Identification of key proteins in early-onset Alzheimer's disease based on WGCNA

Abstract

Introduction: Early-onset Alzheimer's disease (EOAD) is sporadic, highly heterogeneous, and its underlying pathogenic mechanisms remain largely elusive. Proteomics research aims to uncover the biological processes and key proteins involved in disease progression. However, no proteomic studies to date have specifically focused on EOAD brain tissue.

Method: We integrated proteomic data from brain tissues of two Alzheimer's disease (AD) cohorts and constructed a protein co-expression network using weighted gene co-expression network analysis (WGCNA). We identified modules associated with EOAD, conducted functional enrichment analysis to understand the biological processes involved in EOAD, and pinpointed potential key proteins within the core modules most closely linked to AD pathology.

Results: In this study, we identified a total of 2,749 proteins associated with EOAD. Through protein co-expression network analysis, we discovered 41 distinct co-expression modules. Notably, the proteins within the core module most closely linked to AD pathology were significantly enriched in neutrophil degranulation. Additionally, we identified two potential key proteins within this core module that have not been previously reported in AD and validated their expression levels in 5xFAD mice.

Conclusion: In summary, through a protein co-expression network analysis, we identified EOAD-related biological processes and molecular pathways, and screened and validated two key proteins, ERBB2IP and LSP1. These proteins may play an important role in the progression of EOAD, suggesting they could serve as potential therapeutic targets for the disease.

Keywords: WGCNA; biological process; early-onset Alzheimer disease; proteome; therapeutic target.

Figure 1

Workflow to identify the key proteins of EOAD.

Figure 2

Proteomic characteristics of EOAD brain tissues. (A) Differentially expressed proteins were analyzed between groups within the cohort using volcano plots. p-values were calculated using students’ t test. A bold horizontal dashed line indicates the adjusted p-value threshold of <0.05. The number of significantly downregulated (blue) and upregulated (red) proteins is displayed at the top. (B,C) GO enrichment (B) and pathway enrichment (C) by GSEA.

Figure 3

Protein network analysis of EOAD brain tissues. (A) Soft threshold (power = 13.5) and mean connectivity (R² = 0.8) in the cohort. (B) Protein hierarchy tree-clustering diagram. The diagram represents different proteins horizontally and the correlation between them vertically. The lower the branch, the stronger the protein correlation within the branch. (C) The protein co-expression network consists of 41 protein modules, composed of 5,612 proteins from two cohorts. (D) Biweight mid-correlation (BiCor) analysis was conducted to assess the relationship between module eigenproteins and neuropathological features of AD (top), including CERAD scores and Braak staging. The strength of positive correlations is represented in red, while negative correlations are depicted in blue, with asterisks indicating statistical significance (p < 0.05). Cell-type associations for each protein module were evaluated using hypergeometric Fisher exact test (FET) (bottom). p-values resulting from FET were adjusted using the Benjamini-Hochberg (BH) method. The intensity of red shading reflects the degree of cell type enrichment, with asterisks indicating statistically significant differences (p < 0.05).

Figure 4

Identification of key proteins for EOAD (A) Boxplot show the level of eigenprotein between the CN and EOAD groups of M3. Student’s t test was used for comparison between the two groups. (B) Functional enrichment analysis of M3 module protein. Functional enrichment was performed using Metascape. For analysis, the top 20 most significantly enriched items were selected for presentation. Darker colors indicate a higher degree of enrichment. (C,D) Correlation of M3 module with CERAD score (C) and BRAAK staging (D). Correlation coefficients and p-values were calculated by Pearson correlation analysis. (E,F) The protein significance of CERAD score (E) and BRAAK stage (F) in M3 module ranked in the top 10 proteins, respectively.

Figure 5

Expression of key proteins in 5xFAD mice brain tissue. (A) Representative immunoblots of ERBB2IP and LSP1 in the brain tissue. Internal protein was normalized to β-actin or GAPDH. (B) Quantification of ERBB2IP and LSP1 in the brain tissue. * means a p-value ≤0.05, **p means a p-value ≤0.01 (Students’ t test, n = 6 per each group).