Transcriptomic profiling of the human brain reveals that altered synaptic gene expression is associated with chronological aging

Abstract

Aging is a biologically universal event, and yet the key events that drive aging are still poorly understood. One approach to generate new hypotheses about aging is to use unbiased methods to look at change across lifespan. Here, we have examined gene expression in the human dorsolateral frontal cortex using RNA- Seq to populate a whole gene co-expression network analysis. We show that modules of co-expressed genes enriched for those encoding synaptic proteins are liable to change with age. We extensively validate these age-dependent changes in gene expression across several datasets including the publically available GTEx resource which demonstrated that gene expression associations with aging vary between brain regions. We also estimated the extent to which changes in cellular composition account for age associations and find that there are independent signals for cellularity and aging. Overall, these results demonstrate that there are robust age-related alterations in gene expression in the human brain and that genes encoding for neuronal synaptic function may be particularly sensitive to the aging process.

Figure 1

WGCNA derived modules of gene expression. (a) An overall view of the network of gene expression. Height on the y-axes represents the dissimilarity of detected transcripts from the overall topology of gene expression. Colors below each line show the modules to which each transcript was assigned. (b) The eigengenes for each module are enriched for specific biological processes. For each module, we performed gene ontology (GO) enrichment, listing the term.id and name for each module (see also supplementary file 4 for p values). The module eigengenes showed correlation with each other, as indicated in the heatmap on the right. (c,d) Examples of modules enriched for biological terms. We used cytoscape to visualize organization of genes assigned to the GO category mitochondria organization in the cyan module (c) or the GO category ‘ensheathment of neurons’ in the red module (d). Circles are sized by the within module connectivity (Kwithin).

Figure 2

Genes that accumulate with aging in the brain include RNA metabolism and longevity genes. (a) The green module, with a correlation between module eigengene and age of 0.7 (p = 2 × 10⁻⁹), contained multiple transcripts enriched for GO terms related to RNA splicing (upper panel) and the nucleus (lower panel). Horizontal axes show the −log10 of the p value for enrichment for the listed terms on the vertical axis and bars are shaded by the number of genes overlapping between the green module and the number of genes in the GO term. (b) Visualization of the relationship of genes in the category GO:0000377 ‘RNA splicing, via trans-esterification reactions with bulged adenosine as nucleophile’ in the green module. Nodes are sized by the correlation with age for the most highly expressed transcript per given gene and linewidths indicate the strength of correlation between genes within the WGCNA analysis. (c,d) Individual associations with age for multiple transcripts for the genes NOL3 (c) and HNRNPA1 (d) with each transcript given a different color. Shaded regions indicate the 95% confidence intervals for the regression for each transcript. (e) The tan module (R for age 0.7, p = 2 × 10⁻⁹) showed enrichment for insulin signaling and longevity regulating genes. (f) Visualization of the kegg:04910 ‘Insulin signaling pathway’ within the tan module, with the addition of the highly associated gene RHBDL3. Nodes are sized by the correlation with age for the most highly expressed transcript per given gene and line widths indicate the strength of correlation between genes within the WGCNA analysis. (g,h) Individual associations with age for multiple transcripts for the genes PPP1R3E (g) and PPARGC1A (h), with each transcript given a different color. Shaded regions indicate the 95% confidence intervals for the regression for each transcript.

Figure 3

Diminished expression of transcripts encoding synaptic proteins with age. (a) GO enrichment for the brown module shows multiple strong enrichments with aspects of neuronal function, with the −log(P) for each GO term shown on the x-axis for multiple terms on the y-axis. Bars are shaded by the number of genes from the blue module that overlap with the GO term for each category. (b) Visualization of the GO:0048167 term ‘regulation of synaptic plasticity’ within the brown module. Nodes are sized by the inverse of the correlation with age for the most highly expressed transcript per given gene and line widths indicate the strength of correlation between genes within the WGCNA analysis. (c–e) Individual associations with age for multiple transcripts for the genes GRIA1 (c), GRIN2A (c) and NCAM1 (d), with each transcript given a different color. Shaded regions indicate the 95% confidence intervals for the regression for each transcript.

Figure 4

Validation of age associations against GTEx data and regional events (a) Overall consistency between the correlations of gene expression and age in our dataset (x axis) and the GTEx frontal cortex (y axis). Each point represents a different gene and is colored by the WGCNA module in our data and sized by −log10 of the p value for association with age, (b) Correlations between age associations in different brain regions in the GTEx dataset. For each pair of brain regions, the overall correlations between age associations are colored by strength; actual values of R are given in each box. (c) Counts of numbers of genes showing significant (nominal p < 0.05) associations with age in each brain region along the horizontal axis, with negative associations in red and positive associations in blue. (d–f) Specific genes showing significant associations with aging from our dataset tested in the GTEx brain regions, specifically excitatory glutamate receptor (d), inhibitory GABA genes (e) and genes related to dopamine neurotransmission (f). The vertical axis shows the correlation with age for each of three genes in different colors for different brain regions in the GTEx dataset along the horizontal axis. Points are sized to the mean expression (as FPKM) in each brain region and filled circles indicate significant (nominal p < 0.05) associations with age.

Figure 5

Population-specific expression analysis disambiguates age-related changes in cellularity from cell-dependent gene expression. (a) Correlation (GS) between age and transcript expression for genes representative of the major classes of cells in the CNS (horizontal axis) or a similar number of randomly selected genes. Genes are sized by the −log10 of the p value for association with age and colored by the WGCNA module assignments in our RNA-Seq dataset; note that all available transcripts were used for each gene. (b) Violin plot of the distribution of values of R with age (vertical axis) for all transcripts in a given WGCNA module (horizontal axis) either without (lighter colors outlined in black) or after correction for the proportion of neuronal cells in each sample (darker colors). (c) Comparison of uncorrected (horizontal axis) and neuron-corrected (vertical axis) associations with age for all transcripts in the brown, synaptic, module, that showed an overall negative correlation with age. Points are sized to the p value for association with age in the corrected samples. (d,e) Associations with age remain for specific transcripts after correction for neuronal proportions. The residuals after linear regression for neuronal markers are plotted on the vertical axis for each sample of different ages on the horizontal axis for transcript uc011dcy.2 (GRIA1, d) and uc010uym.2 (GRIN2A, e).