Where the Metabolite Enters Single-Cell MS Studies
Single-cell sequencing in multiple sclerosis (MS) produces transcript counts, and a metabolite is not a transcript. The gap between them is usually crossed by pathway enrichment, which tells you that a cell is equipped for a metabolic programme without telling you what that cell is making or consuming. Three recent studies all put single-cell data at the centre of a metabolic claim about a named cell type in MS, and what separates them is where the metabolite enters the argument. One infers it from transcripts in human lesion tissue and then tests two genes. One infers it from pathway enrichment and writes down plainly that the operative molecule is still unknown. One measures it in serum in living people and uses the sequencing to read what happened next.
A Lipid Programme Named, Then Tested With Two Genes
Feng and colleagues built a single-cell molecular and spatial atlas of chronic active MS lesions, pairing single-nucleus RNA sequencing with MERFISH, an imaging method that counts selected transcripts inside intact tissue. Within lesion rims they found CD8 T cell niches sitting beside inflamed microglia that carried an interferon response and raised lipid metabolism. The lipid half of that description is a transcript signature, and MERFISH adds coordinates rather than molecules, so what the atlas delivers is a cell type, a state and a position. The test came next and changed species. Deleting the cholesterol efflux transporters ABCA1 and ABCG1 in the microglia of mice with experimental autoimmune encephalomyelitis raised the number of lipid-storing phagocytes and amplified inflammation, while targeting sterol metabolism pharmacologically reduced foam cell formation and inflammatory demyelination. Two transporters are a narrow handle on a programme the sequencing could only name in aggregate.
Positioning a Gene Along a Differentiation Trajectory
Xiao and colleagues came at the problem from human genetics first. Patients with 6-pyruvoyl-tetrahydropterin synthase deficiency show high-intensity signal in cerebral white matter and widespread delayed myelination, which is what pointed the group at the gene; its product is the enzyme for de novo tetrahydrobiopterin synthesis. In brain sections from seven MS donors and five matched controls, PTS immunoreactivity rose in Olig1-positive cells in normal-appearing white matter, in a punctate rather than an even pattern. Re-analysis of public human and mouse single-cell datasets then mapped the gene onto the oligodendrocyte lineage: PTS is low in precursor cells under normal conditions and climbs steadily through committed progenitors, newly formed and myelin-forming oligodendrocytes, peaking at immune oligodendroglia in the human data. In precursors specifically, PTS expression tracked with downregulated proliferation pathways.
A Pathway Switched On, With the Metabolite Left Open
Deleting Pts in precursor cells under a Pdgfra driver lessened disease severity in the mouse model, while the same deletion in mature oligodendrocytes did nothing, which is a useful piece of cell-type resolution. Immune infiltration, cytokine levels and astrocyte and microglial markers were all unchanged, so the benefit was not immune. Bulk sequencing of the cord gave 496 downregulated and 386 upregulated genes, with cell cycle terms up and steroid and terpenoid backbone biosynthesis enriched, driven by HMGCS1 and CYP51. Knocking down HMGCS1 abolished the proliferation gain, and atorvastatin cut proliferation genes. Then the authors state their own open question: whether the effect is attributable to cholesterol itself, to an intermediate metabolite, or to one of the enzymes. Earlier work had also found the astrocytic PTS effect running independently of tetrahydrobiopterin synthesis, so even the enzyme's nominal product is unsettled.
Measuring the Metabolite in People, Then Sequencing the Response
Falk and colleagues ran the one design here in which somebody measured a metabolite in a living person. Eleven relapsing-remitting MS patients and seven healthy donors received four weeks of narrowband ultraviolet B, followed by single-cell transcriptomics of immune cells from skin, blood and cerebrospinal fluid, giving 384,399 cells after quality control across 48 lineage subclusters. Serum 25-hydroxyvitamin D was measured before and after. Healthy donors went from a mean of about 21 to 37 ng/mL, a gain near 16, while patients went from about 24 to 33, a gain near 9. So the molecule was measured, in people, and the two groups differed in how much of it they gained. The VDR pathway came up in CD8 and CD4 T cells and myeloid cells in skin, and across all lineages in blood.
Where the Single-Cell Signal Stopped Carrying
In skin, irradiation lowered CD8 T cell proportions (P < 0.001) and raised myeloid and natural killer cells. Classical monocytes rose in patients (P = 0.00098) and intermediate monocytes went from 1 to 9 per cent (P = 0.002) as macrophages fell from 27 to 13 per cent (P = 0.002). Th17 and Th22 proportions rose in healthy donors but fell in patients, Th22 from 9 to 4 per cent (P = 0.02). In blood, B cells fell from 13 to 10 per cent (P = 0.02), mostly naive cells. The authors read the pattern as a blunted translation from skin to blood in MS, with exaggerated cutaneous activation and stress-response genes NR4A1, JUN and HSPA1B in patients set against homeostatic EEF1A1, TPT1 and TMSB4X in donors. The cerebrospinal fluid arm rests on ten samples and is reported as preliminary.
Three Entry Points for One Kind of Molecule
Lined up, the three studies put the metabolite in three different positions relative to the sequencing. Feng infers it from transcripts in human lesion tissue and converts the inference into two mouse genes. Xiao infers it from pathway enrichment, perturbs the pathway, and records that the operative molecule remains open. Falk measures it in serum in living people and uses the single-cell data to read the response, which is the closest of the three to having a metabolite and a transcriptome in one experiment, though serum and the sequenced cells are still separate compartments. The common thread for a geneticist is that all three finished with named genes: ABCA1 and ABCG1, PTS with HMGCS1 and CYP51, and VDR. Those are testable. In two of the three, the molecules they move are still inferred.
Disclaimer: This blog post is based on the cited studies and is intended for informational purposes only. It is not intended to provide medical advice. Please consult with a healthcare professional for any health concerns.
References:
Feng, R., Spieth, L., Liu, L., et al., Groh, J., & Simons, M. (2025). Single-cell spatial transcriptomic profiling defines a pathogenic inflammatory niche in chronic active multiple sclerosis lesions. Immunity, 58(12), 2989-3005.e10. https://doi.org/10.1016/j.immuni.2025.10.003
Xiao, H., Long, Y., Guo, Y.-S., et al., Gao, J., & Zhou, J.-W. (2025). Suppression of 6-pyruvoyl-tetrahydropterin synthase promotes remyelination in multiple sclerosis. BMC Medicine, 23(1), 618. https://doi.org/10.1186/s12916-025-04351-0
Falk, S., Krämer, J., Leson, S., et al., Wiendl, H., & Schwab, N. (2026). Single-cell analysis of skin and blood reveals systemic immune responses to ultraviolet B and their impairment in MS. Journal of Neuroinflammation, 23(1). https://doi.org/10.1186/s12974-026-03887-z
