Our results show that mitochondrial DNA deletion ratio (mtDNADR) is higher in arrested GV stage oocytes compared to mature human oocytes in an IVM protocol (using G-IVF™ PLUS media for 48 h). The cross-sectional study design limits inference based on causality, but the most plausible interpretation of these findings is that mtDNA deletions impair oogenesis. Biologically, this is plausible given that following MI, and without an intervening round of DNA replication, the oocyte proceeds to meiosis II, rapidly reforms a spindle and arrests at metaphase MII, a cellular activity requiring massive organelle and cytoskeletal reorganization with concomitant expenditure of ATP. Intriguingly, in oocytes mitochondria are underdeveloped, primitive structures with truncated cristae with only minimal oxygen uptake [20]. Presumably ATP generated at earlier stages of oogenesis is essential to meiotic progression.

We assume that contributions from our IVM protocol to promote ROS damage and therefore to confound our findings to be negligible. Recent studies employing biochemical and functional assays have confirmed that mitochondrial metabolism in oocytes is suppressed by inhibition of complex I. This is understood as an evolutionarily conserved strategy to protect long-lived oocytes [21].

Our data also shows that progression from MI to MII is positively correlated with mtDNA deletion load (OR = 0.10). The statistical significance of such a small OR may result from the fact that ORs tend to exaggerate the effect size compared with a relative risk [22], thus representing a Type I error. Alternatively, the minimal checkpoint control after exit from meiotic prophase I, and absence of DNA damage checkpoint before the first meiotic division [23], may permit oocytes containing damaged mtDNA to progress.

The positive correlation between age and time to polar body extrusion (PBE) is intriguing, since age-associated meiotic aneuploidy is one of the most important processes affecting human fertility. Expedited progression through the first meiotic division in oocytes from older women presumably would reduce the time available for proper chromosome congression prior to chromosome segregation [24]. This has clinical consequences, suggesting a shorter interval from trigger to oocyte retrieval in older patients is warranted to prevent early ovulation and loss of oocytes.

In our study we did not look at the resulting functional effects of the mitochondrial DNA deletions on the oocytes. However, what is known is that most of the mitochondrial deletion breakpoints occur within two directly repeated sequences, which are thought to cause most large-scale mtDNA deletions [25, 26]. Among the 263 large-scale deletions these frequently locate in the major arc, especially the ND4 and ND5 gene. Furthermore, these deletions occur within perfect repeats (68%) (class I deletions), whereas 12% deletions are flanked by imperfect repeats (class II deletions) and 20% deletions are flanked by no direct repeats (class III deletions) [27]. mtDNA 4977 bp deletion has been frequently reported within the category of class I deletions [28]. The two 13 bp perfect repeats (ACCTCCCTCACCA) have been found at mtDNA nucleotide positions 8470–8482 bp (in the ATPase8 gene) and 13,447–13,459 bp (in the ND5 gene) surrounding this deletion breakpoint. One repeat remains, whereas the other is removed. This deletion removes two complex V subunits, one complex IV subunit, four complex I subunits and five intervening tRNAs. The deleted 5 kb subgenomic fragment therefore lies in the hotspot of the distribution of deletions in the major arc [28].

Recent advances in live imaging techniques have revealed a functional heterogeneity of mitochondria with respect to mitochondrial redox state, membrane potential, respiratory activity, uncoupling proteins, mitochondrial ROS and calcium [29] which could be a consequence of deletions/mutations. The heterogeneity of mitochondrial function demonstrates an additional level of mitochondrial complexity. For example, mitochondria can be classified according to their membrane potential, and because the magnitude of potential is directly correlated to important mitochondrial functions, including ATP production, there is the awareness that mitochondria are not a homogeneous population and that specific localization of subpopulations characterized by different membrane potential may correspond to precise oocyte regulatory needs. It is notable that the localization of highly polarized, and therefore more functional mitochondria occurs in a timely fashion in a specific oocyte domain where they are believed to be crucially needed. Several lines of evidence in fact suggest the notion that hyperpolarized mitochondria are essential for supporting the very first steps of fertilization at the cortical level [30]. Future studies should establish whether mtDNADR load can affect mitochondrial function and how and if the mtDNADR load can predict other downstream oocyte competency indicators such as fertilization rate, good blastocyst development rate, implantation rate and live birth rates.

Aging is associated with a general cellular and mitochondrial impairment affecting tissue function. Tissues with slow turnover of mitochondria, such as the ovary, develop physiologic deficits via mitochondria-regulated apoptosis. mtDNA also becomes compromised as a tissue ages due to the acquisition of mutations. The 4977 base pair common deletion has been detected in oocytes from reproductively older women [16]. In this particular study, it was shown that unfertilized oocytes obtained from women undergoing IVF presented deleted mtDNA (common deletion). However, IVF pregnancy rates did not differ between those patients whose unfertilized oocytes harbored mtDNA deletions and those patients whose oocytes did not harbor mtDNA deletions. Nevertheless, this can be explained in that the high-frequency deletions could arise in individual oocytes rather than the whole patient pool of oocytes. Unfertilized oocytes were used in that particular study, and it is unknown whether those findings can be generalized to those embryos transferred to patients.

Arbeithuber et al. recently detected low-frequency, de novo point mutations in mtDNA from oocytes and somatic tissues. Mutation frequencies and patterns differed between germline and somatic tissues and among mtDNA regions, suggestive of distinct mutagenesis mechanisms. Additionally, they discovered more pronounced genetic drift of mitochondrial genetic variants in the germline of older versus younger mice, arguing for mtDNA turnover during oocyte meiotic arrest [31].

The possible impact of mitochondrial aging on female reproductive capacity was investigated by Duran and colleagues [32]. The study involved examination of the ATP content, mitochondria number and the presence of the 4977 bp deletion in individual human oocytes at different stages of maturation (i.e. germinal vesicle [GV]; meiosis I [MI]; meiosis II [MII]). Mitochondrial numbers, estimated by mtDNA copy number, were more closely associated with reproductive age (as determined by measurement of FSH levels), than chronological age. Additionally, ATP content increased as oocytes matured, but its amount was not related to chronological age. The 4977 bp deletion was identified only in arrested or degenerate oocytes [32].

Clinically significant heteroplasmy, in which cells contain mixtures of mtDNA with different sequences, is frequent. The likelihood of heteroplasmy persisting across generations is reduced by a mechanism regulating mitochondrial segregation during oogenesis known as the “genetic bottleneck” [33]. This involves the elimination of the majority of the oocyte mtDNA molecules, with only a small population ultimately passed to the next generation. In most cases, this genetic bottleneck results in either the removal of mtDNA variants harboring mutations or their increase past a critical level of heteroplasmy, causing tissue dysfunction and possibly mitochondrial disease. We cannot rule out the possibility that meiotic maturation is contributing to this genetic mechanism.

One possible limitation of our study is its generalizability as oocytes were derived from stimulated cycles and oocytes were matured in vitro. Importantly, our study applied the same culture conditions to all oocytes stages to focus on the possible role of mtDNA deletions on development. We are aware that we cannot extrapolate our findings to other IVM protocols.

But we have no reason to believe that meiotic maturation in vivo would be immune from the effects of high levels of mtDNA deletions. Also, we do not understand the mechanism underlying variation in the levels of the common mtDNA deletion among oocytes. We are not aware of evidence that gonadotropin stimulation influences mtDNA deletion ratio load, though definitive conclusions on this matter await studies on oocytes obtained from natural cycles.

We did not find a statistically significant difference in mtDNA copy number among the oocyte maturation stages, though we likely were underpowered to detect such differences. A post hoc power analysis showed that the number of oocytes needed to achieve statistical power was 6180, a number that was not feasible for this study. We recognize the limitation of our sample size in this respect; however, mtDNA deletion ratio appears to have a more robust effect than mtDNA copy number, since we detected statistically significant differences, even with the relatively modest sample size studied. mtDNA copy number has provided the metric most commonly used to estimate mitochondrial function in zygotes but has been questioned by many contradicting studies [34, 35]. mtDNA deletion ratio thus appears to provide a more sensitive metric to assess mitochondrial function, at least during meiotic maturation. Future studies should assess mtDNA deletion ratio during later stages of development.

With the increased emphasis on three-person reproduction as a method to curb the inheritance of mitochondrial disease, further understanding of mitochondrial mutational load on oocyte biology may lead to better outcomes for patients in the future.