The need to understand the underlying mechanisms associated with mitochondrial therapies in assisted reproduction before further clinical trials are performed

Abstract

Over a number of years, there has been growing interest in the introduction of more invasive ARTs, such as nuclear transfer, otherwise referred to as mitochondrial donation, and mitochondrial supplementation/transfer into clinical medicine. They have been proposed to overcome repeated failed fertilization or developmental arrest or to prevent carriers of mitochondrial DNA disease from having affected children. These technologies require considerable manipulation of the oocyte, which can affect its epigenetic programming that was established as it grew and developed into a fertilizable oocyte. Consequently, when a nucleus is transferred into an enucleated oocyte or pronuclei are transferred into an enucleated zygote, the nucleus must adapt to its new cytoplasmic environment in readiness for the waves of DNA demethylation and methylation that take place during preimplantation development. As a result, some key developmental gene networks are affected. Additionally, these approaches also affect patterns of mitochondrial DNA inheritance, with some embryos and offspring possessing mitochondrial DNA carried over into the oocyte with the nucleus, as well as the mitochondrial DNA from the donor oocyte. Similar outcomes result from the addition of extra mitochondrial DNA into oocytes through mitochondrial supplementation. We provide a background as to how these technologies evolved and discuss recent outcomes associated with clinical work so far undertaken within these approaches and their consequences for the offspring. We conclude that these technologies are not simply replacing or replenishing defective ooplasms with new or extra mitochondria but rather induce a series of genomic and epigenomic events that we do not yet fully understand. To our minds, these issues should be first addressed before clinical trials are continued.

Introduction

Mitochondrial donation to potentially overcome the transmission of mitochondrial DNA (mtDNA) disease and the use of mitochondrial supplementation to resolve reproductive disorders associated with gamete dysfunction have received increasing attention. However, these invasive ARTs can result in extensive reconstruction and remodelling of the oocyte. Critically, they can perturb the interactions between the nuclear and mitochondrial genomes that contribute to the offspring’s genetic identity (Chaffer, 2023). Therefore, a full understanding of the consequences of the processes is required before these ARTs are implemented on a larger scale into clinical medicine. This is especially necessary given the limited number of large animal models that have been generated in pre-clinical testing, which is deemed to be essential prior to clinical trials. We focus specifically on the science and the readiness of these technologies for clinical application using knowledge gained from humans and large animal models, which are regarded as the most appropriate pre-clinical models (Larsen and Rolin, 2004; Perleberg et al., 2018).

Supplementing the cytoplasm

In the early 1990s, the introduction of ICSI revolutionized infertility treatment (Esteves et al., 2018). It was originally designed to help overcome poor sperm quality, but has become an ever-increasing first-line tool to resolve male- and female-factor infertility. In the late 1990s, this approach was extended by introducing extra cytoplasm donated from a younger woman’s oocyte into the patient’s oocyte as ICSI is performed, namely cytoplasmic transfer. It was developed to promote fertilization and overcome early embryonic developmental arrest that is common in some older patients and resulted in the birth of one child (Cohen et al., 1997) with subsequent births leading to a total of 17 babies from 13 couples (Chen et al., 2016). At the time, it could have been regarded as a sensible approach given that the ageing cytoplasm can lead to late-stage aneuploidy and other developmental abnormalities (Verdyck et al., 2023). However, it was not clear which cytoplasmic factors improve the quality of the oocyte and is, thus, a blunt rather than refined tool. More critically, two populations of mtDNA can be transmitted to the children (heteroplasmy), one from the patient’s oocyte and one from the oocyte donor (Brenner et al., 2000). This contravenes the normal pattern of maternal mtDNA inheritance (Giles et al., 1980), which is from the mtDNA present in the oocyte at fertilization. It has been described as ‘three-parent IVF’ and was subsequently prohibited in many jurisdictions, including the USA through the Food and Drug Administration and the UK by the Human Fertilization and Embryology Authority (Johnson, 2016).

Other approaches have focussed on the isolation of purified populations of mitochondria (El Shourbagy et al., 2006), with some arguing their addition to the oocyte alone would increase energy production and promote developmental outcome (Tilly and Sinclair, 2013). However, the choice of extra mitochondria has been contentious. Mitochondria isolated from the patient’s own fertilizable oocytes would be counter-productive since clinical protocols emphasize the use of all mature oocytes for clinical outcomes. Mitochondria have been isolated from the patient’s cumulus cells and used to generate live births (Tzeng et al., 2004). They have also been isolated from putative oogonial stem cells resident in the patient’s ovary, and through patented technology, namely AUGMENT™ (Chappel, 2013), have led to live births (Morimoto et al., 2023). The effectiveness of AUGMENT™ has been weakened by clinical trials that were aborted due to their being no perceived effect (Labarta et al., 2019). This raises issues about the design of such trials, which might be based on perceived mechanisms of action rather than on solid scientific validation before trialling begins.

Exchanging the cytoplasm

The most invasive tool is nuclear transfer, where nuclear genetic material is transferred into the cytoplasm of an oocyte that has had its own nuclear genetic material removed. It was first demonstrated when somatic nuclei were transferred to generate frogs, i.e. somatic cell nuclear transfer (SCNT) (Gurdon et al., 1958). It was furthered when live sheep were generated using nuclei from embryonic (Campbell et al., 1996) and somatic (Wilmut et al., 1997) cells.

Over the last 20 years, there has been considerable interest in the use of ‘oocyte’ or ‘zygotic’ nuclear transfer to overcome poor oocyte quality (Subira et al., 2025). Several approaches have been proposed, which include transfer of the germinal vesicle (GVT) (Zhang, 2015), the metaphase II spindle (MST) (Costa-Borges et al., 2023), and pronuclei (PNT) (Zhang et al., 2016). Interestingly, an early application of PNT was to generate heteroplasmic mice to understand the segregation of mtDNA in embryos (Meirelles and Smith, 1998) and offspring (Meirelles and Smith, 1997). Other approaches, such as polar body I or polar body II nuclear transfer (Tang et al., 2019), are in early stages of development. In each case, the nucleus or nuclei are transferred to an oocyte or zygote of the same stage of development. Importantly, with GVT, a diploid nucleus (Zhang, 2015) is transferred, whilst a haploid nucleus is transferred with MST (Costa-Borges et al., 2023). In the case of PNT, both pronuclei can be transferred to an enucleated zygote, or the maternal only pronucleus is transferred to a donor zygote that had been fertilized by the patient’s sperm (Subira et al., 2025). With GVT and MST, fertilization is performed in the reconstructed oocyte, whilst this has already taken place with PNT. Nevertheless, genomic recombination takes place post-transfer in each approach.

The role of the two genomes

We understand that the nuclear genome defines the vast majority of our characteristics. However, the mitochondrial genome plays a significant role in our well-being. Its genes encode 13 of the subunits of the electron transfer chain along with 22 tRNAs and 2 rRNAs that reside in mitochondria (Anderson et al., 1981). However, the vast majority of genes associated with mitochondria, including the electron transfer chain, which generates cellular ATP through oxidative phosphorylation (OXPHOS) (Pfeiffer et al., 2001), are encoded by the nuclear genome. Nevertheless, varying degrees of mutation or deletion to mtDNA, also referred to as heteroplasmy, or depletion of mtDNA to low levels, can result in some severe and lethal diseases, for which careful clinical management is required, especially as therapies are still in development (Gorman et al., 2015). This situation provides the rationale for the development of the nuclear transfer tools, misnamed as ‘mitochondrial donation’, to overcome the transmission of mtDNA disease. In addition, we are beginning to understand that mtDNA might define more of our traits since specific groupings, mtDNA haplogroups, characterize our common origins and are associated with protection from or predisposition to certain diseases (Wallace et al., 2003). These include type II diabetes (Fuku et al., 2007; Shen et al., 2022), Parkinson’s disease (Liou et al., 2016; Marom et al., 2017), and osteoarthritis (Blanco et al., 2018; Koo et al., 2019). In a reproduction context, they also influence human sperm motility (Ruiz-Pesini et al., 2000); the size of the human ovarian reserve (May-Panloup et al., 2014); and fertilization and embryo development rates, oocyte mtDNA copy number, and reproductive outcomes, including litter size in pigs (Tsai et al., 2016; St John and Tsai, 2018). Consequently, both genomes appear to contribute to our phenotypic traits individually or collectively.

Importantly, it appears that cooperation between the two genomes is critical at any stage of development, as shown in Fig. 1, where a number of factors can influence one genome that in turn affects the other. For example, and in the context of this discussion, mitochondrial metabolism generates by-products that act as co-factors to regulate DNA methylation of both genomes (Haseeb et al., 2014; Lee et al., 2017; Sun et al., 2018). This co-operative state, namely ‘genomic balance’ (St John, 2019) (Fig. 1), is likely essential for effective cellular function.

Figure 1.

Genomic balance. Throughout development, cells establish genomic balance through continual flow of regulatory information between the nuclear and mitochondrial genomes. The nuclear genome contributes by mediating epigenetic regulation, e.g. the levels of DNA methylation, to control gene expression. Other factors include rearrangements (mutations and deletions) and copy number variants. The mitochondrial genome contributes by modulating mitochondrial DNA (mtDNA) copy number, and through haplotypes and rearrangements, all of which regulate cell metabolism. Some by-products of metabolism are co-factors that modulate DNA methylation of both the nuclear and mitochondrial genomes that regulate gene expression of the nuclear genome and potential replication of the mitochondrial genome (Sun et al., 2018). Indeed, DNA demethylation of the nuclear-encoded mtDNA replication factors appears to be in synchrony with DNA demethylation of the mitochondrial genome and increases in mtDNA copy number. It does not appear that the changes in mtDNA methylation relate to transcription of the mitochondrial genome. Black arrows in the direction of either genome reflect the impact of a specific factor or a series of factors on that genome, whilst black arrows pointed away from a genome indicate outputs from that genome. Dashed green arrows indicate that the by-products of mitochondrial metabolism affect both genomes. The figure is reproduced and modified from (St John, 2019) under an open access Creative Commons CC BY 4.0 license.

Co-dependence of the two genomes

The relationship between the nuclear and mitochondrial genomes in the mature oocyte is very different to that found in somatic cells. On a nucleotide per nucleotide basis, the human haploid metaphase II oocyte contains 3234.83 Mb of nuclear DNA and >200 000 copies of mtDNA, which equate to ∼3313.8 Mb of mtDNA (St John, 2019). Therefore, just prior to fertilization, mtDNA and nuclear DNA are near equal genomic contributors. Indeed, from the primordial germ cell onwards, a relationship is built between the two genomes during development. As the highly DNA methylated primordial germ cells initiate differentiation into metaphase II oocytes, they undergo waves of global DNA demethylation through TET-mediated hydroxylation (Gkountela et al., 2015; Zhu et al., 2021). However, by the germinal vesicle stage in the humans, CpG methylation is reset (Yu et al., 2017). Concomitantly, mtDNA copy per cell increases from 1400 (Floros et al., 2018) to ∼250 000 (Santos et al., 2006) (Fig. 2). From then until metaphase II, CpG methylation remains stable (Ivanova et al., 2020) and mtDNA content is marginally refined (Barritt et al., 1999).

Figure 2.

Mitochondrial DNA (mtDNA) replication during development. Throughout development, there are large-scale changes in mtDNA copy number, e.g. during oogenesis and preimplantation development. At the blastocyst stage, replication is initiated only in the trophectoderm. The inner cell mass cells continue to reduce copy number and establish the ‘mtDNA set point’, the founder population of mtDNA which, when replicated, populates differentiating cells with sufficient mtDNA to match their needs for OXPHOS-derived ATP. If mutated or deleted copies are present at the mtDNA set point, they can be selected for amplification, preferentially or neutrally, and contribute to the mtDNA content of one or more of the offspring’s tissues. The mtDNA genetic bottleneck filters mtDNA molecules to primordial germ cells as they form. They provide the population of mtDNA that would be inherited by the next generation. OXPHOS: oxidative phosphorylation; Δ: mutated/deleted; WT: wild type. The figure is reproduced and modified from (Kelly and St John, 2010) with permission.

In humans, pigs, and cows, as fertilization takes place, the sperm’s nuclear genome is highly DNA methylated whilst the oocyte’s genome exhibits intermediate levels of methylation (Ivanova et al., 2020). To maintain intermediate levels of DNA methylation, the sperm genome undergoes DNA demethylation, which is complete by the 2- to 4-cell stage. These levels are maintained until the 8- to 16-cell stage. Global DNA demethylation then ensues to the blastocyst stage, and reprogramming is set by the hatching blastocyst stage (Ivanova et al., 2020). At the same time, mtDNA copy number decreases significantly by the 2-cell stage with a minor mtDNA replication event, most likely a checking or turnover event, that takes place between the 2- and 4-cell stages (Spikings et al., 2007; Cagnone et al., 2016; Hashimoto et al., 2017). Low levels of mtDNA copy number per cell persist until the morula stage, with significant increases by the blastocyst stage. This increase is restricted to the trophectodermal cells, whilst the cells of the inner cell mass continue to dilute out their mtDNA copy number with each cell division (Fig. 2).

These key events highlight how both genomes reset genomic balance throughout early development, and there is an interdependence between them. This interdependence has been shown in other cellular systems where the use of DNA demethylation agents results in demethylation of the nuclear genome coupled with increased mtDNA copy number. The changes in mtDNA copy number are mediated by changes in DNA methylation and gene expression of the nuclear-encoded mtDNA replication factors (Sun et al., 2018). Indeed, the interdependence likely continues as naïve cells commit to specific lineages. During post-implantation development, mtDNA copy number continues to be diluted out until cells initiate lineage commitment at gastrulation. Just prior to this, mtDNA copy number per cell reaches perilously low levels (Facucho-Oliveira et al., 2007), a stage described as the ‘mtDNA set point’ (Kelly et al., 2012) (Fig. 2). As naïve cells initiate differentiation, these copies of mtDNA serve as the template or ‘founder’ molecules that contribute to the developing foetus since mtDNA transcription and replication are concomitantly initiated at this stage (Larsson et al., 1998; Hance et al., 2005). Then, as cells commit to a specific fate, such as neurons or beating cardiac cells, they replicate their mtDNA in a cell-specific manner (Fig. 2) to meet their demands for OXPHOS-derived ATP (Facucho-Oliveira et al., 2007; Kelly et al., 2012).

Similarly, as the primordial germ cells are established, an mtDNA genetic bottleneck takes place, with a small population of mtDNA used to populate the offspring of the next generation. It also accounts for the variability in heteroplasmy that can exist in the mature oocytes of carriers of mtDNA disease (Wei and Chinnery, 2020). Evidence for the bottleneck exists in a range of species, including mouse (Jenuth et al., 1996), cow (Hauswirth and Laipis, 1982), and human (Marchington et al., 1997). Nevertheless, once cells commit to a specific fate, they will be governed by epigenetic changes, and expression of the nuclear-encoded mtDNA replication factors will regulate the amount of mtDNA copy number at any given stage. Any heteroplasmy persisting to gastrulation can be selected for or against as differentiating cells replicate mtDNA in a cell-specific manner (Fig. 2). Thus, high OXPHOS-requiring cells, e.g. neurons, and heart and muscle cells, will acquire higher numbers of mtDNA copy and, potentially, heteroplasmic molecules as they are the tissues primarily affected in mtDNA disease (McFarland et al., 2007) (Fig. 2).

What do we still need to know?

There is a drive amongst some in the ART clinical community to promote the use of mitochondrial supplementation (Morimoto et al., 2023) and nuclear transfer to enhance still seemingly low ART outcomes (Subira et al., 2025). We understand that recent reports related to mitochondrial supplementation (Morimoto et al., 2023), MST for fertility reasons (Costa-Borges et al., 2023), and MST (Zhang et al., 2017) and PNT (Hyslop et al., 2025; McFarland et al., 2025) for mtDNA disease have indicated that healthy babies have been born and in some cases report improved embryo quality as an outcome. In infant follow-up assessments, they further report that the offspring met developmental ‘norms’ (Costa-Borges et al., 2023; Morimoto et al., 2023). Nevertheless, mtDNA diseases can arise at various stages of an individual’s lifespan (Gorman et al., 2015), and understanding the associated molecular mechanisms will provide significant insights into potential onsets of mtDNA disease or unanticipated effects that could lead to complications, and potentially explain the reduced pregnancy rate following embryo transfer (Hyslop et al., 2025). The critical questions relate to the mechanism of action, efficacy, and safety, and these should really be answered before further implementation. Importantly, at the mtDNA, genomic, and epigenomic level, there are three key areas common to both approaches. They are (a) mtDNA carryover, (b) mtDNA haplotype matching, and (c) nucleo-mitochondrial interactions, and they appear to be interconnected.

The problem with mtDNA carryover

Typically, when a nucleus or nuclei is/are transferred from an oocyte or zygote harbouring mutant mtDNA to an unaffected donor oocyte or zygote, a small number of mitochondria, containing mtDNA, are carried over with the nucleus (Fig. 3; red mtDNA). Evidence to determine how much carryover persists and is amplified from clinically relevant large animal models is very limited. Nevertheless, in four MST-derived monkeys, mtDNA carryover was detectable at ≤3% in blood and skin samples (Ma et al., 2021). In one offspring, in the 18 other tissues and organs analysed, the carried-over mtDNA ranged from 0% to 10.1% in the kidney, 11.8% in the stomach, 12.4% in the liver, and 16.6% in the small intestine. In a second offspring, carryover ranged from 2.5% to 14.8% in skin, kidney, heart, thymus, optic nerve, and lung, whilst the two other offspring exhibited trace levels. Based on mtDNA carryover representing 0.6% of the oocyte’s original mtDNA population (average 250 000 copies), this represents a significant selective replicative advantage for the carried-over mtDNA during foetal development. It was as high as 27.7-fold in the intestinal tissue. Although MST was performed with oocytes containing unaffected mtDNA, the mixing of mtDNA genotypes is undesirable as each mtDNA genotype contains differences in sequences that alter the amino acid composition of some subunits of the electron transfer chain, as demonstrated in porcine-derived SCNT offspring (St John et al., 2005) and the MST-derived child exhibiting high levels of heteroplasmic load (Costa-Borges et al., 2023). As a result, each mtDNA genotype exhibits differences in OXPHOS efficiency (Yu et al., 2015), and mixing of genotypes will affect tissue function. Indeed, it is argued that the transmission of two distinct mitochondrial genomes would result in competition between the two genomes, and the one harbouring variants that exhibited greater rates of replication efficiency would become the dominant molecule (Beekman et al., 2014). In some cases, this would be advantageous whilst in some, it could affect the organism’s fitness since some mtDNA genotypes are more efficient at OXPHOS generation than others (Gomez-Duran et al., 2010).

Figure 3.

Mitochondrial DNA (mtDNA) carryover as demonstrated through metaphase II spindle transfer. The metaphase II (MII) spindle containing nuclear DNA is removed from a MII oocyte of a carrier of mtDNA disease. An unaffected donor oocyte is enucleated. The carrier’s MII spindle is transferred to the enucleated oocyte and fused. The reconstructed oocyte is fertilized and cultured in vitro. mtDNA containing either mutation only or mutation and wild type copies (heteroplasmy) accompanying the nucleus (red) are carried into the donor oocyte. The same principle applies to patients who have defective ooplasms, and wild-type only mtDNA is transferred. In both cases, the offspring can inherit two populations of mtDNA. The figure is reproduced and modified from (St John, 2014).

The outcomes from MST to overcome the transmission of affected mtDNA, in the only case so far documented, a carrier of Leigh Syndrome gave birth to a child possessing between 2.36% and 9.23% mtDNA carryover in skin and body fluid samples (Zhang et al., 2017); whilst the biopsied trophectoderm from the respective blastocyst possessed 5.1% heteroplasmic load. This contrasts with the use of MST to overcome infertility, where five out of the six babies did not possess any detectable levels of carried-over mtDNA. However, one child showed levels of carryover at 36% in cord blood, 42.4% in the cord, 37.7% in peripheral blood, 51.1% in urine, and 60.06% in saliva (Costa-Borges et al. 2023). Preimplantation genetic testing is regarded as a reliable tool to assess heteroplasmic load in embryos and to predict outcome in offspring from carriers of mtDNA disease (Rubens et al., 2025), however, discrepancies have been reported, as discussed in (Li et al., 2025). In the case of the heteroplasmic offspring, preimplantation genetic testing identified heteroplasmic load at 0.8% in the trophectodermal cells (Costa-Borges et al., 2023). For the nuclear transfer embryo, we do not know if the trophectodermal cells are as indicative of the whole of the embryo as they might be for IVF-/ICSI-derived embryos and, thus, not necessarily inform us of the cells that give rise to the embryo proper, foetus and offspring, namely the inner cell mass cells (see Fig. 2). In other words, we still need to understand patterns of mtDNA segregation in the nuclear transfer embryo. Prenatal diagnosis may have provided an indication of outcome, given that mtDNA replication would have been initiated by this stage in the developing foetus (see Fig. 2) but was not undertaken by this couple (Costa-Borges et al., 2023). Again, this approach, as with preimplantation genetic testing, only samples one source of cells, such as those present in the amniotic fluid. These cells may not have been fully representative, especially as the segregation of heteroplasmic molecules of mtDNA is not uniform (Marchington et al., 1997; Wei and Chinnery, 2020). Indeed, it is likely that the two heteroplasmic children in these studies (Zhang et al., 2017; Costa-Borges et al., 2023) would carry higher levels of heteroplasmic load in their other tissues and organs, which may affect quality of life.

Initial studies on PNT in human embryos showed levels of mtDNA carryover ranging from 0% to 30.2% and, after refinement, from 0% to 11.4% (Craven et al., 2010). There was also uneven distribution amongst blastomeres. Further studies suggested levels ranged from 2% to 4% in blastocysts (Hyslop et al., 2016); however, a more recent report suggests levels as low as 0.74–2.8% (Hyslop et al., 2025). Nevertheless, since human blastocysts possess ∼300 000 copies of mtDNA, carryover equates to 2220–8400 copies. Furthermore, the recent report showed mtDNA carryover ranged from undetectable in five babies to 5–9%, 12–13% and 16–20% in three other newborns from blood and urine samples, respectively (McFarland et al., 2025). Of the PNT- and MST-derived babies carrying heteroplasmic load, we do not know if the levels in tissues would be upregulated as the offspring aged, given that heteroplasmic load increases with age in affected tissues such as muscle and neural tissue (Chinnery et al., 2002; Grady et al., 2018).

Although the recent human study suggests that the levels of mtDNA carryover are below an acceptable threshold and, thus, reduce the risk of disease transmission (Hyslop et al., 2025; McFarland et al., 2025), we have not seen sufficient data to support this. We do not know whether the carried-over mtDNA would segregate to the other tissues in a uniform manner or unequally since peripheral blood and urine were the only samples examined (McFarland et al., 2025). Whilst many tissues would not be available from a newborn to determine overall levels of heteroplasmy, other sources were accessible, for example, buccal epithelium or hair, as in (Zhang et al., 2017). This would have provided evidence from each of the embryonic cell lineages and would be more indicative of early segregation outcomes. Such information would also have provided further confidence in the researchers’ findings, given that we do not know if mtDNA segregates in the same way in the PNT-derived embryo and foetus as those derived from natural conception. If different methods of segregation following MST or PNT are at play then the patterns of segregation may be different. It is important to have that information.

Currently, it appears from each of the human studies using either MST (Zhang et al., 2017; Costa-Borges et al., 2023) or PNT (Hyslop et al., 2025) that elimination or even regulation of mtDNA carryover has not been effectively overcome. Moreover, there appears to be preferential amplification for the carried-over mtDNA in some offspring. This process is referred to as ‘reversion’, i.e. reverting to the nucleus’s partnering population of mtDNA. Although it is not a complete reversion, as occasionally observed in some embryonic stem cell models of MST (Kang et al., 2016; Yamada et al., 2016) and PNT (Hyslop et al., 2016), it is partial. In those cases where the carried-over mtDNA was present in offspring, the increased levels ranged from 6.8-fold to 27-fold (McFarland et al., 2025); 3.9-fold to 15.4-fold (Zhang et al., 2017); and 60-fold to 100.2-fold (Costa-Borges et al., 2023). Consequently, there is the propensity for the carried-over mtDNA to be replicated and transmitted to the offspring, and impact on offspring health and well-being. In the monkey studies, similar amplification of the carried-over mtDNA was observed (Ma et al., 2021), as mentioned earlier. Putting the human studies together, 5 out of 15 offspring generated using either form of nuclear transfer represent a 33.3% chance of a resultant offspring carrying two populations of mtDNA. That is not a good outcome if the rationale is to try and prevent the transmission of heteroplasmic mtDNA from mother to offspring.

Nucleo–mtDNA interactions

Transferring nuclei between oocytes and zygotes, as with MST and PNT, respectively, requires the nucleus to adapt to its new cytoplasmic environment, and reestablishment of the interactions between the nuclear and mitochondrial genomes if embryos are to successfully develop.

It is not understood how the transfer of a metaphase II spindle from one oocyte to another or pronuclei between zygotes would affect the de novo DNA methylation programming and mtDNA reduction events that take place in the developing embryo (St John, 2019). SCNT-derived embryos exhibit incomplete reprogramming (Morgan et al., 2005) that results in the continued expression of the mtDNA replication factors, which can preferentially replicate the mtDNA carried over with the somatic cell (Bowles et al., 2007). Indeed, SCNT is associated with a number of developmental abnormalities (Cibelli et al., 2002), and the presence of carryover can be as high as 44% in porcine (Takeda et al., 2006) and 40% in bovine tissues (Takeda et al., 2003). Again, based on the initial carryover into the oocyte being only 0.6%, this represents 67- to 73-fold preferential amplification of the carried-over mtDNA. Furthermore, in somatic cell models, where a nucleus is fused to an enucleated cell carrying heteroplasmic mtDNA, the significant changes to the nucleus’s epigenome are proportional to heteroplasmic load (Kopinski et al., 2019). Indeed, more recently, large scale studies have shown clear indications between mtDNA heteroplasmy and consequential changes in DNA methylation (Lai et al., 2025). Likewise, embryonic stem cell lines possessing the same nuclear genotype but different mtDNA haplotypes exhibit major differences in global DNA methylation and nuclear gene expression (Kelly et al., 2013; Lee et al., 2017). Consequently, as the nuclear genome adjusts to its new cytoplasmic environment, ‘genomic balance’ is reestablished in accordance with the factors that influence each genome (Fig. 1).

The strength of the interaction between the nuclear and mitochondrial genomes was recently illustrated in a pig embryo model of MST. Simply removing the metaphase II spindle and re-injecting it into the same cytoplasm results in a number of nuclear-encoded genes being differentially expressed at the hatching blastocyst stage that affect multiple gene networks and pathways, including cell cycle and epigenetic regulators (Penn et al., 2024). Likewise, in human blastocyst-stage embryos generated through MST, a delay in DNA methylation was identified (Qi et al., 2022). A similar pattern emerges when purified mitochondria from sister oocytes (autologous) are introduced into pig oocytes. When pig metaphase II oocytes with low mtDNA copy number, which typically fail to fertilize (Cagnone et al., 2016), were supplemented with extra mtDNA (∼780 copies) as the oocytes were fertilized, they achieved similar levels of blastocyst development to controls. However, the additional mtDNA induced a mtDNA replication event that increased mtDNA copy number by 4.4-fold prior to the 2-cell embryo stage, most likely through the mtDNA turnover event described above. Additionally, it modulated the DNA methylation and expression profiles of the nuclear-encoded mtDNA-specific replicase, DNA Polymerase Gamma (the catalytic subunit; POLG) (Tsai et al., 2018), as also seen in SCNT embryos (Bowles et al., 2007) and MST-derived embryos (Penn et al., 2024); and the expression profiles of ∼250 nuclear genes in blastocysts (Cagnone et al., 2016). It would be interesting to see similar studies using the same model on PNT to, perhaps, determine if the pronuclei or metaphase II spindle were more adept at adapting to the new cytoplasmic environment.

Effects on nucleo–mtDNA interactions are also illustrated in pig oocytes containing sufficient copies of mtDNA. The addition of ∼780 copies of mtDNA resulted in over 2000 local nuclear genomic regions exhibiting differential levels of DNA methylation, and 52 genes showing significant differences in expression in resultant blastocysts, including cell cycle and inflammatory response networks (Okada et al., 2022). In offspring generated through this approach, some developmental milestones were affected, and there were alterations to gene expression (McIlfatrick et al., 2023) and metabolite pathways (Andreas et al., 2024) in high OXPHOS-requiring tissues, namely brain, heart, and liver. Consequently, the reconstructed oocyte, whether formed by mitochondrial supplementation alone or nuclear transfer, undergoes significant change that results in remodelling of its epigenome and mtDNA replication. This could pose a significant risk to the offspring in terms of epigenetic and mtDNA inheritance and affect subsequent generations. These issues represent a ‘black hole’ that requires considerable scientific investigation.

mtDNA haplotype matching

Whilst mitochondrial donation requires the donor oocyte to be from an unaffected individual, the degree of haplotype matching required between the patient’s and donor’s oocyte is not fully understood. Some argue that mtDNA matching is not essential (Yamada et al., 2016). Others have shown an association with development (Tamassia et al., 2004; Kang et al., 2016) and development to term in SCNT (Bowles et al., 2008). Indeed, matching was clearly ignored in the one child generated from MST as the carrier’s and donor’s oocytes were from very diverse mtDNA haplotypes (Zhang et al., 2017). However, mathematical modelling has resulted in a reference chart to match patient nuclei with donor oocytes for mitochondrial donation (Royrvik et al., 2016). Clearly, outcomes to date are limited, and large-scale studies would be required to achieve appropriate statistical significance. Nevertheless, the role that mtDNA haplotypes may play in the selection of carried-over mtDNA remains to be resolved (Kang et al., 2016; Hudson et al., 2019; Kang et al., 2019). It has been argued that (a) variant(s) within Conserved Sequence Box II (CSBII) of the mitochondrial genome may influence the selection of carried-over mtDNA since this is a regulatory region where the switch from transcription to replication is determined (Agaronyan et al., 2015). Interestingly, there appears to be an association between oocyte mtDNA copy number and a specific variant within CSBII for different mtDNA haplotypes of pigs (Tsai et al., 2016).

Conclusions

Although the various studies of mitochondrial supplementation, PNT, and MST report advances in embryo production and that the offspring met developmental milestones, there are many unknowns still related to these technologies. We are not simply dealing with replacing energy-generating mitochondria in a deficient oocyte, whether the oocyte belongs to a carrier of mtDNA disease or a patient suffering from repeated developmental arrest. In each case, we are dealing with significant genomic and epigenomic events. With mitochondrial donation, the oocyte is being challenged to undo a series of genomic and epigenomic events that were initiated from the primordial germ cell onwards that, when modified, change gene expression and mtDNA content in the oocyte. Likewise, dysfunctional mtDNA associated with infertility will induce altered epigenetic programming of the oocyte’s nucleus that will also require additional epigenomic remodelling. Indeed, even mtDNA carryover alone can alter programming, which means that if any of these technologies are to be implemented into clinical medicine, they require considerable refining. We believe emphasis should be on resolving these issues and ensuring that mtDNA carryover is eliminated and nucleo–mitochondrial interactions are fully understood, and any harmful consequences are also eliminated. We should not be merely reducing risk to offspring generated by these techniques but ensuring their utmost safety and efficacy.

Data availability

No datasets were generated or analysed in the current manuscript.

Authors’ roles

J.C.S.J. and R.J.R. conceived the concepts of the manuscript and wrote and edited the manuscript.

Funding

The studies were supported by the grants from the National Health and Medical Research Council (GNT1136065, GNT1160106, and GNT2000723).

Conflict of interest

The authors declare no competing interests.