Basal forebrain cholinergic neurons are vulnerable in a mouse model of Down syndrome and their molecular fingerprint is rescued by maternal choline supplementation

Abstract

Basal forebrain cholinergic neuron (BFCN) degeneration is a hallmark of Down syndrome (DS) and Alzheimer's disease (AD). Current therapeutics in these disorders have been unsuccessful in slowing disease progression, likely due to poorly understood complex pathological interactions and dysregulated pathways. The Ts65Dn trisomic mouse model recapitulates both cognitive and morphological deficits of DS and AD, including BFCN degeneration and has shown lifelong behavioral changes due to maternal choline supplementation (MCS). To test the impact of MCS on trisomic BFCNs, we performed laser capture microdissection to individually isolate choline acetyltransferase-immunopositive neurons in Ts65Dn and disomic littermates, in conjunction with MCS at the onset of BFCN degeneration. We utilized single population RNA sequencing (RNA-seq) to interrogate transcriptomic changes within medial septal nucleus (MSN) BFCNs. Leveraging multiple bioinformatic analysis programs on differentially expressed genes (DEGs) by genotype and diet, we identified key canonical pathways and altered physiological functions within Ts65Dn MSN BFCNs, which were attenuated by MCS in trisomic offspring, including the cholinergic, glutamatergic and GABAergic pathways. We linked differential gene expression bioinformatically to multiple neurological functions, including motor dysfunction/movement disorder, early onset neurological disease, ataxia and cognitive impairment via Ingenuity Pathway Analysis. DEGs within these identified pathways may underlie aberrant behavior in the DS mice, with MCS attenuating the underlying gene expression changes. We propose MCS ameliorates aberrant BFCN gene expression within the septohippocampal circuit of trisomic mice through normalization of principally the cholinergic, glutamatergic, and GABAergic signaling pathways, resulting in attenuation of underlying neurological disease functions.

Keywords: Alzheimer's disease; Down syndrome; RNA-seq; bioinformatics; laser capture microdissection; maternal choline supplementation; medial septum; selective vulnerability; trisomy.

Figure 1:

Overview of experimental workflow A. The MCS paradigm is shown, with dams and pups fed a normal choline (1.1 g/kg choline) or choline supplemented (+; 5.0 g/kg choline) diet during the perinatal period (E0-P21). Upon weaning (P21), pups are moved to a normal choline diet until sacrifice at 6 MO. B. Brains are immediately flash frozen, cryocut at 20 μm, adhered to PEN membrane slides and immunostained for ChAT-ir. C. LCM is performed on 500 individual MSN BFCNs under a 40X objective Scale bar 50 μm. Multiple sections are collected and combined for RNA purification. D. QC is performed on total RNA. RNA-seq library preparation is done and QC on resulting cDNA library for each sample. E. Illumina HiSeq 4000 single reads were performed by the NYUGSOM GTC. F. FastQ files were generated for the reads for each sample and bioinformatic analysis is initiated.

Figure 2:

Pipeline for bioinformatic workflow using RNA-seq libraries derived from LCM-captured MSN BFCNs. A. Workflow for RNA-seq bioinformatics pipeline. B. PCA shows overall gene expression profiles for each mouse, with mean expression for each group as a larger sized dot. (black= 2N, blue= Ts, grey = 2N+, light blue= Ts+) C. Heatmaps illustrate individual gene expression differences from each individual mouse with DEGs by genotype (2N and Ts; 2,510 DEGs) and disease diet (Ts+ and Ts; 2,098 DEGs) binned by upregulation (pink) or downregulation (green).

Figure 3:

DEGs are presented by genotype (Ts and 2N) and maternal diet (MCS and normal choline). A. Table showing the DEGs at each FDR and p-value cutoff from the total number of analyzed genes. B. Bar graphs highlighting genes per LFC bin upregulated (pink) and downregulated (green) by genotype and disease diet. C. Pie charts show percentage of DEGs for protein coding and ncRNAs. D. Volcano plots show upregulated and downregulated genes with individual genes noted per dot. Light green dots indicate p<0.01 downregulated, dark green dots indicate p<0.05 downregulated, light pink dots indicate p<0.01 upregulated, and dark pink dots indicate p<0.05 upregulated.

Figure 4:

Trisomic protein coding genes do not necessarily match copy number within vulnerable MSN BFCNs. A total of 73 genes (of 88) show quantifiable expression levels with 7 DEGs attaining statistical significance at 6 MO by genotype were rescued in disease diet (A), an additional 7 genes were only affected by genotype (B), and an additional 4 genes were only affected by disease diet (C). Grey vertical dashes delineate DEGs affected by genotype and rescued in disease diet, those significant only by genotype, and those significantly affected by disease diet. D. Violin plot of 73 triplicated HSA21 orthologs expressed in BFCNs via LFC by genotype and disease diet. Each triangle represents individual gene and LFC amount is represented by thickness of the violin. Key: *** p<0.001, ** p<0.01, * p<0.05, t = 0.05>p<0.1. ns= not significant.

Figure 5:

Bioinformatic assessment of vulnerable pathways in trisomic BFCNs by IPA and KEGG. A. Bar charts show select pathways identified by IPA dysregulated by genotype and treatment, with LFC represented by each bar (2N versus Ts white; Ts versus Ts+ dots), with z-score reflected in coloring (pink upregulated, green downregulated) B. Pathways uniquely identified by KEGG analysis are shown. C. Bar graph showing percent gene expression changes within pathways affected by genotype and diet; red = gene changes unique to genotype, purple = gene changes affected by genotype and diet, blue = genes modified by diet in disease. D. Bar graph shows the number of genes are affected by genotype, diet, or both. E. Cholinergic pathway is dysregulated by genotype and partially rescued by diet, highlighting specific cholinergic receptor subunits and associated proteins. F. Glutamate receptor signaling is dysregulated by genotype and partially rescued by disease diet, with NMDA, AMPA, kainite, and metabotropic receptors showing individual subunit gene expression changes. G. GABA receptor signaling is dysregulated by genotype and modulated by MCS. (C-G, colored red for genotype, blue for MCS, and purple for genes impacted by both genotype and diet).

Figure 6:

STRING analysis of select DEGs in Cytoscape A. Chrm2 was selected from the Cholinergic pathway to identify genotype DEGs within the interactome, with 26 DEGs directly linked the Chmr2 interactome. B. A total of 9 trisomic interactome genes are significantly modulated with 18 DEGs within the interactome not significantly modulated in disease diet. C. Grin2A was selected from the glutamatergic neurotransmission pathway to identify genotype DEGs within the interactome. D. A total of 25 trisomic interactome genes are significantly rescued by disease diet and 42 (of 67) genes were not significantly attenuated by disease diet in the Grin2A glutamatergic interactome. E. From the GABAergic pathway, the Gabrg2 interactome contains 54 DEGs. F. A total of 20 trisomic interactome genes are rescued by MCS in Ts BFCNs and 34 of the DEGs within the Gabrg2 interactome are not significantly attenuated following MCS exposure. Key: pink DEGs, upregulated; green DEGs, downregulated.

Figure 7:

DEGs by genotype and diet were interrogated in IPA for disease and functional correlations. A. Bar chart identifies key disease and behavioral functions that have underlying molecular changes due to genotype (no pattern) or disease diet (striped pattern). Z-scores of changes for genotype and disease diet are colored with green indicating downregulation and pink representing upregulation. B. Violin plots show LFC of DEGs unique to genotype (* red), genotype changes that overlap with disease diet (genotype** purple), disease diet that overlap with genotype (disease diet** purple), and disease diet (* blue) in four representative pathways of neurological function. C. Venn diagrams illustrate within motor dysfunction and movement disorders, cognitive impairment, and ataxia neurological functions for DEGs upregulated or downregulated (up arrow or down arrow) by disease in genotype (red) which are modulated by MCS (purple) and those that are only MCS responsive (blue).

Figure 8:

Select DEGs by genotype and diet were validated by qRT-PCR. A. LFC of the means for qPCR synthesis products were correlated with RNA-seq gene expression data for genotype (circles) and disease diet (diamonds). Trend lines were generated to show correlation between qRT-PCR and RNA-seq LFCs. Key: red dashes, genotype; blue dot/dash, disease diet. B. Violin plots show relative gene expression levels for the interrogated genes. Camk2a, Chrm2, Grin2a, and Mapk8 are significantly dysregulated via RNA-seq by genotype and significantly attenuated by MCS in Ts+ MSN BFCNs. Ngfr was trend level dysregulated via RNA-seq by genotype and significantly attenuated by MCS. qPCR validated the directionality of all these DEGs found by RNA-seq. Key: 2N, white; Ts, dark green; Ts+, light green; 2N+, gray.