At the time, Santamaria was working with a mouse model of type 1 diabetes, a disease in which epitope spreading is known to occur. So, he developed a cocktail of particles carrying eight different antigen fragments, or epitopes. For a control, he used a nanoparticle carrying a single epitope, surmising that one wouldnt have any effect because T cells recognize thousands of epitopes in diabetes. Surprisingly, however, both the control and the experimental nanoparticles reversed diabetes symptoms. It totally didnt make any sense, Santamaria says.

It took him years to work out what was going on, but now he thinks he understands2,3. The nanoparticles prompt T cells to multiply and morph into regulatory T cells, which travel to the site of inflammation. There, they bind and deactivate antigen-presenting cells that carry not only the antigen that these cells recognize, but thousands of other antigens that are important in diabetes. So these cells can no longer activate the immune cells that fuel the disease. In a sense, the nanoparticle-fused antigen works as a master switch. It sounds too good to be true sometimes, Santamaria says. And it might be. The strategy hasnt been Nature | Vol 625 | 25 January 2024 | 647 Feature tested in the clinic yet. Santamaria has founded a company called Parvus Therapeutics in South San Francisco; it should begin its first trial in people this year, starting with an autoimmune disease that affects the liver.

A key in the liver The liver is where all the blood carrying foreign antigens from the gut gets filtered. It is also the destination of all the cellular debris left behind when cells and tissues die. It, therefore, has an important role in establishing immune tolerance. Hubbell and his colleagues discovered that cellular debris carries a special sugar tag that directs it to the liver. By adding this sugar tag to other proteins, they realized, they could direct just about any molecule they wanted to the liver, including antigens such as the myelin proteins that activate the immune system in multiple sclerosis. In work published in 2023, they showed that this strategy works to reverse symptoms of a multiple-sclerosis-like disease in mice4. What was so exciting about this paper, Hubbell says, is that the animals had advanced disease, meaning that their immune systems were probably reacting against a variety of antigens. Yet, treatment with just one antigen worked to reverse paralysis.

A company called Anokion in Cambridge, Massachusetts, which Hubbell co-founded, is working on a similar strategy. It has tested it in a phase I trial in people with multiple sclerosis to assess its safety, and is currently enrolling participants in a phase II trial, in which it will start to assess the efficacy. The firms chief executive, Deborah Geraghty, wont say exactly how the therapy works after the antigen travels to the liver. We havent disclosed a lot of that publicly, she says, but she adds, We believe were driving a strong T-regulatory component.