Cells in individual tumors can interconvert into different cell types including reverting into cancer stem cells in order to maintain equilibria in terms of the proportion of cells existing in different states within the cancer, researchers claim.They found that rather than existing as a hierarchical society in which all cells are derived from cancer stem cells, cancers exist as a decentralized society of different cell types that can sense when one type of cell has been depleted and generate new cells of the relevant type to take their place.
The group included scientists at the Massachusetts of Institute of Technology Broad Institute, Tufts University, and Harvard Medical School. Details are published in Cell in a paper titled “Stochastic State Transitions Give Rise to Phenotypic Equilibrium in Populations of Cancer Cells.” Results could have significant implications for cancer cell therapy, claim lead researchers Eric S. Lander, Ph.D., and Piyush B. Gupta, Ph.D., because removing cancer stem cells will just prompt other cell types in the tumor to convert into stem cells to top up the population.
One of the puzzling features of cancer cell populations is their ability to retain phenotypic equilibrium over extended periods of time, the team writes. Populations of cancer cells often harbor subpopulations with specific cell-surface marker profiles, which are stably maintained across many cell divisions in culture.
To investigate the basis of this equilibrium-maintaining phenomenon further, the researchers isolated and separately cultured three cell types—stem-like, basal, and luminal—from two different human breast cancer lines derived from primary tumors. Each of the three cell types was confirmed to display specific morphological and cell surface marker characteristics.
These relatively pure subpopulations of cells, which each represented a given differentiation state, were then allowed to expand in culture, and relevant population dynamics monitored over time. Surprising, the researchers found that when they assessed the relative proportions of stem-like, basal, and luminal cells in each originally ‘pure’ population after expansion, there had been an evident rapid progression back to equilibrium proportions.
Two lines of evidence indicated that this progression was due to interconversion between states, rather than as a result of differential growth rates of cells in the basal, stem-like, or luminal states, they claim. Firstly, there was no difference in the proliferation rates of the stem-like, basal, or luminal subpopulations sorted from either of the two stem cell lines: they all replicated at about the same rate.
Secondly, given the purity of the original sorted populations and the rapid rate of return to equilibrium proportions, some minority subpopulations would need to have been dividing at more than three times per day to achieve the observed proportions through differential growth rate alone. “Such a high proliferation rate is implausible because even the most rapidly dividing human cells— embryonic stem cells—require at least 24 hours to complete a proliferation cycle,” they claim.
Based on the notion that interconversion between cell states was therefore occurring, the team used data from their expanded breast cancer cell populations to developed a Markov model, in which the cell type transition probabilities depend only on a cell’s current state, not on its prior state. The inferred Markov transition probabilities thus make it possible to quantitatively predict how a population of cells evolves over time, given the initial proportions of cells in different states.
The resulting model made several predictions about how the cell populations would develop, and these were confirmed in the cultured breast cancer populations, the researchers note. However, a number of unexpected predictions also emerged. One of these was that basal and luminal cells can transition back into a stem-like state: “that is, cancer stem-like cells can arise from non-stem-like cells." This essentially contradicts current concepts relating to normal tissues, which assume a rigid lineage-hierarchy in which stem cells can give rise to nonstem cells, but not vice versa, they write.
They tested this particular prediction by implanting either freshly sorted, or sorted and then cultured subpopulations of tumor cells in mice. As expected according to traditional dogma, only the stem-like fraction could efficiently seed tumors, and neither the luminal nor basal fraction was capable of doing so.
However, because the lack of tumor-seeding ability displayed by the basal and luminals could have been due to their inability to survive after transplantation, the researchers repeated the exercise by co-inoculating the cells with GFP-labeled, irradiated parental carrier cells from one of the breast cancer lines. Under these conditions, all three fractions (stem-like, basal, and luminal) were equally capable of efficiently seeding tumors.
Moreover, examination of the tumors arising from basal and luminal subpopulations mixed with irradiated carrier cells revealed the presence of significant numbers of stem-like cells. The proportions of basal, stem-like, and luminal cells contained in the resulting tumors were comparable irrespective of the sorted subpopulation initially used to seed the tumor.
“Collectively, these results demonstrated that the luminal and basal fractions can indeed regenerate functional stem-like cells in vivo and suggested that convergence toward equilibrium cell-state proportions could be occurring due to cell-state interconversion within tumors,” the authors write. “A specific prediction of this quantitative model is that any subpopulation of cancer cells will return to a fixed equilibrium of cell-state proportions over time, provided that it is possible through one or more interconversions to transition between any two states.”
The de novo generation of cancer stem cells has implications for the effectiveness of anticancer therapies focused on killing this cell type, because of the ability of other cancer cell types to regenerate cancer stem cells after cessation of therapy and lead to renewed tumor growth, they add. “Therefore, in order to be effective, cancer therapies will need to combine agents that are selectively toxic to cancer stem cells with agents that either target the bulk noncancer stem cell populations within tumors or inhibit transitions from noncancer stem cell to cancer stem cell states.”
The team claims their model could also be extended to other biological settings in which stochastic state transitions occur, either in normal or diseased contexts.