One of my favorite recent ideas wondering through the literature is that of an RNA regulon or post-transcriptional operon. Operons in prokaryotes are groups of genes whose protein products all function in the same biochemical pathway. The genes are coordinated by sticking them all next to each other and transcribing all when you transcribe one. The post-transcriptional operon idea is that RNA motifs allow proteins in the same biochemical pathway to be regulated at the translation step instead. If several proteins were needed, for instance, to build some new architecture sticking off a cell at a specific location far from the nucleus, it wouldn't do to have to coordinate them way back there. Instead, you just throw in an RNA motif, say AUUUA. Then produce an RNA binding protein that is specific for that motif. Now traffic that protein to the location of interest. All of the RNAs will be localized to the right spot.

Of course, localizaton is just one way this could work. Any process better controlled faster or farther away from the nucleus could use an RNA regulon. One notable case is that of the Pumilio family (Puf) RNA-binding proteins in yeast. Melissa J. Moore explains it here:

... each Puf protein exhibited a highly skewed distribution of bound mRNAs: Puf1p and Puf2p bound mostly mRNAs encoding membrane-associated proteins, Puf3p almost exclusively targeted messages for nuclear-encoded mitochondrial proteins, and Puf4p and Puf5p associated primarily with transcripts encoding proteins bound for the nucleus. In several cases, a majority of the subunits comprising a particular multiprotein machine, such as the mitochondrial ribosome and a number of nuclear chromatin modification complexes, were encoded by mRNAs "tagged" by a single Puf protein. Together with earlier data (12), these new results (16) strongly support the idea that the expression of proteins with common functional themes or subcellular distributions is coordinated by large-scale regulatory networks operating at the mRNP level.

Many other examples can be found in this review by Jack Keene. I don't think I've seen an example of this yet, but given the slight wobble in microRNA specificity, one could imagine a single microRNA regulating a whole set of genes. Also, most interesting for my neuro-tastes is the recent report from the Moore lab showing that the immediate-early gene implicated in neuronal homeostasis, Arc, may be part of a regulon defined by introns in the 3'UTR. The mechanism is just too clever but requires an explication on the "pioneer round" of translation. Basically the cell tricks itself into thinking it made a funky RNA and destroys it after one round of synthesis. The other RNAs regulated in this path in neurons must have opposing effects to Arc though because knocking down this negative regulation pathway led to increased excitability (increased Arc reduces neuronal excitability). This raises a more general question. The idea of RNA regulons is nice, but how much can you predict knowing that your gene of interest is part of one? RNAs associate with multiple complexes throughout their lifespan, and complexes gain and lose factors dynamically. Also, how promiscuous are RNA binding proteins for cellular processes? For instance, I originally became aware of the Hu proteins as positive regulators of the pre-synaptic calcium-buffering protein GAP-43, but it turns out that they also regulate proteins involved in immune function. Maybe I am just thinking at too high a level of cellular organization. Perhaps all of those proteins respond to calcium in some way. At any rate, I'm expecting that RNA regulons will be increasingly important in understanding the translational regulation that must take place in dendrites to produce persistent memories. Looking forward to more on that in the next year or so.

Labels: RNA, translation

Kosik and colleagues used laser capture microdissection to get RNA populations from dendrites or cell bodies of cultured rat neurons. They optimized their technique so that mRNAs known to be enriched in dendrites, such as CaMKII and MAP2, showed about equal levels from soma and dendrite. They then performed multiplex PCR for several mRNAs and 187 miRNAs. The distribution of mRNAs and miRNAs is similar with a large somatic population and a gradient going out into the dendrites. Some small proportion of miRNAs have a little bit of dendritic enrichment. One point the authors are trying to get across is that there is no such thing as a 'dendritic RNA' because even the mRNAs and miRNAs that show some dendritic localization usually show just as much in the cell body.

Two nice things are that this paper validates a couple miRNA target prediction programs (Pictar and Targetscan) and that they provide a quantitative view of the miRNA copy number per cell. Both of these prediction programs suggested that miR-26a would target MAP2. This is convenient since both showed a somatodendritic distribution, meaning they hang out together even out at the farthest dendritic reaches. Inhibition of miR-26a with a synthetic oligonucleotide resulted in increases in MAP2 protein expresion, as one would expect from the classic miRNA-target relationship. As far as I am aware this brings the total of known dendrtici miRNA target pairs up to three, the other two being mir-268 and CaMKII (in drosophila) and miR-134 and LIM-Kinase. Quantification was achieved using PCR with known copy number standards. They knew how many cell bodies they captured, so they could get a copy number per cell estimate (probably a minor undershoot since even if they are awesome they probably couldn't save allllll the RNA from degradation). Anyway, they found... well I'll let them explain it:

rno-miR-124a is among the most abundant miRNAs in neurons and fell in the range of 10^4 copies per neuronal cell body. Despite its abundance, rno-miR-124a is enriched in cell bodies. rno-miR-26a and rno-miR-16 are less abundant miRNAs and fell in the range of 10^3 copies per neuronal cell body (Table 6). Because (delta)Ct of 2.61 +/- 0.39 describes the distribution of most miRNAs between the cell body and neurite, the number of copies of many miRNAs distributed along this gradient may be as low as in the hundreds of copies in the dendritic compartment. Even a one-order-of-magnitude error in this number is far below the number of synapses on the dendritic tree, and, therefore, the copy numbers of many miRNAs are likely to fall below one per synapse.

Delta Ct refers to the number of PCR cycles (i.e. doublings) it takes for the dendritic levels to reach the somatic levels. For instance, a delta Ct of 2.61 means that there are 2^2.61 (~6.1) times more somatic copies of the miRNA than there are dendritic copies.

I was particularly intrigued by this last sentence even though I have no idea what it means:

Stochasticity derived from the effects of miRNAs will contribute to the activation barrier for coherent responses, to the utilization of information provided by translational bursting, and to the flexibility needed by dendrites to sample alternative states (Kaern et al. 2005).

Guess I'll have to read Kaern et al. real quick.

Labels: dendrites, microRNA, RNA

RNA interference is a process by which small (20-22 nt) RNAs bind to a fully or partially complementary messenger RNA and reduce the amount of protein product from that mRNA. The general rule is that if the match is perfect (full complementarity) then the target mRNA is cut into two pieces and destroyed forthwith. If the match is imperfect such that there are bulges in the double stranded RNA that forms between the interfering RNA and the target, then the target is sequestered to a newly discovered cellular entity called a Processing Body (P Bodies, PBs). There are enzymes in PBs capable of degrading mRNAs, but sometimes the mRNAs can be released and become translationally competent again.

New research from Kiriakidou et al in Cell provides a mechanism for this translational repression sans degradation. The effects of small interfering RNAs (siRNAs) are mediated by the Argonaute family of proteins (Ago1, Ago2, etc). This family can be subdivided depending on the proteins' ability to cleave RNA and thus carry out the "perfect-match" type of translational repression, but even non-cleaving Agos can do the sequestration route for repression. The latest news is that this can be achieved by blocking interactions between the cap-binding translation initiation factor eIF4E and the 5' cap of mRNAs.

Let me unpack. For efficient initiation of protein synthesis from an mRNA, several proteins must assemble into complexes centered around the mRNA. There are several proteins that bind near the other end of the mRNA where there is a cap. A cap is a modified guanine nucleotide flipped around backward and stuck on the head-end of the mRNA early in its life. One protein in particular, eIF4E recognizes the cap structure and binds to it, recruiting other initiation factors and eventually the small ribosomal subunit. This is an important and highly regulated step in protein synthesis. For instance, there is a family of proteins (4E-BPs) whose sole function is to bind eIF4E and get in the way of cap-binding. If they become highly phosphorylated because of this signaling pathway or that, they let go and translation proceeds. Ago proteins can do the same thing, but on the cap side and without the phosphorylation business.

They showed the effect by first purifying an Ago protein with and without important amino acids for cap-interaction and testing for binding with caps immobilized on a column. Only Ago proteins with the two important (phenylalanine) amino acids could bind. Further assays in vivo showed that the mutant Agos couldn't mediate translational repression.

There are a couple predictions to make based on these findings.

1) Organisms with Agos that lack this domain should be bad at this process.

This domain is not found in Ago proteins of plants, archaea, or fission yeast, in Drosophila AGO2 and in most members of the C. elegans Ago protein family, with the exception of ALG-1 and ALG-2. In addition, the MC domain is absent from proteins of the PIWI family.

I can't recall if any of there is anything already contradictory in that list. I think there is definitely something weird about the way plants handle siRNAs, but the details escape me.

2) RNAs that are capable of cap-independent translation should not be regulated by this process. There is debate about the degree to which mRNAs can undergo cap-independent translation, but the field is moving along as though internal ribosomal entry sites are an important cellular tool, so these RNAs should escape translational repression via this process.

Labels: RNA, translation

"Ya stressed out. Depressed out ya brain." - Baatin

I've been meaning to write about the Costa-Mattioli et al paper in the early April issue of Cell. It's got some very cool findings, but there is a lot of background to get on board. So maybe we can take a running start by covering some of their references and some basic biology. The basic idea is that some proteins are counterintuitively upregulated while the protein synthesis machinery is globally inhibited. The mechanism is pretty clever and it may be used in a relatively large number of eukaryotic mRNAs.

First off, just a little bit about the mechanism of translation. Of course you know the central dogma of genetics. DNA --transcription-> mRNA --translation-> protein. The mRNA is supposed to act an intermediary between the nucleus and the cytoplasm so the two worlds can communicate. An mRNA contains a nucleotide 'recipe' that is decoded by translation machines called ribosomes and a special type of RNAs called transfer RNAs (tRNAs). Numerous cofactors help the process of translation along at its various stages: initiation, elongation, termination, and release/recycling. In eukaryotes, these factors are named in a semi-organized system indicating which stage they have been implicated in. For instance, initiation factors are named eIFsomething for eukaryotic Initiation Factor X. You can't always trust nomenclature systems based on function though, because new knowledge renders the naming system inaccurate. For instance, new studies indicate a role for eEF1a in initiation of translation. I apologize for all the nomenclature, but the names are the names and we all have to live with it.

So I want to talk about a particular initiation factor, eIF2. There are three eIf2 subunits: alpha, beta, and gamma. We are going to pretend gamma doesn't exist. Alpha is crucial to the assembly of ribosomes on an mRNA. In its GTP-bound form, it is responsible for bringing the first amino acid for any given protein (which is always methionine) to the ribosome. As initiation actually occurs, the Guanidine TRI - phosphate is converted to Guanidine DI - phosphate (GDP) releasing energy and allowing the machine to change shapes in the necessary ways to start scootching down the mRNA reading codons. You have to have eIF2alpha-GTP to start synthesizing a new protein, and it is a resource that must be replenished with every round of translation initiation. The job of exchanging the GDP falls to eIF2beta.

All of that is the normal process carried out by cells day-to-day. Under a range of circumstances, cells will want to regulate the amount of new proteins being synthesized on a more-or-less global level. For instance, during viral infection it may be to the cell's advantage to reduce synthesis of new proteins and go into a more protective, stressed-out state. Also, if something in the protein folding process starts going haywire, the cell may want to slow down on creating new proteins until they can get the post-translational processing sorted out. These cellular stress states are communicated to the translation machinery by way of a group of enzymes called eIF2alpha kinases. They are capable of phosphorylating eIF2alpha. I'm not sure how many times I've explained what phosphorylation is, but you can think of it in basic terms as adding a reactive group to a protein to change its shape and electronegative characteristics. It is a very common way of 'throwing a switch' to activate or deactivate a given protein. A kinase, by definition, is an enzyme that phosphorylates other proteins.

EIF2alpha that has been phosphorylated becomes a qualitatively different protein. Rather than promoting translation initiation, it now acts as an inhibitor. It is still bound by eIF2beta, but eIF2beta can no longer load it up with a new GTP. Instead alpha sticks in its craw and ruins it even for other unphosphorylated alphas. The net effect is to reduce the amount of eIF2-GTP and thus the amount of ready-to-roll translation machines. The pathway to remember here is this:

Cellular Stressor -> Cellular Stress Response -> eIF2alpha Kinases -> eIF2alpha phosphorylation > eIF2beta inhibition -> reduced eIF2alpha-GTP -> globally reduced translation initiation.

This is a lot of names and a lot of pathways to get up on. In coming posts I hope to build on this knowledge to examine a specific type of mRNA that can circumvent this global translation reduction. In fact, certain mRNAs gain an advantage during cellular stress states. Generally these mRNAs code for proteins that are important for dealing with the stressor. Once we have the mechanism on board we will be at the very starting point for understanding the Costa-Mattioli paper I mentioned at the beginning. By the way, for our Spanish speaking audience, I think I found an interview with Costa-Mattioli en espanol. Three classes later I still don't know any spanish science terminology, so if anyone listens to it and I am entirely mistaken about the content, lemme know.

Labels: RNA, translation