Fig 1.
Comparing methodology of original paper and present study and Abeta expressing C. elegans transgenic strains.
a) The model expresses human Abeta that results in alteration phenotypes (please refer to Material and Methods section), b) C. elegans strains used by Hassan and colleagues, c) Experimental design and methodology by Hassan and colleagues. Briefly, synchronized eggs of transgenic worms were grown at 16˚C for 36 h and shifted to 25˚C to activate the smg-1 gene. Worm samples were collected at 4 hours intervals from T0 (temperature shift time) to T20 (before worm paralysis), and analysis including microarray, PCR, gene ontology, and RNA interference were performed in the original study. d) Pipelines used in this study to analyze publicly available datasets GSE65851, and datasets for incipient AD including GSE12685 and GSE28146.
Fig 2.
C. elegans transcriptome and gene ontology enrichment analyses.
a) Principle Component Analysis (PCA) of the samples. b) A bar graph for the number of differentially expressed genes at each time-point and the number of up and down-regulated genes. Maximum deregulation of genes occurred at transition from T12 to T16. c) A Venn diagram for the comparison of the DEGs among time-points after Abeta expression. d) Affected biological processes in each time-point comparison. The numbers close to the bars indicate the percentage of DEGs that are involved in each process. e) Enrichment map of biological processes, in which nodes are biological processes and edges show common genes between two nodes. Larger nodes represent those processes that contain more DEGs. Main process in each network is indicated by red star. There was no enrichment network for the T8 vs. T4 comparison. In PCA graph large dots are average of all the sample of that group.
Fig 3.
C. elegans transcriptome analysis of Abeta versus GFP samples, PPI networks, and gene ontology enrichment.
a) DEGs in each time-point. b) Venn diagram of common DEGs among time-points. c) Bar graph of biological processes that DEGs are involved in. d) PPI network of T8. e) PPI network of T12. f) PPI network of T16. g) Table of top three biological processes in the modules from networks. Note—the biological processes with P-value less than 0.05 were selected. In the network, modules are indicated by numbers. Nodes and labels in larger size indicate higher centrality-based degree factor. Thicker edges indicate higher weight. There was no network for time-points T8 and T12.
Fig 4.
PPI networks and gene ontology of module members obtained for each network.
a) PPI network of T8 vs. T4 stage. b) PPI network of T12 vs. T8 stage. c) PPI network of T16 vs. T12 stage. d) PPI network of T20 vs. T16 stage. e) Table contains information regarding the top three biological processes found for the genes in the modules detected in each network. The biological processes with p-value less than 0.05 were selected. In the network, modules are indicated by numbers. Red and blue colors of nodes represent up- and down-regulation, respectively. Nodes and labels in larger size indicate higher centrality based on degree of connectivity factor. Thicker edges indicate higher weight.
Fig 5.
a) Hierarchical clustering of the samples and soft-power detection. The graph indicates that soft-power above 10 meets scale free topology above 0.8. b) Module detection and merging modules with 75 similarities in eigengenes. c) Module-trait relationship heatmap. Only those modules with high relationships (0.3<) were selected for further analysis. d) Module membership vs. gene significance of the highly correlated modules with specific time-points. Only modules with p-value ≤ 1e-5 were considered for further analysis.
Fig 6.
Networks of the top 10 percent of most central genes in selected modules identified by WGCNA.
Identified modules were labeled as a) “antiquewhite4”, b) “brown4”, c) “cyan”, d) “darkgrey”, e) “darkolivegreen2”, and f) “purple”. Red and blue colors of nodes represent up- and down-regulation, respectively. Grey nodes represent no DEGs for that gene. Nodes and labels in larger size indicate higher centrality in the network based on degree factor. Thicker edges indicate higher weight in terms of network analysis. Only fifty percent of edges are shown.
Table 1.
Top five biological processes with lowest p-value for detected modules by WGCNA.
Fig 7.
Clustering of TFs in transcriptome data obtained from Abeta expressing C. elegans.
a) Heatmap showing clustering of all TFs based on similarity approach. We marked two clusters on the heatmap based on their expression pattern through successive stages. Either upregulation at T12 (these DE-TFs show over expression in at least three stages) or marked a group of TFs with low expression at the early time-points and over expressing in later stages. To highlight these differences, we have included b) The clustering of DE-TFs among at least three stages by correlation method. c) Clustering of up-regulated DEGs in T12 vs. T8 and T16 vs. T12 by correlation method. d and e) Show the top three biological processes (p-value ≤ 0.05) that DE-TFs and Over expressing TFs are involved in, respectively.
Fig 8.
WGCNA analysis of the comparison between GFP and Abeta-GFP accumulation.
There are clearly three modules related to Abeta accumulation. The GO of the members of these modules are presented in the S3 Fig Settings and thresholds are as Fig 5.
Fig 9.
Hub genes from modules related to the Abeta expressing C. elegans.
For each gene, expression trends through time-points (every 4 hours) and the mouse and human homologous genes were detected. For identified homologous genes, expression status in two datasets from human incipient Alzheimer’s disease, the direct relationship with Alzheimer’s disease, any potential role in Alzheimer’s or any neurodegenerative disease and Abeta presence were investigated. For each gene, higher intensity in color shows higher expression. NF indicates homologous genes are not found. Color bars indicate the trend of gene expression through time points.
Fig 10.
The top hub genes (Maximum 10) detected in the PPI networks at different stages of Abeta accumulation have been listed.
For each gene, expression trends through time-points (every 4 hours), mouse and human homologous genes were extracted and presented. For identified homologous genes, expression status in two datasets from human incipient Alzheimer’s disease, the direct relationship with Alzheimer’s disease, any potential role in Alzheimer’s or association with Abeta accumulation were also included. For each gene, higher intensity in color shows higher expression. NF indicates homologous genes could not be found. * TFs responding to Abeta accumulation through development, ** TFs uniquely responding to Abeta accumulation when compared to GFP sample at the same developmental stage. Brown and green bars indicate trends of expression in response to GFP and Abeta accumulation, respectively. Darker color indicates more expression.
Table 2.
Encoded protein, gene ontology and anatomical location of hub genes in worm and biological processes of their orthologues in human.
All the Information are extracted from Uniprot (https://uniprot.org) and Wormbase (https://wormbase.org) databases. Anatomical location indicates cell/organ/regions that the genes are expressed in. Dash lines means no information has been found.
Fig 11.
Transcriptome analysis of samples obtained from human data at incipient Alzheimer’s disease.
We have analyzed two datasets, GSE12685 and GSE28146. PCA was performed, which as expected showed noticeable heterogeneity among samples. DEGs were detected using GEO2R tool. 77 common DEGs were detected between the two datasets. Enrichment map of biological processes show that transport (transmembrane) and actin organization are common in both datasets (indicate by red stars). Nodes in larger size contain a larger number of involved DEGs in the biological process. In the PCA graph, large dots are average of all the samples of that group.