Cell 'villages' reveal genetic differences in how cells divide and survive

A cell's willingness to divide and its capacity to survive are not universal traits. They vary from person to person, shaped by genetics and environment in ways that can predispose someone toward developmental disorders, tissue damage, or cancer. Understanding these differences—what scientists call cell fitness—could eventually explain why some people weather a disease or toxic exposure better than others, and guide treatment tailored to individual vulnerability. But measuring cell fitness across many genetic backgrounds has been a stubborn problem. Researchers have had to grow cells from each donor separately, a slow and expensive process vulnerable to tiny, uncontrollable variations in temperature, oxygen, and handling that can obscure the real biological signal.

UCLA scientists have found a way around this bottleneck. In a study published in the American Journal of Human Genetics, they pooled neural progenitor cells—the early building blocks of the developing brain—from dozens of genetically distinct donors into a single shared culture. By growing all the cells under identical conditions in one dish, they eliminated the technical noise that had previously masked genuine genetic differences. Michael F. Wells, an assistant professor of human genetics at UCLA's David Geffen School of Medicine and the platform's creator, describes the approach simply: "By using cell villages, we eliminate a lot of the technical noise that can mask real biology, and it means we can include far more genetic diversity in a single experiment." The pooled cells generate complex data, however, so the team built a companion statistical tool called Townlet to reliably measure each donor's fitness within the crowd.

The math problem was real and consequential. When cells from multiple donors grow together, tracking each person's share of the dish creates a statistical trap. Each donor's cells occupy a slice of a fixed whole, so when one person's cells grow faster, the others' slices automatically shrink—not necessarily because those cells are less fit, but because the math demands it. Most standard statistical tools would mistake this forced shrinkage for genuine biology, sending researchers down false leads. Harold Pimentel, an assistant professor of computational medicine at UCLA, explains the challenge: "The readout is proportional data—essentially a pie chart changing over time. When one slice grows, the others have to shrink to make room, because everything has to sum to one. But most standard statistical tools assume each slice doesn't know about the others." Townlet was built to tell the difference between mathematical artifact and real biological variation.

The team first tested the platform on a puzzle that has divided autism researchers: a deletion on chromosome 16p11.2, a genetic variant that increases autism risk and, in some carriers, is linked to macrocephaly—an unusually large head that may indicate a larger brain. The connection has been debated for years, but head circumference in an infant's first year is one of the strongest early predictors of an autism diagnosis, making the question urgent. The researchers pooled neural progenitor cells from twelve people carrying the deletion and eleven without it, the largest study of its kind. The answer emerged clearly: cells from people with the deletion divided faster. This points to early overgrowth of brain cells as a key driver of the larger heads seen in some autistic individuals—a mechanistic foothold for understanding how the condition takes root, appearing months before behavioral signs emerge.

When the team scanned the genomes of thirty-five donors in a larger village, they traced natural variation in how fast brain cells grow to a region near ZFHX3, a known brake on cell division in the developing brain. They confirmed the link in an independent group of donors. Survival, though, told a different story. When the researchers exposed a thirty-nine-donor village to lead—a well-established threat to developing brains, found in old paint, aging water pipes, and contaminated soil—the damage varied enormously. Some donors' cells showed a twenty percent drop in viability after a week of exposure; others showed a ninety percent drop, with everything in between. The team traced part of this variation to a region near ARNT2, a gene that protects cells from stress. The findings point toward two possibilities: identifying people whose genetics make their cells more vulnerable to lead exposure, and developing therapies to shield them from harm.

The broader ambition is what Wells calls an "atlas of human vulnerability"—a map of genetically influenced fitness across many cell types and exposures. The team has made Townlet's code publicly available so researchers anywhere can use it. Chloe Hanson, a graduate student who helped build Townlet, sees the approach as part of a larger shift toward studying human biology directly, reducing reliance on animal studies. The vision is the personalized medicine that has been promised for years: a future where people understand their own genetic predispositions, know which exposures they should avoid because they are more sensitive than average, and receive treatment designed for their particular vulnerability rather than a population average.