Beyond sequence homology: Cellular biology limits the potential of XIST to act as a miRNA sponge.

INTRODUCTION

The sponging of microRNAs by a long non-coding RNA (lncRNA) away from their coding gene targets is a conceptually-simple, yet biologically-complex method of lncRNA-mediated gene regulation. Currently, predictions of genes that participate in sponge-based regulation are largely based on sequence homology alone, which may not adequately reflect the cellular environment in which lncRNA:miRNA pairs interact. The vast number of potential interactions generated by these predictions impedes the identification of functional gene regulatory relationships, which necessitates an approach that considers biological context. XIST, the female-specific lncRNA canonically involved in silencing the X chromosome, has been suggested by many studies to act as a miRNA sponge. The sex-specificity of XIST provides the opportunity to study the biological feasibility of proposed XIST-miRNA interactions. Here we take a comprehensive approach by considering factors that affect possible regulation through XIST-miRNA sponging.

RESULTS

To identify the most feasible candidates in a particular tissue (lung adenocarcinomas), we considered protein-coding genes that (1) were positively correlated with XIST expression within sexes, (2) were targeted by miRNAs shared with XIST, and (3) expressed in lung adenocarcinoma. This revealed a robust set of 124 genes potentially positively regulated by XIST through the sequestration of 804 shared miRNAs. We then used the basic sex-specific nature of XIST to compare the changes in miRNA-target gene relationships in endogenously high-XIST and low-XIST systems to discover a high-confidence set of only 13 miRNA-gene pairs. As XIST is expressed exclusively in the nucleus, we validated the nuclear presence of several of these high-confidence miRNAs using RT-qPCR, confirming the co-localization required for XIST to interact with these species.

CONCLUSIONS

We use a biology-driven approach to identify genes defended from miRNA-based inhibition by the lncRNA XIST. Importantly, we identify that only a small subset of miRNAs predicted by sequence homology alone have the capacity to mediate the XIST-target gene axis, as they are enriched in the nucleus and able to co-localize with XIST for sponging. Our results reinforce the necessary consideration of biological features in future studies of lncRNA:miRNA interactions.