Glucose Uptake and ATP Synthesis in MS T Cells

The hardest problem in multiple sclerosis (MS) immunology is telling apart what the disease does from what the patient's genome was always going to do. Kavaka and colleagues solved it by design rather than by statistics, studying monozygotic twins discordant for MS. Identical genomes, shared early environment, one twin affected and one not: any difference between them cannot be genetic. Twelve pairs, 24 individuals, drawn from the MS TWIN STUDY at LMU Munich and recruited between 2012 and 2018. The cotwins then split in a way that turns this into a study of timing rather than of disease. Six had no sign of neuroinflammation. Six had subclinical neuroinflammation, meaning MRI or cerebrospinal fluid findings without meeting the clinical and radiological criteria for MS, and a cotwin in this cohort carries up to a 25% lifetime risk of developing it. From these people the team sequenced 193,771 peripheral CD8⁺ T cells with paired T cell receptor data, plus cerebrospinal fluid from four of the pairs.

The First Result Is a Negative, and It Shapes Everything After
Twelve CD8⁺ T cell clusters emerged, running from a naive-like population marked by CCR7, LEF1, TCF7 and SELL, through transitional states, to effector-like clusters carrying CX3CR1, GNLY, GZMH, FGFBP2 and PRF1, alongside natural killer-like and mucosal-associated invariant T cell populations. Analysis of how those clusters were distributed revealed no significant differences across clinical conditions. The fraction of clonally expanded cells per cluster did not differ between affected and healthy twins either. MS, in other words, is not marked by a different mix of CD8 subsets in the blood. Whatever separates these people has to lie in what the cells are doing rather than in which cells are present, and the rest of the paper is an attempt to show that at the level of expression.

Narrowing to the Clones That Actually Reach the Brain
Matching T cell receptor sequences between blood and cerebrospinal fluid in four twin pairs showed up to 21% of clonotypes shared between compartments, concentrated in the activation and effector clusters rather than spread evenly. Cluster 8_CD74 carried both the highest degree of overlap and the highest fraction of expanded clones in the cerebrospinal fluid. The obvious artifact was checked rather than assumed away: mucosal-associated invariant T cells share an invariant receptor chain and would appear matched for trivial reasons, but their blood-to-fluid overlap came in below 5%, so they were set aside. This step is what gives the expression analysis its meaning, since it restricts attention to clones demonstrably present in both compartments.

The Gradient Runs Healthy, Then Subclinical, Then MS
Here is the central observation. Cells scoring highest on the immunological signature were few in healthy cotwins, moderate in cotwins with subclinical neuroinflammation, and markedly more numerous in the twins with MS. The content differs by stage as well as the quantity. IL2RB was predominantly up-regulated in the subclinical subgroup, consistent with early CD8 activation beginning before any clinical disease, while the MS twins additionally up-regulated proteasomal degradation (PSMB7, PSMB8, PSMB9, PSME2) and HLA class II antigen presentation (CD74, HLA-DRA, HLA-DRB5). Across both affected groups the modules covered activation (COTL1, ANXA5, CD27, CD69), receptor signaling (LIME1, JAK3, MAP4K1, RRAGA, LAMTOR2, LAMTOR4), cytotoxicity (GZMH, GZMK, KLRB1, KLRD1), secretion (VAMP5, VAMP8), effector output (CCL5, FLT3LG, IL32) and migration (SELL, CXCR3).

The Metabolic Half, and Why It Points at Treatment
Oxidative phosphorylation came out as the single most enriched pathway in both the subclinical and the MS subgroups. The MS twins raised glucose uptake through SLC2A3, the transporter also known as GLUT3, along with glycolysis (TPI1, GAPDH, LDHB), the citric acid cycle (MDH2, IDH2) and electron transport complexes I, III and IV together with ATP synthase. Antioxidant machinery rose alongside, with GPX4, SOD1, GLRX and PRDX5 up-regulated, which is what a cell running more oxidative phosphorylation has to do about the reactive oxygen species it generates. GPX4 was already among the most highly expressed genes in the subclinical group. The cluster with the highest metabolic enrichment score was 8_CD74, the same cluster carrying the most blood-to-fluid clonal sharing. The therapeutic consequence is specific rather than hand-waving: inhibitors of complexes I and III and of ATP synthase can completely block CD8⁺ T cell activation.

What the Twin Design Controls, and What Was Checked Against It
The decisive methodological step is the pairwise comparison within twin pairs, six pairs in each contrast, holding genetic background constant by construction. Most markers remained up-regulated under it, which is the basis for saying the enhanced activation, migration and effector capacity belongs to the disease process rather than to MS susceptibility genetics. Several other layers sit around it. Interferon-β-treated patients were excluded from the differential expression analysis because their interferon signatures otherwise dominated. An independent cohort of 12 treatment-naive patients at first MS manifestation, against five people with idiopathic intracranial hypertension as non-inflammatory controls, reproduced the immunological and metabolic signatures across 89,859 cells, with blood-to-fluid clonotype sharing reaching almost 50% in cluster 8_CD74. Two public brain single-nucleus datasets then showed T cells inside MS lesions expressing the same modules, and a further negative result: killer cell immunoglobulin-like receptor transcripts, which mark a suppressive CD8 population in autoimmune disease, were absent from both datasets.

What the Study Says It Cannot Settle
The limitations are stated without hedging. The cohort is small, particularly among the twin pairs, which limits statistical power and may hide subtler immunological changes. Only CD8⁺ T cells were examined, so nothing here speaks to the other immune populations involved in MS. Transcript levels need not track protein levels or activity, given post-transcriptional regulation. And although sorting targeted antigen-experienced T cells, naive T cells still turned up in the data, which the authors attribute to low or transient CD45RO expression, CCR7 expression, and the technical limits of the sorting itself. One further constraint follows from the design rather than appearing in that list. The six cotwins with subclinical neuroinflammation may or may not go on to develop MS, and every sample here was collected at a single time point, so the healthy-to-subclinical-to-MS gradient is assembled from three groups of different people rather than observed in anyone passing through it. Following those six cotwins forward is the study that would convert this sequence into a trajectory.

Disclaimer: This blog post is based on the cited research article 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.

Reference:
Kavaka, V., Mutschler, L., de la Rosa del Val, C., Eglseer, K., Gómez Martínez, A. M., Flierl-Hecht, A., Ertl-Wagner, B., Keeser, D., Mortazavi, M., Seelos, K., Zimmermann, H., Haas, J., Wildemann, B., Kümpfel, T., Dornmair, K., Korn, T., Hohlfeld, R., Kerschensteiner, M., Gerdes, L. A., & Beltrán, E. (2024). Twin study identifies early immunological and metabolic dysregulation of CD8⁺ T cells in multiple sclerosis. Science Immunology, 9(99), eadj8094. https://doi.org/10.1126/sciimmunol.adj8094