The cholesterol transporter NPC1 is essential for epigenetic regulation and maturation of oligodendrocyte lineage cells

Abstract

The intracellular cholesterol transporter NPC1 functions in late endosomes and lysosomes to efflux unesterified cholesterol, and its deficiency causes Niemann-Pick disease Type C, an autosomal recessive lysosomal disorder characterized by progressive neurodegeneration and early death. Here, we use single-nucleus RNA-seq on the forebrain of Npc1^-/- mice at P16 to identify cell types and pathways affected early in pathogenesis. Our analysis uncovers significant transcriptional changes in the oligodendrocyte lineage during developmental myelination, accompanied by diminished maturation of myelinating oligodendrocytes. We identify upregulation of genes associated with neurogenesis and synapse formation in Npc1^-/- oligodendrocyte lineage cells, reflecting diminished gene silencing by H3K27me3. Npc1^-/- oligodendrocyte progenitor cells reproduce impaired maturation in vitro, and this phenotype is rescued by treatment with GSK-J4, a small molecule inhibitor of H3K27 demethylases. Moreover, mobilizing stored cholesterol in Npc1^-/- mice by a single administration of 2-hydroxypropyl-β-cyclodextrin at P7 rescues myelination, epigenetic marks, and oligodendrocyte gene expression. Our findings highlight an important role for NPC1 in oligodendrocyte lineage maturation and epigenetic regulation, and identify potential targets for therapeutic intervention.

Fig. 1. snRNA-seq of WT and Npc1^−/− forebrain at P16.

Nuclei were isolated from the forebrains of three mice per group and sequenced using the 10X Genomics Chromium platform. A UMAP plot for WT and Npc1^−/− samples with each point representing a single nucleus. Plot points are clustered into cell type based on gene expression patterns. B Bubble matrix showing the expression of Npc1 (at top) and cluster-defining genes. Delta symbols (Δ) within bubbles indicate the three most highly expressed defining genes for each cluster. Size of the bubble corresponds to the percentage of cells expressing a gene, while bubble color indicates expression level. C Number of up- (red) and down- (blue) regulated differentially expressed (DE) genes (genes with |log₂(fold-change)| of > 0.5 and adjusted p < 0.05) per cell cluster as determined through differential expression analysis utilizing a Wilcoxon rank-sum test. Source data are provided as a Source Data file. D Graph showing the percentage of the total nuclei captured in each cluster. The colors and cell types correspond with (A). Inset shows oligodendrocyte lineage.

Fig. 2. Death of oligodendrocyte lineage cells during development.

A Midline sagittal brain section from P16 mice stained for MBP (green) and DAPI (blue). B Brain sections from P6, P9, and P16 WT and Npc1^−/− mice were stained for oligodendrocyte lineage cells using OLIG2 (green). Nuclei are stained with DAPI (blue). Quantified at right. C Western blot of SOX10, OLIG2, and neurofilament (NF200) from P6 and P16 mouse forebrain. Values were normalized to vinculin and quantified relative to WT. Quantifications at right. D Brain sections were stained for DNA double-strand breaks using a TUNEL assay (red) to detect cell death. Nuclei are stained with DAPI (blue). E Expression of oligodendrocyte stage-specific markers in P16 mouse forebrain as measured by qRT-PCR and normalized to WT. Dashed lines indicate corpus callosum in all images. Scale bars = 200 µm (A) and 150 µm (B and D). Quantified at right. Data are mean ± SEM. n.s. not significant, *p < 0.05,**p < 0.005, ***p < 0.001 by two-tailed unpaired Student’s t-test; (A) Experiment was done on 3 biological replicates per group, all with similar results; (B) n = 5 mice; t = 0.3385, 2.477, 4.182; df = 8; p = 0.7437, 0.0383, 0.0031 (C) n = 5 mice; t = 0.47, 0.90, 4.33, 5.30; df = 8; p = 0.6487, 0.3958, 0.0025, 0.0007; (D) n = 5 mice; t = 0.6961, 4.081, 1.463; df = 8; p = 0.5061, 0.0035, 0.1817; (E) n = 5 mice; t = 0.1813, 0.06927, 3.891, 3.479, 6.470, 3.618; df = 8; p = 0.8606, 0.9465, 0.0046, 0.0083, 0.0002, 0.0068. Source data are provided as a Source Data file.

Fig. 3. Gene expression changes in the oligodendrocyte lineage.

A Gene Ontology analysis was performed on differentially expressed genes (DE) identified by snRNA-seq. The 10 most significant (by p-value) up- (red) and down-regulated (blue) GO terms (biological process) in Npc1^−/− oligodendrocyte lineage cells are shown. The dashed line indicates an adjusted p < 0.05 after Bonferroni correction. B Volcano plots of DE genes in the oligodendrocyte lineage as determined by differential expression analysis of snRNA-seq. The top 5 up- or down-regulated genes by log₂(FC) are labeled. Dashed lines correspond to an adjusted p-value of 0.05 and a |log₂(FC)| of 0.5.

Fig. 4. Epigenetic dysregulation of Npc1^−/− oligodendrocytes.

A H3K27me3 and H3K9me3 increase as oligodendrocytes differentiate. PDGFRa expression peaks earlier in development compared to O4. The color in the bars approximates marker levels. B Brain sections from P16 WT and Npc1^−/− mice were stained for H3K9me3 or H3K27me3 (green) and co-stained for SOX10 (red, oligodendrocytes) or NeuN (neurons). Nuclei stained with DAPI (blue). Images were captured from the corpus callosum (oligodendrocytes) and cortex (neurons). Representative images show H3K27me3 in OLs; images of NeuN and OL H3K9me3 stains in Supplementary Fig. 4. Scale bar = 50 µm. Methylation staining intensity in the cortex (neurons) and corpus callosum (oligodendrocytes) is quantified below. Methylation intensity was quantified in SOX10+ or NeuN+ cells and values were normalized to WT. Western blot of histones extracted from cells purified from P16 mice using microbeads targeting PDGFRa (C), O4 (D), or the flow through cells left over after cell purification (E). Values were normalized to total H3 and quantified relative to WT. F Western blot of EZH2 from O4-purified cells. Data are mean ± SEM. n.s. not significant, *p < 0.05,**p < 0.005, ***p < 0.001 by two-tailed unpaired Student’s t-test; (B) n = 5 mice; t = 3.1, 4.2, 0.4, 0.3; df = 8; p = 0.0146, 0.0031, 0.6765, 0.8028; (C) n = 3 biological replicates; t = 12.9, 0.5; df = 4; p = 0.0002, 0.6418; (D) n = 3; t = 3.0, 2.9; df = 4; p = 0.0415, 0.0451 (E) n = 3 biological replicates; t = 0.03, 0.7; df = 4; p = 0.9754, 0.5059; (F) n = 6 biological replicates; t = 4.0; df = 10; p = 0.0025. Source data are provided as a Source Data file.

Fig. 5. Dysregulation of H3K27me3 and H3K27ac at neuron-related genes.

ChIP-seq was performed on O4-purified cells from P16 WT and Npc1^-/- mice. A Average H3K27me3 occupancy at gene promoters (defined as being within 5 kb of transcription start site (TSS)) for WT (red), Npc1^−/− (blue), and background (green). B H3K27me3 promoter enrichment (Npc1^−/− relative to WT, log₂(fold change), y-axis) plotted against gene expression (Npc1^-/- relative to WT, log₂(fold change), x-axis) for genes with differential H3K27me3 (defined as having a log₂(fold change) <−0.3) and significant changes in expression. Red dots indicate 16 genes associated with neuronal pathways. C Pathways significantly enriched among genes with decreased H3K27me3 and increased expression in Npc1^−/− cells, as determined by Gene Ontology: Cellular Component analysis with Bonferroni correction. D Average H3K27ac occupancy at gene promoters for WT (red), Npc1^−/− (blue), and background (green). E H3K27ac promoter enrichment (Npc1^−/− relative to WT, log₂(fold change), y-axis) plotted against gene expression (Npc1^−/− relative to WT, log₂ fold change, x-axis) for genes with differential H3K27ac and significant changes in expression. Red dots indicate 10 genes associated with neuronal pathways. F IGV genome browser snapshots of H3K27me3 (blue) and H3K27ac (red) bigWig files at promoter regions of Lhfpl4, Cpeb1, and Kcna6. Arrows point to areas of enrichment. A, D Center of plotted lines indicates average occupancy, while line thickness shows the 95% confidence interval.

Fig. 6. Inhibiting H3K27 demethylase promotes oligodendrocyte differentiation in vitro.

Primary OPCs from WT and Npc1^−/− mice were treated with the KDM6A and KDM6B inhibitor GSK-J4 (0.5 µM), the inactive control GSK-J5 (0.5 µM), or DMSO vehicle (Veh) for 4 days. A Npc1^−/− cells treated with GSK-J5 or GSK-J4 and stained for H3K27me3 (green) and DAPI (blue). Intensity of H3K27me3 staining was measured relative to WT cells treated with GSK-J5. B Npc1^−/− cells were treated with GSK-J5 or GSK-J4, and gene expression was quantified by qRT-PCR relative to cells treated with GSK-J5. Lines connect paired samples from biological replicate cultures. C Cells were stained for SOX10 (red) and MBP (green) to assess differentiation. Nuclei stained with DAPI (blue). The percentage of MBP-positive cells is quantified at the right. Scale bars = 50 µM (A) and 100 µM (C). Data are mean ± SEM (A, C). n.s. not significant, *p < 0.05, ****p < 0.0001 by two-tailed unpaired Student’s t-test (A), two-tailed paired Student’s t-test (B), and one-way ANOVA with Tukey post hoc test and multiple comparisons (C). A n = 3 biological replicates; t = 3.510; df = 4; p = 0.0247 (B) n = 5 biological replicates; t = 2.782, 3.479, 3.158; df = 4; p = 0.0497, 0.0254, 0.0343; (C) n = 6 biological replicates; F = 95.44; df = 5. Source data are provided as a Source Data file.

Fig. 7. A single injection of 2-hydroxypropyl-β-cyclodextrin rescues myelination defects.

Mice were administered 2-hydroxypropyl-β-cyclodextrin (4000 mg/kg, i.p.) (HPβCD) or a vehicle control (saline) (Veh) at P7. Brains were collected and myelination assessed at P16. Dashed line in A indicates the outline of the brain section. Midline sagittal sections were stained for (A) MBP (red) and (B) OLIG2 (green). Nuclei stained with DAPI (blue). The number of OLIG2+ cells in the corpus callosum (dashed line in B) quantified at right. C Western blot for SOX10, OLIG2, and MBP from forebrain lysates. Values were normalized to vinculin and relative to WT. Quantified at right. D TEM images of the corpus callosum. G-ratios were quantified (right). E Midline sagittal sections were stained for H3K27me3 (green) and SOX10 (magenta). Images were captured from the corpus callosum, and the intensity of H3K27me3 within SOX10+ nuclei was quantified (right). F Expression of neuronal genes in O4+ cells purified from P16 mice as determined by qRT-PCR. A Scale bars = 1000 µm; (B) 100 µm; (D) 2 µm; (E) 100 µm. A Experiment was performed on 3 biological replicates per group, all with similar results; dashed lines outline the sagittal brain. D Dashed lines in the violin plot indicate quartiles. Data are mean ± SEM. n.s. not significant, *p < 0.05, **p < 0.005, ***p < 0.001, ****p < 0.0001 by one-way ANOVA with Tukey post hoc test and multiple comparisons (B–D, F) or two-tailed unpaired Student’s t-test (E). B n = 3 mice; F = 27.65; df = 2; p = 0.0009 (C) n = 5 mice; F = 35.97, 22.49, 16.55; df = 2; p = <0.0001, <0.0001, 0.0004 (D) Linear regression slopes: F = 1.574; df = 2; p = 0.2078; Linear regression intercepts: F = 256.9; df = 2; p < 0.0001; ANOVA: n = 443, 301, 327 axons (from 3 mice per group); F = 128.1; df = 2; p = <0.0001 (E) n = 3 mice; t = 3.167; df = 4; p = 0.0340 (F) n = 5 mice; F = 78.24, 10.83, 6.170; df = 2; p = <0.0001, 0.0021, 0.0144. Source data are provided as a Source Data file.

Fig. 8. Cell-autonomous defects in H3K27me3 are corrected by 2-hydroxypropyl-β-cyclodextrin.

H3K27me3 staining (green) was performed on sagittal brain sections from (A) Olig2-Cre;Npc1^fl/+ and Olig2-Cre;Npc1^fl/− or (B) Syn1-Cre;Npc1^fl/+ and Syn-Cre;Npc1^fl/− mice at P16. Sections were co-stained for SOX10 (red) and DAPI (blue). Images were taken from the corpus callosum, and H3K27me3 intensity within SOX10 + cells was quantified relative to control. C–E Olig2-Cre;Npc1^fl/+ and Olig2-Cre;Npc1^fl/− mice were administered 2-hydroxypropyl-β-cyclodextrin (4000 mg/kg, i.p.) (HPβCD) or a vehicle control (saline) (Veh) at P7. Brains were collected at P16. Sagittal sections were stained for (C) MBP (red), (D) OLIG2 (green), and (E) H3K27me3 (green). DAPI is in blue. Dashed lines outline the corpus callosum. The experiment in (C) was done on 3 biological replicates per group, all with similar results. OLIG2+ cells in the corpus callosum were quantified normalized to area (D). H3K27me3 intensity in SOX10+ (red) cells of the corpus callosum were quantified relative to Olig2-Cre;Npc1^fl/+ mice treated with vehicle (E). Scale bars = 50 µm (A, B, E), 200 µm (C), and 150 µm (D). Data are mean ± SEM. n.s. not significant, *p < 0.05, **p < 0.005 by two-tailed unpaired Student’s t-test. A n = 5 mice; t = 2.968; df = 8; p = 0.0179; (B) n = 4 mice; t = 0.5540; df = 6; p = 0.5996 (D) n = 4 mice; t = 3.826; df = 6; p = 0.0087; (E) n = 4 mice; t = 3.274; df = 6; p = 0.0170. Source data are provided as a Source Data file.