Comparative analysis of vertebrates reveals that mouse primordial oocytes do not contain a Balbiani body

The earliest stage of a recognizable follicle in the ovary is the primordial follicle, which contains a primordial oocyte. The ovarian reserve consists of primordial oocytes, which are considered dormant as they do not grow or divide (Reddy et al., 2010). They can remain in the ovary for long periods of time, ranging from several weeks in mice to several decades in humans (Flurkey et al., 2007; Wallace and Kelsey, 2010). Each individual has a pool of thousands of primordial oocytes from which only a few are activated to grow at any given time. Upon sexual maturity, some of the growing oocytes mature to produce fertilizable eggs (Grive and Freiman, 2015; Rimon-Dahari et al., 2016).

The most characteristic morphological feature of primordial oocytes of many species is the Balbiani body. The Balbiani body has also previously been referred to as the mitochondrial cloud (Guraya, 1979), and is a non-membrane-bound super-organelle consisting mostly of mitochondria but also Golgi, an endoplasmic reticulum (ER), other vesicles and RNA (Boke et al., 2016; Cox and Spradling, 2003; Hertig, 1968; Kloc et al., 2004; Banani et al, 2017). The Balbiani body is only present in early oocytes and dissociates upon oocyte activation. Therefore, it is closely associated with oocyte dormancy. In lower vertebrates, such as Xenopus and zebrafish, the Balbiani body is held together by an amyloid-like matrix formed by an intrinsically disordered protein (Boke et al., 2016; Krishnakumar et al., 2018), and is essential for the determination of the germline through inheritance of the germ plasm (Jamieson-Lucy and Mullins, 2019; Kloc et al., 2004). Although the function of the Balbiani body in mammals remains elusive, it is proposed to protect mitochondria in the germline (Colnaghi et al., 2021; Jamieson-Lucy and Mullins, 2019; Kloc et al., 2004). This enigmatic super-organelle has been observed in, among others, humans (Hertig and Adams, 1967), chimpanzees (Barton and Hertig, 1972), cats (Amselgruber, 1983), frogs (Al-Mukhtar and Webb, 1971; Boke et al., 2016) and zebrafish (Marlow and Mullins, 2008). Recent research has suggested that mouse primordial oocytes also contain a Balbiani body (Lei and Spradling, 2016; Pepling et al., 2007).

As virtually all of the organelles and cytoplasm of the zygote and hence, the new embryo, are derived from the oocyte, maintenance of oocyte health is imperative for producing healthy offspring (Cafe et al., 2021; Goodman et al., 2020; van der Reest et al., 2021). Although there is growing knowledge on how oocytes activate and interact with their somatic environment (Clarke, 2018; Handel et al., 2014; Li and Albertini, 2013; Matzuk et al., 2002), and how they segregate their chromosomes (Holubcová et al., 2015; Pfender et al., 2015), we know little about the cellular biology of dormant oocytes. Here, for the first time, we characterize and compare the cytoplasmic features of Xenopus, mouse and human primordial oocytes to study their cytoplasmic organization and organelle activity. We find that primordial oocytes of all three species contain active lysosomes, mitochondria and Golgi. In mouse oocytes, unlike human and Xenopus, mitochondria are not clustered within a Balbiani body. Furthermore, the conglomeration of Golgi stacks in mouse, which was previously used as a marker for the Balbiani body, is neither associated with RNA-binding proteins (RBPs) nor functionally connected with oocyte dormancy. Therefore, we provide strong evidence that mouse primordial oocytes, unlike human and Xenopus, do not contain a Balbiani body, thereby highlighting the similarities and differences between different model systems that are used to study oocyte dormancy.

Live characterization of cells can reveal features that are lost after fixation, such as organelle dynamics and activity. However, such studies do not yet exist for primordial oocytes. We began our study by isolating follicle-enclosed oocytes from mouse, human and Xenopus ovaries for live imaging (Fig. S1A). The granulosa cells, which form the somatic component of the follicle, were used as internal controls for comparing oocytes to somatic cells (Fig. 1A-C,G).

We then probed the oocytes with fluorescent markers for mitochondria, lysosomes and Golgi, and imaged them live to investigate the dynamics and activity of these organelles in the oocyte cytoplasm. We chose these three organelles for their fundamental roles in oocyte activation, growth and aging (Cafe et al., 2021; Goodman et al., 2020; van der Reest et al., 2021).

First, we imaged lysosomes with LysoTracker, which is a membrane-permeable dye that preferentially accumulates in acidic, and thus active, lysosomes (Zhang, 1994). Previous studies suggested that lysosomes are inactive in early C. elegans oocytes (Bohnert and Kenyon, 2017; Samaddar et al., 2021). In contrast, lysosomes of all three vertebrate primordial oocytes accumulated LysoTracker (Fig. 1A,B; Fig. S1B). This labelling was lost upon incubation of oocytes with bafilomycin A1, an inhibitor of lysosomal acidification (Bowman et al., 1988), confirming the specificity of the LysoTracker dye (Fig. S1B). LysoTracker intensity was similar between somatic cells and primordial oocytes in all three species (Fig. 1A,B), as well as between primordial and growing oocytes [germinal vesicles (GVs)] in mouse (Fig. S1C,D). Thus, we show for the first time that vertebrate primordial oocytes have active lysosomes distributed in their cytoplasm.

Next, we used nitrobenzoxadiazole (NBD) C₆-ceramide, a fluorescent sphingolipid derivative, to image Golgi in live oocytes. NBD C₆-ceramide is taken up by cells and transported to the Golgi, where it is metabolized and localized to the late Golgi cisternae (Lipsky and Pagano, 1985; Pagano et al., 1989). Xenopus and human oocytes displayed a distributed punctate pattern of Golgi in their cytoplasm, whereas NBD C₆-ceramide showed a specific pattern previously described as a Golgi conglomerate in mouse oocytes (Fig. 1C; Movie 1) (Wischnitzer, 1970). The Golgi conglomerate (herein referred to as Golgi ring) was not present in growing mouse oocytes (Fig. S1E,F). We found that the Golgi ring had polarized Golgi stacks as it contained cis- and trans-Golgi markers [GM130 (also known as GOLGA2) and TGN46 (also known as TGOLN2), respectively], and confirmed previous reports that it is associated with pericentrin (Lei and Spradling, 2016) (Fig. S1F-H), similar to Golgi in somatic cells (Klumperman, 2011). Thus, we conclude that the Golgi ring has structural features of a conventional Golgi.

We investigated whether the Golgi is capable of membrane trafficking and, thus, active in primordial oocytes. For this, we used brefeldin A (BFA), one of the most specific compounds that acts on the Golgi to induce Golgi disassembly (Chardin and McCormick, 1999). BFA induces extensive tubulation of active Golgi cisternae and its ultimate fusion with ER membranes (Lippincott-Schwartz et al., 1990, 1989). Hence, an active Golgi is disassembled by BFA action, whereas a Golgi with impaired trafficking remains unaffected (Duran et al., 2012; Lippincott-Schwartz et al., 1990; Tan et al., 1992). We isolated primordial oocytes from mouse and Xenopus ovaries, treated them with BFA, and labelled their Golgi with NBD C₆-ceramide for live imaging. BFA treatment led to the dissociation of the Golgi in mouse and Xenopus oocytes (Fig. 1D-F). We thus conclude that the Golgi in oocytes is capable of membrane trafficking and its structure is actively maintained.

Finally, we imaged mitochondria in oocytes using tetramethylrhodamine ethyl ester (TMRE), a cell-permeant fluorescent dye that accumulates in active mitochondria dependent on mitochondrial membrane potential (Ehrenberg et al., 1988). All three vertebrate oocytes had detectable mitochondrial membrane potential, as judged by TMRE labelling (Fig. 1G). Treatment of oocytes with carbonyl cyanide m-chlorophenyl hydrazone (CCCP), an ionophore that dissipates the mitochondrial membrane potential (Heytler, 1963), led to the loss of TMRE, confirming its specificity (Fig. S2A). In human and Xenopus oocytes, the majority of the mitochondria was present within the Balbiani body, as reported previously (Boke et al., 2016; Hertig and Adams, 1967) (Fig. 1G; Fig. S2B,C). Indeed, the centre of the mitochondrial mass was mostly located inside the Balbiani body in human and Xenopus oocytes (Fig. S2B,C). In contrast, mitochondria of mouse oocytes were distributed throughout the cytoplasm (Fig. 1G; Fig. S2B-E), with the centre of mitochondrial mass located inside the nucleus, confirming a more dispersed mitochondrial distribution (Fig. 1G; Fig. S2B-E). Finally, we compared mitochondrial distribution of Xenopus, mouse and human oocytes side-by-side using the same scale (Fig. S2F). This revealed that human and Xenopus oocytes have spaces several microns wide that are not occupied by mitochondria in their Balbiani bodies (Fig. S2F). Overall, we conclude that mitochondria are conglomerated in Xenopus and human primordial oocytes, whereas mouse oocytes have their mitochondria distributed in the cytoplasm.

Primordial oocytes are almost exclusively obtained from neonatal mice in the recent literature (Shimamoto et al., 2019; Pepling et al., 2007; Morohaku et al., 2016; Castrillon et al., 2003; Eppig and Wigglesworth, 2000). As Xenopus and human oocytes in this study were isolated from young adults, we repeated our experiments using primordial oocytes obtained from young adult mice. We found that similar to newly formed primordial oocytes, young adult primordial oocytes also contain active organelles. Thus, primordial oocytes isolated from neonatal and young adult mice are similar with regards to the activity of their organelles (Fig. S3A-C).

Taken together, live characterization revealed that primordial oocytes of all three vertebrates contain active lysosomes, Golgi and mitochondria. The cytoplasmic distribution of these organelles was similar in Xenopus and humans, but different in mouse (see schematic in Fig. 1H). We further confirmed this finding by immunostaining of fixed tissue sections (Fig. S1F,G; Fig. S3D-F).

Primordial oocytes of many species contain a Balbiani body, which is characterized by a dense accumulation of mitochondria adjacent to the nucleus (Kloc et al., 2004). The diffuse pattern of mitochondria in mouse oocytes prompted us to investigate whether mouse oocytes contain a Balbiani body. Previous studies have shown that the Balbiani body is held together by an amyloid-like matrix in Xenopus oocytes (Boke et al., 2016). To examine whether human and mouse oocytes contain amyloid-like assemblies in the form of a Balbiani body, we probed ovary sections with the aggresome dye Proteostat, which is widely used to mark amyloid-like proteins previously (Olzscha et al., 2017; Tao et al., 2020; Usmani et al., 2014). Proteostat clearly marked the Balbiani bodies of Xenopus and human oocytes (Fig. 2A,B; Fig. S4A,B). In fact, the accumulation of the Proteostat dye was reminiscent of the mitochondrial distribution in these two species (Fig. 1G,H, Fig. 2A,B; Fig. S4A,B). On the other hand, we did not observe a specific structure in the cytoplasm of mouse oocytes that would indicate a Balbiani body (Fig. 2C,D; Fig. S4A,B).

Live characterization and the Proteostat staining of tissue sections revealed that mouse primordial oocytes are different from human and Xenopus oocytes, such that mouse oocytes do not have any mitochondrial conglomeration in the form of a Balbiani body (Fig. 1G, Fig. 2A-D; Fig. S2B-E, Fig. S4A,B). Our findings contrast with previous reports performed on thin mouse ovary sections, which suggested a degree of mitochondrial conglomeration around the Golgi ring, and these reports used the Golgi ring as a marker for the Balbiani body (Lei and Spradling, 2016; Pepling et al., 2007).

To clarify whether the Golgi ring is indeed associated with mitochondria, we labelled mitochondria and the Golgi ring in mouse primordial oocytes, and imaged them live. We confirmed our previous finding that mitochondria were distributed throughout the cytoplasm but were spatially excluded from the area of the Golgi ring (Fig. 3A; Movie 2). In fact, a mitochondrial exclusion zone (MEZ) was present around the Golgi ring (Fig. 3A, arrowheads). We conclude that the Golgi ring is not associated with mitochondria in mouse primordial oocytes.

It could be possible that mouse oocytes are unique in possessing a Balbiani body-like compartment that lacks mitochondria but comprises the Golgi ring and RBPs, held together by a protein matrix. We hypothesized that upon Golgi ring dissociation by BFA treatment, such a protein matrix/compartment would still occupy a space and would not allow the movement of large organelles, such as mitochondria, through (Fig. 3A,B). Therefore, the MEZ should be maintained under BFA treatment.

To test this hypothesis, we disassembled the Golgi ring with BFA treatment as described above, and performed live imaging of untreated and BFA-treated mouse primordial oocytes to follow their mitochondrial distribution (Fig. 3C). In untreated cells, which had an intact MEZ, mitochondria occupied 19% of the cytoplasm (Fig. 3C,D; Fig. S4C). Upon BFA treatment, the MEZ disappeared and mitochondria redistributed throughout the entire cytoplasm, almost doubling their occupancy to 35% of the cytoplasm (Fig. 3C,D; Fig. S4C). Similar results were obtained when we treated oocytes with nocodazole, which causes the redistribution of the Golgi via a different mechanism than BFA (Cole et al., 1996; Turner and Tartakoff, 1989; Fig. 3E,F). Thus, the redistribution of mitochondria upon Golgi ring disassembly suggests that mouse oocytes lack a proteinaceous matrix holding components of a presumed Balbiani body-like compartment (Fig. 3B).

Finally, the Balbiani body has been implicated in the storage of RNAs complexed with RBPs in organisms such as zebrafish and frogs that harbour a germ plasm (Jamieson-Lucy and Mullins, 2019; Kloc et al., 2004). However, because mammals do not contain a germ plasm, it was assumed that the mammalian Balbiani body would not contain RNAs (Kloc et al., 2004; Marlow, 2010). In contrast to this notion, it has been suggested that two RBPs, namely RNGTT and RAP55B (also known as LSM14B), localize to the Golgi ring, and can be used as a marker for the mouse Balbiani body (Lei et al., 2020 preprint; Pepling et al., 2007). To test whether these RBPs indeed associated with the Golgi ring, we probed mouse oocytes for RNGTT and RAP55. RNGTT is a nuclear mRNA-capping enzyme that interacts with RNA polymerase II (Galloway and Cowling, 2019), whereas RAP55 is an mRNA-binding protein that is localized to RNA granules (Yang et al., 2006) and ER-exit sites (Wilhelm et al., 2005). Neither of these proteins were detected in Xenopus Balbiani bodies (Boke et al., 2016). In mouse primordial oocytes, RNGTT only displayed nuclear localization, whereas RAP55 localized to DDX6⁺ RNA granules (Fig. 4A-D) that have been previously described during oocyte development (Kato et al., 2019). Neither of the two proteins displayed any particular accumulation within the Golgi ring (Fig. 4A-C). Thus, we conclude that the Golgi ring does not associate with RNGTT or RAP55, and does not necessarily host any RBPs.

As the Golgi ring does not associate with mitochondria, is not maintained within a proteinaceous matrix and does not colocalize with RBPs, we conclude that it is not a marker for the Balbiani body in mouse primordial oocytes. Based on this, and together with our previous result showing that mouse oocytes lack mitochondrial conglomeration and an amyloid-like protein matrix, we conclude that mouse oocytes, unlike human and Xenopus, do not contain a Balbiani body.

We next investigated whether the Golgi ring, which has been historically linked to oocyte dormancy (Lei and Spradling, 2016; Pepling et al., 2007; Jamieson-Lucy and Mullins, 2019), is required, and thus functionally relevant for the maintenance of dormancy, in mouse oocytes. To dissect the relationship between the presence of the Golgi ring and oocyte dormancy, we investigated whether primordial oocytes would activate and exit dormancy upon Golgi ring disassembly.

The localization of the transcription factor FOXO3 serves as a dormancy marker in oocytes; it is nuclear in dormant oocytes, and is exported to the cytoplasm upon oocyte activation (Castrillon et al., 2003; Shimamoto et al., 2019). To monitor the relationship between the Golgi ring and FOXO3 in dormant and activated oocytes, in vitro culture of neonatal [postnatal day (P)3] ovaries was performed. Ovaries were fixed immediately after extraction (t=0) or after 1 h or 5 h of in vitro culture, and were processed for whole-mount imaging with GM130 and FOXO3 antibodies to check for the presence of the Golgi ring and the dormancy status of oocytes, respectively. As expected, at t=0, FOXO3 was nuclear in primordial oocytes and a Golgi ring was present in their cytoplasm (Fig. 5A), whereas growing oocytes showed cytoplasmic FOXO3 and a dissociating Golgi ring (Fig. 5A,C,D). After 1 h of in vitro culture, oocytes started activating en masse, as indicated by their cytoplasmic FOXO3 localization, as reported previously (Hayashi et al., 2020; Shimamoto et al., 2019; Fig. 5B,D; Fig. S5A,B). After 5 h of in vitro culture, nuclear FOXO3 localization remained low in all three replicates (Fig. 5B,D). The Golgi ring was present in the oocytes irrespective of FOXO3 localization (Fig. 5B-F; Fig. S5A,B), as more than 90% of oocytes still had a Golgi ring even after 5 h of in vitro culture (Fig. 5B,E; Fig. S5A,B). This suggests that oocyte activation and the corresponding export of FOXO3 to the cytoplasm precedes the disassembly of the Golgi ring.

We next investigated whether artificial disassembly of the Golgi ring would have any impact on oocyte activation. Disassembly of the Golgi ring was induced by treating whole ovaries with brefeldin A during in vitro culture. One hour after brefeldin A treatment, oocytes no longer had the Golgi ring (Fig. 5B,E; Movies 3, 4). Within each replicate, the percentage of oocytes with nuclear FOXO3 was comparable to untreated ovaries at the same timepoint (Fig. 5D). This suggests that artificial disassembly of the Golgi ring does not induce oocyte activation. Finally, ovaries were washed to remove brefeldin A and cultured in fresh medium for an additional 4 h. Surprisingly, almost all oocytes reformed their Golgi ring (Fig. 5B,E; Fig. S5A, Movies 5, 6), although many had already exited dormancy, as determined by their cytoplasmic FOXO3 staining (Fig. 5B,D-F). Thus, we conclude that Golgi ring formation is reversible, and not linked to the dormancy status of the oocyte.

Therefore, we conclude that Golgi ring disassembly follows oocyte activation and does not have a causal function in this process. Moreover, the fact that activated oocytes can have a Golgi ring calls for caution when considering using the Golgi ring as a dormancy marker.

Here, we performed the first live characterization of primordial oocytes from three vertebrate species – Xenopus, mouse and humans. This allowed us to address, for the first time, key questions about the cell biology of primordial oocytes, such as the activity and dynamics of individual organelles, their relationship with each other and their association with dormancy. We showed that, unlike Xenopus and human, mouse oocytes do not have a Balbiani body or a Balbiani body-like compartment. This might seem surprising as humans and frogs are evolutionarily more distant to each other than mouse is to either species. However, a similar phenomenon has also been observed for the inheritance of the centrosome, which is similar between humans and Xenopus but different in mice (Clift and Schuh, 2013; Schatten, 1994). We propose that this discrepancy could be explained by the different lengths of the reproductive lifespans of these animals. Oocytes are considered very long-lived cells. In particular, among the species we examined, human oocytes have the longest lifespan of the three, and can live up to 55 years (Wallace and Kelsey, 2010). Xenopus laevis oocytes can remain several years in the ovaries without growing (Callen et al., 1980; Keem et al., 1979), whereas mouse oocytes have the shortest lifespan, ranging from 8 to 14 months (Rugh, 1968). As early embryogenesis depends on the integrity of the oocyte and its organelles, the oocyte cytoplasm has to remain intact throughout dormancy (Cafe et al., 2021; Goodman et al., 2020). We speculate that the Balbiani body serves to protect the quality of mitochondria and other organelles, and its necessity depends on the length of the reproductive lifespan of the species. This would be particularly true in those animals that do not require the Balbiani body to determine the future germ line in the form of germ plasm. Consistent with this prediction, the primordial oocytes of other mammals with short reproductive lifespans, such as rat, hamster, opossum and bandicoot, also lack an obvious Balbiani body in their cytoplasm, and rather present scattered mitochondria (Falconnier and Kress, 1992; Sotelo, 1959; Ullmann and Butcher, 1996; Weakley, 1966). Other mammalian systems with longer dormancy periods, such as cows, dogs or pigs, or even non-mammalian species, such as frogs or axolotl, may thus be more appropriate than mouse for studying certain aspects of oocyte dormancy.

We showed that the Golgi ring, which was previously used as a marker for the Balbiani body in mouse oocytes (Lei and Spradling, 2016; Pepling et al., 2007), is not associated with oocyte dormancy. The Golgi ring has received attention in the field of oocyte dormancy, likely due to its unconventional and apparently unique shape. However, although textbooks typically display the Golgi as a crescent-shaped ribbon (Klumperman, 2011), several different shapes of Golgi are reported in different cell types (Kreft et al., 2010; Lu et al., 2001; Rao et al., 2018). In particular, a ring-shaped Golgi has also been reported in rat pituitary gonadotrophs and in HeLa cells depleted for a structural Golgi protein (Bassaganyas et al., 2019; Watanabe et al., 2012). We found that the Golgi ring in mouse oocytes contains stacked cis- and trans-cisternae, associates with the centrosome as reported previously (Lei and Spradling, 2016), and is capable of active membrane trafficking. Therefore, the Golgi ring, despite having an unconventional shape, displays features of a conventional Golgi.

Finally, our data indicate that primordial oocytes in vertebrates, including humans, have metabolically active organelles. This is particularly interesting considering the need for the oocyte to keep its cytoplasm healthy for long periods of time. This suggests that oocytes require efficient mechanisms to prevent or reset intracellular damage caused by metabolic activity. Therefore, it will be of great interest in the future to study how vertebrate oocytes protect themselves from the byproducts of metabolic activity during their long-lasting dormancy.

All animals were sacrificed by accredited animal facility personnel before the extraction of their ovaries. Ethical Committee permission to conduct the human oocytes aspect of this study was obtained from the Comité Étic d'Investigació Clínica CEIC-Parc de salut MAR (Barcelona) and Comité Ético de Investigación Clínica CEIC-Hospital Clínic de Barcelona (approval number HCB/2018/0497). Written informed consent to participate was obtained from all participants prior to their inclusions in the study. Any clinical investigations involving human subjects were conducted according to the principles expressed in the Declaration of Helsinki.

Xenopus and mouse colonies used in this article were housed in the Animal Facility of the Barcelona Biomedical Research Park (PRBB, Barcelona, Spain). X. laevis females were purchased from Nasco (NJ, USA). C57BL/6J mice were purchased from Charles River Laboratories and maintained under specific pathogen-free conditions at 22°C, 12 h light-dark cycles, and with access to food and water ad libitum. Female mice aged between 1 day and 7 weeks were used for experiments.

Primordial oocytes were isolated using a protocol modified from Gosden (1990). Briefly, the ovaries were digested in 1.5 mg ml⁻¹ collagenase IA (Sigma-Aldrich, C9891-1G; ovaries from P3-P4 mice) or in 0.2 mg ml⁻ Liberase (Sigma-Aldrich, 5401119001; ovaries from 5-6-week-old mice) in medium M199 (Sigma-Aldrich, 51322C) at 37°C for 30 min on a benchtop shaker. After 30 min, the solution was pipetted up and down to release individual follicles. The resulting suspension was neutralized with an equal volume of medium M199 (Gibco, 41550-020) containing 10% fetal bovine serum (FBS; Gibco, 26140-087), 2.5 mM Na-pyruvate (Thermo Fisher Scientific, 11360070), 0.2% Na-DL-Lactate syrup (Sigma-Aldrich, L7900), 1× penicillin-streptomycin (Gibco, 15070-063) and 25 µg ml⁻¹ DNaseIA (Sigma-Aldrich, CAS9003-08-9). The suspension was filtered through a 100 µm filter (Corning, CLS431752) to remove remaining ovary pieces. The solution was centrifuged at 300 g for 5 min, the supernatant decanted and pellet resuspended in fresh medium (as indicated above, without DNaseIA). The cells were transferred to a Petri dish (33×10 mm, Corning, 351008) and placed in an incubator at 37°C and 5% CO₂.

Primordial oocytes from neonatal (P3 or P4) mice ovaries were isolated using a protocol modified from Eppig and Wigglesworth (2000). Briefly, the ovaries were digested in 0.05% trypsin-EDTA (Gibco, 25300-054) with 0.02% DNase I (Sigma-Aldrich, DN25-100 mg) at 37°C for 30 min. The resulting suspension was neutralized with an equal volume of medium M199 (Gibco, 41550-020) containing 10% FBS (Gibco, 26140-087), 2.5 mM Na-pyruvate (Thermo Fisher Scientific, 11360070), 0.2% Na-DL-lactate syrup (Sigma-Aldrich, L7900) and 1× penicillin-streptomycin, and then centrifuged at 151 g for 3 min. The supernatant was decanted, and cells were transferred to a Petri dish and placed in an incubator at 37°C and 5% CO₂. All mouse oocyte imaging experiments were conducted in the medium mentioned above.

Donations were provided by the gynaecology service of Hospital Clinic Barcelona, from women aged 19 to 34 undergoing ovarian surgery. Women fulfilling the inclusion criteria undergoing ovarian surgery were asked to participate in the study. Informed consent was obtained from all of them. The inclusion criteria were as follows: aged between 18 to 35, fertile (assessed by uninduced menstrual cycles or the presence of antral follicles identified by ultrasound examination), presence of at least one ovary and signed informed consent. Exclusion criteria were as follows: women with menopause, endometriosis or have undergone bilateral oophorectomy. All oocytes incorporated in this study were from women that were free of disease affecting the reproductive system.

Donated ovarian cortex samples were transported in Leibovitz medium (Gibco, 21083-027) containing 3 mg ml⁻¹ bovine serum albumin (BSA, heat shock fraction, Sigma-Aldrich, A7906) and quickly cut into 3-mm cubic pieces. Ovary pieces were transferred to Dulbecco's modified Eagle medium (DMEM) containing 25 mM HEPES (Gibco, 21063-029) and 2 mg ml⁻¹ collagenase type III (Worthington Biochemical Corporation, LS004183), and were left to be digested in a 37°C incubator with a 5% CO₂ atmosphere for 2 h, with occasional swirling of the Petri dishes (100×20 mm, Corning, 353003). After 2 h, the resulting suspension containing individual cells was separated from tissue fragments by sedimentation in a 50 ml falcon tube, and collagenase III was neutralized by adding a 1:1 amount of DMEM/F12 medium (Gibco, 11330-032) containing 15 mM HEPES and 10% fetal calf serum (FCS, Gibco, 10270106). Individual human follicles are several magnitudes larger in volume and thus heavier than other single cells in the suspension. Incorporating this feature of follicles into the isolation protocol vastly improved the efficiency of isolation: after transferring the above supernatant to Petri dishes, oocytes sedimented to the bottom within 15 s. We then removed the top layers of the single-cell suspension by suction to have a primordial follicle-enriched Petri dish, mostly cleaned from other cells of the ovary. Follicles were picked manually under a dissecting microscope with a p10 pipette and transferred to a tissue culture dish. We obtained 60 to 180 primordial follicles from each of our ovary preparations. Leftover fragments of tissue were treated again for 2 h with DMEM containing 25 mM HEPES and collagenase III for a further 2 h, and follicles were picked as before. All human oocyte imaging experiments were conducted in the medium mentioned above.

Oocytes were isolated from young adult Xenopus (aged 3 to 5 years) ovaries according to the protocol described by Boke et al. (2016). Briefly, ovaries were digested using 2 mg ml⁻¹ collagenase IA (Sigma-Aldrich, C9891-1G) in MMR (Marc's modified Ringer; Ubbels et al., 1983), with gentle rocking for 30 to 45 min until dissociated oocytes were visible. The resulting mix was passed through two sets of filter meshes, the first with a 297-µm mesh size and the second with a 250-µm mesh size (Spectra/Mesh, 146424, 146426). All washes were performed in MMR. Oocytes that passed through the 250-µm mesh were washed once more with MMR and transferred to oocyte culture medium (OCM; Boke et al., 2016; Mir and Heasman, 2008). All frog oocyte imaging experiments were conducted in OCM at room temperature and atmospheric air.

Ovaries of 6-week-old mice were dissected in M2 medium (Sigma-Aldrich, M7167) to remove the fat pad and oviducts attached to the ovaries. The ovaries were punctured using an insulin needle to release GV-stage oocytes. The oocytes were collected with an oocyte manipulation pipette and transferred to a new dish containing M2 medium plus 400 µM dbcAMP (Sigma-Aldrich, D0627), and incubated at 37°C in atmospheric air.

Oocytes are classified according to their size and morphology in both mammalian and frog reproduction fields (Dumont, 1972; Gougeon, 1986; Pedersen and Peters, 1968; Westergaard et al., 2007). Xenopus oocytes were classified according to Dumont (1972). The average early stage I X. laevis oocyte was 200 µm in diameter and surrounded by a single layer of somatic granulosa cells. Primordial oocytes in humans were ∼30 µm in diameter and surrounded by a single layer of flattened pre-granulosa cells (Gougeon, 1986; Westergaard et al., 2007). Primordial oocytes in mice were classified according to their size (Pedersen and Peters, 1968) and were 15-17 µm in diameter. They were also surrounded by a single layer of flattened pre-granulosa cells. For cartoon representations of the oocytes and flattened pre-granulosa cells, see Fig. 1H.

Neonatal ovaries were dissected, cleaned of adjoining tissue in M2 medium and placed on Millicell hanging cell culture inserts (Merck, MCSP24H48) in a 24-well plate (Greiner Bio-One, 662160). A 500 µl amount of DMEM/F12 medium (Gibco, 31331-028) supplemented with 10% FBS and 1× penicillin-streptomycin was introduced into each well such that a thin layer of liquid was present above the ovary. The ovaries were cultured for the indicated times in an incubator at 37°C and 5% CO₂.

TMRE, Perchlorate (TMRE) (Thermo Fisher Scientific, T669) was added to oocytes of all species at a final concentration of 500 nM and incubated for 30 min. Oocytes were washed and plated on 35 mm glass-bottomed MatTek dishes (MatTek Corporation, P35G-1.5-20-C) in fresh medium.

LysoTracker Deep Red (Thermo Fisher Scientific, L12492) was added to the oocytes of all three species at a final concentration of 50 nM and incubated for 30 min. Oocytes were washed and plated on glass-bottomed MatTek dishes in fresh medium.

Human and mouse oocytes (including GV oocytes obtained from mouse) were incubated in medium containing NBD C₆-ceramide (Thermo Fisher Scientific, N22651) to a final concentration of 5 µM for 30 min (3 µm only in the case of mouse oocytes obtained through ‘trypsin-mediated digestion’ as mentioned in Materials and Methods) at 37°C and 5% CO₂. The oocytes were then washed and plated on MatTek dishes in fresh medium. Xenopus oocytes were incubated with 5 µM NBD C₆-ceramide at 4°C for 30 min, washed and incubated at 18°C for a further 30 min.

Mouse oocytes were incubated in medium containing 3 µM NBD C₆-ceramide and 100 nM MitoTracker Deep Red FM (Thermo Fisher Scientific, M22426) to visualize the Golgi and mitochondria. After 30 min, the oocytes were washed, plated on MatTek dishes and imaged live.

Ovaries were dissected from Xenopus, neonatal (P4) and adult mice (8 weeks old). Human ovarian cortex pieces were donated by patients. The tissues were fixed in 4% paraformaldehyde (PFA), embedded in paraffin and cut into 5-µm sections. Staining was performed by adapting the manufacturer's instructions (Proteostat Aggresome Detection Kit, ENZ-51035-K100). Formalin-fixed paraffin-embedded tissue sections from neonatal and adult mice, Xenopus and human were deparaffinized and permeabilized for 30 min on ice, as recommended by the manufacturer. Proteostat was added at a 1:2000 final concentration, and Hoechst was added at 1:1000 for 30 min in the dark. The slides were washed, mounted and imaged using a Leica TCS SPE microscope with a 63× oil immersion objective (numerical aperture, 1.40, Leica, 506350). The images were analysed using Fiji/ImageJ. For Fig. S5A, top 10% fluorescent intensity thresholding masks were applied in Fiji to the Proteostat signal in the oocytes of all three species.

Isolated oocytes were incubated in culture medium containing CCCP (Abcam, ab141229) at a final concentration of 30 µM for 15 min, followed by TMRE addition.

Isolated Xenopus and mouse oocytes were incubated in a droplet of medium containing bafilomycin A1 (Abcam, ab120497) at a final concentration of 500 nM (Xenopus) or 100 nM (mouse) for 1 h, followed by the addition of LysoTracker Deep Red, as described above.

Primordial oocytes isolated from neonatal or young adult (5-6 weeks old) mice were incubated in culture medium containing BFA (Abcam, ab120299) at a final concentration of 10 µM for 30 min. Whole P3 ovaries were placed on Millicell hanging cell culture inserts (Merck, MCSP24H48) in 500 µl medium with 10 µM BFA for 1 h in a 24-well plate. The ovaries were then either fixed in 4% PFA and processed for whole-mount immunostaining, or transferred for 4 h to BFA-free medium for 4 h before fixing them. Xenopus oocytes were isolated and incubated in OCM containing a final concentration of 10 µM BFA for 30 min. Then, the BFA was washed and oocytes were incubated with NBD C₆-ceramide, as described above.

Isolated oocytes were incubated in culture medium containing nocodazole (Sigma-Aldrich, M1404) at a final concentration of 5 µM for 45 min to 1 h. The oocytes were then incubated in medium containing NBD C₆-ceramide and MitoTracker Deep Red FM to visualize the Golgi and mitochondria.

Mouse and human oocytes were imaged in their respective culture medium using a Leica TCS SP5 STED microscope with a 63× water immersion objective (numerical aperture, 1.20, Leica, 506279) in an incubation chamber maintained at 37°C and 5% CO₂. Frog oocytes were imaged in OCM at room temperature and atmospheric air using a Leica TCS SP8 microscope with a 40× water immersion objective (numerical aperture, 1.10, Leica, 506357). All images were acquired using Leica Application Suite X (LAS X) software. The images were analysed using Fiji/ImageJ.

Neonatal (P3 or P4) ovaries were dissected in M2 medium to remove the surrounding tissue, and fixed in 4% PFA in PBS at 4°C for 3 h. For preparing frozen sections, ovaries were transferred to 30% sucrose in PBS and incubated overnight at 4°C. The next day, ovaries were placed in optimal cutting temperature medium within a mould, and 20-µm sections were cut using a microtome and transferred onto glass slides. Alternatively, ovaries were embedded in paraffin after fixation and 5-µm sections were cut and transferred onto glass slides.

Fragments of human ovary of ∼3×3 mm were cut from the cortex and fixed in 4% PFA in 100 mM phosphate buffer (pH 7.5) for 4 h at room temperature. Samples were then embedded in paraffin, and 5-µm sections were cut and transferred onto glass slides.

Fragments of Xenopus ovary of ∼1×1 cm were cut from freshly extracted ovaries and fixed in 4% PFA in PBS at 4°C overnight. The next day, the ovary pieces were embedded in paraffin and 5-µm sections were cut and transferred onto glass slides.

Paraffin-embedded sections were deparaffinized, boiled in 10 mM sodium citrate (pH 6.0) and directly blocked in blocking buffer (PBS containing 3% BSA, 2% normal goat serum and 0.05% Tween 20) for 1 h. Frozen sections were equilibrated at room temperature for 10 min and washed in PBS for 15 min in Coplin jars. The sections were then permeabilized (PBS containing 0.2% Triton X-100 and 0.1% Tween 20) for 30 min and blocked in blocking buffer for 1 h before proceeding with antibody incubation.

Isolated primordial follicles were fixed in 2% PFA for 15 min at room temperature in an Eppendorf tube, permeabilized and blocked with the respective buffers mentioned above. The primary antibodies were diluted in blocking solution. Follicles and sections were incubated with primary antibodies overnight at 4°C.

The primary antibodies were against GM130 (1:100, BD, 610822), TGN46 (1:100, Abcam, ab16059), Pericentrin (1:100, Abcam, ab4448), RNGTT (1:100, Abcam, ab201046), RAP55 (1:100, Abcam, ab221041), ATP5A (1:100, Abcam, ab14748), LAMP1 (1:100, Abcam, ab24170), LAMP1 (1:100, Abcam, ab25245), mannosidase II (1:100, Merck, AB3712), DDX4 (1:100, Abcam, ab27591) and DDX4 (1:100, Abcam, ab13840). Secondary antibodies goat anti-mouse-IgG Alexa Fluor 488 (1:1000, Invitrogen, A32723), goat anti-rabbit-IgG Alexa Fluor 647 (1:1000, Invitrogen, A21245), goat anti-mouse-IgG Alexa Fluor 555 (1:1000, Abcam, ab150114) or goat anti-rat-IgG Alexa Fluor 546 (1:1000, Invitrogen, A-11081) were diluted in blocking solution.

Samples were imaged in a droplet of mounting medium containing DAPI (Abcam, ab104139). Imaging was carried out using Leica TCS SP5 or Leica TCS SP8 microscopes with a 63× oil immersion objective (numerical aperture, 1.40, Leica 15506350) in the case of sections, or a 63× water immersion objective (numerical aperture, 1.20, Leica, 506346) in the case of isolated follicles. The images were analysed using Fiji/ImageJ.

Whole-mount immunostaining was performed according to Rinaldi et al. (2018). The ovaries were incubated with primary antibodies rabbit anti-FOXO3 (Cell Signaling Technology, 2497S), mouse anti-GM130 (BD, 610822), rabbit anti-RAP55 (Abcam, ab221041) and mouse anti-DDX6 (Santa Cruz Biotechnology, sc-376433) at a 1:100 ratio in blocking solution for 48 h. After an overnight wash, ovaries were incubated with secondary antibodies goat anti-mouse-IgG Alexa Fluor 488 (1:1000, Invitrogen, A32723) and goat anti-rabbit-IgG Alexa Fluor 647 (1:1000, Invitrogen, A21245) in blocking solution for 48 h. The ovaries were washed overnight with wash buffer, followed by an incubation with 50 µg/ml DAPI in wash buffer for 8 h and another overnight wash. The ovaries were imaged in a droplet of PBS on a MatTek dish using a Leica TCS SP8 microscope with a 20× air (numerical aperture 0.70, Leica, 11506166) or 40× oil immersion (11506358, Leica, N.A 1.30) objective. The images were analysed using Fiji/ImageJ.

Z-stacks of 20 µm were generated by imaging whole-mount ovaries at 1 µm intervals. The oocytes were marked by creating regions of interest (ROIs) in Fiji ROI Manager based on FOXO3, GM130 and DAPI staining. FOXO3 staining was assessed manually as nuclear or cytoplasmic, and the number of oocytes in each case was recorded. Similarly, the presence of the Golgi ring, as seen by GM130 staining, was counted in these oocytes. Nuclear FOXO3 and the presence of the Golgi ring were quantified for ovaries cultured in vitro for 0, 1 and 5 h in the absence of BFA (untreated), for ovaries treated with BFA for 1 h and for ovaries treated with BFA for 1 h followed by a 4-h washout. The corresponding values at each timepoint were plotted by GraphPad software.

Oocytes isolated from neonatal mice (P4) or young adult Xenopus and human ovaries were incubated with TMRE and imaged as described previously. The mitochondrial mass was demarcated manually as shown in Fig. S2B, and the centre of mass function in Fiji was applied to it. The coordinates of the mitochondrial centre of mass were used to determine its location, as indicated in Fig. S2C (see schematic).

Lysosomes on z-sections were used to avoid z-depth-related fluorescent intensity loss. After background subtraction, images were duplicated and subjected to a threshold to detect lysosomes. A mask was created from the thresholded image, which was applied to the original image to detect all lysosomes (Fig. S1C). Fluorescence intensities of lysosomal puncta were measured individually and plotted in Fig. 1B.

We thank Nicholas Stroustrup and Vivek Malhotra for critical reading of the manuscript; Ishier Raote for his insightful comments; the PRBB Animal Facility personnel for their continued support; the Histology Facility (Alexis Rafols Mitjans); and the Advanced Light Microscopy team (especially Timo Zimmerman). We acknowledge support of the Spanish Ministry of Science and Innovation to the EMBL partnership, the Centro de Excelencia Severo Ochoa and the CERCA Programme/Generalitat de Catalunya.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.259394.