Integrated analysis on transcriptome and behaviors defines HTT repeat-dependent network modules in Huntington's disease

Abstract

Huntington's disease (HD) is caused by a CAG repeat expansion in the huntingtin (HTT) gene. Knock-in mice carrying a CAG repeat-expanded Htt will develop HD phenotypes. Previous studies suggested dysregulated molecular networks in a CAG length genotype- and the age-dependent manner in brain tissues from knock-in mice carrying expanded Htt CAG repeats. Furthermore, a large-scale phenome analysis defined a behavioral signature for HD genotype in knock-in mice carrying expanded Htt CAG repeats. However, an integrated analysis correlating phenotype features with genotypes (CAG repeat expansions) was not conducted previously. In this study, we revealed the landscape of the behavioral features and gene expression correlations based on 445 mRNA samples and 445 microRNA samples, together with behavioral features (396 PhenoCube behaviors and 111 NeuroCube behaviors) in Htt CAG-knock-in mice. We identified 37 behavioral features that were significantly associated with CAG repeat length including the number of steps and hind limb stand duration. The behavioral features were associated with several gene coexpression groups involved in neuronal dysfunctions, which were also supported by the single-cell RNA sequencing data in the striatum and the spatial gene expression in the brain. We also identified 15 chemicals with significant responses for genes with enriched behavioral features, most of them are agonist or antagonist for dopamine receptors and serotonin receptors used for neurology/psychiatry. Our study provides further evidence that abnormal neuronal signal transduction in the striatum plays an important role in causing HD-related phenotypic behaviors and provided rich information for the further pharmacotherapeutic intervention possibility for HD.

Keywords: Behaviors; CAG repeat; Huntington's disease; Mice; Single-cell RNA sequencing; Small chemicals; Striatum; Transcriptome.

Figure 1

Overview of the experimental design and data analysis strategy. An allelic series of CAG repeat lengths of the transcriptome data of 445 mRNA samples, 445 microRNA samples, 396 PhenoCube behaviors, 111 NeuroCube behaviors were analyzed by consensus weighted gene coexpression network analysis (WGCNA), followed by gene ontology network analysis, to address the disease mechanisms in the WGCNA detected modules.

Figure 2

Integrated effect of the gene expression and behavioral features in the striatum at 10 months. (A) The gene co-expression network and (B) bar plot of the module significance defined as the mean gene significance across all genes in the module. MEbrown and MEturquoise are the two modules with a correlation of more than 0.6 detected in the striatum at 10 months.

Figure 3

Module–behavioral feature associations in the striatum at 10 months of the PhenoCube behavior (A) and NeuroCube behavior (B). Each row corresponds to a module eigengene and each column to a behavioral feature. Each cell contains the corresponding correlation and P-value. The full names of the behaviors are listed in Table S1.

Figure 4

Networks shared by the CAG repeat length and behaviors. (A) A scatter plot of the gene significance for CAG repeat length vs. module membership in the brown module. (B) Gene ontology networks for the CAG repeat length. (C–G) Gene networks of the significant ontologies for the CAG repeat length. Blue lines, physical interactions. Brown lines, co-localization. Green lines, pathway. Dark pink lines, coexpression. Light green lines, shared protein domains. (C) Postsynaptic density (GO: 0014069), (D) Calcium, (E) Signal transduction inhibitor function, (F) Amphetamine ontology, and (G) Neuronal cell body ontology (GO: 0043025).

Figure 5

The graphical representation of the DAVID FDR of the enriched functional terms for gene modules that are associated with representative PhenoCube and NeuroCubebe behaviors.

Figure 6

Highlighted networks in the grouping behaviors in the PhenoCube platform. (A) Visualization of the eigengene network representing the relationships between the modules and the number of groupings. The upper panel shows a hierarchical clustering dendrogram of the eigengenes in which the dissimilarity of eigengenes EI and EJ is given by 1 - cor (EI; EJ). The heat map in the lower panel shows the eigengene adjacency AIJ = (1 + cor (EI; EJ))/2. (B) A scatter plot of the gene significance for number of groupings vs. module membership in the brown module. (C) Gene ontology networks for number of groupings. (D–H) Gene networks of the significant ontologies for the number of grouping. Blue lines, physical interactions. Brown lines, co-localization. Green lines, pathway. Dark pink lines, coexpression. Light green lines, shared protein domains. (D) Dendritic spine, (E) Synapse, (F) S_TKc domain, (G) Cell junction, and (H) Oxytocin signaling pathway.

Figure 7

Highlighted networks in gait characteristics and number of steps in the NeuroCube platform. (A) Visualization of the eigengene network representing the relationships between the modules and the number of steps. The upper panel shows a hierarchical clustering dendrogram of the eigengenes in which the dissimilarity of eigengenes EI and EJ is given by 1 – cor (EI; EJ). The heat map in the lower panel shows the eigengene adjacency AIJ = (1 + cor (EI; EJ))/2. (B) A scatter plot of the gene significance for number of steps vs. module membership in the brown module. (C) Gene ontology networks for number of steps. (D–F) Gene networks of the significant ontologies for number of steps. Blue lines, physical interactions. Brown lines, co-localization. Green lines, pathway. Dark pink lines, coexpression. Light green lines, shared protein domains. (D) Postsynaptic membrane, (E) Locomotor, and (F) Perikaryon.

Figure 8

Highlighted networks in the gait characteristics of the hind limb angle (hind limb mean angle of orientation of the maximal diameter to the direction of the run [X]) in the NeuroCube platform. (A) A scatter plot of the gene significance for number of hind limb angles vs. module membership in the brown module. (B) Gene ontology networks for the hind limb angle. (C, D) Gene networks of the significant ontologies for the hind limb angle. Blue lines, physical interactions. Brown lines, co-localization. Green lines, pathway. Dark pink lines, coexpression. Light green lines, shared protein domains. (C) Cytoskeleton and (D) Proton acceptor.

Figure 9

Single-cell-based RNA expression patterns of the HTT repeat-dependent behavior genes in the Striatum cell population. The cell population expression patterns of genes in the network of CAG repeat length (A), number of grouping (B), number of steps (C), and hind limb angle (D). Ependy-C: Ependy cilia; NSCs: neuronal stem cells; OPCs: oligodendrocyte precursor cells.