Traditionally, gene expression patterns were seen as mostly dependent on transcription ... yes those nasty bits of DNA that seemed to be ignored by most "science journalists". But the picture that is emerging is that transcription is looking more and more sloppy, and this "sloppyness" is built into the system so that the act of transcription tends to influence the organization of that part of the genome, regardless of what is being transcribed. It also appears that processes upstream of transcription, (such as mRNA processing, mRNA export and translation) play a greater role in determining expression patterns.
In a recent issue of Current Biology there's a nice example of post-transcriptional regulation of genes. In their manuscript, Merritt et al., demonstrate that in the ovaries of C. elegans (a nematode worm), the timing of expression of a given gene product at the protein level is largely determined by RNA elements found in the portion of the transcript that comes after the protein-coding region (also known as the 3' untranslated region or 3'UTR for short).
Here's a remarkable figure from that paper showing the expression of GFP when fused to either the 3'UTR of an oocytes specific gene (left) or under the control of that same gene's promoter (right). Just to orient you, worm oocytes are a long sack that begin on the top left where the oocytes progenitors are found. As the oocytes mature they travel rightwards, then make a sharp U-turn and they finally endup at the bottom left where sperm is created (most C. elegans are hermaphrodites) :
Note that the GFP expression pattern is modulated by being hooked up to the 3'UTRs of the fbf-2, pgl-3, mex-5 and spn-4 genes and not their promoters. Remarkably these UTR-driven expression patterns recapitulate the observed expression patterns. In contrast when the same experiment is performed with elements from the spe-11 gene, the expression of GFP in the sperm seem to be more dependent on the promoter. Now worm oovaries and their oocytes share one curious feature with almost all other oocytes (including ours), they are full of RNA granules. Another curious feature is that for quite a substantial period of time during maturation, the oocytes nuclei are all found in a a common cytoplasm (called a syncytia). These syncytia also occur in other organism at various points in development, such as during drosophila embryogenesis. Interestingly it was demonstrated that about 70% of the mRNAs in the drosophila syncytium were localized, often to same subcellular structures where the encoded protein was normaly present (for more info on this incredible finding, see this post).
So getting back to the question that arises from nthe Current Biology paper, how would RNA granules and 3'UTRs coordinate protein expression patterns?
One possibility is that the localization of the mRNAs themselves to different granules throughout the oocytes could dictate protein expression, the localization of the mRNAs being specified by RNA elements within the 3'UTRs. A second possibility is that the entry and exit of mRNA into the granules at various sites would be dictated by these same UTRs. Most observations suggest that when mRNAs are loaded into these large structures, their translation is silenced. We do know that as the oocytes mature, the content of their associated granules changes. Perhaps these changes cause mRNAs to dissociate with the granules allowing for translation or allow them to be repackaged into the granules and thus turning off protein production. It remains formally possible that some translation could occur within the granules and this could be modulated by their protein composition.
The authors try to clarify these issues by providing some indirect evidence that the mRNAs are found throughout the ovary but that the UTRs help to restrict transcription in distinct regions. So it would appear that the RNA granule transit model is more likely, but more work needs to be done to get at the exact mechanism.
My favorite part about that paper was their "operon" fusion validation that the mechanism was all post-transcriptional; I had no idea that that technique was even possible in metazoans.