Metformin alters mitochondria-related metabolism and enhances human oligodendrocyte function

Abstract

Metformin rejuvenates adult rat oligodendrocyte progenitor cells (OPCs) allowing more efficient differentiation into oligodendrocytes and improved remyelination, and therefore is of interest as a therapeutic in demyelinating diseases such as multiple sclerosis (MS). Here, we test whether metformin has a similar effect in human stem cell derived-OPCs. We assess how well human monoculture, organoid and chimera model culture systems simulate in vivo adult human oligodendrocytes, finding most close resemblance in the chimera model. Metformin increases myelin proteins and/or sheaths in all models even when human cells remain fetal-like. In the chimera model, metformin leads to increased mitochondrial area both in the human transplanted cells and in the mouse axons with associated increase of mitochondrial function/metabolism transcripts. Human oligodendrocytes from MS brain donors treated pre-mortem with metformin also express similar transcripts. Metformin's brain effect is thus not cell-specific, alters metabolism in part through mitochondrial changes and leads to more myelin production. This bodes well for clinical trials testing metformin for neuroprotection.

Fig. 1. Metformin treatment increases MBP⁺ human embryonic stem cell (hESC)-derived monolayer oligodendroglia.

A Schematic representation of hESC-derived monolayer oligodendroglia differentiation protocol. B Illustrative immunofluorescence image of hESC-derived oligodendroglia at day 70 of in vitro differentiation (DIV) showing PDGFRα⁺ oligodendrocyte precursor cells (OPCs) (magenta) and MBP⁺ oligodendrocytes (grey) co-expressing OLIG2 (cyan). Scale bar = 100 μm. C Fold change (FC) difference of OLIG2⁺MBP⁺ oligodendrocytes after treatment with metformin or clemastine compared to their respective vehicle-treated controls (ddH2O or DMSO). n = 5 for metformin and vehicle control, n = 6 for clemastine and vehicle control where n = number of differentiations with 4 technical repeats. Kolmogorov-Smirnov normality test with Dallal-Wilkinson-Lillie for p-value, two-tailed unpaired t test. Mean ± SEM. D UMAP representation of clustered oligodendroglia. E Dot plot of selected marker genes showing cluster segregation. F Integration with adult human post-mortem spinal cord oligodendroglia snRNAseq dataset. G Cosine similarity heatmap showing similarities between the hESC-derived monolayer oligodendroglia (target, labelled mono_) dataset and the adult human spinal cord oligodendroglia (source) dataset. H Integration with the second and third trimester foetal OPC snRNAseq dataset. Source data are provided in the Source Data file. Created in BioRender. Swire, M. (2025) https://BioRender.com/j633z18.

Fig. 2. Metformin treatment increases MBP⁺ human embryonic stem cell (hESC)-derived cortical brain organoid oligodendroglia.

A Schematic representation of hESC-derived cortical brain organoid oligodendroglia differentiation protocol. B Illustrative immunofluorescence images of hESC-derived oligodendroglia in organoids at day 67 of in vitro differentiation (DIV), showing PDGFRα⁺ oligodendrocyte precursor cells (OPCs) (red) co-expressing OLIG2 (cyan) and of day 112 showing MBP⁺ oligodendrocytes (yellow) co-expressing OLIG2 (cyan). Scale bars = 100 μm or 50 μm (zoom). C MBP⁺ area of immunofluorescence per total organoid area after treatment with metformin compared to vehicle-treated controls (ddH2O). n = 5 biological replicates. Kolmogorov-Smirnov normality test with Dallal-Wilkinson-Lillie for p-value, two-tailed unpairedt test. Mean ± SEM. D UMAP representation of clustered organoid-derived oligodendroglia integrated with scRNAseq organoid data from (Marton et al.). E Dot plot of selected marker genes showing cluster segregation. F Integration with adult human post-mortem cortical hemisphere oligodendroglia snRNAseq dataset. G Cosine similarity heatmap showing similarities between the hESC-derived brain organoid oligodendroglia (target, labelled with co_) dataset and the adult human cortical oligodendroglia (source) dataset. Source data are provided in the Source Data file. Created in BioRender. Swire, M. (2025) https://BioRender.com/cl7w2km.

Fig. 3. Metformin treatment increases the percentage of myelinated axons in chimeric mice.

A Schematic representation of green fluorescent protein-positive (GFP⁺) human embryonic stem cell (hESC)-derived PDGFRA⁺ oligodendrocyte precursor cells (OPCs) transplantation into the corpus callosum of Rag2^-/-: Shi/Shi P2-P4 mice. B Illustrative immunofluorescence and electron micrograph (EM) images of chimeric corpus callosum at 70 days post transplantation showing HuNu⁺ cells (cyan)(left) or MBP⁺ oligodendrocytes (cyan) co-localising with OLIG2 (magenta)(middle) and myelinated rodent axons (right). Scale bar = 100 μm (fluorescence) or 5 μm (EM). For control untreated mice, a mean of 16.15% ± 1.88 SEM rodent axons were myelinated. n = 11 animals. C Quantification of myelinated axons showed a significant increase after metformin treatment compared to the vehicle-treated (ddH₂O) controls. Each dot represents an animal, n = 5 animals. Kolmogorov-Smirnov normality test with Dallal-Wilkinson-Lillie for p-value, two-tailed unpaired t test. Mean ± SEM. D UMAP representation of clustered oligodendroglia. E Dot plot of selected marker genes showing cluster segregation. F Integration with adult human post-mortem oligodendroglia snRNAseq dataset (brain and spinal cord). G Cosine similarity heatmap showing similarities between the hESC-derived chimeric oligodendroglia (target, labelled with chi_) dataset and the adult human oligodendroglia (source) dataset. Source data are provided in the Source Data file. Created in BioRender. Swire, M. (2025) https://BioRender.com/ghxthad.

Fig. 4. Metformin treatment increases myelination and alters mitochondria.

A Electron microscopy (EM) of myelinated chimeric animal axons showing a significant increase of mitochondrial area after metformin treatment compared to vehicle-treated controls, quantified (B) as the average of myelinated axons measured per mouse or number per individual axon measured to indicate the spread. n = 5 metformin- and n = 4 vehicle-treated animals. Kolmogorov-Smirnov normality test with Dallal-Wilkinson-Lillie for p-value, two-tailed unpaired t test. Mean ± SEM. Boxes in plots visualise median and 25th and 75th percentiles, and whiskers mark range up to 1.5 * inter-quartile ranges to show potential outliers. See Supplementary Data S11 for exact summary statistics. Scale bar = 1 μm. C EM of chimeric animal corpus callosum showing glia with increased area of mitochondria after metformin treatment compared to vehicle, quantified (D) as average of glia measured per mouse or number per individual glial cytoplasm area measured to indicate the spread. Kolmogorov-Smirnov normality test with Dallal-Wilkinson-Lillie for p-value, two-tailed unpaired t test. n = 6 metformin- and n = 3 vehicle-treated animals. Mean ± SEM. Boxes in plots visualise median and 25th and 75th percentiles, and whiskers mark range up to 1.5 * inter-quartile ranges to show potential outliers. See Supplementary Data S11 for exact summary statistics. Scale bar = 1 μm. E Gene Ontology (GO) analysis of differentially expressed genes between human embryonic stem cell (hESC)-derived oligodendroglia from metformin- and vehicle-treated chimeric animals. F Volcano plot of significantly differentially expressed genes between hESC-derived oligodendroglia from metformin- and vehicle-treated chimeric animals. pval < 0.05 and logFC > 0.5. DGE analysis was performed using MAST within Seurat, a two-sided statistical test that accounts for detection rate and zero inflation typical of single-cell RNA-seq data. Latent variables to correct for batch effects were included. P-values were adjusted for multiple comparisons using the Benjamini-Hochberg false discovery rate (FDR) method. G Violin plot of the most highly expressed specific genes in hESC-derived oligodendrocytes in metformin-treated chimeric animals. H Quantification of in situ hybridisation using NDUFA11 and EIF1 RNA probes and immunofluorescence for OLIG2 on human cells in chimera tissue with and without metformin treatment. n = 3 metformin- and 3 vehicle-treated animals with 2 ROIs of high-density HuNu⁺ nuclei analysed per animal. Composite score = (fraction of OLIG2⁺ cells) x (mean number of RNA probe puncta). Kolmogorov-Smirnov normality test with Dallal-Wilkinson-Lillie for p-value, two-tailed unpaired ttest. Mean ± SEM. I Western blot showing TOMM20 and CHCHD2 levels, compared to housekeeping B-TUBULIN in hESC-derived oligodendroglial monocultures treated for 7 days with metformin versus vehicle, with quantification. Mean ± SEM, two-tailed paired t test. M = metformin treated, C = vehicle treated controls. n = 3 separate differentiations. J Violin plot showing expression of EIF1 in multiple sclerosis (MS) donor oligodendrocytes with and without metformin (met) treatment versus controls (CTR). Source data are provided in the Source Data file.