hits the natural stop codon and the translational termination machinery and the EJC that sits on the 3'UTR collaborate to target degradation of the transcript - a self destructing message!

Next the authors analyzed transcripts from humans, mouse and rat to search for other mRNAs that have conserved introns in the 3'UTR. Specifically the introns have to be at least 50 nucleotides away from the stop codon and thus should activate NMD. Here's what they found:

- 266 alternatively spliced transcripts from (an unknown number of) genes have introns in their 3'UTRs
- 149 genes have introns in their 3'UTR but not in their open reading frame.

Some of these transcripts were found in neurons but many were also found in testis and a large portion were specific for peripheral blood and bone marrow indicating that hematopoietic cells may use this natural NMD process extensively.

As an aside, the Kosik lab show that neurons express miRNAs that inhibit the translation of transcripts that are found in dendrites. Although this process may simply act to inhibit the expression of these transcripts in the cell body, it is possible that miRNAs are playing a similar role to the EJC in that they contribute to a burst of translation. So just like the EJC, miRNAs target transcripts to P-bodies and inhibit translation. Unlike the EJC (and the NMD process) the inhibition of translation is not due to mRNA destruction and is not stimulated by translation itself. For more on this paper, check out this entry by Amnestic.

Corinna Giorgi, Gene W. Yeo, Martha E. Stone, Donald B. Katz, Christopher Burge, Gina Turrigiano, and Melissa J. Moore
The EJC Factor eIF4AIII Modulates Synaptic Strength and Neuronal Protein Expression
Cell (07) 130:179-191 doi:10.1016/j.cell.2007.05.028

Min-Jeong Kye, Tsunglin Liu, Sasha F. Levy, Nan Lan Xu, Benjamin B. Groves, Richard Bonneau, Kaiqin Lao, and Kenneth S. Kosik
Somatodendritic microRNAs identified by laser capture and multiplex RT-PCR
RNA (07) 13:1224-1234 doi:10.1261/rna.480407

More like this

Cool paper, thanks for pointing it out. My first thought is, is the escape from NMD a key to the mechanism, or simply a side effect of early capture of the mRNA for storage? Surely not all stored mRNAs have 3' UTR introns, and it seems overly baroque and costly to use NMD, on top of which depending upon the gain and/or maintenance of a 3' UTR intron. Not that we have any shortage of baroque, costly mechanisms in cells :-) but it seems like it's not a general-enough mechanism for the suite of mRNAs that must be stored away in the cytoplasm. That list of 3' UTR-intron containing transcripts covers a lot of functional ground, only a fraction of which has anything to do with neurons.


It seems like only a few transcripts in these RNA granules are derived from pre-mRNAs with introns in their 3'UTRs - so this NMD dependent burst (if it indeed exists) may only act on a couple of transcript types found in granules. I also agree (and the authors point out) that this process may be utilized in other cell types as well. One extra point is that the Moore group looked for introns within 3'UTRs that were conserved - it would be interesting to analyze which ones were not.

I don't mean to sound like a reviewer, or anything;
"P-body like aggregates are seem in oocytes and these structures are where all the mRNAs that an egg inherits from the mother are stored."

Have you got a reference or two I could read about this?

Thanks for your help ;o)

Hi Mags,

OK maybe not all the maternal mRNAs are in P-bodies, but probably quite a large fraction. As for a ref ... I'll try to get you something after I get back to Boston.