miRNAs are one of the hottest fields of research right now in molecular biology — they’re thought to regulate development, cancer progression, and probably everything else, too, since bioinformatics analyses show that up to 20% of the genome (and maybe more) are targeted by miRNAs. As a one-sentence primer on their function, miRNAs are transcribed off of our genome and processed into short, 20-22nt long single-stranded RNAs, which then attach to mRNAs with complementary sequence to the miRNA and (we thought) downregulate their expression in coordination with the protein complex RISC (RNAi-induced silencing complex). To find out that these miRNAs work the opposite of what we thought (in non-dividing tissue, which is most of the adult human) would be major.

On the one hand, I’m highly skeptical (one of my colleagues in a lab that works on miRNAs are dubious as well). On the other hand, it is plausible considering that a lot of miRNA research has been on development (when cells are dividing more), and that the vast majority of major papers that look for function do so with microarrays on dividing cells (HeLa, A549, etc.).

I read the abstract, skimmed the paper, and wondered why I didn’t see any computational analysis, despite the fact that there’s a LOT of data out there that could easily be used to test this hypothesis. So I did a very quick analysis — I searched Google Scholar for the first miRNA I could find that’s expressed in the brain, since it’s a low-proliferating tissue. I ended up with miR-134 based on this Greenberg paper in Nature. Then I went to TargetScan and found the conserved targets of miR-134. And then plugged them into SymAtlas to find their tissue specificity.

According to Steitz, this miRNA, which is expressed in the brain, should be UP-regulating its targets. So we expect the target list to have enriched expression in brain tissues — this is the opposite of what was previously believed. And what the hell do you know, but it panned out! The three tissues with highest average expression were the amygdala, prefrontal cortex, and whole brain. Here’s the graph, click on it to see it enlarged (note that I didn’t do any cleaning of the SymAtlas data, and I didn’t remove duplicate expression profiles):


On a philosophical level, this paper showcases both the strengths and weaknesses of the current scientific process. On the one hand, it shows how years and millions of dollars can be spent on research — including cutting-edge genomic and bioinformatic research — and we can still miss major aspects of what we’re studying. The biggest reason for this is that, as I mentioned before, the large-scale microarray studies are done on growing cells, not serum-starved cells. And it appears that the up-regulating effect acts on translation rather than transcription, so we probably wouldn’t even see anything on microarrays if we had been studying serum-starved cells. Unfortunately, large-scale proteomics is seriously lagging behind microarrays, so it’s very difficult to amass enough data on the effects of miRNAs at the protein level…unless you have an army of graduate students to do western blots all day.

On the other hand, this paper is how science is supposed to work. Science is by definition not always right. But through openness and transparency, we can eventually come closer to the truth. And as a reminder, this is just one paper, and we haven’t seen the responses from the miRNA community. So we don’t even know if this paper represents a major step toward the truth.

Steitz, J.A., Tong, Y., Vasudevan, S. (). Switching from Repression to Activation: MicroRNAs Can Up-Regulate Translation. Science, , . DOI: 10.1126/science.1149460

(Source: http://www.onerandomscientist.com )