Hippocampal CA1 Pyramidal Neurons Display Sublayer and Circuitry Dependent Degenerative Expression Profiles in Aged Female Down Syndrome Mice

Abstract

Background:

Individuals with Down syndrome (DS) have intellectual disability and develop Alzheimer’s disease (AD) pathology during midlife, particularly in the hippocampal component of the medial temporal lobe memory circuit. However, molecular and cellular mechanisms underlying selective vulnerability of hippocampal CA1 neurons remains a major knowledge gap during DS/AD onset. This is compounded by evidence showing spatial (e.g., deep versus superficial) localization of pyramidal neurons (PNs) has profound effects on activity and innervation within the CA1 region.

Objective:

We investigated whether there is a spatial profiling difference in CA1 PNs in an aged female DS/AD mouse model. We posit dysfunction may be dependent on spatial localization and innervation patterns within discrete CA1 subfields.

Methods:

Laser capture microdissection was performed on trisomic CA1 PNs in an established mouse model of DS/AD compared to disomic controls, isolating the entire CA1 pyramidal neuron layer and sublayer microisolations of deep and superficial PNs from the distal CA1 (CA1a) region.

Results:

RNA sequencing and bioinformatic inquiry revealed dysregulation of CA1 PNs based on spatial location and innervation patterns. The entire CA1 region displayed the most differentially expressed genes (DEGs) in trisomic mice reflecting innate DS vulnerability, while trisomic CA1a deep PNs exhibited fewer but more physiologically relevant DEGs, as evidenced by bioinformatic inquiry.

Conclusions:

CA1a deep neurons displayed numerous DEGs linked to cognitive functions whereas CA1a superficial neurons, with approximately equal numbers of DEGs, were not linked to pathways of dysregulation, suggesting the spatial location of vulnerable CA1 PNs plays an important role in circuit dissolution.

Keywords

Alzheimer’s disease bioinformatics CA1 circuitry Down syndrome hippocampus laser capture microdissection RNA-seq selective vulnerability trisomy

INTRODUCTION

Down syndrome (DS) results from the aneuploidy of human chromosome 21 (HSA21) and occurs in ∼1 in 700 births.1,2, 1,2 DS is the primary genetic cause of intellectual disability and results in the development of Alzheimer’s disease (AD) pathology early in midlife, including deposition of amyloid-β (Aβ) plaques, neurofibrillary tangles, and dysregulation of the endosomal-lysosomal system.3 –8 In addition, individuals with DS develop cognitive deficits related to septohippocampal circuit degeneration,5,9–12 , 5,9–12 and volume reductions in the hippocampus at an early age.13 –17 These pathologies result in neurological deficits, with impairments of language acquisition, attention, and hippocampal dependent learning and memory.18 –21 Age related cognitive decline is associated with hippocampal degeneration in both DS and AD, with dementia the primary cause of death in persons with DS over 35 years of age.5,22–26 , 5,22–26 Females have a higher risk of developing AD with several biological variables postulated to underlie this sex difference. 27 However, the sex risk for developing AD in DS is equivocal.28 –31 Several studies suggest males have higher risk of developing AD in DS,29,31, 29,31 while other reports indicate women have higher dementia prevalence in DS.28,30, 28,30 Taken together, the study of relevant animal models and postmortem human brain tissues by sex is an important factor to understand the progression of AD in DS.

Several mouse models of DS and AD have been developed with partial or full triplication of murine orthologs to HSA21. The most widely used is the Ts65Dn mouse, a segmented trisomy model with ∼90 protein coding genes orthologous to HSA21, as a small translocation chromosome.32 –35 The Ts65Dn mouse model develops cognitive decline, degeneration in the septohippocampal projection system,36 –44 as well as hippocampal dependent spatial learning deficits, context discrimination and attentional memory.41,45–48 , 41,45–48 A novel derivative of the Ts65Dn mouse model of DS, the Ts[Rb(12.1716)]2Cje (Ts2) mouse, has a stable chromosomal rearrangement of the segmental trisomy. 49 The Ts2 model provides fidelity to the Ts65Dn phenotype, but does not carry the reproductive disadvantages,49,50, 49,50 and shows improved transmission of the chromosome through the germline.49 –52

Trisomic mice recapitulate human DS brain volume deficits and decreased neuronal density.53,54, 53,54 During aging, trisomic mice have progressive defects of the cholinergic septohippocampal fibers which innervate the superficial and glutaminergic entorhinal cortex neurons that, in turn, innervate the molecular layer of the dentate gyrus and stratum lacunosum moleculare (SLM) CA1 pyramidal neurons (PNs).55,56, 55,56 The septohippocampal circuit innervates the CA1 hippocampal region, which shows phenotypic loss of basal forebrain cholinergic neurons (BFCNs) and loss of hippocampal cholinergic tone, notably input deficits into the hippocampal formation from the medial septum/ventral diagonal band complex.38,54,57–59 , 38,54,57–59 These deficits begin at ∼6 months of age (MO), with behavioral and physiological deficits prevalent by ∼10 MO.44,46,53,54,60 , 44,46,53,54,60 From a translational perspective, synaptic disconnection of the perforant path system connecting the entorhinal cortex with the hippocampal formation is a hallmark of AD.12,57,61–64 , 12,57,61–64

Murine CA1 PNs have two functionally and circuitry distinct sublayers, a deep and superficial layer along the radial axis of the hippocampus, 65 that also differs along the transverse axis with regard to innervation and function. 66 CA1 deep neurons are more likely to form place cells65,67, 65,67 and in the CA1a region, receive less innervation from the lateral entorhinal cortex (LEC) compared to their CA1 superficial counterparts. 66 Conversely, CA1a superficial neurons receive a stronger innervation from the LEC, with significantly higher spine density than CA1 deep neurons, whereas both deep and superficial neurons in CA1a receive moderate innervation from medial entorhinal cortex (MEC). 66 While this mediolateral innervation pattern is postulated to have an effect on spatial and non-spatial learning and memory,55,68,69 , 55,68,69 there is a lack of knowledge linking gene expression and encoded protein levels to the learning and memory behaviors associated with these subregions.

Single population RNA sequencing (RNA-seq) of CA1 PNs via laser capture microdissection (LCM) classified CA1 PNs into discrete subpopulations based on gene expression markers in an epilepsy model, linking innervation to expression, in a sublayer specific manner. 70 While this murine model of epilepsy shows significant alterations in gene expression and cell type specificity in the hippocampus, no such studies have been conducted in DS/AD models. Importantly, DS/AD models recapitulate key aspects of hippocampal memory impairments, with translation to the human condition.

Trisomic mice have deficits in spatial and non-spatial learning and memory linked to the septohippocampal pathway and CA1 PNs,45,46,53 , 45,46,53 suggesting that both deep and superficial neurons are dysregulated. Place cells, mainly located in the CA1 deep layer 66 may have significant deficits underlying spatial memory impairments observed in trisomic mice during age-related decline. Since the CA1a region is also associated non-spatial memory, 66 gene expression changes may differ with CA1 neurons located in CA1a sublayers in the Ts2 model.

To examine whether gene expression is linked to system-based connectivity, we performed LCM on female Ts2 DS mice and normal disomic (2N) mice for deep or superficial neurons in the CA1a region compared to PNs microisolated from the entire CA1 region. Single population RNA-seq was performed with both overlapping and unique gene expression profiles specific to the DS phenotype, that are postulated to correlate with unique connectivity dysregulation. Previous work has shown sex effects on BFCN number and size in trisomic mice and disomic littermates, 71 and we focus on females in the present study. We hypothesize the Ts2 mouse model of DS/AD will reveal unique dysregulation in CA1 pyramidal neuron gene expression based on connectivity, and that regional CA1 pyramidal neuron sequencing recapitulates only a portion of these subtype-specific gene changes relevant to the pathophysiology of DS and AD.

MATERIALS AND METHODS

Mice

Animal protocols were approved by the Nathan Kline Institute/New York University Grossman School of Medicine (NYUGSOM) IACUC in accordance with NIH guidelines. Breeder pairs (female B6EiC3Sn-Rb(12.Ts17 16 65Dn)2Cje/CjeDnJ, strain ID 004850, herein Ts2 and male B6EiC3SnF1/J, strain ID 001875) mice were purchased from Jackson Laboratories (Bar Harbor, ME) and mated at the Nathan Kline Institute. Breeder pairs had ad libitum access to normal chow diet (AIN-76A, Dyets Inc., Bethlehem, PA) and water, with environmental enrichment in each cage consisting of a t-tube, nestlets, bedding (Biofresh Performance Plus, WF Fisher & Son, Somerville, NJ) and igloos. Pups were weaned at postnatal 21, given ad libitum food and water, with nestlets and igloos in each cage for enrichment.72 –75 Tail clips were taken and genotyped at weaning. 76 Mice were sacrificed at middle-age (∼11 MO), a timepoint at which pathology and behavioral deficits in Ts2 mice are well established.53,54, 53,54

Tissue preparation

Brain tissues were accessed from trisomic (Ts2; n = 11) and disomic (2N; n = 11) female mice (age range: 10.6–11.4 MO). Mice were transcardially perfused with ice-cold 0.15 M phosphate buffer as previously described.72,73,77,78 , 72,73,77,78 Brains were immediately flash frozen and cut on a cryostat (CM1860UV, Leica, Buffalo Grove, IL) at 20μm thickness at –25°C in the coronal plane and mounted on polyethylene naphthalate (PEN) membrane slides (Leica). Slides were kept under desiccant at –80°C until used for immunohistochemistry. RNase-free precautions were employed, and solutions were made with 18.2 mega Ohm RNAse-free water (Nanopure Diamond, Barnstead, Dubuque, IA).

Immunohistochemistry and neuron collection

For each brain, slides containing the rostral CA1 region was identified (Bregma –1.22 to –1.94) on dry ice. Tissue mounted on PEN membrane slides were equilibrated to room temperature (RT) under desiccant (–20°C for 5 min, 4°C for 10 min, RT for 5 min) followed by a rapid Nissl staining protocol to preserve intact RNA in the unfixed tissue. 79 Briefly, tissue sections were incubated ∼30 s at RT with 0.1% thionin in 49.5 mM sodium acetate (C₂H₃O₂Na) and 3.6 mM glacial acetic acid (Sigma, St Louis, MO), then washed twice in ice-cold phosphate buffered saline (PBS, pH 7.4). Slides were dried at RT 3–5 min before immediately taken for laser capture microdissection (LCM, LMD7000, Leica) as previously described. 75 An overview image was collected at 5x to identify the rostral CA1 target region. Isolation of the region of interest (ROI) was done using “draw and cut” feature at 20x with manual outlines drawn of the CA1 and/or CA1 sublayer ROIs. Once the ROIs were drawn, tissue was cut using the laser screw feature cutting in the z-plane twice at a depth of 5μm apart. Cut tissue was dropped via gravity into the collection device containing lysis buffer (Qiazol, Qiagen, Germantown, MD) with a maximum collection time of 1 hour to preserve RNA quality. Tissue collection was randomized with experimenter blind to genotype in two batches, with batch 1 (n = 6 per genotype) collected and processed prior to batch 2 (n = 5 per genotype). In each tissue section, CA1a-Deep (CA1D) and CA1a-Superficial (CA1S) PNs 66 were isolated by LCM (Fig. 1A-Biii-vi). For batch 1 CA1D and CA1S, 2 slides were employed (total of 12 sections; 6 per slide), with ∼1,500 CA1D and CA1S PNs collected for downstream processing. CA1-All (entire CA1 region, CA1A; Fig 1A-Bi-ii) was microisolated from a single slide adjacent to the CA1D/ CA1S dissection from 3 sections, totaling ∼500 CA1A PNs collected per brain before proceeding to RNA isolation and RNA-seq library preparation. Batch 2 samples were collected in the 3 sublayers and done in concert from the same slides in adjacent sections. Sequential slides were collected with 2–4 slides utilized to microisolate ∼500–600 neurons per brain region, per case for each of the CA1A, CA1D and CA1S sublayers. Samples were immediately frozen at –80°C until RNA purification was performed.

Fig. 1

LCM of select regions of CA1 in a mouse model of DS. A) Schematic representation LCM collection of each subset of PNs. B) 5x overview image of CA1A before (i) and after (ii) LCM. 5x overview CA1D and CA1S isolated in parallel from each section (iii, before; iv, after). LCM was performed utilizing a 20x objective identifying CA1D (blue) and CA1S (yellow; v, before; vi, after). Scale bar 5x = 400μm; 20x = 50μm. C) Bioinformatic pipeline schematic, using CombatSeq in the DREAM pipeline for batch effects. D) Heatmap indicates cell type specific markers queried to confirm isolation of excitatory PNs. Color coded scale bar indicates normalized mean gene expression for each cell type. E) Sequence reads and alignment scores for each CA1A, CA1D, and CA1S in 2N and Ts2 (2N-A, 2N-D, 2N-S and Ts-A, Ts-D, Ts-S). F) Pie charts of DEGs indicate percentage of protein-coding and non-coding genes differentially expressed in Ts2 versus 2N for each sublayer.

RNA purification

RNA from each brain for each CA1 sublayer was purified independently using miRNeasy Micro kit (Qiagen) according to manufacturers’ specifications. A DNase digestion was performed twice sequentially before the final washes and RNA purification. RNA quality control (QC) was performed (Agilent 2100; RNA 6000 Pico kit, Agilent, Santa Clara, CA).

Library preparation and RNA-seq

The SMARTer Stranded Total RNA-Seq kit-Pico input Mammalian (Takara Bio, Mountain View, CA) was employed with minor modifications to normalize neuronal numbers for each sample. Batch 1a was done using RNA inputs for CA1D and CA1S samples utilizing 6μl RNA input for LCM >1,500 neurons (∼50% total RNA purified) and 8μl RNA input for LCM <1,500 neurons (∼66% total RNA purified). Batch 1b (LCM, RNA and library processed separately) for CA1A utilized 10μl input volume, with adjustments made to Step A of library preparation to utilize the entire RNA sample collected (after QC). Batch 2 samples (n = 5/genotype/region) were collected, library preparation and sequencing done for all 3 sublayers analyzed with library preparation performed with 8μl of RNA input per sample. Samples were quantified (Agilent 2100 HS DNA kit, Agilent), with all samples reading a minimum of 2 nM of library for QC. For each sample run (Batch 1a, 1b and 2), library samples were pooled in equimolar concentrations and assayed on an HiSeq-4000 (Illumina, San Diego, CA) using a single read 50 cycle protocol at NYUGSOM Genome Technology Center (GTC).

Bioinformatic processing

FastQ files were utilized for both genotypes (Ts2 and 2N) and all 3 sublayers (CA1A, CA1D and CA1S) to analyze data in parallel and uploaded to GEO (http://www.ncbi.nlm.nih.gov/geo; GSE264258). FastQ files were generated and QC of the raw reads was performed by FastQC v0.11.9. 80 Read trimming was performed as necessary by Trimmomatic 0.39. 81 If QC passed and showed no adapter contamination, this step was skipped. Trimmed reads were indexed and aligned to the reference genome (Gencode GRCm39-mm10) using STAR Aligner (2.7.10a). 82 QC of aligned reads was performed using RseQC (v4.0.0). 83 Quantification of aligned reads was performed using Picard (2.27.1) 84 and RSEM (1.3.3). 85 Differential gene expression was conducted using R version R-v4.3.1/Rstudio v1+554 using genes.results produced from RSEM with the mouse reference genome (Gencode GRCm39-mm10).

Statistical analysis

Gene Count matrix obtained from RSEM was analyzed. Genes with over 0.1 counts per million (CPM) were retained followed by TMM normalization implemented by edgeR 86 for downstream analysis. This step removes lowly expressed genes as they provide little evidence of differential expression and increase statistical errors and false discovery rates (FDR).86 –88 After filtering, batch effects were corrected by using CombatSeq 89 from the sva package. 90 Batch correction is required to reduce the negative impact implemented by the variations developed due to the different experimental batches. Subsequently, analysis was performed using the DREAM pipeline, which is built using limma-voom framework from the variancePartition package. 91 In addition to genotype and batch, the following variables were included as covariates: Intergenic percentage, intronic percentage, mRNA base percentage, and correct strand reads percentage. Other than genotype and batch, the covariates are computed from aligned reads by Picard (2.27.1). 84 Multidimensional Scaling (MDS) was used to visualize the distribution of points and identify the presence of outliers within the CA1 data. To determine the nature of the outliers, data previously published utilizing this same bioinformatics analysis protocol examining medial septal nucleus BFCNs from the comparable Ts65Dn DS mouse model 92 was integrated with the CA1 data (Supplementary Figure 1) and MDS was re-performed. Subsequently, any CA1 data points that are determined to be true outliers were systematically removed and R analysis was re-performed after eliminating the outliers from the count matrix. This process ensured robustness and reliability of the data by eliminating erroneous data points and increasing the accuracy of the analysis. CA1A had one 2N outlier (2N, n = 10, Ts2 n = 11; Supplementary Figure 1A) and CA1D had two outliers, one from 2N and one from Ts2 (2N, n = 10, Ts2 n = 10; Supplementary Figure 1B) that were removed. CA1S showed no outliers in analysis (2N, n = 11, Ts2 n = 11; Supplementary Figure 1C). Further outlier removal was deemed not necessary, as all samples passed RNA, library preparation, and RNA-seq QC. Top-Table extracted genes were present for all comparisons. Gene expression differences at (p < 0.05) were considered statistically significant. Protein coding genes were extracted using the AnnotationDbi 93 package from R Bioconductor. Multiple testing corrections were performed by FDR. 94 Statistical analysis was performed in R using the lmer package for cell specific markers comparing normalized cell counts for gene lists for excitatory neuronal markers compared to all other groups 95 modeled as a function of group (DS versus CTR), using mixed effects models with random effects to account for the correlation between repeated assays within the same brain region. 96 Significance was judged at the level α= 0.05, two-sided.

Pathway analyses

Pathway analyses consisted of Ingenuity Pathway Analysis (IPA; Qiagen),97,98, 97,98 Kyoto Encyclopedia of Genes and Genomes (KEGG), 99 Gene Ontology (GO),100,101, 100,101 and STRING 102 in Cytoscape (cutoff 0.4). 103 A Shiny package was utilized to create a web-based app to run GO and KEGG analyses using R version 4.3.1 and Rstudio version 2023.06.0 Build 421. This app was also used to filter keyword targets to identify classes of processes affected by genotype and region. GoPlots for the GO analysis data were generated using the Goplot package version 1.0.2 using the same versions for R and Rstudio. 104 STRING analysis was performed on Ts2 compared to 2N DEGs in CA1A, CA1D and CA1S PNs.

RT-qPCR

LCM was performed on slides for each brain adjacent to tissue collected for RNA-seq analysis, with the CA1A region collected for 3 sections per brain (n = 11 Ts2, n = 11 2N). The miRNeasy Micro kit (Qiagen) was utilized to purify RNA with a DNase digestion performed twice sequentially as above for RNA destined for RNA-seq assays. RNA QC was performed on 2μl from RNA eluted in 14μl final volume (High Sensitivity RNA Assay, TapeStation, Agilent). Due to small amounts of RNA extracted, entire aliquots of RNA for each brain was used for reverse transcription in a 30μl reaction volume to generate cDNA using random hexamers as described previously.72–75,77,78,92,105 , 72–75,77,78,92,105 RT-qPCR was performed using 1μl of cDNA and Taqman PCR primers for a subset of DEGs chosen from STRING, IPA, and GO analysis for CA1A, CA1D, and CA1S. Taqman primers were utilized for the following genes of interest: Amyloid-β precursor protein (App; Mm01344172_m1), ATP synthase peripheral stalk subunit d (Atp5h aka Atp5pd; Mm02392026_g1), 5-hydroxytryptamine (serotonin) receptor 2C (Htr2c; Mm00434127_m1), 5 hydroxytrptamine (serotonin) receptor 4 (Htr4; Mm00434129_m1), NADH:ubiquinone oxidoreductase subunit B8 (Ndufb8; Mm00482663_m1), and Rac family small GTPase 1 (Rac1; Mm01201653_mH). Samples were assayed in triplicate on a real-time qPCR cycler (PikoReal, ThermoFisher) as described previously.72–75,77,78,92,105 , 72–75,77,78,92,105 The ddCT method was employed to determine relative gene level differences between Ts2 and 2N groups.106 –108 Hypoxanthine phosphoribosyltransferase 1 (Hprt1; Mm01318747_g1) and glucuronidase beta (GusB; Mm01197698_m1) qPCR products were utilized as controls, as they are well-established standard controls.74,75,105,109 , 74,75,105,109 Negative controls consisted of the reaction mixture without input RNA. Sample data was compared with respect to PCR product synthesis for each gene tested. Bar charts were generated in Graphpad Prism (10.1.2, GraphPad, Boston, MA). Statistical analysis was performed for each gene with the PCR product synthesis modeled as a function of group (Ts2 versus 2N), using mixed effects models with random effects to account for the correlation between repeated assays within the same brain.72–74,77,78,96,105 , 72–74,77,78,96,105 Significance was judged at the level α= 0.05, two-sided. Correlation coefficient was performed in R using the cor.test() function with Pearson’s rank correlation for rho with significance judged at α= 0.05.

RESULTS

Profiled CA1 PN excitatory neurons

We examined excitatory PNs from 3 hippocampal CA1 sublayers in 11 MO female Ts2 mice compared to 2N disomic controls. We postulate gene expression changes are i) pan-CA1 sublayer as well as those ii) unique to CA1 sublayers reflecting specific circuitry input/outputs within CA1 deep and CA1 superficial PNs. LCM was utilized to isolate individual PN populations, upon which RNA-seq analysis was performed. Specifically, excitatory PNs from the entire CA1 sector, (herein CA1A; Fig. 1A, Bi,ii), CA1a-Deep neurons (herein CA1D; Fig. 1A, Biii-vi), and CA1a-Superficial neurons (herein CA1S; Fig. 1A, Biii,vi) were microisolated via LCM for downstream single population RNA-seq. Bioinformatic analysis was performed using the DREAM pipeline with CombatSeq employed to address batch effects and outliers were removed (Fig. 1C).

Analysis was performed to confirm select isolation of excitatory PNs using normalized gene expression data, querying previously identified cell specific markers in 8 cell types. 95 CA1A, CA1D, and CA1S PNs expressed excitatory neuron (Exc) genes at highly significant gene expression levels (CA1A, p < 4.88E^-6; CA1D, p < 3.57E^-6; CA1S, p < 3.61E^-5) compared to other cell types examined, with expression averages ranging from 1.5–4.5 fold higher than inhibitory neurons (Inh), oligodendrocytes (Oligo), astrocytes (Astro), microglia (MG), oligodendrocyte progenitor cells (OPC), endothelial cells (Endo), and T cells (T) (Fig. 1D). No significant differences were seen in Exc normalized gene expression levels when examining genotype comparisons (Ts2 versus 2N) in CA1A, CA1D, or CA1S. Sample identification for each sublayer is presented in Fig. 1E. Results indicate LCM and single population RNA-seq enriched for excitatory PNs throughout all 3 CA1 sublayers. DEGs revealed significant differences between Ts2 and 2N mice, predominately protein coding genes were significantly dysregulated with 89.2% in CA1A, 96.8% in CA1D, and 94.1% in CA1S with non-coding genes accounting for 10.8%, 3.2%, and 5.9% respectively (Fig. 1F).

Gene expression analysis in CA1 pyramidal neuron sublayers

MDS plots show CA1A (Fig. 2A) PNs had less variability within groups than CA1D PNs (Fig. 2B) and more than CA1S PNs (Fig. 2C) after outliers were removed (Supplementary Figure 1A-C). This did not result in decreased DEGs or alterations in log fold change (LFC) values. Volcano plots showed overall similar gene expression ranges of DEGs, with LFC –4.98 to 3.35 for CA1A PNs (Fig. 2D), –4.74 to 3.47 LFC for CA1D PNs (Fig. 2E), and –4.32 to 4.05 LFC for CA1S PNs (Fig. 2F). When examining DEGs by genotype, CA1A PNs had 880 DEGs with 322 downregulated (36.6%) and 558 upregulated (63.4%) in Ts2 compared to 2N (Fig. 3A, Supplementary Table 1). CA1D PNs had 591 DEGs with 271 downregulated (45.9%) and 320 upregulated (54.1%) in Ts2 compared to 2N (Fig. 3A, Supplementary Table 2). CA1S PNs had 568 DEGs with 254 downregulated (44.7%) and 314 upregulated (55.3%) in Ts2 compared to 2N (Fig. 3A, Supplementary Table 3). Of these DEGs, 122 genes were dysregulated in CA1D PNs and CA1A PNs, 120 genes were dysregulated in CA1S PNs and CA1A PNs, and 99 genes were dysregulated in CA1D PNs and CA1S PNs. A total of 63 DEGs were convergent in the CA1A and both CA1 sublayers examined (Fig. 3A). Scatter plots reveal the majority of DEGs were differentially expressed in 1 sublayer with few DEGs with significant LFCs in other sublayers (Fig. 3B), indicating expression level changes are predominantly sublayer-specific.

Fig. 2

RNA-seq gene expression data Ts2 versus 2N genotype for each subset of the hippocampus examined. A) CA1A MDS plot represents 2N (light blue) and Ts2 (dark blue) for each animal after outlier analysis (2N, n = 10; Ts2, n = 11). B) CA1D MDS plot with 2N in light green and Ts2 in dark green (2N, n = 10; Ts2, n = 10). C) CA1S MDS plot shows 2N in light purple and Ts2 in dark purple with no outlier present (2N, n = 11; Ts2, n = 11). D) Volcano plot for CA1A indicates more upregulated (63.4%) than downregulated DEGs (36.5%; p < 0.05) in CA1A. E) Volcano plot for CA1D indicates 54.1% upregulated and 45.9% downregulated DEGs in Ts2 versus 2N (p < 0.05). F) Similar to CA1D, the volcano plot for CA1S indicates a 55.3% versus 44.7% ratio of upregulated and downregulated DEGs in Ts2 versus 2N (p < 0.05).

Fig. 3

RNA-seq shows DEGs for each sublayer of the CA1 in Ts2 versus 2N have unique expression profiles. A) In CA1A, CA1D, and CA1S genotype comparisons, few DEGs overlap between individual Ts2 CA1 pyramidal neuron populations. A total of 63 genes are convergent to all 3 CA1 sublayers. B) Correlation analysis box plot shows each DEG as a black dot with the convergent genes colored by upregulation or downregulation log fold change (LFC; yellow downregulated; blue, upregulated). C) Convergent DEGs are isolated in a box plot showing upregulation or downregulation for each population’s LFC. D) STRING analysis in Cytoscape shows only a subset of the 63 convergent genes have direct protein-protein interactions (PPIs) with each other, with DEGs without interactions on the right side. E) DEGs of the HSA21 orthologs show similar gene expression for all 3 sublayers, with all DEGs upregulated in Ts2 compared to 2N. F) Top heatmap shows triplicated DEGs significantly upregulated in all 3 (CA1A, CA1D, and CA1S) neuronal populations in Ts2 versus 2N, with color indicating LFC and size indicating significance (-log (p-value)). Bottom heatmap shows triplicated genes that were significant in 1-2 regions, while not significant in 1 or more regions (circle size represents significant (p < 0.05), trend (0.1 > p > 0.05) or not significant). LFC is identified by color (blue downregulated; red upregulated).

Convergent dysregulation in CA1 PNs with links to the triplicated DS region

The 63 DEGs convergently dysregulated in all 3 CA1 sublayers were largely upregulated (53 of 63 DEGs) with 10 DEGs downregulated (Fig. 3C; Supplementary Table 4). STRING in Cytoscape was used to examine the protein-protein interactions (PPI) within this subset of convergently dysregulated DEGs. DEGs including dual-specificity tyrosine phosphorylation regulated kinase 1a (Dyrk1a; 12), transmembrane protein 50B (Tmem50b; 10), DOP1 leucine zipper like protein B (Dop1b; 9), and synaptojanin 1 (Synj1; 9) show the most PPIs (Fig. 3D). Of the 63 DEGs, 15 DEGs had only 1 PPI partner. Another 20 of the 63 DEGs do not have any discernable PPIs. In the convergent DEGs, 31 of the 43 DEGs that had at least 1 PPI were triplicated for DS, with 23 triplicated DEGs with 3 or more PPIs. We examined these 3 CA1 sublayers for trisomic gene expression and found 50 triplicated DEGs upregulated in CA1A PNs, 40 in CA1D PNs, and 35 in CA1S PNs (Fig. 3E, Supplementary Table 4). We found 31 convergent DS triplicated genes significantly upregulated (Fig. 3F, top). Of the remaining 22 DS triplicated genes that were determined to be DEGs in at least 1 sublayer, 14 were significantly or trended to upregulation (0.05 < p < 0.1, Fig. 3F bottom) in 2 or more CA1 PN sublayers (Supplementary Table 5).

Convergent pathway analysis of CA1 pyramidal sublayers

We performed IPA analysis on DEGs for CA1A PNs (880), CA1D PNs (591), and CA1S PNs (568). CA1A PNs had 76 significant neuronal pathways dysregulated (Supplementary Table 6), whereas CA1D PNs had almost double with 155 neuronal pathways significantly dysregulated in Ts2 compared to 2N mice (Supplementary Table 7), while CA1S PNs had the least significant neuronal pathways (72, Supplementary Table 8). Neuronal pathways that were convergently dysregulated included upregulation of gonadotropin releasing hormone (GNRH) signaling and downregulation of phosphate and tensin homolog (PTEN) signaling, with only 10 identified pathways dysregulated similarly. However, 4 were “no call” without giving rise to a z-score, which indicates directionality of pathway change, from any sublayer examined including axonal guidance signaling and Gap junction Signaling (Fig. 4A, top). An additional 7 canonical pathways were dysfunctional across the 3 sublayers, but were divergent in their activity score, including Neurotrophin/Trk signaling, which was downregulated in CA1A PNs and CA1D PNs, but was a “no call” on the z-score in CA1S PNs (i.e., the effect of the DEGs is unclear), and mitochondrial dysfunction, which was downregulated in CA1A PNs (z-score –1.5), minorly upregulated in CA1D PNs (z-score 0.6), and moderately upregulated in CA1S PNs (z-score 1.15; Fig. 4A, bottom and Supplementary Tables 6–8). We examined these overlapping pathways for DEGs that may drive multiple pathways of degeneration and found App, calcium channel, voltage-dependent, T type, alpha 1G subunit (Cacna1g), and Htr2c all were involved in several dysfunctional pathways. Convergent DEGs linked to one dysregulated pathway included interferon (alpha and beta) receptor 1 (Ifnar1), intersectin 1 (Itsn1), and Synj1 (Fig. 4B).

Fig. 4

A) IPA was performed on DEGs for CA1A, CA1D, and CA1S. LFC is indicated by blue-white heatmap, while z-score (pink, upregulation; green, downregulation) indicates direction of pathway change. White bar with X indicates not enough data (NA) to determine z-score upregulation or downregulation. B) Top “driver” gene candidates for convergent dysregulation are limited in number of pathways each affects. C) GO analysis is shown in bar graph with y-axis indicating percentage of unique or co-dysregulated processes, with the letters A, D, or S at top of graph indicating the sublayer analyzed. Each bar represents the region sorted by key word targets for processes. Color indicates overlap between three regions or unique process dysregulated in Ts2 versus 2N. D) Pie chart shows co-dysregulated processes sorted from least number of processes dysregulated (axonal/dendritic in purple) to most dysregulated processes (dark red). E) Circle charts for the top six co-dysregulated processes in the behavior category were further examined for regulation. DEGs are indicated by red (upregulated) or blue (downregulated) dots, while z-score indicates increasing overall gene expression (red) or decreasing overall gene expression (blue). GO terms are as follows: GO:0007611 = learning or memory; GO:0007613 = memory; GO:0007626 = locomotory behavior; GO:0050890 = cognition; GO:0019233 = sensory perception of pain.

GO analysis was performed to examine the biological processes dysregulated in the 3 CA1 PN sublayers in Ts2 compared to 2N mice. Keywords were used to bin neuronal processes into 13 biologically relevant categories with non-neuronal processes binned to the 14th category (Supplementary Tables 9–11 for CA1A/ CA1D/ CA1S sublayers, respectively). We found, similar to IPA analysis, while the DEGs showed overlap of a subset of processes within each neuronal category (light green, Fig. 4C), the majority of the processes dysregulated were unique to each sublayer of CA1 PNs (dark green, Fig. 4C). Of these overlapping processes, the majority were binned into “other cell processes”, which incorporates cellular communication, cellular matrix processes and other maintenance processes, with “protein” processes the next highest number of processes, including post-translational modifications and protein interactions, and the fewest overlapping processes in “axonal/dendritic”, “endo/exosomal” and “behavior” (Fig. 4D). The 6 overlapping GO processes binned into the behavioral category in CA1A PNs and CA1D PNs indicated these behaviors showed increased activity based on DEG LFCs, while in CA1S PNs, were predominantly downregulated (Fig. 4E). GO does not account for known activity effects of each gene when generating z-scores. 104

Limited overlap between sublayers by IPA and GO analysis indicates there are likely unique functional differences in CA1 PNs based on their spatial location and circuitry inputs. Diverse DEGs and pathways in Ts2 mice suggests each sublayer of CA1 PNs drives distinct dysfunctional pathways, with CA1A PNs postulated to recapitulate dysfunctional pathways overlapping either deep and superficial neuron profiles, with few unique pathways, based on the inclusion of these neurons within the CA1A layer. We queried the IPA pathways for unique pathways that were significantly dysregulated in 1 PN sublayer (Fig. 5). CA1A PNs had 26 unique significant neuronal pathways (34.2%; Supplementary Table 12), with neuronal pathways involved in intracellular signaling, including upregulation of AMPK Signaling and downregulation of p38 MAPK Signaling (Fig. 5A). To determine PPI networks between the top driver DEGs, STRING analysis on 880 DEGs was performed, with a secondary STRING analysis examining interactions of the top 21 genes with >11 PPIs (Fig. 5B). Gene-network plots that were generated to examine top DEGs from neuronal pathways revealed moderate connections from the top 5 dysregulated DEGs identified in unique IPA pathways (Fig. 5C). CA1D PNs have the most unique pathways (46; 29.7%, Supplementary Table 13) including upregulation of autophagy and downregulation of oxidative phosphorylation and several unique signaling pathways (Fig. 5D). Driving these pathways, STRING analysis identified the top 21 DEGs with >10 PPIs as candidates for driver genes, with the PPI interactions within these 21 DEGs identified (Fig. 5E). The spiral gene-network plot shows the top 5 DEGs and indicates strong interactions between the DEGs and dysregulated IPA pathways in CA1D PNs. Interestingly, 3 of the top 5 DEGs identified as drivers of multiple pathways of dysregulation were convergently dysregulated in both CA1D PNs and CA1A PNs, including mitogen-activated protein kinase kinase 1 (Map2k1), phosphoinositide-3-kinase regulatory subunit 3 (Pik3r3), and phosphoinositide-3-kinase regulatory subunit 5 (Pik3r5), even though the IPA pathways examined were unique for each sublayer (Fig. 5F). Although the total number of DEGs were comparable to CA1D PNs, CA1S PNs showed only 15 unique neuronal pathways (20.8%) and the majority did not generate a z-score for activity (Fig. 5G, Supplementary Table 14). Examining CA1S PNs with STRING for driver genes, the top 20 DEGs showed fewer PPIs (>7) (Fig. 5H) and gene network plots show no overlap between DEGs and unique pathways, indicated by 1 representative gene for each pathway shown (Fig. 5I). A full list of all unique DEGs for the pathways listed is shown in Supplementary Table 13.

Fig. 5

IPA shows unique pathways significantly affected by DEGs for CA1A, CA1D, and CA1S. A) A subset of neuronal pathways affected only in CA1A, with both upregulation and downregulation seen by z-score (green, downregulation, pink, upregulation). B) STRING by Cytoscape analysis was performed on all CA1A DEGs, with top 21 DEGs shown as a PPI plot. C) Gene network analysis shows top “driver” genes dysregulated in the Ts2 CA1A each interacting with 6–8 IPA canonical pathways. D) A select number of CA1D neuronal pathways are significantly dysregulated, with both upregulation and downregulation seen by z-score (green, downregulation, pink, upregulation). E) STRING analysis shows the top 20 DEGs with the largest number of interactions as a PPI network. F) Driver genes were isolated with gene network analysis showing spiral graph of top 5 targets having high overlap in dysregulation with each gene interacting with 12–30 IPA canonical pathways. G) All CA1S unique neuronal pathways are shown with heatmaps indicating LFC (white-blue) and z-score (green-pink). H) STRING analysis was performed on all CA1S DEGs, with top 20 interactors isolated and reanalyzed for PPI between top hits. I) Gene network analysis was performed on unique IPA pathways to isolate “driver” genes dysregulated in the Ts2 CA1S, however, no genes were drivers for more than one IPA canonical pathway, so 5 representative target genes are shown.

Functions and diseases interactions with driver DEGs in CA1 sublayers

To examine functional interactions of dysregulated gene expression in Ts2 mice, 25 key neurological functions and diseases (F/Ds) were selected from the bio-functions and diseases identified by IPA. These F/Ds are dysregulated in all 3 CA1 PN sublayers (Supplementary Figure 2). The top 10 DEG interactors were selected and examined for their functional links in these 25 F/Ds (Fig. 6). In CA1A PNs, the top 10 gene interactions were selected from the top 28 DEGs identified in >9 F/Ds (Fig. 6A), with App and synuclein alpha (Scna) exhibiting the greatest number of gene-pathway interactions, associated with 23 and 22 F/Ds respectively (Fig. 6A and Supplementary Figure 2A). In CA1D PNs, these same F/Ds were queried using the 591 CA1D PN DEGs. App was also the top driver DEG with 23 F/Ds followed by insulin like growth factor 1 (Igf1) with 22 F/Ds and calcium/calmodulin-dependent protein kinase II alpha (Camk2a) and reelin (Reln) with 15 F/Ds (Fig. 6B and Supplementary Figure 2B). The top 10 DEG drivers in CA1S PNs included both overlapping and unique DEGs. App gene was again a top driver with 23 F/Ds interactions, followed by Igf1 (22 F/Ds interactions), and apolipoprotein E (Apoe; 21 F/Ds interactions; Fig. 6C, Supplementary Figure 2C). Although some overlapping DEGs exist across CA1 PN sublayers (e.g., App), unique drivers of pathology also were observed, indicating the circuitry and likely mechanistic underpinnings driving onset of behavioral alterations are distinct.

Fig. 6

Top F/Ds were isolated from IPA analysis for each subset of DEGs. A) Spiral plot of top ten CA1A DEGs with circle size indicating relative number of functions driven by each DEG. Function circle size indicates number of top DEGs associated with arrow showing gene-function interaction. B) CA1D spiral chart for the same F/Ds show overlapping and unique driver DEGs. C) CA1S ‘driver’ DEGs include both overlapping and unique drivers, with unique DEGs Ntrk3 and Ntf3 drivers in CA1S. Ntrk3 dysregulation is unique to the CA1S sublayer, while Ntf3 is simply not a top driver in CA1A and not significant in CA1D.

Validation by RT-qPCR

Select DEGs were validated by RT-qPCR using LCM captured CA1A PNs from tissue sections adjacent to those examined for single population RNA-seq analysis. The housekeeping genes Gusb and Hprt1 served as control primers.74,75,77,78,92,105 , 74,75,77,78,92,105 A total of 6 genes of interest dysregulated in 1 or more of the sublayers were selected for validation. Correlation analysis between single population RNA-seq LFC and RT-qPCR LFC revealed highly significant correlations between the two approaches, validating expression level changes and directionality.75,92, 75,92 Specifically, significant correlation for GusB and trend level for Hprt1 was observed for CA1A PNs (Fig. 7G), CA1D PNs (Fig. 7H), and CA1S PNs (Fig. 7I), validating directionality of expression level changes seen in both single population RNA-seq and RT-qPCR, consistent with our findings in BFCNs in trisomic mice.75,92, 75,92

Fig. 7

RT-qPCR was performed on CA1A neurons using adjacent tissue sections for Ts2 (n = 11) and 2N (n = 11). A) App gene expression was not significant. B) Atp5h showed no significant difference by RT-qPCR, with no significant differences in CA1A and CA1S and a significant decrease in CA1D RNA-seq. C) Htr2c shows significant increases in Ts2 relative to 2N values with both Gusb (^* p < 0.05) and Hprt1 (^*** p < 0.001) housekeeping genes, matching RNA-seq with significant increases in CA1A/D/S. D) Htr4 was not significantly altered in RT-qPCR. E) Ndufb8 was unchanged in RT-qPCR, matching CA1A and CA1S, although CA1D showed significant decreases in RNA-seq. F) Rac1 was not significantly upregulated with Gusb but was significant (^* p < 0.05) with Hprt1. G) Correlation analysis between LFC of RNA-seq (x-axis) and LFC of RT-qPCR (y-axis) show moderate (Hprt1 control; black dashes) and high (GusB; grey dashes) correlation in CA1A. H) Correlation between RNA-seq (x-axis) and RT-qPCR (y-axis) in CA1D indicates high correlation with GusB (grey) and moderate non-significant correlation with Hprt1 as control (black) I) Correlation between RNA-seq (x-axis) and RT-qPCR (y-axis) in CA1S indicates high correlation with GusB (grey) and moderate non-significant correlation with Hprt1 as control (black).

RT-qPCR results show few significant differences across the 3 sublayers. For example, App (Fig. 7A) was significantly upregulated in all 3 sublayers in the single population RNA-seq analysis but was not upregulated by RT-qPCR (p = 0.137 GusB control; p = 0.119 Hprt1 control). Atp5h, necessary for oxidative phosphorylation complex V, was downregulated in CA1D PNs, but was not significantly altered by RT-qPCR (Fig. 7B). Serotonin receptor Htr2c was significantly upregulated in both RT-qPCR (p < 0.036, GusB; p < 0.0001, Hprt1; Fig. 7C) and single population RNA-seq in all 3 sublayers. Serotonin receptor Htr4 (Fig. 7D) and Ndufb8, part of complex 1 of oxidative phosphorylation, (Fig. 7E) were not significantly dysregulated by RT-qPCR. Rac1, a GTPase highly expressed in brain, was upregulated in CA1D PNs and CA1S PNs by single population RNA-seq and significantly upregulated by qPCR using the Hprt1 control (p < 0.047) but did not reach statistical significance with GusB (Fig. 7F).

DISCUSSION

We interrogated 3 PN populations of the rostral CA1 hippocampus and found both overlapping and unique DEGs and pathways in trisomic mice compared to disomic controls. We postulate gene expression changes underlie behavioral and circuitry deficits in this murine model of DS/AD. Few genes (<15%) of the total DEGs were convergently dysregulated in all CA1 sublayers in 11 MO female Ts2 mice. Single population transcriptomic examination was performed in a single sex (female mice) in the present study, eliminating any sex effects that may confound interrogating DS and AD pathophysiology.28 –31 Approximately half (49.2%) of the convergently dysregulated DEGs were triplicated orthologs of the HSA21 chromosome. This finding affirms copy number overexpression of these ‘DS region’110,111, 110,111 genes throughout the CA1 sector and validates use of the Ts2 mouse to model CA1 pyramidal neuron defects in DS.

To examine hippocampal CA1 PN dysregulated genes that may drive the early onset of AD in DS, bioinformatic inquiry was performed. CA1D PNs showed the most dysregulated pathways by IPA and the most unique processes by GO. By these criteria, CA1D PNs may be the most actively disrupted in 11 MO Ts2 CA1 PNs, which has translational relevance for DS and AD. Moreover, DEG and bioinformatic pathway analysis suggests circuitry inputs and unique cellular functions of each sublayer drives the Ts2 phenotype.

Previous work has shown prominent heterogeneity in gene expression across the dorsal-ventral CA1 axis, with fewer gene expression differences in the proximal-distal axis in the CA1 sector in wildtype mice. 112 Normative rodent models also showed relatively little variability in deep and superficial CA1 PN gene expression, with only a few genes showing differential regulation in deep and superficial neurons.112 –114 In contrast, profound differences in gene expression were seen in Ts2 mice in both CA1D PNs and CA1S PNs, which were largely divergent or unique to each sublayer. The CA1 distal sector is thought to encode non-spatial memory, although recent evidence also suggests some spatial memory is encoded for object displacement and global spatial cues in the distal CA1 sector.115,116, 115,116 Thus, distal CA1 DEGs found in Ts2 mice likely underlie specific deficits in both spatial and non-spatially encoded memory.42,46,48,117–119 , 42,46,48,117–119 Significant differences in CA1D PNs compared to CA1S PNs led to the interrogation of target changes and functional consequences in the Ts2 mouse model.

CA1D PNs had a larger number of dysregulated pathways and unique processes, suggesting that CA1D PNs are more resilient to disease onset in the DS hippocampus, whereas CA1S neurons were more vulnerable to degeneration, with fewer active functional inputs and consequently little pathway activity. Functional input differences from the entorhinal cortex may underlie trisomic CA1D resilience PNs compared to CA1S PNs. Specifically, the MEC is relatively spared compared to the LEC early in DS/AD pathology. 120 CA1D PNs have relatively higher MEC inputs through direct perforant path connections compared to CA1S PNs. 66 Moreover, CA1D PNs have increased firing rates compared to superficial PNs, 121 suggesting CA1D PNs have greater functional connectivity compared to CA1S PNs in 11 MO Ts2 mice. Olfaction is associated with the LEC and the CA1 distal sector, 122 and is essentially non-functional by ∼13 MO in trisomic mice, 123 coinciding with a loss of function in CA1S PNs. DS mouse models have behavioral deficits in both novel object recognition and spatial learning,34,124, 34,124 involving place cells in the deep PN layer.55,68, 55,68

Quantitatively, more DEGs were found in the CA1A sublayer than either CA1D or CA1S sublayers, but this increased amount of DEGs did not result in the highest number of functional alterations. This finding suggests some sublayers compensate for functional consequences or PNs along the transverse or radial axis that have competing functions. 121 Interestingly, mitochondrial dysfunction was dysregulated in all three PN populations, however, CA1D and CA1S PNs showed increased dysfunction, whereas CA1A showed slightly decreased dysfunction. Early and prolonged mitochondrial dysfunction has been associated with AD pathology,125 –127 suggesting distal CA1 neurons show more prominent early pathological changes associated with AD in trisomic mice than CA1A PNs, corresponding to increased LEC inputs in the distal CA1 region. 122 Increased mitochondrial dysfunction and associated downregulation of oxidative phosphorylation is specifically seen in CA1D PNs. Oxidative stress is an early marker of AD progression, occurring prior to plaque formation or tau pathology.128,129, 128,129 In CA1D neurons, uniquely dysregulated pathways, including oxidative phosphorylation and superoxide radical degradation, suggest in 11 MO trisomic mice CA1D PNs are actively undergoing early dysfunction associated with prodromal AD and DS.

To determine targets that may underlie circuitry dependent functional differences in trisomic mice, the 3 sublayers were interrogated for convergent and unique regulators of dysfunction in the Ts2 mice. App, triplicated in DS, was indicated as the top gene hit for the 25 F/Ds dysregulated in all 3 CA1 sublayers. App and the pathogenic β-CTF fragments derived from App are integral in plaque formation, a primary marker of disease onset in DS and AD.17,25,130 , 17,25,130 However, to date, clinical trials and FDA approved treatments targeting the APP protein and its metabolites have equivocal efficacy for AD.131 –133 No information on APP therapeutic effects in DS+AD patients is currently available. We postulate that despite App expression upregulation in CA1 PNs, disease onset is driven by the contribution of other factors in DS and AD.

Several additional DEGs were top drivers, including Htr2c, encoding the serotonin 2C receptor, which was upregulated in trisomic CA1D PNs and CA1S PNs. Serotonin receptor signaling by IPA analysis was convergently upregulated in all CA1 sublayers. Htr2c is associated with depressive behavior, with abnormal gene activity seen in over 80% of AD cases. 134 By contrast, the Htr4 receptor expression was downregulated in the 3 CA1 sublayers, which alteration in Htr4 expression is associated with dysregulation of calcium signaling. 135 Studies in hippocampal circuits with serotonergic antidepressants suggest upregulation of Htr4 plays a critical role in “dematuration” of hippocampal circuits and increased HTR4 will affect proper functioning in mature hippocampal circuits. 136 This indicates dysfunction of CA1 hippocampal PNs as a key target for therapeutic intervention in aged DS.

Rac1 was also upregulated in DS CA1D and CA1S PNs. RAC1B, a constitutively active form of RAC1, was increased in postmortem cholinergic basal forebrain neurons (p75^NTR-positive) in mild to severe AD but not in persons with mild cognitive impairment or no cognitive impairment. 137 Hippocampal Rac1 expression is activated in young and reduced in old 3xTg-AD mice and postmortem human frontal cortex in end-stage AD patients. 138 Rac1 activity has also been shown to trigger generation of pathogenic Aβ fragments, 138 indicating a significant role of Rac1 for disease progression in DS and AD. Upregulation of Rac1 in CA1D and CA1S sublayers may be due to septohippocampal denervation in DS and trisomic mouse models,44,139–142 , 44,139–142 linking gene expression changes in CA1S and CA1D PNs to DS/AD pathology.

Neuroinflammation has been strongly associated with AD pathological progression 143 and unsurprisingly, in DS+AD as well, as HSA21 contains several key inflammatory genes144,145, 144,145 including Infar1, Ifnar2, and Ifngr2, which were all upregulated in the CA1 PNs (Fig. 3F). Further, CA1D PNs exhibited altered signaling of multiple interleukin pathways. Recently, downregulation of interleukin-2 (IL-2) has been found in the hippocampus of AD patients, 146 which correlates with the decreased IL-2 signaling seen in the CA1D PNs (Supplementary Table 7). Top drivers of several interleukin pathways were present in CA1D including two phosphoinositide-3 kinases, Pik3r3 and Pik3r5, along with the phosphoinositol-4 phosphate-3 kinase catalytic subunit 2 alpha, Pik3c2a (Fig. 5F). Phosphoinositides are postulated to play a role in neurodegeneration and neuroinflammation. 147 For example, a report demonstrates ibrutinib, an FDA-approved therapeutic to treat B-cell lymphoma, attenuates Aβ and tau pathology in two mouse models of AD via phosphorylation of phosphoisoditide 3-kinase. 148 Taken together, these results suggest top DEG driver genes are active in the neuroinflammatory response seen in DS CA1 PNs and may be relevant towards treatment in the context ofDS+AD.

Future studies may include adding a cohort of male mice as well as additional time points and examination of the medial (CA1b) and proximal (CA1c) pyramidal sublayers for comparison, all of which are beyond the scope of this report. Assessment of male trisomic mice at 11 MO and different age points would allow for identification of sex-specific gene expression differences, although it remains unclear as to sex-based risk for AD development in individuals with DS, reiterating the need for human and animal/cellular DS studies stratified by sex. Examining CA1 PN sublayers in younger and older mice may elucidate whether CA1D PNs changes are age-related in DS. Perhaps targeting key genes dysregulated in CA1D PNs would ameliorate key pathological consequences of premature aging and dysfunction in the Ts2 mouse model. Future research examining sublayer specific DEGs in DS hiPSCs may validate current findings and enable mechanistic understanding at the molecular and cellular levels as well as at the physiological (functional readout)level.

In conclusion, CA1D PNs in the distal CA1 sublayer revealed significant deficits in key DEGs and canonical pathways along with associated F/Ds in older female Ts2 compared to 2N mice, which result in more functional alterations than CA1A PNs or CA1S PNs. The present study identified putative CA1D PNs cellular targets for the amelioration of circuitry dependent deficits related to the aging hippocampus and targets for drug discovery in individuals with DS with translation to AD.

AUTHOR CONTRIBUTIONS

Melissa J. Alldred (Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Supervision; Validation; Visualization; Writing – original draft; Writing – review & editing); Harshitha Pidikiti (Data curation; Formal analysis; Software; Validation; Visualization); Kryillos W. Ibrahim (Data curation; Software; Validation; Visualization); Sang Han Lee (Data curation; Formal analysis; Writing – review & editing); Adriana Heguy (Methodology; Resources); Gabriel E. Hoffman (Methodology; Resources; Software); Elliott J. Mufson (Funding acquisition; Writing – review & editing); Grace E. Stutzmann (Funding acquisition; Project administration; Writing – review & editing); Stephen D. Ginsberg (Conceptualization; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Supervision; Writing – original draft; Writing – review &editing).

Footnotes

ACKNOWLEDGMENTS

We would like to thank Arthur Saltzmann, M.S. and Paul Zappile, M.S. for technical support. We would like to thank Eva Petkova, Ph.D. and Arjun V. Masurkar, M.D., Ph.D. for support in conception and statistical suggestions. This special issue of Journal of Alzheimer’s Disease is dedicated to the memory of Dr. Mark Smith, whose seminal work on the involvement of oxidative stress was fundamental to understanding the cellular biology of AD.

FUNDING

Funding was provided by support from grants AG014449, AG072599, AG074004, AG085572, AG081286, and AG077103 from the National Institutes of Health, Arizona Alzheimer’s Consortium, Barrow Neurological Foundation, and BrightFocus Foundation.

CONFLICT OF INTEREST

Elliott J. Mufson is an Editorial Board Member of this journal but was not involved in the peer-review process of this article nor had access to any information regarding its peer-review.

All other authors have no conflict of interest to report.

DATA AVAILABILITY

Single population RNA-seq data analyzed within this study are available from GEO (; GSE264258) or from the corresponding author upon request. LCM images for all brain tissue, RNA/library metadata and RT-qPCR data is available from the corresponding author upon request.

The supplementary material is available in the electronic version of this article: .

References

Parker

, Mai

, Canfield

, et al. Updated National Birth Prevalence estimates for selected birth defects in the United States, 2004-2006. Birth Defects Res A Clin Mol Teratol 2010; 88: 1008–1016.

Mai

, Isenburg

, Canfield

, et al. National population-based estimates for major birth defects, 2010-2014. Birth Defects Res 2019; 111: 1420–1435.

Cataldo

, Peterhoff

, Troncoso

, et al. Endocytic pathway abnormalities precede amyloid beta deposition in sporadic Alzheimer’s disease and Down syndrome: Differential effects of APOE genotype and presenilin mutations. Am J Pathol 2000; 157: 277–286.

Hartley

, Blumenthal

, Carrillo

, et al. Down syndrome and Alzheimer’s disease: Common pathways, common goals. Alzheimers Dement 2015; 11: 700–709.

Perez

, Miguel

, He

, et al. Frontal cortex and striatal cellular and molecular pathobiology in individuals with Down syndrome with and without dementia. Acta Neuropathol 2019; 137: 413–436.

Startin

, Hamburg

, Hithersay

, et al. Cognitive markers of preclinical and prodromal Alzheimer’s disease in Down syndrome. Alzheimers Dement 2019; 15: 245–257.

Sendera

, Ma

, Jaffar

, et al. Reduction in TrkA-immunoreactive neurons is not associated with an overexpression of galaninergic fibers within the nucleus basalis in Down’s syndrome. J Neurochem 2000; 74: 1185–1196.

Mufson

, Ginsberg

, Ma

, et al. Editorial: Down syndrome, neurodegeneration and dementia. Front Aging Neurosci 2021; 13: 791044.

Chen

, Dyakin

, Branch

, et al. In vivo MRI identifies cholinergic circuitry deficits in a Down syndrome model. Neurobiol Aging 2009; 30: 1453–1465.

10.

Head

, Lott

, Hof

, et al. Parallel compensatory and pathological events associated with tau pathology in middle aged individuals with Down syndrome. J Neuropathol Exp Neurol 2003; 62: 917–926.

11.

Iulita

, Do Carmo

, Ower

, et al. Nerve growth factor metabolic dysfunction in Down’s syndrome brains. Brain 2014; 137: 860–872.

12.

Sadowski

, Wisniewski

, Tarnawski

, et al. Entorhinal cortex of aged subjects with Down’s syndrome shows severe neuronal loss caused by neurofibrillary pathology. Acta Neuropathol 1999; 97: 156–164.

13.

Lott

and Dierssen

. Cognitive deficits and associated neurological complications in individuals with Down’s syndrome. Lancet Neurol 2010; 9: 623–633.

14.

Millan Sanchez

, Heyn

, Das

, et al. Neurobiological elements of cognitive dysfunction in down syndrome: exploring the role of APP. Biol Psychiatry 2012; 71: 403–409.

15.

Pinter

, Brown

, Eliez

, et al. Amygdala and hippocampal volumes in children with Down syndrome: a high-resolution MRI study. Neurology 2001; 56: 972–974.

16.

Rachidi

and Lopes

. Mental retardation and associated neurological dysfunctions in Down syndrome: a consequence of dysregulation in critical chromosome 21 genes and associated molecular pathways. Eur J Paediatr Neurol 2008; 12: 168–182.

17.

Coyle

, Oster-Granite

and Gearhart

. The neurobiologic consequences of Down syndrome. Brain Res Bull 1986; 16: 773–787.

18.

Chapman

and Hesketh

. Behavioral phenotype of individuals with Down syndrome. Ment Retard Dev Disabil Res Rev 2000; 6: 84–95.

19.

Mann

, Yates

and Marcyniuk

. Alzheimer’s presenile dementia, senile dementia of Alzheimer type and Down’s syndrome in middle age form an age related continuum of pathological changes. Neuropathol Appl Neurobiol 1984; 10: 185–207.

20.

Hook

. Issues pertaining to the impact and etiology of trisomy 21 and other aneuploidy in humans; a consideration of evolutionary implications, maternal age mechanisms, and other matters. Prog Clin Biol Res 1989; 311: 1–27.

21.

Hodgkins

, Prasher

, Farrar

, et al. Reduced transferrin binding in Down syndrome: a route to senile plaque formation and dementia. Neuroreport 1993; 5: 21–24.

22.

Zigman

and Lott

. Alzheimer’s disease in Down syndrome: neurobiology and risk. Ment Retard Dev Disabil Res Rev 2007; 13: 237–246.

23.

Hithersay

, Startin

, Hamburg

, et al. Association of dementia with mortality among adults with down syndrome older than 35 years. JAMA Neurol 2019; 76: 152–160.

24.

Prasher

, Farrer

, Kessling

, et al. Molecular mapping of Alzheimer-type dementia in Down’s syndrome. Ann Neurol 1998; 43: 380–383.

25.

Mann

, Yates

, Marcyniuk

, et al. The topography of plaques and tangles in Down’s syndrome patients of different ages. Neuropathol App Neurobiol 1986; 12: 447–457.

26.

Roth

, Sun

, Greensite

, et al. Premature aging in persons with Down syndrome: MR findings. AJNR Am J Neuroradiol 1996; 17: 1283–1289.

27.

Snyder

, Asthana

, Bain

, et al. Sex biology contributions to vulnerability to Alzheimer’s disease: A think tank convened by the Women’s Alzheimer’s Research Initiative. Alzheimers Dement 2016; 12: 1186–1196.

28.

Lai

, Kammann

, Rebeck

, et al. APOE genotype and gender effects on Alzheimer disease in 100 adults with Down syndrome. Neurology 1999; 53: 331–336.

29.

Schupf

, Kapell

, Nightingale

, et al. Earlier onset of Alzheimer’s disease in men with Down syndrome. Neurology 1998; 50: 991–995.

30.

Raghavan

, Khin-Nu

, Brown

, et al. Gender differences in the phenotypic expression of Alzheimer’s disease in Down’s syndrome (trisomy 21). Neuroreport 1994; 5: 1393–1396.

31.

Mhatre

, Lee

, Pang

, et al. The association between sex and risk of Alzheimer’s disease in adults with Down syndrome. J Clin Med 2021; 10: 2966.

32.

Davisson

, Schmidt

, Reeves

, et al. Segmental trisomy as a mouse model for Down syndrome. Prog Clin Biol Res 1993; 384: 117–133.

33.

Akeson

, Lambert

, Narayanswami

, et al. Ts65Dn – localization of the translocation breakpoint and trisomic gene content in a mouse model for Down syndrome. Cytogenet Cell Genet 2001; 93: 270–276.

34.

Reeves

, Irving

, Moran

, et al. A mouse model for Down syndrome exhibits learning and behaviour deficits. Nat Genet 1995; 11: 177–184.

35.

Sturgeon

and Gardiner

. Transcript catalogs of human chromosome 21 and orthologous chimpanzee and mouse regions. Mamm Genome 2011; 22: 261–271.

36.

Granholm

, Sanders

and Crnic

. Loss of cholinergic phenotype in basal forebrain coincides with cognitive decline in a mouse model of Down’s syndrome. Exp Neurol 2000; 161: 647–663.

37.

Holtzman

, Santucci

, Kilbridge

, et al. Developmental abnormalities and age-related neurodegeneration in a mouse model of Down syndrome. Proc Natl Acad Sci U S A 1996; 93: 13333–13338.

38.

Ash

, Velazquez

, Kelley

, et al. Maternal choline supplementation improves spatial mapping and increases basal forebrain cholinergic neuron number and size in aged Ts65Dn mice. Neurobiol Dis 2014; 70: 32–42.

39.

Kelley

, Ash

, Powers

, et al. Effects of maternal choline supplementation on the septohippocampal cholinergic system in the Ts65Dn mouse model of Down syndrome. Curr Alzheimer Res 2016; 13: 84–96.

40.

Strupp

, Powers

, Velazquez

, et al. Maternal choline supplementation: a potential prenatal treatment for Down syndrome and Alzheimer’s disease. Curr Alzheimer Res 2016; 13: 97–106.

41.

Moon

, Chen

, Gandhy

, et al. Perinatal choline supplementation improves cognitive functioning and emotion regulation in the Ts65Dn mouse model of Down syndrome. Behav Neurosci 2010; 124: 346–361.

42.

Hunter

, Bimonte

and Granholm

A-CE

. Behavioral comparison of 4 and 6 month-old Ts65Dn mice: Age-related impairments in working and reference memory. Behav Brain Res 2003; 138: 121–131.

43.

Kelley

, Ginsberg

, Alldred

, et al. Maternal choline supplementation alters basal forebrain cholinergic neuron gene expression in the Ts65Dn mouse model of Down syndrome. Dev Neurobiol 2019; 79: 664–683.

44.

Kelley

, Powers

, Velazquez

, et al. Maternal choline supplementation differentially alters the basal forebrain cholinergic system of young-adult Ts65Dn and disomic mice. J Comp Neurol 2014; 522: 1390–1410.

45.

Driscoll

, Carroll

, Moon

, et al. Impaired sustained attention and error-induced stereotypy in the aged Ts65Dn mouse: a mouse model of Down syndrome and Alzheimer’s disease. Behav Neurosci 2004; 118: 1196–1205.

46.

Powers

, Velazquez

, Kelley

, et al. Attentional function and basal forebrain cholinergic neuron morphology during aging in the Ts65Dn mouse model of Down syndrome. Brain Struct Funct 2016; 221: 4337–4352.

47.

Velazquez

, Ash

, Powers

, et al. Maternal choline supplementation improves spatial learning and adult hippocampal neurogenesis in the Ts65Dn mouse model of Down syndrome. Neurobiol Dis 2013; 58: 92–101.

48.

Powers

, Velazquez

, Strawderman

, et al. Maternal choline supplementation as a potential therapy for Down syndrome: assessment of effects throughout the lifespan. Front Aging Neurosci 2021; 13: 723046.

49.

Villar

, Belichenko

, Gillespie

, et al. Identification and characterization of a new Down syndrome model, Ts[Rb(12.1716)]2Cje, resulting from a spontaneous Robertsonian fusion between T(171)65Dn and mouse chromosome 12. Mamm Genome 2005; 16: 79–90.

50.

Levine

, Saltzman

, Levy

, et al. Systemic pathology in aged mouse models of Down’s syndrome and Alzheimer’s disease. Exp Mol Pathol 2009; 86: 18–22.

51.

Kaur

, Sharma

, Xu

, et al. Glutamatergic transmission aberration: a major cause of behavioral deficits in a murine model of Down’s syndrome. J Neurosci 2014; 34: 5099–5106.

52.

Valbuena

, García

, Mazier

, et al. Unbalanced dendritic inhibition of CA1 neurons drives spatial-memory deficits in the Ts2Cje Down syndrome model. Nat Commun 2019; 10: 4991.

53.

Rueda

, Florez

and Martinez-Cue

. Mouse models of Down syndrome as a tool to unravel the causes of mental disabilities. Neural Plast 2012; 2012: 584071.

54.

Gautier

, Kelley

, Lee

, et al. Maternal choline supplementation protects against age-associated cholinergic and GABAergic basal forebrain neuron degeneration in the Ts65Dn mouse model of Down syndrome and Alzheimer’s disease. Neurobiol Dis 2023; 188: 106332.

55.

Ito

and Schuman

. Functional division of hippocampal area CA1 via modulatory gating of entorhinal cortical inputs. Hippocampus 2012; 22: 372–387.

56.

Fyhn

, Molden

, Witter

, et al. Spatial representation in the entorhinal cortex. Science 2004; 305: 1258–1264.

57.

Belichenko

, Masliah

, Kleschevnikov

, et al. Synaptic structural abnormalities in the Ts65Dn mouse model of Down syndrome. J Comp Neurol 2004; 480: 281–298.

58.

Cooper

, Salehi

, Delcroix

, et al. Failed retrograde transport of NGF in a mouse model of Down’s syndrome: reversal of cholinergic neurodegenerative phenotypes following NGF infusion. Proc Natl Acad Sci U S A 2001; 98: 10439–10444.

59.

Sofroniew

, Galletly

, Isacson

, et al. Survival of adult basal forebrain cholinergic neurons after loss of target neurons. Science 1990; 247: 338–342.

60.

Powers

, Kelley

, Velazquez

, et al. Maternal choline supplementation in a mouse model of Down syndrome: Effects on attention and nucleus basalis/substantia innominata neuron morphology in adult offspring. Neuroscience 2017; 340: 501–514.

61.

Hyman

, Van Hoesen

and Damasio

. Memory-related neural systems in Alzheimer’s disease: an anatomic study. Neurology 1990; 40: 1721–1730.

62.

Hyman

, Van Hoesen

, Damasio

, et al. Alzheimer’s disease: cell-specific pathology isolates the hippocampal formation. Science 1984; 225: 1168–1170.

63.

Kurt

, Kafa

, Dierssen

, et al. Deficits of neuronal density in CA1 and synaptic density in the dentate gyrus, CA3 and CA1, in a mouse model of Down syndrome. Brain Res 2004; 1022: 101–109.

64.

Hyman

, West

, Rebeck

, et al. Neuropathological changes in Down’s syndrome hippocampal formation. Effect of age and apolipoprotein E genotype. Arch Neurol 1995; 52: 373–378.

65.

Mizuseki

, Diba

, Pastalkova

, et al. Hippocampal CA1 pyramidal cells form functionally distinct sublayers. Nat Neurosci 2011; 14: 1174–1181.

66.

Masurkar

, Srinivas

, Brann

, et al. Medial and lateral entorhinal cortex differentially excite deep versus superficial CA1 pyramidal neurons. Cell Rep 2017; 18: 148–160.

67.

Geiller

, Fattahi

, Choi

, et al. Place cells are more strongly tied to landmarks in deep than in superficial CA1. Nat Commun 2017; 8: 14531.

68.

Danielson

, Zaremba

, Kaifosh

, et al. Sublayer-specific coding dynamics during spatial navigation and learning in hippocampal area CA1. Neuron 2016; 91: 652–665.

69.

Masurkar

. Towards a circuit-level understanding of hippocampal CA1 dysfunction in Alzheimer’s disease across anatomical axes. J Alzheimers Dis Parkinsonism 2018; 8: 412.

70.

Cid

, Marquez-Galera

, Valero

, et al. Sublayer- and cell-type-specific neurodegenerative transcriptional trajectories in hippocampal sclerosis. Cell Rep 2021; 35: 109229.

71.

Kelley

, Powers

, Velazquez

, et al. Sex differences in the cholinergic basal forebrain in the Ts65Dn mouse model of Down syndrome and Alzheimer’s disease. Brain Pathol 2014; 24: 33–44.

72.

Alldred

, Lee

, Petkova

, et al. Expression profile analysis of vulnerable CA1 pyramidal neurons in young-middle-aged Ts65Dn mice. J Comp Neurol 2015; 523: 61–74.

73.

Alldred

, Lee

, Petkova

, et al. Expression profile analysis of hippocampal CA1 pyramidal neurons in aged Ts65Dn mice, a model of Down syndrome (DS) and Alzheimer’s disease (AD). Brain Struct Funct 2015; 220: 2983–2996.

74.

Alldred

, Lee

, Stutzmann

, et al. Oxidative phosphorylation is dysregulated within the basocortical circuit in a 6-month old mouse model of Down syndrome and Alzheimer’s disease. Front Aging Neurosci 2021; 13: 707950.

75.

Alldred

, Penikalapati

, Lee

, et al. Profiling basal forebrain cholinergic neurons reveals a molecular basis for vulnerability within the Ts65Dn model of Down syndrome and Alzheimer’s disease. Mol Neurobiol 2021; 58: 5141–5162.

76.

Duchon

, Raveau

, Chevalier

, et al. Identification of the translocation breakpoints in the Ts65Dn and Ts1Cje mouse lines: relevance for modeling Down syndrome. Mamm Genome 2011; 22: 674–684.

77.

Alldred

, Chao

, Lee

, et al. CA1 pyramidal neuron gene expression mosaics in the Ts65Dn murine model of Down syndrome and Alzheimer’s disease following maternal choline supplementation. Hippocampus 2018; 28: 251–268.

78.

Alldred

, Chao

, Lee

, et al. Long-term effects of maternal choline supplementation on CA1 pyramidal neuron gene expression in the Ts65Dn mouse model of Down syndrome and Alzheimer’s disease. FASEB J 2019; 33: 9871–9884.

79.

Alldred

and Ginsberg

. Microisolation of spatially characterized single populations of neurons for RNA sequencing from mouse and postmortem human brain tissues. J Clin Med 2023; 12: 3304.

80.

Andrews

. FastQC: A quality control tool for high throughput sequence data. 2010.

81.

Bolger

, Lohse

and Usadel

. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 2014; 30: 2114–2120.

82.

Dobin

, Davis

, Schlesinger

, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 2013; 29: 15–21.

83.

Wang

, Wang

and Li

. RSeQC: quality control of RNA-seq experiments. Bioinformatics 2012; 28: 2184–2185.

84.

Picard Toolkit, https://broadinstitute.github.io/picard/ (2019, accessed 8/24/22).

85.

and Dewey

. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics 2011; 12: 323.

86.

Robinson

, McCarthy

and Smyth

. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 2010; 26: 139–140.

87.

Zehetmayer

, Posch

and Graf

. Impact of adaptive filtering on power and false discovery rate in RNA-seq experiments. BMC Bioinformatics 2022; 23: 388.

88.

Rau

, Gallopin

, Celeux

, et al. Data-based filtering for replicated high-throughput transcriptome sequencing experiments. Bioinformatics 2013; 29: 2146–2152.

89.

Zhang

, Parmigiani

and Johnson

. ComBat-seq: batch effect adjustment for RNA-seq count data. NAR Genom Bioinform 2020; 2: lqaa078.

90.

Leek

, Johnson

, Parker

, et al. SVA: Surrogate Variable Analysis, R package version 3.50.0, https://bioconductor.org/packages/sva. (2023).

91.

Hoffman

and Schadt

. variancePartition: interpreting drivers of variation in complex gene expression studies. BMC Bioinformatics 2016; 17: 483.

92.

Alldred

, Pidikiti

, Heguy

, et al. Basal forebrain cholinergic neurons are vulnerable in a mouse model of Down syndrome and their molecular fingerprint is rescued by maternal choline supplementation. FASEB J 2023; 37: e22944.

93.

Pagès

, Carlson

, Falcon

, et al. AnnotationDbi: Manipulation of SQLite-based annotations in bioconductor, https://bioconductor.org/packages/release/bioc/html/AnnotationDbi.html (2019).

94.

Broberg

. A comparative review of estimates of the proportion unchanged genes and the false discovery rate. BMC Bioinformatics 2005; 6: 199.

95.

Mathys

, Peng

, Boix

, et al. Single-cell atlas reveals correlates of high cognitive function, dementia, and resilience to Alzheimer’s disease pathology. Cell 2023; 186: 4365–4385.e27.

96.

McCulloch

, Searle

and Neuhaus

. Generalized, Linear, and Mixed Models. Second Edition. New York: John Wiley & Sons, 2011.

97.

Qiagen. https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis (2020).

98.

Krämer

, Green

, Pollard

, et al. Causal analysis approaches in Ingenuity Pathway Analysis. Bioinformatics 2013; 30: 523–530.

99.

Kanehisa

and Goto

. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res 2000; 28: 27–30.

100.

Ashburner

, Ball

, Blake

, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2000; 25: 25–29.

101.

The Gene Ontology resource: enriching a GOld mine. Nucleic Acids Res 2021; 49: D325–D334.

102.

Szklarczyk

, Gable

, Lyon

, et al. STRING v11: protein–protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res 2018; 47: D607–D613.

103.

Shannon

, Markiel

, Ozier

, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 2003; 13: 2498–2504.

104.

Walter

, Sánchez-Cabo

and Ricote

. GOplot: an R package for visually combining expression data with functional analysis. Bioinformatics 2015; 31: 2912–2914.

105.

Alldred

, Lee

and Ginsberg

. Adiponectin modulation by genotype and maternal choline supplementation in a mouse model of Down syndrome and Alzheimer’s disease. J Clin Med 2021; 10: 2994.

106.

Ginsberg

, Alldred

, Counts

, et al. Microarray analysis of hippocampal CA1 neurons implicates early endosomal dysfunction during Alzheimer’s disease progression. Biol Psychiatry 2010; 68: 885–893.

107.

Jiang

, Mullaney

, Peterhoff

, et al. Alzheimer’s-related endosome dysfunction in Down syndrome is Abeta-independent but requires APP and is reversed by BACE-1 inhibition. Proc Natl Acad Sci USA 2010; 107: 1630–1635.

108.

ABI. Guide to performing relative quantitation of gene expression using real-time quantitative PCR. Applied Biosystems Product Guide 2004: 1–60.

109.

ThermoFisher. https://www.thermofisher.com/us/en/home/life-science/pcr/real-time-pcr/real-time-pcr-assays/taqman-gene-expression/single-tube-taqman-gene-expression-analysis/taqman-endogenous-controls.html (accessed 12-05-2023).

110.

Zhou

, Zhi

, Chen

, et al. Single-nucleus RNA sequencing reveals cell type-specific transcriptome alterations of Down syndrome hippocampus using the Dp16 mouse model. Genes Genomics 2023; 45: 1305–1315.

111.

Gupta

, Dhanasekaran

and Gardiner

. Mouse models of Down syndrome: gene content and consequences. Mamm Genome 2016; 27: 538–555.

112.

Cembrowski

, Bachman

, Wang

, et al. Spatial gene-expression gradients underlie prominent heterogeneity of CA1 pyramidal neurons. Neuron 2016; 89: 351–368.

113.

Cembrowski

, Wang

, Sugino

, et al. Hipposeq: a comprehensive RNA-seq database of gene expression in hippocampal principal neurons. Elife 2016; 5: e14997.

114.

Dong

, Swanson

, Chen

, et al. Genomic-anatomic evidence for distinct functional domains in hippocampal field CA1. Proc Natl Acad Sci USA 2009; 106: 11794–11799.

115.

Vandrey

, Duncan

and Ainge

. Object and object-memory representations across the proximodistal axis of CA1. Hippocampus 2021; 31: 881–896.

116.

Deshmukh

. Distal CA1 maintains a more coherent spatial representation than proximal CA1 when local and global cues conflict. J Neurosci 2021; 41: 9767–9781.

117.

Belichenko

, Belichenko

, Kleschevnikov

, et al. The “Down syndrome critical region” is sufficient in the mouse model to confer behavioral, neurophysiological, and synaptic phenotypes characteristic of Down syndrome. J Neurosci 2009; 29: 5938–5948.

118.

Fernandez

and Garner

. Episodic-like memory in Ts65Dn, a mouse model of Down syndrome. Behav Brain Res 2008; 188: 233–237.

119.

Escorihuela

, Fernandez-Teruel

, Vallina

, et al. A behavioral assessment of Ts65Dn mice: a putative Down syndrome model. Neurosci Lett 1995; 199: 143–146.

120.

Khan

, Liu

, Provenzano

, et al. Molecular drivers and cortical spread of lateral entorhinal cortex dysfunction in preclinical Alzheimer’s disease. Nat Neurosci 2014; 17: 304–311.

121.

Masurkar

, Tian

, Warren

, et al. Postsynaptic integrative properties of dorsal CA1 pyramidal neuron subpopulations. J Neurophysiol 2020; 123: 980–992.

122.

Igarashi

, Lu

, Colgin

, et al. Coordination of entorhinal-hippocampal ensemble activity during associative learning. Nature 2014; 510: 143–147.

123.

Bianchi

, Bettini

, Guidi

, et al. Age-related impairment of olfactory bulb neurogenesis in the Ts65Dn mouse model of Down syndrome. Exp Neurol 2014; 251: 1–11.

124.

Fernandez

, Morishita

, Zuniga

, et al. Pharmacotherapy for cognitive impairment in a mouse model of Down syndrome. Nat Neurosci 2007; 10: 411–413.

125.

Moreira

, Carvalho

, Zhu

, et al. Mitochondrial dysfunction is a trigger of Alzheimer’s disease pathophysiology. Biochim Biophys Acta 2010; 1802: 2–10.

126.

Moreira

, Siedlak

, Wang

, et al. Autophagocytosis of mitochondria is prominent in Alzheimer disease. J Neuropathol Exp Neurol 2007; 66: 525–532.

127.

Castellani

, Hirai

, Aliev

, et al. Role of mitochondrial dysfunction in Alzheimer’s disease. J Neurosci Res 2002; 70: 357–360.

128.

Moreira

, Zhu

, Liu

, et al. Compensatory responses induced by oxidative stress in Alzheimer disease. Biol Res 2006; 39: 7–13.

129.

Ogawa

, Perry

and Smith

. The “Down’s” side of mitochondria. Dev Cell 2002; 2: 255–256.

130.

Wisniewski

, Dalton

, Crapper McLachlan

, et al. Alzheimer’s disease in Down’s syndrome: clinicopathologic studies. Neurology 1985; 35: 957–961.

131.

Solopova

, Romero-Fernandez

, Harmsen

, et al. Fatal iatrogenic cerebral β-amyloid-related arteritis in a woman treated with lecanemab for Alzheimer’s disease. Nat Commun 2023; 14: 8220.

132.

Ebell

, Barry

, Baduni

, et al. Clinically important benefits and harms of monoclonal antibodies targeting amyloid for the treatment of Alzheimer disease: a systematic review and meta-analysis. Ann Fam Med 2024; 22: 50–62.

133.

Digma

, Winer

and Greicius

. Substantial doubt remains about the efficacy of anti-amyloid antibodies. J Alzheimers Dis 2024; 97: 567–572.

134.

Cacabelos

, Carril

, Corzo

, et al. Pharmacogenetics of anxiety and depression in Alzheimer’s disease. Pharmacogenomics 2023; 24: 27–57.

135.

Torres

, Chaput

and Andrade

. Cyclic AMP and protein kinase A mediate 5-hydroxytryptamine type 4 receptor regulation of calcium-activated potassium current in adult hippocampal neurons. Mol Pharmacol 1995; 47: 191–197.

136.

Kobayashi

, Ikeda

, Sakai

, et al. Reversal of hippocampal neuronal maturation by serotonergic antidepressants. Proc Natl Acad Sci U S A 2010; 107: 8434–8439.

137.

Perez

, Getova

, He

, et al. Rac1b increases with progressive tau pathology within cholinergic nucleus basalis neurons in Alzheimer’s disease. Am J Pathol 2012; 180: 526–540.

138.

Borin

, Saraceno

, Catania

, et al. Rac1 activation links tau hyperphosphorylation and Aβ dysmetabolism in Alzheimer’s disease. Acta Neuropathol Commun 2018; 6: 61.

139.

Granholm

, Ford

, Hyde

, et al. Estrogen restores cognition and cholinergic phenotype in an animal model of Down syndrome. Physiol Behav 2002; 77: 371–385.

140.

Mufson

, Bothwell

and Kordower

. Loss of nerve growth factor receptor-containing neurons in Alzheimer’s disease: a quantitative analysis across subregions of the basal forebrain. Exp Neurol 1989; 105: 221–232.

141.

Mufson

, Ma

, Dills

, et al. Loss of basal forebrain p75(NTR) immunoreactivity in subjects with mild cognitive impairment and Alzheimer’s disease. J Comp Neurol 2002; 443: 136–153.

142.

Mufson

, Counts

, Perez

, et al. Cholinergic system during the progression of Alzheimer’s disease: therapeutic implications. Expert Rev Neurother 2008; 8: 1703–1718.

143.

Rauf

, Badoni

, Abu-Izneid

, et al. Neuroinflammatory markers: key indicators in the pathology of neurodegenerative diseases. Molecules 2022; 27: 3194.

144.

Wilcock

and Griffin

. Down’s syndrome, neuroinflammation, and Alzheimer neuropathogenesis. J Neuroinflammation 2013; 10: 84.

145.

Wilcock

, Hurban

, Helman

, et al. Down syndrome individuals with Alzheimer’s disease have a distinct neuroinflammatory phenotype compared to sporadic Alzheimer’s disease. Neurobiol Aging 2015; 36: 2468–2474.

146.

Alves

, Churlaud

, Audrain

, et al. Interleukin-2 improves amyloid pathology, synaptic failure and memory in Alzheimer’s disease mice. Brain 2017; 140: 826–842.

147.

Ernest James Phillips

and Maguire

. Phosphoinositides: roles in the development of microglial-mediated neuroinflammation and neurodegeneration. Front Cell Neurosci 2021; 15: 652593.

148.

Lee

, Jeon

, Kim

, et al. Ibrutinib modulates Aβ/tau pathology, neuroinflammation, and cognitive function in mouse models of Alzheimer’s disease. Aging Cell 2021; 20: e13332.