Mitochondrial genome inheritance and replacement in the human germline

Abstract

Mitochondria, the ubiquitous power packs in nearly every eukaryotic cell, contain their own DNA, known as mtDNA, which is inherited exclusively from the mother. The number of mitochondrial genomes varies depending on the cell's energy needs. The mature oocyte contains the highest number of mitochondria of any cell type, although there is little if any mtDNA replication after fertilization until the embryo implants. This has potential repercussions for mitochondrial replacement therapy (MRT; see description of currently employed methods below) used to prevent the transmission of mtDNA‐based disorders. If only a few mitochondria with defective mtDNA are left in the embryo and undergo extensive replication, it might therefore thwart the purpose of MRT. In order to improve the safety and efficacy of this experimental therapy, we need a better understanding of how and which mtDNA is tagged for replication versus transcription after fertilization of the oocyte.

mtDNA is a circular molecule coding for only 37 genes that are critical to energy production by oxidative phosphorylation. mtDNA has extensive interactions with nuclear DNA (nDNA), which contains additional genes involved in oxidative phosphorylation as well as factors needed for mtDNA replication, transcription, and translation. mtDNA is also more susceptible to mutation than nuclear DNA, owing in part to its proximity to reactive molecules, a susceptibility to replication errors and the absence of histones. Several hundred diseases are known that are related to mutations of mtDNA, some inherited via the germline and others acquired through somatic mutations. Mitochondrial dysfunctions caused by inherited mtDNA mutations underlie complex, multifaceted conditions that affect the brain, the neuromuscular system, the heart, and other highly energy‐dependent organs. Clinical symptoms often vary depending on the particular mutation and its heteroplasmy, that is, its penetration through the mitochondrial DNA population, which is expressed as a ratio of mutant to wild‐type mtDNA within a cell. In contrast, homoplasmy applies if all mtDNA molecules contain the mutation. The vast majority of mtDNA‐related diseases are caused by heteroplasmic mutations, and the level of heteroplasmy can drastically change during development; consequently, it is difficult to forecast the risk that a child will inherit a mtDNA‐related disease from its heteroplasmic mother. Moreover, our understanding of the underlying mechanisms of intergenerational mtDNA transmission is largely incomplete.

MRT to prevent second‐generation passage of mtDNA‐based disease

The lack of understanding of the intergenerational mtDNA transmission mechanisms has direct relevance for MRT to prevent mtDNA‐related disease. Conventional preimplantation genetic diagnosis (PGD) is based on the genetic analysis of several cells taken from multiple cleaving embryos after in vitro fertilization. Embryos with low mutation levels are then selected for uterine transfer to initiate pregnancy. PGD can obviously not be used to prevent the transmission of mtDNA‐related disorders if the woman carries homoplasmic mtDNA mutations since all oocytes and all resulting embryos would inherit the homoplasmic mutation. In the case of heteroplasmic mutations, it creates the inevitable dilemma of what is regarded as a safe mutation level in the biopsied embryo, that is, which level of heteroplasmy will not likely cause disease in the child.

There are conflicting estimates of acceptable mutation loads for PGD selection, ranging from as high as 30% to as low as < 5%. Whatever the ratio, it limits the number of available embryos for uterine transfer. Another problem is that the proportion of mutant to wild‐type mtDNA in the biopsied cells may not represent the mutation levels in the other cells owing to the uneven segregation of heteroplasmic mtDNA in early embryos (Lee et al, 2012). Another challenge for the success of PGD is that the mutation load in the preimplantation embryo must stay at the same level during subsequent development. However, several studies observed that heteroplasmy levels determined in mature oocytes, zygotes, or preimplantation embryos can change drastically later during fetal development. One clinical case involves a 30‐year‐old patient with mitochondrial encephalomyopathy and a 35% heteroplasmy level, who had an affected daughter with a 84% mutation load. The patient underwent IVF/PGD treatment to have a second, unaffected child; 23 oocytes were recovered that resulted in seven blastocysts after fertilization. A blastocyst with 12% heteroplasmy was transferred and eventually resulted in the birth of a male baby. However, subsequent analysis by a specialized diagnostics laboratory on samples collected from the boy at 6 weeks and 18 months after birth demonstrated mutant heteroplasmy loads of 47% and 46% in blood and 52% and 42% in urine, respectively (Mitalipov et al, 2014). Such observations question the clinical validity of PGD to select against mtDNA‐related diseases.

Given that mtDNA is inherited exclusively from the mother, MRT is a viable strategy to prevent the passage of any mtDNA mutation from mother to child by using mitochondria from healthy egg donors. Two main approaches have been proposed: spindle transfer (ST) replaces mitochondria with mutated mtDNA in unfertilized patient oocytes with wild‐type mtDNA from donor oocytes; pronuclear transfer (PNT) replaces mitochondria in one‐cell embryos with healthy ones from donor embryos (Wolf et al, 2015).

A common problem of both procedures is that the replacement of mutant mtDNA is not complete. A small fraction of maternal mtDNA, termed carryover, will inevitably persist. A carryover amount of 1% or less is expected after the ST procedure (Kang et al, 2016) while it can be as high as 4% in PNT (Hyslop et al, 2016). However, although these initial, low levels of mutant mtDNA are considered insignificant, its persistence and expansion during subsequent embryo development is a potential problem.

First evidence of genetic drift or reversal after fertilization came from heteroplasmic monkey oocytes that were experimentally generated with a defined mixture of two wild‐type mtDNA haplotypes. After fertilization to establish pregnancies or to derive embryonic stem cell (ESC) lines, the resulting offspring and ESCs unexpectedly carried predominantly one mtDNA haplotype (Lee et al, 2012). This suggests a shift to the homoplasmic condition during early development of a zygote, without actual passage of mtDNA through the ovarian germline.

Similarly, when human MRT embryos were generated, carryover for the maternal mtDNA was, as expected, below 1%. However, during subsequent culture to ESCs, some of the cell lines showed rapid reversal back to the homoplasmic maternal mtDNA independent of donor mtDNA haplotype or sequence (Kang et al, 2016; Yamada et al, 2016). A similar phenomenon has been reported for human ESCs derived from embryos generated by somatic cell nuclear transfer (SCNT): some ESCs reversed back to the somatic mtDNA haplotype of the skin fibroblasts from which the nuclei were taken (Kang et al, 2016). This reversal to homoplasmy and the complete loss of donor mtDNA were observed in 15% of ESC lines and it was much higher than the 1% expected based on stochastic genetic drift. These observations raise the possibility that some children, who were conceived after MRT, might revert to the original, maternal mutant mtDNA (Fig 1). Therefore, a better understanding of the mechanisms of mtDNA transmission is essential for improving the success rate of MRT.

Figure 1. Mitochondrial replacement in human oocytes by spindle transfer and subsequent reversal to the maternal mtDNA in ESCs.

Mitochondrial replacement by spindle transfer isolates and transfers a meiotic spindle apparatus with the chromosomes from an unfertilized maternal oocyte into a donor oocyte cytoplasm containing healthy mtDNA. Blue and orange dots depict normal and mutant mtDNA, respectively. Replacement is not absolute and results in a small (less than 1%) carryover of mutant mtDNA from the maternal oocyte. After fertilization, preimplantation embryo development, or expanded culture of ESCs, the proportion of carryover mtDNA may increase resulting in complete reversal to the mutant maternal mtDNA.

Intergenerational mtDNA transmission and a post‐fertilization bottleneck

The observations of rapid shifts in mtDNA heteroplasmy between generations, with a return to homoplasmy, have spawned the hypothesis of a mtDNA bottleneck during development. Although the precise mechanism and the timing when it occurs during development are poorly understood, there is general agreement that the bottleneck occurs in the ovary during oocyte development, possibly by a marked reduction in mtDNA copy number in primordial germ cells followed by mtDNA segregation into mature oocytes (Stewart & Chinnery, 2015). Once oocytes have matured and ovulated, their heteroplasmy levels remain steady. This ovarian mtDNA bottleneck appears passive and random and results in different levels of heteroplasmy in individual oocytes.

However, this model does not explain the dramatic mtDNA switch between the oocyte/zygote stage and ESCs or offspring as we described above. Available data suggest that, in contrast to the ovarian bottleneck, the post‐fertilization mtDNA bottleneck is a more active process and involves selective replication of a small subpopulation of mitochondria in unfertilized, mature oocytes. Further investigations are necessary to clarify the exact nature of this phenomenon.

During preimplantation development, mtDNA is not replicated and the original number of mitochondria, along with their mtDNA, that are present in the oocyte are distributed to the embryonic blastomeres during cell divisions. Active mtDNA replication is evident at the late blastocyst stages when the reversal likely occurs (Fig 2). Some evidence suggests that preferential replication of specific mtDNA haplotypes may be a result of a sequence polymorphism within the control region, called conserved sequence box II (CSBII). Analysis of donor and maternal mtDNA replication in reversed ESC lines suggested that certain haplotypes of the CSBII sequence may indeed confer replicative advantage. However, other reversed ESCs did not have CSBII sequence differences, which indicates other, mtDNA sequence‐independent, mechanisms in selective replication (Kang et al, 2016).

Figure 2. Model of the post‐fertilization mtDNA bottleneck.

During the first ovarian bottleneck, individual mature oocytes may acquire various heteroplasmy levels of mtDNA mutation. These mutation levels in oocytes may change again after fertilization during subsequent embryonic development resulting in a dramatic increase in mutant mtDNA levels in offspring. This rapid shift in mtDNA heteroplasmy is likely due to preferential replication of a small selected population of mtDNAs in mature oocytes. The vast majority of mtDNA molecules (99%) in an oocyte are not replicated and will be lost during subsequent embryonic development. Similarly, paternal, sperm mtDNA introduced during fertilization will be passively lost due to lack of replication. Replication‐competent mtDNA (1%) is likely marked epigenetically by the time of oocyte maturation and co‐localized within perinuclear compartment in human oocytes. PGC: primordial germ cell. Blue, orange, and green dots represent normal, mutant, and sperm mtDNA, respectively. Star‐marked red dots depict replication‐competent mtDNA.

We propose an alternate hypothesis on the post‐fertilization genetic bottleneck in which a few mtDNA molecules in mature oocytes are epigenetically tagged for subsequent replication, while most of the remaining mtDNA is marked for transcription only (Fig 2). It is likely that these replication‐competent mtDNAs in human oocytes or zygotes are co‐localized in close proximity to the spindle apparatus or to pronuclei. The MRT procedures described above would therefore transfer some of this replication‐tagged population of maternal mtDNA along with the nucleus while the enucleation of donor oocytes/zygotes would deplete the cytoplasm of replicative mtDNA from the donor oocyte. This model could explain how < 1% of maternal carryover mtDNA can be selectively propagated leading to complete reversal after MRT. We also postulate that lack of replication‐tagged mtDNA in sperm can similarly explain why paternal mtDNA, which is introduced during fertilization, is excluded from replication during subsequent embryonic development. Post‐MRT reversal was not reported in mice and monkeys suggesting that this post‐fertilization bottleneck is more distinct in humans compared to other animals implying species‐specific differences in the underlying mechanism.

Epigenetic regulation of mtDNA replication

Epigenetic mechanisms that could be involved in selective labeling and preferential amplification of specific mtDNA molecules in the oocyte could be mediated by nuclear‐encoded transcription and replication factors, which, along with mtDNA, are packed into mitochondrial nucleoids (Milenkovic et al, 2013). Some evidence suggests that mtDNA replication and transcription are mutually exclusive, which further supports the notion that only selected mtDNAs are replicated while the remaining molecules are utilized for transcription.

Some of the nucleoid components involved in mtDNA replication include TFAM, ATAD3 (ATPase family AAA Domain‐containing protein 3), PolG, TWINKLE, mtSSB (mitochondrial single‐stranded DNA‐binding protein), and 7SDNA. TFAM belongs to the HMG box family of genes that are associated with chromosomal DNA packaging. Since there is no histone chromatin structure in mitochondria, TFAM and ATAD3 together serve as the mtDNA backbone. TFAM is a 25‐kDa protein and has two important functions: binding and bending mtDNA. The binding function suggests a direct relationship between TFAM and mtDNA copy number, while bending suggests a role in mtDNA replication initiation. Upregulation of TFAM expression results in an increased expression of PolG and an increase in mtDNA copy number. ATAD3 is also implicated in mtDNA replication: it binds to mtDNA at the D‐loop region and is involved in nucleoid formation and segregation.

7S DNA (or D‐loop chain) is a short mtDNA strand—approximately 650 bases long in humans—which is formed through premature termination of heavy‐strand replication. Its 5′end coincides with the origin of heavy‐strand replication, Ori‐H. The function of 7S DNA is still unknown, but it may serve as a primer for mtDNA replication under the strand‐displacement mtDNA replication model (Nicholls & Minczuk, 2014). As not all mtDNAs carry 7S, it is possible that only 7S‐containing mtDNA are primed for replication. The D‐loop has also been proposed as the binding site for proteins that regulate the dynamics of mitochondrial nucleoids.

PolG mediates the replication, recombination, and repair of mtDNA and is essential for mtDNA maintenance. PolG consists of two subunits: PolGA, the catalytic subunit with exonuclease activities, and PolGB, the accessory subunit that stabilizes the enzyme to increase its efficiency. The hexameric DNA helicase TWINKLE has been shown to unwind DNA during replication and was recently demonstrated to be essential for 7S DNA synthesis in the D‐loop. MtSSB mediates the unwinding of mtDNA through its physical interaction with TWINKLE and co‐localizes with TFAM.

If and how these mtDNA replication factors and nucleoid proteins are involved in selective mtDNA replication is unknown. Meanwhile, the proposed hypothesis provides a context for future discussions and investigations that could ultimately help to better understand mtDNA transmission and replication in early development and thus increase the efficiency of MRT.

Conclusion

The efficacy and safety of ST for MRT has been established in monkeys and humans. Nonetheless, the reversal of maternal mtDNA post‐MRT has recently become apparent. As discussed above, up to 15% of MRT babies could experience reversal to the mother's mutant mtDNA. While applications of MRT have been legally regulated and approved in the UK, births of MRT babies in Mexico and in the Ukraine have been reported in the press. We would argue that such unregulated applications are premature, as MRT requires specific expertise and skills that are largely absent in most private IVF clinics. Moreover, MRT should still be carried out as an experimental treatment under strict oversight and with extensive follow‐up of born children. Families undergoing MRT must be consented and made aware of the experimental nature of this treatment and its potential adverse effects.

Possible solutions to address the problem of mtDNA reversal include developing methods to identify and visualize replication‐tagged mtDNA in oocytes/zygotes during MRT to select against or to avoid this population of mitochondria during nuclear transfer and enucleation. Another, more radical, approach is using gene editing tools that could selectively destroy carryover mtDNA. It has already been shown that mtDNA‐targeted nucleases, such as mitoTALENs, mitoREs, and mitoZFNs, can cleave specific sequences in heteroplasmic mtDNA to eliminate mutant molecules. However, low levels of mutant mtDNA may still remain and there is potential for unintended off‐target mutations. In general, it is imperative to further optimize MRT procedures so as to minimize or completely eliminate the carryover of maternal MRT.

Conflict of interest

The authors declare that they have no conflict of interest.