Diseases like cancer or neurodegeneration are known to arise from genetic misfires. But treating such complex conditions hasn't been simply a matter of identifying the malfunctioning genes involved. With hundreds of genetic mutations spanning diverse pathways at play, connecting the dots between a constellation of mutations and a specific outcome has proved to be enormously difficult.
Now, a new study in Nature describes a novel way forward. Researchers developed a platform called PerturbFate that can systematically map how diverse disease-associated genetic variations reshape cells, and where those paths converge. By tracking gene regulation in single cells over time, the team thus identified regulatory nodes common to many different variations. Using melanoma drug resistance as a proof-of-concept, they then showed that these shared points of control offer a path toward combination therapies that can target disease across its many genetic causes.
"We focus here on cancer drug resistance, but the paper really starts from a broader question: once you know that a disease is associated with hundreds of genes, how do you design one therapy to target it?" says Junyue Cao, head of the Laboratory of Single-Cell Genomics and Population Dynamics. "We wondered whether all these different genes may be mediated by some shared downstream signaling that we can discover and target instead."
A new bottleneck
Advances in genomic sequencing and genetic screening have transformed biomedicine, allowing researchers to identify hundreds of genetic mutations linked to disease. But this progress has created a new bottleneck. The offending genes often span diverse pathways, with responsibilities from gene regulation to cell signaling, making them difficult to target as a group. The result is that, even as our understanding of disease expands, our ability to treat it lags.
Cao wondered whether these seemingly unrelated mutations were as independent as they appear. If these mutations instead converged on shared downstream programs that ultimately determine how a cell behaves, the challenge would not be to target each mutation individually, but to identify common control points, known as regulatory nodes. "We wanted to develop a technology to identify these shared regulatory nodes as targets in and of themselves," says Cao.
Such technology would need to compare many different genetic disruptions at once and track, in detail, how each one changes a cell. Existing approaches could capture only part of that picture, often measuring a single molecular layer at a time or missing the real-time dynamics of gene activity. So Zihan Xu, a graduate student in Cao's lab, developed PerturbFate. This platform allows researchers to watch, in real time, how different genetic changes reshape a cell by tracking which parts of the DNA are accessible and how RNA is produced and processed. By capturing these changes in the same single cell, the system reveals the networks of genes that control cell behavior and shows where very different genetic variations end up having the same effect.
"This technology lets us perturb hundreds to thousands of genes in parallel and then measure the detailed molecular changes in each individual cell," says Cao. "That allows us to link many different genetic perturbations to their downstream effects and identify regulatory nodes."
Perturbing fate, in cancer and beyond
To test the platform, the researchers focused on melanoma drug resistance, a case in which many different mutations lead to the same outcome. Using PerturbFate, they selected 143 genes linked to resistance to the common melanoma drug Vemurafenib, and systematically shut them down in melanoma cells. PerturbFate then tracked how each disruption reshaped the cell. By labeling newly made RNA, the team could distinguish real-time gene activity from older signals, while single-cell profiling captured which genes were active, which regions of DNA were accessible, and how those changes unfolded over time. This approach provided a cell-by-cell view of how different mutations alter gene regulation, and where those paths converge.
"We're capturing not just gene expression, but also RNA dynamics and chromatin state," says Cao. "That's critical for identifying the upstream regulators that drive these disease states."
A companion computational pipeline developed by Xu then brought these layers together, reconstructing the gene regulatory networks that drive each cell's response over time. The result linked early changes in the activity of transcription factors to alterations in DNA accessibility and bursts of new RNA and, eventually, stable gene expression patterns.
After analyzing more than 300,000 cells, the researchers found that very different genetic perturbations all pushed melanoma cells into the same drug-resistant state. When the team targeted these common control points, drug resistance dropped significantly, pointing to a promising strategy for combination therapies.
The platform also revealed an important nuance involving the Mediator Complex, a cellular machine that regulates gene activity. The researchers found that disrupting different parts of this same complex could trigger drug resistance through entirely distinct routes. Yet even these divergent paths ultimately converged on the same survival signal in melanoma cells, called VEGFC. When the researchers blocked that signal, the resistant cells could no longer grow.
Taken together, the PerturbFate proof-of-concept demonstrates that complex genetic diversity does not necessarily require complex treatments. With PerturbFate, researchers can now focus on common regulatory nodes when designing drugs, rather than targeting each mutation individually.
The team has made both the experimental and computational tools behind PerturbFate openly available, and plans to extend the approach from cultured cells to living systems. By applying PerturbFate to conditions such as aging and Alzheimer's disease-both major focuses of his lab-Cao and colleagues hope to uncover shared vulnerabilities that can guide more effective treatments.
"This is just a starting point," says Cao. "Now that we've demonstrated the approach in a simple model, we're working to extend it into living systems to study even more complex diseases."
