Stem cells

Low-risk reprogramming

New techniques circumvent a roadblock to the production of embryonicstem-cell-like lines from adult tissue. Such reprogrammed cell lines should be much safer to use for therapy.

Shinya Yamanaka's amazing discovery1 that cells from differentiated tissues can be reprogrammed into induced pluripotent stem (iPS) cells — cells that can potentially differentiate into any cell type — has transformed research in stem-cell biology and regenerative medicine. The breakthrough provided both a deft approach to the production of patient-specific stem-cell lines with which to study disease, and a practical means of developing large banks of stem-cell lines suitable for tissue matching in transplantation therapy. But the original protocols for producing iPS cells relied on the integration of foreign 'reprogramming' genes into the host-cell genome, a process associated with risks including mutation, dysregulation of native gene expression, and the development of cancers after iPS-cell transplantation. Four studies2,3,4,5, including two in this issue2,3, now show that iPS cells can be produced without any permanent modification to the host-cell genome.

The first-generation iPS cells were produced from culture-grown mouse and human somatic (non-germ) cells, most often skin fibroblasts1,6,7. The protocol involved the introduction into the host cell of reprogramming genes crucial for the establishment or maintenance of the pluripotent state using various lentiviral or retroviral constructs as vectors. The constructs integrated into the host genome at multiple sites.

Although in most cases the foreign genes eventually became inactive in the iPS cell lines, this was not always the case. Moreover, even if expression of the transcription factors that the genes encoded was stopped, the integrated foreign DNA remained in the host genome. This could, in principle, disrupt host genes, alter gene expression at nearby genomic loci or, if subsequently reactivated in the differentiated cells, result in these cells becoming cancerous. Indeed, chimaeric mice generated from normal cells and some iPS cell lines developed tumours8. Work with viral vectors that integrate into the host genome also left open the daunting possibility that integration and genetic modification of the host cell per se might be required for reprogramming.

Subsequently, several groups showed that, depending on the host cell type, reprogramming can be achieved using fewer foreign genes. But the goal of completely eliminating the need for genomic integration of foreign sequences remained a priority.

Given the intense activity in the field of reprogramming, many groups pursued solutions to this particular obstacle. Two groups4,5 report using non-integrating adenoviral vectors or plasmids to achieve transient expression of reprogramming factors without disturbing the host genome. But such an approach presents two immediate problems: the requirement for prolonged expression of the pluripotency factors to achieve reprogramming, and the difficulty of repeatedly delivering the full complement of factors using a different vector for each one.

To address the first problem, the authors4,5 used hepatocytes — which, compared with other cell types, are more amenable to both reprogramming and infection with adenoviruses — and introduced the genes into the cells repeatedly over a period of days. The second problem was solved in one study5 by borrowing a clever trick from the foot-and-mouth virus. By inserting a virally derived oligopeptide sequence called 2A as a spacer between four reprogramming genes, the researchers made a multiprotein expression vector.

The key feature of the 2A sequence is its ability to undergo self-splicing and be removed from a peptide undergoing translation, possibly through a mechanism in which the ribosome skips over one codon without forming a peptide bond, thus allowing the production of several peptides from one transcript. Using this approach, several reprogramming genes can be introduced and efficiently expressed using a single adenoviral or plasmid construct.

Both groups obtained mouse iPS cells that express cell-surface markers and genes characteristic of embryonic stem cells, and that could undergo differentiation in vitro. In vivo, these cell lines contributed to teratomas (benign tumours containing cells from various differentiated tissues) in chimaeric mice, although their ability to form germ cells was not assessed. The main drawback of this approach was its relative inefficiency: at least 100-fold fewer iPS colonies were obtained than with the retroviral or lentiviral vectors. Given that the production of iPS cells from human cells is generally less efficient than from mice, it is questionable how practical it would be to use these methods4,5 for human cells. But the studies established one point clearly — viral integration is not necessary for reprogramming.

Two other elegant studies2,3 provide an alternative, more efficient, strategy that involves virus-free integration of reprogramming genes, followed by their removal. Woltjen et al. (page 766)2 and Kaji et al. (page 771)3 combined powerful technologies, developed independently, to overcome many of the difficulties others encountered in attempting virus-free reprogramming. These groups also made use of the virally derived 2A-peptide sequence5 to create multi-protein expression vectors incorporating all of the reprogramming genes. Instead of retroviruses or plasmids, however, they used the piggyBac transposon/transposase gene-delivery system. This vector can easily integrate into the genome. But more importantly, the integrated DNA can also be removed from the genome — through transient expression of the transposase enzyme — in a highly efficient and seamless fashion, leaving no trace of the integration in the genome of the iPS cells. The use of the 2A peptide is crucial, not just because it allows delivery of all of the required reprogramming genes in a single construct, but also because it makes complete excision of the foreign constructs much easier. What's more, the efficiency of this approach2,3 is much higher than that of transient transduction of cells using adenoviral vectors4,5.

Kaji et al.3 also generated chimaeric mice using their iPS cells and found that these cells contributed to tissues derived from all three embryonic germ layers. The researchers do not mention, however, whether the animals could give rise to iPS-cell-derived offspring — the gold-standard test of pluripotency in mice. Moreover, they did not extensively characterize the human iPS cells, although these cells did have the expected features of pluripotent stem cells.

Together, these studies2,3,4,5 remove a major barrier to the generation of iPS cell lines that are safe for clinical use, showing beyond doubt that transient expression of reprogramming factors in somatic cells is sufficient to reset their gene expression to the pluripotent state. The piggyBac technology, in particular, will find broader use for transiently introducing genes — such as those encoding 'reporter' proteins or master regulatory transcription factors — into embryonic stem cells and subsequently removing them. It could, in addition, be applied to the reprogramming of one differentiated cell type into another, such as the reprogramming of pancreatic exocrine cells into insulin-producing islet cells9, rather than reverting a cell back to an embryonic-like state.

It remains to be seen whether alternative reprogramming methods currently under development — such as reprogramming by simply exposing differentiated cells to small molecules, either alone or in combination with gene introduction — will prove more efficient than the techniques described so far. And crucial questions about the equivalency of human iPS cells to embryonic-stem-cell lines remain: do these cell lines have the same developmental capacity as embryonic stem cells, and will they prove to be stable genetically and epigenetically? Rapid progress in this exciting field hints that the answers to these questions will soon become clear.


  1. 1

    Takahashi, K. & Yamanaka, S. Cell 126, 663–676 (2006).

  2. 2

    Woltjen, K. et al. Nature 458, 766–770 (2009).

  3. 3

    Kaji, K. et al. Nature 458, 771–775 (2009).

  4. 4

    Stadtfeld, M., Nagaya, M., Utikal, J., Weir, G. & Hochedlinger, K. Science 322, 945–949 (2008).

  5. 5

    Okita, K., Nakagawa, M., Hyenjong, H., Ichisaka, T. & Yamanaka, S. Science 322, 949–953 (2008).

  6. 6

    Takahashi, K. et al. Cell 131, 861–872 (2007).

  7. 7

    Yu, J. et al. Science 318, 1917–1920 (2007).

  8. 8

    Okita, K., Ichisaka, T. & Yamanaka, S. Nature 448, 313–317 (2007).

  9. 9

    Zhou, Q. & Melton, D. A. Cell Stem Cell 3, 382–388 (2008).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Pera, M. Low-risk reprogramming. Nature 458, 715–716 (2009).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.