NEW YORK – Researchers from MIT built family trees of lung cancer cells by using barcoded single cells to trace the patterns underlying tumor metastasis in a mouse model of lung adenocarcinoma.
The resulting phylogenies, or family trees, demonstrated the generation-by-generation development of rare subclones that adopted distinct transcriptional programs driving tumor expansion.
These insights, the authors argue, provide a framework for improving future cancer models and identifying new therapeutic strategies.
"We'd like to see what the vulnerabilities of a tumor are as it evolves not after it's already become truly aggressive," said Jonathan Weissman, a professor of biology and MIT and the study's corresponding author.
The idea that such rare subclones enable tumor growth and metastasis is not new. "The clonal expansion model of tumorigenesis was proposed almost 50 years ago and has been verified by many orthogonal methods," Cheng-Zhong Zhang, a professor of biomedical informatics at the Dana-Farber Cancer Institute, said in an email.
But Weissman and his group bring an unprecedented level of granularity to the process, enabling other researchers to now watch tumor evolution unfold in a nearly step-by-step manner.
"The technical achievement is remarkable," Zhang wrote.
Understanding how tumors progress and metastasize will enable scientists and physicians to develop more rational strategies to limit their growth and prevent metastases.
Weissman and his colleagues combined CRISPR-Cas9 and single-cell readout techniques to simultaneously insert their barcoded lineage tracing system, as well as KRAS and Trp53 oncogenic mutations into individual lung epithelial cells of a mouse cancer model.
"We engineer into a mouse genome, neutral pieces of DNA that Cas9 can then mark, and once it makes these marks, because it's DNA, it's inherited by all the progeny," Weissman said.
The progression of acquired marks can then be graphed to show what Weissman refers to as a "flight recorder" of genetic changes over time.
After allowing several months of growth, single-cell analysis showed significant diversity among subpopulations of cells within the same tumor, and that this diversity arose quickly, early on in tumor development, as cells evolved through various transient transcriptional states before finding more stable ones. Stable states that conferred particularly good growth and survival advantages resulted in more aggressive subpopulations that grew to dominate their tumor.
Cells that became dominant generally adopted one of two evolutionary paths, both of which could lead to aggressive mesenchymal cell states that associate with metastasis.
To test how additional cancer-linked mutations might affect the observed evolutionary paths, Weissman and his colleagues deactivated one of two frequently mutated tumor suppressors in human lung adenocarcinoma: LKB1 and APC.
Although the loss of either tumor suppressor triggered increased tumor growth, each one altered cells' transcriptional states in distinct ways. Cells that lost LKB1, for instance, quickly progressed to and stabilized in a pre-endothelial-to-mesenchymal transition state, whereas those losing APC largely moved through APC-specific states, ending in a mesenchymal state.
Moving forward, Weissman plans to adapt this technique to a 3D system to observe how tumors interact with stroma and with the immune system. To do this without losing the cells' native 3D environment, however, would require different approaches to single-cell analysis.
"There are a number of spatial transcriptomic approaches that are being developed that should be that can be applied to this," Weissman said.
He also plans to investigate which branches of tumor evolution give rise to therapeutic resistance.
"We'd like to identify where resistance comes from and find a way of either pushing [cells] away from those states that give rise to resistance or killing those states that give rise to resistance," he said.
Finally, Weissman plans to more comprehensively study and refine this technique by extending it into different model organisms as well as different cancers, especially leukemias.
"There are new therapeutics for AML," he said, "which is one of the most deadly forms of leukemia, and we'd like to understand how they work and how resistance emerges."