Cancer biologist Cyrus Ghajar was gearing up to study how the immune system can fight breast cancer when he hit a snag: The supposedly fast-spreading cancer cells he implanted in mice stayed put and sometimes even disappeared after about 11 days. Then postdoc Candice Grzelak identified the culprit: the green fluorescent protein (GFP) the researchers were using to track the cells. The marker itself was stimulating the rodents’ immune system to attack the tumor cells.
Ghajar’s lab at the Fred Hutchinson Cancer Research Center got around this unexpected problem, which it described in a paper last month. But he and others say the lab’s experience reflects a broader issue in mouse studies of immunotherapies, powerful treatments that harness the immune system to vanquish tumors: The glowing proteins biologists use to track the cancer cells, often borrowed from fireflies or jellyfish, may be provoking their own immune attack on the cells.
Other foreign proteins that are workhorses of lab studies, such as components of the genome editor CRISPR, could have the same effect. And the phenomenon could explain why labs sometimes can’t reproduce immunotherapy findings from other groups, suggests Glenn Merlino, a cancer biologist at the National Cancer Institute (NCI).
As immunotherapy becomes more and more important, he adds, scientists need to be aware of confounding factors like this. “So many preclinical experiments do not end up telling you anything useful in the clinic,” says Merlino, a co-author on commentary on the issue today in Cancer Cell.
Although it’s long been known that the immune system can sense marker proteins such as GFP as foreign, it didn’t much matter for cancer studies. That’s because most labs used mice lacking an immune system so they would not reject the transplanted human cancer cells often used to assess treatments.
But as immunotherapies have taken off in past decade, more labs are working with mice that have intact immune systems. Ghajar and others shifted to mouse cancer cells, which aren’t immediately rejected. Others use mice that have humanized immune systems and accept human cancer cells.
Ghajar’s lab realized its mice were producing immune sentinels called T cells that attacked the GFP-labeled cells, blocking their growth. They lowered the levels of GFP, but the cancer cells still didn’t metastasize. In the end, the group found the best solution was to trick the mouse immune system into thinking the GFP was a natural protein, by using mice engineered to produce GFP in certain immune cells known as dendritic cells, which induce tolerance. In these rodents, the breast cancer cells grew as expected, they reported in Cancer Cell.
“We wanted to draw attention to the problem and provide the field with reagents and metrics necessary to solve it,” Ghajar says.
Merlino and his NCI co-authors warn in their commentary that the same problematic immune response could arise in experiments using other glowing proteins from various species, viruses used to introduce cancer genes, and even Cas9, CRISPR’s DNA-cutting enzyme, which comes from a bacterium. Researchers may need to find workarounds, such as mouse strains modified so they tolerate the foreign proteins, like the mice turned to by Ghajar’s lab. Merlino calls for other researchers to share similar experiences, perhaps in a database.
Peter Friedl, who studies metastasis at Radboud University and the MD Anderson Cancer Center, says he, too, has had experiments fail because of an immune reaction to a nonmouse marker protein. Researchers have fingered other causes for cancer biology’s replication problem, such as variations in mouse colony microbiomes. But the unexpected immune responses, Friedl says, “absolutely” could contribute.