News & Views | Published:


Replacing the cell's power plants

Nature volume 540, pages 210211 (08 December 2016) | Download Citation

Nuclear DNA from human eggs that harbour mutations in the DNA of organelles called mitochondria has been successfully transferred to donor eggs, bringing the prospect of therapy for mitochondrial diseases a step closer. See Letter p.270

A revolution in DNA-sequencing technology in the past few years has allowed relatively easy identification of the underlying causes of many genetic diseases. However, preventing transmission of disease-causing defects remains a formidable challenge. Our cells contain two genomes — one in the nucleus and another much smaller, semi-autonomous one in organelles called mitochondria, which are essential for the production of energy. Diseases caused by mutations in mitochondrial DNA (mtDNA) affect about 1 in 5,000 people1, have an extraordinarily broad spectrum of symptoms and are currently untreatable. Kang et al.2 show on page 270 that it is possible to prevent harmful mutations in human mtDNA from being transmitted to offspring, using a mitochondrial replacement technique.

We inherit all of our mitochondria, and so all of our mtDNA, from our mothers. Female eggs carry a few hundred thousand copies of mtDNA, and most cells in an adult contain hundreds to thousands of these copies. In individuals with mtDNA disease there is usually a mixture of normal and mutated mtDNAs, and the severity of disease generally correlates with the proportion of mtDNA copies that carry the disease-causing mutation.

This proportion can vary widely between eggs from the same woman. One way to avoid the transmission of mutations is to combine in vitro fertilization (IVF) with genetic diagnosis to identify and implant only embryos harbouring normal mtDNAs. Although this approach has been successful3, it is not always possible to recover suitable embryos.

Another strategy might be to swap the mitochondria containing mutated mtDNA for healthy ones (Fig. 1). This is the idea behind mitochondrial replacement techniques (MRTs). In one MRT, known as meiotic-spindle transfer, nuclear DNA from the egg of a 'carrier' mother harbouring mutated mtDNA would be transferred to a donor egg that lacks a nucleus and contains only normal copies of mtDNA. This egg would then be fertilized in vitro and implanted. A variation on the theme is pronuclear transfer, in which the nuclei from the sperm and egg would be transferred to a nucleus-free donor egg immediately after fertilization, before the two have fused. In both cases, the reconstituted embryo would carry the same genes as its biological parents, except for the few on mtDNA that come from an unrelated donor.

Figure 1: Putative techniques for mitochondrial replacement.
Figure 1

Mutations in the DNA of organelles called mitochondria cause disease. A possible therapy would involve replacing mutated mitochondrial DNAs (mtDNAs) with healthy ones. One such therapy would be transfer of pronuclei — sperm and egg nuclei post-fertilization, before the two fuse — from an egg that harbours some mutated mtDNAs into a donor cell that has healthy mtDNA, from which the nucleus has been removed. In another technique, known as meiotic-spindle transfer, nuclear DNA from an egg harbouring some mutant mtDNAs would be transferred into a nucleus-free egg, which would subsequently be fertilized. Kang et al.2 provide evidence that meiotic-spindle transfer can produce healthy embryos at the blastocyst stage of development that are free of mutant mtDNA.

The group that conducted the current study has previously performed proof-of-principle tests to show that meiotic-spindle transfer can produce healthy macaque offspring4 and human embryos that develop normally up to the blastocyst stage at five days of development5. Earlier this year, another group had a similar success with pronuclear transfer6. However, these studies used carrier eggs that contained normal mtDNA only.

In the current study, Kang et al. turned to eggs from women carrying single-nucleotide mtDNA mutations. The authors successfully used meiotic-spindle transfer to produce embryos that developed to the blastocyst stage and that carried a virtually undetectable proportion of mutant mtDNAs. So, it seems that the precise timing of nuclear transfer (before or after fertilization) is unimportant. However, because pronuclear transfer has not been attempted using eggs from women carrying mutant mtDNA, a direct comparison would be required to confirm this speculation.

Concerns have been raised about possible nuclear–mitochondrial incompatibilities resulting from MRT — the idea that certain combinations of nuclear and mitochondrial genomes could have adverse effects on cell fitness. However, evidence to support such concerns is based on studies of inbred mice7,8, which have no genetic variation in their nuclear genomes. Human populations, by contrast, are highly heterogeneous, and so paternal genes find themselves in a potentially novel mtDNA environment with each generation. Thus, it seems unlikely that any kind of tight coupling would have evolved between nuclear and mitochondrial DNA in humans. Indeed, Kang and colleagues show that embryo development is independent of the genetic distance between the mtDNA of the carrier and that of the donor, corroborating observations5 in a macaque MRT experiment that used donors and carriers with highly divergent mtDNAs.

Another concern is the fate of the 2% or so of mtDNA copies that hitch-hike with the nucleus from the carrier egg into the reconstituted embryo. To investigate the potential consequences of such hitch-hiking, Kang et al. analysed embryonic stem cells that they isolated from the reconstructed embryos and grew in culture. Their analyses showed that, in rare instances, donor mtDNAs can be outcompeted by even small proportions of carrier mtDNA after a large number of cell divisions — in line with the results of two previous studies6,9.

Two possible mechanisms could be in play here: genetic drift (random changes to genetic make-up that occur over generations) and replicative advantage. If the changes seen in mtDNA proportions are due entirely to genetic drift, the probability that a rare mutation in a mixed mtDNA population will become predominant is directly proportional to its initial frequency — as has been demonstrated in vivo for cells in crypt structures in the colon, which are continuously regenerated from a stem-cell population10. This probably would not be much of a worry for therapeutics, because most cells would lose the rare mtDNAs carrying the mutation.

But Kang and colleagues' study shows that, in some cases, the carrier mtDNA could have a replicative advantage over mtDNA from the donor, probably owing to sequence variation in the region of mtDNA that regulates the replication rate of the mitochondrial genome. This phenomenon could perpetuate a systematic return to a majority of carrier mtDNA. As the authors suggest, it should be possible to select donors on the basis of relative efficiencies of mtDNA replication when studying the process in model systems. But the problem cannot be directly studied in human embryos, because mtDNA replication is stalled until after the blastocyst stage. It would, however, be useful to investigate the phenomenon in other mammals, where it is possible to study replication of mtDNA throughout early embryonic development.

Aside from concerns about the fate of hitch-hiking mtDNA from the carrier egg, what is holding back widespread application of MRTs? The procedures are technically challenging and so not likely to be licensed widely in IVF clinics. It is also currently unclear how many carrier or donor eggs would have to be used to reliably produce healthy reconstructed embryos for implantation. A recent report of a baby boy born after an MRT (see Nature; 2016), performed by a US physician in Mexico, made headlines, highlighting the need for carefully supervised clinical trials. Legislation has so far legalized the procedure in a single jurisdiction — the United Kingdom. Until that changes, it seems unlikely that many women will benefit.



  1. 1.

    et al. Ann. Neurol. 63, 35–39 (2008).

  2. 2.

    et al. Nature 540, 270–275 (2016).

  3. 3.

    et al. J. Med. Genet. 43, 244–247 (2006).

  4. 4.

    et al. Nature 461, 367–372 (2009).

  5. 5.

    et al. Nature 493, 627–631 (2013).

  6. 6.

    et al. Nature 534, 383–386 (2016).

  7. 7.

    et al. Nature 535, 561–565 (2016).

  8. 8.

    et al. Nature Genet. 38, 1261–1268 (2006).

  9. 9.

    et al. Cell Stem Cell 18, 749–754 (2016).

  10. 10.

    , & Nature Genet. 16, 93–95 (1997).

Download references

Author information


  1. Eric A. Shoubridge is in the Department of Human Genetics, Montreal Neurological Institute of McGill University, Montreal, Quebec H3A 2B4, Canada.

    • Eric A. Shoubridge


  1. Search for Eric A. Shoubridge in:

Corresponding author

Correspondence to Eric A. Shoubridge.

About this article

Publication history




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.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing