Resilience as a Target in MS
Most mechanistic work in multiple sclerosis (MS) asks what destroys tissue. A smaller line asks the inverse, which cells survive conditions that kill their neighbours and what they are doing differently. Three recent papers sit along that line. Two find survival mechanisms by starting from cells that lived rather than cells that died, and both go past the correlation they started with to a manipulation that tests it. The third sets out, as a framework, what this class of cell-level data can establish and what it can only suggest, which is the constraint the other two are working against.
Which Neurons Survive, and a Protein Doing an Unexpected Job
Mayer and colleagues used retinal ganglion cells, a mixed neuronal population in the eye that degenerates in MS, as a way to ask why some neuronal subtypes withstand inflammatory injury while others do not. Single-nucleus RNA sequencing of those cells from MS donors and from controls showed expression of CFH, the gene encoding complement factor H, correlating strongly with resilience to degeneration. The function they then found is not the one complement factor H is known for. It is induced in retinal and in other CNS neurons under inflammatory and oxidative stress, where it limits accumulation of reactive oxygen species and limits lipid peroxidation, and it sits at the endoplasmic reticulum, a major site of lipid peroxidation during neuronal ferroptosis.
How That Claim Was Secured
Claiming a new function for a well-studied protein requires ruling out the old one, and the paper does that by subtraction three times over. The protective activity depends on the C-terminal SCR20 domain, which localizes the effect to one part of the molecule rather than to the protein as a whole. It does not require complement factor H to be secreted, which separates it from everything the protein does outside the cell. And it survives in the absence of complement component C3, which removes the pathway complement factor H is classically defined by. Each subtraction on its own would leave an alternative reading open; together they leave the job inside the cell standing as something separate rather than as a variation on the canonical one. The result holds in mice and in humans, which is why the authors present it as a general mediator of neuronal resilience rather than a feature of one model.
Axons That Live a Year Without Myelin
Yamada and colleagues started from an animal that does better than it should. Mice lacking Teneurin-4 show severe dysmyelination of small-diameter axons, the population that degenerates first in demyelinating disease, yet they survive beyond a year. Immunohistochemistry found abnormally diffuse and intense neurofilament staining alongside raised nonphosphorylated neurofilament and amyloid precursor protein, both markers of axonal stress, so these are not healthy axons. Electron microscopy then showed them still present at one year with no compact myelin around them. Mass spectrometry of the tissue pointed at a candidate, with four of the seven 14-3-3 isoforms highly expressed in those axons, and the test that followed was pharmacological rather than correlational: inhibiting 14-3-3 function with two separate compounds produced progression of small-diameter axon damage. That is what puts 14-3-3 proteins among the factors actively keeping dysmyelinated axons alive, rather than among proteins that merely happen to be present.
What This Class of Data Can and Cannot Establish
Du and Yan wrote the framework these results operate inside, working on peripheral monocytes and monocyte-derived macrophages in neurodegenerative disease. Their starting problem is that these cells converge phenotypically with resident microglia, so lineage attribution across studies is inconsistent and conclusions depend heavily on which model and which disease stage was used. They separate three linked processes: a recruitment gate controlling access to CNS borders and lesions, a lesion-decoding hub where aggregate, lipid, cytokine, complement, antigenic, hypoxic and danger-associated inputs are interpreted, and a state-persistence layer in which metabolic and epigenetic reinforcement stabilizes programs that are inflammatory, repair-supportive, or a hybrid that restricts repair. The part that reaches beyond myeloid biology is their explicit separation of what single-cell or spatial data can infer from what only origin-resolving approaches such as fate mapping can establish, together with proposed terminology for studies that can do neither. Their worked example makes the stage problem concrete: one recruitment pathway may be harmful while a lesion expands and useful during debris clearance.
Where Metabolism Enters, and Where It Does Not
Metabolism sits at the centre of two of these papers and at the edge of the third, which is worth stating rather than smoothing over. In the retinal ganglion cell work it is the mechanism itself, since what complement factor H does inside the neuron is restrain reactive oxygen species and lipid peroxidation, and the organelle it was found in is where that peroxidation happens during ferroptosis. In the myeloid framework it is the stabilizer, with metabolic and epigenetic reinforcement named as what locks a cell into one program rather than another, which makes metabolism the reason a state persists rather than the reason it begins. In the axonal work metabolism is not the subject at all; the 14-3-3 result is about protein-level maintenance of axons under a condition that should have killed them.
The Move These Three Share
What connects these papers is not a pathway but a refusal to stop where the data first points. The retinal ganglion cell study did not rest on a correlation between expression and survival; it removed secretion, removed C3 and mapped the domain, so the claim about a separate function stands on three independent subtractions rather than on one association. The axon study did not rest on proteomics showing which proteins were abundant in surviving axons; it inhibited them and watched the damage appear. And the myeloid framework is largely an argument that the field has been claiming cellular origin from data that cannot resolve origin, with a vocabulary proposed for saying less when less is known. None of the three is a finding about MS metabolism in the sense of a measured metabolite in a patient. Two are findings about what keeps a cell alive, and all three are arguments about how much weight a cell-level observation can carry.
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:
Mayer, C., Woo, M. S., Sonner, J. K., et al., Schirmer, L., & Friese, M. A. (2026). Intracellular complement factor H protects neurons during CNS inflammation. Nature, 658(8136), 765–773. https://doi.org/10.1038/s41586-026-10981-y
Yamada, N., Saito, H., Takahashi, R., et al., & Suzuki, N. (2026). The maintenance of dysmyelinated small-diameter axons by 14-3-3s in the central nervous system. Glia, 74(11), e70221. https://doi.org/10.1002/glia.70221
Du, L., & Yan, J. (2026). Lineage-calibrated peripheral monocyte-derived myeloid states in neurodegenerative disease: recruitment, lesion decoding, and state persistence. Neurobiology of Disease, 228, 107556. https://doi.org/10.1016/j.nbd.2026.107556
