The birth of Dolly the sheep in 1996 produced great excitement among researchers. This first cloned mammal had been created1 by introducing the nucleus of a somatic (non-germ) cell into an egg cell from which the genomic DNA had been removed, and transferring the resulting embryo into a foster mother. One implication of this achievement was that similarly cloned embryos could be used to produce stem cells that would be genetically identical to the cells of the somatic-cell donor, so that if these stem cells, or cells or tissues derived from them, were transplanted into the donor for treatment purposes they would not be rejected by the donor's immune system. In the years that followed, this technique of somatic-cell nuclear transfer (SCNT) was successfully used to produce stem cells from cloned mouse embryos2,3, but all attempts in humans had failed — until the publication of a study in Cell by Tachibana et al.4.

The study is noteworthy for several reasons. First, it explains why all previous attempts at cloning human embryos have failed. The human egg (oocyte), and that of most mammals, is released from the ovary at the metaphase II stage of meiotic cell division. The cell resumes and completes meiosis only after it is fertilized. Removal of the meiotic spindle — the cellular structure that ensures the faithful distribution of chromosomes between dividing cells — is an integral part of SCNT. Tachibana et al. realized that this induces premature completion of meiosis in human eggs and subsequent loss of their capacity to reprogram somatic cells to a pluripotent state, which would allow them to differentiate into all cell types in the body. Crucially, the addition of caffeine to the culture medium slowed meiotic completion, ensuring the success of the authors' procedure.

The paper also shows that blastocysts (roughly 100-cell embryos) derived using this modified SCNT protocol were healthy enough to be used for generating embryonic stem (ES) cells that were genetically identical to the donor nucleus. And, notably, the authors have managed to generate these SCNT-ES cells using nuclei not only from fetal cells but also from post-natal cells. This approach could therefore be used to create cellular models of the genetic disease that a somatic-cell donor might carry.

These technical achievements, for which researchers worldwide have strived for at least a decade, should be celebrated. Nonetheless, parallel advances made in the field of stem-cell research somewhat dampen the excitement that the present paper might have received — the sort of excitement that was generated some nine years ago by Woo Suk Hwang's report of similar results, before it was discovered that those data had been fabricated5.

In our opinion, the discovery6 in 2006 that differentiated adult cells can be directly reprogrammed to a stem-cell-like state called induced pluripotent stem (iPS) cells was a more significant breakthrough for this research field. iPS cells can be generated by introducing just four transcription factors into differentiated cells of an individual, without the need for the ethically sensitive step of creating embryos from oocytes as intermediates (Fig. 1). Indeed, many laboratories now routinely generate iPS cells from patients, bypassing the practical and regulatory difficulties associated with obtaining human oocytes.

Figure 1: Generation of pluripotent stem cells in vitro.
figure 1

a, Tachibana et al.4 show that human embryonic stem (ES) cells can be generated by a technique called somatic-cell nuclear transfer (SCNT). The authors removed the meiotic spindle from an oocyte arrested at the metaphase II stage of meiotic cell division. The oocyte had been incubated with caffeine to prevent premature completion of meiosis. They then inserted a somatic cell into the enucleated oocyte. Oocyte activation and cellular reprogramming followed, leading to blastocysts from which SCNT-ES cell lines were derived. b, By comparison, the generation of induced pluripotent stem (iPS) cells involves the introduction of four pluripotency-related transcription factors into a differentiated cell to induce its direct reprogramming.

Full size image

But an intriguing question now is how similar human iPS and SCNT-ES cells are. One difference is immediately apparent. In iPS cells, mitochondria (organelles that are the main source of cellular energy), as well as all other organelles, originate from the donor cell. In SCNT-ES cells, the mitochondria are derived from the oocyte and not from the donor of the nucleus. Apart from the nucleus, mitochondria are the only organelles that contain DNA, which encodes around ten genes. This means that SCNT-ES cells might activate the immune system of an individual who is ostensibly being treated with their 'own' SCNT-ES cells and cause them to be rejected.

On the other hand, it makes these cells suitable for studying mitochondrial diseases, which are maternally inherited. However, to create their SCNT-ES cells, Tachibana et al. used nuclei from the cells of a patient with Leigh syndrome — a disorder that can be caused by mutations in a mitochondrial gene. Because mitochondria in SCNT-ES cells originate from the oocyte, these cells would not carry the same mutation, nor model Leigh syndrome. Nonetheless, the cells generated are a proof of principle that adult somatic cells can be used in human SCNT.

The present study shows that the authors' SCNT-ES cells meet the main criteria for pluripotency: they can differentiate in vitro; they express pluripotency genes; and, when injected into an immune-deficient mouse, they form a teratoma — a type of tumour that contains many different cell types. Nevertheless, other properties of these cells were not extensively explored.

For instance, although human iPS cells are known to accumulate mutations without the necessary care7, overall they are very similar to ES cells derived from normal 'surplus' human embryos obtained by in vitro fertilization (IVF) treatment, and they are genetically and epigenetically stable under careful culture conditions over long periods8. Tachibana et al., however, did not compare the efficiency of SCNT-ES cells, human IVF ES cells and iPS cells at differentiating in vitro under optimal conditions. Similarly, it is unclear whether SCNT-ES cells remain stable over time. Further investigation along these lines would be beneficial.

What this study provides is an excellent source of reference. Direct reprogramming of human iPS cells takes several weeks, whereas SCNT-ES cells are reprogrammed within a few hours by the natural factors present in the oocyte, and could in principle give rise to new offspring. A head-to-head comparison of these cell types over a long culture period would be ideal, not least to identify factors that might improve the efficiency and yield of direct reprogramming.

A cautionary note: since the paper's publication, there has been ongoing discussion about some errors, such as possible figure duplication and mislabelling, that it contains9. We therefore eagerly await experimental confirmation of Tachibana and co-authors' results by others, as well as the outcome of an investigation by Cell to determine how such errors occurred and whether they affect the study's overall conclusions (as Nature went to press, the results of this investigation had not been released). These concerns notwithstanding, the present findings are a major development, particularly for those studying human reproduction and IVF. Whether it is a game changer for research into understanding disease, regenerative medicine and drug discovery is debatable.