Involvement of microRNAs and their targeted genes in the pharmacogenomics of lithium response in bipolar disorder: a genome wide study

Bipolar disorder (BD) is a frequent mood disorder characterized by manic and depressive episodes. The disease affects approximately 1-2% of the population. Lithium is a mood stabilizer used as first line treatment of BD episodes and for maintenance. Despite 30% of patients show full remission of symptoms under chronic treatment, a high percentage of patients do not respond sufficiently. Moreover, lithium is a drug with a particularly narrow therapeutic window and wide range of mild to severe side effects. Lithium response has an apparently strong genetic background and lithium responders consist in a group with more homogenous manifestation of the disease. However, a vast body of research on the field was not enough to explain the existing variability, suggesting that factors other than DNA variants could be involved. Thus, more and more studies turn towards epigenetic research in order to explain the mechanisms of lithium response and identify predictive biomarkers. We attempted to investigate the total miRNAs expression profiles through a Next Generation Sequencing (NGS) approach. Lymphoblastoid cells lines (LCLs) derived from 20 BD patients, 10 Full Responder (FR) and 10 Non-Responder (NR) to chronic lithium treatment according to the “Retrospective Criteria of Long - Term Treatment Response in Research Subjects with Bipolar Disorder”, Sardinians for four generations and part of the Consortium on Lithium Genetics (ConLiGen) sample. The LCLs from both groups were cultured either in presence or in the absence of 1mM LiCl. Total RNA was extracted enriched in miRNAs and sequenced with MiSeq instrument (Illumina) and the reads were mapped using miRBase database. A comparative analysis between FR vs NR groups was performed and the effect of lithium treatment in vitro on the miRNA expression was estimated in FR and NR. Available transcriptomic data from the same cohort were used for a correlation analysis between the top hits of the two datasets. Subsequently, target prediction with online miRNA target prediction software helped to narrow down the list of the inverse correlated couples (miRNA-mRNA) and select miRNAs and target mRNAs for validation with qRT – PCR. Two out of four selected miRNAs were validated as significantly differentially expressed, one was downregulated and the other was upregulated. Three of the corresponding mRNA targets were also found significantly differentially expressed and inversly correlated to the associated miRNA.