Electron transport chain remodeling by GSK3 during oogenesis connects nutrient state to reproduction

Summary

Reproduction is heavily influenced by nutrition and metabolic state. Many common reproductive disorders in humans are associated with diabetes and metabolic syndrome. We characterized the metabolic mechanisms that support oogenesis and find that mitochondria in mature Drosophila oocytes enter a low activity state of respiratory quiescence by remodeling the electron transport chain (ETC). This shift in mitochondrial function leads to extensive glycogen accumulation late in oogenesis and is required for the developmental competence of the oocyte. Decreased insulin signaling initiates ETC remodeling and mitochondrial respiratory quiescence through Glycogen Synthase Kinase 3 (GSK3). Intriguingly, we observed similar ETC remodeling and glycogen uptake in maturing Xenopus oocytes, suggesting that these processes are evolutionarily conserved aspects of oocyte development. Our studies reveal an important link between metabolism and oocyte maturation.

Introduction

Reproduction is intimately associated with nutrition and the metabolic state of the organism. Moreover, reproduction imposes tremendous demands on female metabolism to provide the nutrients and building blocks required for embryogenesis. Women suffering from metabolic syndrome and diabetes frequently suffer from infertility, polycystic ovary syndrome (PCOS), and elevated rates of ovarian cancer (Ben-Shlomo and Younis, 2014; Tania et al., 2010). Intriguingly PCOS, the most common cause of female infertility, blocks both oocyte maturation and ovulation suggesting a link between metabolic regulation and oocyte development (Mayer et al., 2015; Velez and Motta, 2014). Furthermore, metabolic disruptions may be associated with cases of impaired developmental competence of oocytes of women undergoing in vitro fertilization (Van Blerkom, 2011). Despite these links the precise metabolic mechanisms that support normal oocyte development and maturation remain unknown.

Drosophila oocyte development provides an ideal system to characterize the links between metabolic regulation and oocyte development. During Drosophila oogenesis, germline and somatic stem cells continuously generate new follicles comprising an oocyte and 15-associated nurse cells surrounded by somatic follicle cells (Spradling, 1993). Follicles develop through 14 morphologically distinct stages over 7 days. During the final stages of oogenesis, yolk proteins and lipids normally accumulate in a process called vitellogenesis (Buszczak et al., 2002; Sieber and Spradling, 2015) (Raikhel et al., 2005). At the end of vitellogenesis, carbohydrates accumulate within the oocyte (Gutzeit et al., 1994), transcription in the oocyte ceases, translation is reduced, and quiescence ensues (Lovett and Goldstein, 1977) (Mermod et al., 1977). The egg’s stored lipids and carbohydrates function as a fuel depot for the developing embryo (Tennessen et al., 2014b). Similarly in mammals, where the placenta reduces the burden of nutrient storage, oocytes still enlarge greatly and accumulate large amounts of nutrient-rich yolk for use during the early developmental stages (Dunning et al., 2014a; Dunning et al., 2014b; Dunning et al., 2010). Furthermore, carbohydrates facilitate mammalian oocyte maturation and developmental competence after fertilization (Sutton-McDowall et al., 2010; Zheng et al., 2001) suggesting a highly-conserved link between carbohydrate accumulation and oocyte development.

One key factor dictating the balance of carbohydrate catabolism and storage is mitochondrial oxidative metabolism. In most cells, with active insulin signaling, glucose is broken down through glycolysis into pyruvate that is consumed by the mitochondria to supply the tricarboxylic acid (TCA) cycle. The TCA cycle then provides intermediates such as NADH and succinate to fuel the electron transport chain (ETC) and stimulate ATP production. During oogenesis, mitochondria increase greatly in number in mature oocytes (Cox and Spradling, 2003; Pepling et al., 2007; Wallace and Selman, 1990), however, mature oocytes display low mitochondrial activity (Dumollard et al., 2007; Trimarchi et al., 2000; Van Blerkom, 2011).(Cox and Spradling, 2003; Pepling et al., 2007; Wallace and Selman, 1990). After fertilization, calcium is released from the endoplasmic reticulum and mitochondrial activity increases during embryogenesis (Campbell and Swann, 2006; Dumollard et al., 2007; Hansford, 1994; Trimarchi et al., 2000; Van Blerkom, 2011). Surprisingly, the mechanism behind these major changes in mitochondrial function during oocyte development, fertilization, and early embryogenesis remain unclear.

The insulin-signaling pathway is a key regulator carbohydrate metabolism, mitochondrial function, and growth (Saltiel and Kahn, 2001; Stump et al., 2003). Insulin stimulates the PI3K/AKT kinase cascade to activate targets that affect both growth and metabolism. Studies in Drosophila have shown that insulin signaling regulates oogenesis by controlling germline stem cell division, promoting follicle growth, and controlling the transition to vitellogenesis during stage 8 (LaFever and Drummond-Barbosa, 2005) (Drummond-Barbosa and Spradling, 2004) (Drummond-Barbosa and Spradling, 2001). Similarly, insulin is also thought to control mammalian oocyte development (Acevedo et al., 2007; Liu et al., 2006) (Pan et al., 2005) (Fisher et al., 1998). However, very little is known about the metabolic role for insulin signaling during germline development.

In this study we identified a dynamic shift in carbohydrate utilization during late oogenesis that leads to extensive glycogen accumulation. This shift in metabolism occurs as mitochondria depolarize, remodel their ETC, and enter the low-activity state of mitochondrial respiratory quiescence. We have found that entry into mitochondrial respiratory quiescence is controlled by a reduction in insulin signaling during the final stages of oocyte development. More specifically, the Akt target gene GSK3, which promotes remodeling of the electron transport chain components in the mitochondria, triggers mitochondrial respiratory quiescence. Overall these studies define the dynamic changes in mitochondrial function that couple nutrient state to oocyte development.

Results

A global shift in carbohydrate utilization promotes glycogen accumulation.

To examine carbohydrate accumulation during Drosophila oogenesis we dissected ovaries from adult Oregon R females and stained developing egg strings (ovarioles) with Periodic-Acid-Schiff’s reagent (PAS). Oocyte glycogen content began increasing near the end of oogenesis during stage 12, and grew dramatically up until stage 14 (Figure 1 A). We quantified this glycogen accumulation via colorimetric assay using dissected follicles and confirmed that glycogen rose ~38-fold between stage 10 and 14 (Figure S1 A).

Figure 1. A shift in glycolysis and gluconeogenesis drives glycogen accumulation in oocytes.

A) Periodic acid Schiff’s staining of glycogen in Drosophila follicles. Scale bar = 50 μm. B) GC/MS measurements of select glycolytic intermediates and sugars from stage 10 and stage 14 follicles. C) GC/MS measurements of TCA cycle intermediates from stage 10 and stage 14 follicles. D) GC/MS measurements of the by products of purine breakdown from stage 10 and stage 14 follicles. E) Model of both the glycolysis and TCA pathways. Highlighted in red are the compounds that increased at least 3 fold in stage 14 oocytes. F) Colorimetric glycogen measurements of stage 14 oocytes from pepck^GS/pepck^GS mutant females. G) Colorimetric glycogen measurements of stage 14 oocytes from DNaseII^lo/DNaseII^lo mutant females. H) PAS staining of oocytes from DNaseII^lo/DNaseII^lo mutant females Error bars represent standard deviation. * P<.05 **P <.005.

We investigated the cause of glycogen accumulation by subjecting isolated stage 10 and 14 follicles to gas chromatography/mass spectroscopy (GC/MS) metabolomic analysis. While most compounds, such as free fatty acids (Figure S1B) were unchanged, stage 14 follicles showed a pronounced increase in several saccharides including fructose, ribose, mannose, and galactose (Figure 1B). Additionally, a number of key glycolytic/gluconeogenic intermediates including pyruvate, phosphoglycerate, 3-phophoglycerate, glucose-1-phophate, and glucose-6-phophate increased 3–14 fold in stage 14 follicles (Figure 1 B and E). Lactate, a byproduct of high glycolytic activity, remained at a constant low level between these stages suggesting a dramatic reduction in glycolysis in stage 14 oocytes. Our GC/MS metabolomics also revealed a significant accumulation of a subset of TCA-cycle intermediates including: 2-ketoglutarate, succinate, malate, fumarate, and 2-hydroxyglutarate suggesting an increased amino acid input into the TCA cycle in stage 14 follicles (Figure 1C).

Collectively, the increases in both TCA cycle and glycolytic intermediates, particularly pyruvate, are consistent with a decreased glycolytic input into the TCA cycle and a corresponding increase in amino acid input. To further support this model, we compared the amino acid composition of stage 10 to stage 14 follicles and found that the most abundant amino acid present at stage 10, aspartate, decreased 60% by stage 14 consistent with aspartate’s role as a glucogenic amino acid (Figure S1C). To determine if gluconeogenesis contributes to glycogen accumulation in stage 14 oocytes we assayed glycogen levels in pepck^GS/pepck^GS mutant oocytes, which lack gluconeogenesis, and found that oocyte glycogen levels are significantly reduced (Figure 1F). In addition, using pgdⁿ³⁹,zw¹⁰²⁹ double mutants of the pentose-phosphate pathway, which antagonizes gluconeogenesis by catabolizing glucose-6-phosphate to produce NADPH and ribose, significantly increased oocyte glycogen content supporting a role for gluconeogenesis in germline glycogen accumulation (Figure S1D).

Interestingly, the timing of glycogen accumulation precisely coincides with the period when nurse cell nuclei are digested and the oocyte becomes a closed-metabolic system. These observations suggest that degraded nurse cell contents may provide additional substrates for glucose synthesis and glycogen storage. Consistent with this idea, the purine degradation products ribose, xanthine, and hypoxanthine increased substantially in stage 14 oocytes (Figure 1D). Moreover, when we inhibited the breakdown of nurse cell nuclei using previously characterized mutations in DNAseII (Bass et al., 2009)(Figure 1G) or spinster (Bass et al., 2009)(Figure S1E) stage 14 glycogen content decreased significantly (Figure 1 G, H).

Entry into mitochondrial respiratory quiescence promotes glycogen accumulation.

We examined the distribution of mitochondria throughout oogenesis using two mitochondrial markers, ATP5a and mito-GFP. Mitochondria display perinuclear localization during the early stages of oogenesis (Figure 2 A and E)(Figure S2 A–F and J). However, beginning at stage 10, mitochondria become more widely dispersed (Figure 2C) (fig S2 J). To assess the significance of this mitochondrial dispersal, we measured mitochondrial membrane potential in dissected ovarioles using the mitochondrial membrane potential dye, TMRE. Mitochondria exhibit a strong membrane potential signal beginning early in oogenesis. The signal is strongest in the perinuclear region of the cell where mitochondria are enriched (Fig 2 B, F). Strikingly, mitochondrial membrane potential becomes strongly reduced during stage 10B (Fig 2D, F, G, H)(N>40 follicle/stage) suggesting that dispersal coincides with a decrease in mitochondrial activity.

Figure 2. Mitochondria enter a low energy state of respiratory quiescence to promote glycogen storage in late oogenesis.

A, C, E) Ovaries from Oregon R females fixed and stained with ATP5a antibodies (red) and DAPI (blue). B, D,F) Ovaries from Oregon R females were stained with the mitochondrial membrane potential stain TMRE G) Stage-specific quantification of the % of TMRE Nurse cells from wild type ovarioles. H) quantification of TMRE fluorescence from stage 8 and stage 14 follicle cells and nurse cells. I) Complex I and III compound activity from early follicles, stage 14 oocytes, and 12–16hr embryos. J) Complex II and III compound activity from early follicles, stage 14 oocytes, and 12–16hr embryos. (K) Complex II and III compound activity from early follicles, stage 14 oocytes, and 12–16hr embryos. GC/MS measurements of glycolytic (L) and TCA cycle (M) intermediates from stage 8 and stage 10 follicles. (N) qPCR quantification of mitochondrial genome number from early follicles and stage 14 oocytes. Error bars represent standard deviation.

* P<.05 **P <.005.

We examined mitochondrial activity directly by measuring the specific enzymatic activity of ETC complexes using compound assays for complex I/III and complex II/III. Early follicles (germaria – stage 8) displayed high levels of both Complex I/III and Complex II/III activity (Figure 2 I, J). A significant amount of this activity was inhibited by rotenone (complex I inhibitor) or by malonate (complex II inhibitor), respectively. Both complex I/III and complex II/III activity decreased dramatically in stage 14 follicles, despite the fact that the amount of mitochondrial DNA increased two fold between stage 10 and stage 14 (Figure. 2N). Furthermore, virtually all of the rotenone-sensitive or malonate-sensitive activity was lost. The low levels of complex I/III and complex II/III activity in stage 14 oocytes are consistent with our previous observation that mitochondrial membrane potential is lost. In contrast, complex IV activity remained constant in stage 14 oocytes suggesting complexes I, II, &III are specifically down regulated during the final stages of oogenesis (Figure 2 K).

We examined developing embryos to learn when and where mitochondrial activity is restored after fertilization. Mitochondrial membrane potential was re-established during embryonic cellularization (Figure S2G), dorsal closure (Figure S2H), and in developing muscle (Figure S2I). Moreover, complex I/III and complex II/III enzymatic activity both recovered during embryogenesis (Figure 2 I, J).

If decreased mitochondrial activity is a primary cause of glycogen accumulation, then glycolytic intermediates should begin to increase in stage 10 oocytes. Consistent with our model, GC/MS metabolomics of stage 8 and stage 10 follicles, revealed that many glycolytic (Figure 2L) and TCA cycle intermediates (Figure 2M) that were high in stage 14, had already begun to increase during stage 10. Thus, oocyte mitochondrial respiratory quiescence is likely to promote glycogen accumulation.

Mitochondria enter respiratory quiescence due to ETC remodeling

In mitochondria from early-stage follicles (germaria – stage 8) we observed robust bands corresponding to assembled respiratory chain complexes I, III, IV, and V via blue native gel electrophoresis (Figure 3A). Strikingly, complexes I and V were significantly decreased in mitochondria from purified stage 14 oocytes, while complex IV was unchanged (Figure 3A). Complex I and V reappeared during embryogenesis (Figure 3B), consistent with the return of mitochondrial activity after fertilization. We quantified these effects by calculating the ratio of complexes I:IV (Figure 3C), III:IV (Figure S3A), and V:IV (Figure 3D). The ratios of complexes I:IV and V:IV both decrease in stage 14 oocytes only to increase again during embryogenesis. Using native western blots for ATP5a (Figure 3E), we observed a reduction in the amount of assembled complex V and a corresponding increase in lower molecular weight complexes during late oogenesis.

Figure 3. ETC remodeling drives mitochondrial respiratory quiescence.

A) Blue native PAGE of mitochondria from early-stage follicles and stage 14 oocytes. B) Blue native PAGE of isolated mitochondria from stage 14 oocytes and 12–16 hr embryos. C) Quantification of the ratio of complex I /complex IV from early follicles, stage 14 oocytes, and 12–16hr embryos. D) Quantification of the ratio of complex V /complex IV from early follicles, stage 14 oocytes, and 12–16 hr embryos. E) Native western blot of ATP5a from early-stage follicles, stage 14 oocytes, and 12–16 hr embryos. F) Whole oocyte western blots of select mitochondrial proteins: ATP5a, NDUF3, COX IV, CYTOCHROME C, PORIN and the loading controls ACTIN and TUBULIN from early-stage follicles, stage 14 oocytes, and 12–16 hr embryo samples. G) Western blots of isolated mitochondria from early-stage follicles, stage 14 oocytes, and 12–16 hr embryo samples. H) Electron microscope images of mitochondria from stage 8, stage 14 oocytes, and embryos. Error bars represent standard deviation. * P<.05 **P <.005.

To determine the cause of the decline in assembled complexes I and V at the end of oogenesis we performed western blot analysis for NDUFS3 (complex I), ATP5a (complex V), CYTOCHROME C, PORIN, COXIV, as well as ACTIN and TUBULIN as loading controls (Figure 3F). We found the levels of these proteins did not change in lysates from stage 14 oocytes (Figure S3E). Western blots of purified mitochondria from stage 14 oocytes, however, did show a dramatic reduction in NDUFS3 and PORIN (VDAC1) levels (Figure. 3G) (Figure S3 F) indicating that these proteins are no longer present within mitochondria. Western blots of cytosolic fractions showed an increase in NDUFS3 levels in stage 14 oocytes (Figure S3C). SDS Coomassie gel electrophoresis of purified oocyte mitochondria showed that many proteins decrease in stage 14 (Figure S3B), suggesting that many mitochondrial proteins in addition to NDUFS3 and PORIN are excluded from the organelle during quiescence. These data suggest that during quiescence, respiratory components remain disassembled and are stored in the oocyte.

When we examined mitochondrial structure by electron microscopy we observed that oocyte mitochondria appear smaller and less differentiated during stage 14 than they do during stage 8 or later during embryogenesis. Furthermore, mitochondria from stage 14 oocytes lack the pronounced cristae seen in stage 8 oocytes, further supporting a global shift in mitochondrial function and remodeling of the ETC at the end of oogenesis (Figure 3H)(Figure S3D).

Insulin controls the timing of respiratory quiescence and glycogen accumulation.

To determine if insulin signaling is required for glycogen accumulation, we used the germline-specific nos-GAL4VP16 driver to express UAS-RNAi transgenes against either the insulin receptor (InR) or the insulin effector gene Akt. In both cases, inhibition of the insulin signaling pathway early in oogenesis caused a severe developmental arrest consistent with previous studies (LaFever and Drummond-Barbosa, 2005) (Drummond-Barbosa and Spradling, 2001). Moreover, inactivation of insulin signaling caused premature glycogen accumulation in arrested follicles at stage 6 and 7 (Figure 4 A&C) (Figure S4A). When we examined mitochondrial membrane potential in ovarioles from nos–InR-RNAi(Figure 4 B &D) or nos–Akt-RNAi (Figure S4B) animals we found that mitochondria depolarize prematurely in the germline under these conditions (Figure 4D). Moreover, we discovered that lowering insulin secretion by starving the animal causes germline mitochondria to depolarize very early in oocyte development (Figure S4C and D). Re-feeding the animals for 12–24 hours partially reversed starvation-induced depolarization (Figure S4D). When insulin signaling was mildly inhibited in the germline, the animals produced stage 14 oocytes that contain substantially higher glycogen levels (Figure 4E) and corresponding decreases in oocyte triglyceride levels (Figure 4F). This would be expected if under these conditions mitochondria become quiescent prematurely and glycogen accumulates over a longer period of time.

Figure 4. Insulin regulates the timing of glycogen accumulation and mitochondrial depolarization.

A) Periodic acid Schiff’s staining (glycogen) of nos-GAL4/+ control and nos->InR-RNAi ovarioles. B) TMRE staining and DIC images of nos-GAL4/+ control and nos->InR-RNAi ovarioles. C) Depicts the % of ovarioles that display premature glycogen accumulation from nos-GAL4/+ control, nos–Akt-RNA, nos–InR-RNA#1, and nos–InR-RNA#2 females. D) Depicts the percentage of ovarioles that displayed a high level of mitochondrial membrane potential in nos-GAL4/+ control, nos–Akt-RNA, nos–InR-RNA#1, and nos–InR-RNA#2 ovarioles. E) Normalized glycogen levels measured by colorimetric assay from nos-GAL4/+ control and nos–InR-DN stage 14 oocytes. F) Normalized triglyceride levels measured by colorimetric assay from nos-GAL4/+ control and nos–InR-DN stage 14 oocytes. G) p-AKT and AKT westerns from early-stage follicles, stage 14 oocytes, and 12–16 hr embryo samples. (H) Quantification of p-AKT/AKT ratios from early stage follicles, stage 14 oocytes, and embryos. Error bars represent standard deviation. * P<.05 **P <.005.

We examined the dynamics of insulin signaling activity during oogenesis by assessing phoso-AKT levels in early follicles and stage 14 oocytes. AKT is active during early-stages of oogenesis but was barely detected in stage 14 oocytes (Figure 4G, H). Collectively these data indicate that insulin signaling is significantly reduced in oocytes during late oogenesis, and this leads to mitochondrial depolarization, respiratory quiescence, and to glycogen accumulation.

The AKT target genes foxo and gsk3/(sgg) are required for glycogen accumulation and embryonic development.

To identify AKT target proteins that promote mitochondrial quiescence we investigated the transcription factor FOXO. Foxo is an insulin effector gene that regulates growth and metabolism (Arden, 2004; Barthel et al., 2005; Neufeld, 2003) (Junger et al., 2003)(Ogg et al., 1997) (Burgering and Kops, 2002; Paradis and Ruvkun, 1998; Stitt et al., 2004) (Burgering and Kops, 2002; Stitt et al., 2004). Stage 14 oocytes from foxo²¹/foxo²⁵ mutants displayed a 50–60% decrease in oocyte glycogen content relative to heterozygous mutant controls (Figure 5A)(Figure S5C), while triglyceride content remained unaffected (Figure 5B). Consistent with these data, inhibiting foxo in the germline using two different UAS-foxo-RNAi transgenes yielded a similar reduction in oocyte glycogen content (Figure 5A) (Figure S5B). Additionally, foxo²¹/foxo²⁵ mutant females displayed significant defects in oocyte production (Figure S5A).

Figure 5. Akt target genes are required for glycogen storage in stage 14 oocytes.

A) Glycogen levels were measured by colorimetric assay and normalized to total protein from the indicated genotypes. The results are displayed as a percentage of the wild type control of each experiment. B) Colorimetric triglyceride measurements of stage 14 oocytes from foxo²¹/foxo²⁵ mutant females. Glycogen (C) and triglyceride (D) levels were measured by colorimetric assay and normalized to total protein from nos-GAL4/+, UAS-gsk3-RNAi/+, and nos–gsk3-RNAi stage 14 oocytes. E) Percent hatching of embryos from nos-GAL4/+, UAS-gsk3-RNAi/+, and nos–gsk3-RNAi stage 14 females. Error bars represent 1xSD. * P<.05 **P <.005.

The serine/threonine kinase GSK3 is another highly conserved target of AKT that regulates protein turnover, systemic metabolism, growth control, proliferation and other processes (Cohen and Frame, 2001)(Cross et al., 1995; Papadopoulou et al., 2004). We knocked down gsk3 function in the germline using the nos-Gal4VP16 driver and found that, relative to UAS-gsk3-RNAi/+ or the nos-Gal4/+ controls, stage 14 oocytes from nos–gsk3-RNAi females contained substantially less glycogen (Figure 5C)(Figure S5B) and triglycerides (Figure 5D). These reductions were not due to altered oocyte growth since total protein levels in nos–gsk3-RNAi oocytes were unaffected (Figure S5E). Consistent with previous studies (Perrimon et al., 1984), embryos from nos-gsk3-RNAi animals were inviable (Figure 5E). Thus, the low glycogen and triglyceride content of oocytes lacking functional GSK3 may cause the maternal-effect lethality and meiotic abnormalities previously seen in gsk3 Drosophila germline mutants (Perrimon et al., 1984; Takeo et al., 2012).

GSK3 promotes mitochondrial quiescence and ETC remodeling.

To determine if GSK3 regulates mitochondrial respiratory quiescence, we dissected ovarioles from nos-GAL4/+ and nos–gsk3-RNAi females and stained them with TMRE. Control stage 10 follicles showed no mitochondrial membrane potential in the germ cells. Stage 10 oocytes dissected from nos–gsk3-RNAi animals, however, displayed a high level of mitochondrial membrane potential (Figure 6A, B), suggesting that these mitochondria are still active. Interestingly, when we starved gsk3-RNAi females for 24hrs we found that a significant number (~40–50%) maintain high levels of TMRE fluorescence in early stage germ cells whereas TMRE fluorescence was consistently reduced in starved early stage control follicles (Figure 6C)(Figure S4C). To determine if GSK3 regulates mitochondrial oxidative metabolism, we measured mitochondrial ETC activity in animals where gsk3 expression had been silenced in the germ line using RNAi. Complex I/III activity in nos-GAL4/+ control stage 14 oocytes was almost undetectable, whereas nos–gsk3-RNAi oocytes showed significant mitochondrial activity, which was readily suppressed by the complex I inhibitor, rotenone (Figure S6A). Similarly nos–gsk3-RNAi oocytes showed elevated complex II/III activity, which was readily suppressed by the complex II inhibitor, malonate (Figure S6B), further suggesting that GSK3 controls mitochondrial respiratory quiescence. In contrast, the mitochondrial membrane potential shut down normally during oogenesis in germ cells from foxo²¹/foxo²⁵ mutant females (Figure S5F).

Figure 6. GSK3 promotes mitochondria respiratory quiescence.

A) TMRE staining and DIC images of nos-GAL4/+ control and nos–GSK3-RNAi stage 10B follicles. B) Quantification the percentage of stage 10B follicles that exhibit high nurse cell mitochondrial membrane potential in control and gsk3-RNAI follicles. C) Mitochondrial membrane potential staining (TMRE) from: control fed, control starved, and gsk3-RNAi starved ovarioles. D) Blue native PAGE of mitochondria from nos-GAL4/+ and nos–gsk3-RNAi stage 14 oocytes and dissected flight muscle (FM). E) Quantification of the ratio of complex I/complex IV from nos-GAL4/+ and nos–gsk3-RNAi stage 14 oocytes. (F) Venn diagram of processes involving proteins found to be decrease in mitochondria from stage 14 oocytes. Each process is proportional to the number of effected proteins. (G) Venn diagram processes involving proteins found to be stabilized in mitochondria from stage 14 GSK3-RNAi oocytes. Error bars represent standard deviation. * P<.05 **P <.005.

We further examined whether gsk3 is required for ETC remodeling by purifying mitochondria from nos-GAL4/+ control and nos–gsk3-RNAi stage 14 oocytes and subjecting the samples to native gel electrophoresis. Consistent with our mitochondrial activity data, mitochondria from gsk3-RNAi oocytes retained a higher level of assembled Complexes I and V (Figure 6D see asterisk). Quantitation showed that the ratios of both complex I:IV (Figure 6E) and V:IV (Figure S6D) were substantially increased, however the complex III:IV ratio was unaffected (Figure S6E). Interestingly, stage 14 oocytes from nos–GSK3-RNAi animals showed ratios of these complexes similar to early stage follicles, suggesting that GSK3 functions during the late stages of oogenesis to promote ETC remodeling. Moreover, when we conducted western blots on mitochondria isolated from gsk3-RNAi oocytes we found that NDUFS3 was stabilized in the mitochondria upon gsk3 inhibition (Figure S6 F). To rule out a contribution from foxo, we assessed the ETC distribution in stage 14 oocyte mitochondrial fractions from foxo²¹/foxo²⁵ mutant oocytes (Figure S5D) and found no defect in ETC remodeling.

To identify candidate targets of GSK3 that may be involved in ETC remodeling and mitochondrial quiescence we made mitochondrial extracts from early stage follicles, stage 14 oocytes, and gsk3-RNAi oocytes. Proteins were identified by LC/MS/MS proteomics, yielding about 250–300 detectable proteins per sample. We found that 60–70% of the 150 most abundant proteins identified in stage 8 samples were known mitochondrial proteins while 25–30% were known ER and Golgi proteins. Interestingly, we found that 61 of these mitochondrial proteins assayed were reduced at least 3-fold in stage 14 mitochondria (Figure S6G). Moreover, 33 of these 62 proteins were completely absent in stage 14 samples (Figure 6F), including 20 that are factors associated with the electron transport chain. These include several components of ETC complex I and complex V, complexes that are reduced in stage 14 mitochondria via western blots and native western blots. Additionally, 10 proteins associated with the TCA cycle were also absent in stage 14 mitochondria. Ten proteins involved with various aspects of amino acid metabolism in the inner mitochondria matrix were also found to be decreased in stage 14 mitochondria. Interestingly, we found that 6 proteins involved with protein synthesis and transport into the mitochondria were also decreased in stage 14 mitochondria suggesting that global mitochondrial synthesis and uptake may be suppressed in stage 14 mitochondria.

To identify candidate targets of GSK3 that are involved with mitochondrial respiratory quiescence we looked for proteins that were stabilized in gsk3-RNAi mitochondria. 28 of the 61 proteins that are decreased in stage 14 mitochondria were stabilized in mitochondria from gsk3-RNAi oocytes. These candidate GSK3 targets include proteins involved with the ETC, TCA cycle, amino acid metabolism and mitochondrial protein transport and synthesis (Figure 6G). Strikingly, 5 of 6 proteins involved with mitochondrial protein synthesis and import that were down regulated in stage 14 mitochondria were stabilized in gsk3-RNAi oocytes suggesting that GSK3 may mediate its role in mitochondrial respiratory quiescence by causing a remodeling of the mitochondrial proteome.

Insulin/AKT regulation of mitochondrial ETC remodeling is conserved in vertebrate oocytes.

Using Xenopus we investigated the evolutionary-conservation of glycogen accumulation and ETC remodeling in vertebrate oocytes. All experiment using Xenopus were done with approval of an institutional animal care review board. Mature stage 6 Xenopus oocyte contain much higher levels of glycogen than stage 3 oocytes (Figure 7A). When we purified mitochondria we found that stage 3 Xenopus oocytes displayed a distribution of ETC complexes similar to that seen in early Drosophila follicles (Figure 7B). Consistent with what happens in late-stage Drosophila oocytes, mitochondria isolated from stage 6 Xenopus oocytes also showed ETC remodeling. However, stage 6 Xenopus oocytes displayed a different distribution of ETC complexes than that observed in Drosophila mature oocytes (Figure S7 A–C). This difference in mitochondrial ETC remodeling may stem from the fact that, similar to mature mammalian oocytes, mature Drosophila oocytes are both transcriptionally and translationally quiescent, whereas Xenopus oocytes remain active as stage 6 oocytes. Thus, Xenopus oocytes may still require ATP and may remodel their ETC in a manner that allows for sufficient ATP production but also promotes glycogen storage.

Figure 7. Insulin regulation of ETC remodeling is conserved in vertebrate oocytes.

A) Total glycogen levels measured by colorimetric assay from stage 3 and stage 6 Xenopus oocytes. Data are expressed as fold change relative to stage 3 glycogen levels. Error bars represent 1xSD. * P<.05 **P <.005. B) Blue native PAGE of mitochondria from stage 3 and stage 6 Xenopus oocytes. C) Blue native PAGE of mitochondria from vehicle-treated and 10uM wortmannin (PI3-Kinase inhibitor)-treated stage 3 Xenopus oocytes. (D) Western blots assaying p-AKT and AKT levels in stage 3 and stage 6 Xenopus oocytes. E) Quantification of the ratio of p-Akt/Akt calculated from western blots of stage 3 and stage 6 Xenopus oocytes. F) Model of germline mitochondrial quiescence and glycogen accumulation.

We hypothesized that insulin/AKT signaling has a conserved role in controlling glycogen accumulation and ETC remodeling in vertebrate oocytes. Consistent with this idea we found that PI3-kinase inhibition (using wortmannin) in Xenopus stage 3 oocytes was sufficient to induce remodeling of the ETC (Figure 7C). Intriguingly, we observed that PI3-kinase inhibition stimulated a remodeling of the ETC that resembled the ETC distribution in stage 14 Drosophila oocytes. When we quantified these results by calculating the ratios of complexes I:IV (Figure S7D), III:IV (Figure S7E), and V:IV (Figure S7F) we found that PI3-kinase inhibition induced ETC ratios very similar to those we observed in Drosophila stage 14 oocytes. Consistent with the high transcriptional and translation activity of stage 6 Xenopus oocytes we found that AKT remained active in stage 6 oocytes (Figure 7D,E).

The differences we observed in ETC remodeling of quiescent Drosophila oocytes and, the more active, Xenopus oocytes suggest that the degree and distribution of ETC remodeling correspond to the metabolic demands of the cell. In agreement with this hypothesis, the distributions of ETC complexes in mitochondria isolated from Drosophila flight muscle (high mitochondrial activity) and intestine (lower metabolic activity) were found to be very different (Figure S7I). Moreover we observed substantially higher ratios of complex I:IV (Figure S7G) and V:IV (Figure S7H) in flight muscle, consistent with its well-characterized high metabolic capacity.

Discussion

Mitochondria enter a reversible low respiratory state that helps establish and maintain oocyte quiescence

Our data describe a dramatic shift in mitochondrial metabolism during late oogenesis. This shift is caused by mitochondria entering into a low activity state that promotes glycogen storage and helps establish and maintain oocyte quiescence prior to fertilization. Mitochondria enter this metabolic state by remodeling the ETC during the late stages of oogenesis, a process that is conserved in Xenopus oocytes. Reducing the amounts of assembled complexes I and V likely decreases intracellular redox potential and oxidative damage during oocyte quiescence. This is consistent with studies that show complexes I and V are major sources of reactive oxygen species production (Hirst et al., 2008; Sanz et al., 2010; Scheibye-Knudsen et al., 2015). Moreover, our findings define how oocytes acquire and maintain a low energy state while stored prior to fertilization. (Dumollard et al., 2007) (Van Blerkom, 2011).

Our studies indicate that the decreased mitochondrial oxidative metabolism associated with ETC remodeling is likely to cause an accumulation of glycolytic and TCA cycle intermediates and promote glycogen storage. Furthermore, our data indicate that this shift in metabolism is coupled with nurse cell degradation. Our data suggests that the byproducts (nucleotides and glucogenic amino acids) from the degrading nurse cells likely provide the substrates used to generate and store glycogen. Taken together these data indicate that a shift in the balance of glycolysis and gluconeogenesis drives the accumulation of glycogen in mature oocytes.

We have found that Drosophila oocytes acquire a low mitochondrial activity state through insulin-dependent remodeling of the ETC, which maintains nutrient storage and likely prevents oxidative damage prior to fertilization. These conclusions were supported by experiments that show gsk3-RNAi oocytes display persistent mitochondrial activity, decreased nutrient storage, and result in a strong maternal effect lethal phenotype in the resulting embryo. Thus, the absence of mitochondrial quiescence in stage 14 oocytes impairs the embryos’ competency to complete development. Collectively, by reducing mitochondrial activity and entering into a state of mitochondrial quiescence, stage 14 oocytes reduce the amount glycolytic consumption by the mitochondria leading to a build-up in glycolytic/gluconeogenesis intermediates that drive glycogen storage. The increase in stored glycogen then functions as a fuel reserve to support embryogenesis (Tennessen et al., 2014b).

ETC remodeling may be a widespread mechanism of mitochondrial regulation

While mitochondrial oxidative metabolism can vary dramatically between tissue types and in response to physiological stimuli, much of the dynamic nature of mitochondria has been thought to be controlled by the biochemical pathways that feed the TCA cycle (Heineman and Balaban, 1990; Mela-Riker and Bukoski, 1985) (Brown, 1992; Millar et al., 2011). In contrast, our data indicate that a programed loss of membrane potential and remodeling of the ETC causes mitochondrial respiratory quiescence in Drosophila oocytes. ETC remodeling may occur during the development of many tissues and play an important role in setting their mature metabolic capacity. For instance, during proliferative stages of development aerobic glycolysis is thought to be high (Tennessen et al., 2014b) (Fraga et al., 2013), and oxidative metabolism is low (Trimarchi et al., 2000) (Magnusson et al., 1986), to promote the synthesis of the building blocks required for growth. However as tissues differentiate ETC remodeling may lead to a greater oxidative potential, as it does in the transition from oocyte to embryo, and be one of the driving forces that helps set the balance between glycolysis and mitochondrial oxidative metabolism that is required for mature tissue function and homeostasis.

Insulin signaling and GSK3 control mitochondrial activity

Our experiments indicate that insulin functions in the germline to promote mitochondrial oxidative metabolism and promote growth. We found that insulin pathway activity represses mitochondrial quiescence and glycogen accumulation by inhibiting the AKT target gene GSK3. When insulin activity decreases in late oocytes, GSK3 functions to stimulate mitochondrial quiescence and prepare the oocyte for fertilization. Previous studies defined key roles for insulin signaling in controlling the division rate of germline stem cells (Drummond-Barbosa and Spradling, 2001) (Hsu and Drummond-Barbosa, 2011; LaFever and Drummond-Barbosa, 2005), the progression of early cyst development in the germarium (Drummond-Barbosa and Spradling, 2001), and the arrest of follicles at the stage 8 checkpoint in conjunction with ecdysone (Buszczak et al., 2002; LaFever and Drummond-Barbosa, 2005). Our data suggest that maintaining mitochondrial activity and promoting oxidative metabolism underlies many of these previously recognized functions of insulin signaling during germline development.

We have found that decreased AKT activity is the trigger for mitochondrial respiratory quiescence. This down regulation during stage 10 may be due to repression of PI3 kinase activity by the ecdysone signaling pathway, which has well-established links to insulin regulation. Interestingly, the formation of the vitelline membrane and eggshell deposition during this time may also contribute to the loss of AKT activity. The formation of these physical barriers both blocks nutrient uptake and also would make the oocyte unable to sense secreted factors such as insulin.

Studies in Drosophila indicate that insulin pathway mutant oocytes arrest and die at defined cell death checkpoints in the germaria and during stage 8 (LaFever and Drummond-Barbosa, 2005). However, once oocytes progress past this stage 8 checkpoint cell death is not observed consistent with the quiescent nature of the oocyte. In mammalian cells, a hallmark of apoptosis is cytochrome c release from mitochondria (Xiong et al. 2014). Intriguingly, previous in vitro studies have suggested that GSK3 binds to outer mitochondrial membrane proteins such as porin/VDAC1 and regulates mitochondrial membrane potential and permeability(Tanno et al., 2014). Furthermore, inhibition of GSK3 is sufficient to block cell death in vitro (Maurer et al., 2006). We observed that multiple proteins, including NDUFS3 and Porin/VDAC1, are excluded from the mitochondria following GSK3 activation during oocyte maturation suggesting GSK3’s action on mitochondria during oocyte maturation may also cause changes in porin and protein shuttling. However, rather than going into apoptosis, mature oocytes become quiescent and can be efficiently reactivated following ovulation. Our mitochondrial proteomics analysis data suggest that GSK3 controls mitochondrial function in by regulating of mitochondrial protein content possibly through suppression of mitochondrial protein import, turnover, and synthesis. Furthermore, we have identified 28 candidate mitochondrial targets of GSK3 some of which are thought to be present in the outer mitochondrial membrane suggesting that GSK3 may have several direct targets that are important for ETC remodeling.

Mitochondrial respiratory quiescence and mammalian oocyte development

Studies in both Drosophila and mammals have shown that mature oocytes enter a quiescent state where transcription has ceased and translation is suppressed. We have found that, during Drosophila oocyte maturation, Akt activation is reduced and mitochondria remodel their ETC and enter into a low activity state that allows for nutrient storage and may support the quiescent state of the oocyte. Based on our data, a reduction in Akt activation and remodeling of the ETC may promote metabolic quiescence in mammalian antral follicles as they are stored prior to ovulation, similar to what we observe in Drosophila. In addition, these processes may promote the quiescent state of young mammalian primordial follicles promote reproductive longevity. Consistent with our findings, numerous studies have shown that mitochondrial activity is very low in mature mammalian oocytes and early embryos, however as the embryo reaches the blastocyst stage mitochondrial activity increases dramatically (Trimarchi et al., 2000) (Dumollard et al., 2007) (Van Blerkom, 2011). Moreover, recent studies that have shown that major events of ovulation and egg activation are conserved suggesting that many aspects of late oogenesis arose early on in evolution (Sun and Spradling, 2013) (Deady et al., 2015). As a result it is possible that that both the entry and exit from mitochondrial respiratory quiescence are highly conserved aspects of sexual reproduction.

Collectively, our studies of mitochondrial quiescence and ETC remodeling may be applicable to understanding human infertility. PCOS is strongly associated with diabetes and insulin resistance suggesting that the defects in oocyte maturation seen in this disease arise from abnormal ETC remodeling and mitochondrial activity as a result of defective insulin signaling. Second, embryos generated by in vitro fertilization (IVF) treatment frequently display developmental abnormalities (Barri et al., 2014) (Simon and Laufer, 2012). Defective development may be due to abnormalities in mitochondrial respiratory quiescence and mitochondrial reactivation in oocytes subjected to in vitro maturation prior to use in IVF. These mitochondrial abnormalities may also explain the low developmental capacity and abnormal morphology of in vitro fertilized oocytes from sub-fertile women (Van Blerkom, 2011). Overall, our studies of mitochondrial respiratory quiescence have established a molecular link between nutritional health and oocyte development that may underlie many reproductive disorders.

Methods

Fly stocks and media

The following Drosophila stocks were used in this study: Oregon R (Bloomington stock center# 25211), nos-GAL4VP16 (Van Doren et al., 1998), foxo²¹/TM6B (Junger et al., 2003), foxo²⁵/TM6B (Junger et al., 2003), y¹, sc*, v¹, UAS-InR-RNAi(GL00139) (Ni et al., 2008), y¹, sc*, v¹, UAS-InR-RNAi(HMS03166) (Ni et al., 2008), y¹, sc*, v¹, UAS-Akt-RNAi(HMS00007) (Ni et al., 2008), y¹, sc*, v¹, UAS-gsk3 -RNAi(GL00277)(Ni et al., 2008), w¹¹¹⁸, UAS-InR(K1409A)dominant negative (Ikeya et al., 2009), y¹w*, pepck^GS22646/SM1 (Kyoto stock center #204216), and Pgd^N39, Zw^lo2a(Bloomington stock center #6033)

Flies were grown on a standard cornmeal, molasses, and yeast media (Bloomington stock center). Adult flies were matured on media supplemented with fresh yeast paste for at least 3 days.

Metabolite measurements

For both triglyceride and glycogen measurements females of the desired genotypes were dissected and 200 oocytes of the corresponding stage were collected. The samples were then homogenized in 120 μl of 1xPBST(0.1% triton) and heat-treated (3 minutes @100°C). The resulting heat-treated lysate was then cleared by centrifugation (2500xg for 3 minutes). Glycogen was then assayed using the glucose oxidase kit (Sigma, cat.# GAGO20–1kt) and amyloglucosidase (Sigma, cat.# 1602) as described in (Tennessen et al., 2014a) (Sieber and Thummel, 2009). Triglyceride levels were measured using triglyceride reagent (Sigma, cat. #T2449) and free glycerol reagent(Sigma, cat.# F6428) as described in (Tennessen et al., 2014a) (Sieber and Thummel, 2012) (Sieber and Thummel, 2009). Protein levels were assayed by using Bio-Rad protein assay reagent (cat.# 500–0006). All glycogen and triglyceride measurements were then normalized to total protein. All data presented is derived from at least 5 replicate samples and each experiment was repeated at least 3 times.

TMRE staining

Dissect ovarioles were incubated in 1xPBS with10 nM TMRE for 5 mins. Oocytes were then washed 4 times in 1xPBS and mounted on slides in 1xPBS. The oocytes were then imaged immediately using fluorescence and differential interference contrast (DIC) on a Leica SP5 confocal microscope.