Tuesday, October 10, 2006

Global inhibition, specific activation   posted by Coffee Mug @ 10/10/2006 11:06:00 PM

Another aspect of the Raab-Graham et al. paper that might be missed in the flurry of different pharmacological agents and signaling pathways, is that the synthesis of Kv1.1 is actually increased in response to a manipulation that normally inhibits translation. Rapamycin is the drug of interest here. It inhibits a protein called mTOR (mammalian target of rapamycin). There is still some debate about what mTOR does exactly, but suffice it to say that it activates other signaling molecules downstream. The end result of mTOR signaling is to promote cap-dependent translation. Most mRNAs have a structure on the head end called a 5' cap. It is basically a nucleotide put on backwards, but it can be recognized by proteins involved in the initiation of protein synthesis. Recognition of the cap by these proteins is usually necessary to get the mRNA hooked up with the ribosomes. So rapamycin should be inhibiting this process and thus inhibiting translation initiation for most cellular mRNAs. Note that other protein synthesis inhibitors that act further downstream actually do decrease the rapamycin-induced increase in Kv1.1.

We don't get an explanation for this counter-intuitive regulation mechanism. The simplest explanation would be that Kv1.1 has an internal ribosomal entry site (IRES). IRESs are common in viral transcripts. Their mechanism is not fully understood, but they somehow allow ribosomes to jump in and bind mRNAs without having to deal with caps or cap-binding proteins. When cap-dependent translation is inhibited, cap-independent translation-capable transcripts would get all the goodies. RNAs with IRESs will be at an advantage. It is not unprecedented for important neuronal mRNAs to carry these elements. For instance, Pinkstaff et al. found functional IRESs in five dendritic mRNAs (including Arc and CaMKII). Unfortunately for this hypothesis, Raab-Graham et al. found no evidence for an IRES in Kv1.1 and showed that the 5' UTR (where IRESs usually hang out) was not necessary for the rapamycin regulation. This suggest that rapamycin is having its effect through some signaling pathway besides that responsible for cap-dependent translation.

Rather than dwell on Kv1.1 since we don't know the answer. I thought I'd show you a couple more examples where translation is inhibited, but certain RNAs do extremely well. I think the mRNA for CaMKII must contain every type of regulatory element, and must be controlled by every possible pathway. In this paper, Scheetz et al. used a fairly unusual in vitro preparation that should mostly just carry the synaptic compartments of dendrites to study the effects of NMDA receptor activation on protein synthesis. As you may know, the classic story is that when two neurons fire at the same time NMDA receptors are activated on the post-synaptic neuron allowing calcium into the cell and initiating signaling that will lead to plasticity. Scheetz et al. found that in the first two minutes following NMDAR stimulation there was actually a drop in global protein synthesis accompanied by an increase in activity of a protein called eEF2 kinase, and an increase in CaMKII synthesis. The activation of eEF2 kinase slows down the elongation phase of translation. The idea here is that mRNAs that are good at initiating translation may actually be at a disadvantage during this state because they get stuck further down the road. This would shift the balance to mRNAs that have a hard time with initiation, potentially like CaMKII. One cause of initiation difficulty may be complicated secondary structures near the beginning of the transcript that make it hard for ribosomes to scan down to the start codon (the codon that signals the first amino acid in the newly forming protein).

Speaking of start codons, one further example where global protein synthesis inhibition can actually be good for specific translation is that of the GCN4 upstream open reading frames. This mechanism hasn't been shown to play a role in plasticity-related protein synthesis yet. Most of the details have been worked out in yeast. In response, to amino acid starvation, an enzyme called GCN2 is activated. It in turn phosphorylates (sticks a electronegative, function-altering phosphate group on) a protein called eIF2alpha. When eIF2alpha is phosphorylated it serves to inhibit production of a key component in translation initiation (called the ternary complex). So it is harder to initiate translation. The ternary complex carries the first amino acid of every protein, methionine, which matches up to the start codon, AUG. The mRNA for GCN4 has several AUGs that are not the start codon for the actual protein. These AUGs are associated with upstream open reading frames (uORFs), sections of mRNA sequence that code for little chunks of protein that don't do any good. These are like 8 or 9 amino acid peptides we're making here. When amino acids are around and all is well with the cell, ribosomes start at the cap of GCN4, find uORF1, translate it, find another uORF downstream, translate that, and fall off. They don't make it to the actual protein-coding part of the transcript. When initiation is inhibited, they can't get a new ternary complex in time to translate the later uORFs. They finally get a new ternary complex in time to read the actual open reading frame of the GCN4 gene. So once again, global protein synthesis is turned down, but an mRNA with a funky 5' end gets the advantage.

I find it interesting to consider that dendritic protein synthesis following synaptic activity might actually contain two components. In the first couple minutes after stimulation, global synthesis might be downregulated and certain RNAs with special attributes that normally weaken their translation are upregulated. In the minutes to hours afterward, global synthesis is upregulated and more general synapse building proteins are manufactured. The synthesis could come in waves coordinated by complex secondary structure and ribosome-obstructing elements in the 5' untranslated regions of certain key mRNAs. Strangely, the intriguing Scheetz et al. finding released back in 2000 has not been followed up. One wonders if others have tried and been unsuccesful or what.