A popular method of editing genes in research labs could trigger an immune reaction when used in people, according to a new study, which has not yet been published in a peer-reviewed journal. But it is too soon to know how serious a problem this could pose for gene therapy, which aims to stop diseases caused by defective genes.

“The big question will be: What impact does it actually have therapeutically?” says Amy Wagers, a stem cell biologist at Harvard University and the Joslin Diabetes Center, who was not involved in the study. In mice, she says, the gene-editing tool triggers an immune response, but is still safe and effective. No one knows what will happen in people. “It’s something that needs to be investigated,” she says.

The CRISPR–Cas9 system, which functions as a genetic scissors and tape for editing DNA, is generally derived from either Staphylococcus aureus or Streptococcus pyogenes bacteria. Most people have been exposed to staph or strep by the time they reach adulthood, which their bodies are likely to remember and may mount an immune attack when reexposed to them, says Matt Porteus, a pediatrician and stem cell scientist at Stanford University who led the research, which was posted to the preprint server bioRxiv last week.

This prior exposure could potentially render the gene editing ineffective, with the body quickly eliminating all the CRISPR–Cas9 proteins. Or worse, it could trigger the kind of immune storm that killed a young gene therapy patient named Jesse Gelsinger in 1999, derailing the field for more than a decade. “We share everyone’s excitement about doing Cas9 genome editing, but we want to make sure we have learned from what happened in the gene therapy world and not ignore the possibility that this could become a problem,” Porteus says. “As we’re all thinking about developing Cas9-based therapeutics, we should think carefully about this potential problem.”

CRISPR–Cas9, a tool adapted from bacteria, has become the darling of the biomedical community since an explosion of research in 2013. CRISPR, which stands for clustered regularly interspaced short palindromic repeats, can be programmed to find specific stretches of genetic code. Then, the Cas9 enzyme attaches to the targeted DNA and cuts it, shutting off the gene.

There are several other approaches to gene editing that predate CRISPR, including so-called zinc finger nucleases and TALENs (transcription activator-like effector nucleases). But each has challenges that have made CRISPR the favorite in research labs—and, many hoped, in people.

The new study suggests it may take longer to make CRISPR gene editing safe to use in patients. But the problem is not insurmountable, bioengineer Feng Zhang, who helped develop the technology, wrote in an e-mail. “There are many open questions about the safety and efficacy of CRISPR-based therapeutics,” wrote Zhang, a core member of the Broad Institute and a Massachusetts Institute of Technology professor. “Currently a number of different strategies are being pursued to develop Cas9 as a therapeutic…and each design requires a unique consideration of safety and efficacy,” Zhang added. “As these designs enter advanced stages of preclinical testing and early clinical trials, we will learn a lot more about the best way to further advance genome editing as therapies.”

Harvard geneticist George Church, who was also involved in early CRISPR work, says he is already working to overcome the problem. “My lab and others have explored many species and enzymes” that could be used in lieu of Cas9, he wrote in an e-mail. “We are also looking into various approaches to immune tolerance.”

Current gene therapies rely on a virus called adeno-associated virus (AAV) to deliver the gene-editing tool to every cell. People who already have immunity to AAV have been excluded from trials or therapies because their immune systems are likely to clear the treatment before it can have a therapeutic effect. That strategy of exclusion would not work with Cas9, Stanford’s Porteus says, because too many adults have been exposed to strep and staph. “What I think is surprising is the high proportion of people who already have immunity,” he says.

The Stanford researchers first looked for and found Cas9 antibodies in human cord blood samples. Their presence shows that human B cells, part of the so-called innate immune system, can secrete antibodies that recognize Cas9. Next they looked in blood samples from 13 adults to see if they had T cells—part of the adaptive cellular immune system that responds to specific infections—designed to attack cells expressing staph or strep Cas9s. About half had T cells that recognized staph Cas9 but none had detectable T cells that recognized strep Cas9. The immune system’s ability to recognize the enzyme and the fact many people have T cells aimed at fighting it, Porteus says, suggest CRISPR may not be as safe and effective in people as it has been in mice.

Porteus notes he pushed to have the study published online as a preprint while the manuscript works its way through the standard peer-review process. He hopes the faster publication will lead more rapidly to solutions—perhaps finding new Cas9s from bacteria that do not normally infect people. The team included all its data and figures so others can independently evaluate their findings.

Wagers, who raised the issue of immune responses to CRISPR in a 2016 paper, cautions against reading too much into the new study, however. It relies on a tiny number of blood samples, and peer reviewers might still find flaws in the methodology and results, she says.

Wagers thinks it should be feasible to design Cas9s that are based on more than just the two types of bacteria, and it is unlikely that a patient in need of gene editing would have immunity to the full range of possibilities. The paper should serve as more of a reminder than a shocker. “You’re introducing a foreign protein,” she says, referring to Cas9. “The immune system is an important component of how our bodies work, and we have to pay it the respect it deserves.”