Risks inherent to mitochondrial replacement.

T he UK Parliament recently debated and then approved legislation to allow mitochondrial replacement (MR) to be used in the clinic. However, we are concerned that some of the science of MR has been misunderstood or otherwise given only fleeting consideration. We set out our arguments below and offer a way forward to ensure that MR can safely deliver the health benefits it promises for those suffering from mitochondrial-related diseases. Recent innovations that enable mitochondrial DNA (mtDNA) mutations to be eliminated from the germline, by replacing mutated mitochondria within an oocyte with mitochondria from a healthy donor female [1–3], offer hope for the eradication of several debilitating and lethal mitochondrial diseases. The potential for clinical application of MR has received widespread support [4,5], but has also provoked safety and ethical concerns from the public and biomedical practitioners [4,6]. In addition to currently addressed safety concerns related to technical details of the procedures [7], a further safety concern exists that cannot be easily addressed by methodological refinements. Embryos produced by all variants of MR (pronuclear transfer, maternal spindle transfer, polar body transfer) will acquire genetic material from three different individuals (nuclear DNA from the prospective parents and mtDNA from a donor female), and some of these novel combinations of genetic material may not be fully compatible with one another (i.e. may be mismatched). For example, various combinations of donor mtDNA and recipient nuclear genomes have experimentally been shown to negatively affect offspring health and fitness in vertebrate and invertebrate models, even though the donated mitochondria were putatively healthy [8]. This evidence has, however, been suggested to have low relevance to humans [3,7,9,10] for three proposed reasons (Box 1). Here we address each of those reasons and explain why none of them refute compellingly the potential for mitochondrial–nuclear (mito-nuclear) mismatches to affect the outcomes of MR in humans.