For a long time researchers figured the body had a tidy way of dealing with immune cells that might trigger diabetes, lupus or other autoimmune diseases—it must kill off these rogue cells early in life, before the immune system matures. New research published on May 19 in Immunity challenges this age-old thinking. Instead, the body seems to keep these so-called self-reactive T cells in benign form to fight potential invaders later.

That conclusion comes from a comprehensive set of immune analyses in mice and people, in which a team at Stanford University has found surprisingly large numbers of self-reactive T cells lurking in the bloodstream through adulthood. The cells are not easily activated, though, suggesting the presence of “a built-in brake,” says immunologist Mark Davis, the paper’s senior author. The findings renew debate about how the immune system manages to marshal its forces against myriad foreign invaders all the while leaving our own tissues alone.

The controversy emerged decades ago when researchers learned the secret to the immune system’s incredible versatility. They discovered that a special gene-shuffling process makes millions of antibodies and receptors. Their sheer number and variety allow our immune cells to recognize any conceivable pathogen, in principle. But the explanation also posed a puzzle: Those random gene rearrangements also produce T cells that could attack the body’s own tissues. As a solution, some scientists proposed that the body wipes out those self-reactive cells while the immune system is developing.

Subsequent experiments by several labs supported this proposal. In one study published in 1988 researchers in Switzerland genetically engineered a mouse so that most of the animal’s T cells recognized the same antigen—a snippet of protein called H-Y that is found only in males. In female mice, which lack this protein, H-Y–specific T cells developed normally, just like T cells that recognize flu viruses or other foreign material. But in male mice H-Y–specific T cells hardly made it into circulation. The results appeared consistent with the longstanding theory about the elimination of self-reactive cells during development.

Still, some scientists were not convinced. The mice in the 1988 study were an artificial system. The female animals’ T cells were predominantly specific for a single protein, H-Y, whereas the 200 billion T cells in a typical human adult recognize millions of different substances. And therein lay the challenge: how to fish out of that huge mix the few cells of interest.

Davis and his colleagues overcame this hurdle in the 1990s when they figured out how to place a fluorescent label on specific, individual T cells. That enabled the researchers to take batches of immune cells and use conventional sorting procedures to isolate the rare cells they wanted to study. With later refinements by Marc Jenkins’s lab at the University of Minnesota Medical School, the method became sensitive enough to examine specific naturally occurring T cells in the context of a normal immune system.

In the current study Davis’s group used this approach to determine the frequency of H-Y–specific T cells in a group of blood donors. In the women about one in 68,000 killer T cells were specific for H-Y. (About a third of our T cells are “killer” T cells, which fight cancer cells and other invaders. The other two thirds are “helper” T cells that help initiate the fight.) In the men the frequency of H-Y–specific cells was only a little lower (one in 200,000). That meant a sizeable number of their H-Y-specific killer T cells had escaped deletion.

A bigger surprise came when the scientists surveyed the blood samples for killer T cells specific to other foreign peptides. The number of foreign-specific T cells was essentially the same as T cells recognizing various “self” peptides. They did not behave the same, though. Cultured in petri dishes, foreign-specific cells grew easily whereas self-specific cells languished. Plus, foreign-specific T cells turned on a set of proliferation-related genes that were expressed at much lower levels in self-specific T cells. “There was something funky about the self-specific cells,” Davis says. “More was required to get them going.”

Considering that infectious diseases are historically the number-one killer, the results arguably make sense. “You still want [self-reactive cells] to be there in case a pathogen comes along with that specificity,” Davis says. In support of that idea his team modified a pivotal section of a hepatitis C viral peptide by substituting in each of 20 possible amino acids. They found that human blood samples contained T cells specific for all 20 versions of the virus.

Some researchers think the new findings could shift the field’s view on how the body handles self-reactive T cells. Instead of killing them wholesale, Jenkins says, it’s more like the system “stuns them so they’re present but don’t function.”

Others see things differently. Philippa Marrack of National Jewish Health, Denver, says the immune system “has got to let self-reactive cells through anyway” because some become specialized regulatory T cells that help the body by suppressing other immune cells.

Whatever their fate, self-reactive cells could also hold clues to cancer immunotherapy, Davis says. This therapeutic approach uses the body’s own T cells to attack tumors—but often the therapy is unreliable. T cells may be inhibited “because they think the cancer cell is a self antigen,” Davis says. Figuring out how to lessen that inhibition could motivate them to attack the cancer.