Global state measures of the dentate gyrus gene expression system predict antidepressant-sensitive behaviors

Abstract

Background: Selective serotonin reuptake inhibitors (SSRIs) such as fluoxetine are the most common form of medication treatment for major depression. However, approximately 50% of depressed patients fail to achieve an effective treatment response. Understanding how gene expression systems respond to treatments may be critical for understanding antidepressant resistance.

Methods: We take a novel approach to this problem by demonstrating that the gene expression system of the dentate gyrus responds to fluoxetine (FLX), a commonly used antidepressant medication, in a stereotyped-manner involving changes in the expression levels of thousands of genes. The aggregate behavior of this large-scale systemic response was quantified with principal components analysis (PCA) yielding a single quantitative measure of the global gene expression system state.

Results: Quantitative measures of system state were highly correlated with variability in levels of antidepressant-sensitive behaviors in a mouse model of depression treated with fluoxetine. Analysis of dorsal and ventral dentate samples in the same mice indicated that system state co-varied across these regions despite their reported functional differences. Aggregate measures of gene expression system state were very robust and remained unchanged when different microarray data processing algorithms were used and even when completely different sets of gene expression levels were used for their calculation.

Conclusions: System state measures provide a robust method to quantify and relate global gene expression system state variability to behavior and treatment. State variability also suggests that the diversity of reported changes in gene expression levels in response to treatments such as fluoxetine may represent different perspectives on unified but noisy global gene expression system state level responses. Studying regulation of gene expression systems at the state level may be useful in guiding new approaches to augmentation of traditional antidepressant treatments.

Figure 1. Principal components analysis generates a robust meaure (PCA1) of gene expression system state.

Panel (a) plots the proportion of variance explained by principal components 1–4. Panel (b) is a bivariate plot of PCA1 values obtained from PCA done on all samples grouped together (x-axis) versus PCA1 from separate PCAs on samples grouped by region (y-axis) (r = 0.99, p<0.001). Panel (c) shows that the data processing algorithm used has no effect on values for PCA1 (r = 1.00, p<0.001). Panel (d) compares PCA1 values of dorsal and ventral dentate from the same mice (r = 0.97, p<0.001). Panels (e–f) are cross-correlation tables for PCA1 values obtained from PCA on independent groups of 2000 expression levels/group in dorsal (e) and ventral (f) dentate samples. Panel (g) plots gene connectivity (x-axis) against the correlation coefficient of genes with PCA1 demonstrating that more connected genes follow more closely with PCA1. Panel (h) plots PCA1 versus the Euclidean distance of transcriptomes from the transcriptome with the lowest value for PCA1 (open circle).

Figure 2. FLX-induced gene expression changes were the same in dorsal and ventral dentate samples.

Gene expression profiling of the dorsal and ventral dentate gyrus of mice treated with FLX+CORT were compared to samples from mice treated with CORT only. Figure plots the log2-fold changes in gene expression levels from FLX+CORT samples relative to CORT only samples in dorsal (x-axis) versus ventral (y-axis) dentate gyrus. Log2 fold changes in gene expression levels were highly correlated (Pearson r = 0.89, p<0.0001) across regions.

Figure 3. PCA1 is inversely related to levels of antidepressant-sensitive behaviors.

Panels (a–b) plot immobility in the FST (y-axis) against PCA1 (x-axis) for the dorsal (a, Spearman r = −0.63, p = 0.004) and ventral (b, Spearman r = −0.63, p = 0.004) dentate gyrus samples. Panels (c–d) plot latency to eat in the NSF (y-axis) against PCA1 (x-axis) for the dorsal (c, Spearman r = −0.81, p = 0.004) and ventral (d, Spearman r = −0.79, p = 0.004) dentate gyrus samples.