Hervé Seitz, Institut de Génétique Humaine (IGH), Centre National de la Recherche Scientifique (CNRS), Montpellier, France
MiRNAs are small regulatory RNAs that repress specific target genes. According to the current view, each miRNA regulates hundreds of genes. Computational tools aim at identifying miRNA targets, looking for seed matches in 3’ UTRs (Bartel, 2009). Usually, this is performed by selecting evolutionarily conserved miRNA binding sites. However, such predictions are often biologically irrelevant, since short matches are very frequent. Sixty per cent of human coding genes seem to be targeted (Friedman et al, 2009), and miRNAs are implicated in every physiological process in animals. Nevertheless, miRNA-mediated gene repression is usually very modest, i.e. less than two-fold, which is lower than well-tolerated fluctuations in gene expression. Focusing on miR-223-guided gene repression, Baek et al (2008) observed that, in wild-type mice, gene repression was smaller than the inter-individual variability for 150 out of 192 predicted targets. Hence, most predicted targets appear functionally insensitive to the miRNAs. Moreover, many conserved miRNA binding sites appear to be conserved in a miRNA-independent fashion. It is unclear, why miRNA binding sites have been conserved since they do not appear to be functional. Sequence elements may be conserved for other reasons, while being fortuitously complementary to miRNAs. Accordingly, comparative genomics may yield a high proportion of false positive results.
In revisiting the definition of a ‘miRNA target’ it is concluded that not every measurable change in gene expression translates into a macroscopic, evolutionarily selectable phenotype. The role of miRNAs in normal and pathological conditions may have been over-estimated, and the very notion of ‘gene regulation’ should be reconsidered taking into account the robustness of cellular homeostasis to external insults. It appears difficult to reconcile such observations with the extreme sensitivity required for genetic fine-tuning. Most likely, the ‘butterfly effect’ (indicating that a small dose may trigger a substantial consequence) has been counter-selected for.
How similar are ncRNA targets in rats and mice, i.e. the predominant animal species used in toxicology?
– Generally, they appear to be very similar, but concordances are not yet fully understood.
Is it necessary to address macroscopically observable effects for RA (which may counteract the modern toxicological paradigm to strive for short-term assays that focus on mechanisms of toxicity and MIEs)? – One should search for dose-sensitive effects that seem to be tightly controlled during gene regulation.