From Metabolite Signature to Drug Target: Glycolysis in Relapsing-Remitting MS

Most blood metabolomics studies in multiple sclerosis (MS) stop once they have a signature. Zahoor and colleagues kept going. They ran discovery metabolomics in patients, used network analysis to find the pathway sitting upstream of everything that changed, tested whether patient cells actually behave that way, blocked that pathway pharmacologically in animals, and then worked out which cell type carried the effect. The serum came from treatment-naive patients with relapsing-remitting MS (RRMS) alongside 14 age- and sex-matched healthy subjects, drawn from a biobank at Henry Ford Hospital in Detroit and diagnosed by the 2010 revised McDonald criteria. Two numbers are given for the patient group: the abstract reports 35 patients, while the results section reports 33 entering the profiling. The treatment-naive part carries real weight, because disease-modifying therapies alter lipid and energy metabolism directly, so any difference found here is not drug exposure in disguise. Of 632 known metabolites measured, 60 differed significantly between patients and controls.

Four Pathways, and Only One Survived Correction
The 60 altered metabolites came out at P < 0.05 with a false discovery rate below 0.10, and their direction was lopsided: 53 were raised in patients and 7 reduced. Lipids made up 46% of them, followed by xenobiotics at 20% and peptides at 12%. Pathway analysis returned four candidates: glycerophospholipid metabolism, the citrate (TCA) cycle, sphingolipid metabolism, and pyruvate metabolism. The detail worth carrying forward is that only glycerophospholipid metabolism cleared multiple testing correction, with an FDR of 3.97E-02. The citrate cycle, sphingolipid, and pyruvate pathways carried FDR values of 0.65, 0.66, and 0.79. They ranked highly on pathway impact scores of 0.10, 0.04, and 0.14, but impact measures a metabolite's position in the network topology rather than the strength of the evidence. The framework the rest of the paper builds on therefore rests partly on pathways that did not survive correction.

Glycolysis Is Not a Fifth Pathway; It Is the One Feeding the Other Four
The argument that makes this paper work is a convergence argument. Dihydroxyacetone phosphate, an intermediate of glycolysis, is converted to glycerol 3-phosphate by glycerol-3-phosphate dehydrogenase, and glycerol 3-phosphate is required for de novo synthesis of glycerophospholipids, sphingolipids, and lysolipids. Pyruvate, the end product of glycolysis, either becomes lactate through lactate dehydrogenase or enters the TCA cycle as acetyl-CoA through pyruvate dehydrogenase. One pathway upstream feeds all four downstream ones. Causal network analysis added independent support by naming sphingosine-1-phosphate receptor 2 as an activated master regulator with an activation z score of 2.449, and TGF-β1 as a second with a z score of 2.236. The network predicted that RRMS could be repressed by targeting sphingosine-1-phosphate receptor 2 through intermediate regulators including interferon-β1 and fingolimod. Both are approved MS drugs, so the algorithm recovered existing therapy from metabolite data alone, which is a fair sign the network is not noise.

The Cells Are Glycolytic Without the Genes Changing
Before testing the pathway, the authors checked whether the change was transcriptional, and found it was not. Messenger RNA for the key glycolytic and TCA enzymes showed no difference between patient and healthy peripheral blood mononuclear cells, including LDHA, LDHB, the malate dehydrogenases, the malic enzymes, and the succinate dehydrogenase subunits. Three public expression datasets agreed, with only LDHA and ASL altered in one of them. So the pathway is running differently while its genes sit still, which points to regulation at the level of enzyme activity rather than expression. The functional test then confirmed the prediction directly. Seahorse measurement of how fast the cells acidify their medium, the standard readout of glycolytic rate, showed significantly higher basal glycolysis in cells from 17 patients with RRMS than in cells from 14 healthy subjects (P < 0.05), a result reproduced in cells from the mouse model. A metabolite pattern predicted a cellular behaviour, and the cells did it.

Blocking Glycolysis Across Four Mouse Models
The therapeutic test used 2-deoxy-D-glucose, a glucose analog that hexokinase phosphorylates into 2-deoxy-glucose-6-phosphate, which cannot be metabolised further and so stalls the pathway. In the SJL relapsing-remitting model, daily treatment at 50 mg/kg from day 6 after immunisation reduced peak severity from 2.7 ± 0.22 to 1.2 ± 0.54 (P < 0.01) without delaying disease onset, and abolished the relapse, with scores of 1.9 ± 0.11 against 0.2 ± 0.22 at day 25 (P < 0.001). The B6 model gave 2.6 ± 0.21 against 0.3 ± 0.14 and the 2D2 T cell receptor transgenic model 2.4 ± 0.27 against 0.3 ± 0.22, both at P < 0.001. Delivery in drinking water rather than by injection also delayed onset and reduced severity. Histology backed the clinical scores, with less immune cell infiltration, smaller lesions, and myelin preserved on Luxol fast blue staining. Four models and two routes of administration is more than most preclinical efficacy claims carry.

Monocytes Carry the Effect, and Transferring Them Transfers the Protection
Treatment cut infiltrating CD4 T cells and their production of IFN-γ, IL-17, and GM-CSF while raising IL-10. In monocytes it reduced glucose uptake, lactate secretion, and expression of GLUT1, HK2, TPI, PKM, LDHA, and MCT1. Total cellular ATP did not change, which says the raised glycolysis in these cells was never supplying energy. It was supplying biomass, the fatty acids needed to expand endoplasmic reticulum and Golgi for the protein output that antigen presentation requires. Macrophages shifted towards an anti-inflammatory profile, with arginase 1 and Ym1/2 up and iNOS and IL-1β down. The experiment that settles the mechanism is the adoptive transfer: monocytes taken from treated animals and given to recipient mice with active disease reduced severity on their own. That moves the finding from an association between drug and outcome to the monocyte phenotype itself carrying the protection. The authors also report a clean negative, no effect on regulatory T cells, which contradicts an earlier study they cite and then discuss rather than bury.

What This Establishes, and What 2-Deoxy-D-Glucose Cannot Be
The authors describe the work as proof of principle and say directly that the compound has poor drug-like properties, with rapid metabolism and a short half-life, and that any therapeutic use would have to be timed around inflammatory episodes. Two further constraints sit alongside that. The compound inhibits glycolysis wherever glycolysis happens, brain and muscle included, which is the specificity problem every glycolysis inhibitor runs into. And the human side of this study is modest: 33 patients from one Detroit biobank for the metabolomics and 17 for the functional work, with 88% of the altered metabolites moving in the same direction, a pattern that in a cohort this size deserves the caution the FDR values already imply. The therapeutic evidence is entirely murine. What deserves copying here is the design, which treats a metabolomic signature as a hypothesis to be tested rather than a result to be reported. The open question the paper leaves is whether glycolytic rate in patient cells tracks relapse activity within individuals over time. That measurement needs only the assay they already ran, and it would separate a treatment target from a marker of active inflammation.

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:
Zahoor, I., Suhail, H., Datta, I., Ahmed, M. E., Poisson, L. M., Waters, J., Rashid, F., Bin, R., Singh, J., Cerghet, M., Kumar, A., Hoda, M. N., Rattan, R., Mangalam, A. K., & Giri, S. (2022). Blood-based untargeted metabolomics in relapsing-remitting multiple sclerosis revealed the testable therapeutic target. Proceedings of the National Academy of Sciences, 119(25), e2123265119. https://doi.org/10.1073/pnas.2123265119