“And start to be at the beginning. Accept swift working of the plan: Then, following eternal norms, You move through multitudinous forms, To reach at last the state of man.“ —Johann Wolfgang von Goethe, Faust II, Act 2

It sounds like alchemy: cells within an organism are genetically almost identical, yet they form cell types as disparate as pulsing neurons, engulfing macrophages and enzyme-secreting villus cells. Recently developed techniques appear able to prompt cells from a terminally differentiated state into one in which they not only divide indefinitely but can, in theory, become any cell type found in adults. Last year's advances in generating such cells from mice and humans have opened what could be a new era of pluripotent stem cell biology.

It is difficult to underestimate these advances; they may stimulate a paradigm shift away from the concept of the embryonic stem cells as the sole source of pluripotent stem cells. Conceivably, in a hundred years, the time between 1981 and 2007 will be considered the era of the embryonic stem cell, which supplanted a previous era of pluripotent stem cells isolated from tumours (that is, embryonic carcinoma cells) and gradually gave way to an era of induced pluripotent stem cells generated from differentiated precursors. What, then, might an era of induced pluripotent stem cells yield to? Though their applications will be valuable, I propose that the questions induced pluripotent stem cells engender will prove just as important as the cells themselves.

Asking embryonic stem cells for reprogramming secrets

The search for factors that reprogram differentiated cells has been under way for at least two decades. Embryonic stem (ES) cells harbour reprogramming elements that could be dysfunctional or absent in healthy differentiated cells, so the field's initial focus on ES cells seems logical. In early work, the introduction of two transcription factors identified in ES cells hinted that somatic cells could respond to pluripotency signals. When Nanog was expressed in cells fused with ES cells, reprogramming was much more efficient. Also, ectopic misexpression of Oct4 in adult mouse tissue caused hyperproliferation1.

Thomas P. Zwaka

Shinya Yamanaka from Kyoto University in Japan sought a core set of factors that would initiate reprogramming in mouse cells. He and his colleagues identified 24 factors essential to pluripotency in ES cells and introduced them into mouse embryonic fibroblast cells using genetically modified retroviruses2. Using a clever selection scheme, they found that some of the somatic cells were in fact reprogrammed. By a series of deductive steps, they narrowed the original set of factors to four (Oct4, Sox2, Myc and Klf4). Nanog, surprisingly, was absent.

Though remarkably similar to ES cells, the first-reported induced pluripotent stem (iPS) cells were not identical to them either in their gene-expression profile or in their behaviour. For instance, when injected into preimplantation mouse embryos, they failed to contribute to normal development and so failed the most rigorous test for pluripotency. More detailed analysis implicated incomplete reprogramming as the likely reason for this. Follow-up reports with ever-improving screening techniques have yielded more ES cell–like cells including mice in which reprogrammed cells contribute to every sort of body cell3,4,5.

Do iPS cells make ES cells less useful?

Reprogramming somatic cells into embryonic-like stem cells by expressing a defined set of transcription factors seems far simpler and more efficient than generating ES cells, whether starting from left-over preimplantation embryos or using somatic cell nuclear transfer (SCNT). It requires only a small-scale laboratory effort. There are, apart from those for standard tissue culturing, no requirements for special techniques, animals, eggs or embryos to be used in the process. The direct derivation of ES cells from preimplantation embryos requires a high level of technical skill, a large number of embryos and an array of special equipment.

Despite its simplicity, the four-factor method of reprogramming used by Yamanaka is relatively inefficient: less than one percent of treated fibroblasts acquire pluripotency. Also, in contrast to reprogramming by SCNT, the acquisition of pluripotency requires multiple days, and it is still unclear which sorts of cells can be reprogrammed. Not only does SCNT currently yield higher-quality pluripotent cells, it also possesses a significant advantage because it does not use genomic alteration to introduce reprogramming factors. Generating iPS cells requires the forced expression of tumour-promoting factors. And mice derived from iPS cells appear more prone to cancer than their normal counterparts.

Suppose, though, that iPS cells can be generated without genetic manipulation. Many labs are pursuing this goal, and it could be achieved soon. Does this mean that SCNT can be discarded, that it should become a relic like mouth pipetting? Absolutely not!

Though pressure from religious and other groups might shift the field away from ES cells and toward iPS cells regardless of scientific merit, the scientific questions SCNT can address overlap only marginally with those that can be tackled through direct reprogramming. SCNT allows the study of how epigenetic and genetic components contribute to the earliest steps of development. Introducing pluripotency factors, on the other hand, does not produce an embryo, nor does it allow one to study the events occurring as an embryo forms and the embryonic epigenome and transcriptome become established.

Applying iPS

Still, iPS cells promise several practical applications. Procedures already validated for ES cells will investigate iPS cells' potential to differentiate into functioning, specialized tissues. Working in sickle-cell anaemia, Rudolf Jaenisch and co-workers have already shown, in an elegant proof-of-concept study in the mouse, how reprogramming, tissue-specific differentiation and gene therapy can be used to cure inherited disorders6.

The clearest potential advantage of using this technology to reprogram human cells will be using it to generate disease-specific cells from many patients or to subject derivatives of reprogrammed cells to diagnostic or pharmacological tests in an individualized medicine approach. In contrast to the term 'therapeutic cloning' coined for SCNT-derived applications, this would be 'therapeutic reprogramming'. As with other biomedical research discoveries, the fields that will benefit most from these recent discoveries are probably those that seem obvious now.

Indeed, the potential applications for iPS technology are endless. An overlooked application of iPS cells would be using them to bypass the difficulty of working with species for which establishing ES cells is difficult or impossible. The ability to perform genetic manipulations would help to engineer traits such as disease resistance or greater muscle mass in domestic or threatened animals. Freezing batches of iPS cells from endangered species may also help to preserve them.

Jumping from iPS to greater reprogramming knowledge and power

Although practical applications may be the easiest to explain, iPS cells could become invaluable for studies that distinguish between epigenetic and genetic effects. Both genetic mutations and epigenetic dysregulation contribute to tumourigenesis, for example. iPS cells could help establish the contributions of these aberrations. Though SCNT-derived ES cells could also serve this purpose, the necessary resources and techniques make using them in this way less practical. Despite the remaining uncertainty concerning both iPS cells and techniques to distinguish between epigenetic and genetic effects, iPS cells can move the cancer field forward by allowing more types of experiments.

Late in 2007, Yamanaka showed that the factors that reprogram mouse cells can do the same in human cells. Reporting at the same time, Jamie Thomson and colleagues from the University of Wisconsin-Madison developed a protocol to reprogram human cells. He also used four factors, but only two (Oct4 and Sox2) were the same as those used by Yamanaka7. This highlights our poor understanding of reprogramming. Why can different sets of genes have the same outcome?

The ultimate question in the field of iPS research is whether identified sets of reprogramming factors truly represent a 'core' regulatory circuit.

The ultimate question in the field of iPS research is whether identified sets of reprogramming factors truly represent a 'core' regulatory circuit. Several lines of evidence indicate that they may not. The recent successes in this area reasonably limited study only to factors important to ES-cells' self-renewal; however, this strategy omits candidate genes not exclusively active in ES cells, and these may initiate reprogramming far more efficiently. Some of the factors discovered so far could just be bystanders of reprogramming; they could simply activate other genes that are the 'real' reprogramming factors. It has already been shown that some of these reprogramming factors are more powerful than others. Myc is useful but ultimately dispensable. In the mouse, iPS cells that are capable of contributing to the gamete are only obtained when a weak selection marker (Fbx15) is exchanged for a more predictive one (Nanog).

Ultimately, reprogramming studies will allow us to ask questions such as “What is a cellular state?” Until now, we have assumed that a differentiating cell goes through distinct, defined steps to terminally differentiate: that a synchronized cascade of specific transcription and epigenetic factors moves a cell from one state to another, and that the order of steps is critical for successful differentiation. Both SCNT and other methods to induce pluripotency suggest that these transitions are not as linear as we had thought. However, the fact that cells can 'jump' from a state of terminal differentiation to a flexible state indicates that the concept of lineage linearity can be bypassed, at least artificially in the laboratory. Nevertheless, genetic reprogramming seems to involve a series of intermediate states in which cells' pluripotency machinery reactivates over several cell divisions; in contrast, reprogramming using SCNT may not be immediate, but it is sufficient to generate an embryo. Not knowing why this is so is a fundamental problem of biology.

Paradoxically, the success of techniques to induce pluripotency indicate that converting one cell type into another may not be necessary to generate pluripotent cells at all. Once we fully understand how to establish the epigenetic state of one cell type, it might be possible to impose that state on another cell type, for example, to generate neurons or haematopoietic stem cells directly from fibroblasts without first reprogramming the cells to a pluripotent state. At present, only a few cell types can be successfully differentiated in a culture dish; 'jumping' from one cell type to another might expand this repertoire. Even if this most extreme form of reprogramming does not work, it may also be possible to program pluripotent cells to differentiate into neurons or haematopoietic stem cells in vitro without mimicking development.

The discovery of reprogramming teaches us an important lesson — cellular states can indeed change quite markedly within a narrow window of time. Given that the number of factors that can accomplish reprogramming is limited, it is almost certain that similar phenomena occur when tumours form or adapt to new requirements (such as in metastasis). Such uncontrolled reprogramming may not only promote certain events in cancer development but may also be essential for disease progression. Thus, a detailed knowledge of the events that occur during reprogramming could help answer fundamental questions about cancer stem cells.

Although it may be a decade or more before the current work on pluripotency affects work in the clinic, the progress stimulated by the pioneers in the field of reprogramming has moved cellular reprogramming from the alchemist's shelf to the forefront of research. Dust will not settle on this field for many years hence.