A microarray gene expression study of the molecular pharmacology of lithium carbonate on mouse brain mRNA to understand the neurobiology of mood stabilization and treatment of bipolar affective disorder

Abstract

Lithium is the most widely prescribed and effective mood-stabilizing drug used for the treatment of bipolar affective disorder. To understand how lithium produces changes in the brain, we studied brain mRNA from 10 mice after treatment with lithium and compared them with 10 untreated controls. We used the MAS 5.0, Smudge miner, GC-RMA and FDR-AME packages of software (Bioconductor, Seattle, Washington, USA) to determine gene expression changes using Affymetrix MOE430E 2.0 microarrays after 2 weeks of lithium treatment. We used both a false discovery rate (FDR-AME) assessment of significance and the Bonferroni method to correct for the possibility of false-positive changes in gene expression among the 39,000 genes present in each array. Our primary method of analysis was to use t-tests on normalized gene expression intensities. By using a Bonferroni correction of P<1.28x10(-6), we found that 121 genes showed significant changes in expression. The three genes with the most changed mRNA expression were alanine-glyoxylate aminotransferase 2-like 1 (Agxt2l1), c-mer proto-oncogene tyrosine kinase (Mertk) and sulfotransferase family 1A phenol-preferring member 1 (Sult1a1). Also among the group of 121 genes with significant changes in gene expression that survived Bonferroni correction () were the genes encoding the Per2 period gene (Per2 P=1.33x10(-8), 2.47-fold change), the metabotropic glutamate receptor (Grm3, P=9.48x10(-7), 0.7-fold change) and secretogranin II (Scg2, P=9.48x10(-7), 1.28-fold change) as well as several myelin-related genes and protein phosphatases. By taking a significance value of P<0.05 without Bonferroni or FDR-AME correction, we identified a total of 4474 genes showing changed mRNA expression in response to lithium. FDR-AME analysis showed that 1027 out of these 4474 genes were significantly changed in expression. Among the mRNAs that were significantly changed with t-tests and FDR-AME were several that had already been implicated in response to lithium such as increased brain-derived neurotrophic factor mRNA ( t-test P=0.0008-0.0005, FDR-AME P=0.0396-0.0393, 1.44-fold change) beta-phosphatidylinositol transfer protein (Pitpnb, t-test P<0.0000, FDR-AME P=0.003, 1.26-fold change) and inositol (myo)-1(or 4)-monophosphatase 1(Impa1, t test P<0.0000, FDR-AME P=0.004, 1.22-fold change). Of interest in relation to the side effect of hypothyroidism, which is caused by long-term lithium treatment was the fact that we observed changes in mRNA expression in five genes related to thyroxine metabolism. These included deiodinase (Dio2 t-test P=0.000003-0.004, FDR-AME P=0.0048-0.061, 1.53-fold change) and thyroid hormone receptor interactor 12 (Trip12, t-test P=0.003, FDR-AME P=0.075, 1.19-fold change). Of relevance to multiple sclerosis was the observed upregulation of the long isoform of myelin basic protein (t-test P=0.00013, FDR-AME P=0.0169). Changes in mRNA expression were found in 45 genes related to phosphatidylinositol metabolism using uncorrected t-tests but only 13 genes after FDR-AME. Thus, our work confirms the considerable previous research implicating this system. Gene ontology analysis showed that lithium significantly affected a cluster of processes associated with nucleotide and nucleoside metabolism. The analysis showed that there were 170 genes expressing RNA described as having ATP-binding or ATPase activity that had changed mRNA expression. The changes found have been discussed in relation to previous experimental work on the pharmacology of lithium.