Researchers drew blood from nine people with multiple sclerosis, coated their own red blood cells with seven fragments of the very protein their immune systems attack, and put the cells back. Three months after a single infusion, the immune signature had shifted toward calm.
That is the headline result from a phase Ib trial published September 8 in the Proceedings of the National Academy of Sciences ↗. It is a first human test of a simple idea with a long history and a stubborn failure record: teach the immune system to leave one specific target alone, without shutting the whole thing down.
Reverse vaccination
Multiple sclerosis is an autoimmune disease. The immune system attacks myelin, the fatty insulation around nerve fibers in the brain and spinal cord, and as the insulation frays the signals it carries slow or stall. Most approved drugs work by broadly dampening immune activity, which raises the risk of infection and cancer over time. The goal that has eluded the field for decades is antigen-specific tolerance: switching off only the cells that attack myelin and leaving the rest of the defense intact.
The therapy tested here, called CLS12311 and run under the RED4MS trial, tries to reach that goal through the body's own disposal system. The team took autologous red blood cells (the patient's own) and chemically attached seven immunodominant myelin peptides to them, short protein pieces drawn from myelin basic protein, myelin oligodendrocyte glycoprotein, and proteolipid protein. Reinfused, the coated cells reach the end of their natural lifespan and get cleared, and the immune cells that do the clearing read the myelin fragments as routine self, not threat. Roland Martin, whose group at the University of Zurich spun out the developer Cellerys, has described it as a vaccination in reverse: instead of training the body to attack, train it to tolerate.
What the trial measured
The study was open-label and dose-escalating, the standard shape of a first-in-human safety trial. The peptide-coupled cells were safe and well tolerated, the primary thing a phase Ib is built to show. Clinical and imaging measures stayed stable over the follow-up, and blood levels of neurofilament light chain, a protein that leaks out when nerve fibers are damaged and a widely used marker of ongoing injury, dropped.
The mechanistic readouts are the real substance. Using single-cell RNA sequencing, which reads gene activity one cell at a time, the team saw the population of myelin-specific CD4+ memory T cells expand into regulatory phenotypes, the calming rather than the attacking kind, while proinflammatory cells fell. Autoreactive T-cell responses declined, and there was an early rise in interleukin-10, Tr1 regulatory T cells, and tolerance-promoting monocytes and dendritic cells, the messengers and cell types that enforce restraint. Biodistribution work in mice and humans confirmed the mechanism's plumbing: the infused cells were processed by Kupffer cells in the liver and marginal zone macrophages in the spleen, the two sites where the body routinely retires old red cells and, apparently, learns from what they carry.
The size of the claim
Nine patients, no control group, a single dose, and a follow-up measured in months. This is a mechanism proof, not evidence that the disease was altered. Stable clinical and imaging scores in an uncontrolled trial can reflect the natural quiet between relapses as easily as a drug effect, and the neurofilament drop, while encouraging, carries the same caveat without a comparator arm. What the trial does show, and shows in humans rather than mice, is that a single infusion of self-antigen on the patient's own red cells can move the specific immune populations that drive the disease in the direction tolerance requires.
The approach has a graveyard behind it. Earlier attempts to induce myelin tolerance, including a 2013 trial from the same lineage that used white blood cells as the carrier, showed the biology was plausible but never converted into a marketed therapy. Red cells are cheaper to prepare and exploit a clearance pathway the body runs anyway, which is the bet here. Novartis has taken an option to acquire Cellerys pending a phase II readout, so the next trial, with a control arm and hard clinical endpoints, is the one that matters.
For a platform built mostly on metabolic peptides, this is a reminder that peptides do therapeutic work far outside appetite and glucose. Here the peptides are not the drug in the usual sense but the immune ↗ instruction set, and whether they protect the nervous system is a neuroprotective ↗ question the phase II will have to answer.