Extended Data Fig. 5: Transitional Th cells from MS-affected twins respond with increased IL-2 expression upon ex vivo reactivation.

a, UMAP showing CITE-seq data of Th cells (78 531 cells) derived from 8 twin pairs discordant for MS (left panel). Color coding indicates cell type yielded by clustering using the transcriptome. Heatmap showing the expression profile of the top differentially expressed genes across the different clusters (middle panel). UMAP showing the expression overlay for surface CD45RA and CD45R0 and FOXP3 and CX3CR1 transcripts. b, Lollipop plot showing p-values for increase in CD25 surface epitope expression between MS-affected twins and unaffected twins for the indicated clusters determined by CITE-seq (45 146 cells). c, UMAP of CITE-seq and CyTOF data with an overlay showing the transitional Th cell nodes that demonstrated a differential CD25 expression in MS-affected versus unaffected twins. d, Diffusion map Th cells (78 531 cells) with pseudotime overlay (left panel). Line plot (right panel) showing the indicated surface markers expression across the computed pseudotime. Horizontal lines indicate mean pseudotime for Tna 4 and Tna 5 respectively. e, UMAP of CITE-seq data (78 531 cells) showing expanded cells (orange; > 2 TCR sequences in same patient) and non-expanded cells (grey). f, Heatmap and violin plots of selected genes within the IL2RA signaling pathway in Tna 4 and Tna 5 (11 343 cells) for MS-affected and unaffected twins. g, Volcano plots showing the differentially expressed genes of the indicated clusters (16 439 total cells) of MS-affected twins compared to unaffected twin siblings. The top 20 significant genes were annotated. Red color indicates increased and green color decreased gene expression in MS-affected twins. h, Network visualization (top left panel) of unsupervised FlowSOM clusters (beige); highlighted are MS twin differential nodes for which CD25 and IL-2 expression appeared significantly (red) and nodes of the manually-annotated reference framework (green). Dot size corresponds to population frequency among total leukocytes. Heatmap (bottom panel) displays marker expression profile of differential state naïve Th cell node and manually-annotated reference nodes in phenotypic proximity. Violin plots (right panel) showing the median CD25 and IL-2 expression levels in cells within the differential state naïve Th cell nodes in untreated twin pairs (n = 6). i, 2D lollipop plot showing the median expression of CD25 and IL-2 for unaffected twin siblings (green) and MS-affected twin siblings (red; n = 25). Twin pairs are indicated by solid line. j, Violin plots showing the inter-twin-pair difference in median IL-2 expression level in the differential state naïve Th cell node, separated according to disease stage of the MS-affected twin. RRMS = relapsing-remitting MS (n = 21); SPMS = secondary progressive MS (n = 4). k, Biaxial plot showing the IL-2 and CD4 expression levels for cells in the differential state naïve Th cell node. The dashed line indicates the gate to determine IL-2-positive cells. l, m, Effect size for the twin-pair difference of indicated cytokines in IL-2 expressing cells in the MS twin differential state node for untreated twin pairs (n = 6; l, left panel). Violin plots showing the median IL-17A and IL- 3 (l, right panel) or IL-10 and IL-9 (m) expression level in IL-2-expressing cells within the differential state transitional Th cell node for MS-unaffected twins and untreated MS-affected twins. n, Heatmap displaying the transcription factor regulon activity for STAT4 and RELA in Tna 4 and Tna 5 (11 343 cells) determined by CITE-seq for individual twins. o, Lollipop plot showing p-values for increase in VLA4 surface epitope expression between MS-affected and unaffected twins for the indicated clusters (75 856 cells) determined by CITE-seq. Violin plots contain a bold horizontal line depicting the respective group mean and dashed line indicates twinship. If not indicated, the differences between experimental groups were statistically not significant (p > 5%) using an two-sided unpaired nonparametric Mann-Whitney-Wilcoxon test (j) or a two-sided paired non-parametric Wilcoxon signed-rank test (h, l, m) with false discovery correction according to the Benjamini-Hochberg approach or a logistic regression model using the twinship as a latent variable and comparing each model to a null model with a likelihood ratio test and applying a Bonferroni correction (b, g, o).