Patient iPSC models reveal glia-intrinsic phenotypes in multiple sclerosis

Abstract

Multiple sclerosis (MS) is an inflammatory and neurodegenerative disease of the central nervous system (CNS), resulting in neurological disability that worsens over time. While progress has been made in defining the immune system's role in MS pathophysiology, the contribution of intrinsic CNS cell dysfunction remains unclear. Here, we generated a collection of induced pluripotent stem cell (iPSC) lines from people with MS spanning diverse clinical subtypes and differentiated them into glia-enriched cultures. Using single-cell transcriptomic profiling and orthogonal analyses, we observed several distinguishing characteristics of MS cultures pointing to glia-intrinsic disease mechanisms. We found that primary progressive MS-derived cultures contained fewer oligodendrocytes. Moreover, MS-derived oligodendrocyte lineage cells and astrocytes showed increased expression of immune and inflammatory genes, matching those of glia from MS postmortem brains. Thus, iPSC-derived MS models provide a unique platform for dissecting glial contributions to disease phenotypes independent of the peripheral immune system and identify potential glia-specific targets for therapeutic intervention.

Keywords: astrocyte; glia; induced pluripotent stem cell; multiple sclerosis; oligodendrocyte.

Figure 1.. An iPSC-derived model to study CNS cell intrinsic dysfunction in MS.

(A) Schematic representation of iPSC reprogramming from people with MS and HC skin biopsies and differentiation of iPSCs into glial CNS cultures. (B) Select demographic information for the 16 iPSC lines used for scRNAseq analysis. (C) Representative images of iPSC-derived CNS cultures from HC, RRMS, SPMS, and PPMS. Cultures are stained for the mature oligodendrocyte marker MBP (teal), astrocyte marker GFAP (yellow), and neuron marker MAP2 (pink). Scale bar is 100μm. (D) UMAP of integrated single-cell analysis from 4 HC, 4 RRMS, 4 SPMS, and 4 PPMS lines, showing major cell type clusters. (E) Heatmap of the top 4 enriched genes for each cluster in (D). (F) Heatmap depicting the correlation between clusters in (D) and cell types from scRNAseq analysis of MS brain tissue (PMID: 30747918). Spearman correlation values generated using the R package ClustifyR. (G) Distribution of cell types within iPSC-derived CNS cultures from HC, RRMS, SPMS, and PPMS.

Figure 2.. iPSC-derived cultures from people with MS contain fewer oligodendrocytes.

(A) UMAP of 47,487 oligodendrocyte lineage cells subset and re-clustered. (B) Heatmap depicting the scaled expression of the top 5 enriched genes for each oligodendrocyte lineage cell cluster in Figure 2A. (C) The proportion of each oligodendrocyte lineage cell type in iPSC-derived cultures from all samples. (D) The proportion of each oligodendrocyte lineage cell type in iPSC-derived cultures from HC, RRMS, SPMS, and PPMS lines. (E) Percentage of OPCs in iPSC-derived cultures from HC, RRMS, SPMS, and PPMS lines. The percentage of OPCs in PP cultures is higher than HC cultures. Data presented as mean +/− s.e.m. for n = 4 per group. p-value generated by Welch’s ANOVA with Dunnett’s T3 correction for multiple comparisons against HC. Data are complementary to panel F. (F) Percentage of newly formed oligodendrocytes (nfOLs) or mature oligodendrocytes (mOLs). in iPSC-derived cultures from HC, RRMS, SPMS, and PPMS lines. The percentage of combined nfOLs and mOLs in PP cultures is lower than HC cultures. Data presented as mean +/− s.e.m. for n = 4 per group. p-value generated by Welch’s ANOVA with Dunnett’s T3 correction for multiple comparisons against healthy controls (HC). Data are complementary to panel E. (G) Percentage of O4⁺ early oligodendrocytes in iPSC-derived cultures from HC, RRMS, SPMS, and PPMS lines. The percentage of O4⁺ oligodendrocytes is lower in PP cultures than in HC cultures. Error bars show mean ± standard deviation (n= 3 wells per line for 4–5 lines per group). p-values generated by one-way ANOVA with Dunnett’s correction for multiple comparisons. (H) Representative images of iPSC-derived glial cultures stained for OLIG2 (yellow) and for MBP (teal). Nuclei in grey (DAPI). Scale bar is 50μm. (I) Percentage of MBP⁺ oligodendrocytes in iPSC-derived cultures from HC, RRMS, SPMS, and PPMS lines. The percentage of MBP⁺ oligodendrocytes is lower in PP cultures than in HC cultures. Error bars show mean ± standard deviation (n= 3 wells per line for 4–5 lines per group). p-values generated by one-way ANOVA with Dunnett’s correction for multiple comparisons. (J) Pseudotime plot of oligodendrocyte lineage trajectory from OPCs to mature oligodendrocytes. (K) Cell density plot that shows the distribution of cells across the oligodendrocyte lineage trajectory from OPCs to mature oligodendrocytes for iPSC-derived cells from HC or PPMS. P-value generated with a two-sample Kolmogorov-Smirnov test. (L) Heatmap depicting the scaled expression of genes that were determined to have pseudotime specific expression profiles. Pseudotime gene modules were determined using oligodendrocyte lineage cells from HC lines only. (M) Comparison of pseudotime gene expression profiles between HC and PPMS oligodendrocyte lineage cells. Pseudotime gene modules were determined using oligodendrocyte lineage cells from HC lines only. (N) Percentage of MBP⁺ cells in all cultures treated with vehicle (DMSO) or ketoconazole. Error bars show mean ± standard deviation (n= 17 lines; each data point corresponds to the average of 2 technical replicates per line). p-values generated by two-way paired t-test. (O) Fold-change in the percentage of MBP⁺ cells in HC, RRMS, SPMS, and PPMS cultures treated with either vehicle (DMSO) or ketoconazole. Data is presented as mean +/− standard deviation for n = 3–6 lines per group; each data point corresponds to the average of 2 technical replicates per line. p-values generated by one-way ANOVA with Dunnett’s correction for multiple comparisons. (P) Representative images of HC, RRMS, SPMS, and PPMS cultures treated with either vehicle (DMSO) or ketoconazole. (Q) Expression of MHC Class I genes in HC, RRMS, SPMS, and PPMS cultures. Expression of MHC Class I genes in oligodendrocyte lineage cells (OLCs) and immunological OLCs from MS postmortem brain (PMID: 30747918). p-value generated by Wilcoxon ranked sum test within the Seurat R package. * p < 0.05. (R) Representative immunofluorescence images of oligodendrocytes from HC and PPMS cultures stained for the live markers O4 (green) and HLA-ABC (purple). (S) Quantification of HLA-ABC area within O4 area shows that oligodendrocytes from MS cultures have increased HLA-ABC proteins on their surface. Error bars show mean ± standard deviation (n= 4 HC and 10 MS lines with 3 wells per line). Each datapoint represents the per-well average of 25 fields of view. p-values generated by one-way un-paired t-test.

Figure 3.. scRNAseq reveals a reactive astrocyte subtype enriched in iPSC-derived MS cultures.

(A) UMAP of 37,927 astrocytes subsetted and reclustered. Eight unique astrocyte subclusters were identified. Only cells identified as astrocytes in the initial clustering (Fig 1D light blue) were subsetted and used for reclustering. (B) Heatmap showing the scaled expression of the top three genes enriched in each astrocyte subcluster. (C) UMAP plots of HC, RRMS, SPMS, and PPMS iPSC-derived astrocytes. The Astro.6 cluster is enriched only in iPSC-derived astrocytes from MS and not HC. (D) Distribution of HC, RRMS, SPMS, and PPMS iPSC-derived astrocytes in the astrocyte subcluster 6. (E) Gene ontology analysis of genes significantly increased in astrocytes subcluster 6 compared to all other astrocyte subclusters. (F) UMAP plot overlayed with the expression of HLA-DRA, a gene significantly increased in Astro.6 compared to all other astrocyte clusters. (G) Representative images of RNAscope in situ hybridization for GFAP and HLA-DRA in iPSC-derived cultures from HC and PPMS. Images show localization of HLA-DRA to GFAP⁺ astrocytes; HOECHST in grey. Scale bar, 50μm. (H) Quantification of RNAscope in situ hybridization for HLA-DRA within GFAP⁺ cells in iPSC-derived cultures from HC and PPMS. Error bars show mean ± standard deviation (n= 5 lines per group with 2–3 wells per line). Each datapoint represents the per-well average of 9 fields of view. p-value generated by one-way unpaired t-test. (I) Representative images of healthy control (HC02) and primary progressive (PP04) cultures immunostained for GFAP (green) and HLA-DRA (magenta), with DAPI in blue. Scale bar, 100μm. (J) Quantification of HLA-DRA⁺ immunostaining within the GFAP⁺ population in iPSC-derived cultures from HC and PPMS lines showed increased protein levels in PPMS lines. Error bars show mean ± standard deviation (n= 4 lines per group with 3–5 wells per line). Each datapoint represents the per-well average of 9 fields of view. p-value generated by one-way unpaired t-test. (K) Representative immunofluorescence images of unstimulated and TIC-reactive astrocytes to highlight the morphological changes. Cells are stained with GFAP; scale bar: 50μm. (L) Most representative cells from each MS type and controls. For each model, we ranked the cells by probability of belonging to the class. Here the top scoring ones are presented. (M) Binary prediction area under the curve (auc) of a logistic regression model trained on healthy versus each MS line individually in astrocytes. The error bar represents the standard deviation of the auc between cross-validation folds. The analysis is performed at site-level averages, with 1 site per well held out for training.

Figure 4.. iPSC-derived astrocytes from people with MS mirror astrocytes from MS brains.

(A) UMAP plot of iPSC-astrocytes integrated with astrocytes from post-mortem brains (PMID: 31316211). (B) UMAP plots showing the distribution of cells from healthy control brains and MS brains. Cluster 5 (red circle) is enriched for cells from MS brains. (C) UMAP plots showing the distribution of cells from iPSC-derived astrocytes from HC and MS cultures. Cluster 5 (red circle) is enriched for iPSC-derived astrocytes from MS cultures. (D) Distribution of cells from healthy control brains (yellow), MS brains (red), HC iPSC-derived astrocytes (light blue), and MS iPSC-derived astrocytes (dark blue) in Cluster 5 of the integrated data sets. (E) Dot plot showing the scaled expression of some MHC Class I and Class II genes in each of the clusters from the integrated data sets. (F) Gene ontology analysis of the top 100 genes enriched in Cluster 5 of the integrated data set compared to all other clusters in the integrated data set.