Much of this work has been done on the hardy arabidopsis—“the lab rat of plants,” as He puts it. There are a few things that make it the perfect test subject. One is that the humble weed’s genome is fairly short, part of the reason it was the first plant to be fully sequenced. Another is the unique way its code can be modified. For most plants, the process is painstaking. New genetic material is introduced in a petri dish, borne by bacteria that slip into the plant’s cells. Once that happens, those modified cells must be cultured and coaxed into new roots and stems. But arabidopsis offers a shortcut. Biologists need only dip the plant’s flowers into a solution filled with gene-bearing bacteria and the messages will be carried straight to the seeds, which can simply be planted. In the painstakingly slow field of botany, that’s going at warp speed.

Still, it took years to figure out what all those SA-producing genes did in perfect greenhouse conditions. Only then could He’s team start tampering with the environment to test what goes wrong. Their mission: find a gene (or genes) that control whatever step was holding up SA production when it got hot. It took 10 years to find the answer. They modified gene after gene, infecting the plants and looking at the effects. But no matter what they did, the plants still withered from disease. “You wouldn’t believe how many failed experiments we had,” He says. Major leads, such as another’s lab identification of heat-responsive genes that affect flowering and growth, ended in crushing disappointment. Generations of grad students kept the project going. “My job is mainly to be their cheerleader,” he says.

Eventually, the lab found a winner. The gene was called CBP60g, and it seemed to act as a “master switch” for a number of the steps involved in making SA. The process of taking those genetic instructions and producing a protein was being stifled by an intermediate molecular step. The key was to bypass it. The researchers could do that, they found, by introducing a new stretch of code—a “promoter” taken from a virus—that would force the plant to transcribe the CBP60g and restore the SA assembly line. There was another apparent benefit: The change seemed to also help restore less-understood disease-resistance genes that were being suppressed by heat.

He’s team has since begun testing the gene modifications on food crops like rapeseed, a close cousin of arabidopsis. Apart from the genetic similarities, it’s a good plant to work with, he says, because it grows in cool climates where the plant is more likely to be affected by rising temperatures. So far, the team has had success turning the immune response back on in the lab, but they need to do field tests. Other potential candidates include wheat, soybeans, and potatoes.

Given the ubiquity of the SA pathway, it’s not surprising that He’s genetic fix would work broadly across many plants, says Marc Nishimura, an expert in plant immunity at Colorado State University who wasn’t involved in the research. But it’s only one of many climate-sensitive immune pathways biologists need to explore. And there are variables other than heat waves that will affect plant immunity, he points out, such as increasing humidity or a sustained heat that lasts through the entire growing season. “It may not be the perfect solution for every plant, but it gives you a general idea of what goes wrong and how you can fix it,” he says. He considers it a win for using basic science to decipher plant genes.

But for any of this to work, consumers will need to accept more genetic tinkering with their food. The alternative, Nishimura says, is more crop loss and more pesticides to prevent it. “As climate change accelerates, we’re going to be under pressure to learn things in the lab and move them into the field faster,” he says. “I can’t see how we’re going to do this without more acceptance of genetically modified plants.”