Independent regulation of mitochondrial DNA quantity and quality in Caenorhabditis elegans primordial germ cells

  1. Aaron ZA Schwartz
  2. Nikita Tsyba
  3. Yusuff Abdu
  4. Maulik R Patel
  5. Jeremy Nance

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    Mitochondria have their own DNA, which is much more likely to gain mutation (due to error-prone DNA polymerase). It is widely appreciated that there are quality control mechanisms such that functional mitochondria are passed from one generation to the next. The proposed mechanisms include a passive mechanism (generation of the bottleneck) as well as an active mechanism (selective removal of non-functional mitochondria), but the processes are not fully understood. This manuscript presents fascinating observations as to how C. elegans germline may remove mitochondria by creating bottlenecks as well as selectively removing non-functional mitochondria. Building upon the authors' previous finding that the C. elegans primordial germ cells (PGCs) shed much of cytoplasm during embryogenesis through 'cannibalism', they now describe that a bulk of mitochondria are removed from PGCs through this process.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Mitochondria harbor an independent genome, called mitochondrial DNA (mtDNA), which contains essential metabolic genes. Although mtDNA mutations occur at high frequency, they are inherited infrequently, indicating that germline mechanisms limit their accumulation. To determine how germline mtDNA is regulated, we examined the control of mtDNA quantity and quality in C. elegans primordial germ cells (PGCs). We show that PGCs combine strategies to generate a low point in mtDNA number by segregating mitochondria into lobe-like protrusions that are cannibalized by adjacent cells, and by concurrently eliminating mitochondria through autophagy, reducing overall mtDNA content twofold. As PGCs exit quiescence and divide, mtDNAs replicate to maintain a set point of ~200 mtDNAs per germline stem cell. Whereas cannibalism and autophagy eliminate mtDNAs stochastically, we show that the kinase PTEN-induced kinase 1 (PINK1), operating independently of Parkin and autophagy, preferentially reduces the fraction of mutant mtDNAs. Thus, PGCs employ parallel mechanisms to control both the quantity and quality of the founding population of germline mtDNAs.

Article activity feed

  1. Version published to 10.7554/elife.80396 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    1. The relationship between lobe cannibalism and mtDNA reduction seems to be too mild. The authors first show that about half of mitochondria are removed in PGCs between the embryo and L1 stages. At this point, the number of mitoDNA/cell decreases by half compared to the embryonic stage, and based on this result, they propose that this is a bottleneck. To me (intuitively) 50% reduction does not seem strong as a bottleneck. Perhaps it is better to tone down the claim a bit here (unless they can provide stronger evidence, such as modeling, that a 50% reduction is sufficient to cause a bottleneck. Textual editing would suffice, though (unless they already have the evidence for bottleneck).

    We used the term ‘bottleneck’ to indicate the point when mtDNA numbers are at their lowest point (prior to mtDNA expansion) that we could detect in the early germ line lineage, which results from a combination of reductive cell divisions to get to the PGC lineage, followed by lobe cannibalism and autophagy for a final two-fold reduction. The ~150-fold total reduction we detect in PGC mtDNA number relative to the estimated number of mtDNAs present in the oocyte (inferred from analyzing whole early embryos) is comparable to the ~100-fold reduction in mtDNAs that occurs between the oocyte and early PGCs in mouse, which has been proposed to be a germline mtDNA genetic bottleneck based on computer simulation studies (Cree et al., 2008, NCB 40: 249-254). In addition, the number of mtDNAs we detect per PGC (~200) at its low point in L1 larvae is comparable to the number of mtDNAs in mouse PGCs purified using two different reporter transgenes (280 mtDNAs per PGC, Wai et al., 2008, NCB 40: 1484-1488; 203 mtDNAs per PGC, Cree et al., 2008, NCB 40: 249-254). However, we agree with Reviewer 1 that our data does not show whether the 150-fold total reduction in mtDNA number we detect in PGCs relative to the oocyte has a functional consequence on segregation of mtDNA mutations in the germ line. To clarify what we have shown and what we propose based on findings and simulations in other systems, we refer to the number of mtDNAs in L1 PGCs as a ‘low point’ and introduce in the discussion how this reduction could affect segregation and inheritance of mtDNA mutations (pg. 14, lines 333-339).

    1. Overall, one thing that struck me was that, when they assay 'selection' by mtDNA (e.g. the number of mtDNA, frequency of mutant mtDNA, reduction by autophagy pathway, reduction by pink1, etc), the effect seems to be way too mild. However, in Fig1c, d, and Fig2c, the amount of mitoGFP that goes to the lobe seems to be at least 80-90%. Is this because the 'striking' images were selected for presentation? Alternatively, I wonder if mito with more mtDNA actually end up surviving, and mito with fewer mitoDNA goes to the lobe (as a result, the amount of mito removed to the lobe is much higher than the amount of mtDNA removed). If so, is this actually THE selection that happens during embryo-to-L1 transition? Is there any way to measure the amount of mito and amount of mtDNA simultaneously?

    Thank you for bringing up this point. Reviewer 1 is correct that 80% of mitochondria are in lobes initially, so the images in Figures 1 and 2 are representative. However, prior to lobe scission, some mitochondria move back into the cell body, such that only 60% are in lobes at the two-fold stage (we have not performed this analysis even later, just as cannibalism begins between the two and three-fold stage, because of embryo rotation in the eggshell). This was shown and quantified in our original Fig. S1. To stress this point, the quantification of this data has been moved to Figure 1, while representative images remain in the supplement (Figure 1—figure supplement 1) (See Figure 1F). The movement of mitochondria back into the cell body is an avenue that we plan to explore in future studies, although we feel that it is beyond the scope of the current manuscript.

    Although we have not quantified mtDNA distribution due to the challenges of imaging late embryos, we have no evidence that there is a significant asymmetry in mtDNA density (mtDNAs per unit of mitochondrial mass) between lobe and cell body mitochondria; mtDNAs are distributed among mitochondria in both lobes and the cell body (see Figure 1I). Our experiments on uaDf5 mutant mtDNA also show that even if a small asymmetry in uaDf5 were present, it is not responsible for selecting against uaDf5 mtDNAs since uaDf5 mtDNA heteroplasmy in PGCs still decreases between embryo and L1 in nop-1 mutants, though we cannot exclude the possibility that a small asymmetry exists.

    Reviewer #2 (Public Review):

    Major points:

    1. I wish that the authors provided more direct evidence to support their conclusion that there is no mtDNA replication in embryonic PGCs and mtDNA only starts to replicate before the first division of PGCs in early L1.

    See essential revisions.

    1. It will also be interesting to show how compromising cannibalism (e.g. using nop1 mutant) affects the replication of mtDNA after the first division of PGCs in L1.

    See essential revisions.

    3)Finally, given that the total mtDNA copy number in later GSCs is similar between worms with and without the PGC cannibalism (wt vs nop-1 mutant) (Fig 3), and cannibalism does not selectively eliminate detrimental mtDNA mutation, I also wonder why PGCs need a bottleneck for the mtDNA population.

    We also do not fully understand why PGC lobe cannibalism is necessary. However, PGCs are born with a relatively high number of mtDNAs, as they arise from a small number of invariant cell divisions during embryogenesis (5) relative to somatic cells (on average ~8); lobe cannibalism could be a way to eliminate this excess to reach ~200 before PGCs differentiate into GSCs in larvae. Our experiments on nop-1 mutants clearly show that this number is important, as it is achieved through an independent mechanism even when lobe cannibalism is blocked. We have dedicated an entire paragraph to discussing these important points (pg. 13-14, lines 326-339).

  3. eLife

    Evaluation Summary:

    Mitochondria have their own DNA, which is much more likely to gain mutation (due to error-prone DNA polymerase). It is widely appreciated that there are quality control mechanisms such that functional mitochondria are passed from one generation to the next. The proposed mechanisms include a passive mechanism (generation of the bottleneck) as well as an active mechanism (selective removal of non-functional mitochondria), but the processes are not fully understood. This manuscript presents fascinating observations as to how C. elegans germline may remove mitochondria by creating bottlenecks as well as selectively removing non-functional mitochondria. Building upon the authors' previous finding that the C. elegans primordial germ cells (PGCs) shed much of cytoplasm during embryogenesis through 'cannibalism', they now describe that a bulk of mitochondria are removed from PGCs through this process.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    This is exciting work, with elegant experimental designs that are rigorously executed. The work will appeal to a broad readership. Specific comments are listed below.

    The relationship between lobe cannibalism and mtDNA reduction seems to be too mild. The authors first show that about half of mitochondria are removed in PGCs between the embryo and L1 stages. At this point, the number of mitoDNA/cell decreases by half compared to the embryonic stage, and based on this result, they propose that this is a bottleneck. To me (intuitively) 50% reduction does not seem strong as a bottleneck. Perhaps it is better to tone down the claim a bit here (unless they can provide stronger evidence, such as modeling, that a 50% reduction is sufficient to cause a bottleneck. Textual editing would suffice, though (unless they already have the evidence for bottleneck).

    Overall, one thing that struck me was that, when they assay 'selection' by mtDNA (e.g. the number of mtDNA, frequency of mutant mtDNA, reduction by autophagy pathway, reduction by pink1, etc), the effect seems to be way too mild. However, in Fig1c, d, and Fig2c, the amount of mitoGFP that goes to the lobe seems to be at least 80-90%. Is this because the 'striking' images were selected for presentation? Alternatively, I wonder if mito with more mtDNA actually end up surviving, and mito with fewer mitoDNA goes to the lobe (as a result, the amount of mito removed to the lobe is much higher than the amount of mtDNA removed). If so, is this actually THE selection that happens during embryo-to-L1 transition? Is there any way to measure the amount of mito and amount of mtDNA simultaneously?

  5. eLife

    Reviewer #2 (Public Review):

    In this manuscript, Schwartz et al. examined the role of their previous finding regarding a programmed bulk cytoplasmic remodeling of the embryonic PGCs in regulating the mtDNA copy number and heteroplasmy level during the PGC to GSC transition in C. elegans. The authors first showed that the mitochondria of embryonic PGCs accumulated in a lobe, most of which were cannibalized/degraded. This was evidenced by the acidification of mitochondria in the evicted cytoplasm. The authors then measured the mtDNA copy number change between the embryonic and larval PGCs by counting TFAM-GFP puncta and FACS and ddPCR, showing that the cytoplasmic remodeling halves the mtDNA copy number per PGC. Finally, the authors showed that compromising the cytoplasmic remodeling does not affect the heteroplasmic level of uaDf5, and suggested pink-dependent mechanisms to be responsible for mtDNA heteroplasmy regulation during the PGC-GSC transition.

    Overall, this work provides a detailed characterization of mitochondrial network and mtDNA copy number changes during the cytoplasmic remodeling process. It also showed that the remodeling reduces mtDNA copies in PGC stochastically. However, I wish that the authors provided more direct evidence to support their conclusion that there is no mtDNA replication in embryonic PGCs and mtDNA only starts to replicate before the first division of PGCs in early L1. It will also be interesting to show how compromising cannibalism (e.g. using nop1 mutant) affects the replication of mtDNA after the first division of PGCs in L1. Finally, given that the total mtDNA copy number in later GSCs is similar between worms with and without the PGC cannibalism (wt vs nop-1 mutant) (Fig 3), and cannibalism does not selectively eliminate detrimental mtDNA mutation, I also wonder why PGCs need a bottleneck for the mtDNA population.

  6. eLife

    Reviewer #3 (Public Review):

    Schwartz et al investigate how mitochondria genomes (mtDNA) are inherited in C. elegans. They identify two distinct processes in primordial germ cells (PGCs) that shape mtDNA inheritance -one that influences the quantity of mtDNA, and the other the quality.

    In the first part of the study, they show that during PGC development mtDNA is reduced two-fold. Building and extending on their previous work, they find that both cannibalization and autophagy contribute to this reduction. The authors show that this does not preferentially eliminate mutant mtDNA and they do not report any long-term, overt phenotypic consequences to inhibit this decrease in mtDNA. Indeed, they find that copy number returns to control levels when PGCs become germline stem cells, even though excess mtDNA is present at the onset of germline expansion. The authors do not explore to what degree this reduction in mtDNA contributes to the stochastic drift of mtDNA variants in the germ line.

    In the second part of the study, they identify a stage of PGC development during which mtDNA quality is assessed. They find that the proportion of mutant mtDNA in PGCs decreases by approximately 5 percent in L1 PGCs, but increases again at later stages of development, presumably due to the selfish nature of the mutant genome they are using. Interestingly, they find that this selective decrease in mutant mtDNA requires the mitochondrial stress kinase PINK1, independent of its canonical role in mitochondrial autophagy.

    Strengths: This study is well conducted with rigorous experimentation and thoughtful data interpretation. The authors make substantial gains in understanding mtDNA inheritance in the C. elegans germline. To my knowledge, they are the first to define a precise developmental stage during which mtDNA purifying selection occurs in the C. elegans germline and to implicate a non-canonical role for PINK1 in this process in C. elegans.

    Weaknesses: The main weakness of this study, in my opinion, is that the authors do not directly measure mtDNA replication in PGCs, which has important implications with respect to how they interpret their data. Based on ddPCR and counting TFAM positive nucleoids, they conclude that little mtDNA replication likely occurs until after PGCs differentiate into GSCs and selection has taken place. While this may be true, it is hard to be certain without directly measuring mtDNA replication.

  7. Version published to 10.1101/2022.05.06.490954v1 on bioRxiv