Single-cell dissection of the human motor and prefrontal cortices in ALS and FTLD

Abstract

Amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration (FTLD) share many clinical, pathological, and genetic features, but a detailed understanding of their associated transcriptional alterations across vulnerable cortical cell types is lacking. Here, we report a high-resolution, comparative single-cell molecular atlas of the human primary motor and dorsolateral prefrontal cortices and their transcriptional alterations in sporadic and familial ALS and FTLD. By integrating transcriptional and genetic information, we identify known and previously unidentified vulnerable populations in cortical layer 5 and show that ALS- and FTLD-implicated motor and spindle neurons possess a virtually indistinguishable molecular identity. We implicate potential disease mechanisms affecting these cell types as well as non-neuronal drivers of pathogenesis. Finally, we show that neuron loss in cortical layer 5 tracks more closely with transcriptional identity rather than cellular morphology and extends beyond previously reported vulnerable cell types.

Keywords: ALS; Betz cell; FTLD; frontotemporal dementia; motor neuron; neurodegeneration; single cell; spindle neuron; von Economo.

Figure 1:. Experimental design and characterization of cell types in the affected cortices.

A. Experimental design for transcriptional profiling of 73 ALS, FTLD, and PN individuals (625,973 cells). B-D. UMAP subplots of annotated (B) excitatory neurons, (C) inhibitory neurons, and (D) non-neuronal subtypes. Insets: clusters of full UMAP visualized in each subplot. E. Taxonomy of annotated subtypes. Values represent average log-transformed marker expression normalized to maximum per column. F. Transcriptional similarity of neuronal populations across brain regions.

Figure 2:. Global gene expression changes in ALS and FTLD.

A. Similarity of disease-induced gene expression changes across the profiled cohort. Examples of: (1) conserved intra-phenotype, intra-region alterations within and across cell types; (2) conserved intra-phenotype, cross-region alterations within and across cell types; (3) non-conservation of changes across dissimilar phenotypes and brain regions; (4) conserved cross-phenotype, intra-region alterations predominantly within cell types; (5) conserved gene expression changes across excitatory neurons within a single phenotype and brain region. B. Comparison of excitatory neuron DEGs across genotypes. In each panel, the Pearson correlation is shown in the top-left and the number of overlapping DEGs (FDR < 0.05 in both genotypes) is shown in the bottom-right. C. Select top-ranking GO terms enriched in DEGs of excitatory neurons in ALS and FTLD cases. D. Top up- and downregulated genes shared across diseases in excitatory neurons, per brain region. E. Differential expression of heat shock genes broadly dysregulated across excitatory neurons.

Figure 3:. Non-neuronal disease mechanisms and vascular changes.

A. Top up- and downregulated genes shared across diseases in astrocyte populations. B. Differential expression of disease-associated astrocyte markers. C. Top up- and downregulated genes shared across diseases in microglia. D. Differential expression of microglial genes associated with disease response and homeostasis. Combined legend for A-D. E. Top up- and downregulated genes shared across diseases in oligodendrocytes. F. Differential expression of select junctional, endothelial identity, and immune response genes in endothelial cells. G: Immunofluorescent labeling of ArpC3 with lectin in MCX. Lectin selectively labels blood vessels. H: Quantification of lectin and ArpC3 signal (n = 5 PN and n = 5 SALS donors). I: Immunofluorescent co-labeling of claudin-5 and VE-cadherin in MCX. J: Quantification of junctional claudin-5 and VE-cadherin signals (n = 5 PN and n = 5 SALS donors). K. Immunofluorescent labeling of vascular HLA-E in MCX. L. Quantification of HLA-E signal (n = 6 PN and n = 6 SALS donors). In G, I, and K, the scale bar is 50 μm. In H, J, and L, the error bars represent standard error of the mean. (*) p < 0.05; (**) p < 0.01; unpaired one-tailed t-test.

Figure 4:. Drivers of cell type-specific differential vulnerability.

A. Heatmap of disease susceptibility scores based on ALS GWAS-linked gene enrichment. Scores superimposed on PN cells of UMAP from Figure 1A. B. Heatmap on the left shows the median susceptibility scores per subtype. Bar plot on the right shows the number of GWAS-linked genes enriched in each subtype. Wilcoxon rank sum test (log₂-fold change z-score > 1 and FDR < 0.001). C. Transcriptional divergence (TxD) score of subtypes by disease and brain region. D. Left: Pearson correlation between susceptibility (from B) and TxD (from C) scores. Right: FDR-adjusted correlation p-values of corresponding comparisons on the left. E. TxD association analysis of ALS and FTD-linked genes. Genes with a significant association with more than one disease group are shown. See extended Figure S3A for all genes. F. Differential enrichment of ALS and FTD-linked genes across excitatory neuron subtypes. Differentially vulnerable (by TxD score) subtypes are denoted in red. Genes enriched in at least one vulnerable subtype are shown. G. Differential expression of select ALS and FTD-linked genes across excitatory neurons in disease. H. Select top-ranking GO terms (left) and KEGG pathways (right) enriched among TxD-associated genes. Terms have positive NES if enriched among positively correlated genes and negative otherwise.

Figure 5:. Characterization of vulnerable L5 excitatory populations

A. Relative expression of known markers of L5 ET neurons, including those with Betz and VEN morphology, from various brain regions. B. Heatmaps superimposed on Figure 1B showing relative expression of known and novel markers of differentially vulnerable excitatory neuron subtypes (whose location in the UMAP is shown on the left). Inset shows markers that distinguish subtypes of L5 VAT1L+ neurons. C. UMAP of excitatory neurons (same as Figure 1B) colored by brain region of origin. Inset: region-specific heterogeneity of L5 VAT1L+ subtypes. D. Differential expression of ALS and FTD-linked genes in L5 VAT1L+ neurons between human and mouse MCX. Only statistically significant (z-score > 1 and FDR < 0.001) genes are marked. NS: not significant. E. Select top-ranking GO terms enriched in marker genes of human L5 VAT1L+ subtypes. See Figure S4B for extended results. F. Differential enrichment of cilia-associated genes across excitatory neuron subtypes. Differentially vulnerable subtypes are denoted in red. See Figure S4C for extended results. G. Representative images of IHC labeling of VAT1L (brown) in PN MCX (left), PFC (middle), and ACC (right). Black arrows denote cells with Betz (in MCX), or spindle (in PFC and ACC) morphology. Scale bar is 100 μm.

Figure 6:. Impact of disease in differentially vulnerable cell types

A. Representative images of VAT1L and TDP-43 pS409/410 in diseased tissue samples by IHC. Black arrows denote intraneuronal inclusions. Scale bar is 50 μm. (B and C) Histological quantification of VAT1L+ neuron density in L5 across brain regions and diseases. In (B), all L5 VAT1L+ neurons are quantified. In (C), only neurons with Betz (left) or non-Betz (right) morphology are considered. Box and whisker plots show median and interquartile range values. (*) p < 0.05; (**) p < 0.01; (***) p < 0.001; one-tailed t test. See Figure S6A for a table of results. D. Enrichment of top-ranking GO terms in DEGs of L5 VAT1L+ neurons by brain region. E. Top DEGs of overrepresented GO term families enriched in L5 VAT1L+ neurons. F. Top ranking ChEA3-predicted regulators of L5 VAT1L+ neurons DEGs. Value is the mean rank across ChEA3 libraries.