Molecular and cellular processes disrupted in the early postnatal Down syndrome prefrontal cortex

Ryan D Risgaard; Kalpana Hanthanan Arachchilage; Sara A Knaack; Masoumeh Hosseini; Rachel J Chen; Pubudu Kumarage; Danielle K Schmidt; Xiang Huang; Jie Sheng; Carlos J Wang; Elisa Giusti; Shuang Liu; Su-Chun Zhang; Daifeng Wang; Anita Bhattacharyya; Andre M M Sousa

doi:10.1126/science.aea1549

. Author manuscript; available in PMC: 2026 Jun 23.

Published in final edited form as: Science. 2026 Apr 23;392(6796):eaea1549. doi: 10.1126/science.aea1549

Molecular and cellular processes disrupted in the early postnatal Down syndrome prefrontal cortex

Ryan D Risgaard ^1,^2,^†, Kalpana Hanthanan Arachchilage ^1,^†, Sara A Knaack ^1,^2,^†, Masoumeh Hosseini ^1,^2,^†, Rachel J Chen ^1,², Pubudu Kumarage ¹, Danielle K Schmidt ^1,², Xiang Huang ¹, Jie Sheng ¹, Carlos J Wang ^1,², Elisa Giusti ^1,², Shuang Liu ¹, Su-Chun Zhang ^1,^2,³, Daifeng Wang ^1,^4,⁵, Anita Bhattacharyya ^1,⁶, Andre M M Sousa ^1,^2,^*

PMCID: PMC13286256 NIHMSID: NIHMS2177336 PMID: 42024756

Abstract

Down syndrome is a genetic condition that causes intellectual disability and is characterized by early-onset delays in motor, cognitive, and language development. The molecular mechanisms underlying these neurodevelopmental impairments remain poorly understood. Here, we utilized single-nucleus multiomic sequencing to simultaneously profile gene expression and chromatin accessibility in the Down syndrome prefrontal cortex during early postnatal development, a critical period for synaptogenesis, neural maturation, and developmental neuroimmune interactions. Our findings reveal widespread dysregulation of chromatin accessibility and gene expression, with deficits spanning metabolic and synaptic pathways, oligodendrocyte lineage progression, and a pronounced neuroinflammatory signature. We present a molecular atlas of Down syndrome neuropathology at a critical stage of brain development, highlighting convergent neurodevelopmental and neurodegenerative pathways and informing potential targeted therapies for Down syndrome-associated neuroinflammation.

Introduction

Down syndrome (DS), resulting from trisomy 21 (Ts21), is one of the most common genetic causes of intellectual disability (1). Individuals with DS present in early infancy with a discernible pattern of developmental milestone delays in motor, cognitive, and language domains. This neurodevelopmental phenotype encompasses deficits in executive function, working memory, motor coordination, and linguistic processing - impairments that emerge within the first postnatal months and profoundly influence academic, occupational, and adaptive functioning trajectories throughout the lifespan (2–5). The early onset of these deficits implicates disruptions in fundamental neurodevelopmental processes during prenatal and early postnatal corticogenesis (6).

Despite advances in modeling Ts21, murine systems have limited translational relevance to human-specific aspects of neurodevelopment and cortical circuitry (7), whereas stem cell models have faced inherent constraints in recapitulating in vivo cellular diversity and the organization of neuronal-glial interactions (8, 9). To address these constraints, we performed single-nucleus gene expression and accessible chromatin (snMultiome) sequencing of the DS dorsolateral prefrontal cortex (dlPFC) - a cortical area central to higher-order cognition and working memory (10) - during the early postnatal period (0–3 years), a critical window of synaptogenesis, neuronal and glial maturation, and developmental neuroimmune interactions (11).

Collectively, our study reveals a global dysregulation of chromatin accessibility and gene expression in the Ts21 dlPFC. This dysregulation extends beyond simple gene-dosage effects and involves complex interactions between transcription, gene regulatory networks, and the accessible chromatin landscape of discrete cell types. Our analysis reveals a pervasive dysregulation of metabolic and neuronal synaptic gene expression programs, impaired oligodendrocyte lineage progression with deficits in myelin-related transcription, and a pronounced neuroinflammatory signature marked by microglial activation and cytokine dysregulation. We reveal convergent pathways of neurodevelopmental disruption and hallmarks of incipient neurodegeneration in the early postnatal Ts21 dlPFC. This snMultiome resource (available at: https://daifengwanglab.shinyapps.io/DS_PFC/) provides a detailed, cell-type-specific molecular atlas of gene expression and chromatin accessibility in the Ts21 brain, establishing a foundation for the development of targeted therapeutic interventions addressing both developmental and degenerative aspects of Ts21 neuropathology in the early postnatal period.

RESULTS

Transcriptomic classification reveals selective vulnerability of cell populations in the early postnatal Ts21 dlPFC.

We performed single-nucleus multiomic RNA+ATAC-sequencing (snMultiome) of the postmortem dorsolateral prefrontal cortex (dlPFC) from 5 individuals with trisomy 21 (Ts21) and 5 age- and sex-matched controls during the early postnatal period (0–3 years of age) (Fig. 1, A and B, and table S1). After implementing stringent quality control (Methods), a total of 220,956 nuclei were retained for analysis (Fig. 1C and fig. S1, A to J).

Fig. 1. — (A) Sampling strategy of dorsolateral prefrontal cortex, encompassing Brodmann areas (BA) 9 and 46, used for single-nucleus multiomic sequencing. (B) Ages of samples profiled in this study. Each dot represents one biological sample. (C) Transcriptomic UMAP visualization of all Control (left panel) and Ts21 (right panel) nuclei analyzed in this study using a joint UMAP embedding of both Control and Ts21 nuclei, colored by cell subclass. ExN, excitatory neurons; IT, intratelencephalic; NP, near-projecting; CT, corticothalamic; ET, extratelencephalic; InN, inhibitory neurons; VLMC, vascular leptomeningeal cells; Endo, endothelial cells; Micro, microglia; OPC, oligodendrocyte precursor cells; Oligo, oligodendrocytes; Astro, astrocytes. (D) Differential cell type composition analysis with age and sex as covariates at the cell subclass annotation level. Statistically significant compositional changes for individual cell subclasses are highlighted in red (FDR <0.05). (E) Distribution and proportion of cell subclasses (bar graph) and cell classes (inset pie chart) for Control (left panel) and Ts21 (right panel). Cell subclasses present only in control are highlighted in red. (F) Upset plot depicting differentially expressed genes at cell subclass level. Highly downregulated genes in Ts21 subclasses (adjusted p<0.01, FC>2, left panel) and highly upregulated genes in Ts21 subclasses (adjusted p<0.01, FC>2, right panel) are shown. Bars are shaded based on the cell-type specificity of the differentially expressed gene set. (G) Bar plot of the percentage of differentially accessible regions (DARs) per cell type with an adjusted p-value threshold of 0.05 from logistic regression, defined as the number of DARs relative to the number of peaks with observed fragments in at least 1% of barcodes for each cell type. (H) Genomic region annotations of DARs by cell type relative to hg38 gene annotations for Ts21 increased accessibility DARs (right panel) and Ts21 decreased accessibility DARs (left panel).

A fundamental question in the field is how an extra copy of chromosome 21 impacts gene expression profiles both within and between cell types. To address this, we performed unsupervised clustering (12, 13) to annotate major cell classes (excitatory neurons [ExN], inhibitory neurons [InN], non-neuronal cells [NNC]), subclasses (for example, InN SST), and subtypes (for example, InN SST NPY CHODL) (fig. S1, K and L, and Methods). Following batch effect correction (14), Ts21 and control nuclei were annotated independently to minimize integration bias and identify putative Ts21-specific cell subpopulations or states. We revealed a general and broad conservation of cell subclasses, with the exception of three rare (≤0.3% representation) subclasses that were detected exclusively in control samples (Fig. 1, C and E, fig. S1, K to M). To determine whether these cell types were truly absent or simply unsampled due to their rarity, we performed immunofluorescence and in situ hybridization and identified the presence of all three subclasses in the Ts21 cortex (fig. S2).

Next, we sought to address how Ts21 influences transcriptomic cell-type identity. To identify principal drivers of biological variability and genes that maximally discriminate control and Ts21 cellular states, we independently identified the top 3,000 highly variable genes (HVGs) in each condition. Whereas most HVGs were shared (71.8%), a subset were condition-specific (fig. S3A). Gene Ontology (GO) (15) analysis revealed that control-specific HVGs were enriched for homeostatic processes such as gene expression, translation, and metabolism, whereas Ts21-specific HVGs were enriched for chronic inflammation and reactive oxygen species (ROS) response (fig. S3B). These condition-specific HVGs suggest disrupted homeostasis, oxidative stress, and inflammation as key drivers of biological variation in the early postnatal Ts21 dlPFC.

We next sought to uncover which cell types were the most affected in Ts21. Using HVG sets calculated independently for control and Ts21, we performed unsupervised clustering to characterize the relationship of cell subclasses between conditions (fig. S3C and Methods). This revealed approximately half (12/21) of Ts21 cell subclasses clustered based on cell identity rather than based on condition. In contrast, nine Ts21 cell subclasses demonstrated condition-specific clustering, irrespective of whether control or Ts21 HVGs were used. These subclasses predominantly comprised glial cells, intratelencephalic-projecting (IT) excitatory neurons, and caudal ganglionic eminence (CGE)-derived inhibitory neurons (fig. S3C). Aberrant glial and ExN populations clustered more closely with microglia subclasses, suggesting activation of inflammatory/immune-related transcriptional programs.

To delineate transcriptomic variability within cell subclasses, we employed an entropy-based approach that revealed elevated heterogeneity in the majority (14/19) of Ts21 subclasses (fig. S3D and Methods). We found that all Ts21 ExNs exhibited greater transcriptomic heterogeneity, whereas InNs displayed subclass-specific differences (fig. S3D). This increased heterogeneity in Ts21 parallels observations in Alzheimer’s disease (AD), where transcriptomic variability scores correlate with clinical severity and neurofibrillary tangle burden (16).

Cell composition alterations in the Ts21 dlPFC

Disruptions in cellular composition and cortical arealization have been hypothesized to contribute to the pathogenesis of neurodevelopmental disorders (17, 18). To systematically evaluate shifts in both cell subclasses and transcriptomically defined subtypes, we employed differential cell composition (19) at both hierarchical levels (Fig. 1, D and E, fig. S3, E to G, and table S2). Cell-subclass differential abundance revealed a significant increase (FDR < 0.05) in the proportion of upper layer excitatory neurons (ExN L2–3 IT) in Ts21 (Fig. 1, D and E, fig. S3F). This observation aligns with midfetal snRNA-seq data from the companion study (20), while contrasting with previous findings in mouse and in vitro models (6).

Whereas no significant differences were observed within InN or glial subclass proportions, subtype-level analysis revealed compositional shifts in transcriptomically-defined astrocyte, microglia, and InN subtypes (fig. S3, E and G). These changes reflect likely Ts21-associated perturbations in developmental patterning, cell maturation, or subtype-specific pathological responses. Together, our results demonstrate a selective expansion of upper-layer IT ExNs in the early postnatal Ts21 dlPFC, concurrent with subtype-specific reorganization of glial and inhibitory neuronal populations. These laminar and cellular imbalances may contribute to the circuit-level pathophysiology observed in Ts21 (21), potentially reflecting disrupted, aberrant cortical connections and altered neuroglial interactions during the early postnatal period.

Genome-wide transcriptional and chromatin remodeling in Ts21

To identify both shared and cell-type-specific patterns of gene expression changes, we performed differential gene expression analysis (22) at the cell class and subclass levels (table S3). Glial cells exhibited the most pronounced upregulation of gene expression, whereas vascular and glial populations demonstrated the greatest number of highly downregulated differentially expressed genes (DEGs; adj. P<0.01, FC>2) (fig. S4, A and B). Although many Ts21 DEGs were cell-type specific, we identified a subset of highly upregulated DEGs that were present across all subclasses (Fig. 1F). Only 4/121 of these shared upregulated genes mapped to chromosome 21, implicating secondary regulatory mechanisms - rather than direct gene dosage effects - in driving a conserved set of pathological gene expression. GO analysis of the shared Ts21-upregulated DEGs revealed enrichment in metabolic processes, consistent with documented mitochondrial and bioenergetic deficits in Ts21 (23) (fig. S4, C and D).

The simultaneous profiling of gene expression and chromatin accessibility in single nuclei allowed us to link regulatory elements to genes, providing insights into Ts21 gene regulatory dysfunction. After implementing stringent quality control (Methods), snATAC-seq data from 124,105 nuclei were retained for downstream analyses (fig. S1, E to H, table S1). To assess changes in Ts21 chromatin accessibility, we calculated differentially accessible regions (DARs) at the cell subclass level (table S4). Ts21-associated chromatin accessibility exhibited a striking asymmetry with widespread increases in accessible regions across cell types, a pattern robust to various statistical thresholds, DAR-calling methodologies, and select matched-sample analyses at the earliest study timepoint (Fig. 1G, fig. S4, E to J; Methods). Ts21 DARs with increased accessibility predominantly localized to promoter regions, whereas regions with decreased accessibility were primarily distal intergenic (Fig. 1H). The predominance of promoter-associated DARs suggests broad dysregulation of transcriptional activation, whereas distal accessibility changes reflect putative disrupted enhancer-promoter interactions. Ts21 regions with increased accessibility demonstrated cell-type specificity in genomic region enrichment (table S4). Ts21 ExN populations had preferentially increased accessibility in intergenic and intronic regions, whereas InNs primarily exhibited an increase in DARs at promoters (Fig. 1H, table S4). Our multiomic analysis of Ts21 gene expression and chromatin accessibility revealed a global and widespread disruption of gene regulation along with discrete, cell-type-specific perturbations at a critical period of neurodevelopment. These findings underscore a global epigenetic rewiring beyond localized chromosomal effects, aligning with prior observations of genome-wide Ts21 chromatin reorganization (24–26).

Cell-cell interaction analyses reveal Ts21 dysregulated neuronal and vascular-associated signaling networks

To delineate higher-order signaling perturbations, we constructed inferred paracrine and autocrine cell-cell communication networks independently for control and Ts21 (table S5) (27). Network-level quantification of total communication probability identified significant enrichment (p<0.05) of Ts21 inflammatory signaling mediators, including IL-6, CCL, CXCL, and SELE pathways (fig. S5A). Conversely, cell-cell adhesion and neuronal adhesion pathways were downregulated in Ts21 (fig. S5A). Across cell classes, the majority of Ts21 populations exhibited an increased number of intercellular interactions (fig. S5B). The differential strength of these interactions, however, revealed cell-type specific patterns. Neuronal classes demonstrated a general pattern of attenuated outgoing and incoming interaction strength, whereas endothelial and vascular leptomeningeal cells (VLMCs) displayed pronounced increases in outgoing interaction strength (fig. S5C). These findings implicate cell-type specific regulatory mechanisms, with Ts21-associated suppression of synaptic adhesion gene networks in neurons contrasting with amplified vascular-associated paracrine signaling.

Chromosome 21 chromatin accessibility and gene expression in Ts21

Chromosome 21 (Chr21) is the genomic epicenter of Ts21 pathophysiology. To systematically evaluate Ts21-mediated genomic dysregulation, we analyzed both chromosome-specific and genome-wide perturbations in accessibility and gene expression. Consistent with well-established gene dosage effects, we found that Chr21 DEGs exhibited, on average, a ~1.5-fold increased expression with significantly higher expression (p < 0.001) compared to non-Chr21 DEGs (Fig. 2A).

Fig. 2. — (A) Differential gene expression of chromosome 21-encoded protein-coding and lncRNA transcripts (blue) vs. non-chromosome 21 protein-coding and lncRNA transcripts (orange) across cell classes. Boxes represent the 25th and 75th percentiles and the median plotted as a central line. Whiskers represent the extension from the 25th and 75th percentile extend to the smallest and largest values within 1.5 times the interquartile range (IQR) from the quartiles and statistical significance tested using the Mann-Whitney U test (*** p<0.001). (B) Heatmap showing the percentage of differentially accessible regions (DARs) with increased accessibility in Ts21, grouped by cell subclass and autosome number (adjusted p<0.05). The numerator is the number of DARs, and the denominator is the number of combined peaks where a 1% fraction of barcodes showed a >0 number of fragments aligned. (C) Heatmap showing the percentage of highly upregulated transcripts in Ts21, grouped by cell subclass and autosome number (adjusted p<0.01, FC>2). The numerator is the number of Ts21 highly upregulated transcripts, and the denominator is the number of transcripts per chromosome. (**D to F**) Integrated visualization of gene expression and chromatin accessibility at the chromosomal scale for intratelencephalic excitatory neurons (ExN L2–3 IT, ExN L3–5 IT, ExN L6 IT) in pseudobulk. Center, ideograms for Chr21 (D), Chr 19 (E), and Chr 18 (F) depicting chromosome length in megabase pair (Mbp) and cytoband patterns. Below, snATAC FPKM coverage are depicted for control and Ts21 (green and purple, respectively) for 10 kbp bins with a rolling mean across 40 bibs. Above, the log fold change of accessibility (from FPKM units) is visualized for uniform 10 kbp bins with a rolling mean (red line) and standard deviation across 40 bins (orange shading). Mean ExN IT differential gene expression log fold change for individual genes are depicted at gene transcription start site (TSS) coordinates (blue dots). (G) Dot plot of Chr21-encoded transcription factor (TF) motif enrichment plotted by cell subclass (x-axis) and JASPAR motif (y-axis). Dot size represents motif fold enrichment in Ts21, and dot color represents statistical significance (adjusted p-value, hypergeometric test). (H) TF footprinting analysis depicting the positional distribution of Tn5 insertions within ±250 bp of TF motif of all BACH1 motif sites in peaks in select cell subclasses for control (green) and Ts21 (purple).

To delineate genome-wide alterations in chromatin accessibility and transcription, we quantified the proportion of Ts21 (1) DARs with increased accessibility and (2) highly upregulated DEGs across cell subclasses, stratified by chromosome (Fig. 2, B and C, table S4). As expected, Chr21 displayed the most pronounced elevation in both chromatin accessibility and highly upregulated DEGs. However, this also revealed widespread non-Chr21 perturbations, with Chr19 exhibiting substantial increases in accessibility and transcriptional activity that paralleled or even exceeded Chr21 increases for specific cell subclasses. In contrast, Chr18 demonstrated relative stability, with minimal Ts21-associated upregulation in gene expression or chromatin accessibility across cell subclasses (Fig, 2, B and C, fig. S4 D and E). These findings are of particular importance given the altered nuclear architecture in trisomic cells, suggesting inter-chromosomal regulatory interactions or shared mechanisms of epigenetic dysregulation at the chromosomal scale.

We next sought to examine patterns of gene expression within Chr21. Given the high correlation of Chr21 gene expression in IT ExNs (fig. S5, F and G), we grouped IT neurons in pseudobulk to examine accessibility and gene expression along the entire chromosome (Fig. 2D). Interestingly, we found that Chr21 chromatin accessibility exhibited loci-specific heterogeneity. Regions spanning Chr21q22.13-q22.3 contained markedly increased accessibility and gene expression, whereas genic-sparse regions (Chr21q21.2) demonstrated minimal alterations in accessibility (Fig. 2D and fig. S6). This demonstrates the non-uniformity of accessibility and gene expression within Chr21, underscoring both localized dosage effects and broader trans-acting regulatory disruptions in Ts21. Chromosomal-scale visualizations of Chr18 and Chr19 further emphasized the heterogeneity of accessibility and expression across the Ts21 genome (Fig. 2, E and F).

Chr21-encoded transcription factors (TFs) act as pivotal regulatory nodes in Ts21, bridging localized gene-dosage effects with genome-wide transcriptional and epigenetic disruptions. To identify TFs that may be driving chromatin accessibility changes, we performed motif enrichment for Chr21 TFs (table S4). This identified distinct, genome-wide enrichment patterns, with members of the ETS TF family (ERG, ETS2, GABPA) exhibiting pronounced overrepresentation across cell types (Fig. 2G). Motifs for BACH1 - a TF associated with impaired Ts21 antioxidant defenses (28) - were significantly enriched (adj. p < 0.01) in accessible chromatin of ExN, microglia, and oligodendrocyte lineage cells (Fig. 2G), with TF footprinting confirming elevated BACH1 activity in these populations (Fig. 2H). Correlating with motif enrichment, Chr21-encoded TF gene expression showed increased expression in relevant cell types (fig. S7A). Collectively, these findings underscore how triplication of specific Chr21-encoded TFs drives both cell type-specific and genome-wide regulatory disruptions, highlighting the complex relationship between chromosomal aneuploidy and global transcriptional networks.

Metabolic and synaptic deficits in Ts21 excitatory neurons

To evaluate the effects of Ts21 on maturing neurons, we subclustered ExNs and analyzed them independently (Fig. 3, A and B, and fig. S8A). Given our identification of pronounced transcriptomic alterations in IT ExNs (fig. S3, C and D) and compositional alterations of upper layer CUX2-expressing neurons (Fig. 1D and fig. S3E), we focused our analyses on L2–3 IT ExNs.

GO analysis of Ts21-upregulated DEGs (Methods) revealed an overrepresentation of metabolic pathways (Fig. 3C and table S6). DEGs associated with oxidative stress, immune response, and apoptosis were also enriched, including increased expression of pro-apoptotic regulators and the executioner caspase CASP3 (Fig. 3D). These findings are especially salient given the observed compositional expansion of upper-layer IT neurons in the early postnatal Ts21 prefrontal cortex (Fig. 1D). This is consistent with the increased proportion of upper-layer CUX2-expressing excitatory neurons and decreased proportions of FOXP2-expressing L3–5 IT neurons we have found in a published adolescent and adult Ts21 dataset (29). However, the more pronounced increase in L2–3 IT neuron composition we observed during the early postnatal period, together with elevated expression of pro-apoptotic factors, suggests prenatal overproduction of upper-layer IT neurons followed by apoptotic elimination during later postnatal maturation.

Ts21-downregulated DEGs in L2–3 IT neurons converged at the level of the synapse and neurite, with enrichment in axon guidance, synaptic organization, and dendritic transport processes (Fig. 3E, fig. S9). SynGO (30) analysis identified Ts21-downregulated DEGs encoding proteins predominantly localized to the postsynaptic density (PSD) membrane, with functional roles enriched for synaptic organization (Fig. 3F, fig. S9 and table S6). These synaptic deficits were conserved across IT subtypes, although cell-type-specific expression patterns suggest heterogeneous manifestations of a widespread synaptic pathology (Fig. 3G). To further delineate the shared functional pathways and identify putative network-level biology, we employed gene co-expression networks (hdWGCNA; Methods) (31) to identify modules of Ts21 DEGs with highly correlated expression (table S7). For L2–3 IT ExN, hdWGCNA identified upregulated modules enriched for metabolic and calcium channel complexes, whereas the most highly downregulated module corresponded to PSD membrane components, including scaffolding proteins critical for receptor clustering and synaptic stability (Fig. 3, H and I, fig. S8, B to F).

We found minimal representation of Chr21 genes among DEGs within key dysregulated pathways in L2–3 IT neurons: metabolism (0.79%), translation (0.88%), immune response (3.7%), and synaptic function (0%). Consequently, these molecular disruptions likely arise from indirect Chr21 regulatory mechanisms or downstream cellular dysfunction rather than direct Chr21 gene dosage effects. Supporting this, analysis of Chr21-encoded TF motif enrichment revealed significant enrichment (adj. p < 0.05) near metabolic and translational DEG gene bodies but not near synaptic or immune genes (fig. S7B). These findings suggest that Chr21 TF-mediated dysregulation predominantly affects metabolic and proteostatic pathways in Ts21, whereas immune and synaptic gene perturbations reflect downstream pathological processes.

To further identify cell-cell interaction patterns of synaptic and neurite deficits, we performed differential signaling analysis for all ExNs (Methods). This revealed enhanced extracellular matrix-related signaling along with pronounced upregulation of APP signaling in Ts21 ExNs (fig. S8G). We found an increased expression of APP across ExNs, and an enhanced APP network for all major cell types (fig. S8, H to J). This finding is of particular relevance given the localization of APP on Chr21 and its role in the pathogenesis of AD (32). Previous work has identified substantial intraneuronal amyloid-beta 42 protein accumulation in Ts21 during the early postnatal period (33, 34). This observation is pertinent, as amyloid has been shown to activate apoptotic pathways by disrupting calcium homeostasis and inducing the release of pro-apoptotic factors (35), both of which are corroborated by findings of this study.

Additionally, the most highly downregulated Ts21 signaling pathways involved synaptic adhesion molecules, with many components of the neurexin, neuroligin, and neuregulin families downregulated in L2–3 IT neurons (fig. S8K, fig. S9). Interaction analyses indicated that these deficits are mediated, at least in part, by attenuated neuron-glia synaptic adhesion interactions (fig. S8, L and M). Together, these findings reveal the molecular underpinnings for the reduced dendritic complexity, synaptic density, and aberrant dendritic spine morphology identified in previous Ts21 postmortem histochemical analyses (36–41). Immunoblotting for the synaptic-localized calcium channel CACNA1C further confirmed the downregulation of synaptic proteins (fig. S8N). The convergence of metabolic and ROS stress, apoptotic activation, and synaptic disorganization during a key period of synaptogenesis underscores the vulnerability of early postnatal Ts21 ExNs. Our findings link the developmental overproduction of L2–3 IT neurons with subsequent molecular, cellular, and synaptic dysfunction.

Conserved ExN:InN neuron ratio and global synaptic dysfunction in the Ts21 dlPFC

We next sought to systematically profile the transcriptome of inhibitory neurons (InN), a cell population critical for neural circuit formation, maturation, and refinement that has been previously implicated in Ts21 pathogenesis (42). We subclustered all InNs and annotated them based on putative developmental origin along with classical markers such as SST, PVALB, and CALB2 (Fig. 4, A and B, fig. S10A). Whereas Ts21 mouse models have identified alterations in the balance of ExN and InN (43–45), we found striking conservation in the ratio of ExN:InN (2.17:1) and the ratio of MGE:CGE InNs (1.4:1) in both control and Ts21 samples (Fig. 4B).

Fig. 4. — (A) Transcriptomic UMAP visualization of all Control (left panel) and Ts21 (right panel) inhibitory neurons (InN) analyzed in this study using a joint UMAP embedding of both Control and Ts21, colored by cell subclass. (B) Upper pie charts: proportion of Control (left) and Ts21 (right) inhibitory neurons based on putative developmental origin or cell subclass. Lower pie charts: ratio of Control (left) and Ts21 (right) excitatory and inhibitory neurons. (**C and D**) SynGO (30) plot of genes downregulated in CGE InN (C) or MGE InN (D). Each bar represents a synaptic cellular component ontology term with concentric rings representing progressively more specific subclasses of synaptic terms. Color intensity in each segment reflects the enrichment statistical significance of gene associations (−log10 Q value) for each term compared to genome-wide expectations with hypergeometric testing with Benjamini-Hochberg correction. (E) Upset plot of all downregulated SynGO synapse-associated genes plotted by neuronal subclass (adjusted p<0.01). (F) Pie chart of downregulated SynGO synapse-associated genes for ExN and InN subclasses.

As InNs transcriptomically cluster based on developmental origin (12), we grouped MGE and CGE InNs and performed hdWGCNA to identify network-level perturbations in gene expression (table S7). Ts21-upregulated gene modules in MGE (Module 1) and CGE (Module 2) neurons exhibited enrichment for metabolic and protein translation processes, whereas Ts21-downregulated modules in MGE (Module 4) and CGE (Module 8) neurons were associated with PSD components, including AMPA receptor complexes and calcium channel signaling (fig. S10, B to K). SynGO (30) analysis of downregulated DEGs in both MGE and CGE InNs confirmed significant enrichment (q-value < 0.01) for PSD membrane-associated components (Fig. 4, C and D, fig. S11). Due to this converging phenotype of downregulated genes encoding PSD proteins in both inhibitory and excitatory neurons, we compared the shared and cell-type specific patterns of Ts21-downregulated synapse-associated genes. We found that many synapse-associated DEGs were shared across neuronal subclasses (Fig. 4E, fig. S9, fig. S11). Comparative analysis of synaptic-associated DEGs revealed a high degree of overlap with 65.5% of downregulated synaptic genes shared between excitatory and inhibitory neurons (Fig. 4F).

These results reveal shared molecular perturbations across inhibitory and excitatory neuron subclasses, uncovering a pan-neuronal signature of metabolic and synaptic gene dysregulation. Validation via digital droplet PCR (ddPCR) with additional independent biological samples further corroborated upregulation of ribosomal-associated genes and downregulation of synaptic genes (fig. S10 L to P). These results demonstrate that whereas the generation, developmental patterning, and migration of InNs to the postnatal dlPFC are conserved in Ts21, widespread molecular dysregulation occurs during postnatal maturation with synaptic and metabolic impairments during interneuron circuit integration.

Cell-type-specific regulatory deficits in Ts21 oligodendrocyte maturation and myelination

Developmental hypomyelination and oligodendrocyte deficits have emerged as defining morphological hallmarks of Ts21 pathogenesis (46). To resolve lineage-specific perturbations, we annotated oligodendrocyte subpopulations from MKI67+ proliferative oligodendrocyte precursor cells (prOPCs) to mature myelinating oligodendrocytes (Oligos) (Fig. 5A and fig. S12A; Methods). To investigate alterations in proliferation, differentiation, and/or maturation, we utilized pseudotime (47) coupled with PHATE (48) to order and visualize nuclei along a reconstructed developmental trajectory (Fig. 5, B to D; Methods). This uncovered significantly increased pseudotime values (p < 0.0001) in Ts21 OPCs and Oligos - with OPCs displaying the most pronounced differences - indicative of impairments in proliferation and differentiation (Fig. 5E).

Fig. 5. — (A) Transcriptomic UMAP visualization of all Control (left panel) and Ts21 (right panel) oligodendrocyte lineage cells analyzed in this study using a joint UMAP embedding of both Control and Ts21, colored by cell subclass. Pie charts demonstrate the proportion of each oligodendrocyte lineage cell subclass for control (left) and Ts21 (right) along with the total number of oligodendrocyte lineage cells per condition. prOPC, proliferative oligodendrocyte precursor cell; OPC, oligodendrocyte precursor cell; nfOligo, newly formed oligodendrocyte; Oligo, oligodendrocyte. (**B to D**) PHATE (48) visualization of all oligodendrocyte lineage cells, colored by cell annotation (B), pseudotime (47) value (C), or condition (D). (E) Violin plot of pseudotime (47) values for OPCs and Oligos colored by condition (**** p<0.0001, unpaired t-test). (F) Waterfall plot of genes highly upregulated in Ts21 OPCs (adjusted p<0.01, FC>2). Gene Ontology: Biological Process (GO:BP) terms are plotted by term rank and normalized enrichment score. Select GO:BP terms are highlighted within the plot and dots are colored by generalized biological process. (G) Oligodendrocyte lineage gene expression dot plots of select genes from generalized biological processes; dot size represents the percentage of nuclei expressing the gene and color represents the average expression value. (H) Waterfall plot of genes highly downregulated in Ts21 OPCs (adjusted p<0.01, FC>2). Gene Ontology: Biological Process (GO:BP) terms are plotted by term rank and normalized enrichment score. Select GO:BP terms are highlighted within the plot and dots are colored by generalized biological process. (I) Oligodendrocyte lineage gene expression dot plots of select genes from generalized biological processes; dot size represents the percentage of nuclei expressing the gene and color represents the average expression value. (J) Combined heatmap and dot plot depicting select SCENIC+ (51) enhancer-driven regulon (eRegulon) activities from the oligodendrocyte lineage cell SCENIC+ gene regulatory network. Heatmap colors indicate the gene expression-based AUC scores, and dot sizes are scaled based on epigenomic region-based AUC scores. The eRegulons are grouped based on whether they are transcriptional activators (top, + regulons) or repressors (bottom, − regulons). (**K to M**) Split violin plots of genes encoding transcriptional myelin regulators (K), myelin-associated proteins (L), or glycolipid synthesis components (M) plotted by oligodendrocyte lineage cell subclass. Control and Ts21 gene expression violin plots are green and purple, respectively. (* p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001, Wilcoxon Rank Sum test).

DEG analysis of Ts21 OPCs revealed enrichment for GO terms associated with metabolism, inflammation, aging, hormone transport, and development (Fig. 5F). We identified redox metabolism-associated genes upregulated in Ts21 OPCs and Oligos - PINK1 and PARK7 - that have been implicated in cellular senescence and neurodegenerative diseases (49, 50) (Fig. 5G). Conversely, downregulated DEGs in Ts21 OPCs were enriched for the regulation of proliferation and differentiation, including cell cycle, cell polarity, and DNA repair processes, highlighting deficits in the regulation and fidelity of mitosis (Fig. 5H). Indeed, kinetochore components KNL1 and NDC80 and key mitosis regulators CDC25C and DTL were specifically downregulated in Ts21 prOPCs (Fig. 5I).

To elucidate the regulatory mechanisms underlying these deficits, we constructed a single-nucleus multiomic gene regulatory network (GRN) (51) integrating gene expression and chromatin accessibility data from oligodendrocyte lineage cells (table S8; Methods). This analysis revealed an attenuation of TF activator activity across the Ts21 oligodendrocyte lineage, particularly affecting cell cycle-related (GLI3, E2F7, BRCA1) and oligodendrocyte lineage-specific (OLIG2, SOX10, TCF7L2, MYRF) regulons, whereas Ts21 TF repressor activity was increased in a cell-type specific manner (Fig. 5J). These findings uncover the regulatory logic and underpinnings of the observed pseudotime and differentiation impairments. We identified an increase in histone deacetylase (HDAC8) repressor activity across all oligodendrocyte lineage cell types, highlighting both shared and cell-type specific gene regulatory perturbations along the developmental trajectory (Fig. 5J).

Components of the myelin sheath (52) exhibited pronounced molecular dysregulation in Ts21 oligodendrocytes (fig. S12B). Genes involved in cholesterol and glycolipid biosynthesis, as well as myelin structural components, were significantly downregulated (p < 0.05) (Fig. 5, K to M). These transcriptional deficits were most pronounced in mature oligodendrocytes relative to newly formed (nfOligos), indicating disrupted maturation from newly formed to myelinating cells (Fig. 5, K to M, and fig. S12, C to I). This transition was accompanied by reduced activity of myelin-associated gene networks and diminished regulon activity of key transcriptional regulators, including MYRF and NKX6–2, which serve as central regulators of myelin gene expression programs (53–55) (Fig. 5J).

Impaired maturation of Ts21 oligodendrocytes is likely to be mediated, at least in part, by non-cell autonomous mechanisms as evidenced by the presence of inflammation-associated DEGs across the Ts21 oligodendrocyte lineage (Fig. 5F and fig. S3, B and C). Interaction profiling revealed further attenuation of Ts21 oligodendrocyte-neuronal adhesion signaling with marked reductions in axon-glial adhesion molecule interactions (fig. S12, J to R). Together, our findings implicate dysregulated oligodendrocyte lineage progression and maturation-dependent repression of myelin gene programs as central mechanisms contributing to Ts21-associated hypomyelination, highlighting critical nodes of vulnerability in myelin gene regulatory networks.

Molecular signatures of Ts21 astrocyte dysfunction reveal regulators of reactivity, stress, and cell survival

Accumulating evidence has underscored the critical role of astrocytes in the pathophysiology of both DS (56) and AD (57). Using the pan-astrocyte marker AQP4, we identified and subset astrocytes from Ts21 and control. Dimensionality reduction and unsupervised hierarchical clustering demonstrated unique clustering of Ts21 astrocytes, implying fundamental alterations in cellular states or disease-associated gene expression (Fig. 6A).

Fig. 6. — (A) Upper panel: astrocytes were identified and subset by expression of the pan-astrocyte marker *AQP4*. Bottom panel: Transcriptomic UMAP visualization of all Control and Ts21 astrocytes analyzed in this study using a joint UMAP embedding, colored by condition. (**B and C**) Split gene expression violin plots of control (green) and Ts21 (purple) astrocytes for consensus genes upregulated (B) or downregulated (C) in reactive astrocytes (58) (**** p<0.0001, Wilcoxon Rank Sum test). (D) Transcriptomic UMAP visualization of all Control (left) and Ts21 (right) astrocytes analyzed in this study using a joint UMAP embedding, colored by astrocyte state. (E) Proportion of control (left) and Ts21 (right) homeostatic and reactive astrocytes. (F) Hallmark gene set (81) enrichment plot for genes upregulated in control reactive astrocytes as compared to Ts21 reactive astrocytes (DEGs: adj. p<0.01, Wilcoxon Rank Sum test). Normalized enrichment score is plotted as bar height and bar color represents hallmark gene set enrichment adjusted p value (Benjamini-Hochberg procedure). (**G to I**) Astrocyte gene expression dot plots of select genes from Hallmark gene sets, (P53 pathway (G), Apoptosis (H), Hypoxia (I)) significantly enriched in control reactive astrocytes; dot size represents the percentage of nuclei expressing the gene and color represents the average expression value. **(J and K**) Astrocyte gene expression dot plots of apoptosis inhibitors stratified by condition (J) or condition and astrocyte state (K). H, homeostatic astrocytes; R, reactive astrocytes. **(L)** Co-expression network visualization of astrocyte Module 8 with the top 25 module hub genes depicted. Hub genes highlighted in red are reactive astrocyte marker genes with increased expression in Ts21 that have been validated with ddPCR in this study (fig. S9). **(M)** Bar plot depicting gene set enrichment results for Module 8 of the astrocyte co-expression network with the top Gene Ontology: Cellular Components (GO:CC) enrichment terms visualized. **(N)** Combined heatmap and dot plot depicting AP-1 complex family SCENIC+ enhancer-driven regulon (eRegulon) activities in astrocytes. The SCENIC+ (51) gene regulatory network was obtained using all non-neuronal cells. Heatmap colors indicate the gene expression-based AUC scores and dot sizes are scaled based on epigenomic region-based AUC scores. All depicted eRegulons are transcriptional activators (+ regulons). (**O and P**) Astrocyte gene expression dot plots of cellular stress-related proteins stratified by condition (O) or condition and astrocyte state (P). H, homeostatic astrocytes; R, reactive astrocytes. (Q) Western blotting for HSPB1. Bar graph demonstrates densitometric analysis of HSPB1 band intensity normalized to Revert^™ 700 Total Protein Stain (LICORbio). Individual dots represent blotting technical replicates (* p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001, two-tailed unpaired t-test).

Transcriptomic analysis revealed pronounced molecular perturbations in Ts21 astrocytes, including significant upregulation (p < 0.0001) of reactivity-associated markers (58) and concurrent downregulation of astrocytic homeostatic genes (Fig. 6, B and C, fig. S13A). We confirmed the expression of these upregulated reactive astrocyte markers and attenuated homeostatic genes with ddPCR validation in biologically independent, age-matched samples (fig. S13, B to G). Our results show that 51.8% of Ts21 astrocytes displayed reactive molecular features in the early postnatal period compared to only 0.5% in controls (Fig. 6, D and E, and fig. S13A).

Following the identification of reactive astrocytes in both control and Ts21 samples, we sought to elucidate the conserved and divergent gene expression programs governing reactivity. DEG analysis revealed enrichment for P53 pathway, hypoxia, and apoptosis pathways in control reactive astrocytes as compared to Ts21 reactive astrocytes (Fig. 6F). We identified enrichment of TP53 and executioner caspase CASP7 expression specifically in control reactive astrocytes, suggesting that astrocyte reactivity in control samples is likely related to postmortem interval, with enrichment for hypoxia-dependent gene expression and activation of the TP53-dependent apoptotic pathway (Fig. 6, G to I). In contrast, Ts21 astrocytes - both homeostatic and reactive - displayed molecular signatures consistent with apoptosis resistance, including upregulation of apoptotic inhibitors (Fig. 6, J and K). Together, these findings uncover the recruitment of astrocytic gene expression programs for the inhibition of apoptosis and adaptive response to Ts21 pathology in the developing dlPFC.

hdWGCNA revealed the Ts21-specific downregulation of key genes associated with astrocyte homeostasis (Module 7), along with upregulation of a gene module consisting of established reactive astrocyte markers - CRYAB and HSPA6 - and numerous heat shock protein and molecular chaperone genes (Module 8) (Fig. 6L and fig. S13, H to K). Module 8 was also highly enriched for targets of the transcription factor AP-1 complex, a dimeric protein assembly critical for immune regulation and a recently identified Ts21 therapeutic target (25) (Fig. 6M).

Given the drastic transcriptomic and chromatin accessibility alterations in Ts21 glia, we constructed a multiomic GRN encompassing all non-neuronal cells in our dataset to uncover the regulatory architecture underlying these deficits (table S8 and fig. S14). This revealed a marked increase in the activity of canonical transcription factors governing reactivity (STAT3, ETS2, NKFB1) and pronounced activation of AP-1 family members in Ts21 astrocytes, except for ATF7, which exhibited reduced activity (Fig. 6N). This Ts21-specific reduction in ATF7 regulon activity is particularly relevant, given the role of ATF7 in epigenetic repression of inflammatory pathways and inhibition of NF-Kβ signaling (59), aligning with exacerbated inflammatory signaling in Ts21 astrocytes. Our findings reveal the Ts21 AP-1 complex as a key mediator of reactivity with TF-target gene links to dysregulated networks of reactivity and inflammation-associated gene expression, including JAK3, CRYAB, and APOD (fig. S13L).

Furthermore, AP-1 targets in Ts21 astrocytes were enriched for components of cellular stress responses, particularly heat shock proteins (HSPs) and molecular chaperones. We identified a broad upregulation of stress-responsive HSPs, of which we found both Ts21-specific and reactivity-associated enrichment (Fig. 6, O and P). Stress response activation reflects astrocytic plasticity in Ts21, balancing pro-survival adaptations with stress-related transcriptional programs. Of particular interest was the prominent enrichment of HSPB1, a key stress response protein and reactive astrocyte marker that has recently been shown to be secreted proximal to amyloid plaques where it exerts non-cell-autonomous neuroprotection through the stabilization of proteostasis (60). We confirmed significantly increased protein expression of HSPB1 in the early postnatal Ts21 cortex via immunoblotting (p < 0.05) (Fig. 6Q).

Collectively, these findings demonstrate that Ts21 astrocytes exhibit profound reactivity, driven by AP-1 factor complex regulation and stress response-mediated adaptation. The identification of AP-1 and stress response components as central regulators of astrocyte pathology highlights their potential as therapeutic targets for modulating astrocyte dysfunction in DS and related neurodegenerative disorders.

Cell intrinsic gene regulation in Ts21 neuroinflammatory microglial activation

Microglia play a critical role in activity-dependent synaptic pruning and the refinement of neural circuits during key periods of synaptogenesis (61). To systematically analyze microglia, we annotated and subset these cells using the pan-microglial markers APBB1IP and P2RY12 (Fig. 7A). Both control and Ts21 cells were enriched for microglia-specific markers and lacked molecular features of non-resident immune cells (fig. S15A). Ts21 microglia demonstrated marked upregulation of activation-associated markers (ITGAX, SOCS3, FCGR2A) accompanied by downregulation of homoeostatic genes (SORL1, P2RY12, CX3CR1) (Fig. 7, B and C). Classification of microglial states based on these markers revealed a shift in Ts21 microglia toward an activated phenotype, with all analyzed cells exhibiting molecular features of activation (Fig. 7, D and E, and fig. S15, B to E). ddPCR validation further confirmed pronounced neuroinflammatory molecular signatures (fig. S15, F to J).

Fig. 7. — (A) Upper panel: microglia were identified and subset by expression of the pan-astrocyte marker *APBB1IP*. Bottom panel: Transcriptomic UMAP visualization of all Control and Ts21 microglia analyzed in this study using a joint UMAP embedding, colored by condition. (**B and C**) Split gene expression violin plots of control (green) and Ts21 (purple) microglia for genes upregulated (B) or downregulated (C) in activated microglia (**** p<0.0001, Wilcoxon Rank Sum test). (D) Transcriptomic UMAP visualization of all Control (left) and Ts21 (right) microglia analyzed in this study using a joint UMAP embedding, colored by microglia state. (E) Proportion of control (left) and Ts21 (right) activated and homeostatic microglia. (**F to H**) Circle plot visualizing the inferred paracrine and autocrine signaling networks for IL-1 (F), P-selectin (G), and E-selectin (H). Each node represents a distinct cell subclass. Directed edges indicate statistically significant ligand–receptor-mediated interactions between cell groups where edge color corresponds to the sender (source) cell subclass and edge thickness reflects the relative interaction strength between cell groups. (**I to L**) Microglia gene expression dot plots of select genes from complement component (I), complement regulator (J), MHC class I (K), or MHC class II (L) HUGO gene families; dot size represents the percentage of nuclei expressing the gene and color represents the average expression value, (M) Circle plot visualizing the inferred APOE/TREM2 interaction network for control (left) and Ts21 (right). Directed edges indicate statistically significant ligand–receptor-mediated interactions between cell groups where edge color corresponds to the sender (source) cell subclass and edge thickness reflects the relative interaction strength between cell groups with equivalent maximum edge weight between control and Ts21. (**N to P**) Gene expression violin plots for complement receptors (N), complement components (O), and MHC class II (P). Violin plots are stratified by microglia subtypes: Micro *CD200 PRKAA2* and Micro *CTSC* DOCK1 for control microglia (green) and Ts21 microglia (purple). (Q) Representative images depicting AIF1+ positive microglia in the dorsolateral prefrontal cortex of control (upper panel) and Ts21 (lower panel) samples. The Ts21 panel highlights an AIF1+/HLA-DRB1+ microglial cell (white arrow), while the control panel shows AIF1+/HLA-DRB1− microglia (white arrows) and an AIF1−/HLA-DRB1+ intravascular immune cell (black arrow). Scale bar, 20μm. (R) Quantification of the percentage of AIF1+ microglia expressing HLA-DRB1 in dlPFC grey matter or white matter of control (age: 2y, 4d postnatal) and Ts21 (age: 2y, 57d postnatal) samples. Each dot represents quantification from one tissue section (**** p<0.0001, two-tailed unpaired t-test).

To investigate the potential mechanisms underlying microglial activation, we employed a multiomic GRN analysis to identify regulons underlying the activated microglial state (table S8). This revealed increased activity of central inflammatory regulators, including PPARG, NFKB1, and the Chr21-encoded TF RUNX1 (fig. S14). We identified decreased activity of the PPARG repressor regulon, indicating diminished suppression of inflammatory gene expression alongside the activation of pro-inflammatory pathways (fig. S14). Network analysis further revealed key interactions associated with inflammation, including regulation of the Chr21-encoded interferon receptor IFNGR2 by both RUNX1 and NFKB1, suggesting a Chr21-driven positive feedback loop wherein RUNX1 enhances IFNGR2 expression, thereby exacerbating interferon signaling and inflammation (fig. S16A).

Given the dysregulated interferon signaling observed across multiple Ts21 organ systems (62, 63), we characterized the expression of interferon receptors, along with downstream interferon response genes. This revealed many interferon receptors highly upregulated across Ts21 cell types (fig. S16, B and C). Furthermore, all cell classes demonstrated marked upregulation of interferon response genes, underscoring a broad and systemic interferon response within the early postnatal dlPFC (fig. S16D). VLMCs exhibited the highest number of upregulated interferon response genes, many of which were cell-type-specific, highlighting meningeal cells as a critical node for Ts21 inflammatory signaling (fig. S16D).

Dysregulated neuroimmune crosstalk and subtype-specific molecular features of Ts21 microglia

Given the concurrent dysregulation of microglia and astrocytes, we investigated cytokine-mediated signaling as a non-cell-autonomous driver of Ts21-associated neuroinflammation. Interaction analyses identified an enrichment of pro-inflammatory IL-1 and IL-6 signaling pathways, revealing inflammatory paracrine signaling between Ts21 astrocytes, microglia, and vascular-associated cells (Fig. 7F and fig. S17, A and B). Vascular signaling networks demonstrated marked upregulation of pro-inflammatory selectin and galectin pathways (Fig. 7, G and H, fig. S17, C and D), which promote feedforward stimulation of interleukin production (64–66). E-selectin (SELE), a cytokine-induced endothelial adhesion molecule, was uniquely expressed in Ts21 vascular cells (Fig. 7G), whereas P-selectin signaling was elevated in microglia–vascular interactions (Fig. 7H), consistent with its role in immune cell modulation and trafficking (66). These networks delineate coordinated vascular-glial-immune crosstalk that amplify neuroinflammatory cascades via cytokine signaling and adhesion molecule–mediated interactions. Together, these findings identify key regulators of microglial inflammation and interferon activation, highlighting both Chr21-encoded factors, such as RUNX1 and IFNGR2, and broader non-cell-autonomous inflammatory signaling as central drivers of Ts21 neuroinflammation.

Inflammatory signaling and microglial activation are particularly consequential during the early postnatal period, a critical window for synaptogenesis and complement-dependent synaptic pruning. Ts21 microglia exhibited pronounced upregulation of complement components - critical for synaptic tagging and engulfment (67) - alongside elevated expression of MHC class I/II receptors (Fig. 7, I to L). Subsequent microglial subclustering identified two predominant subtypes: Micro CD200 PRKAA2 and Micro CTSC DOCK1 – present in both control and Ts21 (fig. S17, E to I). Both Ts21 microglial subtypes showed enhanced TREM2-APOE interactions, a pathway central to apoptotic neuron clearance and the transition to activated microglial states (Fig. 7M). Dysregulation of the TREM2/APOE axis has been implicated in heightened neuroinflammation, neuronal injury, and complement-mediated synaptic clearance, representing a critical therapeutic target in AD pathology (68).

Transcriptomic profiling further revealed divergent molecular properties between microglial subtypes. Ts21 Micro CTSC DOCK1 displayed increased expression of complement receptors critical for phagocytosis, including CR3 (ITGAM/ITGB2), CR5 (ITGAX/ITGB2), and C3aR (C3AR1), along with MHC class II components essential for antigen presentation (Fig. 7, N to P). This subtype also exhibited increased interferon receptor expression (fig. S16C) and upregulation of MHCII HLA-DRB1 (Fig. 7P), a marker consistently elevated in AD (69). Immunohistochemical analyses confirmed a significant increase (p < 0.0001) in the proportion of HLA-DRB1+/AIF1+ microglia and the immunointensity of HLA-DRB1 in the early postnatal Ts21 dlPFC, with more pronounced alterations observed in white matter (Fig. 7, Q and R, and fig. S17, J to L). Additionally, multiplex immunoassays (Methods) in an expanded early postnatal cohort further revealed dysregulation of inflammatory cytokines in the Ts21 cortex (fig. S17, M to Q).

Together, these results demonstrate unique molecular signatures of Ts21 activated microglial subtypes, highlighting key functional differences in complement-mediated phagocytic activity and antigen presentation. These findings underscore the role of neuroimmune dysfunction in disrupted synaptic refinement during Ts21 early brain development and identify key pathways and cell subtypes for potential therapeutic intervention.

Conclusions

This study presents a comprehensive single-nucleus multiomic analysis of the early postnatal dlPFC in Down syndrome, revealing a complex landscape of molecular and cellular dysregulation that underpins both the neurodevelopmental and the neurodegenerative features of this disorder. Our analyses identified broad and asymmetric increases in chromatin accessibility and gene expression across diverse cell types and chromosomes in Ts21. We uncovered shared patterns of differential accessibility, highlighting a pervasive epigenetic reorganization that extends beyond classical “gene dosage effects” and implicates both local and trans-acting regulatory mechanisms in molecular pathogenesis.

We further identified a proportional expansion of excitatory L2–3 intratelencephalic (ExN L2–3 IT) neurons in the neocortex, suggesting a potential cellular substrate for the cortico-cortical hyperconnectivity reported in individuals with Down syndrome (70–74). Additionally, our data revealed multicellular dysfunction within the prefrontal cortex converging at the level of neurites and synapses. We identified molecular correlates of intrinsic pan-neuronal synaptic deficits, oligodendrocyte lineage progression and functional myelination defects, impaired neuron-glia adhesion, and activated antigen-presenting microglia during a critical period of synaptogenesis. These findings establish a molecular framework for the circuit-level dysfunction observed in Down syndrome and highlight the vulnerability of neurite outgrowth, synaptic formation, and functional maturation during early postnatal brain development.

Accumulating evidence supports the classification of DS as a multisystem interferonopathy (62, 63, 75). We identified a robust interferon response signature across all dlPFC cell types analyzed in this study, highlighting interferon activation at an early and critical stage of brain development (fig. S16). In addition, we identified a pronounced neuroinflammatory signature encompassing glial, immune, and vascular-associated cells along with cell-cell interaction networks and upstream transcriptional inflammatory mediators. Ts21 astrocytes exhibited a distinct shift toward a reactive phenotype, with upregulation of stress-responsive heat shock proteins and molecular chaperones. We identified that this reactivity is driven in part by the activation of the key transcription factor AP-1 complex and is accompanied by the suppression of homeostatic gene expression and upregulation of cell-survival gene expression programs. The AP-1 complex has emerged as a potential therapeutic target bridging interferon signaling and downstream cellular pathology in Ts21 and represents a potential target for the modulation of astrocytes in Down syndrome (25).

Furthermore, Ts21 microglia displayed marked activation in the early postnatal period, with increased expression of inflammatory markers, complement cascade components, and MHC class I/II receptors, as well as subtype-specific enrichment for phagocytic and antigen presentation pathways. We identified inflammatory mediators that drive this activation and further potentiate interferon receptor expression through regulation of the pro-inflammatory interferon receptor IFNGR2. This is especially relevant given interferon-responsive microglia have been shown to be highly phagocytic and actively shape cortical development (76), suggesting that interferon-induced microglia pathologically and excessively prune synapses critical for physiological circuit maturation in the Ts21 dlPFC. Our analyses identify a Ts21 microglial subtype - Micro CTSC DOCK1 - as an interferon-responsive microglial therapeutic target. We further identified glial and immune state alterations to be amplified by dysregulated cytokine signaling and enhanced vascular-immune-glial crosstalk, establishing a multicellular feedforward inflammatory axis in the Ts21 brain.

DS is among the select few conditions that encompass both neurodevelopmental and neurodegenerative clinical features, with most individuals demonstrating progressive cognitive impairment by mid-adulthood (77, 78). Whereas it is well-established that the neuropathological substrates of dementia precede clinical onset (79), a central question is the timing of onset of these molecular changes. This study identified molecular signatures corresponding to six of eight established neurodegeneration hallmarks (80) - including metabolic dysfunction, oxidative stress, synaptic loss, neuroinflammation, impaired proteostasis, and activation of cell death pathways - during the early postnatal period in the Ts21 PFC. The concurrence of neurodevelopmental deficits and neurodegenerative signatures supports a model in which these processes are mechanistically interconnected, potentially compounding cognitive and functional impairment from the earliest stages of life in DS. These findings shift the temporal onset of DS-associated neurodegenerative processes to the early postnatal period and underscore the necessity for early biomarker development, therapeutic innovation, and clinical trials targeting disease modification within this critical developmental window.

Materials and Methods

Postmortem brain samples

This study was performed in accordance with the ethical and legal guidelines of the University of Wisconsin-Madison Institutional Review Board. Tissue samples were handled under ethical guidelines and regulations for the research use of human brain tissue set forth by the National Institutes of Health (NIH) and the WMA Declaration of Helsinki (https://www.wma.net/policies-post/wma-declaration-of-helsinki-ethical-principles-for-medical-research-involving-human-subjects/). A detailed list of all specimens used in this study is available in table S1.

Nucleus isolation for single-nucleus multiome (snATAC-seq and snRNA-seq)

Frozen dorsolateral prefrontal cortex (dlPFC) specimens were cryogenically pulverized using liquid nitrogen-chilled mortar and pestle (Fisherbrand FB961A/FB961K). Pulverized dlPFC tissue (25–35 mg) was subsequently processed for nuclear extraction. All solutions were prepared immediately prior to use and maintained at 4°C throughout the procedure. For nuclear isolation, 3 mL of ice-cold Iodixanol buffer was prepared consisting of 50% Optiprep (Iodixanol) (Sigma#D1556) solution (v/v), 25 mM KCl (Sigma #60142), 5mM MgCl2 (Sigma #M1028), 20mM Tris-HCl (pH 7.5) (Invitrogen #15567–027), 1% cOmplete^™ Mini, EDTA-free Protease Inhibitor Cocktail (Roche#11836170001), 1% BSA (GeminiBio #700–100p), RNase inhibitor (80U/ml) (Roche #03335402001), and 1mM DTT (Sigma #43186) and transferred into a 15 mL centrifuge tube on ice. 1mL of lysis buffer - 250 mM sucrose (Sigma #S0389), 25 mM KCl2 (Sigma #21115), 5 mM MgCl2 (Sigma #63052), 20mM Tris-HCl (pH 7.5) (Invitrogen #15567–027), 1% cOmplete^™ Mini, EDTA-free Protease Inhibitor Cocktail (Roche#11836170001), RNase inhibitor (40U/μl) (Roche #03335402001), 1mM DTT (Sigma #43186), and 0.1 % IGEPAL CA-630 (v/v) (Sigma#I8896) - was then added to the pulverized tissue. Another 1mL of ice-cold lysis buffer was added to the douncer. An additional 1 mL of lysis buffer was then added to the pulverized tissue tube to rinse and collect all tissue. The suspension was subsequently homogenized using a 15 mL RNase-free Dounce tissue grinder (Wheaton #357544) with 30 strokes each of loose and tight pestles, applying consistent pressure while avoiding air introduction. The homogenate was transferred into the 15 mL centrifuge tube containing the Iodixanol buffer and mixed by gentle inversion (10 times), then passed through a pre-wetted 40 μm cell strainer (Falcon #352340). The filtered homogenate was centrifuged at 1,000 × g for 30 minutes at 4°C (Eppendorf #5910 Ri). The supernatant was carefully removed, and the pellet was resuspended in 1 mL of wash buffer consisting of 10 mM NaCl (Sigma #60142), 3 mM MgCl₂, 10 mM Tris-HCl (pH 7.5), 1% BSA, RNase inhibitor (1,000 U/mL), and 1 mM DTT. This suspension was filtered again through a 40 μm strainer and centrifuged at 500 × g for 5 minutes at 4°C. The resulting supernatant was removed, and the nuclei pellet was gently resuspended in 200 μL followed by 300 μL of Buffer C (10 mM NaCl, 3 mM MgCl₂, 10 mM Tris-HCl [pH 7.5], 0.01% Tween-20 [Bio-Rad #1662404], 0.001% Digitonin [Thermo Fisher #BN2006], 0.01% IGEPAL CA-630, 1% cOmplete^™ Mini EDTA-free Protease Inhibitor Cocktail, 1% BSA, RNase inhibitor [1,000 U/mL], 1 mM DTT). The nuclei suspension was incubated on ice for 5 minutes, followed by addition of 500 μL of Buffer D (identical to Buffer C but containing 0.1% Tween-20 and no Digitonin or IGEPAL). The nuclei were gently resuspended and assessed for quality and concentration using a hemocytometer under 40× magnification. Based on the initial counts, the nuclei were resuspended to a final concentration of 5 × 10⁶ nuclei/mL. Remaining nuclei were pelleted at 500 × g for 5 minutes at 4°C, then resuspended in 1× Nuclei Buffer (10x Genomics, #2000207) supplemented with 1,000 U/mL RNase inhibitor and 1 mM DTT.

snMultiome library generation and sequencing

Single-nucleus multiome (snMultiome) library preparation and sequencing was performed following the 10x Genomics Chromium Next GEM Single Cell Multiome ATAC + Gene Expression protocol with sample-specific targeting of 10,000 nuclei per reaction. Nuclei were incubated with a transposition mix, enabling concurrent chromatin fragmentation and adaptor tagging. The transposed nuclei were then combined with barcoded gel beads and partitioning oil (10x Genomics, PN-2000190), loaded into a Next GEM Chip J (10x Genomics, PN-2000264), and processed using the Chromium Controller (10x Genomics, PN-120270) to generate Gel Bead-in-Emulsions (GEMs), each containing a single barcoded nucleus. After GEM generation and incubation, cleanup was performed to recover barcoded DNA and RNA fragments. Preamplification of the recovered material followed, with ATAC libraries constructed using the Single Index Kit N Set A (10x Genomics, PN-1000212). Simultaneously, snRNA-seq libraries were generated using the Library Construction Kit (10x Genomics, PN-1000190) and Dual Index Kit TT Set A (10x Genomics, PN-1000215), according to the manufacturer’s instructions. Library quality and concentration were evaluated at each stage using the Agilent 4200 TapeStation to ensure the integrity of cDNA, ATAC, and gene expression libraries. Final libraries were sequenced on an Illumina NovaSeq X Plus system, targeting a sequencing depth of 50,000 raw reads per nucleus.

Alignment and processing of snMultiome data

The 10x multiomic data was processed using BRAIN Initiative Cell Atlas Network (BICAN) default methods with the human hg38 genome assembly and BICAN filtered annotations (https://storage.googleapis.com/gcp-public-data--broad-references/hg38/v0/star/v2_7_10a/modified_v43.annotation.gtf ). Specifically, the snRNA-seq data was aligned with STAR v2.7.11a (76) and aggregated into count matrices with STARSolo. Alignments were performed with default arguments from the WARP Multiome v5.9.0 pipeline [https://broadinstitute.github.io/warp/docs/Pipelines/Multiome_Pipeline/README]. In particular, the STAR alignment was performed with the –GeneFull_Ex50pAS argument to include alignments overlapping exonic regions, per the nuclear RNA biology present in these experiments. Quality control checks included assessment of the percentages of uniquely- and multiply-mapped reads, and statistics of corrected barcodes and UMI. The snATAC-seq data was processed with the WARP Multiome v5.9.0 pipeline. This includes barcode correction with the fastqProcessing processing tool from WarpTools (https://github.com/broadinstitute/warp-tools/tree/develop/tools), and subsequent trimming with Cutadapt v4.4 (77). Alignment was performed with BWA-MEM2 (78). Finally, the generation of fragment files and initial QC with SnapATAC v2.3.1. This was enabled using an updated version of the snATAC component from the WARP Multiome v5.9.0 pipeline, adapted for implementation on Amazon AWS EC2 instances. Docker images for Cutadapt v4.4 and the BWA alignment were prepared to leverage AVX2 processing for improved performance. All Ts21 and control samples were treated identically in this portion of the analysis.

snRNA-seq data processing

Due to the presence of an additional copy of chromosome 21 in Ts21 samples, we modified the standard preprocessing pipeline to account for genotype-specific differences in transcript abundance. The extra chromosome results in an increased number of unique molecular identifier (UMI) counts per nucleus. Typically, quality control (QC) procedures impose upper and lower thresholds on per-cell UMI counts, with the upper threshold primarily used to remove potential homotypic doublets: instances where nuclei of the same cell type are combined to make a doublet. However, applying the same upper threshold across both Ts21 and control samples would disproportionately filter out high-quality Ts21 nuclei while retaining low-quality control nuclei. To address this, we implemented an adaptive thresholding approach tailored to each genotype to ensure equitable retention of high-quality nuclei across Ts21 and control samples.

Beyond direct gene dosage effects of the additional copy of chromosome, Ts21 also influences transcriptomic signatures across other chromosomes. These genotype-driven differences are expected to introduce transcriptomic variability, potentially confounding clustering and cell type annotation. For example, if highly variable genes are identified from a merged dataset comprising both Ts21 and control samples, some of these genes may reflect genotype-specific differences rather than true cell type distinctions or batch effects. To mitigate this, we identified highly variable genes independently for each genotype, thereby enhancing the detection of cell type differences and improving clustering accuracy. Consequently, we implemented snRNA-seq processing in two phases: (1) separate processing of control and Ts21 samples and (2) integration of both datasets.

Phase 1: Quality Control, Batch Correction, and Cell Type Annotation

In Phase 1, control and Ts21 samples were processed independently. First, we employed SoupX v1.6.2 (79) to remove ambient RNA, which could otherwise distort cell type identification and downstream analyses. Next, low-quality nuclei were filtered using stringent QC criteria: fewer than 200 expressed genes, ribosomal gene content exceeding 40%, mitochondrial gene content exceeding 5%, and UMI counts outside the range of 800 to Q3 + 3(Q3 − Q1), where Q1 and Q3 represent the lower and upper quartiles, respectively (13, 80) (fig. S1, A to D).

Following initial QC, we performed doublet removal using an ensemble approach to account for the inherent uncertainties in doublet detection methods. We incorporated three tools: DoubletFinder v2.0.4 (81), Scrublet v0.2.3 (82), and scDBIFinder v1.18.0 (83) and classified a nucleus as a doublet if it was flagged by at least two of the three methods. These doublets were removed from the dataset (fig. S1, I and J). Notably, both genotypes underwent the same QC pipeline, with the caveat of genotype-specific upper thresholds for UMI counts. Additionally, steps such as SoupX correction and doublet removal were performed per sample, making them independent of genotype. After filtering low-quality nuclei, we log-normalized the SoupX-corrected UMI counts using the NormalizeData() function in Seurat v5.1.0 (84). We then identified the top 3,000 highly variable genes independently for each genotype using the variance-stabilizing method in Seurat. Gene expression data for these highly variable genes were subsequently scaled, and dimensionality reduction was performed via principal component analysis (PCA). To correct for batch effects, we applied Harmony v1.2.3 (14) within Seurat.

Cell type annotations were performed using an iterative, hierarchical clustering approach as previously described (13). Ts21 and control nuclei were annotated independently to minimize integration bias and identify putative Ts21-specific cell subpopulations or states. First, nuclei were clustered via Louvain algorithm (85) using batch corrected Harmony components and annotated at (1) the major cell type hierarchical level comprising cortical excitatory neurons (ExN – RBFOX3, SLC17A6, SATB2), inhibitory neurons (InN – RBFOX3, GAD1, GAD2), or non-neuronal cells (NNC – lack of expression of aforementioned markers). We then independently subclustered the three major class cell types to transcriptomically annotate (2) the cell subclass level.

Glutamatergic ExN subclasses were defined by expression of canonical cortical layer and projection markers: layer 2–3 intratelencephalic (L2–3 IT – CUX2), layer 3–5 intratelencephalic (L3–5 IT – RORB, FOXP2), layer 5 extratelencephalic (L5 ET – POU3F1), layer 6 intratelencephalic (L6 IT – OPRK1), layer 6 corticothalamic (L6 CT – SYT6), layer 6 near-projecting (L6 NP – HTR2C), layer 6B (L6B – NR4A2), and Cajal-Retzius cells (RBFOX3) (fig. S8A). GABAergic InN subclasses were defined by the expression of marker genes corresponding to their inferred developmental origins from the medial ganglionic eminence (MGE – SOX6, LHX6), caudal ganglionic eminence (CGE – PROX1, NR2F2), or dorsal lateral ganglionic eminence (dLGE – MEIS2, TSHZ1, FOXP2) (Schmitz et al. 2022), along with canonical InN markers (fig. S10A). Non-neuronal cell subclasses were defined by the expression of principal marker genes: astrocytes (Astro - AQP4), microglia (Micro - PTPRC, APBB1IP), oligodendrocyte precursor cells (OPC – PDGFRA, PCDH15), oligodendrocytes (Oligo – MBP, MOG, OPALIN), endothelial cells (Endo – CLDN5, FN1), and vascular leptomeningeal cells (VLMC – COL1A1, COL1A2) (fig. S1, K and L). Cellular states were annotated following the cell-type specific subclustering of astrocytes (Fig. 6A) and microglia (Fig. 7A). Reactive astrocytes were annotated based on increased expression of HSPB1, GFAP, VIM, and CRYAB, accompanied by reduced expression of SLC1A3 and KCNJ10. Activated microglia were annotated based on increased expression ITGAX, SOCS3, FCGR2A accompanied by reduced expression of SORL1, P2RY12, CX3CR1.

Following cell subclass annotation, we defined the final hierarchical level of annotation: (3) transcriptomically-defined cell subtypes for select cell subclasses (fig. S3, E and G). Cell subtype marker genes were defined by comparing the gene expression of a given subtype with the residual subtypes of its respective subclass and class. The FindMarkers() function in Seurat was used to perform the Wilcoxon rank sum test, and genes with expression in more than 25% of cells within the respective subtype, and Bonferroni corrected p-value less than 0.01 were identified as potential subtype marker genes. A subtype-restricted marker gene was identified from this list following a comparison of gene expression levels and expression fold change among all subtypes of the given subclass and class.

Phase 2: Data Integration

Following cell annotation, the retained control and Ts21 samples were merged within Seurat and preprocessed. First, the SoupX-corrected UMI counts were log-normalized, and the top 3,000 highly variable genes were identified across both genotypes. Gene expression data for these highly variable genes were then scaled, followed by principal component analysis (PCA). Batch effects were removed using Harmony, ensuring better integration of samples. Finally, a joint UMAP embedding of control and Ts21 nuclei was generated (Fig. 1C).

Evaluation of clustering stability and cell type separability

The accuracy of cell type annotation is highly dependent on the ability of the clustering method to identify transcriptomically similar cells. Since the clustering method primarily uses a shared neighbor graph obtained at reduced dimensions, the cluster stability or the accuracy of the cell type annotations can be determined by investigating cell neighbors. Thus, a neighbor voting-based approach was adopted (12, 13). Because cell types were annotated separately for each genotype, the analysis was conducted independently for control and Ts21 samples. First, we subsampled the original dataset by randomly selecting a subset of cells without replacement (considering 5% and 10% of the original sample). The subsampled dataset was then preprocessed, including log-normalization, selection of the top 3,000 highly variable genes, scaling, principal component analysis, and batch correction using Harmony. Next, the 20 nearest neighbors for each nucleus were identified in the reduced-dimensional space. New cell-type predictions were made using the neighbor voting approach, where each nucleus was labeled based on the most abundant subclass among its neighbors (requiring at least 60% agreement). Predictions were performed for only 2,500 randomly selected nuclei from the subsampled data. To quantify the consistency of the annotations, we computed the area under the receiver operating characteristic (AUROC) scores for each cell subclass, comparing the original and predicted annotations. AUROC values were highly concordant for the two subsampled datasets across genotypes underscoring inherent cluster stability (fig. S1M).

Transcriptomic similarities/differences across cell type and genotype

To systematically assess transcriptomic similarities and differences across cell types and genotypes, we performed unsupervised hierarchical clustering on pseudobulked nuclei, stratified by cell subclass and genotype. This analysis was conducted independently for two sets of the top 3,000 highly variable genes, each identified separately within Control and Ts21 samples. Hierarchical clustering was implemented using the hclust() function in R with the “complete” linkage method, which is best suited to finding similar clusters. Additionally, to investigate the contribution of Chr21 gene expression to cell type separability, we conducted a parallel analysis focusing on genes located on Chr21. Pseudobulk nuclei were generated by cell subclass, and pairwise Pearson correlations were computed to quantify transcriptomic similarity. The resulting correlation matrices were visualized as heatmaps for each genotype (fig. S5, F and G), facilitating direct comparison of transcriptomic patterns attributable to genotype and cell type.

Transcriptomic heterogeneity within cell types

To evaluate the transcriptomic variability within cell types across genotypes, we implemented an entropy-based approach (12). Briefly, this method estimates transcriptomic heterogeneity by subtracting the unstructured entropy (permuted data) from the actual entropy, which is evaluated using real data. For each cell type, the calculation includes (1) down-sampling both control and Ts21 samples to have the same number of nuclei. The number of retained nuclei was taken as 75% of nuclei from the smallest dataset. For cell types that have more than 10,000 nuclei (L2–3 IT and L3–5 IT subclasses), we only selected 30% of the nuclei to reduce the computational burden. (2) Obtaining the union of the top 2000 highly variable genes estimated on Control and Ts21 samples for downstream analysis. (3) The gene expression of the unified highly variable genes (HVGs) was then scaled and dimensionality reduction was performed using principal component analysis (PCA). These components were then used to correct batch effects using Harmony v1.2.3 (14). (4) Estimating 1-D UMAP embeddings for Control and Ts21 samples using the batch corrected Harmony components. (5) Split the cells into 20 equally distanced bins along 1-D UMAP embedding dimension and obtain their average gene expressions of the HVGs. (6) Calculate the Shannon entropy for each gene using the binned gene expression. This result was considered as the actual entropy in the calculation. (7) Permutate the gene expression matrix (scaled data) five times and follow steps 2 through 6 without the batch correction step in step 3. The average of 5 permutations was considered as the permutated or unstructured entropy. (8) Take the difference between actual and permutated entropy (from steps 5 and 6, respectively) and get the average of genes that have an entropy difference greater than 0.5. (9) Follow steps 1–7, 100 times, starting with a new subsampled dataset with a different set of HVGs for Control and Ts21 samples. These values can be visualized using violin plots as depicted (fig. S3D). The statistical significance between Ts21 and Control entropy values for each cell subclass was obtained with unpaired Wilcoxon rank sum test using the stat_compare_means() function in ggpubr package v0.6.0.

Differential cell composition analysis

To quantitatively assess statistically significant alterations in cellular composition across distinct hierarchical levels, we employed scCODA(19), a Bayesian framework designed for the analysis of compositional single-cell data, as implemented in the pertpy package (v0.9.4). Specifically, we investigated compositional variations at the (1) cell subclass and (2) cell subtype hierarchical levels. Each analysis incorporated discrete experimental covariates, including genotype (control vs. Ts21), sex (male vs. female), and categorical age groups (≤365 days vs. >365 days). A stable reference cell type was designated for each analysis: endothelial cells (Endo) for the subclass level and InN SST NPY CHODL for the subtype level. Statistical significance of compositional changes was determined using a false discovery rate (FDR) threshold of <0.05. Statistically significant changes in cell population proportions, along with their associated log2 fold changes, are detailed in table S2.

Differential gene expression analysis

Differential gene expression analysis was performed using the NEBULA (22) R package (v1.5.3), which fits gene-wise negative binomial gamma mixed models (NBGMM) to account for both cell-level overdispersion and sample-level variability. For each gene, expression was modeled using a linear formulation with fixed effects for genotype (Ts21 and control), age (in days as a continuous variable), and sex (male and female), along with a random effect to account for sample level variation. We employed the NBGMM-HL (hierarchical likelihood) estimation mode, which provides an accurate estimation of cell-level overdispersion, particularly in large-scale single-cell datasets. Genes with poor model convergence (defined as convergence code ≤ −20) were excluded from downstream analysis to ensure reliability. Differential gene expression analyses were conducted at two hierarchical levels, (1) cell class (excitatory neurons, inhibitory neurons, glial cells, and vascular cells) and (2) cell subclass level (ExN L2–3 IT, InN SST, etc.). P-values from NEBULA were adjusted using the Benjamini-Hochberg (BH) method with the p.adjust() function in R. Genes with an adjusted p-value < 0.01 were considered differentially expressed (table S3). In addition to the genome-wide NEBULA analysis, DEGs for select cell types and cell states (reactive astrocytes) were calculated using the FindAllMarkers() function in Seurat with default parameters (Wilcoxon rank sum test).

Gene Set Enrichment Analysis

Gene set enrichment analysis (GSEA) for ExN L2–3 IT and OPCs was conducted using the Molecular Signatures Database (MSigDB; msigdbr v7.5.1) C5 ontology gene set (Homo sapiens; category: Gene Ontology (15) Biological Processes [GO:BP]) and the fgsea (86) (v1.24.0) R package. Differential gene expression results from NEBULA analysis served as input, with enrichment calculations restricted to genes exhibiting statistically significant (adjusted p-value < 0.01) and highly upregulated (log2FC > 1) or highly downregulated (log2FC < −1) expression. Permutation-based significance testing was performed using 5,000 iterations (nPermSimple = 5000). The top and bottom 50 GO:BP terms, ranked by normalized enrichment score (NES), were visualized to highlight pathways with the strongest enrichment signals.

For comparative analysis of control reactive astrocytes versus Ts21 reactive astrocytes, GSEA was performed using the MSigDB Hallmarks (75) gene set (Homo sapiens; category H) and the fgsea R package. Input genes were filtered to include those with an adjusted p-value < 0.01 and expression detected in >10% of control cells (pct.1 > 0.1), as identified by the Wilcoxon rank sum test (FindAllMarkers()). Permutation-based significance testing was performed using 5,000 iterations (nPermSimple = 5000). Enriched terms were ranked by NES, and the 15 top pathways enriched in control-reactive astrocytes were selected for visualization.

To investigate alterations in synaptic gene expression, we performed Synaptic Gene Ontologies (SynGO(30)) enrichment analysis. For the ExN L2–3 IT population, all genes identified as downregulated (adjusted p < 0.01) from NEBULA analysis were used as the input gene set, with the standard SynGO “brain expressed” gene set serving as the background. For analysis of MGE-derived neurons, we utilized the union of all downregulated genes (adjusted p < 0.01) identified across MGE cell subclasses (InN PVALB, InN SST, InN ETV1, InN HGF, InN LAMP5 LHX6) in NEBULA analysis. Similarly, for CGE-derived neurons, the union of all downregulated genes (adjusted p < 0.01) from CGE cell subclasses (InN CALB2, InN LAMP5, InN SP8) was used. The standard “brain expressed” gene set was employed as the background gene set for both MGE and CGE analyses. For a broadened analysis of synaptic gene dysregulation, we identified sets of upregulated or downregulated genes (adjusted p < 0.01) at the broad ExN or InN major class and performed SynGO enrichment using the standard “brain expressed” background gene set (figs. S9 and S11). Statistical significance of enrichment results was determined by controlling the False Discovery Rate (FDR) using the Benjamini-Hochberg procedure, and Q values reported. Gene set enrichment analyses are summarized in table S6.

Inferred Trajectories of Oligodendrocyte Development

To investigate transcriptional dynamics within the oligodendrocyte lineage, we utilized the Potential of Heat-diffusion for Affinity-based Trajectory Embedding (48) (PHATE v1.0.11) for high-dimensional data visualization and Palantir v1.3.3 (47) to infer pseudotime, approximating the developmental progression of individual cells. Both analyses were performed on the top 50 batch-corrected Harmony (14) components to ensure robust integration across samples. A neighborhood graph was computed using 50 Harmony components with the setting use_rep='X_harmony' and 50 nearest neighbors via scanpy.pp.neighbors function of scanpy (v1.9.8)(87), which served as input for both PHATE and Palantir. First, PHATE embeddings were computed with parameters k=5, a=20, and t=150, providing a low-dimensional representation of the developmental landscape. Next, we applied Palantir (47) to infer pseudotime. Proliferative oligodendrocyte precursor cells (prOPCs) from the earliest time point (postnatal day 60) were selected as root cells to initialize the trajectory. Palantir was run via scanpy.external.tl.palantir, using the precomputed neighborhood graph (use_adjacency_matrix=True) to build diffusion maps and compute pseudotime values.

Inference, analysis, and visualization of differential cell-cell communication

Cell-cell communication networks were inferred, analyzed, and visualized using the CellChat R package (v2.1.2) (27). For each condition, separate CellChat objects were initialized in label-based mode using normalized expression data and pre-annotated cell types, independently for Ts21 and control. To capture both broad and specific patterns of cell-cell communication we performed CellChat using two sets of annotated cell types: (1) major cell classes (ExN, InN, OPC, Oligo, Astro, Micro, Endo, VLMC) and (2) major cell classes with defined microglia subtypes (ExN, InN, OPC, Oligo, Astro, Micro CTSC DOCK1, Micro CD200 PRKAA2, Endo, VLMC).

The CellChatDB v2 human ligand-receptor interaction database, comprising 3,234 literature-supported molecular interactions, was employed for network inference using default parameters. Overexpressed genes and interactions were identified for each condition using the identifyOverExpressedGenes() and identifyOverExpressedInteractions() functions, respectively, with default settings. Communication probabilities and intercellular networks were computed using the computeCommunProb() function, employing a truncated mean (trim = 0.05) to calculate average gene expression per cell group. For integrated analysis, CellChat objects from Ts21 and control were then merged. The differential number of interactions and interaction strengths (edge weights) were visualized using the netVisual_heatmap() function. Changes in signaling within specific cell populations were analyzed with the netAnalysis_signalingChanges_scatter() function. Relative information flow, defined as the sum of total network weights, was quantified using the rankNet() function, with statistical significance assessed by paired Wilcoxon test (p<0.01). Chord and circle plots were generated to visualize signaling networks, with maximum edge weights standardized across control and Ts21 conditions.

snATAC-seq data processing

The Signac (v1.14.0) (88) and Seurat (84) (v5.2.0) packages were used to analyze snATAC-seq data. We began our analysis with the nuclei retained in the snRNA-seq dataset, ensuring that the identities of nuclei in the snATAC-seq dataset were known a priori. Accurate chromatin accessibility analysis depends on precise peak calling, as calling peaks across the entire cell population can inadvertently exclude those specific to smaller subpopulations (85). To address this, the CallPeaks() function in Signac, utilizing the MACS2 method (89), was applied after grouping by cell subclass and genotype. This approach ensured the retention of peaks associated with all cell subpopulations across genotypes. The blacklisted regions curated in the ENCODE project (90) were removed from the identified peak set. Once the consensus peaks were obtained, the peak count matrix was generated using ATAC-seq fragment files. Then, standard quality control metrics such as transcriptional start site (TSS) enrichment, nucleosome signal, and the number of peaks were calculated using the TSSEnrichment(), NucleosomeSignal(), and CountFragments() functions in Signac, respectively. The low-quality nuclei were identified and filtered based on whether the per-cell peak count was less than 1,000 or greater than 100,000, nucleosome signal greater than 4, and TSS enrichment less than 2 (fig. S1, E to H). The retained data were then normalized using the term frequency inverse-document frequency (TF-IDF) method(85), and the latent semantic indexing (LSI) components were calculated. The top 2–30 LSI components were used to remove batch effects using Harmony. Peak calling was performed on both the entire dataset and on a selected pair of samples (HSBPY0.16; 60 days postnatal, HSBDPY0.8; 65 days postnatal) at the earliest study timepoint as an additional quality control step to assess the robustness of the findings.

snATAC peak GREAT analysis

Peak sets identified across all samples, along with the selected sample pair, were independently analyzed using the Genomic Regions Enrichment of Annotations Tool (GREAT) to assess the enrichment of chromatin state annotations. These annotations were derived from a 25-state ChromHMM model on interpolated data from the Epigenome Roadmap Consortium (91, 92). Specifically, annotations for the adult dlPFC (E073) sample with liftOver (93) to the hg38 genome assembly were utilized (https://egg2.wustl.edu/roadmap/web_portal/imputed.html#chr_imp ) for reference. This analysis assessed the per-bp overlap of peak regions with regions associated with each of the 25 annotated chromatin states. Fold enrichment and statistical significance (p-value, calculated using a negative binomial distribution) were determined for each state relative to the combined peak set. This analysis was performed for all combined peaks, as well as for subsets of peaks uniquely identified in control or Ts21 samples. The same approach was applied to the peak set obtained from the selected pair of samples at the earliest study timepoint (HSBPY0.16, HSBDPY0.8) (fig. S4I). Notably, there was marked enrichment for active enhancer and promoter states across all peak sets, while regions annotated as heterochromatin were underrepresented among identified peaks. These results serve as a quality control measure, supporting the validity of the genomic regions identified in the peak analysis.

snATAC differentially accessible region analysis

Differentially accessible regions (DARs) were called for each cell subclass. The set of aggregated counts of combined peaks was compared between Ts21 and control and results with an adjusted p-value of 0.05 or less from the logistic regression as implemented via Seurat (v5.2.1) FindMarkers() method was selected. Three latent variables were applied: 1) number of counts in peaks per barcode, 2) sex of each donor, and 3) age of each donor. The inclusion of sex and age as latent variables was found to reduce the statistics of DARs by ~15.8% (from 30,874 DA regions - in any cell-type - without sex and age as latent variables to 26,010 with sex and age as latent variables), and these variables were consequently taken as important for this analysis. Lastly, the minimum percentage of barcodes with > 0 fragments for a given peak to be considered in the DA analysis was min.pct=0.05, per the Signac recommendations for DA peak analysis.

The general statistics of DARs are summarized by cell type and chromosome (Fig. 1G and Fig. 2B) where the numerator is the number of DARs and the denominator is the number of combined peaks where a 1% fraction of barcodes showed a >0 number of fragments aligned. These results present a striking ~50-fold asymmetry of DARs in favor of peaks gained in the Ts21 karyotype compared to disomic control. The robustness of these findings was tested in multiple ways: the statistics of DARs were compared for adjusted p-value thresholds of 0.05 and 0.01 as well as log2-fold-change thresholds of 0 or 0.25 and not found to impact the striking asymmetry in favor of peaks gained in Ts21 over control. Several DAR-calling methods were tested, including a Poisson generalized linear model approach and Wilcoxon rank sum approach, as implemented in Seurat (v5.2.1). The same latent variables were applied in the Poisson analysis approach as noted above for the logistic regression analysis. The general asymmetry of DARs was well recapitulated in the results of these methods and the overall statistics of DARs were found comparable (Poisson) or greater (Wilcoxon) than those from logistic regression analysis (fig. S4, E to G). The DA peak analysis was repeated using the peak set derived from the earliest sample pair in the study (HSBPY0.16, 60 days postnatal; HSBDPY0.8, 65 days postnatal), applying the same analytical framework as for the full cohort. In this case, the latent variable set was restricted to the number of counts in peaks per barcode, given the single age and sex represented. This analysis demonstrated a comparable asymmetry in DARs and recapitulated the trends observed across all three statistical methods, thereby supporting the robustness and generalizability of the primary findings (fig. S4J).

Genomic region annotations for the peaks called by cell-type were generated for the hg38 genome, using ChIPSeeker (v1.42.1) (94, 95) (TxDb data used as input: DOI: 10.18129/B9.bioc.TxDb.Hsapiens.UCSC.hg38.refGene) annotatePeak() function, and these results are presented for the full analysis of all samples (Fig. 1H). Fisher’s exact test was utilized to interrogate the statistical significance of the association of promoter and distal intergenic regions among Ts21 and control-favoring DARs, respectively. This was done to assess the over-representation of each genomic annotation category among DAR’s in each cell type, relative to defined background sets of regions. Two definitions of background regions were tested: 1) the set all regions called across all cell types and 2) the cell type specific set of regions with detected fragments in at least 1% of barcodes for that cell type. In both cases, annotations were generated for the (inclusive) background set(s) with ChIPseeker::annotatePeaks(), and statistics by annotation category were compiled for DARs relative to all background regions per cell type (table S4). The broad trend of over-representation of promoter regions among Ts21-favoring DARs and distal intergenic regions among control-favoring DARs was found to be significant (p<0.05) across most cell types (table S4).

snATAC Motif finding, enrichments and foot-printing

Motifs were found for the hg38 genome assembly using motifmatchr (v1.26.0) for the general (redundant) set of 1518 JAPASR2024 (96) position weight matrices for human transcription factors. Genome assembly package used DOI: 10.18129/B9.bioc.BSgenome.Hsapiens.UCSC.hg38. Enrichments for these motifs among DARs were obtained using the Signac (v1.14) FindMotifs() workflow. For each cell type, a set of 50k background peak regions was matched and selected for GC content percentage relative to the DARs with MatchRegionStats(), to mitigate GC content bias in the motif enrichment comparison with FindMotifs(). For two cell types, the statistics of peaks/barcodes led to fewer than 50k regions being selected, 3,151 regions for the Endo cell subclass and 30,186 for VLMC, respectively. Here, we utilized the set of DARs more accessible in Ts21 as our query set, due to the small number of DARs more accessible in control, and this being more specific. Motif enrichments were selected by adjusted p-value from the hypergeometric test result from FindMotifs() with a threshold of <=0.05. Genome-wide foot-printing analysis of this motif was performed with Signac (v1.14) Footprint() for all cell types with correction for Tn5 insertion bias.

To further investigate chromosome 21-encoded TFs, motif enrichment was assessed in gene sets of key dysregulated biological pathways in Ts21 ExN L2–3 IT neurons. These sets included upregulated differentially expressed genes associated with metabolism, immune response, and protein translation (Fig. 3C), as well as downregulated synapse-related genes (Fig. 3F) in ExN L2–3 IT. Peaks in ExN L2–3 IT neurons were selected if located within defined distances upstream or downstream of gene bodies within each set, forming test peak sets. Motif enrichment analysis followed the same Signac workflow with 50,000 GC content-matched background peaks and hypergeometric testing. Significant enrichment (adjusted p < 0.05) of ERG, ETS2, and GABPA motifs was observed near translation- and metabolism-associated (fig. S7B). This enrichment pattern was robust across varying genomic distance thresholds (2–20 kbp) around gene bodies. Collectively, these results indicate chromosome 21-encoded TF specific regulatory effects with notable biological specificity in Ts21 gene dysregulation.

Chromatin-accessibility and gene-expression at the chromosome-scale

To visualize the snMultiome data at a chromosomal (Mbp) scale, we employed karyoploteR (97) v1.30.0, summarized in the accompanying Shiny app (Fig. 2, D to F, fig. S6, https://daifengwanglab.shinyapps.io/DS_PFC/ ). To prepare this visualization, snATAC data was first aggregated into counts for uniform 10 kbp bins of the hg38 genome for each cell-type and condition. The uniform bins were generated using bedtools (98) makewindows (v2.31.0). Count matrices for those bins were compiled for 2x21 conditions using the Signac (v1.14) FeatureMatrix() function. For each cell type, the row-sum of counts (across barcodes) was saved, and FPKM normalized based on bin size and the total number of counts on chromosomes 1–22. Per-bin log2-fold-change (LFC) values between Ts21 and control were then calculated from the FPKM values, representing differential accessibility in each cell type.

The gene expression data were simultaneously considered via LFC values from NEBULA differential gene expression analysis per cell type. For each gene, the hg38 transcription start coordinate and chromosome were matched to the accompanying LFC value for each cell type. Control and Ts21 snATAC profiles in FPKM units are shown in the bottom half of each plot (green and purple, respectively). On top, the rolling mean and standard deviation of per-bin snATAC LFC values in 400 kbp windows is shown. To directly visualize trends of gene expression, each gene is plotted against the snATAC LFC as a dot according to the associated transcription start site (TSS) coordinate and LFC. For Fig. 2, specifically, we present the data as combined for the three ExN IT cell types (L2–3 IT, L3–5 IT, and L6 IT), where the snATAC uniform bin count data were first summed across these three cell types before applying the same FPKM and LFC treatment, and the represented gene expression LFC values are the mean of the Nebula LFC values across the three IT cell types.

Chromosome 21 is one of five human acrocentric chromosomes (13, 14, 15, 21, and 22) that exhibit high sequence conservation and are characterized by an abundance of repetitive elements within their short arms (99, 100). A substantial part of chr21 is consequently masked in the hg38 assembly, notable in the chromosome-wide visualization (Fig. 2D and fig. S6A). In this view, chr21 presents distinct genome sequence biology on the short arm compared to the rest of the chromosome.

Weighted gene co-expression networks

We calculated weighted gene co-expression networks for cell types at multiple hierarchical levels (cell class and cell subclass levels) using the high-dimensional weighted gene co-expression network analysis (hdWGCNA) R package (31) (v0.4.05). We first filtered out the genes not expressed in at least 5% of the cells. The retained genes were used to generate co-expression networks. First, metacells were obtained via K-nearest neighbors in cells grouped by the sample ID and genotype. Co-expression networks were then constructed using the lowest soft power threshold that has a scale free topology model fit ≥ 0.8. Then module eigengenes were identified. Then, the differential module co-expression values across genotypes were obtained using the FindDMEs() function. We also conducted gene set enrichment using enrichR (101) v3.4 and obtained the GO cellular components [v2025, https://maayanlab.cloud/Enrichr/#libraries ] corresponding to each gene module. The module assignments, top 10 eigengenes, differential module enrichment across genotypes, and their gene set enrichments were tabulated (table S7).

Multiomic gene regulatory network inference using SCENIC+

We inferred two independent multiomic gene regulatory networks (GRNs) for (1) all non-neuronal cells (OPC, Oligo, Astro, Micro, Endo, and VLMCs) and (2) the combined oligodendrocyte lineage (prOPC, OPC, nfOligo, Oligo) using SCENIC+ (51) (v1.0a1). Since we already processed the snATAC-seq data using Signac and obtained the consensus peaks, we directly used the ATAC peak count matrix generated from Signac instead of peak calling within the SCENIC+ platform. We then performed topic modeling using pycisTopic (51) (v2.0a0) and obtained the optimum number of topics based on log-likelihood metrics. Candidate enhancer regions were then obtained using 1) binarizing topics using Otsu method, 2) the top 3000 regions per topic and 3) differentially accessible peaks obtained across cell subclasses and genotypes with default parameters. We then created a custom cistarget database following the standard functions in SCENIC+ using our consensus peaks. Finally, the SCENIC+ Snakemake pipeline was used with the human hg38 BICAN filtered annotations BICAN pipeline from https://storage.googleapis.com/gcp-public-data--broad-references/hg38/v0/star/v2_7_10a/modified_v43.annotation.gtf ). We only retained direct eRegulons that function either as (1) an activator that opens chromatin of the target regions and induces gene expression of the target gene or (2) a repressor that closes the chromatin peak and represses target gene expression. Metadata from the entire regulatory networks were tabulated in table S8. SCENIC+ outputs were depicted in terms of combined dotplot/heatmap plots, where eRegulon specificities across genotype and cell types are visualized, and in terms of network diagrams using the Cytoscape (102) package. Cell type-specific network visualizations were obtained by retaining cell type specific eRegulons or curating biologically relevant eRegulons. The networks were simplified by isolating the top 10 up/down-regulated target genes for each eRegulon based on their log(FC) across genotypes (obtained using our NEBULA (22) differentially expressed gene analysis).

RNA extraction, cDNA synthesis, and droplet digital polymerase chain reaction (ddPCR)

Frozen specimens (table S1) were cryogenically pulverized using a liquid nitrogen-chilled mortar and pestle (Fisherbrand FB961A/FB961K). Pulverized tissue (10–15 mg) was used for RNA extraction. Total RNA was purified using the RNeasy^® Plus Mini Kit (Qiagen, Cat. No. 74134) according to manufacturer instructions and RNA eluted in 30 μL of RNase-free water. Complementary DNA (cDNA) synthesis was performed using the SuperScript^® VILO kit (Invitrogen, Cat. No. 11754050) in a 20 μL reaction volume using 400 ng of total RNA, following the manufacturer’s protocol. No-template and no-reverse transcription (no-RT) controls were included for each reverse transcription run. cDNA was diluted 1:40 in RNAse-free water for ddPCR experiments.

ddPCR was performed with the QX200 Droplet Digital PCR System (Bio-Rad #1864001) using TaqMan probes: FAM (target genes) and HEX (ACTR3 housekeeping gene). The ddPCR reaction mixture consisted of 10 μL of 2 × ddPCR^™ Supermix for Probes (No dUTP) (BioRad #1863024), 1 μL of each 20 × primers/probes mix, 1–3uL of template cDNA, and nuclease-free H2O up to a final volume of 22 μL. Droplet generation and transfer of emulsified samples to PCR plates was performed according to manufacturer’s instructions (Instruction Manual, QX200^™ Droplet Generator, Bio-Rad). In summary, 20 μL of the reaction mixture was loaded into disposable DG8^™ Cartridges (Bio-Rad ##1864008) for QX200^™Droplet Generator together with 70 μL of droplet generation oil (Bio-Rad #1863005) and placed into the droplet generator (Bio-Rad #1864002). After processing, droplets generated from each sample were transferred to a ddPCR^™ 96-Well Plates (Bio-Rad #12001925). PCR amplification was performed with a C1000 Touch thermal cycler (Bio-Rad, USA) using a thermal profile beginning at 95 °C for 10 min, followed by 40 cycles of 95 °C for 30 s, and 60 °C for 60 s, and ending of 98 °C for 2 min at a ramp rate of 2 °C/s. After PCR, the plate was loaded on the droplet reader (Bio-Rad #1864001) and acquired data were analyzed with QuantaSoft Analysis Pro software (Bio-Rad, USA). Target gene concentrations were normalized to the housekeeping gene (ACTR3) and statistical significance of age-matched Ts21 and control samples was tested using a two-tailed unpaired t-test (GraphPad Prism).

Protein extraction and Western blotting

Frozen specimens (table S1) were pulverized and homogenized using a liquid nitrogen-chilled mortar and pestle (Fisherbrand FB961A/FB961K). The resulting homogenate was incubated with a protein lysis buffer (10 mM Tris, pH 7.4; 150 mM NaCl; 1% Triton X-100; 10% glycerol; 1X cOmplete^™, Mini, EDTA-free Protease Inhibitor Cocktail [MilliPore Sigma #11836170001]) for 1 hour on ice, with intermittent pipette mixing every 15 minutes. Lysates were centrifuged at 15,000 × g for 30 minutes at 4°C, and the soluble supernatant collected. The soluble protein fraction was washed with protein lysis buffer and pelleted by centrifugation. Protein concentrations were quantified using the colorimetric DC protein assay (Bio-Rad #5000112) according to the manufacturer’s instructions, with a standard curve for calibration.

For each sample, 20 μg of total protein was separated on 4–15% Mini-PROTEAN^® SDS-PAGE gels (Bio-Rad 4568085) at 60 V for 30 minutes, followed by 200 V for 35 minutes in 1X Tris/Glycine/SDS buffer (Bio-Rad #1610732). Proteins were subsequently transferred to a 0.45 μm nitrocellulose membrane (Bio-Rad 1620115) and air-dried overnight. Total protein was visualized using the Revert^™ 700 Total Protein Stain for Western Blot Normalization (LICORbio #926–11016), following the manufacturer’s single-color Western blot protocol. Imaging was performed using the Li-Cor Odyssey system. Membranes were then rinsed, blocked in 5% nonfat milk, and incubated with primary antibody against HSPB1 (ProteinTech #18284–1-AP, RRID:AB_2295540) diluted 1:1000 or CACNA1C (Abcam #ab84814, RRID:AB_1860052) diluted 1:500 in 5% nonfat milk. Detection was achieved using a secondary antibody conjugated to a near-infrared dye (IRDye 800CW, LICORbio #926–32211) diluted 1:10,000 in 5% nonfat milk, and visualized with the Li-Cor Odyssey imaging system. Densitometric analysis of total protein (700 nm channel) and HSPB1 or CACNA1C (800 nm channel) was performed using Li-Cor Image Studio software. Target protein intensity was normalized to total protein intensity and statistical significance of age-matched Ts21 and control samples was tested using a two-tailed unpaired t-test (GraphPad Prism).

In situ hybridization (ISH)

For multiplex fluorescent ISH, the RNAscope Multiplex Fluorescent Reagent Kit v2 Assay was used (ACD, Bio-Techne, 323110). The manufacturer’s pretreatment protocol for fixed frozen tissue was carried out with modifications (ACD, Bio-Techne, document no. 323100-USM). Sections were rinsed in PBS, dehydrated in progressive solutions of 50%, 70%, and 100% EtOH, and air-dried for 15 min at room temperature. Antigen unmasking was performed with the Retriever 2100 (Electron Microscopy Services) using buffer A (EMS, 62706–10). Sections were rinsed in PBS and incubated in 3% hydrogen peroxide for 15 min at room temperature. Sections were then washed in PBS and incubated in RNAscope Protease Plus (ACD, Bio-Techne, 32330) for 10 min at 40 °C. RNAscope probes used in this study include: Hs-CHST9-C1 (ACD, Bio-Techne, 1096741-C1), Hs-LAMP5-C2 (ACD, Bio-Techne, 487691-C2), and Hs-RELN-C2 (ACD, Bio-Techne, 413051-C2). RNAscope multiplex fluorescent v2 detection protocol was performed with modifications (ACD, Bio-Techne, document no. 323100-USM). All washes were performed with PBST (1X PBS + 0.3% Triton X-100). Tyramide signal amplification was performed with TSA Plus Fluorescein (Akoya Biosciences, NEL741001KT) or TSA Plus Cy3 (Akoya Biosciences, NEL744001KT) diluted 1:200 in RNAscope Multiplex TSA Buffer. Following the RNAscope detection protocol, sections were washed, treated with Autofluorescence Eliminator Reagent (Millipore, 2160) according to manufacturer instructions, and cover-slipped with Vectashield Plus Antifade Mounting Medium (Vector Laboratories, H-1000).

Immunohistochemistry and confocal microscopy

Fixed dorsolateral prefrontal cortex (dlPFC) tissue blocks (table S1) were embedded in optimal cutting temperature (OCT) compound and sectioned at a thickness of 50μm using a Leica CM1950 cryostat. Tissue sections were rinsed with PBS to remove residual OCT and antigen retrieval performed using the Retriever 2100 system (Electron Microscopy Services) with buffer A (EMS #62706–10), according to the manufacturer’s protocol. Following antigen retrieval, sections were incubated for 30 minutes at room temperature in blocking solution containing 5% (v/v) normal donkey serum (Jackson ImmunoResearch Laboratories) and 0.3% (v/v) Triton X-100 in PBS. Primary antibodies against SCGN (R&D Systems # AF4878, RRID:AB_2269934, 1:500) HLA-DRB1 (Abcam #ab133578, RRID:AB_3678597, 1:100), and/or AIF1 (Abcam #ab5076, RRID:AB_2224402, 1:100) were diluted in blocking solution and tissue sections incubated with gentle agitation for 24 hours at 4°C. Sections were then washed with PBST (PBS containing 0.3% Triton X-100) and incubated with secondary antibodies diluted 1:250 in blocking solution. After three additional washes with PBST (5 minutes each), sections were treated with Autofluorescence Eliminator Reagent (Millipore #2160) following the manufacturer’s instructions and coverslipped with Vectashield Plus Antifade Mounting Medium (Vector Laboratories #H1000). To quantify the number of HLA-DRB1+/AIF1+ microglia, images from five independent regions of the grey matter and five independent regions of the white matter were acquired from each tissue section (n=3 sections per condition) using a Nikon A1 confocal microscope. High resolution (1024 × 1024 pixels, zoom 1.00, pixel size 0.21 μm) z-stack images were acquired with 0.70 μm step size and high and low z-position limits set above and below the section focal plane, respectively, using an oil immersion 60X objective (Plan Apo, numerical aperture 1.4, working distance 130 μm). For each image, identical laser power and detector gain settings were used. The number of AIF1+ and HLA-DRB1+/AIF1+ double positive cells were counted per image, the average HLA-DRB1+/AIF1+ percentage calculated per section, and statistical significance tested using a two-tailed unpaired t-test (GraphPad Prism). To quantify the immunointensity of HLA-DRB1, Z-stack images were projected using “sum of all slices” (ImageJ) and ROIs of AIF1+ microglia cell bodies and processes were drawn per image. The HLA-DRB1 integrated density (sum of all pixel intensities) of each microglia ROI along with one background ROI was recorded and the cell total corrected fluorescence (CTCF) was calculated after background subtraction (CTCF = integrated density of signal – [area of signal ROI * mean intensity of background ROI]). The CTCF was compared between Ts21 and control dlPFC and significance tested using a two-tailed unpaired t-test (GraphPad Prism).

Multiplexed immunoassay

The ProcartaPlex^™ Human Inflammation Panel, 20plex bead-based multiplex immunoassay (Thermo Fisher Scientific, Catalog No. EPX200–12185-901) was used to simultaneously profile inflammatory cytokines and adhesion molecules in human cortical tissue. An expanded cohort of control (n=10) and Ts21 (n=10) frozen frontal cortex samples (table S1) were lysed to obtain soluble protein fractions following the manufacturer’s protocol (Thermo Fisher Scientific, Invitrogen, Publication No. MAN0028348, Revision C). Total protein concentrations were determined using the Bradford Protein Assay (Bio-Rad, #5000201), and lysates were normalized to 3 μg/μL. A subsequent Bradford assay was performed to confirm uniform protein concentrations across all samples following dilution. Assays were performed following the manufacturer’s protocol (ThermoFisher Scientific, Invitrogen, Publication number: MAN0028348, revision C). Briefly, 25 μL of each sample lysate was incubated with magnetic beads conjugated to target-specific capture antibodies, followed by sequential incubation with biotinylated detection antibodies and streptavidin-phycoerythrin (SA-PE). The immunoassay was conducted in the 96-well black microplate included in the kit, and fluorescence intensity was measured using a MAGPIX instrument under the specified acquisition parameters (50 μL volume, minimum bead count of 50). Standard curves were generated for each analyte using the provided standard mixes (A, B, and F), and sample concentrations were interpolated using a five-parameter logistic (5-PL) curve fit within the ThermoFisher ProcartaPlex Analysis App. All samples and standards were analyzed in duplicate, and mean values were used for subsequent analyses. Assay quality was evaluated based on intra-assay coefficients of variation (CV) and adherence of standard values to expected ranges. Analytes with measured concentrations below the lowest detectable standard were excluded from downstream analyses. Statistical comparisons between groups were performed within the ProcartaPlex Analysis App using the Mann–Whitney rank test (M-statistics algorithm), which compares the distribution of concentrations across groups.

Supplementary Material

Table S1

NIHMS2177336-supplement-Table_S1.xlsx^{(19KB, xlsx)}

Table S2

NIHMS2177336-supplement-Table_S2.xlsx^{(26.6KB, xlsx)}

Table S3

NIHMS2177336-supplement-Table_S3.xlsx^{(42.8MB, xlsx)}

Table S4

NIHMS2177336-supplement-Table_S4.xlsx^{(7.7MB, xlsx)}

Table S5

NIHMS2177336-supplement-Table_S5.xlsx^{(1.9MB, xlsx)}

Table S6

NIHMS2177336-supplement-Table_S6.xlsx^{(2MB, xlsx)}

Table S7

NIHMS2177336-supplement-Table_S7.xlsx^{(3.3MB, xlsx)}

Table S8

NIHMS2177336-supplement-Table_S8.xlsx^{(32MB, xlsx)}

Supplementary Materials

NIHMS2177336-supplement-Supplementary_Materials.pdf^{(10.4MB, pdf)}

Supplementary Materials

Figs. S1 to S17

Captions for Tables S1 to S8

Acknowledgements:

We thank H. Thurston, A. Osterman, H. Doll, M. Jandy, N. West, and M. Russo for technical assistance and critical discussion of the data; K. Knobel at the Waisman IDD Model Core for core services. We also thank Dr. de la Torre Ubieta and members of his lab for feedback on the manuscript.

Funding:

National Institutes of Health 1R01HD106197 to AMMS, DW, S-CZ, and AB

National Institutes of Health R01AG067025 to DW

National Institutes of Health RF1MH128695 to DW

Brain Research Foundation BRFSG-2023-11 to AMMS

National Institutes of Health 1F30MH140382-01 to RDR

Medical Scientist Training Program grant T32 GM140935 to RDR

Morse Society Fellowship to RDR

National Institutes of Health P50HD105353 to the Waisman Center

National Institutes of Health P30 EY016665 to the UW-Madison Core Grant for Vision Research

Footnotes

Competing interests: SC-Z is a co-founder of BrainXell, Inc. The other authors declare that they have no competing interests.

Data and materials availability:

The genomic data has been submitted to dbGaP (accession number: phs004098.v1.p1). The data can be interactively visualized at https://daifengwanglab.shinyapps.io/DS_PFC/. The code and data used for generating figures can be accessed at Zenodo: https://zenodo.org/records/17527842. All other data are available in the main paper or supplementary information.

References and Notes

1.Karam SM, Riegel M, Segal SL, Félix TM, Barros AJD, Santos IS, Matijasevich A, Giugliani R, Black M, Genetic causes of intellectual disability in a birth cohort: a population-based study. Am J Med Genet A 167, 1204–1214 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Lott IT, Dierssen M, Cognitive deficits and associated neurological complications in individuals with Down’s syndrome. The Lancet Neurology 9, 623–633 (2010). [DOI] [PubMed] [Google Scholar]
3.Jain PD, Nayak A, Karnad SD, Doctor KN, Gross motor dysfunction and balance impairments in children and adolescents with Down syndrome: a systematic review. Clin Exp Pediatr 65, 142–149 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Martin GE, Klusek J, Estigarribia B, Roberts JE, Language Characteristics of Individuals with Down Syndrome. Top Lang Disord 29, 112–132 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Antonarakis SE, Skotko BG, Rafii MS, Strydom A, Pape SE, Bianchi DW, Sherman SL, Reeves RH, Down syndrome. Nat Rev Dis Primers 6, 9 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Russo ML, Sousa AMM, Bhattacharyya A, Consequences of trisomy 21 for brain development in Down syndrome. Nat Rev Neurosci 25, 740–755 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Zhao X, Bhattacharyya A, Human Models Are Needed for Studying Human Neurodevelopmental Disorders. Am J Hum Genet 103, 829–857 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Saha K, Jaenisch R, Technical challenges in using human induced pluripotent stem cells to model disease. Cell Stem Cell 5, 584–595 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Qian X, Song H, Ming G, Brain organoids: advances, applications and challenges. Development 146, dev166074 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Friedman NP, Robbins TW, The role of prefrontal cortex in cognitive control and executive function. Neuropsychopharmacol. 47, 72–89 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Silbereis JC, Pochareddy S, Zhu Y, Li M, Sestan N, The Cellular and Molecular Landscapes of the Developing Human Central Nervous System. Neuron 89, 248–268 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Ma S, Skarica M, Li Q, Xu C, Risgaard RD, Tebbenkamp ATN, Mato-Blanco X, Kovner R, Krsnik Ž, de Martin X, Luria V, Martí-Pérez X, Liang D, Karger A, Schmidt DK, Gomez-Sanchez Z, Qi C, Gobeske KT, Pochareddy S, Debnath A, Hottman CJ, Spurrier J, Teo L, Boghdadi AG, Homman-Ludiye J, Ely JJ, Daadi EW, Mi D, Daadi M, Marín O, Hof PR, Rasin M-R, Bourne J, Sherwood CC, Santpere G, Girgenti MJ, Strittmatter SM, Sousa AMM, Sestan N, Molecular and cellular evolution of the primate dorsolateral prefrontal cortex. Science 377, eabo7257 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Gao Y, Dong Q, Arachchilage KH, Risgaard RD, Syed M, Sheng J, Schmidt DK, Jin T, Liu S, Sandoval SO, Knaack S, Eckholm MT, Chen RJ, Guo Y, Doherty D, Glass I, Levine JE, Wang D, Chang Q, Zhao X, Sousa AMM, Multimodal analyses reveal genes driving electrophysiological maturation of neurons in the primate prefrontal cortex. Neuron, S0896–6273(25)00308–3 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh P, Raychaudhuri S, Fast, sensitive, and accurate integration of single cell data with Harmony. Nat Methods 16, 1289–1296 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT, Harris MA, Hill DP, Issel-Tarver L, Kasarskis A, Lewis S, Matese JC, Richardson JE, Ringwald M, Rubin GM, Sherlock G, Gene Ontology: tool for the unification of biology. Nat Genet 25, 25–29 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Dong Z, Yan Q, Wang X, The Variation of Transcriptomic Perturbations is Associated with the Development and Progression of Various Diseases. Dis Markers 2022, 2148627 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Cadwell CR, Bhaduri A, Mostajo-Radji MA, Keefe MG, Nowakowski TJ, Development and Arealization of the Cerebral Cortex. Neuron 103, 980–1004 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Gandal MJ, Haney JR, Wamsley B, Yap CX, Parhami S, Emani PS, Chang N, Chen GT, Hoftman GD, de Alba D, Ramaswami G, Hartl CL, Bhattacharya A, Luo C, Jin T, Wang D, Kawaguchi R, Quintero D, Ou J, Wu YE, Parikshak NN, Swarup V, Belgard TG, Gerstein M, Pasaniuc B, Geschwind DH, Broad transcriptomic dysregulation occurs across the cerebral cortex in ASD. Nature 611, 532–539 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Büttner M, Ostner J, Müller CL, Theis FJ, Schubert B, scCODA is a Bayesian model for compositional single-cell data analysis. Nat Commun 12, 6876 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Vuong CK, Weber A, Seong P, Matoba N, Shafie B, Salinda A, Zhang P, Chen Y-J, Nichterwitz S, Younesi S, Qi L, Gandal MJ, Geschwind DH, Stein JL, de la Torre-Ubieta L, A single cell multi-omic analysis identifies molecular and gene-regulatory mechanisms dysregulated in the developing Down Syndrome neocortex. (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Bartesaghi R, Brain circuit pathology in Down syndrome: from neurons to neural networks. Reviews in the Neurosciences 34, 365–423 (2023). [DOI] [PubMed] [Google Scholar]
22.He L, Davila-Velderrain J, Sumida TS, Hafler DA, Kellis M, Kulminski AM, NEBULA is a fast negative binomial mixed model for differential or co-expression analysis of large-scale multi-subject single-cell data. Commun Biol 4, 629 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Tan K-L, Lee H-C, Cheah P-S, Ling K-H, Mitochondrial Dysfunction in Down Syndrome: From Pathology to Therapy. Neuroscience 511, 1–12 (2023). [DOI] [PubMed] [Google Scholar]
24.Meharena HS, Marco A, Dileep V, Lockshin ER, Akatsu GY, Mullahoo J, Watson LA, Ko T, Guerin LN, Abdurob F, Rengarajan S, Papanastasiou M, Jaffe JD, Tsai L-H, Down Syndrome Induced Senescence Disrupts the Nuclear Architecture of Neural Progenitors. Cell Stem Cell 29, 116–130.e7 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Guo Z, Zhu Y, Xiao H, Dai R, Yang W, Xue W, Zhang X, Hao B, Liao S, Integration of ATAC-seq and RNA-seq identifies MX1-mediated AP-1 transcriptional regulation as a therapeutic target for Down syndrome. Biological Research 56, 67 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Marderstein AR, De Zuani M, Moeller R, Bezney J, Padhi EM, Wong S, Coorens THH, Xie Y, Xue H, Montgomery SB, Cvejic A, Single-cell multi-omics map of human fetal blood in Down syndrome. Nature 634, 104–112 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Jin S, Plikus MV, Nie Q, CellChat for systematic analysis of cell-cell communication from single-cell transcriptomics. Nat Protoc 20, 180–219 (2025). [DOI] [PubMed] [Google Scholar]
28.Perluigi M, Tramutola A, Pagnotta S, Barone E, Butterfield DA, The BACH1/Nrf2 Axis in Brain in Down Syndrome and Transition to Alzheimer Disease-Like Neuropathology and Dementia. Antioxidants (Basel) 9, 779 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Palmer CR, Liu CS, Romanow WJ, Lee M-H, Chun J, Altered cell and RNA isoform diversity in aging Down syndrome brains. Proceedings of the National Academy of Sciences 118, e2114326118 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Koopmans F, van Nierop P, Andres-Alonso M, Byrnes A, Cijsouw T, Coba MP, Cornelisse LN, Farrell RJ, Goldschmidt HL, Howrigan DP, Hussain NK, Imig C, de Jong APH, Jung H, Kohansalnodehi M, Kramarz B, Lipstein N, Lovering RC, MacGillavry H, Mariano V, Mi H, Ninov M, Osumi-Sutherland D, Pielot R, Smalla K-H, Tang H, Tashman K, Toonen RFG, Verpelli C, Reig-Viader R, Watanabe K, van Weering J, Achsel T, Ashrafi G, Asi N, Brown TC, De Camilli P, Feuermann M, Foulger RE, Gaudet P, Joglekar A, Kanellopoulos A, Malenka R, Nicoll RA, Pulido C, de Juan-Sanz J, Sheng M, Südhof TC, Tilgner HU, Bagni C, Bayés À, Biederer T, Brose N, Chua JJE, Dieterich DC, Gundelfinger ED, Hoogenraad C, Huganir RL, Jahn R, Kaeser PS, Kim E, Kreutz MR, McPherson PS, Neale BM, O’Connor V, Posthuma D, Ryan TA, Sala C, Feng G, Hyman SE, Thomas PD, Smit AB, Verhage M, SynGO: An Evidence-Based, Expert-Curated Knowledge Base for the Synapse. Neuron 103, 217–234.e4 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Morabito S, Reese F, Rahimzadeh N, Miyoshi E, Swarup V, hdWGCNA identifies co-expression networks in high-dimensional transcriptomics data. Cell Reports Methods 3, 100498 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Hampel H, Hardy J, Blennow K, Chen C, Perry G, Kim SH, Villemagne VL, Aisen P, Vendruscolo M, Iwatsubo T, Masters CL, Cho M, Lannfelt L, Cummings JL, Vergallo A, The Amyloid-β Pathway in Alzheimer’s Disease. Mol Psychiatry 26, 5481–5503 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Mori C, Spooner ET, Wisniewsk KE, Wisniewski TM, Yamaguch H, Saido TC, Tolan DR, Selkoe DJ, Lemere CA, Intraneuronal Abeta42 accumulation in Down syndrome brain. Amyloid 9, 88–102 (2002). [PubMed] [Google Scholar]
34.Gyure KA, Durham R, Stewart WF, Smialek JE, Troncoso JC, Intraneuronal Aβ-Amyloid Precedes Development of Amyloid Plaques in Down Syndrome. Archives of Pathology & Laboratory Medicine 125, 489–492 (2001). [DOI] [PubMed] [Google Scholar]
35.Kumari S, Dhapola R, Reddy DH, Apoptosis in Alzheimer’s disease: insight into the signaling pathways and therapeutic avenues. Apoptosis 28, 943–957 (2023). [DOI] [PubMed] [Google Scholar]
36.Marin-Padilla M, Structural abnormalities of the cerebral cortex in human chromosomal aberrations: a Golgi study. Brain Research 44, 625–629 (1972). [DOI] [PubMed] [Google Scholar]
37.Marin-Padilla M, Pyramidal cell abnormalities in the motor cortex of a child with Down’s syndrome. A Golgi study. J Comp Neurol 167, 63–81 (1976). [DOI] [PubMed] [Google Scholar]
38.Takashima S, Becker LE, Armstrong DL, Chan F, Abnormal neuronal development in the visual cortex of the human fetus and infant with down’s syndrome. A quantitative and qualitative Golgi study. Brain Res 225, 1–21 (1981). [DOI] [PubMed] [Google Scholar]
39.Becker LE, Armstrong DL, Chan F, Dendritic atrophy in children with Down’s syndrome. Ann Neurol 20, 520–526 (1986). [DOI] [PubMed] [Google Scholar]
40.Petit TL, LeBoutillier JC, Alfano DP, Becker LE, Synaptic development in the human fetus: a morphometric analysis of normal and Down’s syndrome neocortex. Exp Neurol 83, 13–23 (1984). [DOI] [PubMed] [Google Scholar]
41.Weitzdoerfer R, Dierssen M, Fountoulakis M, Lubec G, Fetal life in Down syndrome starts with normal neuronal density but impaired dendritic spines and synaptosomal structure. J Neural Transm Suppl, 59–70 (2001). [DOI] [PubMed] [Google Scholar]
42.Contestabile A, Magara S, Cancedda L, The GABAergic Hypothesis for Cognitive Disabilities in Down Syndrome. Front. Cell. Neurosci. 11 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Pérez-Cremades D, Hernández S, Blasco-Ibáñez JM, Crespo C, Nacher J, Varea E, Alteration of inhibitory circuits in the somatosensory cortex of Ts65Dn mice, a model for Down’s syndrome. J Neural Transm (Vienna) 117, 445–455 (2010). [DOI] [PubMed] [Google Scholar]
44.Chakrabarti L, Best TK, Cramer NP, Carney RSE, Isaac JTR, Galdzicki Z, Haydar TF, Olig1 and Olig2 triplication causes developmental brain defects in Down syndrome. Nat Neurosci 13, 927–934 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Hernández-González S, Ballestín R, López-Hidalgo R, Gilabert-Juan J, Blasco-Ibáñez JM, Crespo C, Nácher J, Varea E, Altered distribution of hippocampal interneurons in the murine Down Syndrome model Ts65Dn. Neurochem Res 40, 151–164 (2015). [DOI] [PubMed] [Google Scholar]
46.Olmos-Serrano JL, Kang HJ, Tyler WA, Silbereis JC, Cheng F, Zhu Y, Pletikos M, Jankovic-Rapan L, Cramer NP, Galdzicki Z, Goodliffe J, Peters A, Sethares C, Delalle I, Golden JA, Haydar TF, Sestan N, Down Syndrome Developmental Brain Transcriptome Reveals Defective Oligodendrocyte Differentiation and Myelination. Neuron 89, 1208–1222 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Setty M, Kiseliovas V, Levine J, Gayoso A, Mazutis L, Pe’er D, Characterization of cell fate probabilities in single-cell data with Palantir. Nat Biotechnol 37, 451–460 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Moon KR, van Dijk D, Wang Z, Gigante S, Burkhardt DB, Chen WS, Yim K, van den Elzen A, Hirn MJ, Coifman RR, Ivanova NB, Wolf G, Krishnaswamy S, Visualizing structure and transitions in high-dimensional biological data. Nat Biotechnol 37, 1482–1492 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Quinn PMJ, Moreira PI, Ambrósio AF, Alves CH, PINK1/PARKIN signalling in neurodegeneration and neuroinflammation. Acta Neuropathologica Communications 8, 189 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Dolgacheva LP, Berezhnov AV, Fedotova EI, Zinchenko VP, Abramov AY, Role of DJ-1 in the mechanism of pathogenesis of Parkinson’s disease. J Bioenerg Biomembr 51, 175–188 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Bravo González-Blas C, De Winter S, Hulselmans G, Hecker N, Matetovici I, Christiaens V, Poovathingal S, Wouters J, Aibar S, Aerts S, SCENIC+: single-cell multiomic inference of enhancers and gene regulatory networks. Nat Methods 20, 1355–1367 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Poitelon Y, Kopec AM, Belin S, Myelin Fat Facts: An Overview of Lipids and Fatty Acid Metabolism. Cells 9, 812 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Bujalka H, Koenning M, Jackson S, Perreau VM, Pope B, Hay CM, Mitew S, Hill AF, Lu QR, Wegner M, Srinivasan R, Svaren J, Willingham M, Barres BA, Emery B, MYRF Is a Membrane-Associated Transcription Factor That Autoproteolytically Cleaves to Directly Activate Myelin Genes. PLOS Biology 11, e1001625 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Hornig J, Fröb F, Vogl MR, Hermans-Borgmeyer I, Tamm ER, Wegner M, The Transcription Factors Sox10 and Myrf Define an Essential Regulatory Network Module in Differentiating Oligodendrocytes. PLoS Genet 9, e1003907 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Southwood C, He C, Garbern J, Kamholz J, Arroyo E, Gow A, CNS Myelin Paranodes Require Nkx6–2 Homeoprotein Transcriptional Activity for Normal Structure. J Neurosci 24, 11215–11225 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Ponroy Bally B, Murai KK, Astrocytes in Down Syndrome Across the Lifespan. Front. Cell. Neurosci. 15 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Habib N, McCabe C, Medina S, Varshavsky M, Kitsberg D, Dvir-Szternfeld R, Green G, Dionne D, Nguyen L, Marshall JL, Chen F, Zhang F, Kaplan T, Regev A, Schwartz M, Disease-associated astrocytes in Alzheimer’s disease and aging. Nat Neurosci 23, 701–706 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Escartin C, Galea E, Lakatos A, O’Callaghan JP, Petzold GC, Serrano-Pozo A, Steinhäuser C, Volterra A, Carmignoto G, Agarwal A, Allen NJ, Araque A, Barbeito L, Barzilai A, Bergles DE, Bonvento G, Butt AM, Chen W-T, Cohen-Salmon M, Cunningham C, Deneen B, De Strooper B, Díaz-Castro B, Farina C, Freeman M, Gallo V, Goldman JE, Goldman SA, Götz M, Gutiérrez A, Haydon PG, Heiland DH, Hol EM, Holt MG, Iino M, Kastanenka KV, Kettenmann H, Khakh BS, Koizumi S, Lee CJ, Liddelow SA, MacVicar BA, Magistretti P, Messing A, Mishra A, Molofsky AV, Murai KK, Norris CM, Okada S, Oliet SHR, Oliveira JF, Panatier A, Parpura V, Pekna M, Pekny M, Pellerin L, Perea G, Pérez-Nievas BG, Pfrieger FW, Poskanzer KE, Quintana FJ, Ransohoff RM, Riquelme-Perez M, Robel S, Rose CR, Rothstein JD, Rouach N, Rowitch DH, Semyanov A, Sirko S, Sontheimer H, Swanson RA, Vitorica J, Wanner I-B, Wood LB, Wu J, Zheng B, Zimmer ER, Zorec R, Sofroniew MV, Verkhratsky A, Reactive astrocyte nomenclature, definitions, and future directions. Nat Neurosci 24, 312–325 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Huang Y, Ge M-X, Li Y-H, Li J-L, Yu Q, Xiao F-H, Ao H-S, Yang L-Q, Li J, He Y, Kong Q-P, Longevity-Associated Transcription Factor ATF7 Promotes Healthspan by Suppressing Cellular Senescence and Systematic Inflammation. Aging Dis 14, 1374–1389 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Yang F, Beltran-Lobo P, Sung K, Goldrick C, Croft CL, Nishimura A, Hedges E, Mahiddine F, Troakes C, Golde TE, Perez-Nievas BG, Hanger DP, Noble W, Jimenez-Sanchez M, Reactive astrocytes secrete the chaperone HSPB1 to mediate neuroprotection. Sci Adv 10, eadk9884 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Paolicelli RC, Bolasco G, Pagani F, Maggi L, Scianni M, Panzanelli P, Giustetto M, Ferreira TA, Guiducci E, Dumas L, Ragozzino D, Gross CT, Synaptic pruning by microglia is necessary for normal brain development. Science 333, 1456–1458 (2011). [DOI] [PubMed] [Google Scholar]
62.Sullivan KD, Lewis HC, Hill AA, Pandey A, Jackson LP, Cabral JM, Smith KP, Liggett LA, Gomez EB, Galbraith MD, DeGregori J, Espinosa JM, Trisomy 21 consistently activates the interferon response. eLife 5, e16220 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
63.Galbraith MD, Rachubinski AL, Smith KP, Araya P, Waugh KA, Enriquez-Estrada B, Worek K, Granrath RE, Kinning KT, Paul Eduthan N, Ludwig MP, Hsieh EWY, Sullivan KD, Espinosa JM, Multidimensional definition of the interferonopathy of Down syndrome and its response to JAK inhibition. Sci Adv 9, eadg6218 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
64.Uchino Y, Woodward AM, Mauris J, Peterson K, Verma P, Nilsson UJ, Rajaiya J, Argüeso P, Galectin-3 is an amplifier of the interleukin-1β-mediated inflammatory response in corneal keratinocytes. Immunology 154, 490–499 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Shirakawa K, Endo J, Kataoka M, Katsumata Y, Yoshida N, Yamamoto T, Isobe S, Moriyama H, Goto S, Kitakata H, Hiraide T, Fukuda K, Sano M, IL (Interleukin)-10–STAT3–Galectin-3 Axis Is Essential for Osteopontin-Producing Reparative Macrophage Polarization After Myocardial Infarction. Circulation 138, 2021–2035 (2018). [DOI] [PubMed] [Google Scholar]
66.McEver RP, Selectins: initiators of leucocyte adhesion and signalling at the vascular wall. Cardiovasc Res 107, 331–339 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Stevens B, Allen NJ, Vazquez LE, Howell GR, Christopherson KS, Nouri N, Micheva KD, Mehalow AK, Huberman AD, Stafford B, Sher A, Litke AM, Lambris JD, Smith SJ, John SWM, Barres BA, The classical complement cascade mediates CNS synapse elimination. Cell 131, 1164–1178 (2007). [DOI] [PubMed] [Google Scholar]
68.Krasemann S, Madore C, Cialic R, Baufeld C, Calcagno N, El Fatimy R, Beckers L, O’Loughlin E, Xu Y, Fanek Z, Greco DJ, Smith ST, Tweet G, Humulock Z, Zrzavy T, Conde-Sanroman P, Gacias M, Weng Z, Chen H, Tjon E, Mazaheri F, Hartmann K, Madi A, Ulrich J, Glatzel M, Worthmann A, Heeren J, Budnik B, Lemere C, Ikezu T, Heppner FL, Litvak V, Holtzman DM, Lassmann H, Weiner HL, Ochando J, Haass C, Butovsky O, The TREM2-APOE pathway drives the transcriptional phenotype of dysfunctional microglia in neurodegenerative diseases. Immunity 47, 566–581.e9 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
69.Hopperton KE, Mohammad D, Trépanier MO, Giuliano V, Bazinet RP, Markers of microglia in post-mortem brain samples from patients with Alzheimer’s disease: a systematic review. Mol Psychiatry 23, 177–198 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
70.Csumitta KD, Gotts SJ, Clasen LS, Martin A, Raitano Lee N, Youth with Down syndrome display widespread increased functional connectivity during rest. Sci Rep 12, 9836 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
71.Jacola LM, Byars AW, Hickey F, Vannest J, Holland SK, Schapiro MB, Functional magnetic resonance imaging of story listening in adolescents and young adults with Down syndrome: evidence for atypical neurodevelopment. J Intellect Disabil Res 58, 892–902 (2014). [DOI] [PubMed] [Google Scholar]
72.Anderson JS, Nielsen JA, Ferguson MA, Burback MC, Cox ET, Dai L, Gerig G, Edgin JO, Korenberg JR, Abnormal brain synchrony in Down Syndrome. Neuroimage Clin 2, 703–715 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Vega JN, Hohman TJ, Pryweller JR, Dykens EM, Thornton-Wells TA, Resting-State Functional Connectivity in Individuals with Down Syndrome and Williams Syndrome Compared with Typically Developing Controls. Brain Connect 5, 461–475 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Jacola LM, Byars AW, Chalfonte-Evans M, Schmithorst VJ, Hickey F, Patterson B, Hotze S, Vannest J, Chiu C-Y, Holland SK, Schapiro MB, Functional magnetic resonance imaging of cognitive processing in young adults with Down syndrome. Am J Intellect Dev Disabil 116, 344–359 (2011). [DOI] [PubMed] [Google Scholar]
75.Waugh KA, Minter R, Baxter J, Chi C, Galbraith MD, Tuttle KD, Eduthan NP, Kinning KT, Andrysik Z, Araya P, Dougherty H, Dunn LN, Ludwig M, Schade KA, Tracy D, Smith KP, Granrath RE, Busquet N, Khanal S, Anderson RD, Cox LL, Estrada BE, Rachubinski AL, Lyford HR, Britton EC, Fantauzzo KA, Orlicky DJ, Matsuda JL, Song K, Cox TC, Sullivan KD, Espinosa JM, Triplication of the interferon receptor locus contributes to hallmarks of Down syndrome in a mouse model. Nat Genet 55, 1034–1047 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
76.Escoubas CC, Dorman LC, Nguyen PT, Lagares-Linares C, Nakajo H, Anderson SR, Barron JJ, Wade SD, Cuevas B, Vainchtein ID, Silva NJ, Guajardo R, Xiao Y, Lidsky PV, Wang EY, Rivera BM, Taloma SE, Kim DK, Kaminskaya E, Nakao-Inoue H, Schwer B, Arnold TD, Molofsky AB, Condello C, Andino R, Nowakowski TJ, Molofsky AV, Type-I-interferon-responsive microglia shape cortical development and behavior. Cell 187, 1936–1954.e24 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
77.Larsen FK, Baksh RA, McGlinchey E, Langballe EM, Benejam B, Beresford-Webb J, McCarron M, Coppus A, Falquero S, Fortea J, Levin J, Loosli SV, Mark R, Rebillat A, Zaman S, Strydom A, Age of Alzheimer’s disease diagnosis in people with Down syndrome and associated factors: Results from the Horizon 21 European Down syndrome consortium. Alzheimers Dement 20, 3270–3280 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
78.Iulita MF, Garzón Chavez D, Klitgaard Christensen M, Valle Tamayo N, Plana-Ripoll O, Rasmussen SA, Roqué Figuls M, Alcolea D, Videla L, Barroeta I, Benejam B, Altuna M, Padilla C, Pegueroles J, Fernandez S, Belbin O, Carmona-Iragui M, Blesa R, Lleó A, Bejanin A, Fortea J, Association of Alzheimer Disease With Life Expectancy in People With Down Syndrome. JAMA Network Open 5, e2212910 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
79.DeTure MA, Dickson DW, The neuropathological diagnosis of Alzheimer’s disease. Molecular Neurodegeneration 14, 32 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
80.Wilson DM, Cookson MR, Van Den Bosch L, Zetterberg H, Holtzman DM, Dewachter I, Hallmarks of neurodegenerative diseases. Cell 186, 693–714 (2023). [DOI] [PubMed] [Google Scholar]
81.Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P, The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 1, 417–425 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
82.Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR, STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
83.Martin M, Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 17, 10–12 (2011). [Google Scholar]
84.Li H, Durbin R, Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
85.Young MD, Behjati S, SoupX removes ambient RNA contamination from droplet-based single-cell RNA sequencing data. Gigascience 9, giaa151 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
86.Mathys H, Peng Z, Boix CA, Victor MB, Leary N, Babu S, Abdelhady G, Jiang X, Ng AP, Ghafari K, Kunisky AK, Mantero J, Galani K, Lohia VN, Fortier GE, Lotfi Y, Ivey J, Brown HP, Patel PR, Chakraborty N, Beaudway JI, Imhoff EJ, Keeler CF, McChesney MM, Patel HH, Patel SP, Thai MT, Bennett DA, Kellis M, Tsai L-H, Single-cell atlas reveals correlates of high cognitive function, dementia, and resilience to Alzheimer’s disease pathology. Cell 186, 4365–4385.e27 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
87.McGinnis CS, Murrow LM, Gartner ZJ, DoubletFinder: Doublet Detection in Single-Cell RNA Sequencing Data Using Artificial Nearest Neighbors. Cell Syst 8, 329–337.e4 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
88.Wolock SL, Lopez R, Klein AM, Scrublet: Computational Identification of Cell Doublets in Single-Cell Transcriptomic Data. Cell Syst 8, 281–291.e9 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
89.Germain P-L, Lun A, Garcia Meixide C, Macnair W, Robinson MD, Doublet identification in single-cell sequencing data using scDblFinder. F1000Res 10, 979 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
90.Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, Molla G, Madad S, Fernandez-Granda C, Satija R, Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol 42, 293–304 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
91.Stuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck WM, Hao Y, Stoeckius M, Smibert P, Satija R, Comprehensive Integration of Single-Cell Data. Cell 177, 1888–1902.e21 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
92.Korotkevich G, Sukhov V, Budin N, Shpak B, Artyomov MN, Sergushichev A, Fast gene set enrichment analysis. bioRxiv [Preprint] (2021). 10.1101/060012. [DOI] [Google Scholar]
93.Wolf FA, Angerer P, Theis FJ, SCANPY: large-scale single-cell gene expression data analysis. Genome Biol 19, 15 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
94.Stuart T, Srivastava A, Madad S, Lareau CA, Satija R, Single-cell chromatin state analysis with Signac. Nat Methods 18, 1333–1341 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
95.Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, Nusbaum C, Myers RM, Brown M, Li W, Liu XS, Model-based analysis of ChIP-Seq (MACS). Genome Biol 9, R137 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
96.Amemiya HM, Kundaje A, Boyle AP, The ENCODE Blacklist: Identification of Problematic Regions of the Genome. Sci Rep 9, 9354 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
97.Roadmap Epigenomics Consortium, Kundaje A, Meuleman W, Ernst J, Bilenky M, Yen A, Heravi-Moussavi A, Kheradpour P, Zhang Z, Wang J, Ziller MJ, Amin V, Whitaker JW, Schultz MD, Ward LD, Sarkar A, Quon G, Sandstrom RS, Eaton ML, Wu Y-C, Pfenning AR, Wang X, Claussnitzer M, Liu Y, Coarfa C, Harris RA, Shoresh N, Epstein CB, Gjoneska E, Leung D, Xie W, Hawkins RD, Lister R, Hong C, Gascard P, Mungall AJ, Moore R, Chuah E, Tam A, Canfield TK, Hansen RS, Kaul R, Sabo PJ, Bansal MS, Carles A, Dixon JR, Farh K-H, Feizi S, Karlic R, Kim A-R, Kulkarni A, Li D, Lowdon R, Elliott G, Mercer TR, Neph SJ, Onuchic V, Polak P, Rajagopal N, Ray P, Sallari RC, Siebenthall KT, Sinnott-Armstrong NA, Stevens M, Thurman RE, Wu J, Zhang B, Zhou X, Beaudet AE, Boyer LA, De Jager PL, Farnham PJ, Fisher SJ, Haussler D, Jones SJM, Li W, Marra MA, McManus MT, Sunyaev S, Thomson JA, Tlsty TD, Tsai L-H, Wang W, Waterland RA, Zhang MQ, Chadwick LH, Bernstein BE, Costello JF, Ecker JR, Hirst M, Meissner A, Milosavljevic A, Ren B, Stamatoyannopoulos JA, Wang T, Kellis M, Integrative analysis of 111 reference human epigenomes. Nature 518, 317–330 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
98.Ernst J, Kellis M, Chromatin-state discovery and genome annotation with ChromHMM. Nat Protoc 12, 2478–2492 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
99.Hinrichs AS, Karolchik D, Baertsch R, Barber GP, Bejerano G, Clawson H, Diekhans M, Furey TS, Harte RA, Hsu F, Hillman-Jackson J, Kuhn RM, Pedersen JS, Pohl A, Raney BJ, Rosenbloom KR, Siepel A, Smith KE, Sugnet CW, Sultan-Qurraie A, Thomas DJ, Trumbower H, Weber RJ, Weirauch M, Zweig AS, Haussler D, Kent WJ, The UCSC Genome Browser Database: update 2006. Nucleic Acids Research 34, D590–D598 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
100.Wang Q, Li M, Wu T, Zhan L, Li L, Chen M, Xie W, Xie Z, Hu E, Xu S, Yu G, Exploring Epigenomic Datasets by ChIPseeker. Curr Protoc 2, e585 (2022). [DOI] [PubMed] [Google Scholar]
101.Yu G, Wang L-G, He Q-Y, ChIPseeker: an R/Bioconductor package for ChIP peak annotation, comparison and visualization. Bioinformatics 31, 2382–2383 (2015). [DOI] [PubMed] [Google Scholar]
102.Rauluseviciute I, Riudavets-Puig R, Blanc-Mathieu R, Castro-Mondragon JA, Ferenc K, Kumar V, Lemma RB, Lucas J, Chèneby J, Baranasic D, Khan A, Fornes O, Gundersen S, Johansen M, Hovig E, Lenhard B, Sandelin A, Wasserman WW, Parcy F, Mathelier A, JASPAR 2024: 20th anniversary of the open-access database of transcription factor binding profiles. Nucleic Acids Res 52, D174–D182 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
103.Gel B, Serra E, karyoploteR: an R/Bioconductor package to plot customizable genomes displaying arbitrary data. Bioinformatics 33, 3088–3090 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
104.Quinlan AR, Hall IM, BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
105.Nurk S, Koren S, Rhie A, Rautiainen M, Bzikadze AV, Mikheenko A, Vollger MR, Altemose N, Uralsky L, Gershman A, Aganezov S, Hoyt SJ, Diekhans M, Logsdon GA, Alonge M, Antonarakis SE, Borchers M, Bouffard GG, Brooks SY, Caldas GV, Chen N-C, Cheng H, Chin C-S, Chow W, de Lima LG, Dishuck PC, Durbin R, Dvorkina T, Fiddes IT, Formenti G, Fulton RS, Fungtammasan A, Garrison E, Grady PGS, Graves-Lindsay TA, Hall IM, Hansen NF, Hartley GA, Haukness M, Howe K, Hunkapiller MW, Jain C, Jain M, Jarvis ED, Kerpedjiev P, Kirsche M, Kolmogorov M, Korlach J, Kremitzki M, Li H, Maduro VV, Marschall T, McCartney AM, McDaniel J, Miller DE, Mullikin JC, Myers EW, Olson ND, Paten B, Peluso P, Pevzner PA, Porubsky D, Potapova T, Rogaev EI, Rosenfeld JA, Salzberg SL, Schneider VA, Sedlazeck FJ, Shafin K, Shew CJ, Shumate A, Sims Y, Smit AFA, Soto DC, Sović I, Storer JM, Streets A, Sullivan BA, Thibaud-Nissen F, Torrance J, Wagner J, Walenz BP, Wenger A, Wood JMD, Xiao C, Yan SM, Young AC, Zarate S, Surti U, McCoy RC, Dennis MY, Alexandrov IA, Gerton JL, O’Neill RJ, Timp W, Zook JM, Schatz MC, Eichler EE, Miga KH, Phillippy AM, The complete sequence of a human genome. Science 376, 44–53 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
106.Guarracino A, Buonaiuto S, de Lima LG, Potapova T, Rhie A, Koren S, Rubinstein B, Fischer C, Gerton JL, Phillippy AM, Colonna V, Garrison E, Recombination between heterologous human acrocentric chromosomes. Nature 617, 335–343 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
107.Kuleshov MV, Jones MR, Rouillard AD, Fernandez NF, Duan Q, Wang Z, Koplev S, Jenkins SL, Jagodnik KM, Lachmann A, McDermott MG, Monteiro CD, Gundersen GW, Ma’ayan A, Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res 44, W90–97 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
108.Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B, Ideker T, Cytoscape: A Software Environment for Integrated Models of Biomolecular Interaction Networks. Genome Res 13, 2498–2504 (2003). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1

NIHMS2177336-supplement-Table_S1.xlsx^{(19KB, xlsx)}

Table S2

NIHMS2177336-supplement-Table_S2.xlsx^{(26.6KB, xlsx)}

Table S3

NIHMS2177336-supplement-Table_S3.xlsx^{(42.8MB, xlsx)}

Table S4

NIHMS2177336-supplement-Table_S4.xlsx^{(7.7MB, xlsx)}

Table S5

NIHMS2177336-supplement-Table_S5.xlsx^{(1.9MB, xlsx)}

Table S6

NIHMS2177336-supplement-Table_S6.xlsx^{(2MB, xlsx)}

Table S7

NIHMS2177336-supplement-Table_S7.xlsx^{(3.3MB, xlsx)}

Table S8

NIHMS2177336-supplement-Table_S8.xlsx^{(32MB, xlsx)}

Supplementary Materials

NIHMS2177336-supplement-Supplementary_Materials.pdf^{(10.4MB, pdf)}

Data Availability Statement

[R1] 1.Karam SM, Riegel M, Segal SL, Félix TM, Barros AJD, Santos IS, Matijasevich A, Giugliani R, Black M, Genetic causes of intellectual disability in a birth cohort: a population-based study. Am J Med Genet A 167, 1204–1214 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R2] 2.Lott IT, Dierssen M, Cognitive deficits and associated neurological complications in individuals with Down’s syndrome. The Lancet Neurology 9, 623–633 (2010). [DOI] [PubMed] [Google Scholar]

[R3] 3.Jain PD, Nayak A, Karnad SD, Doctor KN, Gross motor dysfunction and balance impairments in children and adolescents with Down syndrome: a systematic review. Clin Exp Pediatr 65, 142–149 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] 4.Martin GE, Klusek J, Estigarribia B, Roberts JE, Language Characteristics of Individuals with Down Syndrome. Top Lang Disord 29, 112–132 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] 5.Antonarakis SE, Skotko BG, Rafii MS, Strydom A, Pape SE, Bianchi DW, Sherman SL, Reeves RH, Down syndrome. Nat Rev Dis Primers 6, 9 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] 6.Russo ML, Sousa AMM, Bhattacharyya A, Consequences of trisomy 21 for brain development in Down syndrome. Nat Rev Neurosci 25, 740–755 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] 7.Zhao X, Bhattacharyya A, Human Models Are Needed for Studying Human Neurodevelopmental Disorders. Am J Hum Genet 103, 829–857 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] 8.Saha K, Jaenisch R, Technical challenges in using human induced pluripotent stem cells to model disease. Cell Stem Cell 5, 584–595 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.Qian X, Song H, Ming G, Brain organoids: advances, applications and challenges. Development 146, dev166074 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] 10.Friedman NP, Robbins TW, The role of prefrontal cortex in cognitive control and executive function. Neuropsychopharmacol. 47, 72–89 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R11] 11.Silbereis JC, Pochareddy S, Zhu Y, Li M, Sestan N, The Cellular and Molecular Landscapes of the Developing Human Central Nervous System. Neuron 89, 248–268 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Ma S, Skarica M, Li Q, Xu C, Risgaard RD, Tebbenkamp ATN, Mato-Blanco X, Kovner R, Krsnik Ž, de Martin X, Luria V, Martí-Pérez X, Liang D, Karger A, Schmidt DK, Gomez-Sanchez Z, Qi C, Gobeske KT, Pochareddy S, Debnath A, Hottman CJ, Spurrier J, Teo L, Boghdadi AG, Homman-Ludiye J, Ely JJ, Daadi EW, Mi D, Daadi M, Marín O, Hof PR, Rasin M-R, Bourne J, Sherwood CC, Santpere G, Girgenti MJ, Strittmatter SM, Sousa AMM, Sestan N, Molecular and cellular evolution of the primate dorsolateral prefrontal cortex. Science 377, eabo7257 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Gao Y, Dong Q, Arachchilage KH, Risgaard RD, Syed M, Sheng J, Schmidt DK, Jin T, Liu S, Sandoval SO, Knaack S, Eckholm MT, Chen RJ, Guo Y, Doherty D, Glass I, Levine JE, Wang D, Chang Q, Zhao X, Sousa AMM, Multimodal analyses reveal genes driving electrophysiological maturation of neurons in the primate prefrontal cortex. Neuron, S0896–6273(25)00308–3 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] 14.Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh P, Raychaudhuri S, Fast, sensitive, and accurate integration of single cell data with Harmony. Nat Methods 16, 1289–1296 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] 15.Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, Eppig JT, Harris MA, Hill DP, Issel-Tarver L, Kasarskis A, Lewis S, Matese JC, Richardson JE, Ringwald M, Rubin GM, Sherlock G, Gene Ontology: tool for the unification of biology. Nat Genet 25, 25–29 (2000). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] 16.Dong Z, Yan Q, Wang X, The Variation of Transcriptomic Perturbations is Associated with the Development and Progression of Various Diseases. Dis Markers 2022, 2148627 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R17] 17.Cadwell CR, Bhaduri A, Mostajo-Radji MA, Keefe MG, Nowakowski TJ, Development and Arealization of the Cerebral Cortex. Neuron 103, 980–1004 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Gandal MJ, Haney JR, Wamsley B, Yap CX, Parhami S, Emani PS, Chang N, Chen GT, Hoftman GD, de Alba D, Ramaswami G, Hartl CL, Bhattacharya A, Luo C, Jin T, Wang D, Kawaguchi R, Quintero D, Ou J, Wu YE, Parikshak NN, Swarup V, Belgard TG, Gerstein M, Pasaniuc B, Geschwind DH, Broad transcriptomic dysregulation occurs across the cerebral cortex in ASD. Nature 611, 532–539 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Büttner M, Ostner J, Müller CL, Theis FJ, Schubert B, scCODA is a Bayesian model for compositional single-cell data analysis. Nat Commun 12, 6876 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Vuong CK, Weber A, Seong P, Matoba N, Shafie B, Salinda A, Zhang P, Chen Y-J, Nichterwitz S, Younesi S, Qi L, Gandal MJ, Geschwind DH, Stein JL, de la Torre-Ubieta L, A single cell multi-omic analysis identifies molecular and gene-regulatory mechanisms dysregulated in the developing Down Syndrome neocortex. (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R21] 21.Bartesaghi R, Brain circuit pathology in Down syndrome: from neurons to neural networks. Reviews in the Neurosciences 34, 365–423 (2023). [DOI] [PubMed] [Google Scholar]

[R22] 22.He L, Davila-Velderrain J, Sumida TS, Hafler DA, Kellis M, Kulminski AM, NEBULA is a fast negative binomial mixed model for differential or co-expression analysis of large-scale multi-subject single-cell data. Commun Biol 4, 629 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] 23.Tan K-L, Lee H-C, Cheah P-S, Ling K-H, Mitochondrial Dysfunction in Down Syndrome: From Pathology to Therapy. Neuroscience 511, 1–12 (2023). [DOI] [PubMed] [Google Scholar]

[R24] 24.Meharena HS, Marco A, Dileep V, Lockshin ER, Akatsu GY, Mullahoo J, Watson LA, Ko T, Guerin LN, Abdurob F, Rengarajan S, Papanastasiou M, Jaffe JD, Tsai L-H, Down Syndrome Induced Senescence Disrupts the Nuclear Architecture of Neural Progenitors. Cell Stem Cell 29, 116–130.e7 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] 25.Guo Z, Zhu Y, Xiao H, Dai R, Yang W, Xue W, Zhang X, Hao B, Liao S, Integration of ATAC-seq and RNA-seq identifies MX1-mediated AP-1 transcriptional regulation as a therapeutic target for Down syndrome. Biological Research 56, 67 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] 26.Marderstein AR, De Zuani M, Moeller R, Bezney J, Padhi EM, Wong S, Coorens THH, Xie Y, Xue H, Montgomery SB, Cvejic A, Single-cell multi-omics map of human fetal blood in Down syndrome. Nature 634, 104–112 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R27] 27.Jin S, Plikus MV, Nie Q, CellChat for systematic analysis of cell-cell communication from single-cell transcriptomics. Nat Protoc 20, 180–219 (2025). [DOI] [PubMed] [Google Scholar]

[R28] 28.Perluigi M, Tramutola A, Pagnotta S, Barone E, Butterfield DA, The BACH1/Nrf2 Axis in Brain in Down Syndrome and Transition to Alzheimer Disease-Like Neuropathology and Dementia. Antioxidants (Basel) 9, 779 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Palmer CR, Liu CS, Romanow WJ, Lee M-H, Chun J, Altered cell and RNA isoform diversity in aging Down syndrome brains. Proceedings of the National Academy of Sciences 118, e2114326118 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] 30.Koopmans F, van Nierop P, Andres-Alonso M, Byrnes A, Cijsouw T, Coba MP, Cornelisse LN, Farrell RJ, Goldschmidt HL, Howrigan DP, Hussain NK, Imig C, de Jong APH, Jung H, Kohansalnodehi M, Kramarz B, Lipstein N, Lovering RC, MacGillavry H, Mariano V, Mi H, Ninov M, Osumi-Sutherland D, Pielot R, Smalla K-H, Tang H, Tashman K, Toonen RFG, Verpelli C, Reig-Viader R, Watanabe K, van Weering J, Achsel T, Ashrafi G, Asi N, Brown TC, De Camilli P, Feuermann M, Foulger RE, Gaudet P, Joglekar A, Kanellopoulos A, Malenka R, Nicoll RA, Pulido C, de Juan-Sanz J, Sheng M, Südhof TC, Tilgner HU, Bagni C, Bayés À, Biederer T, Brose N, Chua JJE, Dieterich DC, Gundelfinger ED, Hoogenraad C, Huganir RL, Jahn R, Kaeser PS, Kim E, Kreutz MR, McPherson PS, Neale BM, O’Connor V, Posthuma D, Ryan TA, Sala C, Feng G, Hyman SE, Thomas PD, Smit AB, Verhage M, SynGO: An Evidence-Based, Expert-Curated Knowledge Base for the Synapse. Neuron 103, 217–234.e4 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.Morabito S, Reese F, Rahimzadeh N, Miyoshi E, Swarup V, hdWGCNA identifies co-expression networks in high-dimensional transcriptomics data. Cell Reports Methods 3, 100498 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] 32.Hampel H, Hardy J, Blennow K, Chen C, Perry G, Kim SH, Villemagne VL, Aisen P, Vendruscolo M, Iwatsubo T, Masters CL, Cho M, Lannfelt L, Cummings JL, Vergallo A, The Amyloid-β Pathway in Alzheimer’s Disease. Mol Psychiatry 26, 5481–5503 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R33] 33.Mori C, Spooner ET, Wisniewsk KE, Wisniewski TM, Yamaguch H, Saido TC, Tolan DR, Selkoe DJ, Lemere CA, Intraneuronal Abeta42 accumulation in Down syndrome brain. Amyloid 9, 88–102 (2002). [PubMed] [Google Scholar]

[R34] 34.Gyure KA, Durham R, Stewart WF, Smialek JE, Troncoso JC, Intraneuronal Aβ-Amyloid Precedes Development of Amyloid Plaques in Down Syndrome. Archives of Pathology & Laboratory Medicine 125, 489–492 (2001). [DOI] [PubMed] [Google Scholar]

[R35] 35.Kumari S, Dhapola R, Reddy DH, Apoptosis in Alzheimer’s disease: insight into the signaling pathways and therapeutic avenues. Apoptosis 28, 943–957 (2023). [DOI] [PubMed] [Google Scholar]

[R36] 36.Marin-Padilla M, Structural abnormalities of the cerebral cortex in human chromosomal aberrations: a Golgi study. Brain Research 44, 625–629 (1972). [DOI] [PubMed] [Google Scholar]

[R37] 37.Marin-Padilla M, Pyramidal cell abnormalities in the motor cortex of a child with Down’s syndrome. A Golgi study. J Comp Neurol 167, 63–81 (1976). [DOI] [PubMed] [Google Scholar]

[R38] 38.Takashima S, Becker LE, Armstrong DL, Chan F, Abnormal neuronal development in the visual cortex of the human fetus and infant with down’s syndrome. A quantitative and qualitative Golgi study. Brain Res 225, 1–21 (1981). [DOI] [PubMed] [Google Scholar]

[R39] 39.Becker LE, Armstrong DL, Chan F, Dendritic atrophy in children with Down’s syndrome. Ann Neurol 20, 520–526 (1986). [DOI] [PubMed] [Google Scholar]

[R40] 40.Petit TL, LeBoutillier JC, Alfano DP, Becker LE, Synaptic development in the human fetus: a morphometric analysis of normal and Down’s syndrome neocortex. Exp Neurol 83, 13–23 (1984). [DOI] [PubMed] [Google Scholar]

[R41] 41.Weitzdoerfer R, Dierssen M, Fountoulakis M, Lubec G, Fetal life in Down syndrome starts with normal neuronal density but impaired dendritic spines and synaptosomal structure. J Neural Transm Suppl, 59–70 (2001). [DOI] [PubMed] [Google Scholar]

[R42] 42.Contestabile A, Magara S, Cancedda L, The GABAergic Hypothesis for Cognitive Disabilities in Down Syndrome. Front. Cell. Neurosci. 11 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R43] 43.Pérez-Cremades D, Hernández S, Blasco-Ibáñez JM, Crespo C, Nacher J, Varea E, Alteration of inhibitory circuits in the somatosensory cortex of Ts65Dn mice, a model for Down’s syndrome. J Neural Transm (Vienna) 117, 445–455 (2010). [DOI] [PubMed] [Google Scholar]

[R44] 44.Chakrabarti L, Best TK, Cramer NP, Carney RSE, Isaac JTR, Galdzicki Z, Haydar TF, Olig1 and Olig2 triplication causes developmental brain defects in Down syndrome. Nat Neurosci 13, 927–934 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] 45.Hernández-González S, Ballestín R, López-Hidalgo R, Gilabert-Juan J, Blasco-Ibáñez JM, Crespo C, Nácher J, Varea E, Altered distribution of hippocampal interneurons in the murine Down Syndrome model Ts65Dn. Neurochem Res 40, 151–164 (2015). [DOI] [PubMed] [Google Scholar]

[R46] 46.Olmos-Serrano JL, Kang HJ, Tyler WA, Silbereis JC, Cheng F, Zhu Y, Pletikos M, Jankovic-Rapan L, Cramer NP, Galdzicki Z, Goodliffe J, Peters A, Sethares C, Delalle I, Golden JA, Haydar TF, Sestan N, Down Syndrome Developmental Brain Transcriptome Reveals Defective Oligodendrocyte Differentiation and Myelination. Neuron 89, 1208–1222 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] 47.Setty M, Kiseliovas V, Levine J, Gayoso A, Mazutis L, Pe’er D, Characterization of cell fate probabilities in single-cell data with Palantir. Nat Biotechnol 37, 451–460 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] 48.Moon KR, van Dijk D, Wang Z, Gigante S, Burkhardt DB, Chen WS, Yim K, van den Elzen A, Hirn MJ, Coifman RR, Ivanova NB, Wolf G, Krishnaswamy S, Visualizing structure and transitions in high-dimensional biological data. Nat Biotechnol 37, 1482–1492 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R49] 49.Quinn PMJ, Moreira PI, Ambrósio AF, Alves CH, PINK1/PARKIN signalling in neurodegeneration and neuroinflammation. Acta Neuropathologica Communications 8, 189 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] 50.Dolgacheva LP, Berezhnov AV, Fedotova EI, Zinchenko VP, Abramov AY, Role of DJ-1 in the mechanism of pathogenesis of Parkinson’s disease. J Bioenerg Biomembr 51, 175–188 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] 51.Bravo González-Blas C, De Winter S, Hulselmans G, Hecker N, Matetovici I, Christiaens V, Poovathingal S, Wouters J, Aibar S, Aerts S, SCENIC+: single-cell multiomic inference of enhancers and gene regulatory networks. Nat Methods 20, 1355–1367 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R52] 52.Poitelon Y, Kopec AM, Belin S, Myelin Fat Facts: An Overview of Lipids and Fatty Acid Metabolism. Cells 9, 812 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R53] 53.Bujalka H, Koenning M, Jackson S, Perreau VM, Pope B, Hay CM, Mitew S, Hill AF, Lu QR, Wegner M, Srinivasan R, Svaren J, Willingham M, Barres BA, Emery B, MYRF Is a Membrane-Associated Transcription Factor That Autoproteolytically Cleaves to Directly Activate Myelin Genes. PLOS Biology 11, e1001625 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R54] 54.Hornig J, Fröb F, Vogl MR, Hermans-Borgmeyer I, Tamm ER, Wegner M, The Transcription Factors Sox10 and Myrf Define an Essential Regulatory Network Module in Differentiating Oligodendrocytes. PLoS Genet 9, e1003907 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R55] 55.Southwood C, He C, Garbern J, Kamholz J, Arroyo E, Gow A, CNS Myelin Paranodes Require Nkx6–2 Homeoprotein Transcriptional Activity for Normal Structure. J Neurosci 24, 11215–11225 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R56] 56.Ponroy Bally B, Murai KK, Astrocytes in Down Syndrome Across the Lifespan. Front. Cell. Neurosci. 15 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R57] 57.Habib N, McCabe C, Medina S, Varshavsky M, Kitsberg D, Dvir-Szternfeld R, Green G, Dionne D, Nguyen L, Marshall JL, Chen F, Zhang F, Kaplan T, Regev A, Schwartz M, Disease-associated astrocytes in Alzheimer’s disease and aging. Nat Neurosci 23, 701–706 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R58] 58.Escartin C, Galea E, Lakatos A, O’Callaghan JP, Petzold GC, Serrano-Pozo A, Steinhäuser C, Volterra A, Carmignoto G, Agarwal A, Allen NJ, Araque A, Barbeito L, Barzilai A, Bergles DE, Bonvento G, Butt AM, Chen W-T, Cohen-Salmon M, Cunningham C, Deneen B, De Strooper B, Díaz-Castro B, Farina C, Freeman M, Gallo V, Goldman JE, Goldman SA, Götz M, Gutiérrez A, Haydon PG, Heiland DH, Hol EM, Holt MG, Iino M, Kastanenka KV, Kettenmann H, Khakh BS, Koizumi S, Lee CJ, Liddelow SA, MacVicar BA, Magistretti P, Messing A, Mishra A, Molofsky AV, Murai KK, Norris CM, Okada S, Oliet SHR, Oliveira JF, Panatier A, Parpura V, Pekna M, Pekny M, Pellerin L, Perea G, Pérez-Nievas BG, Pfrieger FW, Poskanzer KE, Quintana FJ, Ransohoff RM, Riquelme-Perez M, Robel S, Rose CR, Rothstein JD, Rouach N, Rowitch DH, Semyanov A, Sirko S, Sontheimer H, Swanson RA, Vitorica J, Wanner I-B, Wood LB, Wu J, Zheng B, Zimmer ER, Zorec R, Sofroniew MV, Verkhratsky A, Reactive astrocyte nomenclature, definitions, and future directions. Nat Neurosci 24, 312–325 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R59] 59.Huang Y, Ge M-X, Li Y-H, Li J-L, Yu Q, Xiao F-H, Ao H-S, Yang L-Q, Li J, He Y, Kong Q-P, Longevity-Associated Transcription Factor ATF7 Promotes Healthspan by Suppressing Cellular Senescence and Systematic Inflammation. Aging Dis 14, 1374–1389 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R60] 60.Yang F, Beltran-Lobo P, Sung K, Goldrick C, Croft CL, Nishimura A, Hedges E, Mahiddine F, Troakes C, Golde TE, Perez-Nievas BG, Hanger DP, Noble W, Jimenez-Sanchez M, Reactive astrocytes secrete the chaperone HSPB1 to mediate neuroprotection. Sci Adv 10, eadk9884 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R61] 61.Paolicelli RC, Bolasco G, Pagani F, Maggi L, Scianni M, Panzanelli P, Giustetto M, Ferreira TA, Guiducci E, Dumas L, Ragozzino D, Gross CT, Synaptic pruning by microglia is necessary for normal brain development. Science 333, 1456–1458 (2011). [DOI] [PubMed] [Google Scholar]

[R62] 62.Sullivan KD, Lewis HC, Hill AA, Pandey A, Jackson LP, Cabral JM, Smith KP, Liggett LA, Gomez EB, Galbraith MD, DeGregori J, Espinosa JM, Trisomy 21 consistently activates the interferon response. eLife 5, e16220 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R63] 63.Galbraith MD, Rachubinski AL, Smith KP, Araya P, Waugh KA, Enriquez-Estrada B, Worek K, Granrath RE, Kinning KT, Paul Eduthan N, Ludwig MP, Hsieh EWY, Sullivan KD, Espinosa JM, Multidimensional definition of the interferonopathy of Down syndrome and its response to JAK inhibition. Sci Adv 9, eadg6218 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R64] 64.Uchino Y, Woodward AM, Mauris J, Peterson K, Verma P, Nilsson UJ, Rajaiya J, Argüeso P, Galectin-3 is an amplifier of the interleukin-1β-mediated inflammatory response in corneal keratinocytes. Immunology 154, 490–499 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R65] 65.Shirakawa K, Endo J, Kataoka M, Katsumata Y, Yoshida N, Yamamoto T, Isobe S, Moriyama H, Goto S, Kitakata H, Hiraide T, Fukuda K, Sano M, IL (Interleukin)-10–STAT3–Galectin-3 Axis Is Essential for Osteopontin-Producing Reparative Macrophage Polarization After Myocardial Infarction. Circulation 138, 2021–2035 (2018). [DOI] [PubMed] [Google Scholar]

[R66] 66.McEver RP, Selectins: initiators of leucocyte adhesion and signalling at the vascular wall. Cardiovasc Res 107, 331–339 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R67] 67.Stevens B, Allen NJ, Vazquez LE, Howell GR, Christopherson KS, Nouri N, Micheva KD, Mehalow AK, Huberman AD, Stafford B, Sher A, Litke AM, Lambris JD, Smith SJ, John SWM, Barres BA, The classical complement cascade mediates CNS synapse elimination. Cell 131, 1164–1178 (2007). [DOI] [PubMed] [Google Scholar]

[R68] 68.Krasemann S, Madore C, Cialic R, Baufeld C, Calcagno N, El Fatimy R, Beckers L, O’Loughlin E, Xu Y, Fanek Z, Greco DJ, Smith ST, Tweet G, Humulock Z, Zrzavy T, Conde-Sanroman P, Gacias M, Weng Z, Chen H, Tjon E, Mazaheri F, Hartmann K, Madi A, Ulrich J, Glatzel M, Worthmann A, Heeren J, Budnik B, Lemere C, Ikezu T, Heppner FL, Litvak V, Holtzman DM, Lassmann H, Weiner HL, Ochando J, Haass C, Butovsky O, The TREM2-APOE pathway drives the transcriptional phenotype of dysfunctional microglia in neurodegenerative diseases. Immunity 47, 566–581.e9 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R69] 69.Hopperton KE, Mohammad D, Trépanier MO, Giuliano V, Bazinet RP, Markers of microglia in post-mortem brain samples from patients with Alzheimer’s disease: a systematic review. Mol Psychiatry 23, 177–198 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R70] 70.Csumitta KD, Gotts SJ, Clasen LS, Martin A, Raitano Lee N, Youth with Down syndrome display widespread increased functional connectivity during rest. Sci Rep 12, 9836 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R71] 71.Jacola LM, Byars AW, Hickey F, Vannest J, Holland SK, Schapiro MB, Functional magnetic resonance imaging of story listening in adolescents and young adults with Down syndrome: evidence for atypical neurodevelopment. J Intellect Disabil Res 58, 892–902 (2014). [DOI] [PubMed] [Google Scholar]

[R72] 72.Anderson JS, Nielsen JA, Ferguson MA, Burback MC, Cox ET, Dai L, Gerig G, Edgin JO, Korenberg JR, Abnormal brain synchrony in Down Syndrome. Neuroimage Clin 2, 703–715 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R73] 73.Vega JN, Hohman TJ, Pryweller JR, Dykens EM, Thornton-Wells TA, Resting-State Functional Connectivity in Individuals with Down Syndrome and Williams Syndrome Compared with Typically Developing Controls. Brain Connect 5, 461–475 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R74] 74.Jacola LM, Byars AW, Chalfonte-Evans M, Schmithorst VJ, Hickey F, Patterson B, Hotze S, Vannest J, Chiu C-Y, Holland SK, Schapiro MB, Functional magnetic resonance imaging of cognitive processing in young adults with Down syndrome. Am J Intellect Dev Disabil 116, 344–359 (2011). [DOI] [PubMed] [Google Scholar]

[R75] 75.Waugh KA, Minter R, Baxter J, Chi C, Galbraith MD, Tuttle KD, Eduthan NP, Kinning KT, Andrysik Z, Araya P, Dougherty H, Dunn LN, Ludwig M, Schade KA, Tracy D, Smith KP, Granrath RE, Busquet N, Khanal S, Anderson RD, Cox LL, Estrada BE, Rachubinski AL, Lyford HR, Britton EC, Fantauzzo KA, Orlicky DJ, Matsuda JL, Song K, Cox TC, Sullivan KD, Espinosa JM, Triplication of the interferon receptor locus contributes to hallmarks of Down syndrome in a mouse model. Nat Genet 55, 1034–1047 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R76] 76.Escoubas CC, Dorman LC, Nguyen PT, Lagares-Linares C, Nakajo H, Anderson SR, Barron JJ, Wade SD, Cuevas B, Vainchtein ID, Silva NJ, Guajardo R, Xiao Y, Lidsky PV, Wang EY, Rivera BM, Taloma SE, Kim DK, Kaminskaya E, Nakao-Inoue H, Schwer B, Arnold TD, Molofsky AB, Condello C, Andino R, Nowakowski TJ, Molofsky AV, Type-I-interferon-responsive microglia shape cortical development and behavior. Cell 187, 1936–1954.e24 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R77] 77.Larsen FK, Baksh RA, McGlinchey E, Langballe EM, Benejam B, Beresford-Webb J, McCarron M, Coppus A, Falquero S, Fortea J, Levin J, Loosli SV, Mark R, Rebillat A, Zaman S, Strydom A, Age of Alzheimer’s disease diagnosis in people with Down syndrome and associated factors: Results from the Horizon 21 European Down syndrome consortium. Alzheimers Dement 20, 3270–3280 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R78] 78.Iulita MF, Garzón Chavez D, Klitgaard Christensen M, Valle Tamayo N, Plana-Ripoll O, Rasmussen SA, Roqué Figuls M, Alcolea D, Videla L, Barroeta I, Benejam B, Altuna M, Padilla C, Pegueroles J, Fernandez S, Belbin O, Carmona-Iragui M, Blesa R, Lleó A, Bejanin A, Fortea J, Association of Alzheimer Disease With Life Expectancy in People With Down Syndrome. JAMA Network Open 5, e2212910 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R79] 79.DeTure MA, Dickson DW, The neuropathological diagnosis of Alzheimer’s disease. Molecular Neurodegeneration 14, 32 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R80] 80.Wilson DM, Cookson MR, Van Den Bosch L, Zetterberg H, Holtzman DM, Dewachter I, Hallmarks of neurodegenerative diseases. Cell 186, 693–714 (2023). [DOI] [PubMed] [Google Scholar]

[R81] 81.Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P, The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 1, 417–425 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R82] 82.Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR, STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R83] 83.Martin M, Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 17, 10–12 (2011). [Google Scholar]

[R84] 84.Li H, Durbin R, Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R85] 85.Young MD, Behjati S, SoupX removes ambient RNA contamination from droplet-based single-cell RNA sequencing data. Gigascience 9, giaa151 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R86] 86.Mathys H, Peng Z, Boix CA, Victor MB, Leary N, Babu S, Abdelhady G, Jiang X, Ng AP, Ghafari K, Kunisky AK, Mantero J, Galani K, Lohia VN, Fortier GE, Lotfi Y, Ivey J, Brown HP, Patel PR, Chakraborty N, Beaudway JI, Imhoff EJ, Keeler CF, McChesney MM, Patel HH, Patel SP, Thai MT, Bennett DA, Kellis M, Tsai L-H, Single-cell atlas reveals correlates of high cognitive function, dementia, and resilience to Alzheimer’s disease pathology. Cell 186, 4365–4385.e27 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R87] 87.McGinnis CS, Murrow LM, Gartner ZJ, DoubletFinder: Doublet Detection in Single-Cell RNA Sequencing Data Using Artificial Nearest Neighbors. Cell Syst 8, 329–337.e4 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R88] 88.Wolock SL, Lopez R, Klein AM, Scrublet: Computational Identification of Cell Doublets in Single-Cell Transcriptomic Data. Cell Syst 8, 281–291.e9 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R89] 89.Germain P-L, Lun A, Garcia Meixide C, Macnair W, Robinson MD, Doublet identification in single-cell sequencing data using scDblFinder. F1000Res 10, 979 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R90] 90.Hao Y, Stuart T, Kowalski MH, Choudhary S, Hoffman P, Hartman A, Srivastava A, Molla G, Madad S, Fernandez-Granda C, Satija R, Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol 42, 293–304 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R91] 91.Stuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck WM, Hao Y, Stoeckius M, Smibert P, Satija R, Comprehensive Integration of Single-Cell Data. Cell 177, 1888–1902.e21 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R92] 92.Korotkevich G, Sukhov V, Budin N, Shpak B, Artyomov MN, Sergushichev A, Fast gene set enrichment analysis. bioRxiv [Preprint] (2021). 10.1101/060012. [DOI] [Google Scholar]

[R93] 93.Wolf FA, Angerer P, Theis FJ, SCANPY: large-scale single-cell gene expression data analysis. Genome Biol 19, 15 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R94] 94.Stuart T, Srivastava A, Madad S, Lareau CA, Satija R, Single-cell chromatin state analysis with Signac. Nat Methods 18, 1333–1341 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R95] 95.Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, Nusbaum C, Myers RM, Brown M, Li W, Liu XS, Model-based analysis of ChIP-Seq (MACS). Genome Biol 9, R137 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R96] 96.Amemiya HM, Kundaje A, Boyle AP, The ENCODE Blacklist: Identification of Problematic Regions of the Genome. Sci Rep 9, 9354 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R97] 97.Roadmap Epigenomics Consortium, Kundaje A, Meuleman W, Ernst J, Bilenky M, Yen A, Heravi-Moussavi A, Kheradpour P, Zhang Z, Wang J, Ziller MJ, Amin V, Whitaker JW, Schultz MD, Ward LD, Sarkar A, Quon G, Sandstrom RS, Eaton ML, Wu Y-C, Pfenning AR, Wang X, Claussnitzer M, Liu Y, Coarfa C, Harris RA, Shoresh N, Epstein CB, Gjoneska E, Leung D, Xie W, Hawkins RD, Lister R, Hong C, Gascard P, Mungall AJ, Moore R, Chuah E, Tam A, Canfield TK, Hansen RS, Kaul R, Sabo PJ, Bansal MS, Carles A, Dixon JR, Farh K-H, Feizi S, Karlic R, Kim A-R, Kulkarni A, Li D, Lowdon R, Elliott G, Mercer TR, Neph SJ, Onuchic V, Polak P, Rajagopal N, Ray P, Sallari RC, Siebenthall KT, Sinnott-Armstrong NA, Stevens M, Thurman RE, Wu J, Zhang B, Zhou X, Beaudet AE, Boyer LA, De Jager PL, Farnham PJ, Fisher SJ, Haussler D, Jones SJM, Li W, Marra MA, McManus MT, Sunyaev S, Thomson JA, Tlsty TD, Tsai L-H, Wang W, Waterland RA, Zhang MQ, Chadwick LH, Bernstein BE, Costello JF, Ecker JR, Hirst M, Meissner A, Milosavljevic A, Ren B, Stamatoyannopoulos JA, Wang T, Kellis M, Integrative analysis of 111 reference human epigenomes. Nature 518, 317–330 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R98] 98.Ernst J, Kellis M, Chromatin-state discovery and genome annotation with ChromHMM. Nat Protoc 12, 2478–2492 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R99] 99.Hinrichs AS, Karolchik D, Baertsch R, Barber GP, Bejerano G, Clawson H, Diekhans M, Furey TS, Harte RA, Hsu F, Hillman-Jackson J, Kuhn RM, Pedersen JS, Pohl A, Raney BJ, Rosenbloom KR, Siepel A, Smith KE, Sugnet CW, Sultan-Qurraie A, Thomas DJ, Trumbower H, Weber RJ, Weirauch M, Zweig AS, Haussler D, Kent WJ, The UCSC Genome Browser Database: update 2006. Nucleic Acids Research 34, D590–D598 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R100] 100.Wang Q, Li M, Wu T, Zhan L, Li L, Chen M, Xie W, Xie Z, Hu E, Xu S, Yu G, Exploring Epigenomic Datasets by ChIPseeker. Curr Protoc 2, e585 (2022). [DOI] [PubMed] [Google Scholar]

[R101] 101.Yu G, Wang L-G, He Q-Y, ChIPseeker: an R/Bioconductor package for ChIP peak annotation, comparison and visualization. Bioinformatics 31, 2382–2383 (2015). [DOI] [PubMed] [Google Scholar]

[R102] 102.Rauluseviciute I, Riudavets-Puig R, Blanc-Mathieu R, Castro-Mondragon JA, Ferenc K, Kumar V, Lemma RB, Lucas J, Chèneby J, Baranasic D, Khan A, Fornes O, Gundersen S, Johansen M, Hovig E, Lenhard B, Sandelin A, Wasserman WW, Parcy F, Mathelier A, JASPAR 2024: 20th anniversary of the open-access database of transcription factor binding profiles. Nucleic Acids Res 52, D174–D182 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R103] 103.Gel B, Serra E, karyoploteR: an R/Bioconductor package to plot customizable genomes displaying arbitrary data. Bioinformatics 33, 3088–3090 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R104] 104.Quinlan AR, Hall IM, BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics 26, 841–842 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R105] 105.Nurk S, Koren S, Rhie A, Rautiainen M, Bzikadze AV, Mikheenko A, Vollger MR, Altemose N, Uralsky L, Gershman A, Aganezov S, Hoyt SJ, Diekhans M, Logsdon GA, Alonge M, Antonarakis SE, Borchers M, Bouffard GG, Brooks SY, Caldas GV, Chen N-C, Cheng H, Chin C-S, Chow W, de Lima LG, Dishuck PC, Durbin R, Dvorkina T, Fiddes IT, Formenti G, Fulton RS, Fungtammasan A, Garrison E, Grady PGS, Graves-Lindsay TA, Hall IM, Hansen NF, Hartley GA, Haukness M, Howe K, Hunkapiller MW, Jain C, Jain M, Jarvis ED, Kerpedjiev P, Kirsche M, Kolmogorov M, Korlach J, Kremitzki M, Li H, Maduro VV, Marschall T, McCartney AM, McDaniel J, Miller DE, Mullikin JC, Myers EW, Olson ND, Paten B, Peluso P, Pevzner PA, Porubsky D, Potapova T, Rogaev EI, Rosenfeld JA, Salzberg SL, Schneider VA, Sedlazeck FJ, Shafin K, Shew CJ, Shumate A, Sims Y, Smit AFA, Soto DC, Sović I, Storer JM, Streets A, Sullivan BA, Thibaud-Nissen F, Torrance J, Wagner J, Walenz BP, Wenger A, Wood JMD, Xiao C, Yan SM, Young AC, Zarate S, Surti U, McCoy RC, Dennis MY, Alexandrov IA, Gerton JL, O’Neill RJ, Timp W, Zook JM, Schatz MC, Eichler EE, Miga KH, Phillippy AM, The complete sequence of a human genome. Science 376, 44–53 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R106] 106.Guarracino A, Buonaiuto S, de Lima LG, Potapova T, Rhie A, Koren S, Rubinstein B, Fischer C, Gerton JL, Phillippy AM, Colonna V, Garrison E, Recombination between heterologous human acrocentric chromosomes. Nature 617, 335–343 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R107] 107.Kuleshov MV, Jones MR, Rouillard AD, Fernandez NF, Duan Q, Wang Z, Koplev S, Jenkins SL, Jagodnik KM, Lachmann A, McDermott MG, Monteiro CD, Gundersen GW, Ma’ayan A, Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res 44, W90–97 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[R108] 108.Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B, Ideker T, Cytoscape: A Software Environment for Integrated Models of Biomolecular Interaction Networks. Genome Res 13, 2498–2504 (2003). [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Molecular and cellular processes disrupted in the early postnatal Down syndrome prefrontal cortex

Ryan D Risgaard

Kalpana Hanthanan Arachchilage

Sara A Knaack

Masoumeh Hosseini

Rachel J Chen

Pubudu Kumarage

Danielle K Schmidt

Xiang Huang

Jie Sheng

Carlos J Wang

Elisa Giusti

Shuang Liu

Su-Chun Zhang

Daifeng Wang

Anita Bhattacharyya

Andre M M Sousa

Abstract

Introduction

RESULTS

Transcriptomic classification reveals selective vulnerability of cell populations in the early postnatal Ts21 dlPFC.

Fig. 1. Single-nucleus multiomic gene expression and chromatin accessibility in the early postnatal Down syndrome dorsolateral prefrontal cortex.

Cell composition alterations in the Ts21 dlPFC

Genome-wide transcriptional and chromatin remodeling in Ts21

Cell-cell interaction analyses reveal Ts21 dysregulated neuronal and vascular-associated signaling networks

Chromosome 21 chromatin accessibility and gene expression in Ts21

Fig. 2. Genome-wide transcriptional and chromatin remodeling in Ts21.

Metabolic and synaptic deficits in Ts21 excitatory neurons

Fig. 3. Metabolic and synaptic deficits in Ts21 intratelencephalic-projecting excitatory neurons.

Conserved ExN:InN neuron ratio and global synaptic dysfunction in the Ts21 dlPFC

Fig. 4. Conserved ExN:InN neuron ratio and global synaptic dysfunction in the Ts21 dlPFC.

Cell-type-specific regulatory deficits in Ts21 oligodendrocyte maturation and myelination

Fig. 5. Cell-type specific regulatory deficits in Ts21 oligodendrocyte maturation and myelination.

Molecular signatures of Ts21 astrocyte dysfunction reveal regulators of reactivity, stress, and cell survival

Fig. 6. Molecular signatures of Ts21 astrocyte dysfunction reveal regulators of reactivity, stress response, and cell survival.

Cell intrinsic gene regulation in Ts21 neuroinflammatory microglial activation

Fig. 7. Dysregulated neuroimmune crosstalk and subtype-specific molecular features of Ts21 microglia.

Dysregulated neuroimmune crosstalk and subtype-specific molecular features of Ts21 microglia

Conclusions

Materials and Methods

Postmortem brain samples

Nucleus isolation for single-nucleus multiome (snATAC-seq and snRNA-seq)

snMultiome library generation and sequencing

Alignment and processing of snMultiome data

snRNA-seq data processing

Phase 1: Quality Control, Batch Correction, and Cell Type Annotation

Phase 2: Data Integration

Evaluation of clustering stability and cell type separability

Transcriptomic similarities/differences across cell type and genotype

Transcriptomic heterogeneity within cell types

Differential cell composition analysis

Differential gene expression analysis

Gene Set Enrichment Analysis

Inferred Trajectories of Oligodendrocyte Development

Inference, analysis, and visualization of differential cell-cell communication

snATAC-seq data processing

snATAC peak GREAT analysis

snATAC differentially accessible region analysis

snATAC Motif finding, enrichments and foot-printing

Chromatin-accessibility and gene-expression at the chromosome-scale

Weighted gene co-expression networks

Multiomic gene regulatory network inference using SCENIC+

RNA extraction, cDNA synthesis, and droplet digital polymerase chain reaction (ddPCR)

Protein extraction and Western blotting

In situ hybridization (ISH)

Immunohistochemistry and confocal microscopy

Multiplexed immunoassay

Supplementary Material

Acknowledgements:

Funding:

Footnotes

Data and materials availability:

References and Notes

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES