Genes and pathways underlying regional and cell type changes in Alzheimer's disease

Abstract

Background: Transcriptional studies suggest Alzheimer's disease (AD) involves dysfunction of many cellular pathways, including synaptic transmission, cytoskeletal dynamics, energetics, and apoptosis. Despite known progression of AD pathologies, it is unclear how such striking regional vulnerability occurs, or which genes play causative roles in disease progression.

Methods: To address these issues, we performed a large-scale transcriptional analysis in the CA1 and relatively less vulnerable CA3 brain regions of individuals with advanced AD and nondemented controls. In our study, we assessed differential gene expression across region and disease status, compared our results to previous studies of similar design, and performed an unbiased co-expression analysis using weighted gene co-expression network analysis (WGCNA). Several disease genes were identified and validated using qRT-PCR.

Results: We find disease signatures consistent with several previous microarray studies, then extend these results to show a relationship between disease status and brain region. Specifically, genes showing decreased expression with AD progression tend to show enrichment in CA3 (and vice versa), suggesting transcription levels may reflect a region's vulnerability to disease. Additionally, we find several candidate vulnerability (ABCA1, MT1H, PDK4, RHOBTB3) and protection (FAM13A1, LINGO2, UNC13C) genes based on expression patterns. Finally, we use a systems-biology approach based on WGCNA to uncover disease-relevant expression patterns for major cell types, including pathways consistent with a key role for early microglial activation in AD.

Conclusions: These results paint a picture of AD as a multifaceted disease involving slight transcriptional changes in many genes between regions, coupled with a systemic immune response, gliosis, and neurodegeneration. Despite this complexity, we find that a consistent picture of gene expression in AD is emerging.

Figure 1

Differential expression results are consistent with previous studies. (a) Genes showing enrichment for CA1 or CA3 in previous studies of AD show similar enrichment in this study (Lein et al. [29], Torres-Muñoz et al. (TM) [31]). The y-axis shows the percentage of genes with consistent results between studies. The x-axis shows the subset of genes used: EXP, 'highly expressed' (average expression >1,000); SIG, 'significant' differential expression (P < 0.05); VERY SIG, 'highly significant' differential expression (P < 0.005). (b) Example mouse in situ hybridizations for common CA1- and CA3-enriched genes were reproduced from the Allen Mouse Brain Atlas (Allen Institute for Brain Science, ©2009 [67]). Bars represent the relative expression levels in CA1 and CA3 in our data. Error bars show standard error. (c) Common region-enriched genes in this study and in the Allen Human Brain Atlas [68]. Points correspond to the ratio of the average CA3 versus CA1 expression in both studies on a log2 scale. See also Figure S3 in Additional file 6. (d) Genes show similar correlations to disease (as measured by Braak stage) between this study and a previous study of similar design [3]. The y-axis shows the correlation of this measure across genes between studies. The x-axis labeled as in (a) (ALL, all genes). Data corresponding to all genes is presented in the inset and Figure S4 in Additional file 6. (e) Genes showing significant disease alteration in CA1 in three previous studies tend also to change in the same direction with AD in this study (Blalock et al. [3], Colangelo et al. (Colang.) [4], Liang et al. [20]). Bars represent the level of consistency between our results and the labeled list of differentially expressed genes (y-axis). Mean gene rank (x-axis) scales from 0.5 (completely opposite results) to 0.5 (perfectly consistent) with chance = 0 (see Materials and methods). P-values: *P < 0.05, ** P < 0.006, *** P < 0.00001, **** P < 10^-45. (f) Genes showing significant region-enrichment in control in three previous studies tend to show similar regional enrichment in this study. Labeling as in (e) (Newrzella et al. (Newr.) [30]).

Figure 2

Gene-by-region study design provides novel insights into AD. (a) Region (x-axis) and disease (y-axis) T-statistics for each gene (point) are plotted, along with the line of best fit. We find a significant correlation between disease and region. The number of these genes differentially expressed with both region and disease (four corners) are displayed in grey, with P-values representing the significance of enrichment (in dark grey) or depletion (in italics). Dashed lines correspond to significant differential expression (P < 0.05). (b) Representative genes for each gene-by-region pattern of expression. Box plots of gene expression levels (y-axis) are displayed for each of the four groups (x-axis): CA1 in control (C1), CA3 in control (C3), CA1 in AD (A1), and CA3 in AD (A3). (c) The four vulnerability genes show higher expression in CA1 than CA3 and also increase with AD to a larger degree in CA1 compared with CA3. Labeling as in (b). (d) The three protection genes show higher expression in CA3 than CA1 and also increase with AD to a larger degree or decrease with AD to a smaller degree in CA3 compared with CA1. Note that there are two significant probes for UNC13C. Labeling as in (b).

Figure 3

Modules for cell type associate with disease-relevant phenotypes. (a-d) Representative modules for four major cell types - pyramidal neuron (a), astrocyte (b), oligodendrocyte (c), and microglia (d) - each show significant association with a disease-relevant trait. In the first column, module association with region and disease is measured using Bayes ANOVA. Box plots are displayed for each of the four groups (x-axis, labeling as in Figure 2b). In the second column, module association with Braak stage in controls is measured using a t-test. Box plots are displayed for Braak stages (Stg) of 1 and 2. In the third column, Pearson correlation between module expression and age is presented, along with the line of best fit. The y-axis in all cases represents module eigengene expression. P-values: +, 0.05 <P < 0.1; **, P < 0.007; ***, P < 0.0004. In the fourth column, mouse in situ hybridizations for the top hub gene in each module were reproduced from the Allen Mouse Brain Atlas (Allen Institute for Brain Science, ©2009, available from [67]). These genes appear to mark the appropriate cell types, although no region-specificity is seen in any case. Note that PPAP2B is the top hub gene for a different astrocyte module (light cyan).

Figure 4

Most genes showing increased expression with Braak stage are confirmed with qRT-PCR. (a) Box plots showing expression levels (y-axis) for five of the top genes differentially expressed with Braak stage in control (x-axis). P-values of differential expression were measured using a t-test (**, P < 0.003; ***, P < 0.0004). (b) Fold changes (y-axis) for each of these genes between Braak stages of 1 and 2 were calculated using three methods (x-axis): microarray (M), qRT-PCR of tissue from hippocampus (HP), and qRT-PCR of tissue from frontal cortex (C). Genes were rated as confirmed (tick; fold change >1.2), marginal (minus sign; 1.1 < fold change < 1.2), and non-confirmed (cross; fold change <1.1).

Figure 5

Genes in the microglial module are related to early NFT pathology. The top 250 gene-gene interactions of the light green module are displayed as measured by topological overlap. Larger dots represent hub genes with at least 15 connections. Circled genes are significantly up-regulated in NFT-bearing neurons (P < 0.03) and also overexpressed in Braak stage 2 controls (P < 0.03). The length of each line and the position of each node were arbitrarily chosen by VisANT to highlight network structure. The top 25 genes based on module membership are also presented.