Promoter Hypomethylation of TGFBR3 as a Risk Factor of Alzheimer's Disease: An Integrated Epigenomic-Transcriptomic Analysis

Abstract

Alzheimer's disease (AD) is characterized by the abnormal deposition of amyloid-β (Aβ) plaques and tau tangles in the brain and accompanied with cognitive impairment. However, the fundamental cause of this disease remains elusive. To elucidate the molecular processes related to AD, we carried out an integrated analysis utilizing gene expression microarrays (GSE36980 and GSE5281) and DNA methylation microarray (GSE66351) in temporal cortex of AD patients from the Gene Expression Omnibus (GEO) database. We totally discovered 409 aberrantly methylated and differentially expressed genes. These dysregulated genes were significantly enriched in biological processes including cell part morphogenesis, chemical synaptic transmission and regulation of Aβ formation. Through convergent functional genomic (CFG) analysis, expression cross-validation and clinicopathological correlation analysis, higher TGFBR3 level was observed in AD and positively correlated with Aβ accumulation. Meanwhile, the promoter methylation level of TGFBR3 was reduced in AD and negatively associated with Aβ level and advanced Braak stage. Mechanically, TGFBR3 might promote Aβ production by enhancing β- and γ-secretase activities. Further investigation revealed that TGFBR3 may exert its functions via Synaptic vesicle cycle, Calcium signaling pathway and MAPK signal pathway by regulating hub genes GNB1, GNG3, CDC5L, DYNC1H1 and FBXW7. Overall, our findings highlighted TGFBR3 as an AD risk gene and might be used as a diagnostic biomarker and therapeutic target for AD treatment.

Keywords: Alzheimer’s disease; Aβ plaque; TGFBR3; methylation; secretase activity.

FIGURE 1

Identification of aberrantly methylated and differentially expressed genes in temporal cortex of patients with AD. GEO2R was used to analyze the DNA methylation profiling of methylation dataset 1 and gene expression profiling of expression dataset 1 and 2. The aberrantly methylated genes and differentially expressed genes were identified according to the cutoff criteria |t| > 2 and p < 0.05. Then Venn diagram analysis was performed to aberrantly methylated and differentially expressed genes in the three datasets (A) Hypermethylated-downregulated genes (B) Hypomethylated-upregulated genes.

FIGURE 2

Gene functional enrichment analysis. GO annotation and KEGG pathway enrichment analysis of the aberrantly methylated and differentially expressed genes was analyzed by Metascape (A) Bar graph of the top 20 enriched terms across, colored by p-values (B) Network of the top 20 enriched terms and (C) MCODE components was identified from the PPI network.

FIGURE 3

Expression cross-validation of the candidate genes in AlzData database (A) NPTX2 (B) RTN1 (C) UBE2N (D) MEF2C (E) IQGAP1 (F) TGFBR3. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 4

High TGFBR3 level was significantly associated with Aβ accumulation (A-F) Pearson correlation analysis was used to evaluate the association between Aβ level and gene expression levels of NPTX2 (A), RTN1 (B), UBE2N (C), MEF2C (D), IQGAP1 (E) and TGFBR3 (F) in 68 brain temporal cortex tissues from the Allen Brain Atlas (G) Representative immunohistochemistry staining data for Aβ. Scale bars, 100 μm.

FIGURE 5

The promoter methylation level of TGFBR3 was reduced in AD and negatively correlated with advanced Braak stage (A) The methylation probes of TGFBR3 (NM_003,243.5) annotated by the USUC genome browser (B) Methylation levels of methylation probes for TGFBR3 in AD tissues and controls from methylation dataset 1 (C) The ROC curves for predicting AD by the cg17074213 (D) The correlation analysis of cg17074213 level and Braak stage (E) The correlation analysis of cg17074213 level and age (F) The ROC curves for predicting AD by the cg09790580 (G) The correlation analysis of cg09790580 level and Braak stage (H) The correlation analysis of cg09790580 level and age. ns: not significant, *p < 0.05, ***p < 0.001.

FIGURE 6

TGFBR3 expression positively correlated with β- and γ-secretase activities (A-F) Pearson correlation analysis was used to evaluate the association between TGFBR3 level and Aβ42 level (A), α-secretase activity (B), β-secretase activity (C), γ-secretase activity (D), Braak stage (E) and Age (F) in 55 brain temporal cortex tissues from the expression dataset 3 (G) TGFBR3 level in male and female (H) TGFBR3 level in different APOE genotypes. ns p > 0.05.

FIGURE 7

TGFBR3-related genomic alterations. A total of 1824 TGFBR3-related genes were identified based on the screening criteria |r|>0.6 and p < 0.05 from expression dataset 3. Then the KEGG pathway enrichment analysis was conducted by Metascape (A) Bar graph of the top 20 enriched KEGG pathways of TGFBR3-related genes, colored by p-values (B) The PPI network of TGFBR3-related genes was constructed by STRING v11.5 and visualized by Cytoscape v3.8.2. hub genes were identified from the network according to the degree value. Then the degree score of each gene in network was calculated by cytoHubba and the 10 genes with the highest degree score were identified as hub genes.

FIGURE 8

Validation of the hub genes (A) Expression levels of hub genes in AD and controls from AlzData database (B) Pearson correlation analysis was used to evaluate the association between and Aβ42 level and hub gene expression level in expression dataset 3 (C) Pearson correlation analysis was used to evaluate the association between and Braak stage and hub gene expression level in expression dataset 3 (D) Venn diagram analysis. NA, data are not available; ns p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001.