Cancer is notoriously hard to treat. In part, this is because the cells making up a tumor are heterogeneous, expressing different genes and molecules that determine their response to treatment. Even if a treatment kills most cancer cells, one survivor is enough for the cancer to persist.
As scientists struggled to find these treatment-resistant cells, they turned to an unexpected tool: sister cells. While human sisters may share clothes or toys, sister cells share their gene expression profiles, which could hint at whether the cells are treatment resistant.
In a study published in Nature Communications, researchers at the University of Helsinki presented a new method called ReSisTrace that utilizes sister cells to identify the molecular states driving treatment resistance in cancer cell lines.1 Guided by these resistance signatures, the researchers devised a method to predict drugs that would sensitize the cells to treatment.
“We can have data [on] both drug sensitivity and transcriptomics at the single-cell level,” said Jing Tang, a bioinformatician at the University of Helsinki and coauthor of the study. “This is unique and novel, and not available by using other techniques.”
Tang and Anna Vähärautio, a cancer biologist at the University of Helsinki and coauthor of the study, wanted to develop a method that combined lineage tracing—the process of tracking cell fate and offspring—with the ability to profile gene expression in individual cells. However, measuring gene expression in a cell typically destroys it, so scientists cannot trace its lineage at the same time. Enter sister cells: a way to achieve both goals in parallel.
Vähärautio's team devised a method to insert unique DNA barcodes into an ovarian cancer cell line using lentiviral transduction. Then, they allowed the cells to undergo a single division to each produce two sister cells, which they found had similar gene expression profiles. The researchers split the pool of cells in half: in one half, they measured gene expression by single cell RNA-sequencing (scRNA-seq) to construct a picture of each cell’s state, and in the other half, they tested whether the cells responded to certain common cancer treatments.
Using the treatment-resistant cells’ barcodes, the researchers matched them with their sister cells in the pre-treatment pool and analyzed their gene expression profiles. This comparison helped them identify genes that might have caused the cell to evade being killed.
At first, the researchers tried to focus on individual genes, but they soon realized this approach might not be enough. “We don't know if [the genes] are really driving the resistance or if they are secondary effects,” Vähärautio said. This inspired the team to search the whole transcriptome for broader gene expression signatures of treatment sensitivity or resistance. Vähärautio and Tang suspected that these signatures could even help predict additional drugs that could sensitize the cells to a subsequent treatment.
Using published gene expression data collected from cell lines treated with a variety of compounds, Tang’s team identified potential drugs that could push treatment-resistant cells’ gene expression toward that of treatment-responsive cells.2 By doing so, the added drug could prime the cells to respond to cancer treatment. Using computational models, the researchers predicted that administering pevonedistat—a drug that inhibits an enzyme involved in protein degradation—before carboplatin chemotherapy would make the cancer cell line that they were studying easier to kill. They tested their predictions and found that pevonedistat pretreatment, and many other predicted compounds, worked synergistically with common cancer therapies to kill the cancer cells.
These findings came as a pleasant surprise to Vähärautio, and they convinced Tang that this could be a new approach for developing more effective cancer treatments to overcome drug resistance.
Amy Brock, a bioengineer at the University of Texas at Austin who was not involved in this study, noted that the authors defined gene expression signatures by comparing all resistant cells to all sensitive cells, but that there might be even more patterns hidden in individual resistant cells. “It would be interesting to further examine whether sister cells become resistant via common or distinct mechanisms,” Brock said.
Brock hopes that, with a slew of similar methods to track cell lineages and single-cell gene expression, researchers will now focus on applying these tools to better understand how cells evade specific treatments.3,4 Vähärautio and Tang are now applying their method to more sample types, including cancer organoids and acute myeloid leukemia cell lines. But Vähärautio thinks this method could even be useful for studying how cells’ states influence their fates in other contexts, such as development or responses to chemicals. With the computational models for drug prediction, ReSisTrace could even identify ways to change these fates.
“I think the method is really widely applicable and can be used to study many different cell state and fate connections,” Vähärautio said.
- Dai J, et al. Tracing back primed resistance in cancer via sister cells. Nat Commun. 2024;15(1):1158.
- Subramanian A, et al. A next generation connectivity map: L1000 platform and the first 1,000,000 profiles. Cell. 2017;171(6):1437-1452.
- Oren Y, et al. Cycling cancer persister cells arise from lineages with distinct programs. Nature. 2021;596(7873):576-582.
- Gutierrez C, et al. Multifunctional barcoding with ClonMapper enables high-resolution study of clonal dynamics during tumor evolution and treatment. Nat Cancer. 2021;2(7):758-772.
