Gender Differences in Gene Expression in Human Brain Using Microarray Model Based Analysis

Vawter, M.P.; Tomita, H.; Watson, S.; Akil, H.; Cox, D.; Jones, E.; Bunney, W.E.
Society for Neuroscience 31st annual meeting. 2001.


We replicated the findings of gender differences in gene expression (Evans et al., SFN abstract). In our current dataset (12 samples) we compared human male and female gene expression levels on Affymetrix U95A with a model based expression analysis (Li and Wong, PNAS, 2001, 98 (1), 31-36). This model based approach to gene expression permits a group comparison of samples in which we found specific differences in gene expression between male and female brain samples. We set specific criteria for acceptance of our findings by examining duplicate Affymetrix chips to establish reliability and false positive rates. Chip to chip variation using duplicate cRNA fragmented samples suggested that two criteria were useful to minimize false positive gene expression differences solely due to chip variations. Our findings using a model based approach agreed with the Evans et al. findings on 4 of the 6 highest gene expression differences which used absolute Affymetrix calls of "Increased", "Decreased", "Present", and "Absent" filtering criteria in pair wise comparisons between all male and female samples. Thus, by using two different data analysis methods, we find evidence of gene expression differences between gender specific to sex chromosome-linked genes. We show that two cortical regions, dorsalateral prefrontal cortex and anterior cingulate cortex, appear similar by clustering while cerebellum samples are separately clustered from cortical samples. We are obtaining in situ hybridization data for probing whether these differentially expressed genes are localized in the neurons, glia, or all cells in the CNS. Although we can currently replicate findings between 2 different laboratories of differential gene expression that exceed 50% we also are examining a larger number of samples to determine normal control variation in gene expression in three brain regions of male and females to determine whether some gene differences may be reliably detected below this difference level. By using these standardization studies, we will embark upon microarray analysis of affective disorders using a case-control brain collection.