When Shinya Yamanaka of Kyoto University reported his transformation of cultured mouse skin cells into a state approximating that of embryonic stem cells1, he was met with plenty of scepticism. Other scientists hadn't anticipated that such a feat was possible. “Nobody else was even close to doing the same experiment,” says Richard Young of the Whitehead Institute in Cambridge, Massachusetts. “That was a very special breakthrough.”

By inserting just four genes — Oct4, Sox2, Klf4 and Myc — into fibroblasts (cultured skin cells), Yamanaka's group had achieved the biological equivalent of making water flow uphill. The resultant induced pluripotent stem (iPS) cells proliferate indefinitely in culture and differentiate into all the tissues necessary to generate a live mouse2,3.

From mouse to human

Scientists raced to understand the process in mice and to make it happen in human cells. Last November, less than 16 months after Yamanaka's initial publication, his group4 and another led by James Thomson5 at the University of Wisconsin–Madison simultaneously reported the generation of iPS cells from human fibroblasts. (In contrast, it took 17 years from the first publications describing the production of embryonic stem (ES) cells from a mouse embryo6,7 for the equivalent research in human cells8 to be published.) Thomson's reprogramming mix differs slightly from Yamanaka's cocktail: it includes Oct4, Sox2 and Nanog, which encode three well-established pluripotency-associated transcription factors, along with a fourth, Lin28, that encodes an RNA-binding protein. That at least two combinations of factors work suggests some flexibility in the process. At least two other groups had papers submitted before Yamanaka's and Thomson's papers were published, showing that the techniques work in multiple hands9,10.

The ten-year head start human ES cells got on human iPS cells has effectively shrunk to zero, according to James Thomson. Credit: Jessica Kolman

Thomson says he hired Junying Yu (first author on their November 2007 Science paper) as a postdoc in 2003 to attempt reprogramming, thinking it would be a 20-year project. “We never just took those three genes and tried them in the first place, which would have saved us five years of work,” he says. “We couldn't believe it would be that easy.”

The fact that making iPS cells does not pose the technical and ethical challenges of working with eggs or embryos is drawing large numbers of researchers into the field and speeding up reprogramming research. “This is definitely the hot thing right now,” says Melina Fan, executive director of Addgene, the Cambridge, Massachusetts–based nonprofit repository that distributes both Thomson's and Yamanaka's viral vectors for the cell-reprogramming genes. As of 17 April, she says, there have been 704 requests from 178 labs at 142 institutions for Thomson's vectors; 514 requests from 131 labs at 113 institutions for Yamanaka's human iPS cell vectors; and over 1,500 requests from 232 labs at 215 institutions for Yamanaka's murine iPS cell vectors. The statistics speak for themselves. Although the Thomson and Yamanaka stem cell plasmids make up only 0.2% of Addgene's total collection, they've accounted for over 10% of Addgene's total plasmid requests since the beginning of 2008, Fan says.

One thing researchers want to know is whether only certain cell types can be reprogrammed. Liver and stomach cells have now been converted to pluripotency11. In April, the Whitehead Institute's Rudolf Jaenisch and colleagues reported their success in generating iPS cells from mature B lymphocytes, each of which makes a unique antibody because of an ability to shuffle its DNA. The genomes of fully differentiated cells have been uniquely, characteristically and irreversibly rearranged and thus allow investigators to show unambiguously that iPS cells need not arise from rare resident stem cells that might be more amenable to reprogramming12.

“Biologically there's no difference” between murine iPS and ES cells, says Jaenisch. Both can generate all the tissues in a mouse. Human iPS cells have not been as rigorously demonstrated to be quasi-equivalent to ES cells, and they won't be, because doing so would require generating human babies or foetuses. Such experimentation is irrelevant anyway, says Douglas Melton, director of the Harvard Stem Cell Institute in Cambridge, Massachusetts, who has derived multiple human ES cell lines. “Nobody's trying to make people.”

Thomson — who, as the first to derive human ES cells, should know — emphasizes that ES cells themselves are an artefact of culture. “Should we spend a lot of time arguing about whether this new tissue-culture artefact is identical to an old tissue-culture artefact?” he asks. What matters is what you can do with the cells.

Ken Chien, a physician-scientist at Massachusetts General Hospital, in Boston, and at the Harvard Stem Cell Institute, is excited by the possibility of being able to simultaneously study iPS cells derived from a particular person and monitor the course of disease in that person. That could reveal powerful new biomarkers and predictors. But like ES cells, he warns, iPS cells present a fair amount of variability. He and Melton recently showed that certain ES cell lines are much more likely to differentiate into, say, cardiac cells than are others13.

ES cells have been around longer and are much better characterized, and the iPS cells still warrant further tinkering to counter a proclivity toward tumourigenicity. Moreover, somatic cell nuclear transfer, also called therapeutic cloning, generates an embryo, so it can address questions about early development that direct reprogramming cannot. Nonetheless, the enthusiasm with which the highest-tier ES cell scientists have turned to reprogramming speaks volumes.

From doing to using

Three applications for iPS cells are in the cards. First, the relatively easily generated iPS cells are ready to be used right now, as ES cells are, for studying cells' differentiation and comparing differences between normal and diseased cells. Second, they should soon be available for drug screening, so that assays once possible only in animals can be performed routinely in human cells.

No one doubts that iPS cells will eventually be generated from the cells of individuals with known medical history. That was the main advantage claimed for somatic cell nuclear transfer, a technically and ethically challenging procedure that has yet to be achieved in humans. For generating person-matched cells, iPS cells may be not only easier to use but perhaps superior, as they would share both nuclear and mitochondrial DNA with the original patient, whereas cells derived by somatic cell nuclear transfer carry only the same nuclear DNA.

The third application, which is more discussed but farther in the future, is regenerative medicine: the production of partially or fully differentiated, immune-compatible (and perhaps gene-repaired) tissues to put back into patients. Last year Jaenisch's group successfully treated transgenic mice carrying the human gene for sickle-cell anemia by giving them haematopoietic stem cells derived from those mice's gene-repaired iPS cells14. In a collaboration with Jaenisch15, a team led by Harvard neuroscientist Ole Isacson recently created dopaminergic neurons from iPS cells and put them into the brains of rats suffering from an induced version of Parkinson's disease. The rats' physical performance improved and, later, their brains showed successful engraftment of the dopaminergic neurons.

Losing the virus

Before iPS cells can be used for regenerative medicine in humans, a couple of major issues have to be resolved. The most immediate one concerns the retroviral vectors currently used to introduce the pluripotency genes into the cell's genome. Retroviruses integrate randomly into chromosomes, sometimes in copy numbers as high as 15 or 20 per genome. The insertion is permanent even though the retrovirus and its cargo gene are needed only temporarily. As reprogramming proceeds, the cell's own silenced pluripotency program whirls into action and the added genes are silenced.

The risk is that the pluripotency genes carried by the virus may reactivate when the cells begin traveling down their differentiation pathways. Jaenisch, who started his career studying viral silencing, says the enzyme that silences retroviral genes in the pluripotent state is different from the enzyme that maintains silence during differentiation, and the latter is a bit more error prone.

Make one mistake with Myc, a gene with a powerful cancer-causing potential, and there's a problem. Up to a third of the mice produced from Myc-containing iPS cells have had tumours associated with the gene's reactivation. Several groups have shown that iPS cells can be generated without Myc, albeit at lower frequencies16,17,18. But with so many cells readily available, that's not a showstopper, says Thomson. “You get enough.”

Another problem with retroviral vectors is insertional mutagenesis. Random insertion of retroviruses into the genome carries the risk of accidentally turning on or off some key gene inappropriately, perhaps later in development.

Can virus-free iPS cells be made? “It's going to happen sooner than people think,” answers Young. Working with Jaenisch, Young found that Oct4, Sox2 and Nanog — three of the four proteins encoded by the retroviral vectors Thomson used — cluster together to coordinately turn on their own genes, which are normally locked down in differentiated cells17. Once expression of the endogenous genes for these key transcription factors reaches a tipping point, the external genes are no longer required, Young says.

In mice, this tipping point occurs roughly 10–12 days after the vectors are introduced, as demonstrated by Jaenisch's team19 and another team led by Harvard's Konrad Hochedlinger20. Both groups used inducible viral vectors that served as combined starter guns and stopwatches, allowing the precise timing of sequences of molecular events involved in pushing the cells being reprogrammed past the point of no return.

Hochedlinger's group and many others are looking for ways to jump-start the expression of endogenous pluripotency genes by transiently delivering the gene products to target cells. One method, he says, might employ nonintegrating adenoviruses whose gene payload, after inducing pluripotency and proliferation, would be diluted.

Another approach might be to replace viral vectors by directly introducing the required proteins with certain amino acid signaling sequences attached to them (such as HIV's Tat epitope) that would permit the proteins to penetrate membranes. Hochedlinger is also collaborating with Sheng Ding, from the Scripps Institute in La Jolla, California, in a systematic search for small nonprotein molecules that mimic the activity of the critical transcription factors.

Using potential starter cells that require fewer pluripotency transcription factors might help, Hochedlinger says. For example, some neurons, although they are not particularly easy to get or maintain in culture, express a bit of Sox2 naturally.

Like a new computer operating system, iPS cells are muscling into the field as their radically fewer barriers to entry, compared with those for ES cells, accelerate the pace of research. The ten-year head start human ES cells got on human iPS cells has effectively shrunk to zero, says Thomson, because so much of the legacy of ES cells — reagents, culture media, hands-on expertise and experimental history — is transferable to iPS cells. Tissue culture is simpler than embryology. Skin is cheap. Creating iPS cells that are innately immunologically compatible with the patient from whose cells they were derived presents an attractive alternative to obtaining human eggs, executing difficult nuclear-transfer protocols and destroying embryos. With the ethical clouds hanging over those procedures lifting, anxieties about funding are receding. A cadre of talented young investigators trained on ES cells and ready to surpass their mentors is chafing at the bit. As a result of this ferment, the convergent view of numerous leaders in the field is that the retroviral delivery problem will be solved within a year or so.

Then it will be on to the second challenge — perhaps the greater one. The goal of all this work, says Melton, is not just to walk a cell up to the pinnacle of pluripotency. The goal is to then get it to roll down a specific lineage ridge to the desired differentiation valley.