The matter of whether human cancer follows a hierarchical or stochastic model of differentiation is one that still remains under debate. However, a new study using a novel microfluidic, single-cell culture method has found compelling evidence that human ovarian cancer mostly follows a hierarchical model, with the occasional stochastic event. Although seemingly rare, instances where a daughter cell gave rise to a stem-like cell were indeed observed (Proceedings of the National Academy of Sciences, doi: 10.1073/pnas.1507899112).
The study used cells from epithelial ovarian cancer tumors that were stained and sorted into four distinct populations based on aldehyde dehydrogenase enzymatic activity (ALDH) and expression of the stem cell marker CD133. The microfluidic device then separated out each cell individually for culture over the course of 3-5 days, after which the resultant daughter cells were then reevaluated for markers.
The vast majority of cell divisions obeyed a hierarchical model, where the stem-like, highly tumorigenic grandmother cells (ALDH+CD133+) generated less-tumorigenic daughter cells (ALDH+CD133+/-), which themselves gave rise to granddaughter bulk cells (ALDH-CD133-). But as a notable exception, one stochastic event was observed where a daughter cell divided to produce a grandmother cell.
The study suggests that targeting these stem-like cells should remain a critical objective for designing new ovarian cancer therapies. However, because non-stem cells may dedifferentiate, they must also be taken into account.
"What we found is that indeed ovarian cancer follows what looks like a hierarchical pattern, and that makes sense because cancer tends to mirror normal biology-it's just a bastardized version," said study author Ronald J. Buckanovich, MD, PhD, the Thomas H. Simpson Collegiate Professor of Internal Medicine at the University of Michigan Medical School. "But consistent with a stochastic or random model, we did see instances when non-stem cell types gave way to a stem cell type."
Cancer Stem Cells
The existence of cancer stem cells remains a controversial topic in cancer research. The hypothesis of a cancer stem cell or hierarchical model of differentiation proposes that, as a parallel to normal biology, certain special cells within the tumor can both proliferate and generate daughter cells. These cancer stem cells could cause disease recurrence if left in a patient's body after primary therapy. In a stochastic or random model, any cell can lead to recurrence.
Studies to define the differentiation capacity of cancer cells from human tumor specimens have been hindered by contamination issues, which can produce untrustworthy results.
"It's been very hard to do these kinds of studies with a very pure cell population," Buckanovich said. "Ideally, you separate a hundred thousand stem cells and a hundred thousand cells of the non-stem cell variety-but the problem is if there is a contamination rate of even 0.1 percent, then you've got a hundred cells in the wrong place."
Research Method
Buckanovich and his colleagues relied on a unique microfluidic chip developed by fellow University of Michigan researcher Euisik Yoon at the College of Engineering. Yoon's device is user-friendly and much more efficient at capturing individual cells, so much so that hundreds of thousands of cells weren't a requirement for the experiment to work-and human ovarian cancer cells could be successfully studied.
Cells from three primary patient specimens with stage III or IV ovarian cancer as well as three ovarian cancer cell lines were sorted into four populations (ALDH+/-CD133+/-). They were then loaded as single cells into separate devices and cultured for 72 hours (cell lines) or 120 hours (primary cells).
The researchers evaluated a total of 3,692 progeny from 2,833 cells. They observed the stem-like grandmother cells (ALDH+CD133+) could either divide into two grandmother cells, or create a grandmother cell and a daughter cell (ALDH+CD133- or ALDH-CD133+). However, an event where a grandmother cell generated a granddaughter bulk cell directly (ALDH-CD133-) did not occur. These granddaughter cells only appeared to arise out of daughter cells in a linear fashion.
While the vast majority of divisions followed this type of hierarchical model, they witnessed a single stochastic event where a daughter cell underwent a division that produced a grandmother cell.
"The results do imply that the cancer stem cell is critical, so getting rid of those will be very important therapeutically," said Buckanovich. "But we do need to consider those non-stem cells that go through a dedifferentiation effect as well-so it will be a twofold step."
Buckanovich and his colleagues also discovered that a protein called bone morphogenetic protein 2 (BMP2) serves as a signal to produce more grandmother cells and less granddaughter cells. This feedback mechanism resulted in suppressing bulk tumor growth in vitro while increasing tumor initiation rates, tumor growth, and chemotherapy resistance in vivo.
"BMP2 is a target, but we have to be intelligent about how we target it. It's going to be a little more complicated than making an antibody that targets BMP2," he said. "We could understand which receptor is found on the stem-like grandmother cell versus the granddaughter cell and selectively target those receptors, but it's going to have to be done in a delicate manner."
Meeta Kim is a contributing writer.