Dynamic regulation of genes during oligodendrocytes lineage progression.

A) Schematic diagram showing magnetic-activated cell sorting (MACS)-based isolation of oligodendrocyte progenitor cells (OPCs), pre-oligodendrocytes (PreOLs), and oligodendrocytes (OLs) from young rat brains (2–3 months). The schema illustrates the order in which these cell types were most effectively isolated: mature OLs were isolated first using the OL-specific marker MOG, followed by OPCs using A2B5, and finally PreOLs using O4. B) Principal component analysis (PCA) (left) and average-linkage hierarchical clustering (right) based on global gene expression show that each cell type forms a distinct cluster. OPCs, PreOLs, and OLs were isolated from four different rats. C) Using gSWITCH, we identified four distinct patterns of switch genes during OL lineage progression. Each coloured line represents one gene and connects its log2fold-change values across the three oligodendrocyte lineage states: OPC, PreOL and OL. The connecting lines are used to visualize gene-wise patterns of expression change across these cell states. (B-C) GEO accession number of RNA-seq data: GSE301578. See also Supplementary Table 1, Supplementary Figure 1, Supplementary text, Supplementary Figure 3A.

Down-regulation of switch genes due to ageing.

A) Principal component analysis (PCA) shows that aged and young OPCs are distinguishable based on global gene expression (meta-analysis of RNA-seq datasets, GEO accession number: GSE303317). B) Volcano plot showing differentially expressed genes in aged OPCs compared to young OPCs. n= 6 (aged rats), 7 (young rats), FC: fold change; Padj: adjusted p-value (Benjamini–Hochberg FDR after Wald test). C) Venn diagram (top) showing the subset of Type 3 and Type 4 switch genes affected by ageing. Their expression patterns are shown in the heat map (bottom), where we observed down-regulation with ageing. * P < 0.05, ** P < 0.001, Chi-square test (two sided) with Yates’ correction. DE: Differentially expressed. DE gene list is accessible through GEO accession: GSE303317. See also Supplementary Table 1, Supplementary Figure 3B.

Transcription factor Bcl11a-mediated network potentially affected duuring ageing.

A) Transcription factor binding site enrichment analysis using Type 3 and Type 4 switch genes as input identified 27 transcription factors (TFs) detected in both ENCODE and ReMap2, which are ChIP-seq–based experimental databases. Among these 27 TFs, we found 5 TFs that are themselves present as Type 3 or Type 4 switches (TF switches extracted from gSWITCH). ** P < 0.001, hypergeometric test. B) Heat map (top) showing expression (RNA-seq, GEO accession number: GSE303317) of these 5 TFs. Their differential expression in aged OPCs relative to young OPCs plotted as log2FC (bottom). FC: Fold change, n= 6 (aged rats), 7 (young rats), mean±SEM, * adjusted p-value < 0.05, ** adjusted p-value < 1×10⁻⁶, Benjamini–Hochberg FDR after Wald test. Note: All five TFs are classified as Type 3 switches, and Bcl11a shows the strongest repression in aged OPCs, with the highest statistical significance. C) Bcl11a, as a Type 3 TF switch, shows significantly repressed expression as OL lineage progresses. DESeq2-normalised values relative to OPCs plotted. n= 4 different rats, mean±SEM, ** adjusted p-value < 1×10⁻⁶, Benjamini–Hochberg FDR after Wald test. GEO accession number of RNA-seq data: GSE301578. See also Supplementary Table 2, Supplementary Table 3, Supplementary Figure 4.

Effect of Bcl11a inhibition on OPCs proliferation and differentiation.

A) Reverse transcription (RT)–qPCR analysis of Bcl11a expression in OPCs transfected with siRNA targeting Bcl11a. siBcl11a: siGENOME Rat Bcl11a; siControl: siGENOME non-targeting control siRNA (Dharmacon™). B) Effect on proliferation. Left: Immunofluorescence analysis using antibodies to OLIG2 and EdU detection (Click-iT protocol), followed by EdU proliferation assay. Right: Quantification of EdU⁺OLIG2⁺ cells plotted as a percentage of OLIG2⁺ cells. P: p-value, Student’s t test (unpaired and two tailed) C) Effect on differentiation. Left: Immunofluorescence analysis using antibody to MBP, OLIG2 upon inhibition of Bcl11a in OPCs. Right: quantification of MBP+ cells were plotted as a percentage of OLIG2+ cells. (A-C) n = 3 independent experiments (each time 4 P7 brains were pooled for OPC isolation). mean±SEM. ** p<0.01, Student’s t test (unpaired, two tailed); The symbol “#” indicates cell count. (B-C) scale bar: 22 µm.

Effect of Bcl11a transient over-expression on proliferation and differentiation of aged OPCs.

A) Effect on proliferation. Left: Immunofluorescence analysis using antibodies to OLIG2 and EdU detection (Click-iT protocol), followed by EdU proliferation assay. Right: Quantification of EdU⁺OLIG2⁺ cells plotted as a percentage of OLIG2⁺ cells. P: p-value, Student’s t test (unpaired, two tailed). B) Effect on differentiation. Left: Immunofluorescence analysis using antibodies to MBP and OLIG2 after transfection of Bcl11a mRNA (mBcl11a) or Gfp mRNA (mGfp; control) in aged OPCs. Right: Quantification of MBP⁺ cells plotted as a percentage of OLIG2⁺ cells. * P<0.05, Student’s t test (unpaired, two tailed). (A-B) n = 3 independent experiments (each using OPCs isolated from pooled brains of 2 aged rats). mean ± SEM. The symbol “#” indicates cell count. scale bar: 22 µm.

OL lineage specific Bcl11a expression in aged mice.

A) Left: SOX10-driven construct carrying either Gfp alone (control) or Bcl11a linked to Gfp via a self-cleaving peptide element (P2A). This construct ensures SOX10-driven expression of both Bcl11a and Gfp. P2A: Porcine teschovirus-1 2A peptide, bGH p(A): bovine growth hormone poly(A) signal. The AAV containing this construct was delivered to 18-month-old mice via tail-vein injection. Right: schematic showing the experimental timeline. B) RNA in situ analysis of Plp1. In the lesioned area, the number of Plp1⁺ cells per mm² was counted and plotted. n=5-6 mice, mean±SEM, ** P<0.01, Student’s t test (unpaired, two tailed). scale bar: 100 µm. The white dotted line demarcates the lesion area.

Test for evolutionary selection on Bcl11a.

Normalised expression values of marker genes (for astrocytes, progenitor-state marker, OPC, PreOL, and OL) from RNA-seq data (GEO accession number: GSE301578) were plotted relative to OPCs.

n=4 (young rats), mean±SEM, * adjusted P<0.05, Benjamini–Hochberg FDR after Wald test. See also Supplementary text.

Schematic representation of the logic of the gSWITCH web application (for detail see ‘Materials and Methods’ section).

The gSWITCH application is accessible at https://altoslabs.shinyapps.io/gSWITCH/. Red circles indicate the start and the end state of a series (time or lineage stages or any other variable dependent series). GLM fit and hypothesis testing followed by contrast analysis between start and the end state, generates a subset of genes on which we apply maximum (max) or minimum (min) criteria within a closed interval while taking account of monotonicity trend as well as further specific contrasts between groups. This allows identification of four types of switches. Detail criteria are shown inside the boxes and the trend pattern are represented by coloured dashed-straight lines. Arc or solid curve lines indicate statistical significance for specific contrast. Adj p value: Adjusted p value [Benjamini–Hochberg FDR after LRT or Benjamini–Hochberg FDR after Wald test for specific contrast analysis]. u.d.: user defined value. Note that to run gSWITCH, RNA-seq data are required (bulk or scRNA-seq flattened to pseudo bulk), where the number of series or time points or group ≥3 and the replicates per time point or group≥2. gSWITCH also identifies transcription factors (TFs) within the switches and provides a summary of the total number of switches and TFs per switch type.

Gene ontology (GO) analysis.

A) Top over-represented GO terms for each switch type shown in Figure 1C. B) Top over-represented GO terms for the 135 overlapping genes shown in Figure 2C, defined as genes that belong to Type 3 switches and are also affected by ageing. Padj: Benjamini–Hochberg adjusted P value. The dotted blue line indicates the significance threshold of Padj = 0.05, corresponding to −log10(Padj) = 1.3. Please note that only 8 Type 4 genes overlapped with differentially expressed genes in ageing. Due to this small number, we could not identify any significant GO-term enrichment, and therefore this was not plotted.

Bcl11a expression in OPCs, OLs and aged OPCs, as determined by RT-qPCR (reverse transcription followed by qPCR).

A) Expression levels of Bcl11a, Pdgfra as an OPC marker, and Mbp as an OL marker were measured in OPCs and OLs and plotted relative to OPCs. B) Expression levels of Bcl11a and Pdgfra were measured in young and aged OPCs and plotted relative to aged OPCs. (A-B) n=3 rats, mean±SEM, * Bonferroni adjusted P<0.05, ** Bonferroni adjusted P<0.01, Student’s t test (unpaired, two tailed).

Fidelity of protein expression from in vitro–transcribed Bcl11a mRNA; mRNA transfection efficiency in aged OPCs; and fidelity of AAV construct expression in the OL lineage and infection efficiency in aged mouse brain.

A) Western blot analysis using antibodies to BCL11A and β-actin (ACTB), with ACTB serving as a loading control, in MCF7 cells transfected with Bcl11a mRNA or Gfp mRNA (control). B) Immunofluorescence analysis using antibodies to OLIG2 and GFP in aged OPCs transfected with Gfp mRNA. Quantification of GFP+OLIG2+ cells were plotted as percentage of OLIG2+ cells. n = 3 independent experiments (each using OPCs isolated from pooled brains of 2 aged rats). mean ± SEM. Scale bar: 22 µm. C) AAV carrying constructs are shown in Figure 6A. Left: Immunofluorescence analysis of OLIG2 and GFP immunostaining. Right top: Quantification of OLIG2+GFP+ cells were plotted as a percentage of GFP+DAPI+ cells. Right bottom: quantification of OLIG2+GFP+ cells were plotted as a percentage to OLIG2+ cells. n= 3 mice, mean+SEM. P: p-value, Student’s t test (unpaired, two tailed). Scale bar: 20 µm.

Immunofluorescence analysis using antibody to OLIG2 after OL lineage specific Bcl11a expression in aged mice.

The AAV containing the Sox10-driven construct carrying either Gfp alone (control) or Bcl11a linked to Gfp via a self-cleaving peptide element (P2A) was delivered by tail-vein injection (see Figure 6A). In the lesioned area, the number of OLIG2+ cells per mm2 was counted and plotted. In the same region, the number of Plp1⁻ OLIG2⁺ cells per mm² was also calculated and plotted (see Figure 6B for Plp1⁺ cells per mm²). n=5-6 mice, mean±SEM, P: p-value, ** P<0.01, Student’s t test (unpaired, two tailed). scale bar: 100 µm. The white dotted line demarcates the lesion area.

27 Transcription factors (TFs) identified by both Encode and Remap2.

Bcl11a DNA and protein sequence accession numbers for 23 vertebrate species used in this study.