The nuclear and mitochondrial genomes need to tango. How is their dance synchronized during development?

Apart from an offspring being at risk of inheriting two population of mtDNA, which could lead to tissue dysfunction, as highlighted in offspring derived through SCNT [ 297 , 298 ], it is likely immune and inflammatory responses also result. The mitochondrial genome is regarded as a ‘foreign’ genome since the mitochondrion originates from α-proteoerium [ 299 ] that used to have full capacity to transcribe and replicate itself given that its genome was significantly larger than its current version [ 300 ]. However, as an evolutionary consequence, many genes translocated to the nucleus and 37 remained in a smaller circular genome. Whether this was out of necessity to ensure critical genes came under more direct control of the nucleus or it enabled a more symbiotic relationship to be established to ensure its ever presence so that the cell would utilize its energetic potential and reduce its potential to clear this foreign body remains unclear. Indeed, the cell possesses a number of mechanisms for sensing foreign bodies. In the context of mtDNA, a number of factors have been postulated for recognising ‘self’ against ‘foreign’ including its different genetic code [ 39 ]; its lack of CpG methylation that generates unmethylated CpG motifs, as argued by some [ 301 ]; its packaging factors that contribute to the nucleoid, including TFAM [ 302 ]; its shedding into the cytoplasmic environment through active and passive methods [ 302 ]; and transcription which generates a RNA–DNA hybrid ‘R-loop’ [ 303 ]. mtDNA can activate various pro-inflammatory signalling pathways (reviewed extensively in [ 299 ]. In the endosome, this is mediated through TLR9, that drives an NF-Kβ-dependent pro-inflammatory signalling program [ 304 ]. In the cytosol, cGAS recognizes mtDNA and triggers endoplasmic reticulum-localized STING that mediates an interferon response [ 302 , 305 , 306 ]. Through the cytosolic inflammasome, for example AIM2 or NLRP3, the mtDNA-dependent inflammasome leads to caspase-1-dependent maturation or pro-inflammatory IL-1 and IL-18 [ 307 , 308 ]. Additionally, pathogens can drive neutrophil recruitment that can extrude mtDNA neutrophil extracellular traps and trigger interferon responses [ 309 ]. mtDNA can also be transferred through exosomes or through intact mitochondria to neighbouring cells [ 240 ], which affects the metabolism and survival of the recipient cell [ 310 ]. Consequently, it might be expected that any foreign mtDNA that is introduced into the oocyte would be eliminated early on during development. Elimination appears to be achieved through inflammasome activity with AIM2 and IPAF activity since it was downregulated in some somatic tissues from animals generated through mitochondrial supplementation suggesting that elimination had previously taken place [ 259 ]. However, mtDNA accompanying the nucleus was still present in heterozygous MST-derived preimplantation embryos suggesting that the nucleus recognized the accompanying mtDNA as ‘self’ and its elimination was not only avoided but this mtDNA was maintained through favourable replication in some instances prior to the blastocyst stage [ 291 ] rather than propagating the desired mtDNA in the reconstructed oocyte. Nevertheless, it appears that the processes leading to elimination of sperm mtDNA [ 82 ] or heteroplasmic molecules at the zygote stage [ 133 ] are not involved in the elimination process in this instance, although this would benefit from extensive investigation.

Concurrent with changes in mtDNA copy number during oogenesis [ 78 , 127 , 163 ], spermatogenesis [ 211 , 212 ], and post-fertilization development [ 129 , 132 , 172 ], there are changes to the methylation status of the gametes, the embryo and the foetus [ 25 , 27 , 213 – 215 ] (Fig.  5A – C ). Collectively, the changes in DNA methylation and mtDNA copy number will affect the balance in genomic content between the two compartments and their state of synchrony at each stage of development [ 48 ]. Furthermore, it is well-recognized that changes to a cell’s mtDNA copy number is concurrent with its metabolic status [ 167 ]. This, in turn, can alter the activity of the citric acid cycle and the by-products generated. These would include α-KG, which acts as a cofactor in mediating the conversion of 5mC to 5hmC [ 21 ]. Indeed, the DNA methylation changes mediated at various stages of development could be applicable to both genomic compartments of the cell given that mtDNA methylation has been detected in maturing oocytes and embryos [ 216 ] and has been shown to be dynamic throughout early development [ 217 ], and in somatic and tumour cells [ 218 , 219 ]. Nevertheless, the presence of mtDNA methylation remains a contentious issue and still requires substantial validation since the mode of detection can significantly influence outcomes [ 220 , 221 ]. Consequently, there are frequent alterations to both genomic compartments at key stages of development, which are likely associated with the fate of the cell at each stage, as predicted by Waddington’s model of the landscape of cell fate [ 222 ]. However, developmental success will require synchrony between the two genomes to ensure ongoing progress at each stage [ 48 ]. Given that oogenesis, spermatogenesis, embryogenesis, and organogenesis are prolonged events that undergo frequent changes to meet developmental milestones, it remains to be determined whether interference at any given stage would perturb outcome and/or the health and well-being of the offspring. If such perturbations exist, it is, therefore, essential to determine if they are transgenerational. Using the pig as the most appropriate model of human germ cell development, since it develops a bilaminar disc prior to onset of gastrulation unlike rodents [ 25 ], it is evident that PGCs are first observed in the early primitive streak stage embryo at the posterior pole of the epiblast [ 223 ]. By the late primitive streak stage, they are present in the yolk sac/allantoic mesoderm and the endoderm followed by migration through the developing gut and then into the genital ridge where colonization takes place [ 223 , 224 ]. Concurrent with premigration of PGCs is their epigenetic reprogramming mediated by the downregulation of DNMT3A, DNMT3B, and UHRF1 coupled with the upregulation of TET2 that increases 5hmC activity and reduces 5mC [ 25 ] with the TET pathway highly dependent on the TCA cycle for co-factors [ 225 ]. These activities initially target imprinted genes and retrotransposons but progress into extensive reprogramming over several weeks of early development with the lowest levels of DNA methylation present in gonadal PGCs. Indeed, the activity of the TET pathway and its oxidized products, and activation of the BER pathway suggests active DNA demethylation in pre-migratory non-replicative PGCs whilst the downregulation of DNMT 3A and 3B and UHRF1 indicates passive DNA demethylation (Fig. 5A ). However, by the germinal vesicle stage in the human, CpG methylation is reset. From then until metaphase II, CpG methylation remains stable [ 27 ]. Furthermore, it appears that de novo DNA methylation does not start until after birth and progresses up to the germinal vesicle stage, which runs concurrently with follicular growth [ 226 ]. Based on this evidence, it appears that DNA methylation is directly linked to oocyte growth especially in the context of imprinted germline DMRs with each locus achieving methylation at specific stages of growth. Interestingly, the sperm genome appears to undergo greater levels of DNA methylation (82.0%) than the oocyte’s genome (54.5%) [ 227 ]. Following fertilization, there are global changes to DNA methylation (Fig. 5B ). There are three distinct waves of global DNA demethylation in preimplantation embryos [ 227 ]. The first wave takes place within 12 h of fertilization with a greater decrease occurring in the sperm methylome compared to the oocyte’s methylome by the pronuclear stage. The second wave takes place between late zygote to the 2-cell stage whilst the third wave takes place between the 8-cell and morula stages. This results in an overall drop in methylation to 35.1%. Nevertheless, DNA demethylation in the preimplantation embryo is countered by de novo methylation which take place in two waves [ 227 ]. The first appears at the early male pronuclear stage and persists to the mid-pronuclear stage whilst the second is indicative of the 4- to 8-cell stage. The second wave marks a larger increase than for the first and takes place in line with what is commonly described as embryonic genome activation [ 27 ]. In all, this highlights two very active processes that modulate the nuclear genome for subsequent developmental events.

mtDNA transcription and replication highlight the need for synchrony between the two genomes. Transcription takes place first by generating a polycistronic molecule that is then cleaved to generate mRNAs, tRNAs and rRNAs [ 54 – 56 ]. Transcription requires a number of nuclear-encoded factors to translocate to the mitochondrion which then interact with the control region of the D-Loop. The key mtDNA transcription factors comprise mitochondrial transcription factors A (TFAM) [ 57 ], B1 (TFB1M), and B2 (TFB2M) [ 58 ], of which TFB2M is deemed to be essential [ 59 ], mitochondrial RNA polymerase (POLRMT) [ 60 ], mitochondrial transcription elongation factor (TEFM) [ 61 ], and mitochondrial termination factor 1 (MTERF1) [ 62 ]. These factors work in unison and, when assessing transcriptional activity, they all should be assayed. Replication is also dependent on the translocation of a number of nuclear-encoded factors (Fig.  3 ). These include the mtDNA-specific replicase, DNA Polymerase Gamma (POLG) [ 63 ]. POLG consists of a catalytic subunit, encoded by POLG , and two accessory subunits encoded by POLG2 . The function of the accessory subunits is to anchor the catalytic subunit to the mitochondrial genome to promote efficient and faithful replication [ 64 ]. Once transcription has generated an elongated polycistronic molecule, cleavage generates a short fragment that is used as a primer to initiate mtDNA replication within CSBII as a result of transcriptional termination [ 65 ] (Fig.  2B ). Replication also necessitates activities mediated by the mtDNA-specific helicase, Twinkle (TWNK), the mitochondrial single-stranded DNA-binding protein (SSBP1) [ 66 ], and the mtDNA-specific topoisomerase ( TOP1MT ) [ 67 ] (Fig.  3 ). Indeed, these factors are ever-present in the mitochondrion where they interact with the mitochondrial genome to form nucleoid structures [ 54 , 68 ]. This process is modulated by TFAM, which also binds and packages mtDNA [ 69 ]. Factors controlling mtDNA replication. mtDNA replication is dependent on nuclear-encoded mtDNA-specific transcription and replication factors that translocate to the mitochondrion. Firstly, mtDNA transcription is initiated, which is a pre-requisite for replication to proceed. The key factors involved in mtDNA replication are POLGA, POLGB, TWNK, TOP1MT, and MTSSB. Reproduced from [ 335 ] under Creative Commons Attribution-NonCommercial-NoDerivatives International 4.0 License (CC BY-NC-ND 4.0). Over the last few years, there has been extensive and even often hostile debate over the mode of mtDNA replication [ 70 , 71 ]. The classical approach has been the strand displacement model. In this instance, replication is initiated at the origin of heavy strand replication (O H ) to produce a nascent H-strand [ 55 ]. As replication extends to two-thirds of the genome, it exposes the origin of light strand replication (O L ), which is located in the minor non-coding region that comprises a stable double-stranded stem region [ 72 ]. As this stage, both strands are then simultaneously synthesized. The other approach is the coupled leading- and lagging-strand synthesis model where the heavy and light strands are proposed to be copied in a bidirectional manner from the same initiation cluster site [ 73 ]. This mechanism has now evolved to become the RNA incorporated throughout the lagging strand (RITOLS) model whereby replication intermediates enable gaps within the mtDNA replicon on the lagging-stand to be filled and that replication is driven from several origins of replication within the vicinity of O H [ 74 ]. The RNA intermediates are then later converted to DNA. It has been further argued that both the strand displacement and RITOLS models may be involved in replication but in different cell types and as cells are (re)populated with mtDNA.

The disparity between high mtDNA copy number and the presence of naïve mitochondria in mature oocytes suggests that oocyte mitochondria are acting as vehicles for the transmission of mtDNA. Coupled with the lack of mtDNA replication up to and including gastrulation, it is likely that the mtDNA content in mature oocytes is also an investment in subsequent developmental outcomes. Across a range of large mammalian species, it is evident that oocytes possessing fewer copies of mtDNA than the anticipated threshold, i.e. mtDNA deficiency, often fail to fertilize or arrest during early embryonic [ 47 , 129 , 131 , 165 , 172 ]. However, since there is no active reduction in mtDNA copy number during murine preimplantation development [ 126 ], the effects of mtDNA depletion are not evident and would account for the disparity in reports regarding the importance of mtDNA copy number to fertilization, preimplantation development and pre-gastrulation outcomes cf. [ 185 , 186 ]. Incidentally, oocyte mtDNA deficiency is associated with ageing as younger women tend to possess oocytes with larger cytoplasmic volumes (<35 versus ≥ 35 years old) and increased mtDNA copy number when compared with women of older reproductive age (<40 versus ≥40 years old) [ 187 ]. mtDNA deficiency is also associated with other human reproductive disorders that include endometriosis [ 188 ] and reduced ovarian reserve [ 189 ].

Since 1997, ART clinics have sought to introduce technologies that manipulate the mitochondrial genome to enhance fertilization outcomes and provide reproductive options for infertile couples or couples at risk of transmitting severe genetic disorders, i.e. the mtDNA transmissible diseases. Initially, this started through a process known as cytoplasmic or ooplasmic transfer [ 241 ]. Many of the outcomes have resulted in the offspring inheriting two populations of mtDNA [ 242 ], which is in a manner very different to that from natural fertilization [ 34 , 85 ] (cf. Fig. 7A – C ). Consequently, the offspring were stigmatized as being ‘3-parent’ children and the technology referred to as ‘3-parent IVF’ [ 243 ], a misnomer given that IVF was not performed but a rather more invasive ART approach. Specifically, the process involves the transfer of ooplasm from a young donor’s mature oocyte into the oocyte of a patient who was suffering from repeated fertilization or embryonic developmental failure [ 241 ]. The overarching aim of providing the extra ooplasm was to boost the defective cytoplasm of the oocytes from older patients. However, this approach involves the transfer of potentially all the factors present within the oocyte’s cytoplasm and was not specifically related to one constituent that may improve developmental outcome. Consequently, it was not evident which factor(s) was/were the enhancing constituent(s). Nevertheless, this approach resulted in the birth of children [ 241 ] that were carriers of two distinct populations of mtDNA, namely from the mother’s oocyte and from the oocyte of the donor ooplasm [ 242 ]. A subsequent, brief report documented a case of pervasive development disorder in one infant and two incidences of XO syndrome, one of which spontaneously aborted and the other was selectively aborted [ 244 ]. A more recent follow study of 13 of the 17 children, when between the ages of 13 and 17, indicated that they had generally met age-related developmental milestones with occasional anomalies [ 245 ], although no in-depth genetic analysis was conducted and the on-going degree of donor mtDNA transmission was not reported. Homoplasmy and heteroplasmy. ( A ) Normally, cells possess one population of mtDNA that originates from our mother’s oocytes, which is a population expanded from a smaller number of copies in her PGCs. This population usually consists of WT copies (unaffected copies) only and, as it is a uniform population, it is described as being homoplasmic. ( B ) Some oocytes harbour mtDNA rearrangements (mutations or deletions) and these can lead to severe forms of disease. The mixing of WT and mutant (Δ) molecules is described as heteroplasmy. ( C ) Some ARTs, for example GVT, MST, PBT, and PNT, carry over mtDNA from the affected oocyte into the donor oocyte. As a result, the offspring can inherit two populations of mtDNA. In this case, they would possess mtDNA from two different mtDNA genotypes, which is also described as being heteroplasmy. Dependent on mtDNA matching procedures, they may be very closely related mtDNA haplotypes or very distant haplotypes. This would be the outcome for patients undergoing nuclear transfer for poor oocyte quality or developmental failure reasons. A similar outcome is associated with CT and SCNT. ( D ) GVT, MST, PBT, and PNT can also be used to restrict the transmission of mutant mtDNA to the offspring. In this instance, mtDNA carryover will consist of mutant only, WT only, or a mixture of mutant and WT mtDNA. As a result, the offspring could inherit three populations of mtDNA from two genotypes (mutant and WT from the mother’s haplotype; and mtDNA from the donor oocyte’s haplotype). This would be a potential outcome for patients undergoing nuclear transfer when associated with mtDNA disease. The overall general good health and well-being of the offspring born from CT is not indicative of the findings from mouse studies where a number of disorders associated pathophysiology [ 246 ], growth [ 247 ], and genomic integrity [ 248 ] were reported. Whilst it has been argued that the intra- and inter-genic transfers in mouse could be a confounding problem [ 245 ], it must also be recognized that the process of mouse fertilization and development of the mouse embryo is not as indicative of human fertilization as would be the case for livestock species such as the pig [ 90 ]. Furthermore, the use of ICSI on mouse oocytes requires additional steps that may or may not affect developmental outcome [ 245 , 246 ]. More recently, in a collection of studies using the pig as a model, it has been shown that oocytes deficient in mtDNA copy number benefited from the addition of extra copies of mtDNA [ 131 , 172 , 249 ]. To this extent, oocytes with a mean mtDNA copy number of <50 000 copies were supplemented with ∼780 copies of mtDNA from mitochondria isolated from sister oocytes (i.e. from the same ovarian source) [ 172 ]. The additional mtDNA appeared to provide a stimulus that modulated a minor mtDNA replication event just after fertilization that resulted in a 4.4-fold increase in mtDNA copy number by the 2-cell stage, thus increasing the mtDNA investment in subsequent developmental outcomes [ 172 ]. This was matched by changes in the expression and DNA methylation profiles of the catalytic subunit of the mtDNA-specific replication factor POLG [ 249 ]. Furthermore, there were changes in gene expression after each cell division throughout preimplantation development [ 202 ]. In addition, blastocysts derived from the supplemented mtDNA-deficient oocytes exhibited global gene expression profiles more similar to blastocysts derived from non-supplemented oocytes carrying the required constituent levels of mtDNA than those derived from non-supplemented mtDNA-deficient oocytes [ 172 ]. Consequently, there appears to be a direct relationship between mtDNA copy number, oocyte developmental competence, and the regulation of nuclear gene expression profiles during early development. This suggests that if the genomic balance is readdressed in mtDNA-deficient oocytes, developmental competence can be restored. However, given the outcomes from CT, an open question is whether other cytoplasmic factors present in the ooplasm would also impact on developmental outcomes. In a clinical context, there has been some success from autologous mitochondrial supplementation. An early report suggested that live offspring can be generated through supplementation of oocytes with mitochondria isolated from cumulus granulosa cells [ 250 ]. The patients comprised those having had a previous failed program of treatment due to fertilization failure, embryo arrest and recurrent implantation failure, and prolonged unexplained infertility, or were aged over 38. A total of 71 cycles resulted in 20 live births of which seven were twins. In a similar approach, improved embryo quality and pregnancy rates were reported in patients over 37 years of age. However, there was no improvement in fertilization outcomes. In all, this work resulted in 7 clinical pregnancies from 18 couples [ 251 ]. Nevertheless, in animal models, others have reported that the transfer of somatic mitochondria would have a detrimental effect on developmental outcome [ 252 ]. Autologous supplementation has also been practised by isolating mitochondria from putative oogonial stem cells within the patient’s own ovaries [ 253 ]. Again, mouse studies reveal differences to human studies in this respect. When founder mice were produced using mitochondria from oogonial stem cells and mated to produce another two generations, there were significant increases in litter size and the number of primordial follicles in the ovary coupled with changes in global gene expression patterns for these early-stage oocytes. To counter this gain, the offspring exhibited increased weight gain during early life; and the first- and second-generation offspring experienced a defect in their cardiac structure, which was corrected in the third generation [ 254 ]. However, a two-centre trial reported four live births from 104 cycles (reported in [ 255 ]), whilst a randomized pilot study ( NCT02586298 ) was stopped as there appeared to be no benefit from the procedure in terms of live birth outcomes [ 256 ]. The failure of this trial may be due to its design rather than the application of the technology given than it was applied to couples who failed to achieve a pregnancy rather than those that had exhibited oocyte mtDNA deficiency. Nevertheless, more recently, the birth of 13 babies has been reported, all of which met developmental milestones [ 186 ]. Long-term follow studies will be necessary to ensure that there are no unanticipated effects. In this context, the generation of pigs using this approach revealed changes in DNA methylation patterns and gene expression in blastocysts [ 257 ] coupled with differences in daily gains for height and weight and some mild biochemical and haematological differences after birth of which some dissipated at sexual maturity [ 258 ]. However, DNA methylation and gene expression were affected in heart, brain, and liver and, in some instances, generically and in others tissue specifically and transmitted from the blastocyst [ 259 ], along with alterations in metabolite profiles that were also reflected in tissue specific gene expression [ 260 ]. Likewise, in a bovine model, the addition of extra oocyte-derived mtDNA as SCNT is performed resulted in significant increases in mtDNA copy number by the blastocyst stage and altered nuclear gene expression profiles [ 261 ]. The endeavours to modulate the cytoplasm have also been followed by far more invasive approaches to manipulate the oocyte, namely the transfer of either the nucleus or the pronuclei from their respective oocytes or zygotes into that of similar stage oocytes or zygotes. Indeed, there is an increasing push to introduce these approaches into clinical medicine to treat infertility and diseases associated with the mitochondrial genome. However, there are a number of questions that remain to be answered in order that we have significant confidence in their application. We will now investigate each of these technologies to determine what we know, what we need to know, and what their current respective safety indications are. The more invasive ARTs developed with a view to their translation into clinical practice include germinal vesicle transfer (GVT), metaphase II spindle transfer (MST), pronuclear transfer (PNT), and polar body transfer (PBT) using either the first or second polar body. These technologies are frequently referred to as mitochondrial donation in the context of mtDNA disease; and nuclear transfer in the context of infertility per se. They have their origins in nuclear transfer which was first successfully performed by John Gurdon and colleagues when they cloned the first frog in the 1950s [ 262 ]. This approach then led to the cloning of the first mammal in the form of sheep generated through the transfer of cultured embryonic cells [ 263 ] and then through a somatic cell (SCNT) in the form of Dolly the sheep [ 264 ] and, subsequently, blastomere (embryonic) cells in monkeys [ 265 ] and cattle [ 266 ]. From a technical perspective, these approaches involve the transfer of a somatic or an embryonic cell into the perivitelline space of an enucleated oocyte that is at the metaphase II stage of development. The two entities are fused together and the reconstructed oocyte is then activated and allowed to develop in culture until the preimplantation embryo is transferred into a recipient surrogate that will carry the embryo to term once implanted [ 263 , 264 ]. Each of the reproductive options that would potentially be available in clinical medicine, for example GVT, MST, and PBT, use a very similar approach except that the maternal chromosomes are fused to an enucleated oocyte. In the case of PNT, the maternal and paternal karyoplasts are transferred into the enucleated zygote [ 267 , 268 ] or the maternal karyoplast is transferred into a zygote from which the maternal chromosomes have been eliminated [ 269 ]. To date, there are a few documented outcomes resulting from the use of MST and PNT to produce children for infertility reasons and to prevent the transmission of affected mtDNA in the context of mtDNA disease. In terms of infertility, a pilot study using MST resulted in six live births, which represents a live birth rate of 31.6% per transfer and a live birth rate per patient of 24% (25 patients) following a clinical pregnancy rate per transfer of 36.8% (7 pregnancies from 19 single embryo transfers) [ 270 ]. In a similar fashion, a live birth was reported of a child from a carrier of a mtDNA disease, namely Leigh Syndrome [ 271 ]. In each case, the children met developmental milestones at birth [ 270 , 271 ] and in infant follow-up studies [ 270 ]. Live births have also been reported from the use of PNT to overcome mtDNA disease [ 272 , 273 ]. In all, 22 patients underwent PNT with eight live births reported from eight clinical pregnancies after 40 single embryo transfers. Of the eight births, five originated from carriers of LHON variants and three were from rare mt–tRNA variants. Each of the children was deemed to be making healthy developmental progress, although three were diagnosed with one of myoclonic epilepsy of infancy, urinary tract infection, and a dilated left ventricle. However, in each case, the offspring recovered to make normal progress [ 273 ]. The introduction of nuclear transfer under specified conditions to overcome mtDNA disease in the UK and Australia arose from Parliamentary reviews and processes that involved a scientific review of the technology. These reviews were conducted under the auspices of the designated authority within the respective jurisdictions, namely the Human Fertilization and Embryology Authority (HFEA) in the UK and a specifically established committee, the Mitochondrial Donation Expert Working Committee in Australia. This Committee was appointed by the Minister for Health through the National Health and Medical Research Council which, through its Embryo Research Licensing Committee in Australia, is the designated authority for embryo research. Indeed, the legislation in Australia still classifies nuclear transfer as a tool that requires further research before implementation. The Australian review was built on knowledge already gained from the UK’s process that had preceded its own and its terms of reference were to determine if any new knowledge had been gained since the HFEA review in 2016 [ 274 ]. In all, three areas were identified as still being inconclusive that would benefit from further research. These include (i) understanding mtDNA carryover, (ii) whether it is necessary to match mtDNA haplotypes, and (iii) the implications for nuclear–mitochondrial interactions. The proposed areas of investigation are important since the oocyte/zygote undergoes significant reconstruction. Indeed, for cytoplasmic and nuclear transfer, as with SCNT, the recipient oocyte is from another maternal source, which results in the offspring inheriting its mtDNA from a third party [ 267 , 275 ] (Fig.  7C ). Consequently, the recycling of the mitochondrial genome through the maternal lineage in a uniparental fashion only is lost and all subsequent offspring will inherit the mtDNA lineage from the third-party source (cf. Figure  7A – D ). Nevertheless, these approaches might be appropriate to replace mtDNA carrying pathogenic mutations or deletions (mitochondrial donation) or to provide a cytoplasm that is far more effective to support fertilization and subsequent developmental events (nuclear transfer), if the outstanding issues could be resolved.

Not only have reproductive scientists and clinicians failed to carefully consider the effects of mtDNA carryover, they have largely ignored the potential consequences of not matching the recipient oocyte that carries the inheritable population of mtDNA. This is especially the case in the context of livestock production. In many instances, oocytes are simply collected from slaughterhouses and collectively matured in vitro , and, once matured, selected based on standard assessments for oocyte quality with little or no attention paid to mtDNA content and, thus, lineage. In the bovine species, oocytes could range from offspring with maternal lineages derived from either Bos taurus or Bos indicus genotypes [ 171 , 311 ]. The benefits of separating oocytes based on mtDNA lineages have been shown in terms of early developmental potential in a bovine model [ 200 ]. Indeed, as the mitochondrial genome is transmitted through the maternal-only germline, it is passed from one generation to the next in a largely unchanged format [ 40 ]. However, the nuclear genotype results from the recombination of maternal and paternal chromosomes at the time of fertilization. Consequently, a Bos taurus cow may well have been generated on a Bos indicus female background but phenotypically exhibits traits associated with Bos taurus cattle, such as enhanced meat or milk quality [ 171 ]. Nevertheless, its mitochondrial genome will pass on relatively unchanged through its oocytes. To this extent, in one study, 55.4% of cattle assessed with Bos indicus phenotypes possessed Bos taurus mtDNA whilst 40.9% of cattle assessed with Bos taurus phenotypes harboured Bos indicus mtDNA [ 171 ]. Early pioneers in reproductive biology most likely paid little heed to the mitochondrial genome as it was considered to be primarily involved in energy production. However, amongst non-mtDNA specialists, there has been a number of misnomers associated with the mitochondrial genome and its function. As stated above, it encodes 13 of the subunits of the ETC, 2 rRNAs, and 22tRNAs [ 39 ] and it is this entity that is maternally inherited [ 34 ]. The mitochondrion per se is largely encoded by nuclear genes as are the majority of subunits, and the chaperones and mediators that assemble the ETC [ 312 ]. Consequently, in order for the ETC to function effectively, a high degree of compatibility is required between the two genomes [ 313 ]. Indeed, within a species, there are a number of haplotypes that have subdivisions [ 314 ]. Each haplotype will possess unique sequences within coding and non-coding regions that provide unique identifiers. In the coding regions, the unique sequences will affect amino acid composition that, in turn, affects protein structure, as demonstrated in porcine offspring derived by SCNT [ 287 ]. As a result, the ability of a cell to assemble efficient ETCs can be compromised, which would affect cellular function through reduced ATP output and increased free radical activity, a by-product of poor ETC function [ 313 ]. Nevertheless, some early investigations alerted our attention to the potential consequences of the effects that different mtDNA genotypes could have on embryo quality and production. For example, the generation of embryos through SCNT derived from the same donor cell source but against two different sources of oocytes indicative of two different mtDNA genotypes demonstrated improved developmental and pregnancy rates for one mtDNA genotype and different levels of ATP production [ 204 ]. Maintaining the same nuclear genome in this fashion demonstrates the effect of the mitochondrial genome alone on developmental outcome. This is further highlighted by the apparent relationship between the mtDNA genetic distance of the donor cell and that of the recipient oocyte and developmental outcome, as shown in ovine SCNT-derived embryos [ 292 ]. Indeed, a small increase in mtDNA genetic distance did not affect developmental outcome whilst a large increase reduced developmental outcome. This appears to be irrespective of whether accompanying mtDNA is present or not. Consequently, there appears to be a window of benefit in terms of the appropriate mtDNA genotype for developmental outcome. This further suggests that mtDNA defines the species barrier given that nuclei less favourably replicate mtDNA from different species and reconstructed cells with these admixtures result in OXPHOS-deficiency [ 315 ]. Nevertheless, from the use of handmade cloning in a bovine model, where a donor cell is fused to one of more oocyte cytoplasts (i.e. the oocyte has been dissected to remove the portion carrying the karyoplast), it is evident that the donor cell preferentially replicated a population of mtDNA that was more genetically diverse to its own [ 205 ]. Furthermore, in live offspring generated using this approach, none were born that possessed a predominant population similar to that of the donor cell’s mtDNA genotype. It could be argued that a more diverse mitochondrial genotype would provide hybrid vigour given that sperm nuclear DNA is not introduced as with natural fertilization, IVF, ICSI, CT, or mitochondrial supplementation. In the context of mitochondrial donation/nuclear transfer, hybrid vigour may not be required through the recipient oocyte as this approach would also rely on the use of the sperm nucleus to provide hybrid vigour. This is likely to be the case with GVT, MST, and PBT. However, this situation may also exist for PNT given that the two pronuclei would not have fused or recombined prior to their transfer. Indeed, it has been argued that pre-pronuclear transfer, i.e. the transfer of very early pronuclei, would yield significantly better developmental outcomes resulting from sperm DNA providing the required hybrid vigour [ 268 , 269 ]. In terms of assessing whether mtDNA haplotypes influence reversion, a study employing a porcine model of SCNT reported that following the transfer of porcine embryonic fibroblast cells from male and female Xiang pigs into enucleated oocytes from Yorkshire or Landrace pigs, donor cell mtDNA was present from 0% to <5% in ear biopsies and blood samples from female offspring [ 275 ]. However, it was detected at levels as high as 95% in the tissues of boars. Analysis of fibroblasts derived from the offspring highlighted how the varying haplotypes used in this study affected cellular capacity for mitochondrial respiration. In human embryonic stem cells derived through MST, the selective replication of mtDNA introduced into the oocyte has been attributed to a variant (G5AG7) that alters the stability of G-quadruplexes present in CSBII of certain mtDNA haplotypes. This, in turn, affects the efficiency of mitochondrial transcription termination and the generation of the primer for mtDNA, thus, leading to preferential selection of accompanying mtDNA [ 276 ]. Although this analysis has been disputed by others [ 316 ], but further endorsed by the proposers of this mechanism [ 317 ], mathematical modelling of G-quadruplex stability shows distinct patterns of replication and segregation bias in the context of heteroplasmic transmission [ 318 ].

mtDNA haplotypes have been associated with a number of phenotypes across a range of species as determined by genome-wide studies and experimental approaches. Indeed, some of the early studies using mouse cybrids, the fusion of a mtDNA-depleted karyoplast (somatic cell retaining its nucleus but depleted of its mtDNA) to an enucleated cytoplast (somatic cell lacking a nucleus but retaining its mtDNA) from different mtDNA genotypes resulted in very different levels of ATP production mediated through OXPHOS [ 313 , 315 , 319 , 320 ]. Whilst it can be argued that the genetic divergence between the two fusion partners ranged from being similar to quite large, this did not affect the ability of similar fusions generated from embryonic stem cells to produce live chimeric mice following their introduction into blastocyst-stage embryos [ 321 , 322 ]. Analysis of the same embryonic stem cells lines in vitro showed subtle differences for the levels of expression of the genes that contribute to the pluripotency network (e.g. Oct4, Sox2, Nanog, Rex1, Dppa5, Pramel7 , and Ndp52L1 ) [ 179 ]. Likewise, there were differences in patterns of differentiation, ability to produce spontaneously beating cardiomyocytes, mitochondrial energy metabolism, and DNA methylation, including occupancy of promoter regions for the key pluripotency genes [ 179 ]. Further studies have indicated that the varying haplotypes result in differential patterns of hypo- and hyper-methylation in undifferentiated cells and in patterns of gene expression associated with neural differentiation [ 323 ]. Likewise, studies in Drosophila have shown that mtDNA haplotype can influence a number of phenotypes including fertility and metabolism [ 104 , 191 ]. Indeed, the use of conplastic strains of mice highlights the affects that different mtDNA haplotypes have on mitochondrial proteostasis, generation of reactive oxygen species, insulin signalling, obesity, and lifespan [ 98 ]. Furthermore, diet can influence the selection for mtDNA variants which, in turn, regulate the fitness of an individual [ 324 ]. Indeed, previous to the view that mtDNA was primarily concerned with energy production, it appears that mtDNA has roles to play in tumorigenesis [ 325 ] and immunity and infection [ 302 ]. Consequently, identifying the most appropriate mtDNA haplotype for the process of nuclear transfer (all forms) would appear to be a complicated task. A systematic review of published data suggested that one in 1330 offspring born through mitochondrial donation would be affected due to mito–nuclear interactions [ 326 ]. Nevertheless, a large-scale modelling exercise revealed the degree of diversity within individual human haplogroup populations, potential selection bias for variants and transmission of carried over mtDNA for selection. As a result, a process was proposed for selecting optimal nuclear and mtDNA partners for nuclear transfer based on mtDNA haplotypes [ 327 ].

Whilst nuclear transfer introduces either one (MST; PBT; GVT; SCNT) or two (PNT) genomes to a new ooplasmic environment, it is likely that the interactions between the nucleus and the mitochondrial genome will need to be re-established to initiate and progress development. For example, during early preimplantation, early patterns of imprinting are established as are global patterns of DNA de/methylation [ 27 ] (Fig.  5B ). Indeed, many of these patterns will use the templates normally established during oogenesis (Fig.  5A ) and spermatogenesis and then be modulated accordingly through preimplantation (Fig.  5B ) and subsequent stages of development (Fig.  5C ) [ 25 – 27 ]. These would include the regulation of imprinted genes which were rightly investigated after the introduction of ICSI into clinical medicine and to examine the effects of culture media on  in vitro culture and hormonal stimulations protocols [ 328 – 330 ]. Furthermore, mtDNA copy number appears to be strictly regulated during each of these stages but in antipathy to DNA methylation. As can be seen from Fig.  5A , as DNA demethylation takes place, mtDNA copy number increases. However, it remains to be determined whether the post-fertilization profiles remain the same once nuclear transfer has taken place. It is evident from SCNT that DNA methylation events post-oocyte reconstruction are, indeed, asynchronous to those following, for example, fertilization [ 331 – 333 ] as are mtDNA replication events coupled with the continued expression of the nuclear-encoded mtDNA replication factors [ 286 , 292 ]. Consequently, not only could the continued expression of these factors result in overall increases in mtDNA copy number during preimplantation development, but it could also promote the replication of the accompanying mtDNA, as described above and indicated in porcine MST where POLG and TFAM were both upregulated during preimplantation development [ 291 ]. Indeed, in the context of SCNT, it appears that the key factors associated with the nuclear-encoded mtDNA replication machinery are not shutdown as observed for ovine and porcine embryos generated through in vitro fertilization procedures [ 129 , 172 , 286 , 292 ]. As a result, the pre-established genomic balance acquired through oogenesis and oocyte maturation will have been perturbed and require resetting. It is further well-documented that there are a number of perturbations associated with SCNT and DNA methylation profiles throughout development [ 334 ]. Indeed, a summary of reports from SCNT offspring highlights organ and tissue dysfunction along with phenotypes similar to the mtDNA diseases [ 297 , 298 ]. It remains to be determined whether other forms of nuclear transfer appropriately establish DNA methylation profiles typical of those resulting from natural fertilization during development and that the internal organs and tissues of the offspring function normally. Consequently, in depth molecular and histopathological assessments are required to determine the extent of any aberrations that might exist. As a point of caution, and as discussed earlier, small amounts of additional mtDNA alone, introduced into the oocyte, for example, through mitochondrial supplementation, can reset the DNA methylation and gene expression profiles of blastocysts [ 257 ] and offspring [ 259 ]. Indeed, these models are effective at studying mtDNA carryover without the confounding other factors associated with nuclear transfer.

Whilst some regulators throughout the world are more accepting of innovative ARTs, albeit under specific conditions, there is still a great deal to be learnt. We do not fully understand how the nuclear and mitochondrial genomes interact following nuclear and mitochondrial transfer protocols and whether they seamlessly generate the appropriate interactions that would enable them to co-exist in normal functional cells, tissues, and organs. Consequently, further studies related to mtDNA carryover, mtDNA haplotype matching and epigenetic regulation need to be performed on clinically relevant large animal models in order that there is a full understanding of the molecular effects of nuclear transfer. Likewise, a series of histopathology studies on founder offspring and, at least, one further generation need to be undertaken to determine if any abnormalities could give rise to concern. Failure to conduct these studies will result in clinics in some jurisdictions undertaking ‘potential treatments’ that are not supported by scientific rigour. The consequences of failing to understand the molecular mechanisms could result in any of the proposed treatments potentially overcoming one issue at the expense of another, often described as a ‘trade off’. However, relevant scientific knowledge would offer opportunities to modify protocols to make them safe. In other words, we still need to determine if our new genomic partners in the reconstructed oocyte, embryo and offspring tango effectively or whether they dance to another tune.

mtDNA is primarily inherited from the population present in the oocyte at the time of fertilization [ 75 , 76 ], which is derived from the PGCs [ 76 – 78 ] and ensures the maternal-only inheritance of this genome [ 34 ]. There are exceptions amongst mammalian species, but this is limited to inter-specific breeding, where the sperm and oocytes originate from two different species or sub-species. To this extent, evidence comes from crossings between the strains of Mus musculus and Mus spretus mice resulting in sperm mtDNA being transmitted to the offspring at levels <1% [ 79 ]. Sperm mtDNA transmission has also been observed following the mating of different sub-species of Macaca mulatta (Indian and Chinese Rhesus macaques) [ 80 , 81 ] and the crossing of Bos taurus and Bos indicus [ 82 ] cattle. However, paternal transmission of mtDNA tends to be the exception to the rule. In humans, there in one reported case of sperm mtDNA transmission that resulted in the offspring suffering from a muscle myopathy originating from a mutation in his father’s sperm mtDNA. Likewise, sperm mtDNA has been detected in human triploid blastocysts [ 83 ], which would fail to fully develop, suggesting that transmission of paternal mtDNA is an aberrant event in humans and does not give rise to biparental transmission as suggested by some [ 84 ] and countered by others [ 85 ]. The significance of these findings is that sperm mtDNA is normally eliminated prior to the onset of embryonic genome activation in each species [ 79 , 82 ] and thus, not normally observed or mistakenly construed for pseudo copies of the sequence within the nuclear genome [ 86 ]. Several mechanisms for sperm mtDNA elimination have been proposed. These include elimination in the fertilized oocyte [ 82 , 87 – 89 ] likely mediated by an autophagy- or, more specifically, a mitophagy-mediated process [ 90 ] through, for example PARKIN- and MUL1-dependent processes [ 91 ], or before fertilization through targeted destruction of the mitochondrial genome mediated by loss of activity of the nuclear-encoded transcription and replication factors such as POLG [ 92 ] and TFAM [ 93 ]. Indeed, it has been previously shown in mouse models that an isoform of TFAM is expressed that prevents its translocation to nucleus and, as spermiogenesis takes place, mtDNA copy number is reduced. Others have shown that TFAM is phosphorylated at a pre-sequence motif that prevents its translocation and import into the mitochondrion and directs TFAM to the nucleus [ 93 ], although this remains to resolved [ 94 ]. Nevertheless, the presence of sperm mtDNA in the offspring of interspecific crosses remains a conundrum that is yet to be seriously addressed. It is conceivable that the recognition motifs associated with sperm mtDNA elimination in intraspecific crosses are not specific enough to overcome its persistence in interspecific crosses. If biparental inheritance were to take place, it has been proposed that the transmission of two distinct genomes, described as heteroplasmy, would result in the two genomes competing with each other. In this respect, the genome harbouring mtDNA variants that exhibited greater rates of replication efficiency would become the dominant genome [ 95 ]. In some cases, this would be advantageous resulting in improved OXPHOS efficiency [ 96 ] but, in others, it could affect the organism’s fitness given that certain mtDNA genotypes are less efficient at OXPHOS generation than others [ 97 ] and can result in a host of pathologies [ 98 ] with heteroplasmic shifts, i.e. selection of one genotype over another, being also age-dependent and transgenerational [ 99 ]. Consequently, uniparental inheritance by-passes the consequences imposed by the ‘selfish genome’ [ 100 , 101 ]. In contrasting arguments, it has been proposed that uniparental inheritance through the female germline can take place at the detriment of the health and well-being of male offspring, otherwise described as the ‘Mother’s Curse’ [ 102 ] or the ‘sex-specific selective sieve’ [ 103 , 104 ]. Consequently, mutations advantageous to female offspring might be disadvantageous to their male counterparts [ 95 ], which could explain the longer lifespan’s associated with females, although they perform metabolically more demanding functions that include pregnancy.

Given that nuclear DNA copy number remains constant (except for cases of aneuploidy and other chromosomal rearrangements), as it divides and replicates itself equally between each daughter cell and for each cell type, the genomic variable within a cell at any given stage of development is mtDNA copy number. This is coupled with epigenetic regulation of nuclear DNA and/or mtDNA that varies according to cell type (Fig. 6 ). However, as these variables result in changes to one of the genomic compartments, the other compartment would need to synchronously change to establish genomic balance and promote cellular function. To understand this, it is perhaps best to draw on other biological systems. The establishment of genomic balance. All cells establish a balance between their two genomes, ‘genomic balance’ to ensure efficient cellular function at any given time and stage in development. The nucleus encodes the mtDNA transcription and replication factors some of which are DNA methylated to match a cell’s specific requirements for mtDNA copy number to meet its requirements for OXPHOS-derived ATP. Likewise, the nuclear genome is epigenetically regulated through, for example DNA methylation, which regulates gene expression. Mutations and deletions and copy number variants also act on the nuclear genome and affect its control of cellular fate. Cellular metabolism is regulated through the mitochondrial genome, which can alter its mtDNA copy number, as well as the cell’s mtDNA genotype (haplotype) influencing fate and phenotype. Additionally, metabolic factors released from the mitochondrion can modulate DNA methylation and other epigenetic modifiers that act on both genomes (dashed green lines). The figure is reproduced and modified from [ 236 ] under an open access Creative Commons CC BY 4.0 license. Tumorigenesis and oogenesis are often regarded as being complementary biological systems [ 228 ]. Both arise from founder cells that undergo exponential proliferative increases in cell number that then invade or implant to form a large mass that has (a degree of) differentiation potential. The tumour will either result in a benign mass that will not affect the individual; or a malignant mass that will result in death, if intervention does not take place. The embryo will give rise to a foetus and then a live offspring but also has the potential to arrest, fail to implant or spontaneously abort. Both primarily utilize glucose metabolism to promote replication and proliferation [ 209 ], which is mediated by low mtDNA copy number that restricts aerobic metabolism through OXPHOS [ 30 , 128 , 184 , 229 ]. As a result, highly proliferative cells have the propensity to use aerobic glycolysis, otherwise described as the Warburg effect [ 230 , 231 ]. They will produce various metabolites that are necessary to support significant increases in cell division and stringent patterns of epigenetic regulation that will influence their fate [ 30 , 209 , 225 , 232 ]. From studies of tumour models, it is evident that there is a direct relationship between mtDNA copy number and DNA methylation of tumour-initiating cells [ 218 , 229 ]. Tumour-initiating cells, such as Glioblastoma multiforme cells, can be treated with DNA demethylation agents, namely 5-azacytidine or vitamin C which act through the DNMT [ 233 ] and TET [ 234 ] pathways, respectively [ 218 ]. These agents induce global DNA demethylation including DNA demethylation of the mtDNA-specific replication factors, and specifically POLG . As a result, these cells acquired the potential to differentiate into mature astrocytes which is indicative of their lineage of origin. This was coupled with significant, but anticipated, increases in mtDNA copy number. Likewise, there are similar occurrences when these cells undergo partial depletion of their mtDNA content through a mtDNA depletion agent [ 183 , 229 ]. This outcome is not just limited to tumour-initiating cells. For example, when somatic cells are incompletely reprogrammed to become induced pluripotent cells, their potential to differentiate stalls and they fail to replicate mtDNA copy number in a manner similar to their embryonic stem cell counterparts [ 235 ]. However, treatment of induced pluripotent stem cells with 5-azacytidine induces differentiation potential and synchronous increases in mtDNA copy number. Consequently, the resetting of one or both genomes alters the cell’s genomic balance and overcomes the trap that cells fall into when the two genomes are in a state of asynchrony [ 236 ]. When depleted tumour-initiating cells are introduced into immune-deficient mice, their potential to invade and give rise to viable, malignant tumours is a function of the levels of mtDNA present in the cell at the time of inoculation [ 183 ]. Cells depleted to 50% of their original mtDNA content form tumours more rapidly than non-depleted cells. However, cells depleted to 2% and 0.2% of their original mtDNA content take significantly longer to form and the frequency of tumour formation significantly declines. In each tumour, mtDNA copy is re-established to pre-depleted levels and similar to tumours formed from non-depleted cells [ 183 , 218 ]. Furthermore, the tumours derived from more extensively depleted cells exhibited remodelling of their nuclear DNA methylation profiles, including the mtDNA-specific replication factors, namely, POLG, TWNK , and TOP1MT [ 237 ]. In addition, cells re-acquired mtDNA variants identified in glioblastoma cells [ 238 ]. The power of mtDNA copy number to influence tumour progression is not just restricted to solid tumours. Similar outcomes were observed in myeloma cells [ 239 ]. In this case, no tumours were formed from cells possessing less than 0.2% of their original mtDNA content. Interestingly, tumour-initiating cells that are completely deplete of their mtDNA content scavenge mtDNA from cells in their neighbouring environment, such as the stroma cells, in order that they can progress to form tumours [ 240 ]. These outcomes further highlight the balance that the two genomic compartments of a cell need to establish before a cell can be functionally competent.

There are large number of documented mitochondrial diseases which are associated with the nuclear and mitochondrial genomes [ 105 ]. In the mtDNA context, we normally inherit wild-type (WT) copies, a situation otherwise described as homoplasmy [ 106 ]. However, in control individuals, heteroplasmy, in this case the mixing of wild type and either pathogenic or non-pathogenic mutations/deletions, can be detected at 0.2%–2% as shown through ultra deep sequencing [ 107 ]. This suggests that homoplasmy, as previously understood, is within the range of 99.8%–98%. Nevertheless, some women harbour pathogenic mtDNA mutations and deletions that can cause severe and even fatal diseases [ 108 ]. In all, 1 in 200 women carries a known pathogenic mtDNA point mutation [ 109 , 110 ] and 1 in 5000 of the population is affected by mtDNA mutations [ 108 ]. The mtDNA diseases are frequently multi-organ or multi-systemic. The key mtDNA diseases associated with point mutations in the coding genes include Leber’s hereditary optic neuropathy (LHON) [ 111 ]; Neuropathy, ataxia, and retinitis pigmentosa (NARP) [ 112 ]; and Leigh Syndrome [ 113 , 114 ]. On the other hand, diseases such as mitochondrial myopathy, encephalomyopathy, lactic acidosis, and stroke-like episodes (MELAS) [ 115 ]; and myoclonic epilepsy with ragged red fibres (MERRF) [ 116 ] are associated with point mutations in the tRNAs. Kearns–Sayre Syndrome and chronic external progressive ophthalmoplegia result from a large scale deletion that removes 4977 bp of the mitochondrial genome including several coding genes and tRNAs [ 117 ]. In all, these mutations and deletions primarily affect tissues that are constituted by high OXPHOS-requiring cells, such as muscle and brain cells which can harbour a high proportion of mtDNA mutations or deletions [ 105 , 108 ]. The onset of a mtDNA disease is very much dependent on the transmission and the random segregation of mtDNA rearrangements through the female germline [ 77 , 118 ], which originates from their prior segregation to the PGCs [ 51 ]. Indeed, the levels of a mtDNA mutation can vary amongst a carrier’s oocytes. For example, in the case of the 8993 (T→G) mutation that gives rise to NARP, a carrier harboured either 0% or >95% mutant load in a cohort of her oocytes [ 119 ] (Fig.  4 ). In other cases, for example, the 3243 (A→G) mutation that gives rise to MELAS, carriers can have a more variable range in mixing of heteroplasmy amongst their oocytes [ 120 ] (Fig.  4 ). The severity of the disease is determined by the amount of mutant load that segregates to affected tissues, which contain high OXPHOS-requiring cells. Consequently, as mutant mtDNA segregates randomly during development and oocytes have variable mutant load, carriers would not know if their child would be affected [ 120 ] (Fig.  4 ). Nevertheless, it appears that pathogenic mutations do not necessarily impact on oocyte quality, fertilization outcome, or embryo-fetal development [ 121 ]. Indeed, it is argued that cells harbouring mutant copies of mtDNA would increase their mtDNA content to overcome negative selection [ 121 ]. The random selection of mtDNA mutations for germline transmission. The molecules present in the PGCs are a result of the random segregation of mtDNA molecules to these very early germ cells and likely originated from the fertilized, mature, metaphase II oocyte that gave rise to them. As the PGCs undergo differentiation to produce the next generation of mature, fertilizable oocytes, mtDNA copy number increases exponentially and mutant copies of mtDNA can be either selected for or against. As a result, there will be differences in mutant load within a cohort of oocytes from an individual, as in the case for carriers of the mutation associated with MELAS. However, for women who are carriers of the NARP mutation, the mutant load in their oocytes can vary more considerably with some oocytes having very high levels and others almost none. WT = wild-type; Δ = mutant. The figure is reproduced and modified from [ 232 ] with permission. The mitochondrial genetic bottleneck proposes that a small population of mtDNA populates the offspring of the next generation and can result in the variability of heteroplasmic load that exists in the mature oocytes of carriers of mtDNA disease [ 122 ]. Consequently, it is likely a filtering mechanism that can incorporate bias by fixing or eliminating mutations in the female germline within a generation, as demonstrated in a bovine model [ 123 ]. It is further argued through studies in mouse models that either good quality mtDNA is subsequently transmitted through the oocyte or the mutation is lethal and the oocytes do not mature [ 124 , 125 ]. However, this is only likely to be the outcome if the mutation is extremely severe, otherwise mtDNA disease would have been eradicated and heteroplasmy continues to be observed in oocytes [ 75 , 76 ] and offspring and persists to adulthood [ 108 ]. As a caveat, it might be pertinent to consider the models employed to study the mitochondrial genetic bottleneck and, specifically, their regulation of mtDNA copy number from fertilization to gastrulation. In the mouse, there is no overall (embryo total) decrease in mtDNA copy during preimplantation development [ 126 ], but there is a decrease in mtDNA copy number per cell [ 127 ] and that mtDNA copy number per cell likely continues to decrease from the hatching blastocyst stage up to gastrulation, as observed in embryonic stem cell models [ 128 ]. In larger mammals, such as pigs [ 129 ] and cattle [ 130 ], mtDNA copy decreases from the 2-cell stage up to the blastocyst stage in both individual cells [ 131 ] and the embryo per se [ 129 ] which is similar to reports for humans [ 132 ]. Consequently, the question that remains to be addressed is whether retention of mutant molecules is greater or equal between the early and late reduction processes. Interestingly, recent findings in a mouse model suggest that mtDNA mutations could be selected against by Ubiquitin-specific peptidase 30 during maternal–zygotic transition [ 133 ]. This would be in line with the pre-embryonic genome activation mtDNA replication event that takes place in the mouse [ 134 ]. Nevertheless, it appears from studies in Drosophila that wild-type molecules tend to be favoured through selective replication rather than elimination [ 135 ], perhaps accounting for the low penetrance of the mtDNA diseases (1:5000) [ 108 ] when compared with the number of carriers (1:200 women) [ 109 , 110 ].

mtDNA rearrangements can be maternally inherited pathogenic mtDNA mutations and deletions or spontaneous, somatic, or de novo rearrangements that naturally occur at low levels within an individual. Indeed, the majority of mtDNA variants are non-pathogenic [ 136 ]. Naturally occurring variants have been identified in humans [ 107 , 137 , 138 ], and it has been argued that they can affect health and well-being and lead to disease [ 139 – 142 ]. Furthermore, it has been suggested that maternal age can influence variant transmission from the mother to the child [ 143 ]. Others have suggested that mtDNA mutations and deletions would be indicative of oocyte quality and would be higher in women of older reproductive age [ 144 , 145 ]. However, the frequency of the 4977 bp deletion, which is often regarded as an age associated indicator, has not produced outright outcomes. In a study of 181 human oocytes and 104 embryos, its frequency was 47.0% and 20.2%, respectively, with no association reported with ageing [ 145 ] whilst two other studies identified links to ageing. In oocytes from older women, there was a greater likelihood of their harbouring the 4977bp deletion than oocytes from younger women [ 146 ]. Likewise, the incidence of 4977 bp deletion was significantly higher in women older than 35 years of age [ 147 ]. Interestingly, multiple mtDNA large scale deletions appear to be more frequently observed in oocytes (50.5%, n = 295) than embryos (32.5%,  n  = 197) with no correlation with ageing [ 144 ]. However, there were significant reductions in the frequency of oocytes possessing mtDNA deletions as oocytes progressed from the germinal vesicle to the metaphase II stage suggesting a cleansing process in the final stages of maturation, whilst others have shown an increase in the 4977 bp deletion associated with meiotic arrest [ 148 ]. When a similar scenario was modelled in cattle, where a series of females were derived by somatic cell nuclear transfer (SCNT), an approach that ensured the same genotype was maintained, it was evident that ovarian ageing (age range = 3 and 10 years old) significantly increased the number of oocytes carrying mtDNA deletions [ 149 ]. However, there were no similar correlations for point mutations. In the context of point mutations, the incidence of a variant at T414G in the D-loop region was lower (4.4%) in the oocytes of women from 26 to 36 years of age compared with the oocytes of women from 37 to 42 years of age (39.5%, P  < 0.01) [ 150 ]. Whilst each of the above studies was associated with frequency, mutations appear to be common amongst oocytes but with large variability in heteroplasmic load (<5% to 50%) [ 151 ]. Although there was an abundance of rearrangements, there were no age-related associations and the outcomes to date are inconclusive in the context of oocyte quality. In a study using the pig as a model, it was shown that naturally occurring variants can persist at high levels in oocytes and preimplantation embryos, and they can be transmitted from one generation to the next through the female germline. Indeed, each of the four variants analysed indicated different levels in oocytes and embryos but their levels were suppressed in somatic tissues, especially those with a high requirement for OXPHOS-derived ATP [ 52 ]. Consequently, the levels did not surpass the threshold associated with the phenotypic onset of mtDNA disease [ 122 , 152 ]. For example, over 60% mutant molecule is required for the onset of LHON [ 153 ], and 85% mutant load for MERRF [ 154 ]. Interestingly, there were gender specific differences for heart and liver tissues for one variant and generational differences for several tissues for the same variant [ 52 ]. Indeed, the consensus opinion is that mtDNA point mutations associated with mammalian aging are insufficient to cause a phenotypic response [ 155 ]. For example, in hair, the exponential increase in mutation with age ranges from 0 to 1.436 ± 0.2086% of total mtDNA content [ 156 ], which, in the context of mtDNA copy number per cell suggests little if no impact. However, there is evidence from mouse models generated through ooplasmic transfer to suggest that there are differences in heteroplasmy between generations [ 99 ], and that the divergence of heteroplasmic molecules in germ and somatic lineages takes place early in development [ 157 ], which suggest two distinct pathways for transmission, one for the germline and one for somatic tissues. Nevertheless, analysis of the breeding lines of commercial pigs in Australia showed that there was no correlation between the total number of mtDNA variants harboured by each of the mtDNA haplotypes investigated and developmental competence, maturation to metaphase II, fertilization rates, blastocyst rates, and litter size [ 158 ]. However, when individual variants were assessed at a presence of >25% heteroplasmic load, there was a negative correlation with oocyte developmental competence; and more specifically with the number of variants present at >25% in the Cytochrome B (Complex III) gene. Furthermore, the level of a variant at position 16 383 in CSBII correlated positively with mtDNA copy number for developmentally competent oocytes. This particular variant is within the site of interaction for the transcription elongation factor TEFM in CSBII [ 65 ]. Termination of transcription at this site within CSBII results in the transition from transcription to replication [ 159 ]. In all, there is no clear-cut evidence to suggest that the presence of somatic mutations and deletions in oocytes is indicative of oocyte quality or the potential of any given oocyte to give rise to an offspring. However, it is worth noting that mutations to POLG can result in a host of deleterious mtDNA mutations in oocytes, as demonstrated in the Polg mutator mouse model (Ma et al. , 2020). Nevertheless, these should be seen in the context of the mutations associated with the nuclear-encoded mtDNA replication factors that can give rise to mtDNA disease [ 160 – 162 ]. Likewise, it is important to note that high heteroplasmic loading of pathogenic variants can lead to spontaneous abortions even at late stages of gestation [ 120 ]. Indeed, the persistence of mtDNA in humans might be explained by the high numbers of carriers of mtDNA rearrangements and efficient medical practice to support affected individuals; and the likely elimination in livestock species since affected animals would have been selected against, based on their commercial non-viability.