Service Communautaire d'Information sur la Recherche et le Développement - CORDIS

Periodic Report Summary 1 - PTRCODE (Systematic study of post-transcriptional regulation mediated by RNA-binding proteins and miRNAs)

Post-transcriptional regulation (PTR) is the control of gene expression at the mRNA level, between transcription and translation. PTR is increasingly recognized as a pervasive and complex system controlling every step of RNA metabolism. PTR is mediated by the interactions of RNA-binding proteins (RBPs) and microRNAs (miRNAs) with target sites on mRNAs. Recent studies show that RBPs and miRNAs function in coordination with each other through cooperative or competitive interactions. Despite this fact, the majority of studies so far have focused on the regulatory effect of individual RBPs and miRNAs. With this project, we aim to develop accurate models of post-transcriptional regulation that incorporate the effects of RBPs, miRNAs and their interactions. Until now, systematic study of PTR mechanisms has been difficult due to lack of experimental results on binding specificities of RBPs and miRNAs and lack of genome-wide studies that measure the post-transcriptional fate of mRNAs in a quantitative fashion. Recently developed high-throughput approaches (e.g. CLIP and RNAcompete) have made significant progress in expanding our knowledge on RBP and miRNA binding sites. Additionally, new data sets on post-transcriptional regulation of mRNAs (e.g., half-life, steady-state expression) are emerging rapidly. In this project we aim to leverage these datasets to achieve the following specific aims: (i) genome-wide mapping of RBP and miRNA binding sites together with the RNA secondary structure and conservation information, (ii) identification of cooperative and competitive interactions between RBPs and miRNAs, (iii) integrative analysis of RBPs, miRNAs and the interactions between them to understand post-transcriptional regulatory events, (iv) combining transcriptional and post-transcriptional regulatory networks to explain differential gene expression in cancer.

In this period, we made significant process in every specific aim mentioned above. We started with predicting the RNA secondary structure of 3’UTRs. In addition to calculating the accessibility of each position, we also identified stable stem-loops. To map RBP binding sites in 3’UTRs, we compiled existing CLIP datasets and RNAcompete-derived motifs. Similarly, we combined the Ago-CLIP datasets with computational methods TargetScan and Pictar to map the binding sites of miRNAs in 3’UTRs. Subsequently, we identified competitive interactions by characterizing the overlaps between binding sites of a pair of factors. To identify cooperative interactions we determined the factors whose sites co-occur more often than expected by random chance. We also identified cases where the secondary structure plays a role in cooperation. Our next step was to analyze genome-wide datasets on PTR to assess the effects of cooperative and competitive interaction. Upon analysis of HuR and PUM1(2) knockdown datasets, we found that competition with other factors decreases the likelihood that the RBP of interest will bind to target sites. As such, presence of overlapping binding sites for other competing factors resulted in distinct functional outcomes. Similarly, we found that the transcripts that contain binding sites of a pair of factors that act in cooperation show greater effect (i.e., increased or decreased stability depending on the factor) than transcripts that contain the site of only one of these factors. We confirmed the existence of cooperative interactions between PUM1(2) and miRNAs, and identified an expanded list of these miRNAs. Next, we developed a logistic regression model that incorporates sites of RBPs and miRNAs to predict stability and steady-state expression of mRNAs in various cell types. We achieved a high accuracy in classifying a transcript asstable (or highly expressed) or unstable (or lowly expressed). Lastly, we investigated regulation mediated by RBPs in cancer. As a first step, we identified the differentially expressed RBPs in several cancer types. We developed a lasso regularized regression that incorporates copy-number variation, DNA methylation, and regulatory effects of TFs, miRNAs and RBPs to predict gene expression in cancer. We applied this model to lung squamous cell carcinoma (LUSC) as we identified a high number of RBPs that are differentially expressed in this cancer type. Performance analysis of this model revealed that RBPs have the highest added predicted value compared to the other feature types. We then applied a feature selection procedure to identify the candidate regulators of gene expression in LUSC. This analysis revealed many RBPs that appear at the top of the list of candidate regulators.

In the second period of the project, we plan to apply our models to additional datasets and organisms to assess the pervasiveness of the cooperative and competitive interactions that we identified. We expect to confirm our current findings on these new datasets. Furthermore, we will utilize our existing collaboration with molecular biologists to experimentally validate some of our findings. We will also check whether there is any disease associated SNP that map to the regulatory regions that we identified. Additionally, we will apply the integrative lasso regularized regression model on other cancer types, and we expect to confirm that RBPs play critical roles in regulating gene expression in several cancer types. Overall, the results of this project will strengthen the fact that RBPs and miRNAs function in coordination with each other. Additionally, our results in predicting gene expression in cancer will reveal the importance of post-transcriptional regulatory effects in carcinogenesis and these regulatory relationships can be used as potential targets for therapies. In summary, the models developed in this project will contribute to a better understanding of gene expression regulation in normal and disease cells.

Project website:

Reported by



Life Sciences