An alternative model for maternal mtDNA inheritance

The exact mechanism preventing paternal mitochondrial DNA (mtDNA) inheritance in humans remains unresolved. In their recent report, ‘Molecular basis for maternal inheritance of human mitochondrial DNA’, Lee et al. propose that this process is controlled by phosphorylation of the mitochondrial transcription factor A (TFAM) on two serine residues located in the matrix targeting sequence (MTS), which blocks this protein’s mitochondrial import, thus uncoating mtDNA and exposing it to attack and destruction by unspecified mitochondrial nuclease(s), which prevents paternal mtDNA from entering the oocyte during fertilization. We argue that the experimental evidence presented in the article is inconclusive and also supports an alternative possibility: that mtDNA turnover following the loss of expression of key mtDNA replication proteins might be responsible. Therefore, additional experimentation is needed to resolve the two competing mechanisms that align with the experimental data presented by Lee et al.

Human mtDNA replication is controlled by several factors, including mtDNA (POLG) and RNA (POLRMT) polymerases and TFAM. Inactivation of the corresponding genes in experimental animals results in embryonic lethality and is accompanied by the loss of mtDNA^1–3. Similarly, cultured cells with CRISPR–Cas9-mediated ablation of these proteins lack mtDNA^4,5. Available evidence indicates that the activities of POLRMT and TFAM contribute primers for mtDNA replication^2,6,7, while POLG participates in extending these primers⁸, both processes being critical for mtDNA maintenance.

In addition to its role in primer synthesis for mtDNA replication, TFAM also serves an architectural role by compacting mtDNA into nucleoids⁹. It is thus sometimes invoked, although without supporting evidence, that TFAM physically shields mtDNA from damage. The empirical data, however, suggest that, to the contrary, mtDNA compacted by TFAM remains accessible to matrix proteins¹⁰. This phenomenon was leveraged as a tool for mtDNA elimination from cultured cells¹¹ and served as the basis for the emerging therapeutic mtDNA-editing technologies¹². In the absence of evidence supporting a protective role of TFAM, it appears unlikely that TFAM redistribution from mitochondria, by itself, would substantially impact mtDNA vulnerability to mitochondrial nucleases, which is one of the cornerstones of the model proposed by Lee et al.

We also believe the four major lines of evidence presented in the paper do not support a definitive conclusion about the lack of TFAM in the mitochondria:

1. Western blots (WBs) of the whole sperm or sperm lacking midpieces show that most TFAM in sperm retains its MTS. However, this technique neither excludes the possibility of unprocessed TFAM localizing to mitochondria nor rules out the presence of a small fraction of processed TFAM in mitochondria.

2. Mass-spectrometry assays also show that the majority of TFAM molecules retain their MTS. However, with a maximum depth of coverage of five peptides, it remains possible that this technique fails to detect as much as 16–20% of processed TFAM. Previous studies indicate that as little as 1–10% of residual TFAM could be sufficient for mtDNA maintenance¹³. Like WB, this technique does not exclude the localization of unprocessed TFAM to mitochondria.

3. Immunogold labeling of cryosections lacks sensitivity to conclusively demonstrate TFAM’s absence in mitochondria.

4. Confocal microscopy is consistent with the redistribution of TFAM between mitochondria, cytosol and nucleus. However, it should not be mistaken for proof of TFAM’s absence in sperm mitochondria. Phosphomimicking experiments presented in Extended Data Fig.8 clearly show the presence of TFAM^31DD/34DD in both mitochondria and the nucleus. In some images (for example, Fig. 4c, panel 6h), faint TFAM fluorescence is clearly visible in the area of the mitochondrion-containing midpiece region of the sperm, undermining the proposed causality between TFAM redistribution and mtDNA loss.

   Furthermore, Scarlet protein targeted to mitochondria with the MTS of TFAM or POLRMT (Extended Data Fig.7b,d) shows diffuse cytoplasmic fluorescence in HeLa cells, which is inconsistent with its mitochondrial localization. Even more strikingly, a TFAM–Scarlet fusion containing the MTS of POLRMT instead of the native TFAM MTS localizes to the nucleus in HeLa cells (Extended Data Fig. 7c), in stark contrast to the mitochondrial localization of the same construct in spermatozoa (Fig. 5b). These contradictions illustrate the limitations of the confocal microscopy.

Therefore, we believe that the enumerated limitations of the experimental approaches used by Lee et al.¹⁴ do not justify the definitive conclusions that no TFAM is present in sperm mitochondria or that the absence of mtDNA in sperm mitochondria is solely due to the lack of TFAM.

Intriguingly, the experimental data presented in the paper are fully consistent with an alternative model that does not depend on TFAM redistribution. Indeed, the authors report the lack of POLRMT and POLG expression in sperm. Inactivation of these proteins in mice or cultured cells invariably results in the loss of mtDNA^1,2,4,5. Furthermore, TFAM MTS processing is compromised in cells lacking mtDNA, as evidenced by the appearance of the corresponding band on WBs^2,4. Therefore, it is plausible that TFAM redistribution from mitochondria could be a consequence of mtDNA loss rather than its cause. Similarly, TFAM MTS phosphorylation could result from extramitochondrial TFAM mislocalization rather than being its cause.

While this alternative model does not necessarily exclude the involvement of nucleases, it does not require the loss of mtDNA protection due to TFAM redistribution. Indeed, mtDNA turnover in postmitotic tissues has half-lives as short as 6.7 d¹⁵ (or perhaps even shorter in maturing sperm). Assuming a 6.7-d half-life for mtDNA in maturing sperm, the 74-d sperm maturation period would represent more than 11 half-lives, which is sufficient to reduce mtDNA content some 2,048-fold (or more, accounting for a few cell divisions during the process). This would be enough to clear all mtDNA from the sperm, initially present in several hundred copies.

Lee et al. criticize this alternative model by noting that the absence of active mtDNA transcription or replication in oocytes (which express both POLG and POLRMT) does not lead to mtDNA loss. To clarify, the alternative model does not attribute mtDNA loss to the absence of detectable mtDNA transcription or replication; it relies solely on the lack of expression of POLG, POLRMT and/or other critical mtDNA replication proteins not expressed in sperm. Therefore, this criticism does not address the model.

Notably, Lee et al.¹⁴ do not consider why sperm cells phosphorylate the MTS of TFAM, a phenomenon not observed in other tissues. Our alternative model suggests that this phosphorylation could result from TFAM mislocalization and accumulation in the cytosol and nucleus following mtDNA loss. This would explain why phosphorylation of the MTS does not occur in most other cells containing mtDNA.

In summary, we suggest that the evidence presented by Lee et al.¹⁴ does not prove the absence of TFAM in sperm mitochondria or establish a causal link between TFAM redistribution from mitochondria and mtDNA loss in sperm. Neither TFAM redistribution from mitochondria nor cell division appears necessary for mtDNA turnover in other cell types. Therefore, we conclude that further experimentation is needed to differentiate between two plausible models, both consistent with the results presented by Lee et al.¹⁴ and each with its limitations.

Data availability

All data are contained within the paper. This study did not use any custom code or software.