MS T Cells Carry More Cholesterol and Fewer Glycosphingolipids

Martin-Gutierrez and colleagues asked what lipid metabolism is doing inside CD4⁺ T cells in relapsing-remitting multiple sclerosis (RRMS), and built the study around one design decision that makes the answer interpretable: every patient was sampled before starting their first disease-modifying therapy. That matters because those drugs alter lipid metabolism directly, which is the very thing being measured. The work combines RNA sequencing of sorted CD4⁺ T cells, flow cytometry of plasma membrane lipids, nuclear magnetic resonance metabolomics of serum, and functional assays using an agonist of the liver X receptors, the lipid-activated nuclear receptors that govern cholesterol handling in immune cells. One number to carry throughout: the cohort size changes with the experiment, from ten against ten for sequencing down to three donors for the functional work.

The Transcriptome, and the Paradox at Its Centre
Sequencing returned 3,940 differentially expressed genes, 1,883 up and 2,057 down. Cytokine signalling led the upregulated pathways, with IKBKG at 6.92-fold (P = 7.43×10⁻⁷⁴) alongside JAK3, TYK2 and IL2RG. Metabolism of lipids also ranked among the top upregulated pathways, carrying FABP5 at 25.61-fold and NR1H2, which encodes LXRβ, at 2.95-fold. Yet signalling by nuclear receptors sat among the most strongly downregulated pathways. That is the paradox the paper is built on: the receptor rises while the pathway it drives falls. Of 66 LXR target genes established in healthy CD4⁺ T cells, 40 were down in patients and 26 up. IDOL, which governs cholesterol uptake by degrading the LDL receptor, fell 2.3-fold, and LDLR itself fell with it, while the classic targets ABCA1 and ABCG1 did not move at all. The authors state the consequence plainly, that the pattern did not resemble a global change in LXR activity, and they report LXR binding motifs enriched among the downregulated genes.

The Membrane Result, and What It Is Not Explained By
Glycosphingolipids were reduced in patient CD4⁺ T cells (P = 0.0264) and cholesterol raised (P = 0.0265), across responder, activated and regulatory subsets alike. The ratio between them fell (P = 0.009) and membrane lipid order, a measure of fluidity that shapes immune signalling, rose (P = 0.027). Two checks lift this above a composition curiosity. The changes were not attributable to differences in CD4⁺ T cell proportion or subset distribution. And the relationship between the two lipids inverts: in healthy donors they are uncorrelated (r = 0.101, P = 0.75), while in patients they correlate negatively (r = −0.620, P = 0.035). Neither lipid tracked with age, body mass index, vitamin D level, disability score or relapse count, all P > 0.23. The membrane changes were also absent in patients with secondary progressive disease, which the authors read as a possible difference in mechanism between disease stages.

Where the Lipid Changes Come From
Two contributors were separated experimentally rather than inferred. Culturing healthy donor T cells for 48 hours in serum from RRMS patients shifted both glycosphingolipid and cholesterol levels. Adding T-cell receptor stimulation on top of patient serum raised cholesterol further but added nothing for glycosphingolipids, while receptor stimulation alone without serum raised both (P = 0.0149 and P = 0.0469). The active ingredient in the serum appears to be lipid rather than inflammatory: serum cytokines did not differ between patients and controls, whereas serum lipid metabolites, measured by nuclear magnetic resonance in 30 controls and 20 patients, were significantly altered and correlated with the differentially expressed genes of the lipid metabolism pathway.

The Rescue Is Half a Rescue
This is where the abstract's language about reversing some abnormalities needs unpacking, because the split is informative. In healthy T cells the LXR agonist GW3965 did all three things it should: raised glycosphingolipids (P = 0.026), lowered cholesterol (P = 0.012), and lowered membrane lipid order (P = 0.019). In RRMS T cells it lowered cholesterol (P = 0.010) and achieved nothing statistically for either glycosphingolipids or lipid order. The endogenous ligand 24S-hydroxycholesterol produced the same asymmetry. Functionally, across three donors, the agonist reduced IL-17A (P = 0.050 by frequency and 0.012 by intensity) and proliferation measured by Ki67 (P = 0.049 and 0.043), while IFN-γ reached neither threshold (P = 0.053 and 0.119). One of two membrane defects corrected, one cytokine moved, and the functional claim rests on three donors.

Two Observations That Widen the Claim
Cross-referencing the LXR-responsive genes against brain tissue expression data found overlapping dysregulation in two regions, the corpus callosum and the optic chiasm, with a small set of genes disturbed in T cells and both brain areas at once. UGCG is the one to note, since it encodes the rate-limiting enzyme of glycosphingolipid biosynthesis and was itself reduced in patient T cells at 1.36-fold, which links the transcriptional finding to the membrane finding through a single enzyme. Separately, CD4⁺ T cells from ten patients sampled before and after three months of interferon-beta showed four lipid metabolism genes shifting back toward healthy control levels: UGCG, IDOL, LDLR and EEPD1, all at P < 0.002. That is a descriptive longitudinal observation rather than a demonstrated mechanism, but it ties the lipid phenotype to a therapy that works.

What the Numbers Will and Will Not Bear
The limitation the authors give the most attention to is the one that deserves it. The sequencing groups differ in age, patients averaging 42.2 years against 25.8 for controls, a gap of more than sixteen years in a comparison of ten against ten. They engage with it at length rather than burying it, noting that age influences membrane fluidity, cholesterol content and fatty acid unsaturation, and concluding that they cannot rule out an age effect while arguing it is unlikely to be the sole contributor, supported by recovering the same pathways in an independent CD4⁺ T cell dataset. That is the right handling, and a sixteen-year gap is still a sixteen-year gap. Their other stated limits: all participants were white, body mass index was unavailable for healthy donors, and sorted cells were not checked for loss of small populations such as CD56⁺ CD4⁺ T cells. To those belongs one more, running through the therapeutic claim rather than the descriptive one: the functional experiments used three donors, the membrane rescue four to six, and no power calculation appears anywhere in the paper. Set against that, the design choice worth keeping is sampling every patient before their first therapy, which removes the confound that would otherwise hollow out a study of lipid metabolism.

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:
Martin-Gutierrez, L., Waddington, K. E., Maggio, A., Coelewij, L., Oppong, A. E., Yang, N., Adriani, M., Nytrova, P., Farrell, R., Pineda-Torra, I., & Jury, E. C. (2024). Dysregulated lipid metabolism networks modulate T-cell function in people with relapsing-remitting multiple sclerosis. Clinical and Experimental Immunology, 217(2), 204–218. https://doi.org/10.1093/cei/uxae032