Nuage is an electron‐dense cytoplasmic structure in germ cells that contains ribonucleoproteins and participates in piRNA biosynthesis. Despite the observation that clustered mitochondria are associated with a specific type of nuage called intermitochondrial cement (pi‐body), the importance of mitochondrial functions in nuage formation and spermatogenesis is yet to be determined. We show that a germ cell‐specific protein GASZ contains a functional mitochondrial targeting signal and is largely localized at mitochondria both endogenously in germ cells and in somatic cells when ectopically expressed. In addition, GASZ interacts with itself at the outer membrane of mitochondria and promotes mitofusion in a mitofusin/MFN‐dependent manner. In mice, deletion of the mitochondrial targeting signal reveals that mitochondrial localization of GASZ is essential for nuage formation, mitochondrial clustering, transposon repression, and spermatogenesis. MFN1 deficiency also leads to defects in mitochondrial activity and male infertility. Our data thus reveal a requirement for GASZ and MFN‐mediated mitofusion during spermatogenesis.
Intermitochondrial cement (IMC) with clustered mitochondria and RNA‐processing enzymes is essential for spermatogenesis. This study shows that mitochondria‐localized GASZ at IMC interacts with MFNs to regulate mitochondrial activities and germ cell development.
The germ cell‐specific gene GASZ contains a mitochondrial localization signal that is crucial for GASZ functions in spermatogenesis.
GASZ interacts with MFNs to promote mitofusion in somatic and germ cells.
MFN1 is required for the development of postnatal spermatocytes.
Mitochondria play important roles in various processes including ATP synthesis, production of reactive oxygen species (ROS), calcium signaling, and apoptosis , , . The multifaceted functions of mitochondria are largely dependent on their dynamically changing structures . For example, in somatic cells, the mitochondrial morphology changes primarily between a tubular type of fused active mitochondria and a spherical shape of fragmented quiescent arrangement to accommodate different energy requirements of cells and to maintain normal cellular homeostasis , , . In adult testes, spermatogonia and leptotene spermatocytes contain orthodox mitochondria, whereas spermatocytes at later stages and spermatids mainly harbor the condensed form that is more efficient for ATP production , . Consistently, spermatogonia exhibit a higher glycolytic activity, while the spermatocytes and spermatids synthesize ATP mainly through mitochondria‐mediated oxidative phosphorylation (OXPHOS) , , , . To date, the physiological significance underlying these changes in mitochondria in germ cell development remains unclear.
In most cells, mitochondria form an extended network that radiates from the nucleus, creating an interconnected system that supplies essential energy and metabolites , , , . In specialized cells, however, mitochondrial distribution is very different. One particular germinal granule that mitochondria are associated with in germ cells is nuage, which is an amorphous electron‐dense structure in the cytoplasm without limiting membrane . In embryonic gonocytes, postnatal spermatogonia, and spermatocytes, clustered mitochondria are found within a specific type of nuage (pi‐body), which is therefore called intermitochondrial cement (IMC) as well , . Many germ cell‐specific proteins are temporarily associated with this type of nuage, including GASZ, PIWI family members for piRNA biosynthesis (including MILI, MIWI, and MIWI2 in mice), MVH/VASA, TDRD1, and TDRD2 , , , , , , . Loss of function of these proteins usually leads to disrupted formation of nuage, defective piRNA biosynthesis, upregulated transposon expression, and male infertility , , , , , , . The nuage has thus been known for its roles in small RNA biosynthesis and the maintenance of genome integrity via repression of transposable elements by piRNAs , , . It is unclear, however, how and why the mitochondria are clustered at IMC, how those germ cell‐specific proteins are localized to IMC, and what role mitochondria may play during germ cell differentiation.
The heterogeneity in the distribution, activity, and morphology of mitochondria in various cell types is dynamically regulated via fusion and fission events , , . Mitochondria are double‐membrane‐bound organelles. Among proteins that participate in these processes, MFN1 and MFN2 are GTPases that are essential for outer membrane fusion, whereas DRP1 is the key enzyme that cooperates with FIS1 and MFF for mitochondrial fission , , . The absolute reliance on mitochondrial fusion and fission has been highlighted in neuronal development. Mutations of MFN2 is associated with Charcot–Marie–Tooth disease type 2A , an autosomal dominant neuropathy in human, and its deletion results in the reduction in dendritic outgrowth and spine formation in murine Purkinje cells as well . It has been reported that Fzo, a Drosophila homolog for MFN, is mainly expressed in spermatids and plays a role at late meiosis II . However, since mice with MFN null mutations are embryonic lethal , their functions in mammalian germ cell development have not been characterized. Recently, two reports demonstrated that MitoPLD, a phospholipase that facilities mitofusion, participates in nuage formation and piRNA biosynthesis during spermatogenesis , . Null mutation of MitoPLD in mice leads to the mislocalization of mitochondria and results in male infertility , , thereby indicating a potential role of mitofusion during germ cell development.
GASZ is a germ cell‐specific protein with four Ankyrin repeats/ANK, a sterile alpha motif/SAM, and a putative leucine zipper/ZIP . Previous studies demonstrated that enforced expression of GASZ promoted SSEA1+ germ cell derivation from both human and mouse embryonic stem cells (ESCs) . Loss of function of GASZ leads to male infertility with defects in nuage formation and piRNA biosynthesis . Yet the mechanisms by which GASZ affects these cellular function in germ cells have not been completely understood. We now report that GASZ contains C‐terminal mitochondrial targeting sequences and that GASZ dimerizes at the outer membrane of mitochondria and interacts with MFNs to promote mitofusion. Additionally, the deletion of mitochondrial targeting sequences of GASZ or MFN1 knockout leads to male infertility, demonstrating the absolute requirement for GASZ/MFN‐mediated mitochondrial function in spermatogenesis.
GASZ contains a mitochondrial localization signal
To explore the relationship of mitochondria to IMC formation, we searched for potential mitochondrial targeting sequences in proteins with known function or nuage localization in germ cells (Fig EV1A). According to the published reports, mitochondrial localization signals (MLS) generally lack acidic but are enriched in basic and hydrophobic amino acids, and they usually have the potential to fold into an amphiphilic α‐helix , . We found that only three out of these proteins with known functions in germ cell development met these criteria and contained MLS (Fig EV1A). MitoPLD and TDRD2 were previously reported to contain MLS and localized at mitochondria , , . The other one was murine GASZ that had a putative MLS at its C‐terminus from amino acids 451 to 475 (Figs 1A and EV1A). The MLS of GASZ is also found in the orthologs from different species (Fig EV1B), including human and non‐human primates, suggesting a functional conservation of GASZ–mitochondrial interactions during evolution.
To confirm the mitochondrial localization of GASZ, we examined its localization by subcellular fractionation assays. We observed that wild‐type GASZ was co‐enriched with MitoPLD in cytoplasmic and mitochondrial fractions (Fig 1B). As controls, nuclear protein NRF1 was largely restricted to the nucleus, whereas MVH, a known nuage protein without MLS, was found in the cytoplasm and mitochondria‐free cytosol (Fig 1B). The mitochondrial localization of GASZ was further supported by direct fluorescence visualization. In this assay, we found that the localization of GFP completely overlapped with MitoTracker, but not with markers for other subcellular compartments when it was targeted at the N‐terminus of GASZ (GFP‐GASZ, Fig EV1C and D). In addition, GFP tagged with MLS alone from GASZ was directly recruited to mitochondria rather than its cytoplasmic localization on its own (Fig 1C). By contrast, mutated MLS (Phe/F to Pro/P mutation in MLS, GFP‐F/P) or MLS with the deletion of its last four amino acids (ΔLQRK) failed to bring GFP to mitochondria (Fig 1C). These results, in large part, match those found in recent findings by Altshuller et al . In common, we identified a functional MLS at the C‐terminus of GASZ protein, and its mutation leads to the mislocalization of GASZ from mitochondria to the cytoplasm in somatic cells.
The mitochondrial localization of GASZ is critical for male reproduction
To further elucidate the role of mitochondrial localization of GASZ in vivo, we deleted its MLS through CAS9/CRISPR‐based gene targeting technique (Fig EV2). Out of the eight mice that were born, five male mice contained different types of MLS mutations (Fig EV2B). One of them harbored mutated MLS on both alleles (with a single‐base pair (bp) insertion on one allele and 2‐bp deletion on the other), and this mouse was later proved infertile (Fig EV2B: #2). Another male mouse with mutated MLS on both alleles appeared to be reproductively normal (Fig EV2B: #5). However, we found that GFP‐GASZ fusion protein containing this type of MLS mutation (#5: LK to Q) was localized at mitochondria in HeLa cells (Fig EV2E), suggesting that the alteration of leucine and lysine residues at the beginning of MLS had no influence on the subcellular localization and functions of GASZ.
We chose a female mouse as the founder for further investigation because the mutations (5‐bp deletion or GTA insertion/2‐bp deletion) started just before the beginning of MLS and in turn totally abolished the downstream mitochondrial targeting sequence (Fig EV2B: #6). In addition, an EcoRI site was disrupted, which could be used for genotyping to detect the mutated MLS (Fig EV2A–D: #6). We confirmed that GASZ with this type of MLS mutations expressed at a slightly reduced level in testes compared to wild‐type controls and abolished the mitochondrial localization of GASZ both in germ cells at day 7 postnatal and in HeLa cells when ectopically introduced (Figs 2A and B, and EV2E). Interestingly, both MILI and MVH displayed a diffused pattern in the cytoplasm of GASZΔMLS/ΔMLS gonocytes at 16.5 days post‐coitum (dpc) instead of perinuclear granular mitochondrial localization in wild‐type controls (Fig 2A), demonstrating mislocalized IMC proteins upon GASZ MLS mutation.
Female GASZΔMLS/ΔMLS mice exhibited no discernible phenotype and were fertile. By contrast, significant smaller testes were observed in adult male mice with MLS disruption at both alleles (Fig 2C). Although no obvious alterations were found in postnatal mutant mice at day 7, blocked development of GASZΔMLS/ΔMLS testes with fewer spermatocytes started to manifest by day 10 postnatal (Fig EV2G, P7 and P10). At P12, most of wild‐type seminiferous tubules advanced to zygotene stage, in contrast to fewer than 20% of mutant tubules containing spermatocytes (Fig 2D). In addition, only spermatogonia and preleptotene spermatocytes were detected in GASZΔMLS/ΔMLS testes at day P14 (Fig 2D). By 8 weeks, few germ cells were left with a Sertoli‐only phenotype in mice containing mutant MLS (Fig 2D).
As GASZ is associated with mitochondria at IMC in germ cells, we next examined the formation of IMC upon MLS disruption by transmission electron microscopy (EM). High electron‐dense structures were clearly seen among clustered mitochondria in spermatogonia from wild‐type neonatal mice (Fig 2E). By contrast, no discernible IMC was observed in GASZΔMLS/ΔMLS germ cells with a much lower degree of mitochondrial clustering as well (Fig 2E). In addition, consistent with the function of IMC as a site for the biosynthesis of piRNAs, which maintains transposon silencing and genomic stability in germ cells, the lack of IMC in turn led to a significant reduction in piRNA expression and an upregulation of transposon components including IAP and Line1 in GASZΔMLS/ΔMLS testes (Figs 2F and EV3A). Taken together, these data demonstrate that mitochondrial localization of GASZ is essential for its function in nuage formation, transposon repression, and spermatogenesis.
GASZ interacts with itself at mitochondrial outer membrane
The experiments above demonstrated that mitochondrial localization of GASZ was required for spermatogenesis, so we next explored the potential functions of GASZ at mitochondria. According to bioinformatic prediction, GASZ possessed essential features as a tail‐anchored membrane protein , with the same region of MLS as a transmembrane motif to target GASZ onto the mitochondrial outer surface (Figs 3A and EV4A). To validate this prediction, membrane‐bound or soluble fractions were extracted from 293T cells with Na2CO3 treatment (which separated membrane‐bound proteins from the soluble fraction) and examined for the existence of ectopically expressed GASZ by Western blots. We found that GASZ and MitoPLD were both enriched in the membrane‐bound fraction, whereas SIRT3, a protein inside of mitochondria, largely existed in the soluble fraction (Fig 3B). This result was further confirmed by treating the extracted mitochondria with proteinase K and/or Triton X‐100 (which permeabilized the mitochondrial membrane). Similar to outer membrane‐bound MitoPLD , digestion of ectopically expressed GASZ by proteinase K alone was observed in the absence of Triton X‐100 (Fig 3C). By contrast, SIRT3 and AIFM1, two proteins inside of mitochondria, were found degraded only in the presence of both proteinase K and Triton X‐100 (Fig 3C). Therefore, these data clearly illustrate that GASZ indeed exists at the outer surface of mitochondria.
Since GASZ contains several protein‐interacting domains including ANK and SAM, we then determined whether GASZ could associate with itself by GST pull‐down assays. We observed that the GST‐GASZ was indeed associated with FLAG‐tagged GASZ in an in vitro binding reaction (Fig 3D), indicating that GASZ interacts with itself. In addition, ectopically transfected FLAG‐tagged GASZ was pulled down with GFP‐GASZ fusion protein from 293T cells (Fig 3E). The interaction was further supported by a Bimolecular Fluorescence Complementation (BiFC) assay, in which YFP was split into half and glowed only when these two halves were in proximity to each other . Indeed, only weak fluorescence was detected when N‐YFP and C‐YFP were transfected into cells (Figs 3F and EV4B). By contrast, when the N‐ or C‐terminal half of YFP was fused to GASZ separately and co‐expressed, both flow cytometry and microscopy showed a much higher percentage of cells with brighter fluorescence (Figs 3F and EV4B). We further used co‐immunoprecipitation (Co‐IP) assays to determine the domains of GASZ that were required for its association and found that the interaction of GFP‐GASZ with its FLAG‐tagged deletion mutants was completely abolished in the absence of both SAM and ZIP motifs (Fig 3G). Taken together, these data demonstrate that GASZ interacts with itself at the mitochondrial outer membrane, likely through the SAM and ZIP domains.
GASZ affects mitochondrial dynamics and metabolism
GASZ interacts with itself at the outer membrane of mitochondria and thus may in turn bring mitochondria into proximity. Indeed, the aggregated mitochondria were observed when GASZ was ectopically expressed (Fig 4A). More importantly, aggregated and elongated mitochondria were clearly visible with transmission electron microscopy (EM) upon GASZ overexpression, and these were not seen when the MLS of GASZ was mutated (Fig 4B). In addition, we found that the formation of tubular mitochondria was increased when GASZ expression was enforced (Figs 4C and EV4C). This pro‐mitofusion ability of GASZ was abolished either when MLS was mutated or when the domains (SAM and ZIP) required for its association were deleted (Figs 4C and EV4C). Mitochondrial dynamic was further monitored directly by visualization and quantification of dilution rate in a mitochondria‐targeted photoactivable GFP (PAGFP) via time‐lapsed confocal imaging. Normalized fluorescence intensity within designated region at photoactivation area was collected and plotted against time. Compared to control HeLa cells or cells with MLS‐mutated GASZ, faster decrease in PAGFP fluorescence within photoactivated mitochondria was observed when wild‐type GASZ overexpressed (Figs 4D and EV4D), reflecting increased mitochondrial fusion rate by GASZ overexpression.
Due to technical challenges to directly monitor the mitochondrial dynamics in primarily cultured gonocytes or in testes in vivo, we investigated whether GASZ affected mitofusion in germ cells using an immortalized spermatogonia cell line, C18‐4, which was established by stable transfection of SV40 large‐T antigen into A‐type spermatogonial stem cells (SSC) . Similar to somatic cells, ectopic expression of GASZ significantly enhanced the percentage of SSCs harboring tubular mitochondria (Figs 4E and EV4E). In addition, we observed an increased copy number of mitochondrial DNA (mtDNA) from GASZΔMLS/ΔMLS testes at day 14 postnatal (Fig 4F). Consistently, significantly higher activities of mtDNA‐contributed respiratory enzyme complex including COX (cytochrome c oxidase in mitochondrial electron transport chain complex IV) and NADH dehydrogenase (a component of complex I) ,  were detected in GASZΔMLS/ΔMLS germ cells at days 10 and 14 postnatal, whereas the staining of SDH (succinate dehydrogenase in complex II), an enzyme entirely encoded by the nuclear genome , , was not obviously altered (Figs 4G and EV4F and G), reflecting that the mislocalization of GASZ in germ cells leads to an increased mitochondrial fission and a dysregulated mitochondrial metabolism.
GASZ promotes MFN‐dependent mitochondrial fusion
Since GASZ does not contain any catalytic motifs to initiate mitofusion by itself, we explored whether GASZ interacted with any known enzymes that engaged in mitofusion. Indeed, we found that GASZ interacted with two key GTPases in mitochondrial fusion, MFN1 and MFN2, both in vitro in GST pull‐down assays (Fig EV5A) and in Co‐IP (Fig 5A) and in testes in vivo (Fig 5B). By contrast, no evidence of interaction of GASZ with MitoPLD (an outer membrane‐bound phospholipase in mitofusion) or DRP1 (a GTPase in mitochondrial fission) was found (Fig EV5C and D). In addition, GASZ appeared to be associated with both MFN1 and MFN2 via its SAM and ZIP domains, and failed to interact with MFNs when these domains were absent (Figs 5C and EV5E). Surprisingly, we observed that the MLS mutants of GASZ still interacted with MFNs (Figs 5C and EV5E), indicating that MLS mutations did not physically block the interaction of GASZ with MFNs.
To further investigate whether mitofusion enhanced by GASZ was functionally dependent on MFNs, we introduced a GASZ‐expressing plasmid into wild‐type mouse embryonic fibroblasts (MEFs), or MEFs with deletion of MFN1, MFN2, or both (Fig EV5F). We found that ectopically expressed GASZ promoted the formation of elongated mitochondria in wild‐type MEFs, but was unable to increase the percentage of cells containing the tubular mitochondria in MFN‐deficient MEFs (Figs 5D and EV5G). We further monitored the mitofusion directly between Su9‐GFP‐ and COX4‐DsRed‐labeled mitochondria . Upon induction by polyethylene glycol (PEG), GASZ significantly increased the percentage of fused mitochondria in wild‐type MEFs, but failed to do so in the absence of MFN proteins (Figs 5E and EV5H). Taken together, these data demonstrate that GASZ promotes mitofusion in a MFN‐dependent manner.
Mice with conditional MFN1 deletion in germ cells are male infertile
So far, it is unknown whether MFN‐mediated mitofusion is required for spermatogenesis. The mice with null mutation of MFNs failed to survive to midgestation . To bypass the embryonic lethality, we used MFN1 conditional knockout mice (MFN1f/f)  and crossed them with a MVH‐Cre line to obtain MFN1 deletion specifically in germ cells (Fig EV6A), as Cre recombinase under the promoter of MVH started to be expressed only in gonocytes at 15.5 dpc . MFN1f/f‐Cre mice were born at appropriate mendelian frequencies and appeared to be physiologically normal compared to their littermate controls. No obvious alteration in reproductive organs was observed in MFN1f/+‐Cre male mice either (data not shown).
We dissected the testes from MFN1f/f‐Cre and MFN1+/+ mice after birth at different time points and found that the size of mutant testes was substantially reduced at 7 weeks compared to their wild‐type littermates (Fig 6A). The defects in MFN1f/f‐Cre testes we observed were very similar to GASZ MLS‐mutant mice. When leptotene/zygotene spermatocytes appeared at day 12 in wild‐type testes, MFN1f/f‐Cre seminiferous tubules mainly contained spermatogonia (Fig EV6C). By day 14 after birth, most MFN1+/+ tubules had advanced to mid‐pachytene and zygotene stage, whereas only a few tubules in MFN1f/f‐Cre testical sections containing spermatocytes were detected (Fig 6B). At 6 weeks postnatal, few germ cells were seen in the seminiferous tubules in MFN1f/f‐Cre testes, along with a complete absence of sperms in epididymis (Fig 6B). In agreement with these findings, MFN1f/f‐Cre male mice were infertile, thereby demonstrating an essential requirement of MFN1 in spermatogenesis. Notably, MFN1 deletion appeared to particularly affect the development of spermatocytes. No reduction in PLZF+ spermatogonia was observed in MFN1f/f‐Cre testes by day 14 postnatal (Fig 6C).
Although no obvious alteration in the localization of GASZ, cytochrome c, MILI, and MVH in MFN1f/f‐Cre germ cells were observed via immunohistofluorescence (IHF) (Fig 6C), electron microscopy revealed two types of mitochondrial distribution in MFN1f/f‐Cre spermatogonia: One contained highly clustered mitochondria at one side of the cells, and another displayed evenly distributed mitochondria in the cytoplasm without perinuclear aggregation. Remarkably, mitochondria from MFN1f/f‐Cre spermatogonia in both cases were larger in diameter, round and swollen with vesiculated cristae inside (Fig 6D). Among clustered mitochondria, a thin layer of electron‐dense IMC‐like structure was observed, but appeared to be thinner and longer than the ones in wild‐type cells (Fig 6D). Similar to GASZ MLS‐mutant mice, MFN1f/f‐Cre testes at day 10 postnatal displayed increased activities of respiratory enzymes, COX and NADH dehydrogenase, with nuclear DNA‐encoded SDH unaffected (Fig 6E), indicating a dysfunctional mitochondrial metabolism with a transient upregulation of mitochondrial fission due to MFN1 null mutation. Taken together, these results reinforce the critical roles of mitochondria and MFN1‐dependent mitofusion in the IMC formation and for the development of p ostnatal germ cells.
In germ cells, structures with ribonucleoprotein aggregates known as IMC form among mitochondrial clusters , . Yet the understanding of the intricate interplay between mitochondria and germ cell development is still limiting and descriptive, especially with regard to the importance and underlying molecular mechanisms of mitochondria in IMC formation and spermatogenesis. So far, most in vivo experiments have focused on removing one component from nuage by loss‐of‐function assays, and thus, reproductive defects caused by which were usually due to the absence of the targeted proteins, but not their mislocalization. We found that a germ cell‐specific protein GASZ formed dimers/oligomers at the outer membrane of mitochondria and promoted MFN‐dependent mitochondrial aggregation and fusion. We thus took an alternative approach by specific deletion of MLS from GASZ to further discern the functional relevance of mitochondria and germ cell development in mice. Here, we demonstrated that mutation of MLS delocalized GASZ from mitochondria to the cytoplasm in germ cells and led to defects of IMC formation, abnormal mitochondrial metabolism, and male infertility. We further disturbed mitochondrial function in germ cells directly by conditional deletion of MFN1, a key GTPase in mitofusion, and this resulted in male infertility as well. Therefore, our results reveal a critical requirement of GASZ/MFN‐mediated mitochondrial functions in germ cell development. Specifically, we propose that the roles of GASZ/MFN1 in these aspects are likely to be multifold.
Firstly, IMC is enriched with small RNAs (including microRNAs and piRNAs) and RNA‐binding proteins (including MVH, an ATP‐dependent RNA helicase, and PIWI family proteins in piRNA biosynthesis) . GASZ may act as a central scaffold protein to recruit these germ cell‐specific factors into close proximity at mitochondria at IMC. To support this hypothesis, multiple nuage‐associated proteins (including MIWI, DAZL, MVH, and TDRD1) have been previously identified as interacting partners for GASZ , , . Unlike a reported GASZ‐knockout study , GASZ and MILI proteins were still found in the GASZΔMLS/ΔMLS gonocytes with the MLS deletion, and both of them displayed a diffused pattern in the cytoplasm instead of a perinuclear granular localization at mitochondria in wild‐type controls. No discernable IMC was formed in the GASZΔMLS/ΔMLS neonatal testes either. Consistently, expression of piRNAs was significantly decreased, whereas transposable elements repressed by piRNAs were upregulated in GASZΔMLS/ΔMLS testes, likely due to disturbed piRNA biosynthesis caused by mislocalized GASZ and MILI. In this case, mitochondria likely function as central anchorages for RNA processing machinery. Mislocalization of GASZ from the mitochondria is sufficient enough to disrupt the normal IMC formation in germ cells and leads to defects in piRNA biosynthesis. Although GASZ interacts with mitochondria‐bound MFNs, localization of GASZ may not rely upon MFNs and likely precedes its interaction with MFNs at mitochondria. The dynamic or weak interactions between GASZ and MFNs may not be strong enough to stabilize GASZ with mutant MLS to the mitochondria with MFNs.
Secondly, self‐association of GASZ and its interaction with MFNs juxtapose mitochondria into clusters and contribute to the formation of IMC. Clustered mitochondria may form a “trap” to temporarily engage multiple RNA‐binding proteins to act in concert needed for small RNA biosynthesis. Notably, MLS mutation of GASZ resulted in reduced mitochondrial clustering and defects in IMC formation, which in turn led to defects in the silencing of transposable components in germ cells. According to our knowledge, most known proteins involved in mitochondrial dynamics are ubiquitously expressed in both germ cells and somatic cells. Yet the mitochondrial clustering is only observed in embryonic gonocytes, postnatal spermatogonia, and spermatocytes. This indicates that some other germ cell‐specific factors may be needed to aggregate mitochondria into proximity at IMC. The role of GASZ is thus unique as so far it is the only protein that has been proved to be germ cell specific, directly targets, and dimerizes/oligomerizes at mitochondria in somatic cells. Interestingly, high sequence similarity of MLS and the transmembrane motif in GASZ exists among different species, thereby suggesting an important functional conservation of GASZ at mitochondria in spermatogenesis during evolution.
Thirdly, GASZ‐induced juxtaposition of mitochondria at IMC may provide the energy required for ATP‐powered small RNA processing and germ cell maturation. It is known that dynamic mitochondrial fusion and fission act to allow cooperation between mitochondria, thereby maintaining their proper distribution, mtDNA stability, and energy requirement , , . Since GASZ is not an enzyme that directly catalyzes mitofusion, it may functionally depend upon MFNs or other undefined cofactors in regulating mitochondrial activities. It is plausible that GASZ, together with MFNs, modulates the morphology and dynamics of mitochondria and in turn regulates the energy production of mitochondria for the surrounding RNA processing machinery in germ cells. In agreement with this notion, dysfunctional mitochondrial metabolism was observed in both GASZ‐ and MFN1‐mutant testes, reflecting dysregulated mitochondrial dynamics in these mutant germ cells, and this may in turn lead to impaired germ cell maturation during spermatogenesis. As previously reported, the morphology, localization, and metabolism of testicular mitochondria change markedly during spermatogenesis, and the energy supply of differentiated spermatocytes relies primarily on mitochondrial metabolism , . Notably, although mitochondrial clustering or morphology appeared to be abnormal in the spermatogonia from neonatal mice, GASZ or MFN1 mutations did not appear to affect PLZF+ germline stem cells at early stages of spermatogenesis, but dramatically blocked the spermatocyte formation.
Although MFN1 and MitoPLD are both required for mitofusion, MFN1 null mutant did not fully phenocopy the MitoPLD deficiency. Notably, no IMC was formed in MitoPLD−/− germ cells and mitochondria were mislocalized to the centrosome , . By contrast, IMC was still observed in MFN1 null spermatogonia, albeit appeared to be malformed compared to wild‐type controls, reflecting a distinct functional contribution of MFN1‐dependent mitofusion and mitochondrial metabolism in spermatogenesis. In addition, MFN1 null mutation resulted in severe defects in mitochondrial morphology, evidenced by swollen and vesicular mitochondria. This phenotype is reminiscent of abnormalities observed in MFN2‐deleted Purkinje cells . Although both MFN1 and MFN2 were key GTPases involved in mitofusion, MFN2, but not MFN1, is required for proper mitochondrial functions of Purkinje cells , suggesting non‐redundant roles of MFNs in neuronal development. It will be of great interest to investigate whether MFN1 and MFN2 are both required for spermatogenesis. Nevertheless, our data reveal an essential requirement of GASZ/MFN‐mediated mitochondrial functions during spermatogenesis.
Materials and Methods
Genes used in this study were amplified from a cDNA library of mouse testis. GASZ cDNA was cloned into the vectors pEGFP‐C1 (in which GFP was fused with N‐terminus of GASZ), pEGFP‐N2 (in which GFP was fused with C‐terminus of GASZ), pcDNA3.1‐CYFP, pcDNA3.1‐NYFP, pGEX‐4T‐1 (in which GST was fused with N‐terminus of GASZ), or pll3.7‐3xFLAG‐IRES‐Hygro through XhoI and BamHI digestion. Domain deletion or MLS mutant for GASZ was constructed by PCR and QuickChange Site‐Directed Mutagenesis Kit (Stratagene). For MitoPLD‐HA and MFN‐HA constructs, MitoPLD, MFN1, and MFN2 cDNAs were cloned into pcDNA3.1‐hygromycin in frame with a C‐terminal HA tag. All primers used for cloning and PCR are listed in Table EV1.
Generation of GASZ MLS‐mutant mice via CAS9/CRISPR
The CAS9 expressing vector pX260 and pGS3‐T7‐gRNA plasmid to express guide RNAs are kind gifts from Dr. Dali Li, East China Normal University (ECNU). pGS3‐T7‐gRNA vector was linearized by DraI and used as a template for in vitro transcription using In Vitro Transcription T7 Kit (Takara). To generate CAS9 mRNA, CAS9‐expressing vector was linearized by NotI and transcribed into mRNA using the SP6 mMESSAGE mMACHINE Kit (Ambion). CAS9 mRNA and gRNAs were co‐injected into zygotes of C57BL/6 mice. To identify the correct gene editing in progenies (all produced from breeding with C57BL/6 mice), tail genomic DNA was amplified by PCR and subjected to EcoRI digestion and/or sequencing. Guide RNA or genotyping primers are listed in Table EV1. All animal experimental procedures were conducted in accordance with the local Animal Welfare Act and Public Health Service Policy (consistent with the WMA Statement on Animal Use in Biomedical Research) and approved by the Committee of Animal Experimental Ethics at ECNU. Animals were housed at the facility with SPF condition and euthanized by carbon dioxide inhalation followed by cervical dislocation before dissection at various time points. The wild‐type (or heterozygote) control and mutant littermates were chosen randomly according to their genotype. No blinding was done for animal experiments in this study.
Generation of conditional MFN1 deletion in germ cells
Conditional MFN1‐knockout mice were created previously . Sperms harboring floxed MFN1 allele were obtained from Mutant Mouse Regional Resource Center (MMRRC at UC Davis). Heterozygote MFN1+/f progenies were derived by intracytoplasmic injection of sperms into oocytes and maintained by crossing with C57BL/6 mice. MFN1f/f female mice were crossed with MVH‐Cre‐MFN1+/f male mice (MVH‐cre mice from Model Animal Research Center at Nanjing University, China) to obtain targeted MFN1 deletion in germ cells.
Cell culture, transfection, and viral infection
HeLa, 293T, MFN1−/−, MFN2−/−, and MFN1&2‐double‐knockout MEFs were maintained in DMEM with 10% fetal bovine serum (FBS, Gibco). For ectopic gene expression, 293T or HeLa cells (originally from ATCC) were transfected with plasmids using calcium phosphate or Lipofectamine®2000 (Invitrogen), respectively. MEFs were infected with retrovirus or lentivirus as previously described . All cell lines were cleared from mycoplasma contamination by PCR with mycoplasma‐specific primers.
Quantification of mitochondrial morphology and immunofluorescence (IF)
HeLa, C18‐4, or MEFs cultured on gelatin‐coated cover slips were transfected with expressing plasmids, incubated with 150 nM MitoTracker Red CMXRos (Invitrogen) at 37°C for 30 min, and fixed with 3.7% paraformaldehyde (PFA). Following permeabilization in ice‐cold acetone or 1% Triton X‐100, cells were directly analyzed or blotted with anti‐FLAG antibody (Sigma) in 5% BSA/PBS and FITC‐conjugated anti‐mouse antibodies (Jackson ImmunoResearch). Mitochondrial morphology was visualized by Leica DM4000B. For each sample, 200 randomly chosen cells were counted and the percentage of different types of mitochondria‐containing cells was calculated according to published reports .
PEG fusion assays and photoactivation assays
PEG fusion assays were performed according to the published protocols . Briefly, 5 × 105 MEFs expressing mitochondria‐targeted DsRed (COX4‐DsRed) were cultured on cover slips overnight at the ratio of 1:1 with MEFs expressing mitochondria‐targeted GFP (Su9‐GFP). The cells were incubated in DMEM with 10% FBS containing 30 μg/ml cycloheximide (CHX) for 30 min and then treated with PEG1500 (Roche) for 2 min at room temperature. After washing with medium three times, the cells were cultured in media with 30 μg/ml CHX again for 5 h and fixed with 3.7% PFA for 15 min before being examined under confocal microscope (Leica TCS SP5) or Leica DM4000B. For photoactivation assay, the cells were grown in 2‐well chambers and transfected with Mito‐PAGFP (mitochondria‐localized photoactivated GFP), COX4‐DsRed, and empty control vector or GASZ‐expressing plasmids via Lipofectamine®3000 (Invitrogen). Live cells were maintained at 37°C in assay solution (10 mM Hepes, 10 mM glucose, 3 mM KCl, 145 mM NaCl, 1.2 mM CaCl2, 1.2 mM MgCl2, pH 7.4), and images were captured with a FW1000 confocal microscope (Olympus) using a 40 × 0.95 objective for time‐lapse imaging and 60 × 1.2W objective for high‐resolution imaging, respectively. Laser light at 405 nm was used for the photoactivation of PAGFP. DsRed was used to monitor the transfection efficiency and morphology of mitochondria in the same photoactivated cells. Normalized fluorescence intensity (dN = PAGFP‐background/DsRed‐background) within designated region at photoactivation area was collected every 10 s and plotted against time. Exponential decay constant λ (dN/dt = −λN, t: time) of PAGFP was calculated. More than twenty cells for each group were randomly selected for quantitative analysis and imaging.
Transmission electron microscopy
EM was performed as described previously . Briefly, samples were fixed with 2.5% EM‐grade glutaraldehyde in 0.1 M PBS (pH 7.4) overnight and then fixed in 1% osmium tetroxide (OsO4) for 2 h at 4°C in the dark. After dehydration in a series of ethanol (30, 50, 70, 90%), the samples were treated in 45% ethanol and acetone for 15 min at 4°C and in 100% acetone three times. The samples were embedded in epoxy resin, processed into 60‐nm sections, and counterstained with uranyl acetate and lead citrate before being examined under a Hitachi H7500 transmission electron microscope.
Subcellular fractionation and membrane extraction assays
The nuclear and cytoplasmic proteins or mitochondria and mitochondria‐free cytosolic fractions of 293T cells were isolated using a Nuclear and Cytoplasmic Protein Extraction Kit or a Cell Mitochondria Isolation Kit (Beyotime) following the manufacturer's protocol. To determine whether target proteins were associated with mitochondrial membrane, the mitochondria‐rich fractions were incubated in the presence or absence of 1% Triton X‐100 with protease inhibitor cocktail (Sigma) in PBS at room temperature and/or following a treatment with 150 μg/ml proteinase K on ice for 15 min. For membrane extraction assay, the cells were harvested and subsequently incubated in 0.1 M Na2CO3, pH 11.5 on ice for 30 min, as described previously . After centrifugation at 10,000 g for 30 min, the supernatants were collected as soluble proteins and the precipitates were enriched for the membrane‐bound proteins. Target proteins were further examined by Western blots.
Western blots and Co‐IP
Cells were harvested in lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 10% glycerol, and 1% Triton X‐100 with protease inhibitor cocktail). The cleared supernatant was incubated with primary antibody, followed by treatment with Flag‐M2 agarose (Sigma, A2220) or Protein A/G‐agarose beads (Santa Cruz Biotech, SC‐2003) to pull down antibody‐interacting complex, which was subsequently separated by SDS–PAGE gel and transferred to nitrocellulose membranes. The membranes were blotted with primary antibodies, followed by secondary antibodies conjugated with IRDye®680 (LI‐COR Biosciences, 926‐68020) or IRDye®800CW (LI‐COR Biosciences, 926‐32211). The bands on the filter were observed by Odyssey infrared imaging system (LI‐COR Biosciences). Primary antibodies used in this study were as follows: HA (Santa Cruz Biotech, SC‐805); FLAG (Sigma, F7425); GFP (Santa Cruz Biotech, SC‐8334); β‐actin (Sigma, A5441); β‐tubulin (Abmart, M30109); DRP1 (Cell Signaling, 8570), MVH (Abcam, ab13840); cytochrome c (Abcam, ab13575); MILI (Cell Signaling Technology, 5940); GASZ ; MFN1 and MFN2 (Abnova ).
GST pull‐down assay
GST‐GASZ fusion protein was induced by IPTG in bacteria at 33°C overnight, purified, and immobilized onto glutathione agarose beads. The protein–bead mixture was blocked with PBS containing 1% Triton X‐100, 10 mg/ml BSA and incubated with 293T cell lysate containing FLAG‐GASZ or MFN1/2‐HA proteins for 3 h at 4°C. The bead‐bound proteins were washed five times with PBS containing 1% Triton X‐100 before being separated by SDS–PAGE and detected by Western blots.
Real‐time RT–PCRs and PCRs
As previously described , total RNAs were extracted with TriZOL (Invitrogen) and cDNAs were synthesized using a PrimeScript® RT reagent kit (TaKaRa Biotechnology Co., Ltd.). For the determination of piRNA expression, reverse transcription was carried out using PrimeScript® RT Kit with gDNA Eraser (Takara) and piRNA‐specific RT primers following the published protocols . Genomic DNA and mtDNA were isolated from fresh mouse testes using PUREGENE DNA Purification kit (QIAGEN) and diluted to 5 ng/μl for the determination of mtDNA copy numbers relative to PECAM gene (a single‐copy gene encoded by nuclear DNA) . Quantitative real‐time PCR was performed on Stratagene MX3000P instrument using SYBR Premix Ex TaqTM II (Takara). The mRNA and piRNA expression levels were normalized to β‐actin or GAPDH and 5S rRNA, respectively. Primers used in this study are listed in Table EV1.
Histology and immunohistofluorescence (IHF)
Mouse testes were fixed in Bouin's fixative at 4°C overnight for staining with hematoxylin and eosin. For IHF, testes were fixed in 4% paraformaldehyde at 4°C overnight and embedded in paraffin. Antibodies used in this studies were as follows: guinea pig anti‐GASZ , PLZF (Santa Cruz Biotech, SC‐28319), MILI (Cell Signaling Technology), FLAG (Sigma), FITC‐conjugated anti‐guinea pig and FITC‐ or TRITC‐conjugated anti‐mouse, or anti‐rabbit (Jackson ImmunoResearch, 706‐095‐148; 115‐095‐146, 115‐025‐146; 111‐095‐144 & 111‐025‐144). Nuclei were stained with 0.5 μg/ml 4′,6′‐diamidino‐2′‐phenylindole (DAPI) for 5 min before being visualized by Leica TCS/SP5 confocal microscope and processed with Image‐Pro Plus.
The activities of COX/NADH/SDH on testical sections were performed according to the published protocols ,  and online procedures from Washington University Neuromuscular Disease Center (http://neuromuscular.wustl.edu/index.html). Briefly, cryostat testical sections at 12 μm were prepared from fresh‐frozen testical samples (n = 3 per genotyping group). Slides were incubated for 40 min/60 min/30 min at 37°C with the COX/SDH/NADH reaction buffer, respectively, before visualization.
Statistical analysis and general method to avoid subjective bias
An online software TMHMM was used to predict MLS in germ cell‐specific proteins (http://www.cbs.dtu.dk/services/TMHMM/), and ClustalX was used to compare the MLS conservation sequences among different species. For all experiments, three or more than three biological replicates have been performed, and the representative one (with others showing similar results) is shown in the figure for Western blot, Co‐IP, HE, IF, IHF, and EM. All statistical analyses were performed using F‐test (to estimate the variance within each group of data) and the unpaired t‐test (for comparison between groups) with GraphPad Prism software. For all experiments, the criteria for inclusion/exclusion of samples were pre‐established. Randomizing of sample allocation and measure as well as blinding of investigator was used when possible.
JZ, QW, MW, MJ, YW, YS, JW, TX, CT, NT, HS, DC, RC, and YW performed the experiments and analyzed the data. SD, BN, and XC contributed reagents and analysis tools. YW conceived and supervised the study. JZ, QW, and YW wrote manuscript.
Conflict of interest
The authors declare that they have no conflict of interest.
Expanded View Figures PDF
We would like to thank Drs. Chuangui Wang and Dali Li at ECNU for plasmids containing FLAG‐SIRT, FLAG‐AIFM1, and pX260/gRNA constructs in CAS9/CRISPR system. We also thank Drs. David Chan (California Institute of Technology) for Su9‐GFP and COX4‐DsRed plasmids, Dr. Quan Chen (Institute of Physics, Chinese Academy of Sciences/CAS) for MFN‐knockout mouse embryonic fibroblasts, Dr. Martin Dym at Georgetown University Medical Center for C18‐4 cell line, and Dr. Martin Matzuk at Baylor College of Medicine for anti‐GASZ antibody. We also thank Drs. Chen Xu at Tongji University, Degui Chen and Muofang Liu from CAS, Mitch Eddy from NIEHS, and Richard Mailman from Penn State College of Medicine for technical help in mouse work. This work was supported by grants from the Ministry of Science and Technology of China (2010CB945403 and 2014CB964800), the National Natural Science Foundation of China (31271589 and 30971522), the Science and Technology Commission of Shanghai Municipality (11DZ2260300 and 13JC1406402), China.
FundingMinistry of Science and Technology of China2010CB9454032014CB964800
- © 2015 The Authors