Thursday, August 23, 2007

RNA regulons   posted by amnestic @ 8/23/2007 10:59:00 PM

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: ,

Tuesday, June 19, 2007

A mechanism for miRNA-mediated repression   posted by amnestic @ 6/19/2007 07:13:00 PM

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: ,

Tuesday, May 08, 2007

Nice 'N Slow: how to make more GCN4   posted by amnestic @ 5/08/2007 08:58:00 PM

Let me take you to a place nice and quiet. There ain't no one there to interrupt. Ain't gotta rush. I just wanna take it nice and slow. - US-HER RA-YM-OND

I just summarized in a previous post how eIF2alpha kinases can reduce the amount of a crucial resource needed for initiation of protein synthesis, namely eIF2-GTP. There are four well-characterized eIF2alpha kinases: PKR, PERK, GCN2 and HRI. I was previously a little vague in my characterization of the inducing conditions for this pathway. The four kinases are activated by double-stranded RNA (representing viral infection), ER stress (i.e. protein misfolding), amino acid starvation, and low heme (iron scarcity), respectively. Regardless of the specific cause, the general effect is to suppress protein synthesis cell-wide while the cell deals with some perturbation. Still, for the cell to go about the business of handling its biz, it has to make a few key proteins. Indeed, it might even want to make more of these proteins than it was making before.

By far, the most heavily studied of these pathways is the GCN2 response to amino acid starvation. Recall that proteins are a string of amino acids. The secret decoder ring to translate the language of nucleotides into that of amino acids is the transfer RNA (tRNA). Transfer RNAs usually have a triplet of nucleotides at one end and an amino acid at the other, like an adapter. If a cell is low on amino acids, tRNAs might get made with on amino acid on the other end. if GCN2 finds out this type of tomfoolery has been going on it brings the hammer down and phosphorylates eIF2alpha, putting a large restraint on protein synthesis. That's all well and good; it's like putting your state under martial law for a minute during a crisis. It's not enough though because if the authoritarian GCN2 regime just repressed every attempt to get a message out, the cell would lose its vitality. Instead, the global repression of protein synthesis has an immediate positive effect: GCN4 synthesis increases. GCN4 is a transcription factor and can thus affect exactly which types of programs the cell is putting its resources toward. GCN4 is a practical transcription factor. Given the situation of amino acid shortages, and it drives the cell to produce more genes in the amino acid biosynthesis pathway.

So how can GCN4 escape the crackdown and curry the favor of the ribosomes? The answer lies in the first ~600 nucleotides of the mRNA coding for GCN4. The sequences in this area, called the 5' untranslated region, allow GCN4 to play it kind of coy with the ribosome. If the ribosome gets things started with GCN4 mRNA, it might get to keep going for a little bit, but it eventually hits roadblocks. The only way for a ribosome to get to the main task of really translating GCN4 is to take it nice and slow and sometimes skip opportunities to make a move. You may not know about the primary structure of mRNAs though, so let's take a brief look.

The main thing to note is that I am an artistic genius. After that, you can note that the whole RNA doesn't code for protein. There are big chunks that hang off the 5' and 3' ends of the protein-coding region (AKA the open reading frame). 5' and 3' refer to specific features of nucleic acid structures, but all you need to know is that ribosomes read from 5' to 3', and since we speak American around here, we will put the 5' end on the left. To initiate translation the ribosome assembles with several other factors including eIF2-GTP at the very 5' end of the mRNA. There is a structure here called the 5' cap that is recognized by the initiation complex. The assembled ribosome et alia start scanning the mRNA from 5' to 3'. As it moves from left to right it will first encounter the 5' untranslated region (UTR). The 5'UTR is made up of nucleotides, but the ribosome does not translate them into amino acids yet. The signal for a ribosome to start translating the genetic code and creating a new protein is a Start Codon. Start codons have the sequence AUG. The scanning ribosome carries a tRNA with it (called the initiator tRNA) that can base-pair with AUG and which carries the amino acid, methionine, on the other end. I don't want to get involved with the mechanism for adding amino acids onto the chain. Suffice it to say that the once the ribosome "opens" a "reading frame" it reads the nucleotide code triplet by triplet and builds a protein. The final feature of the open reading frame is the Stop Codon. There are three nucleotide triplets that do not code for any amino acid. When the ribosome reaches these, it usually pauses for a while and then stops making proteins. I would've said that it falls off, but I am about to describe a process that depends on it continuing along in the 3' direction. The last two features of your average mRNA are not of great importance to the current discussion. They are the 3' UTR (more nucleotides not coding for proteins) and the polyA tail. The polyA tail promotes translation and RNA stability.

So to be very clear: an Open Reading Frame (ORF) is the part of the RNA that a ribosome actually reads between the Start and Stop codons. The role of eIF2-GTP is to bring the initiator tRNA to the ribosome, so that it can be used as soon as the ribosome finds a Start Codon. When the peptide is started, the eIF2-GTP is used up, and if that ribosome wants to start any other peptides, it has to have a new eIF2-GTP.

Now imagine that I lied to you about 5' UTRs. You don't have to imagine very hard. There are quite a few 5' UTRs that contain open reading frames. They just aren't the main course as it were. They are called upstream open reading frames (uORFs). A ribosome can start at the cap, read an uORF and synthesize a short protein, and then scan further down the mRNA to the real ORF. GCN4 mRNA, for instance, has four uORFs before the ORF that codes for the GCN4 protein. If you artificially construct a GCN4 mRNA that lacks these uORFs, you get a lot more GCN4 protein. The uORFs thus have the effect of inhibiting translation of the downstream ORF under normal conditions. Many of the experiments detailing the 5' UTR pretty much ignore the middle uORFs and focus on uORF1 and uORF4 because these seem to be enough to get eIF2 kinase dependent expression.

The ribosome always reads uORF1 first because that's what it encounters first during scanning. Under normal conditions, it is reloaded with eIF2-GTP and an initiator tRNA relatively rapidly and it can read ORF4. ORF 4 is a hangup. It causes the ribosome to tarry right at the end. This is partially because its last amino acid is proline, which is relatively rare and difficult to find, so the ribosome is losing momentum by the time it reaches the end of ORF4. As it tries to scan even further down to the GCN4 ORF, it encounters a little more opposition, throws up its hands, and dissociates from the mRNA. As a result, we get no GCN4. The magic of eIF2 kinases is that they reduce the availability of eIF2-GTP and thus delay re-initiation after translation of uORF1. If the ribosome takes just a little bit longer to be ready again, it an skip ORF4. See? Isn't that cool? If it only reads ORF1 and takes its sweet time getting ready it can get all the way to the GCN4 start codon. In this way, GCN4 protein is increased while translation of the majority of other mRNAs is inhibited.

The protein of interest in Costa-Mattioli et al is not GCN4, but it has a similar mechanism. Next I hope to describe some of the experiments one has to do to discover that an mRNA is controlled in this manner. I will focus on the articles that showed that the production of ATF4 protein is regulated by eIF2 kinases in mammalian cells.

Labels: , ,

Sunday, May 06, 2007

Intro to stress-induced translation regulation   posted by amnestic @ 5/06/2007 01:57:00 PM

"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: ,