Multimodal single-cell profiling reveals neuronal vulnerability and pathological cell states in focal cortical dysplasia

Abstract

Focal cortical dysplasia (FCD) is a neurodevelopmental condition characterized by malformations of the cerebral cortex that often cause drug-resistant epilepsy. In this study, we performed multi-omics single-nuclei profiling to map the chromatin accessibility and transcriptome landscapes of FCD type II, generating a comprehensive multimodal single-nuclei dataset comprising 61,525 cells from 11 clinical samples of lesions and controls. Our findings revealed profound chromatin, transcriptomic, and cellular alterations affecting neuronal and glial cells in FCD lesions, including the selective loss of upper-layer excitatory neurons, significant expansion of oligodendrocytes and immature astrocytic populations, and a distinct neuronal subpopulation harboring dysmorphic neurons. Furthermore, we uncovered activated microglia subsets, particularly in FCD IIb cases. This comprehensive study unveils neuronal and glial cell states driving FCD development and epileptogenicity, enhancing our understanding of FCD and offering directions for targeted therapy development.

Keywords: Biological sciences; Developmental neuroscience; Neuroscience; Omics.

Graphical abstract

Figure 1

Overview of the multimodal single-nuclei sequencing (snRNA-seq + snATAC-seq) in FCD type II (A) Schematic of FCD samples and study design. Cortical tissues from lesions and adjacent non-lesion areas in individuals with FCD IIa and IIb were profiled by multiome single-nuclei ATAC and RNA sequencing. Created with Biorender.com. (B) Uniform manifold approximation and projection (UMAP) visualization of joint modalities (ATAC + RNA) after sample integration. Nuclei are colored by sequencing batch. (C) UMAP joint visualization of nuclei colored by broad cell type annotation. (D) UMAP joint visualization colored using a high-resolution annotation with classification of neuronal subtypes. (E) UMAP plots depicting nuclei by tissue condition. (F) Dot plot showing gene expression of known marker genes for major cortical cell types. Dot size corresponds to the fraction of cells expressing the gene, and color denotes normalized expression levels. (G) Coverage plots of marker genes depicting chromatin accessibility across cell types from snATAC-seq. The normalized ATAC signal is depicted in a region of +/− 1 kb around gene start/end coordinates. (H) Boxplots showing cell type proportion by tissue condition. The center line of the boxplot shows the median of the data; the box limits show the upper and lower quartiles; the whiskers show 1.5 times interquartile ranges. Overlay dots represent cell type proportions in individual samples. Cell type changes between conditions were detected using a linear model implemented in Speckle. ˙p < 0.1, ∗p < 0.5, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 2

Excitatory and inhibitory neuron subtypes affected in FCD (A) Boxplots showing excitatory neurons (ENs) subtype frequency in FCD and controls. The center line of the boxplot shows the median of the data; the box limits show the upper and lower quartiles; the whiskers show 1.5 times interquartile ranges. Overlay dots represent proportions in individual samples. Cell type changes between conditions were detected using a linear model implemented in Speckle. ˙p < 0.1, ∗p < 0.5, ∗∗p < 0.01, ∗∗∗p < 0.001. (B) Bar graphs depicting the number of differentially expressed genes (DEGs) in EN subtypes. DEGs between FCD and controls were computed using MAST with adj. p < 0.05. Only the five neuronal subtypes with the most transcriptional changes are shown. Left: FCD IIa vs. controls. Color indicate the proportion of up- or downregulated genes. Right: FCD IIb vs. controls. (C) Boxplots showing inhibitory neurons (INs) subtype frequency in FCD and controls. Boxplots legends and significance levels are defined as in (A). (D) Bar graphs depicting the number of DEGs in IN subtypes. DEGs between FCD and controls were computed using MAST with adj. p < 0.05. Bar graphs legends are defined as in (B).

Figure 3

Characterization of a disease-specific (DS) cluster containing dysmorphic neurons (A) UMAP highlighting the DS cluster. The pie chart indicates the percentage of nuclei from FCD IIa, IIb, and controls within the cluster. (B) Pathway enrichment of DS cluster markers. Marker genes were identified using Seurat’s FindMarkers and KEGG pathway enrichment was performed using g:Profiler. The bars represent the significance score of pathway enrichment, and the color indicates the number of genes in each pathway. (C) Dot plot denoting expression of marker genes associated with neurons in the DS cluster. Neuronal genes are grouped by biological function. Dot size corresponds to the fraction of cells expressing the gene, and color denotes normalized expression levels. (D) Violin plots denoting gene expression of NEFM and NEFL neurofilaments in the various cell types. (E) Coverage chromatin accessibility plots of neuronal genes expressed in the DS cluster from snATAC-seq. The normalized ATAC signal is depicted in a region of +/− 1 kb around gene start/end coordinates. (F) UMAP showing the average expression of a dysmorphic neuron (DN) signature obtained from Baldassari et al. Signature expression was calculated using the AddModuleScore function from Seurat. (G) UMAP RNA plots showing the DS cluster population in FCD IIa and IIb lesions, histologically normal tissue from FCD individuals (internal controls), and healthy cortex (autopsy controls). The leftmost UMAP depicts annotated cell types in the integrated dataset, which was created by integrating snRNA-seq data from this study with Siletti et al. using Harmony.

Figure 4

Microglia pathological cell states and activation in FCD type IIb (A) UMAP visualization of microglia and lymphoid subclusters sequenced in FCD lesions. (B) Feature plots depicting expression of key genes distinguishing microglia subsets CD83⁺ (CD83, CCL2, and CCL3) and CD74⁺ (CD74, HLA-DRA, and HLA-DPA1). (C) Dot plot denoting the expression of key marker genes in microglia and lymphoid subclusters. Dot size corresponds to the fraction of cells expressing the gene, and color denotes normalized expression levels. Gene symbols are colored according to their subcluster defined in (A). (D) Boxplots denoting subcluster frequency in FCD and controls. The center line of the boxplot shows the median of the data; the box limits show the upper and lower quartiles; the whiskers show 1.5 times interquartile ranges. Overlay dots represent cell type proportions in individual samples. Cell type changes between conditions were detected using a linear model implemented in Speckle. ˙p < 0.1, ∗p < 0.5, ∗∗p < 0.01, ∗∗∗p < 0.001. (E) Volcano plot depicting differentially expressed genes (DEGs) in FCD IIb microglia. DEGs were obtained using a Wilcox test after SCT transformation with correction for sequencing batch, mitochondrial gene percentage, and brain region. The x axis is the fold-change (log2) expression in FCD IIb vs. controls, and the y axis depicts the significance of the change (−log10 of the adjusted p value). The color indicates the gene status. (F) Dot plots showing top enriched GO terms in up- and downregulated genes from (E) computed using clusterProfiler. Dot size and color correspond to the number of genes associated with the GO term and enrichment p value, respectively. (G) Top five differential motifs in microglia from FCD IIb vs. controls, as measured by chromVAR (see STAR Methods). (H) Feature plots depicting IRF8 and STAT1/STAT2 motif activity in microglia open chromatin regions from FCD IIb and controls. Cells are colored by the chromVAR score.

Figure 5

Open chromatin changes and cell-type specific regulatory links in FCD (A) Barplots depicting differentially accessible chromatin regions (DACRs) between FCD tissues and controls. The number of chromatin regions with opening or closing status (gain or loss of chromatin accessibility, respectively) is indicated for each cell type. (B) Enhancer types overlapping DACRs. The number of opening and closing peaks associated with distal or proximal enhancers from Encode are indicated. (C) Diagram depicting the inference of regulatory links identifying neighboring open chromatin regions (snATAC-seq peaks) associated with gene expression using linear regression and random forests (see STAR Methods). (D) Regulatory links predicted for FKBP5 in astrocytes. ATAC coverage tracks and corresponding expression levels in controls, FCD IIb, and IIa tissues are depicted at the top. Regions highlighted in gray indicate opening peaks in lesions and red asterisks indicate Encode-annotated enhancers. The gene track shows the location of the FKBP5 locus in chromosome 6, and the bottom track shows all peaks contained in the chromosomal region.

Figure 6

Impaired Astrocyte Differentiation in FCD (A) UMAP visualization of astrocytes temporally ordered on a trajectory as a function of pseudotime built by Monocle3. Nuclei are colored according to inferred pseudotime. (B) Feature plots depicting normalized gene expression of immature astrocyte markers CD44, TNC, VCAN, and GFAP (top), and differentiated astrocyte markers SLC1A2, SLC1A3, and GPC5 (bottom). (C) UMAP visualization and density quantification along the trajectory. Density plots indicate the normalized density of cells along the pseudotime according to tissue condition. (D) Feature plots depicting expression of balloon cells and reactive astrocytes gene signatures computed by AddModuleScore from Seurat (top). Bar graphs (bottom) quantify the proportion of active cells for these signatures along the pseudotime trajectory. Active cells were defined using the AUCell method and are colored by tissue condition. (E) Expression levels of genes associated with balloon cells and reactive astrocytes CHI3L1, HSPB1, SERPINA3, GFAP, and VIM in the trajectory and by tissue condition.