Many stem cell biologists have experienced the frustration: you do an experiment; your lab mate at the next bench does the same. But the two of you get different results. It may turn out that some tiny variation—cell passage number, confluence or the company you ordered a reagent from—altered the outcome. Multiply the problem by the number of laboratories doing stem cell science, and “it is a horror story,” said Mark Noble of the University of Rochester School of Medicine in New York in an interview with StemBook.
Stem cell biologists lack clear definitions for cell types and agreed-upon protocols to make them. A variety of methods can be a good thing, allowing scientists to pursue many paths, but could hurt stem cell biology in the clinic. Stem cell standardization is a big enough issue that the American Association for the Advancement of Science held a webinar devoted to the topic on January 28, (http://www.sciencemag.org/webinar). Noble spoke alongside Amy Wagers of Harvard University and Ron McKay of the National Institutes of Health in Bethesda, Maryland. Researchers need to share the nitty-gritty specifics of their methods, the panelists said, so that others can copy them.
There are some standards in stem cell science. Testing for pluripotency by looking for teratoma formation in mice is a common method. Scientists have had time to define markers for hematopoetic stem cells, said Matthias Hebrok of the University of California in San Francisco, but marker sets for other cell types lag behind.
The variability in protocols and outcomes may be part of the growing pains of a young field. “Whenever new techniques are developed, there is a gold rush period,” Hebrok said. Scientists rush to try all kinds of experiments, and only some pass the reproducibility test.
The complexity of cell biology is also a factor; even within an organism there is natural heterogeneity among cells of the same type, McKay said during the webinar. “Working with stem cells is a little more like cooking than like building a car,” Hebrok said. And even master chefs may struggle to provide the right mix of spices to succeed with another’s recipe. Subtle changes in conditions such as the oxygen level in the incubator can throw off cellular differentiation. Some scientists split their cells with trypsin; others scrape up the cells mechanically. Mouse cell feeder layers and media with fetal calf serum add undefined, variable quantities of nutrients and signaling molecules. In addition, each cell type has its own specific requirements; there is no master recipe that works for all.
Standards will really matter when the FDA gets involved. “As we move to medical applications, it is essential that we move from poetry to engineering,” Noble said during the webinar. Cells that are ninety, or even ninety-eight, percent pure will not cut it, he told StemBook. And if the cell types used are uncertain, clinical trials might do more harm than good. For example, Noble told webinar viewers that astrocytes differentiated from progenitors in the presence of bone morphogenetic protein-4 can assist in spinal cord injury repair in rats. But one can also make astrocytes from the same precursors exposed to ciliary neurotrophic factor instead. These astrocytes not only fail to repair, they actually cause pain (Davies et al., 2008: PMID18803859). And that kind of mistake at the trial stage might engender concerns about stem cell therapies that could diminish grant monies and trial approvals, setting back the entire field.
For now, researchers rely on each other to share methodology via publication or website. Wagers frequently hosts visitors who come to learn her techniques. For those who cannot travel, the Journal of Visualized Experiments publishes video protocols. For example, viewers can watch how scientists in Doug Melton’s lab at Harvard University collect human embryonic stem cells (Chen and Melton, 2007: PMID18989406).
Researchers are also adopting technology to purify and define stem cell populations, which might limit the variation human scientists bring to procedures. At the NIH, McKay employs whole-transcriptome characterization to create detailed profiles of gene expression in embryonic stem cell lines. Wagers uses flow cytometry to purify cells based on surface markers. But even machines are not perfect, Wagers told StemBook. For example, differences in laser power can alter the sorting. “There are limitations to the technology, always,” she said.
Noble suggests that a key test for culture purity should be clonal analysis. He advocates separating stem cell cultures into thousands of single cells, and differentiating each individual clone to observe its characteristics. “It is the platinum standard of standardization,” he told StemBook, and the only way to ensure that a purified cell population is not really a mixture.
Settling on standards is no easy task. “Once you say standardization, the question is, whose standards?” Noble said. “Who gets to be the chief?” Democracy would select the most common methods, but not necessarily the best. Perhaps, he suggested, the techniques with the best therapeutic outcomes should win out.
In that case, it may be too early, now, to settle the matter. Variation is not all bad, of course, and many scientists say it is good. “Diversity gives a better chance of success,” said Dieter Egli of the New York Stem Cell Foundation in New York City. “Maybe one method will lead you to the cell type you want.”
Results matter more than recipes, said John Gearhart of the University of Pennsylvania in Philadelphia. Nature allows for plenty of “wobble,” he said, in getting from one cell type to the next. And standards that are too narrow might cause scientists to discard valuable cell types that do not fit the rules, he warned. “It can have a lot of that wobble, and you would still get the outcome you want,” Gearhart said.
Chen, A.E., Melton, D.A. (2007). Derivation of human embryonic stem cells by immunosurgery. J. Vis. Exp., 574.
Davies, J.E., Pröschel, C., Zhang, N., Noble, M., Mayer-Pröschel, M., Davies, S.J.A. (2008). Transplanted astrocytes derived from BMP- or CTNF-treated glial-restricted precursors have opposite effects on recovery and allodynia after spinal cord injury. J. Biol., 24.