microRNAs emerged in the past decade as important regulators of gene expression, essential for many processes including embryonic development, cell differentiation, metabolism and immunity. They appear able to drive, on their own, the induction of pluripotency in somatic cells. Although initially thought to induce translation inhibition without affecting mRNA levels, it has become clear miRNAs do induce mRNA degradation, with some delay relative to translational inhibition. Surprisingly, transfection or induction of miRNA expression appear to have small effects on the target mRNAs. Therefore, one of the open questions in the field is how the dramatic phenotypic effects of miRNAs are brought about by relatively small changes in target mRNA levels. Various hypotheses have been proposed to explain this paradox, one being that miRNAs reduce the variance rather than the average levels of their targets, specifically through regulatory network motifs that they form with transcription factors. Such motifs involve a transcription factor (TF) that activates the expression of a miRNA and of an mRNA target, the miRNA post-transcriptionally down-regulating the same target. This feed-forward loop (FFL) is called 'incoherent' (iFFL) because the TF and miRNA have opposite effects on the common target, the ultimate outcome being a reduction in the fluctuations that the common target experiences. Especially through computational analysis, a number of examples of iFFLs have been discovered.

In the past years we developed methods for identification of miRNA targets and for estimating their effects. Specifically, we obtained binding sites of the Argonaute protein through crosslinking and immunoprecipitation and used them to infer a biophysical model of miRNA-target interaction. We showed that this model enables us to accurately identify targets of individual miRNAs, including those that do not perfectly match the 5' end of the miRNA (also called non-canonical). We further collaborated with the group of Erik van Nimwegen on modeling mRNA levels in terms of transcriptional regulation by transcription factors and post-transcriptional regulation by miRNAs. The model that we developed is able to identify key miRNA and TF regulators in individual cell types (embryonic stem cells) or processes (epithelial-to-mesenchymal transition, EMT). We therefore would like to develop this model further in the coming years by
1. Comprehensively identifying promoters of miRNA genes and predicting the associated transcription regulatory elements.
2. Extending the set of curated transcription factor regulatory motifs based on recently released chromatin immunoprecipitation data.
3. Analyzing a fine-grained time course of miRNA and mRNA expression during EMT (that will be obtained in our lab) in the context of a computational model, to identify and characterize the dynamics of networks involving miRNAs and transcription factors that operate during EMT.
4. Extending the model to the prediction of key splicing regulatory factors, some of which are known to be targeted by miRNAs and/or play important roles in EMT.

I believe that this project will result in new predictive tools that could be generally used to identify key regulators of gene expression in specific cell types and at the transition between cell types. Furthermore, by applying our computational tools to EMT we will on the one hand have the possibility to validate the model against the large body of available data and on the other hand to contribute new insights to this process of high medical relevance.