Since the 2006 discovery by Shinya Yamanaka and Kazutoshi Takahashi at Kyoto University, Japan, that just four transcription factors could reprogram mouse fibroblasts to a pluripotent state, induced pluripotent stem cells have created excitement among researchers who see them as more available, easier to make, and less ethically conflicted than embryonic stem cells (ESCs). Generating several iPSC lines takes about six months, compared with 18 months to obtain one ESC line, said Mahendra Rao, vice president of stem cells at Life Technologies, a biotechnology tools company. "It’s really changed the embryonic stem cell field," Rao said. "It’s made pluripotent cells available to people who don’t have to be experts."
Yet scientists in this field still face numerous challenges. One concern, which has been covered extensively on ARF, is the use of retroviral or lentiviral reprogramming vectors, which integrate into the genome. Most researchers contacted for this story feel that iPS cells reprogrammed with integrating viruses are unsuitable for cell replacement therapy, but are acceptable for cellular disease models. Nonetheless, researchers led by Rudolf Jaenisch at MIT found that residual expression of reprogramming genes in iPS cells may interfere with fully resetting the cells to an embryonic pluripotent state. When Jaenisch and colleagues removed the reprogramming genes using a Cre/Lox system, the iPS cells’ gene expression profile matched up better to that of embryonic stem cells than did iPS cells that contained reprogramming genes (see Soldner et al., 2009). This suggests that removing reprogramming factors may be important even for disease modeling. Researchers have now demonstrated the efficacy of numerous next-generation reprogramming methods that leave no trace of themselves in the cell (see ARF related news story on Stadtfeld et al., 2008; ARF related news story on Okita et al., 2008; ARF related news story on Kaji et al., 2009, Soldner et al., 2009, and Woltjen et al., 2009; and ARF related news story on Yu et al., 2009). The problem with all of these methods is that the already low reprogramming efficiency drops further, usually by a couple of orders of magnitude, causing most scientists interviewed here to continue to opt for the higher-efficiency viral methods for the time being.
Scientists also stew over the quandary of how thoroughly cells truly are reset to pluripotency. Some new research suggests that iPS cells may not be fully reprogrammed to a stem cell-like state. For example, research groups led by George Daley and Konrad Hochedlinger, both at Harvard, independently found that iPS cells can contain an epigenetic "memory" of their tissue of origin, which restricts their differentiation potential (see Polo et al., 2010 and Kim et al., 2010). Hochedlinger’s group found that continuous passaging of the pluripotent cells caused this memory to fade, while Daley’s group was able to reset these cells using serial reprogramming or chromatin-modifying drugs. On the other hand, new work by Jaenisch’s group published last month in Cell Stem Cell analyzed gene expression and epigenetic profiles of iPSCs and ESCs and concluded that there are no significant, consistent differences (see Guenther et al., 2010). This report engendered some debate, as William Lowry and colleagues at the University of California, Los Angeles, contend that some iPSC-ESC differences seen by various labs are indeed significant (see Chin et al., 2010). Lowry’s group concurs, however, that variables such as delivery of the reprogramming factors on a single vector, and high passage number (50-80 passages), reduce the observed differences between iPSCs and ESCs.
Many scientists interviewed for this story agreed that it is important to be aware of the issue of incomplete iPSC reprogramming. At the same time, few consider it a disabling problem, as it can be solved by longer passage times or better reprogramming methodologies. If the differentiation potential is restricted, that is not necessarily a bad thing, said In-Hyun Park at Yale in New Haven, Connecticut. Scientists can use cells of the desired lineage to generate iPSCs, for example, neuronal lineage cells to generate iPSCs for neuronal research. Kevin Eggan of Harvard suggested that for disease models, the question of whether iPSCs and ESCs are exactly the same might be less important than whether iPSCs can make the functional cell types of interest and whether those cells show a phenotype in a culture dish.
Recapitulating Disease in a Dish
Indeed, the latter ability is one of the crucial traits that iPS cells must possess if they are to realize their potential as disease models. So far, very few published lines have demonstrated disease features in culture. Two notable examples that did show disease traits both modeled early-onset, monogenetic disorders, that is, spinal muscular atrophy and familial dysautonomia. Many researchers interviewed for this series voiced concern that sporadic forms of disease may simply not show up in iPS cells, in part because these disorders may result from environmental factors that change the epigenetic profile of the neurons. iPS cells, which have a nearly fresh epigenetic slate and are often derived from a different tissue to boot, would not mimic these disorders.
Scientists also remain skeptical at this point about whether late-onset diseases such as AD will show defects in a cell culture. "I think the concern in the field for Alzheimer’s cells is that you would have to culture them for an enormous amount of time," said Jered McGivern, University of Wisconsin, Madison, who works with Clive Svendsen, now at Cedars-Sinai Medical Center, Los Angeles, California. Some solutions have been proposed, such as genetically modifying AD cell lines to stress or age them more quickly, or using human-animal chimeras (see Saha et al., 2009), but these suggestions still await testing. Disease modeling with iPS cells remains a frontier that scientists have only just begun to explore.
Here is another sticky issue for modeling. Can a cell culture that contains but one or two cell types ever successfully represent a complex disease? Some of these diseases, such as schizophrenia, many forms of epilepsy, and perhaps AD, are probably circuitry disorders, said Rao. "Anything you do on a dish in 2D is not going to reconstruct the normal neuronal circuitry you’d see in a three-dimensional brain," concurred James Ellis, of the University of Toronto in Canada. Disease modeling remains problematic for another reason, Ellis said. That is, scientists lack efficient differentiation procedures for creating adult-phenotype, subtype-specific cells of interest without also getting undesired cell types.
The variability of iPS lines presents one of the biggest obstacles for scientists hoping to use these lines for research, Park said. Because of natural genetic variation between one person and the next, iPS lines from different patients are hard to compare. Park said he also sees differences even among iPS cells derived from the same fibroblast line, indicating that some of the variation is due to the reprogramming process. In addition, Ellis pointed to new research showing that the expression profiles of iPSCs and ESCs are affected by the laboratory that makes them (see Newman and Cooper, 2010), suggesting that culture conditions significantly alter gene expression. All these factors may make iPS data too noisy to recognize a phenotypic difference between a single mutant line and a single wild-type line, Park said, adding, "It’s going to be difficult to generate an in vitro model of a disease."
Asa Abeliovich, of Columbia University in New York City, agreed that this variability is a key concern with using iPS cells to model disease. Abeliovich points to the importance of comparing multiple iPSC-derived lines and using statistical analysis to determine the ratio of signal to noise in the system. Without such analysis, Abeliovich said, you can’t be sure that a phenotype is real. Abeliovich emphasized this point while presenting initial data on his lab’s iPS cell lines at the International Conference on Alzheimer’s Disease in July 2010 in Honolulu, Hawai’i.
"We think what’s going to happen is that researchers will want their own lines, from their specific patient set," Rao said, adding that particularly for complex diseases such as AD, having a clinical history to correlate with the iPSC data will be important. The iPSC line variability means that having proper control lines is critical, but there’s no consensus in the field yet as to what lines will make the best controls. Many institutions try to gather control lines from a patient’s relatives, to minimize the effect of the genetic background. For monogenetic diseases, Jaenisch suggested in a paper, the most stringent control might be to genetically rescue some iPS cells, so that the only variation between rescued and mutant cells is the gene of interest (see Saha et al., 2009). iPSC variability will also make it harder for labs to compare data. Rao said it will be important for the science community to have a commonly used, standard reference line. Adding a plug, he noted that Life Technologies plans to develop such a reference iPS line.
Despite all these hurdles, most researchers remain optimistic about the potential of iPS cells to make useful disease models and research tools. "I think the opportunities that [iPS cells] present outweigh the limitations," said Selina Wray, of University College London, U.K. Rao agreed, saying, "Depending on the question you’re asking, some of these problems are not relevant. We’ve always lived with the fact that we can never have a perfect model." Induced pluripotent stem cells will be particularly valuable for disorders such as motor neuron diseases, Rao said. Because motor neurons can’t be easily cultured or genetically manipulated, scientists have never had a good model system for some of these diseases. "iPSC now offers you a way to do that," Rao said. "Stem cells have the promise of providing us entry into a black box of development that we simply had no other way to look at."