1. Cell Biology
Download icon

Elimination of paternal mitochondria in mouse embryos occurs through autophagic degradation dependent on PARKIN and MUL1

  1. Rebecca Rojansky
  2. Moon-Yong Cha
  3. David C Chan  Is a corresponding author
  1. California Institute of Technology, United States
Research Article
  • Cited 36
  • Views 2,795
  • Annotations
Cite this article as: eLife 2016;5:e17896 doi: 10.7554/eLife.17896

Abstract

A defining feature of mitochondria is their maternal mode of inheritance. However, little is understood about the cellular mechanism through which paternal mitochondria, delivered from sperm, are eliminated from early mammalian embryos. Autophagy has been implicated in nematodes, but whether this mechanism is conserved in mammals has been disputed. Here, we show that cultured mouse fibroblasts and pre-implantation embryos use a common pathway for elimination of mitochondria. Both situations utilize mitophagy, in which mitochondria are sequestered by autophagosomes and delivered to lysosomes for degradation. The E3 ubiquitin ligases PARKIN and MUL1 play redundant roles in elimination of paternal mitochondria. The process is associated with depolarization of paternal mitochondria and additionally requires the mitochondrial outer membrane protein FIS1, the autophagy adaptor P62, and PINK1 kinase. Our results indicate that strict maternal transmission of mitochondria relies on mitophagy and uncover a collaboration between MUL1 and PARKIN in this process.

https://doi.org/10.7554/eLife.17896.001

eLife digest

Mitochondria are commonly referred to as the 'powerhouses' of animal cells because these structures provide the majority of the energy in most cells. People inherit their mitochondria from their mother, and not their father. This is because the father's mitochondria, which are delivered by sperm to the egg, are degraded early on when the embryo starts to develop.

Previous studies with model organisms, like nematode worms, showed that mitochondria delivered via sperm (also known as 'paternal mitochondria') were delivered to structures called lysosomes and broken down by the enzymes contained within. However, it remained controversial whether this process, named mitophagy, also occurred in mammalian cells, and the molecules involved were unknown.

Now, Rojansky et al. have identified key molecules that are essential for the degradation of mitochondria in mouse cells and show that these same molecules are needed to degrade paternal mitochondria in early mouse embryos. These results indicate that paternal mitochondria are indeed degraded by mitophagy in mice. In addition, Rojansky et al. also note that one of the key molecules is a protein called PARKIN, which is mutated in many inherited cases of Parkinson's disease, a major neurodegenerative disorder.

Even though these new findings provide a clearer idea as to how paternal mitochondria are degraded, the question of why remains unanswered. As a result, it is likely that this topic will continue to be heavily debated. Nevertheless, having identified the key molecules involved in degrading paternal mitochondria, it may now be possible to address this question more directly – for example by interfering with this process and then examining the consequences.

https://doi.org/10.7554/eLife.17896.002

Introduction

In most animals, including mammals, mitochondria are inherited strictly through the maternal lineage. Because sperm deliver mitochondria into the egg during fertilization, mechanisms likely exist to eliminate paternal mitochondria from the early embryo. Uniparental inheritance of mitochondria ensures that only one haplotype of mitochondrial DNA (mtDNA) exists in the offspring, a phenomenon with considerable biomedical implications. It underlies the maternal inheritance of diseases caused by mutations in mtDNA (Carelli and Chan, 2014) and enables the use of mtDNA sequences to track human migrations during evolution. Mouse studies suggest that extensive heteroplasmy, the co-existence of more than one haplotype of mtDNA, is genetically unstable and associated with physiological abnormalities (Sharpley et al., 2012).

Although uniparental inheritance is a defining characteristic of mitochondria, there is much speculation about its mechanism in vertebrates (Carelli, 2015). Most of our knowledge has come from invertebrate model organisms. The phenomenon has been most decisively dissected in Caenorhabditis elegans, where paternal mitochondria are eliminated by mitophagy (Al Rawi et al., 2011; Sato and Sato, 2011; Zhou et al., 2011), a process in which mitochondria are engulfed by autophagosomes and delivered to lysosomes for destruction. In Drosophila melanogaster, paternal mitochondrial elimination involves autophagic components but occurs independently of PARKIN (Politi et al., 2014), a Parkinson’s disease-related E3 ubiquitin ligase that is central to the most heavily studied mitophagy pathway (Pickrell and Youle, 2015). However, it is unclear to what extent these insights from invertebrate model organisms extend to mammals. Consistent with a role for autophagy, sperm mitochondria from mice are ubiquitinated (Sutovsky et al., 1999) and, after fertilization, are immuno-positive for P62 and the ATG8 homologs LC3 and GABARAP (Al Rawi et al., 2011). However, a subsequent study in mouse disputed the role of autophagy in elimination of paternal mitochondria (Luo et al., 2013). The association of LC3 with paternal mitochondria was observed to be transient and occurred well before paternal mitochondrial elimination. In addition, it was found that paternal mitochondria were segregated unevenly to blastomeres during early embryonic cell division. Based on these results, the authors rejected the role of autophagy and advocated a passive dilution mechanism whereby murine paternal mitochondria are stochastically lost due to uneven segregation to the cells of the embryo (Luo et al., 2013).

This mechanistic uncertainty highlights the need to move beyond correlative studies relying on co-localization of autophagy markers with paternal mitochondria, and instead to perform functional studies that directly test the role of autophagy. In C. elegans, the functional role of autophagy was revealed by the persistence of paternal mitochondria in embryos depleted for core autophagy genes, such as the ATG8 homologs LGG-1 and LGG-2 (Al Rawi et al., 2011; Sato and Sato, 2011; Zhou et al., 2011). A similar approach is not feasible in mouse, however, because disruption of basal autophagy results in embryonic arrest at the four-cell stage (Tsukamoto et al., 2008), well before paternal mitochondria are normally eliminated.

To circumvent this technical hurdle, we reasoned that a functional test for the role of mitophagy might be possible by focusing on mitophagy-specific genes, whose depletion would be less likely to arrest early embryonic development compared to core autophagy genes. To obtain a set of candidate mitophagy genes, we first characterized the requirements for mitophagy in cultured cells. These experiments led to the realization that two E3 ubiquitin ligases, PARKIN and MUL1, synergistically function in degradation of mitochondria. We then used a gene disruption approach in early embryos to show that mitophagy mediates the degradation of paternal mitochondria.

Results

A functional assay for elimination of paternal mitochondria

To develop an assay to track paternal mitochondria in the early mouse embryo, we utilized male PhAM mice, in which all mitochondria, including those in the sperm midpiece, are labeled with a mitochondrially-targeted version of the photoconvertible Dendra2 fluorescent protein (Pham et al., 2012) (Figure 1A). When male PhAM mice were mated with wild-type females, the resulting embryos contained brightly fluorescent paternal mitochondria. At 12 hr post-fertilization (Figure 1B), the paternal mitochondria were found in a linear cluster, reflecting their original, compact organization in the sperm midpiece. At 36 hr after fertilization (Figure 1C), this cluster began to disperse in cultured embryos, and thereafter, well-separated individual mitochondria were visible within blastomeres. Over the next 2 days, paternal mitochondrial content progressively decreased (Figure 1D–F). At 84 hr after fertilization, the majority of embryos had lost all paternal mitochondria (Figure 1F). Quantification of these results showed a reproducible and progressive loss of paternal mitochondria between 60 and 84 hr post-fertilization (Figure 1G). To determine whether this pattern is specific to paternal mitochondria, we additionally mated PhAM female mice with wild-type males, resulting in embryos with fluorescent maternal mitochondria. In these embryos, there was no reduction in the maternal mitochondrial content between 60 and 84 hr post-fertilization (Figure 1H, Figure 1—figure supplement 1).

Figure 1 with 1 supplement see all
Paternal mitochondria are degraded by 84 hr after fertilization.

(A) Fluorescence of mito-Dendra2 in a live sperm cell isolated from the cauda epididymis of a PhAM mouse. (B–F) Mito-Dendra2 in a 12 hr (B), 36 hr (C), 60 hr (D), 72 hr (E), and 84 hr embryo (F). In (B), note that mito-Dendra2 is circumscribed to a distinct rod-like structure. The mitochondria disperse in later embryos and are lost by 84 hr. (G) Quantification of the mito-Dendra2 signal (see Materials and methods) at 36, 60, 72, and 84 hr after fertilization. Each data point represents the mean of 15 embryos. Error bars indicate SD. (H) Representative maximum intensity projection images of maternal mitochondrial content versus paternal mitochondrial content over time. Embryos with mito-Dendra2-labeled maternal mitochondria were derived from crosses of wildtype males with homozygous PhAM females. Embryos with labeled paternal mitochondria were derived from crosses of wild-type females with homozygous PhAM males, whose sperm donate Dendra2-labeled mitochondria to the embryo upon fertilization. Embryos were cultured in vitro and imaged at the indicated time. Note that paternal Dendra2 signal decreases with time, whereas maternal Dendra2 signal does not. (I) Schematic of paternal mitochondrial elimination assay. Wildtype females are mated with PhAM males. One-cell embryos are microinjected in the perivitelline space with concentrated lentivirus targeting candidate genes. During in vitro culture, embryos are periodically imaged live and monitored for their ability to eliminate paternal mitochondria. (J) Representative images of embryos injected with lentivirus carrying nontargeting shRNA. The left three images show mito-Dendra2, phase-contrast, and mCherry signals at 60 hr; the right three images show the same as 84 hr. (K) Embryos injected with lentivirus carrying Atg3 shRNA. (L) Embryos treated with bafilomycin A1. (M) Embryos injected with lentivirus carrying Parkin shRNA. All scale bars are 10 μm.

https://doi.org/10.7554/eLife.17896.003

We used a lentiviral approach to functionally probe the role of autophagy genes in this process (Figure 1I). We microinjected one-cell stage zygotes with lentivirus encoding mCherry and control shRNA or shRNA targeting the core autophagy gene Atg3. In embryos injected with lentivirus, the mCherry reporter was expressed within 48 hr of injection (60 hr post-fertilization). When nontargeting shRNA was expressed, development of the embryo was unaffected, and Dendra2-positive mitochondria were eliminated by 84 hr with the usual kinetics (Figure 1J). In embryos injected with shRNA against Atg3 (Figure 1K), however, embryo development was arrested at the four-cell stage, consistent with a previous report using Atg5-null oocytes (Tsukamoto et al., 2008). Similarly, treatment of embryos with bafilomycin, an autophagy inhibitor, arrested embryonic development (Figure 1L). In both cases, the treated embryos showed persistence of paternal mitochondria at 84 hr. However, due to the early disruption of embryonic development, it was not possible to conclude if autophagy has a specific role in elimination of paternal mitochondria. This result indicated that disruption of core autophagy genes in this system is not a viable experimental approach. We therefore decided to focus on mitophagy-specific genes. We injected embryos with lentivirus encoding shRNA against Parkin (Park2), an E3 ubiquitin ligase that is central to the most studied pathway for mitopahgy (Durcan and Fon, 2015; Pickrell and Youle, 2015). Such embryos show loss of paternal mitochondria by 84 hr after fertilization, suggesting that the process occurs in the absence of PARKIN (Figure 1M).

FIS1, TBC1D15, and P62 are essential for OXPHOS-induced mitophagy in MEFs

Given the negative results with PARKIN, we turned to cultured cells, where the role of specific proteins in mitophagy could be more readily analyzed. Our strategy was to identify, in cultured cells, a small set of mitophagy genes, which could then be re-analyzed in early embryos. To monitor mitophagy, we constructed a dual color fluorescence-quenching assay based on an EGFP-mCherry reporter localized to the mitochondrial matrix. Normal mitochondria are yellow, having both green and red fluorescence in the matrix, whereas mitochondria within acidic compartments show red-only fluorescence, due to the selective sensitivity of EGFP fluorescence to low pH. A similar approach using a mitochondrial outer membrane EGFP-mCherry reporter has been effective for monitoring mitophagy (Allen et al., 2013). When mouse embryonic fibroblasts (MEFs) were cultured with a moderate concentration (10 mM) of glucose, a condition in which their metabolism relies largely on glycolysis, they showed few red-only mitochondria (Figure 2A). We previously defined a glucose-free, acetoacetate-containing culture formulation that induces MEFs to substantially upregulate OXPHOS activity (Mishra et al., 2014). When cells were cultured for 4 days in this OXPHOS-inducing medium, many cells exhibited numerous red puncta (Figure 2A). This observation is consistent with a study showing that glucose-free conditions promote increased turnover of mitochondria (Melser et al., 2013) and likely reflects the higher turnover of mitochondria when the activity of the respiratory chain is elevated. Atg3 knockout MEFs did not form red puncta under the OXPHOS-inducing condition (Figure 2B–C), indicating that formation of red puncta is dependent on the core autophagy machinery. Consistent with this idea, the level of lipidated LC3, another core component of the autophagy pathway, was elevated (Figure 2D). Moreover, the red-only puncta co-localized extensively with mTurquoise2-LC3B, suggesting that they represent mitochondrial contents within the autophagosome pathway (Figure 2E, arrows). In addition, a subset of the red puncta co-localize with LAMP1, likely indicating later intermediates that have progressed to lysosomes (Figure 2F). In contrast, in glycolytic medium, mTurquoise2-LC3B did not co-localize with mitochondria (Figure 2E). In addition, we found that p62 (SQSTM1), a protein implicated in autophagy (Pankiv et al., 2007) and mitophagy (Seibenhener et al., 2013), localized to mitochondria only under the OXPHOS-inducing condition (Figure 2G). Unlike LC3B and LAMP1, however, P62 was localized to both red punctate mitochondria and elongated yellow mitochondria. These results indicate that the OXPHOS-inducing condition results in an increase in mitophagy intermediates.

Induction of mitophagy by OXPHOS-inducing medium.

Mitophagy was examined in cells stably expressing Cox8-EGFP-mCherry. Wild-type (A) or Atg3 knockout mouse embryonic fibroblasts (MEFs) (B) were grown in Glucose (Glu) or Acetoacetate (Ac) containing medium for 4 days and then imaged by fluorescence microscopy. The red puncta in the bottom panel of (A) represent mitochondrial contents within acidic compartments. (C) Quantification of red-only puncta. Error bars indicate SD of three biological replicates, **p<0.01, p=0.0039 (Atg3+/+ Glu vs. Ac), p=0.0052 (Atg3+/+ vs. Atg3 -/-) (Student’s t-test). (D) Western blot analysis of LC3B expression in MEFs cultured in the indicated medium. The lower band is lipidated LC3B. Actin is a loading control. (E) Co-localization of LC3B with red puncta. MEFs expressing cox8-EGFP-mCherry and mTurquoise2-LC3B were grown in the indicated medium and imaged by fluorescence microscopy. Arrows indicate examples of mTurquoise2-LC3B co-localization with red mitochondrial puncta. (F) Co-localization of LAMP1 with red puncta. MEFs stably expressing cox8-EGFP-mCherry were grown in acetoacetate-containing medium and immunostained with anti-Lamp1 antibody (blue). Arrows indicate red mitochondrial puncta that co-localized with LAMP1. Scale bar in (A) is 10 µm and applies to (A–F). (G) Co-localization of p62 with mitochondria. MEFs were grown in the indicated medium and immunostained with anti-p62 (green) and anti-HSP60 (red, mitochondrial marker). Error bars indicate SD. Scale bar, 10 μm.

https://doi.org/10.7554/eLife.17896.005

With this cellular system, we sought to identify genes required for induced mitophagy. Previous studies suggested that mitochondrial dynamics, particularly mitochondrial fission, is important for efficient mitophagy (Mao et al., 2013; Tanaka et al., 2010). To explore this idea, we examined the efficiency of OXPHOS-induced mitophagy in a panel of MEF cell lines deficient in mitochondrial fusion or fission genes: Mitofusin 1 (Mfn1), Mitofusin 2 (Mfn2), both Mfn1 and Mfn2 (Mfn-dm), Optic atrophy 1 (Opa1), Mitochondrial fission factor (Mff), Dynamin-related protein 1 (Drp1), and Mitochondrial fission 1 (Fis1) (Figure 3A). MEFs deficient in mitochondrial fusion were competent for mitophagy. In fact, Mfn-dm cells and Opa1-/- cells showed substantial mitophagy even under glycolytic culture conditions, consistent with the findings that mitochondrial fusion protects against mitophagy (Gomes et al., 2011; Rambold et al., 2011) and that Mfn-dm cells have constitutive localization of Parkin to mitochondria (Narendra et al., 2008). Among cell lines deficient in mitochondrial fission, Drp1-/- and Mff-/- cells showed normal levels of mitophagy under OXPHOS conditions (Figure 3A).

Figure 3 with 1 supplement see all
Mitophagy under OXPHOS-inducing conditions requires FIS1, TBC1D15, and p62.

(A) Mitophagy in cells with mutations in mitochondrial dynamics genes. MEFs of the indicated genotype were cultured in glucose or acetoacetate medium, and mitophagy was quantified using the Cox8-EGFP-mCherry marker. Neither Mfn-dm cells nor Opa1-/- cells were viable in acetoacetate-containing medium. Error bars indicate SD of three biological replicates, p=0.0078 (Student’s t-test. (B) MEFs stably expressing Cox8-EGFP-mCherry were grown in acetoacetate containing medium and then imaged by fluorescence microscopy. p62 and Tbc1d15 shRNAs were introduced by retroviral infection. (C) Co-localization of mTurquoise2-LC3B with mitochondria. MEFs were grown in acetoacetate containing medium. Note that mTurquoise2 puncta localize to mitochondrial puncta (arrows) only in WT cells. (D) Co-localization of P62 with mitochondria. MEFs were grown in acetoacetate containing medium and immunostained with anti-P62 (green) and anti-HSP60 (red). (E) Quantification of red-only puncta in WT cells and cells containing shRNA against Tbc1d15 or p62 cultured in glucose (Glu) or acetoacetate (Ac) medium. Error bars indicate SD of three biological replicates, p=0.0048 (Tbc1d15), p=0.0053 (p62) (Student’s t-test).

https://doi.org/10.7554/eLife.17896.006

In contrast, Fis1-/- cells had dramatically reduced mitophagy under OXPHOS conditions (Figure 3A–B), and a failure of both LC3 and P62 to co-localize with mitochondria (Figure 3C–D). Although FIS1 is a central player in yeast mitochondrial fission, it does not play a prominent role in mammalian mitochondrial fission (Losón et al., 2013; Otera et al., 2010). Instead, recent studies implicate FIS1 and its interacting protein TBC1D15 (Onoue et al., 2013) in mitochondrial degradation, specifically in PARKIN-dependent mitophagy (Shen et al., 2014; Yamano et al., 2014). Similar to Fis1 deletion, Tbc1d15 knockdown efficiently blocked mitophagy and decreased LC3 and p62 localization to mitochondria (Figure 3C–E). Expression of shRNA-resistant Tbc1d15 in these cells restored red puncta formation (Figure 4—figure supplement 1B,C). Because depletion of either FIS1 or TBC1D15 blocked mitophagy and abolished P62 localization to mitochondria, we tested whether P62 is required for mitophagy. Cells knocked down for p62, as well as p62 knockout cells, were deficient for OXPHOS-induced mitophagy and showed reduced mTurquoise2-LC3B localization to mitochondria (Figure 3C,E; Figure 3—figure supplement 1A–B). Expression of mTurquoise2-p62 restored red puncta formation in p62 knockout cells, and expression of shRNA-resistant p62 restored red puncta formation in p62 shRNA expressing cells, consistent with a role for P62 in OXPHOS-induced mitophagy (Figure 3—figure supplement 1C, Figure 4—figure supplement 1B). Taken together, these results place FIS1 and TBC1D15 upstream of P62 in promoting autophagic engulfment of mitochondria.

PARKIN and MUL1 coordinately regulate OXPHOS-induced mitophagy

Because PINK1 and PARKIN are central components of the most widely studied pathway for mitophagy (Pickrell and Youle, 2015), we tested the role of these molecules in our mitophagy assay. Pink1-/- cells showed a substantial reduction in OXPHOS-induced mitophagy (Figure 4A–B). However, Parkin knockout MEFs had normal mitophagy (Figure 4A–B), a surprising observation given that PINK1 is known to operate upstream of PARKIN (Clark et al., 2006; Park et al., 2006; Yang et al., 2006). This observation suggests that another molecule may compensate for the loss of PARKIN. Recently, the mitochondrial E3 ligase MUL1 (MULAN/MAPL), has been shown to act parallel to the PINK1/PARKIN pathway in ubiquitination and proteasomal degradation of mitofusin (Yun et al., 2014). We hypothesized that MUL1 might work in parallel with PARKIN in OXPHOS-induced mitophagy, such that its presence would maintain mitophagy in the absence of PARKIN. Indeed, knockdown of Mul1 by either of two independent shRNAs in the Parkin knockout cell abolished mitophagy (Figure 4A–B; Figure 4—figure supplement 1A–C). In contrast, knockdown of Mul1 alone did not inhibit mitophagy. Inhibition of mitophagy due to loss of PINK1 or PARKIN/MUL1 prevented co-localization of LC3 with mitochondria (Figure 4C). These results reveal that MUL1 and PARKIN have redundant functions in mitophagy. We found a similar redundancy of MUL1 and PARKIN function in mitophagy induced by depolarization of mitochondria with CCCP (Figure 4—figure supplement 1D)

Figure 4 with 1 supplement see all
MUL1 and PARKIN have redundant functions in OXPHOS-induced mitophagy.

(A) Quantification of red-only puncta in cells grown in acetoacetate-containing medium. Presence (+) or absence (-) of Pink1, Parkin, or Mul1 is indicated. Error bars indicate SD of three biological replicates, p=0.015 (Pink1), p=0.0011 (Parkin-/- Mulan shRNA) (Student’s t-test). (B) Mitophagy in wild-type and mutant cells. Cells stably expressing Cox8-EGFP-mCherry were grown in acetoacetate-containing medium and imaged by fluorescence microscopy. (C) Co-localization of LC3B with mitophagy intermediates. Wild-type and mutant cells were retrovirally transduced with mTurquoise2-LC3B, grown in acetoacetate-containing medium and imaged by fluorescence microscopy. Examples of LC3B co-localization with mitophagy intermediates are indicated by arrows. (D) Accumulation of polyubiquitinated proteins in mitochondria. Cells were grown in the indicated medium, and mitochondria were isolated by differential centrifugation. Mitochondrial lysates were analyzed by Western blot for pan-Ubiquitin. HSP60 is a loading control. (E) Quantification of polyubiquitinated proteins in mitochondria. Three independent experiments were quantified by densitometry and averages are shown. Ubiquitin level was normalized to HSP60. Error bars indicate SD, p=0.0003 (WT Glu vs. Ac), p=0.0011 (Pink1-/-), p=0.0016 (Parkin-/- Mulan shRNA), p=0.0206 (Parkin-/-) (Student’s t-test).

https://doi.org/10.7554/eLife.17896.008

Mitochondria from cells grown in OXPHOS media are ubiquitinated ~six-fold more than cells grown in glycolytic media (Figure 4D,E). Loss of MUL1 or PARKIN alone had modest or no effect on the induction of mitochondrial ubiquitination under OXPHOS conditions. However, loss of both MUL1 and PARKIN, or PINK1 alone, substantially reduced the ubiquitination of mitochondria (Figure 4D,E). Taken together, these data suggest that MUL1 and PARKIN act in concert to ubiquitinate mitochondrial substrates, and that a threshold level of ubiquitination may be required to trigger mitophagy under OXPHOS conditions. The level of mitochondrial ubiquitination is known to dynamically regulate mitophagy (Bingol et al., 2014; Cornelissen et al., 2014).

Mitophagy genes are required for elimination of paternal mitochondria in embryos

With these molecular insights from the cellular assay, we re-visited the embryonic system to test whether the same pathway is involved in elimination of paternal mitochondria. We found that embryos expressing shRNA against p62, Tbc1d15, or Pink1 showed strong suppression of paternal mitochondrial loss, compared to embryos expressing a non-targeting shRNA (Figure 5A). When these mitophagy genes were knocked down, the majority of embryos retained substantial paternal mitochondria at 84 hr post-fertilization (Figure 5B). In contrast, less than 20% of embryos containing non-targeting shRNA retained significant paternal mitochondria, with the majority of embryos showing complete loss of paternal mitochondria. Depletion of either Parkin or Mul1 alone modestly reduced paternal mitochondrial elimination, but depletion of both had a severe and highly significant effect. Over 60% of Parkin/Mul1-depleted embryos showed retention of paternal mitochondria at 84 hr (Figure 5A–B, Figure 5—source data 1).

Figure 5 with 1 supplement see all
Clearance of paternal mitochondria in preimplantation embryos requires mitophagy genes.

(A) Impaired elimination of paternal mitochondria upon inhibition of mitophagy genes. Embryos were injected with lentivirus expressing shRNA against the indicated genes. The mitochondrial Dendra2 signal is shown for live embryos at 60, 72, and 84 hr after fertilization. Images are maximum intensity projections. Scale bar, 10 μm. (B) Quantification of paternal mitochondrial elimination at 84 hr post-fertilization. Maximum intensity z-projection images were analyzed encompassing the full embryo with z-slices overlapping. Embryos were scored as having no paternal mitochondria (black bar), less than five mitochondrial objects (white bar), or five or more mitochondrial objects (grey bar). Averages of at least three independent injection experiments are shown with 32–200 embryos quantified. Error bars indicate SD, *p<0.05; **p<0.01; ***p<0.001 (Chi-squared test). p-Values compare experimental embryos to control embryos with non-targeting shRNA. Chi-squared values: 75.386 (Tbc1d15), 155.784 (p62), 58.064 (Parkin shRNA, Mulan shRNA), 1.484 (Mulan shRNA), 8.074 (Parkin shRNA). (C) Clearance of paternal mitochondria in embryos expressing mCherry (control) or Fis1-DN. Same scale as (B). (D) Quantification of 84 hr results from (D). Error bars indicate SD. ***p<0.001 (Chi-square test). p-Values compare experimental embryos to mCherry control embryos. Chi-squared value: 125.584.

https://doi.org/10.7554/eLife.17896.010

Although FIS1 is a key molecule in the OXPHOS-induced mitophagy pathway, the relevant FIS1 molecules are likely to be contributed by the sperm and not the egg. Our shRNA approach can only knockdown proteins synthesized within the embryo. To circumvent this issue, we developed a dominant negative version of FIS1 (FIS1-DN) that lacks the C-terminal transmembrane domain that is essential for mitochondrial outer membrane localization. Retroviral overexpression of the cytosolic FIS1-DN protein in MEFs strongly inhibits OXPHOS-induced mitophagy (Figure 5—figure supplement 1A–C). When FIS1-DN was expressed in embryos, we found that loss of paternal mitochondria was strongly inhibited, with the majority retaining substantial paternal mitochondria (Figure 5C–D, Figure 5—source data 1).

The signal for selective degradation of paternal mitochondria in mammals is unknown, but some other forms of mitophagy are triggered by loss of mitochondrial membrane potential. Using the cationic dye TMRE (tetramethylrhodamine ethyl ester), we found robust staining of sperm isolated from the caudal epididymis of PhAM male mice, indicating intact mitochondrial membrane potential (Figure 6A). At 18 hr after fertilization, paternal mitochondria remained in a linear cluster in the embryo and stained robustly with TMRE. However, over the next 36 hr, paternal mitochondria gradually lost TMRE staining, such that at 48 hr and later, nearly all paternal mitochondria failed to stain with TMRE (Figure 6B–C). In the same experiment, maternal mitochondria always maintained TMRE staining, indicating that there is selective loss of membrane potential in paternal mitochondria

Figure 6 with 1 supplement see all
Loss of membrane potential in paternal mitochondria after fertilization.

(A) Mitochondrial membrane potential in live sperm cell. Spermatozoa were isolated from the cauda epididymis of a PhAM mouse, stained with 20 nM TMRE, washed, and imaged by fluorescent microscopy. Red signal is TMRE; green signal is mito-Dendra2. The boxed region is enlarged below. Scale bar, 10 μm. (B) Membrane potential of paternal mitochondria in early embryos. Embryos, generated by mating wildtype females with PhAM males, were collected at 12 hr after fertilization and cultured in vitro. At 18, 48, or 72 hr after fertilization, the embryos were incubated in 20 nM TMRE, washed, and imaged by fluorescent microscopy. Dashed box indicates region enlarged below. Arrows indicate examples of mito-Dendra2-positive spots lacking TMRE signal. Scale bar, 10 μm. (C) Fluorescence line analysis of the boxed regions in (A) and (B). Each plot measures the TMRE and mito-Dendra2 signals along a one-pixel width line through the center of the boxed region. Note that the mito-Dendra2 and TMRE signals are co-incident at 18 hr after fertilization but not at 48 or 72 hr.

https://doi.org/10.7554/eLife.17896.013

We also examined whether paternal mitochondria fused with maternal mitochondria. To assess mitochondrial fusion, we utilized photo-conversion of Dendra2. We generated embryos in which either maternal mitochondria or paternal mitochondria were labeled with Dendra2. At 36 hr after fertilization, we photo-converted a subset of mitochondria and tracked their fate by confocal microscopy. At 60 hr post-fertilization, the photo-converted signal in maternally labeled embryos had spread widely into other mitochondria, resulting in a dramatic reduction in mean pixel intensity (Figure 6—figure supplement 1A–B). In contrast, the photo-converted signal from paternal mitochondria did not diffuse and clearly did not undergo fusion with other mitochondria in the embryo. This segregation of paternal mitochondria is likely to be important for their eventual degradation.

Discussion

Our results provide two major insights about mitophagy in mammals. First, we find in two biological systems—OXPHOS-induced mitophagy in cultured cells and paternal mitochondrial elimination in pre-implantation embryos—that PARKIN and MUL1 work synergistically to promote degradation of mitochondria by autophagy. Previous work showed that PARKIN and MUL1 have partially redundant roles in controlling the ubiquitin-dependent degradation of mitofusin (Yun et al., 2014). Our results show that this collaboration extends to the process of mitophagy. In MEFs, we find that both PARKIN and MUL1 regulate the levels of ubiquitin on mitochondria in response to OXPHOS conditions, and removal of both is necessary to cause a substantial reduction of ubiquitination. In a mitophagy assay where PARKIN is overexpressed, polyubiquitination of mitochondrial outer membrane proteins leads to their proteasomal degradation, which in turn is required for turnover of mitochondria by autophagy (Chan et al., 2011).

The redundant function of MUL1 likely explains why PARKIN knockout mice show surprisingly mild and inconsistent mitochondrial phenotypes (Palacino et al., 2004; Perez and Palmiter, 2005). Similarly, the redundant role of MUL1 may also explain why, in Drosophila, PARKIN is dispensable for paternal mitochondrial elimination (Politi et al., 2014). In future work, this insight may help to uncover the in vivo functions of PARKIN.

Second, we show that mitophagy is likely to be the mechanism underlying the elimination of paternal mitochondria in the early mouse embryo. It is unclear whether MEFs cultured under OXPHOS conditions bear any physiological relation to the early embryo. Nevertheless, we find that the genetic requirements for removal of paternal mitochondria in the embryo mirror those of MEFs undergoing mitophagy in response to OXPHOS induction. Although previous studies had shown that paternal mitochondria in mouse embryos co-localized with autophagy markers (Al Rawi et al., 2011), the functional relevance of these localization studies has been challenged and a passive mechanism for loss of paternal mitochondria has been proposed (Carelli, 2015; Luo et al., 2013). By identifying several molecules necessary for paternal mitochondrial elimination, our studies provide functional evidence for the role of mitophagy in this process.

Because we find that paternal mitochondria lose membrane potential shortly after entering the oocyte, it is tempting to speculate that this membrane depolarization may be the trigger for mitochondrial degradation. Previous studies indicate that PARKIN is recruited to mitochondria upon membrane depolarization (Narendra et al., 2008, 2010), and our results also suggest that PARKIN and MUL1 work together to degrade mitochondria that are depolarized (Figure 4—figure supplement 1D). However, we do not have direct evidence that membrane depolarization has a functional role in paternal mitochondrial degradation.

Although uniparental inheritance of mitochondria is nearly universal in animals, its physiological function is mysterious and difficult to address. One recent idea is that uniparental inheritance of mitochondria ensures that offspring contain only one haplotype of mtDNA. When mice with approximately equal proportions of two wild-type haplotypes of mtDNA were generated, they were found to have behavioral and cognitive abnormalities compared to homoplasmic counterparts (Sharpley et al., 2012). However, it is unclear to what extent this experimental result is relevant for a case in which paternal mitochondria were not eliminated. Sperm contain many fewer mitochondria (at least a thousand fold) compared to the oocyte, and therefore, the ensuing heteroplasmy levels would be very low. The identification of molecules essential for paternal mitochondrial elimination may facilitate further examination of this issue.

Materials and methods

Antibodies

The following commercially available antibodies were used: anti-Actin (Mab1501R, Millipore), anti-HSP60 (SC-1054, Santa Cruz Biotech), anti-LAMP1 (1D4B, Developmental Studies Hybridoma Bank), anti-P62 (PM045, MBL), anti-LC3B (2775 s, Cell Signaling), anti-c-Myc (C3956, Sigma), anti-Ubiquitin (P4D1, Cell Signaling), anti-PINK1 (75488, Abcam), anti-TBC1D15 (121396, Abcam), anti-PARKIN (15954, Abcam), and anti-MUL1 (HPA017681, Sigma).

For Western analysis, densitometry was done using ImageJ. The intensity of the ubiquitin signal was normalized to that of HSP60, and the average of three separate experiments was taken.

Immunostaining

For immunofluorescence experiments, cells were fixed with 10% formalin, permeabilized with 0.1% Triton X-100 and stained with the primary antibodies listed above and with the following secondary antibodies: goat anti-mouse Alexa Fluor 633, donkey anti-goat Alexa Fluor 546, goat anti-rabbit Alexa Fluor 488, goat anti-rabbit Alexa Fluor 633 (Invitrogen, Carlsbad, CA). When used, DAPI (d1306, Invitrogen) was included in the last wash.

shRNA virus design and production

For experiments in MEFs, the retroviral vector pRetroX-H1, which contains the H1 promoter, was used to express shRNAs. shRNAs were cloned into the BglII/EcoRI sites. For embryo injection experiments, a third-generation lentiviral backbone was used to express shRNAs. The lentiviral vector FUGW-H1 (Fasano et al., 2007) was modified by replacing the GFP reporter gene with mCherry and changing the shRNA cloning sites from Xba/SmaI to BamHI/EcoRI, generating FUChW-H1. For dual knockdown experiments in embryos, a second H1 promoter was added, along with XbaI/NheI cloning sites 3’ to the original H1 promoter, generating FUChW-H1H1.

The shRNA target sequences were:

p62: TGGCCACTCTTTAGTGTTTGTGT

Tbc1d15: GTGAGCGGGAAGATTATAT

Mul1 sh1: GAGCTAAGAAGATTCATCT

Mul1 sh2: GAGCTGTGCGGTCTGTTAA

Pink1: GGCTGACAGGCTGAGAGAGAA

Parkin: CCTCCAAGGAAACCATCAA

Non-targeting: GACTAGAAGGCACAGAGGG

Lentiviral vectors were cotransfected into 293T cells with plasmids pMDLG/pRRE, pIVS-VSVG, and pRSV-Rev. Retroviral vectors were cotransfected into 293T cells with plasmids pVSVG and pUMVC. All transfections were done using calcium phosphate precipitation. For microinjection, virus was collected, filtered, concentrated by ultracentrifugation at 25,000 rpm for 2 hr, resuspended in PBS and stored at −80°C as described previously (Lois et al., 2002; Pease and Lois, 2006). Viral titers were measured by infecting MEFs with serial dilutions of viral preparations, followed by flow cytometric analysis after 48 hr. Virus was used at 1×107 transducing units/μL.

Embryo microinjection

All mouse work was done according to protocols approved by the Institutional Animal Care and Use Committee at the California Institute of Technology. For each experiment, four C57/Bl6J wild-type female mice at 21–25 days old were superovulated by hormone priming as described previously (Pease and Lois, 2006), and then each was caged with a PhAM male (Pham et al., 2012) (RRID:IMSR_JAX:018397). After euthanization of females by CO2 asphyxiation, the embryos were harvested and placed in M2 medium (MR-015-D, Millipore) at 12 hr after fertilization as described in (Pease and Lois, 2006). Approximately 60 to 100 embryos were collected per experiment. Embryos were divided into two equal groups and microinjected with 10 to 100 pl of viral stock into the perivitelline space as described in (Lois et al., 2002; Pease and Lois, 2006). Embryos were washed with KSOM+AA medium (MR-106-D, Millipore) and cultured in that medium covered by oil (M8410, Sigma) at 37°C and 5% CO2. For each construct, at least three separate microinjection sessions were performed. In preparation for imaging, embryos were transferred to 10 μl droplets of KSOM+AA medium on glass-bottom dishes (FD35-100, World Precision Instruments).

Imaging and quantification

All images were acquired with a Zeiss LSM 710 confocal microscope with a Plan-Apochromat 63X/1.4 oil objective. All live imaging was performed in an incubated microscope stage at 37°C and 5% CO2. The 488 nm and 561 nm laser lines were used to excite cox8-EGFP-mCherry and imaging was done in line mode to minimize movement of mitochondria between acquisition of each channel. The 405 nm laser line was used to excite mTurquoise2 and DAPI. Alexa 488, Alexa 546, and Alexa 633, conjugated dyes were excited by the 488 nm laser, 561 nm laser, and the 633 nm laser, respectively.

In experiments tracking paternal mitochondrial degradation, all viable embryos from each experiment were imaged. Only embryos that were fragmented, lysed, or developmentally delayed were not imaged. The top and bottom of the embryo was set as the top and bottom z slices for z-stack image acquisition. Optical slices were acquired at 1.1 μm thickness, and z stacks were oversampled at 0.467 μm to ensure that all mitochondria were captured. Maximum intensity projections were created with Zen 2009 software and used for quantification.

For quantification of paternal mitochondria, control and experimental embryo images were randomized and counted blind. The number of mitochondria within each embryo was counted manually. In cases where two or more mitochondria were clustered together and could not be definitely resolved as distinct objects with separable borders, the cluster was counted as one object. Each maximum intensity z-projection was categorized as having either no mitochondria, less than five mitochondrial objects, or five or more mitochondrial objects. Embryos from four females were pooled per experiment, and three or more independent replicate experiments were averaged.

For photo-conversion of Dendra2, a region of interest was illuminated with the 405 nm line (4% laser power) for 90 bleaching iterations. The 488 nm laser line (5% laser power) and the 561 nm laser line (6.5% laser power) were used to excited Dendra2 in the unconverted state and photo-converted state, respectively. The pinhole used was 159 microns. Bandpass filters were used for detection of unconverted and photo-converted Dendra2 from 494 to 547 nm and 566 to 735 nm, respectively. The mercury lamp was not used to avoid inadvertent photoconversion.

For quantification of photo-converted Dendra2, maximum intensity z-stacks encompassing the entire embryos were analyzed in Matlab. For total intensity measurement, positive pixels were defined as those having an intensity greater than 10 (a low threshold designed to remove background), and the sum of these pixel intensities was calculated. For mean intensity measurement, this sum was divided by the total number of positive pixels.

Images were cropped when appropriate, and image contrast and brightness were globally adjusted in Photoshop (Adobe). Replicates are as indicated in figure legends.

Isolation of spermatocytes

Sperm were isolated from 4-month-old PhAM male mice. Longitudinal cuts were made in the cauda epididymis, and the tissue was incubated in PBS at 37°C to enable motile, mature sperm to swim out.

Membrane potential measurements

TMRE fluorescence was used to monitor mitochondrial membrane potential in spermatocytes and embryos. Samples were loaded with 20 nM TMRE for 20 min at 37°C and then washed into PBS (spermatocytes) or KSOM+AA (embryos). Samples were imaged live. Line analysis was performed using ImageJ.

Isolation of mitochondria

Mitochondria were isolated by differential centrifugation. Cells were washed in PBS, collected by scraping in isolation buffer (220 mM mannitol, 70 mM sucrose, 80 mM KCl, 5 mM MgCl2, 1 mM EGTA, 10 mM K+HEPES, pH7.4, and HALT protease inhibitors), and lysed on ice. Lysates were cleared of cell debris and nuclei with four 600 g spins. A crude mitochondrial fraction was isolated with a 10,000 g spin for 10 min and washed three times in isolation buffer.

Retroviral expression constructs

The Cox8-EGFP-mCherry retroviral vector (kindly provided by Drs. Prashant Mishra and Anh Pham) consists of the Cox8 mitochondrial targeting sequence placed N-terminal to an EGFP-mCherry fusion. To clone mTurquoise2 fusion proteins, mTurquoise2 was amplified from pmTurquoise2-Mito (Addgene plasmid # 36208, Dorus Gadella, [Goedhart et al., 2012]). Human LC3B was amplified from pFCbW-EGFP-LC3. Mouse p62 was amplified from pMXs-puro GFP-p62 (Addgene plasmid # 38277, Noboru Mizushima, [Itakura and Mizushima, 2011]). mTurquoise2 fusion proteins were cloned into the retroviral vector, pBABEpuro. The FIS1 dominant negative construct was cloned into pBABEpuro and consists of amino acids 1–121 of mouse FIS1, with 9 Myc tags at the N-terminus. The corresponding control construct consists of mCherry cloned into the pBABEpuro vector. All plasmids were verified by DNA sequence analysis. Stable cell lines were generated by retroviral infection followed by selection with 2 μg/μl puromycin.

Cell culture

MEFs were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin and 100 U/mL streptomycin at 37°C and 5% CO2. Glucose and acetoacetate containing media were made as previously described (Mishra et al., 2014). For mitophagy experiments, cells were plated on Nunc Lab-Tek II Chambered Coverglass slides (155409, Thermo) in DMEM-based media. After cells had adhered, they were washed with PBS and glucose- or acetoacetate-containing medium was applied, after which cells were allowed to grow for 4 days and then imaged. Because cells grow more slowly in acetoacetate medium, a four-fold excess of cells was plated relative to glucose medium so that both samples were at the same density on the day of imaging.

Cell lines

The cells used included: Atg3-null MEFs (Sou et al., 2008) (kindly provided by Yu-Shin Sou and Masaaki Komatsu), p62-null MEFs (Ichimura et al., 2008) (kindly provided by Shun Kageyama and Masaaki Komatsu), Pink1-null, Parkin-null (both kindly provided by Clement Gautier and Jie Shen), and Drp1-null (Ishihara et al., 2009) (kindly provided by Katsuyoshi Mihara). Mfn1-null (ATCC Cat# CRL-2992, RRID:CVCL_L691), Mfn2-null (ATCC Cat# CRL-2993, RRID:CVCL_L693), Mfn-dm (ATCC Cat# CRL-2994, RRID:CVCL_L692), Opa1-null (ATCC Cat# CRL-2995, RRID:CVCL_L694), Mff-null, Fis1-null MEFs have been described previously (Chen et al., 2005; Losón et al., 2013). The identity of MEF cell lines was authenticated by PCR genotyping of the relevant gene. Cell lines were negative for mycoplasma by DAPI staining.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47

Decision letter

  1. Serge Przedborski
    Reviewing Editor; Columbia University Medical Center, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Elimination of paternal mitochondria in mouse embryos occurs through autophagic degradation dependent on PARKIN and MUL1" for consideration by eLife. Your article has been favorably evaluated by Sean Morrison as the Senior Editor and three reviewers, one of whom is a member of our Board of Reviewing Editors. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission. The reviewers had some serious concerns but believe your work is conceptually important and would be willing to consider a revised manuscript if you are able to address their concerns. The reviewer comments are shown below in full to provide context, but the essential points you would have to address are described here.

One main difficulty that comes through the comments of all three reviewers is the challenge to assess the specificity of the mitophagy process you describe for paternal mitochondria. Are maternal mitochondria in zygotes also subjected to a similar process or is it specific to paternal mitochondria? This question may be addressed by using the same mutant mouse line but this time using female mutants crossed with wild-type males and while the number of maternal mitochondria may be large, you may be able to limit your quantitation by photoconverting Dendra in discrete areas of the zygote or by quantifying a fixed number of fields.

The use of glucose-free, acetoacetate medium also raised a number of questions. Not much is known about the effects of this medium on the biology of the cell beyond its action on OXPHOS. Could this medium have other effects that confound the interpretation of the results in MEFs? Are these culture conditions relevant to the more physiological conditions in the zygote? If the zygote has a more glycolytic metabolism, and the machinery identified here required an enhanced OXPHOS metabolism, how accurate can the identified machinery be? Clearly, some core elements such as Parkin and Mulan have been identified and are sufficient to modulate mitophagy in zygotes but these questions regarding the appropriateness and relevance of the culture medium must be addressed.

This led us to raise the question of membrane potential. Was the machinery of mitophagy that you identified validated under normal membrane potential? In the last paragraph of the Results section you indicate that zygotes show a time-dependent loss of membrane potential, suggesting that this trigger is important to the autophagy of paternal mitochondria. From your data you do not demonstrate that the loss of membrane potential is instrumental here. It is possible that the loss of membrane potential is an epiphenomenon, unrelated to the autophagy of paternal mitochondria, since the machinery of mitophagy you report herein has not been identified in the context of loss of membrane potential. It would thus be helpful to at least validate your identified mitophagy machinery under an induced membrane potential defect. Formal assessment in the zygotes of the membrane potential in maternal mitochondria should also be provided.

Your cellular and subcellular quantification methods were also questioned, and it would be important to better define how many elements were counted, how they were selected, and in how many cells from the embryos. Along this line it may be helpful to more clearly stress in the results that you counted all of the cells in each embryo and all paternal mitochondria in each cell. If, in some instances, this was not the case, then you must clarify how cells and/or mitochondria were selected and whether corrections were made for different number of cells per embryos of different ages. In addition, it was not clear what the level of detection is of Dendra-containing mitochondria in the microscope. Is the precision of your method and probe sufficient to confirm that one counted object equals one organelle? Also not clear from the manuscript was how you deal with detection of two mitochondria that are the fragmented/fissioned daughters of a parental organelle. Lastly, is it possible to photoconvert a single mitochondrion and follow its fate (sounds technically daunting but would be truly informative)?

As you consider revising your manuscript, please also provide more detailed information about the statistical treatment of the data as requested by reviewer #1. Also, it would also be important to provide further detail and additional controls for the knockdown experiments to assure that the correct targets were indeed silenced and the effects do not reflect off-target effects, as discussed by reviewer #2.

Finally, we would like to provide our editorial position with respect to two more points. First, with respect to the novelty of the study, as questioned by reviewer #3 ("The MEF experiments are convincing, but do not illustrate novel mechanisms"), it is our position that the novelty of the study does not reside in the "novel mechanism" of mitophagy, but in the demonstration that mitophagy (even if it uses a known mechanism) is involved in the elimination of paternal mitochondria.

Second, with respect to some of the mechanistic gaps, primarily pertaining to how and when paternal mitochondrial are recognized (e.g. reviewer #3, points #1, 2, and 6), these are valid questions but are beyond the scope of this specific study. However, a valuable approach to at least provide support to the specificity of the mechanism for the paternal mitochondria would be to compare intra- vs. inter-species mitophagy. Indeed, the phenomenon of paternal mitochondrial incompatibility applies only to intra-specific matings in that in inter-specific matings the paternal mitochondria survive, at least in mice (Kaneda et al., PNAS 92:4542, 2015). Consequently, an inter-specific mating should essentially abrogate all the relevant phenomena found here in the intra-specific mating (e.g. the decay of Dendra signal), thereby strongly supporting your conclusions.

Reviewer #1:

In this new study, the authors report on the fundamental, unresolved question of how the paternal mitochondria, which come from the spermatozoid and which enter the oocyte upon fertilization, are eliminated, leaving the blastocyst cells populated with only maternal mitochondria. In mammals, the debate remains as to whether paternal mitochondria are eliminated by stochastic dilution (related to the random partition of mitochondria during cell division) or mitophagy, and while these two processes are not mutually exclusive, it has been quite difficult to demonstrate a role for mitophagy in the destruction of paternal mitochondria. To clarify this fundamental issue, the authors used a mutant mouse in which all mitochondria are labelled with a fluorescent protein. By using a mutant male mouse crossed with a female wild-type mouse, the authors were able to readily identify and monitor paternal mitochondria in mouse zygotes. The authors confirmed that the number of paternal mitochondria declined over time and that by ~84 hrs post-fertilization, zygotes were devoid of any paternal mitochondria. As anticipated, using cells deficient in core autophagy factors such as ATG3 did stop embryo development and thus was not an unusable strategy to determine whether mitophagy was involved. Thus, the authors thought to use MEF cells to acquire further molecular insights into mitophagy to ultimately be able to target mitophagy rather than global autophagy, which they believed would be better tolerated by the embryo. Accordingly, they took a variety of MEF lines which were either constitutively deficient or were subjected to silencing strategy to knockout or knockdown key factors of mitochondrial fission/fusion as well as mitophagy such as Parkin and PINK1. To induce mitophagy, the authors exposed the different MEF lines to a glucose-free acetoacetate containing culture medium, and under these experimental conditions they found that: fusion/fission factors may modulate mitophagy but were not key factors except for Fis1 which was found to be a regulator. They also found Parkin in a redundant manner with another E3 ligase, MUL1, which was also instrumental in mitophagy induced by this particular culture medium. Other known factors such as PINK1, P62, and Tbc1d15 were all also implicated. With this information in hand, the authors went back to the zygotes and by emulating each of the molecular situations that were attenuating mitophagy in MEF, they consistently found that the elimination of paternal mitochondria was reduced. The authors concluded that their study provides compelling evidence that in mammalian cells, the elimination of paternal mitochondria in zygote and maternal mitochondria in MEFs used the same machinery of mitophagy and involved the cooperation between Parkin and MUL1.

This is quite an elegant study that provides an important demonstration about the involvement of mitophagy in the elimination of paternal mitochondria. The data are, for the most part, convincing, as are the interpretation and discussion of the results. Yet, a few weaknesses have been identified that reduced the enthusiasm for this work.

1) The authors demonstrate that targeting key factors of mitophagy does alter the decline of paternal mitochondria in zygote. However, since it is safe to assume that over ~84 hr primordial cells continue to divide, and that none of the proposed strategies abrogates but rather attenuates the decline, one cannot exclude a contribution of stochastic dilution in this process. One is thus left with the question of, if both processes are involved, what is the relative contribution of each to the elimination of paternal mitochondria? Unless overlooked, could the authors assess the decline by blocking cell division so that only mitophagy could be monitored?

2) While the strategy to find factors of mitophagy is original, it is somewhat confusing as it is not clear what this glucose-free acetoacetate containing medium does to the cell. While the authors claim that this culture medium promotes OXPHOS, the link between this effect and mitophagy is unclear. If as mentioned by the authors, loss of membrane potential is known to induce mitophagy, why did they not use this strategy to make their data perfectly comparable to those in most of the literature on the topic? Was there a problem on their hand in profoundly altering membrane potential?

3) Along this line, it is surprising that while the authors elected to move away from the membrane potential, in the last paragraph they examine this question in the embryos. This small piece of data reads as a last minute add-on with no connection to the rest of the study. Furthermore, these data do not demonstrate causality as they are strictly correlative and as such misleading.

4) It is also confusing to see that while the authors had to force a more OXPHOS-based metabolism to detect a meaningful factor of mitophagy, zygotes are cells that rather rely on glycolysis. If this is correct, there seems to be a disconnect in the logic of the experimental design.

5) More of the data are presented in a manner that the quantitative dimension of the observation is often difficult to assess. This manuscript is written more as a semi-quantitative/qualitative study, making it quite challenging to properly judge the significance of the results: are the data representing an exception or a rule?

6) Along this line, statistics are often too cursory presented. The authors must indicate for each of the tests used the N or DF, the results of the test and the actual p values. Also unclear is the number of technical and biological replicates that were used for each experiment.

7) There is an overall lack of technical details which prevents the reader from fairly evaluating the data.

Reviewer #1 (Additional data files and statistical comments):

As indicated above, there is an overall lack of details about the statistical treatment of the data.

Reviewer #2:

Rojansky and Chan confirm that genes already implicated in mitophagy, including PINK, p62, Fis1 and TBC1D15, behave similarly in their ox/phos driven mitophagy assay. They also suggest that Parkin and Mulan (redundantly), PINK1 and p62 are required to eliminate paternal mitochondrial DNA in pre-implanted embyros up to 84 hours. While the pre-implantation and mitophagy assay are important tools to determine the factors necessary to eliminate paternal mitochondrial DNA, the conclusions could be better substantiated by taking the pre-implantation assay out longer if the point is to link the results to maternal inheritance. How PINK1 functions upstream of Mul1 warrants further exploration. In addition, rescue experiments are required for both cell culture and embryo assays. Major methodological problems need to be addressed to build confidence in the conclusions.

1) In Figures 1 and 5 the authors knock down various genes in the embryo using a single shRNA, but fail to show the efficiency of the knockdown. Assessment of knockdown is essential to make any conclusions regarding what genes are actually necessary for the elimination of paternal mitochondrial DNA.

2) Again, a key problem in Figure 3 and 4 is that it is not shown to what extent the shRNAs knock down the expression of p62, Mul1 or TBC1D15. Rescue experiments with Parkin and Mul1, for example, would shore up the conclusions and address off target Mul1 shRNA effects.

3) It is not clear how accurate/quantitative the bar graphs of cell counting are (what is the lower cut off point for red dot number to count as mitophagy?) because in the one western blot shown (Figure 4D) the ubiquitination in the Mul1 shRNA lane 10 is not less than wild type in lane 6 whereas Parkin loss shows a clear effect – inconsistent with the two right bars in the cell counting results in Figure 4A and a main conclusion of the manuscript. A more quantitative approach that would assess total red dot number, such as FACS, would likely improve confidence in the comparisons. Or other corollaries of mitophagy based on western blotting and not imaging could be examined, such as ubiquitination of Mfn1.

4) The most important conclusion is that Mul1 and Parkin both function downstream of PINK1. Parkin has been shown to be activated by the PINK1 substrate phospho-ubiquitin. Does phospho-ubiquitin activate Mul1?

5) PINK1 and p62 knockout mice are viable and fertile and whether they properly eliminate paternal mitochondrial upon fertilization could yield a rigorous corroboration of those conclusions.

Reviewer #3:

The manuscript by Rojansky and Chan investigates the role of mitophagy in the process of elimination of paternal mitochondria, upon oocyte fertilization and early embryogenesis. They show that Parkin and MUL1 synergize in triggering sperm mitochondria degradation, likely by ubiquitination of mitochondrial proteins. PINK1, Fis1, Tbc1d15, and p62 are also shown to be necessary, since silencing of any of these components of the mitophagy pathway causes impairment of sperm mitochondria degradation. The findings were somehow similar to those in MEF cells subjected to OXPHOS stress in acetoacetate medium. The main significance of the findings is in the reaffirmation of the role of mitophagy in maternal inheritance of mtDNA, which had been proposed and then negated in previous literature. The structure of the study is somehow unusual, because sperm and oocyte experiments are alternated with MEF experiments, which almost constitute two different stories. While the general mechanism of mitophagy may be conserved in oocytes and MEFs in OXPHOS medium, the specificity of the process for sperm mitochondria elimination is unexplained by this mechanism. On the other hand, some key information missing in this study could help clarifying the process. Overall, this is an interesting study, but mechanistic details are still lacking.

1) The rationale of why the ubiquitination process should only occur in the embryo and not in the sperm, before fertilization is essentially based on the fact that silencing in the embryo results in delayed elimination. However, it would be important to know if some components of the pathway are already tagging sperm mitochondria, even though the whole process is terminated after fertilization.

2) The levels of the key components of the mitochondrial ubiquitination pathway in sperm or in embryos before and after silencing are not shown. The former at least should be easy to investigate, in relation to the first point.

3) The MEF experiments are convincing, but do not illustrate novel mechanisms. It is unclear why they belong in this study on sperm mitochondria elimination.

4) The driver of the loss of function of sperm mitochondria hours after fertilization was not investigated. This is a fundamental question, under the premises that these mitochondria become targeted for mitophagy after fertilization. If some triggering components of the machinery were assembled prior to fertilization, in the sperm, the explanation would have to be searched in these cells.

5) If an unexplained loss of membrane potential were the trigger for mitophagy, then one would expect to see it occurring even if mitophagy is blocked by silencing of it components. This experiment was not performed. The opposite would suggest that depolarization could be the effect of a damage inflicted onto sperm mitochondria.

6) It should be determined if the loss of sperm mitochondria function could be triggered by insufficient maintenance by the embryo. For example, don't sperm mitochondria fuse with oocyte mitochondria? If not, why? Could the fusion machinery of the sperm mitochondria be insufficient?

https://doi.org/10.7554/eLife.17896.015

Author response

[…] One main difficulty that comes through the comments of all three reviewers is the challenge to assess the specificity of the mitophagy process you describe for paternal mitochondria. Are maternal mitochondria in zygotes also subjected to a similar process or is it specific to paternal mitochondria? This question may be addressed by using the same mutant mouse line but this time using female mutants crossed with wild-type males and while the number of maternal mitochondria may be large, you may be able to limit your quantitation by photoconverting Dedra in discrete areas of the zygote or by quantifying a fixed number of fields.

We agree that the specificity of the mitophagy process is important to address. To examine this issue, we monitored the fate of mito-Dendra2 in the embryo at 36-84 hours post- fertilization depending on whether the fluorophore originated from the sperm or the egg. These new data are presented in Figure 1H and quantified in Figure 1—figure supplement 1 of the revised manuscript. When mito-Dendra2 arose from the male, the total fluorescence signal declines sharply at 72 and 84 hours, consistent with our earlier observations. In contrast, when mito- Dendra2 arose from the female, the total fluorescence signal shows no decline throughout the monitoring period. These results clearly indicate that the mitophagy process is specific for paternal mitochondria.

Related to this point, we also found that paternal mitochondria, unlike maternal mitochondria do not fuse with other mitochondria in the embryo (Figure 6—figure supplement 1). This segregation of paternal mitochondria is likely relevant to their eventual degradation.

The use of glucose-free, acetoacetate medium also raised a number of questions. Not much is known about the effects of this medium on the biology of the cell beyond its action on OXPHOS. Could this medium have other effects that confound the interpretation of the results in MEFs? Are these culture conditions relevant to the more physiological conditions in the zygote? If the zygote has a more glycolytic metabolism, and the machinery identified here required an enhanced OXPHOS metabolism, how accurate can the identified machinery be? Clearly, some core elements such as Parkin and Mulan have been identified and are sufficient to modulate mitophagy in zygotes but these questions regarding the appropriateness and relevance of the culture medium must be addressed.

This is an interesting issue – we did not intend to imply that OXPHOS-inducing media conditions bear a physiological relationship to early embryonic growth. We used the OXPHOS- inducing condition simply as an experimental tool to promote higher levels of mitophagy, whose molecular basis could be assessed. We then asked whether the molecules identified (MUL1, PARKIN, P62, FIS1, TBC1D15) are relevant to paternal mitochondrial degradation. The two conditions may in fact bear little direct relevance (in terms of metabolism) to each other. This issue is intriguing, because we may have identified mitophagy machinery that is relevant to mitochondrial degradation in a broad range of cell contexts.

In the revised manuscript, we have added the following text to the Discussion make this point clear:

"It is unclear whether MEFs cultured under OXPHOS conditions bear any physiological relation to the early embryo. Nevertheless, we find that the genetic requirements for removal of paternal mitochondria in the embryo mirror those of MEFs undergoing mitophagy in response to OXPHOS induction."

This led us to raise the question of membrane potential. Was the machinery of mitophagy that you identified validated under normal membrane potential? In the last paragraph of the Results section you indicate that zygotes show a time-dependent loss of membrane potential, suggesting that this trigger is important to the autophagy of paternal mitochondria. From your data you do not demonstrate that the loss of membrane potential is instrumental here. It is possible that the loss of membrane potential is an epiphenomenon, unrelated to the autophagy of paternal mitochondria, since the machinery of mitophagy you report herein has not been identified in the context of loss of membrane potential. It would thus be helpful to at least validate your identified mitophagy machinery under an induced membrane potential defect. Formal assessment in the zygotes of the membrane potential in maternal mitochondria should also be provided.

We agree – loss of membrane potential is an attractive trigger for mitophagy, but its functional role in paternal mitochondrial degradation cannot be definitively concluded from the available evidence. Given the known role of PARKIN in membrane-potential-dependent mitophagy, and the data showing loss of membrane potential in paternal mitochondria, it is reasonable to suggest that loss of membrane potential may play a role in triggering mitophagy in the early embryo. But it remains possible that it is an epiphenomenon. In the revised text, we state this point explicitly in the Discussion:

"Because we find that paternal mitochondria lose membrane potential shortly after entering the oocyte, it is tempting to speculate that this membrane depolarization may be the trigger for mitochondrial degradation. […] However, we do not have direct evidence that membrane depolarization has a functional role in paternal mitochondrial degradation."

In the cell culture assay, we found that CCCP-induced mitophagy is substantially reduced in MEFs lacking PARKIN and MUL1, but not cells lacking only one of these proteins. This result is presented in Figure 4—figure supplement 1D.

Figure 6B shows that maternal mitochondria retain membrane potential, in contrast to paternal mitochondria.

Your cellular and subcellular quantification methods were also questioned, and it would be important to better define how many elements were counted, how they were selected, and in how many cells from the embryos. Along this line it may be helpful to more clearly stress in the results that you counted all of the cells in each embryo and all paternal mitochondria in each cell. If, in some instances, this was not the case, then you must clarify how cells and/or mitochondria were selected and whether corrections were made for different number of cells per embryos of different ages. In addition, it was not clear what the level of detection is of Dendra-containing mitochondria in the microscope. Is the precision of your method and probe sufficient to confirm that one counted object equals one organelle? Also not clear from the manuscript was how you deal with detection of two mitochondria that are the fragmented/fissioned daughters of a parental organelle.

All viable embryos from each experiment were imaged in their entirety. First, the top and bottom positions for each embryo were identified and used to capture z-stacks that recorded the entire embryo. Optical slices were taken at 1.1μm thickness, and z stacks were oversampled at 0.467 μm to ensure that all mitochondria were captured. Maximum intensity projections of these z-stacks were used for quantification. While it is true that embryos from the same litter can contain different numbers of cells, by utilizing maximum intensity z-projections of the full embryo, all mitochondria were counted regardless of slight variations in developmental stage.

The control and experimental embryo images were randomized and counted blind. Green fluorescent signals were counted manually. A mitochondrial object was defined as a green fluorescent structure with visibly distinct borders. In cases where two or more mitochondria could not be resolved definitely as separate organelles, they were counted as one mitochondrial object. As indicated in the representative images in Figure 5A, C, the size of mitochondrial objects was within the range of 0.4 μm to 4 μm. Each maximum intensity z-projection was categorized as having either no mitochondria, less than five mitochondrial objects, or five or more mitochondrial objects.

These methodological details are now included in the Materials and methods section.

Lastly, is it possible to photoconvert a single mitochondrion and follow its fate (sounds technically daunting but would be truly informative)?

We performed photoconversion experiments with paternal and maternal mitochondria and found that they behave very differently. The photoactivated signal in maternal mitochondria diffuses over time and positive pixels decreases in intensity, indicating extensive fusion with adjacent mitochondria. In contrast, the photoactivated signal in paternal mitochondria does not change over time, indicating a lack of fusion activity. In the revised manuscript, these data are included in Figure 6—figure supplement 1A-B.

As you consider revising your manuscript, please also provide more detailed information about the statistical treatment of the data as requested by reviewer #1. Also, it would also be important to provide further detail and additional controls for the knockdown experiments to assure that the correct targets were indeed silenced and the effects do not reflect off-target effects, as discussed by reviewer #2.

In the revised manuscript, we have included detailed information about the statistical analysis and have included an excel file with the relevant statistical details for the figures.

In Figure 4—figure supplement 1B, C, we present the control experiments discussed by reviewer #2. These results indicate that the shRNA targets are effectively silenced. In addition, we show that mitophagy is robustly rescued by shRNAi-resistant constructs, thereby addressing the issue of off-target effects.

Finally, we would like to provide our editorial position with respect to two more points. First, with respect to the novelty of the study, as questioned by reviewer #3 ("The MEF experiments are convincing, but do not illustrate novel mechanisms"), it is our position that the novelty of the study does not reside in the "novel mechanism" of mitophagy, but in the demonstration that mitophagy (even if it uses a known mechanism) is involved in the elimination of paternal mitochondria.

Second, with respect to some of the mechanistic gaps, primarily pertaining to how and when paternal mitochondrial are recognized (e.g. reviewer #3, points #1, 2, and 6), these are valid questions but are beyond the scope of this specific study. However, a valuable approach to at least provide support to the specificity of the mechanism for the paternal mitochondria would be to compare intra- vs. inter-species mitophagy. Indeed, the phenomenon of paternal mitochondrial incompatibility applies only to intra-specific matings in that in inter-specific matings the paternal mitochondria survive, at least in mice (Kaneda et al., PNAS 92:4542, 2015). Consequently, an inter-specific mating should essentially abrogate all the relevant phenomena found here in the intra-specific mating (e.g. the decay of Dendra signal), thereby strongly supporting your conclusions.

We appreciate the editorial position concerning mechanism.

The suggested inter-specific experiment is very insightful. In the Kaneda et al. study, M. spretus males were mated to M. m. domesticus females to generate F1 hybrid embryos. PCR-based analysis showed that 25 of 45 neonates had detectable spretus mtDNA. Thus there is a variable amount of paternal transmission that occurs in this interspecific cross. As discussed in the Kaneda et al. paper, this result suggests that domesticus factors existing in the egg likely operate with reduced efficiency on the spretus mitochondria.

We were intrigued by this result and pursued it further. However, the only practical way to perform this experiment would be in the reverse direction of the Kaneda et al. study. We mated our domesticus males (containing mito-Dendra2) with commercially available spretus females, which we primed with hormone. Unfortunately, when we flushed the oviducts of such females, we found only a few abnormal, nonviable embryos.

We examined further the literature on interspecific mouse matings and found that the direction of the cross is crucial: domesticus males do not mate successfully with spretus females. For example, Zechner et al. (Nature Genetics 12: 398-403, 1996) states that "While the (mus x spr) F1 offspring are readily obtained, the reciprocal (spr x mus) F1 are notoriously difficult to breed." [Note: In this notation, the female parent is listed first.] A review (Dejager et al. Trends in Genetics 25: 234-241, 2009) on spretus interspecific genetics states: "Although viable F1 hybrids can be easily produced by mating a M, spretus male with a female of most classical inbred strains, the reciprocal crosses are almost invariably sterile."

Moreover, F1 males from mus x spr crosses are sterile due to lack of sperm cells. As a result, it is not possible, through mating, to construct a spretus male hybrid with the mito-Dendra2 allele.

https://doi.org/10.7554/eLife.17896.016

Article and author information

Author details

  1. Rebecca Rojansky

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    RR, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article
    Competing interests
    The authors declare that no competing interests exist.
  2. Moon-Yong Cha

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    M-YC, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.
  3. David C Chan

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    DCC, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    dchan@caltech.edu
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0191-2154

Funding

National Institute of General Medical Sciences (GM08042)

  • Rebecca Rojansky

National Institutes of Health (GM119388)

  • David C Chan

National Institutes of Health (GM083121)

  • David C Chan

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We are grateful to Shirley Pease (Director, Transgenic Core at Caltech) for training and advice on embryo injection. We thank Katherine Kim for preliminary work with MUL1 knockdown experiments, Kurt Reichermeier for advice on the ubiquitin assay, Ruohan Wang for technical assistance with p62 overexpression, and Hsiuchen Chen for advice on animal work. RR is supported by an NIH NIGMS training grant (GM08042) and the UCLA Medical Scientist Training Program.

Reviewing Editor

  1. Serge Przedborski, Columbia University Medical Center, United States

Publication history

  1. Received: May 17, 2016
  2. Accepted: November 14, 2016
  3. Accepted Manuscript published: November 17, 2016 (version 1)
  4. Version of Record published: November 29, 2016 (version 2)

Copyright

© 2016, Rojansky et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,795
    Page views
  • 969
    Downloads
  • 36
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)