To determine both proteome and phosphoproteome of Xenopus oocytes during meiosis resumption, meiotic maturation was triggered by progesterone, and oocytes were collected at regular intervals after progesterone exposure (Video 1, Methods): prophase oocytes (PRO) at the time of progesterone addition, then 2hr after progesterone treatment, with hourly timepoints past that, up to 9hr post-treatment. Groups of ten or fifty oocytes were collected at each timepoint for protein and phospho-proteomics measurements, respectively. After oocyte lysis, digestion into peptides and phospho-peptide enrichment in the case of phosphoproteomics analysis, samples were processed for TMT-MS3 LC/MS. After isolating phosphorylated proteins, they were chemically tagged with barcoding labels (TMT) to run multiple samples through the instrument simultaneously, thus minimizing artifacts. The resulting merged sample is then analyzed using liquid chromatography combined with a mass spectrometry technique (MS3) to accurately measure relative phosphorylation levels across the barcoded samples (Methods).

To correlate the time points of this experiment with the cytological events known to occur during meiosis resumption, a set of proteins whose accumulation had been extensively studied during Xenopus oocyte meiosis progression were used as markers. One of the first events following progesterone stimulation is the accumulation of Cyclin B1 that begins before and independently of Cdk1 activation (Santoni et al., 2024; Frank-Vaillant et al., 1999). In our experiments, Cyclin B1 accumulation starts at about 2hr and increases linearly until MII arrest at 6.5hr (Figure 2A). It has been reported that Mos translation is stimulated by progesterone concomitant with the accumulation of Cyclin B1; however, the Mos protein does not accumulate before NEBD because it turns over rapidly (Santoni et al., 2024; Frank-Vaillant et al., 1999). Another critical well-documented event that occurs downstream to Cdk1 activation is the degradation of CPEB1 (Santoni et al., 2024; Mendez et al., 2002), an RNA-binding protein whose degradation activates protein translation in MI. Hence, Mos accumulation and the degradation of CPEB1 are well-established markers for the timing of NEBD. After NEBD, activation of the anaphase-promoting complex/cyclosome (APC/C) leads to the ubiquitination and degradation of Cyclin B2, which marks the entry into anaphase I. During the MI-MII transition, multiple additional events take place: Cyclin B3 is degraded (Bouftas et al., 2022), Cdc6 (Lemaître et al., 2002; Whitmire et al., 2002; Daldello et al., 2015), Wee2 (Nakajo et al., 2000), and Cyclin E (Rempel et al., 1995) accumulate, reaching their maximal levels in MII. These four events mark the entry into MII. Based on these markers, the timing of cytological events in our experiment is as follows (Figure 2A): NEBD and MI occur between 3 and 4hr, the MI-MII transition takes place between 5 and 6hr, and entry and arrest in MII are at 7hr. This time course is in agreement with the cytological previously described events (Huchon et al., 1981; Gard, 1992). A physiological measure of MII arrest can be ascertained by the ability of the oocytes to be activated by an electric shock, a property that is acquired only in MII (Video 2).

The quality of oocytes resuming meiosis is also confirmed by the dynamics of protein accumulation and degradation that occur during meiotic progression (Figure 2B). A set of 34 proteins, whose changes in concentration have been previously studied during meiotic divisions, was used to validate the quality of our quantitative proteomics data (Figure 2B—figure supplement 1A). Our proteome correctly classifies 15 out of 18 proteins accumulating during Xenopus meiosis resumption (FC(MII/PRO)>1.5), 11 out of 12 proteins whose concentration is constant (0.75<FC(MII/PRO)<1.5), and 4 out of 4 proteins whose accumulation was reported to decrease (FC(MII/PRO)<0.75) (Figure 2B—figure supplement 1A). The concordance between our dataset (30/34 proteins) with the changes in the protein levels reported in the literature is substantially higher than in a previously published system-wide proteomic dataset from Xenopus oocytes where only 7 out of these 34 proteins were correctly measured (Peuchen et al., 2017; Figure 2—figure supplement 1A–B). This also may explain why changes in protein concentration reported in that study poorly correlate with the one measured in our proteomic study (Figure 2—figure supplement 1B). The high quality of the new proteomic data is also confirmed by the strong correlation of the translational pattern of Xenopus allo-alleles (Figure 2—figure supplement 1C–D), which were derived by genome endoduplication and are called ‘S’ or ‘L’ alleles (Session et al., 2016).

We performed phosphoproteomic mass spectrometry analysis in order to measure the overall dynamics of protein phosphorylation taking place during meiotic divisions and to identify which specific proteins and which sites are phosphorylated during this process. We identified 6783 different phospho-peptides and quantified the dynamics of relative change across all time points (Supplementary file 3). These peptides came from 2308 distinct proteins. Our phosphosites dataset contains 80% Ser, 19.9% Thr, and 0.01% Tyr. Phospho-Tyr is slightly less abundant than what has been described in most cells (up to 0.05% Sharma et al., 2014). The same observation was made regarding the distribution of phosphorylated amino acids in mouse oocytes, where phospho-Tyr abundance is relatively diminished in oocytes compared to mouse organs (Sun et al., 2024). When respective protein and unphosphorylated peptide were also measured, we were able to compute the stoichiometry (see Methods). A list of all 415 phosphosites for which it was possible to calculate the phospho-occupancy is provided (Supplementary file 4).

We identified the human homologous counterparts of 5901 (87%) among the 6783 sites found to be phosphorylated during meiotic divisions. Interestingly, only 177 (3%) out of 5901 conserved sites were already known to be phosphorylated on phosphosite.org database (https://www.phosphosite.org/homeAction). Thus, our phosphosite dataset identifies a very large number of novel phosphosites, never previously identified and characterized. Such a small number of previously reported phosphorylations among those detected by our phosphoproteome might suggest that a large portion of these phosphorylation events are related to cell division. Dividing cells generally account for a small fraction of tissue mass. This leads to a high dilution of the phosphorylation signature of cell division, as compared to the highly synchronous meiotic divisions of oocyte maturation. Interestingly, 60% of the phosphosites detected in oocytes are dynamically regulated during meiotic maturation, highlighting the importance of this post-translational regulation in controlling this last step of oogenesis and meiosis. Unbiased clustering reveals that changes in protein phosphorylation are pervasive and far more dynamic than changes in protein abundance (Figure 2—figure supplement 3), which was also noted during yeast meiosis (Koch et al., 2024). Comparing the three stages of meiotic division (PRO, MI and MII) and based on the main cellular events of these stages, we can group phospho-peptide stoichiometry into 5 classes (Figure 3A; Supplementary file 3). Class I includes proteins bearing a phosphosite whose occupancy decreases during the first 2hr following progesterone stimulation. This class includes only 136 phospho-peptides (2%) which might be involved in the early signaling pathway induced by progesterone and PKA downregulation. This small number was expected since there is a low level of phosphorylation detectable in prophase-arrested oocytes (Maller et al., 1977). This makes sense since there are presumably only a few substrates of PKA sufficient to initiate meiosis resumption. Class II is larger (39% of the phospho-peptides) and includes proteins whose phosphorylation increases between NEBD and MI and which remain highly phosphorylated in both MI and MII. The massive amount of Class II phosphosites presumably reflects the catalytic power of Cdk1 as well as its downstream kinases, to generate the thousands of substrates essential for cell division. We used the online server (https://kinase-library.phosphosite.org/kinase-library/score-site) to score the phosphosites in the Class II and predict the kinases likely responsible for their phosphorylation (Johnson et al., 2023). Using a percentile score threshold of 90, we identified 303, 304, and 267 peptides predicted to be phosphorylated by Cdk1, Erk1/2, and Plk1, respectively (Figure 3—figure supplement 1A). Among these, 166 sites were predicted to be efficiently phosphorylated by either Cdk1 or Erk1/2, consistent with the similarity in their phosphorylation consensus motifs. In contrast, there was minimal overlap between the sites predicted to be phosphorylated by Plk1 and those targeted by Cdk1 or Erk1/2. Consequently, Class II phosphorylations are largely dependent on Cdk1 activity, whether they are directly catalyzed by this kinase or indirectly through kinases under its control, such as Erk1/2 and Plk1 (see below). Our measurements show that most of the sites, which are phosphorylated in MI, remain stably phosphorylated during MI to MII transition is consistent with the observations made in yeast showing that most of Cdk1 motifs remain phosphorylated at the end of meiosis I (Celebic et al., 2024). Class III (5%) consists of proteins whose phosphorylation peaks in MI then decreases while oocytes progress throughout meiotic maturation. This is the most predominant class of phosphosites present during meiotic maturation in starfish (Swartz et al., 2021) and follows the peculiar pattern of Cdk1 activity peaking at MI in these oocytes (Okano-Uchida et al., 1998; Kishimoto, 2003). Class IV (9%) is composed of proteins whose phosphorylation progressively increases from MI to MII, hence potentially implicated in MII entry. Finally, Class V (5%) comprises proteins whose phosphorylation increases from MII and which could play a role in establishing the MII arrest. The phosphosites that are specific for either MI (Class III) or MII (Classes IV and V) are much less abundant (5%, 9%, and 5%, respectively, Figure 3A) and might distinguish the biochemical state of the two meiotic divisions. These conclusions were supported by the identification of the kinases whose predicted phosphorylation sites are most enriched within each class, by using the same server (https://kinase-library.phosphosite.org/kinase-library/score-site). Each class was associated with a distinct set of enriched kinases (Figure 3—figure supplement 1B). It confirms that Cdk1 and its downstream kinases, Erk1/2 and Plk1 are the master regulators in Class II. The kinases identified in Classes IV and V may represent novel regulators of entry into and arrest at MII.

Two waves of translation take place in the oocytes: the first depends on PKA inactivation. It occurs upstream and independently of Cdk1 activity and is required for Cdk1 activation. The second translation wave takes place downstream Cdk1 activation (Santoni et al., 2024). Hence, in contrast to mitosis, a period when protein synthesis is repressed (Ross, 1997), translation is activated during meiotic maturation, indicating the existence of meiosis-specific controls of the process. However, the regulation of the two waves of protein translation and the identity of the newly translated proteins had not been unraveled. Interestingly, protein translation components are extensively regulated at the level of phosphorylation as demonstrated by the enrichment of the proteins regulating these processes among 3 of the 5 phospho-peptide classes (Figure 4A).

The activation of Cdk1 during oocyte maturation relies on an intricate network of feedforward and feedback phosphorylation pathways mediated by kinases and phosphatases (Figure 1). Indeed, components of the cell division machinery are highly enriched among Class II, III, IV, and V (Figure 6A). Some of the phosphorylation sites implicated in cell division control have been identified previously through detailed studies using site-specific mutagenesis and phospho-specific antibodies. Our phosphoproteomic analysis of oocyte maturation provides new regulatory elements (Figure 6B). An important regulator of the Cdk1 auto-amplification loop is the kinase Plk1 that contributes to the phosphorylation of the two direct Cdk1 regulators, Cdc25 and Myt1. Plk1 is activated by Aurora-A, a kinase itself under the indirect control of Cdk1 in Xenopus oocyte (Maton et al., 2003), through the phosphorylation of T210 (T201 in Xenopus) (Macůrek et al., 2008), a residue localized in the activation T-loop of the kinase domain. We show here that Plk1 phosphorylation at T210 increases in MI and is then constant throughout meiotic maturation (Class II) (Figure 6B, Figure 6—figure supplement 1A). This is consistent with the function of Plk1, which phosphorylates multiple proteins required for the mechanics of division, and which must logically be kept active throughout both meiotic divisions (Hansen et al., 2006; Solc et al., 2015). Interestingly, the phosphoproteome also detects additional phosphorylation sites of Plk1, S326 (S335 in human) and S340 (not conserved in human), which were previously found to be upregulated in response to okadaic acid (Wind et al., 2002), an inhibitor of PP2A that strongly induces M-phase. These two residues are located between the two functional domains of Plk1, the kinase domain and the Polo-box domain (Cheng et al., 2003). This suggests that the phosphorylation of other critical residues than T210 of Plk1 could be essential for its catalytic activation. Aurora-A kinase is activated during meiotic maturation downstream of Cdk1 (Maton et al., 2003; Castro et al., 2003; Maton et al., 2005). While our phosphoproteome does not detect Aurora-A autophosphorylation at T288 (T295 in Xenopus), it registers the extensive phosphorylation of Bora, an activator of Aurora-A involved in Plk1 activation (Thomas et al., 2016; Figure 6B—figure supplement 1B). Several phosphorylation sites of Bora have been reported to be important for its function as an Aurora-A activator: S41/38, S112/110, S137/135 and S252/S252 (human/Xenopus) (Thomas et al., 2016; Vigneron et al., 2018; Tavernier et al., 2015; Tavernier et al., 2021). Among these sites, S110 phosphorylation of Bora by Cdk1-Cyclin A is critical for mitotic entry in Xenopus egg extracts (Vigneron et al., 2018; Tavernier et al., 2015; Tavernier et al., 2021). Our phosphoproteome reveals that Bora belongs to Class II, being phosphorylated starting NEBD, but surprisingly on distinct sites that were never reported in the literature: S191/189, T294/287, S552/549, S-/521, T12/9, and T15/12 (human/Xenopus), with the exception of S278/285. In contrast to the situation in mitosis, Cyclin A is very weakly expressed in the prophase oocyte and accumulates during meiotic maturation (Kobayashi et al., 1991). Our proteome dataset reveals that Cyclin A accumulates after MI. Hence, the phosphorylation of Bora at MI cannot be catalyzed by Cdk1-Cyclin A in the Xenopus oocyte. Cdk1-Cyclin B is probably involved in these phosphorylations that take place concomitantly with its activity and at Cdk1 phosphorylation consensus sites for 5 of the 7 phosphosites (Figure 6B—figure supplement 1B). Thus, this analysis reveals that Bora activation correlates with its phosphorylation at residues not previously identified in earlier studies of the oocyte. We also detect additional phosphorylation events on other players of the Cdk1 amplification loop, such as Cdc25 and Gwl/MASTL (Figure 6B—figure supplement 1C–D), whose extensive phosphorylation during M-phase is usually detected by a large electrophoretic shift (Dupré et al., 2014).

In Xenopus oocytes, Cdk1 is not fully inactivated at anaphase I and during extrusion of the first polar body. Indeed, the low Cdk1 activity that persists during the MI-MII transition is essential to avoid the reformation of a nucleus, chromosome decondensation and DNA replication (Huchon et al., 1993; Thibier et al., 1997). This short period is followed by a further increase in Cdk1 activity, due to a sustained synthesis of B-Cyclins; the increased Cdk1 activity allows the formation of the MII spindle (Gerhart et al., 1984; Figure 1). Although this bimodal subtle regulation of Cdk1 activity is essential for the success of meiosis, i.e., linking two successive divisions without intervening DNA replication, it has never been quantified precisely. Cdk1 activity can be measured by looking at the phosphorylation of one of its direct substrates, Cdc27, a core APC component, at S428 (S426 in human, Patra and Dunphy, 1998). Since PP2A, the phosphatase responsible for dephosphorylating Cdc27, is inactive during the MI-MII transition, the phosphorylation levels of Cdc27 can be directly attributed to Cdk1 activity during this period (Deak et al., 2003; Torres et al., 2010; Lemonnier et al., 2021; Labbé et al., 2021). Our precise quantification of the phospho-occupancy of this site shows that its phosphorylation is virtually absent in prophase, increases by 40% in MI, decreases during the MI-MII transition to 20%, then increases again in MII by 60% (Figure 6—figure supplement 1E). Three other residues of Cdc27, T205/207, S435/437, and S438/440 (human/Xenopus), also known to be phosphorylated at M-phase (Kraft et al., 2003), follow the same bimodal pattern as S428. The subsequent increase in Cdk1 activity essential for MII entry requires that the kinase escapes inhibitory phosphorylations by Myt1 and Wee2. Both Myt1 and Wee2 are reported to be inhibited by phosphorylations, although the sites responsible for this inhibition are not precisely defined. In Xenopus, Wee2 is not expressed in prophase oocytes and accumulates after NEBD (Nakajo et al., 2000; Charlesworth et al., 2012; Figure 1). Our analysis reveals that newly synthesized Wee2 is immediately phosphorylated at four sites (Classes II and IV), one of them being not conserved in human, and none of them corresponding to the proposed inhibitory S53 and S123 sites (Figure 6, Figure 6—figure supplement 1F). Hence, these data provide exciting new avenues of research to discover unexpected new regulations of old players in Cdk1 activation.

Mos plays a well-known role at the origin of a critical pathway controlling meiotic maturation (Yew et al., 1992; Furuno et al., 1994). It phosphorylates Mek1/2, which in turn phosphorylates Erk1/2 (also known as MAPK) which leads to Rsk1/2 activation (Figure 6B). These play multiple essential roles, such as preventing the reformation of a replicating nucleus between MI and MII (Dupré et al., 2002), enabling arrest in MII (Sagata et al., 1989b), but also participating in Cdk1 activation (Sagata et al., 1989a) and the functioning of meiotic spindles (Araki et al., 1996; Choi et al., 1996; Bodart et al., 2005). The phosphoproteome detects the activation of the Mos-MAPK module, with the phosphorylation of Erk2 on T185 and Y187 (T188 and Y190 in Xenopus) by MEK1/2 (Payne et al., 1991), an event that begins in MI (Figure 6B—figure supplement 2A). Interestingly, additional phosphorylation sites are identified in MEK1 and Rsk1/2, the latter being hyperphosphorylated during meiosis as detected by a large electrophoretic shift (Bhatt and Ferrell, 2000; Figure 6B—figure supplement 2B–E). Rsk1 auto-phosphorylation at S381 (S380 in human) is required for its activity (Vik and Ryder, 1997). Our phosphoproteome measures a highly confident phospho-occupancy for this site, demonstrating the switch-like activation of the MAPK pathway at NEBD, followed by the maintenance of a constant level of its activity (Figure 6B—figure supplement 2F). The timing of activation of the Mos-MAPK cascade has been controversial, one hypothesis being that it is activated before Cdk1 and participates in the activation of this kinase (Sagata et al., 1988; Gavin et al., 1999), while another is that it is under the control of the starter Cdk1 activity and is only involved in the autoamplification loop (Santoni et al., 2024; Dupré et al., 2002; Fisher et al., 1999). The phosphoproteomic data clearly support the latter.

As a result of the massive activation of kinase activities under the control of Cdk1, hundreds of proteins are phosphorylated and orchestrate the structural events, choreographing the mechanics of meiotic divisions. The first of these is the massive vesiculation of the nuclear envelope surrounding the enormous oocyte nucleus. During mitosis, nuclear pore proteins are extensively phosphorylated by Cdk1-Cyclin B, Plk1, and Nek6/7 to promote the breakdown of the nuclear envelope (Kutay et al., 2021). The Xenopus oocyte has the unusual feature of having an enormous nucleus (475μm in diameter), and, therefore, a considerable surface area of the nuclear envelope (0.96 mm²), as well as large stocks of annulate lamellae in the cytoplasm. Annulate lamellae are cytoplasmic stack cisternae of nuclear envelope perforated with nuclear pores, 10 times more abundant than in the nucleus in the Xenopus oocyte, but devoid of lamins (Bement and Capco, 1990; Miller and Forbes, 2000; Cordes et al., 1995). Their surface area also far exceeds that of the nuclear envelope surrounding the nucleus (Cordes et al., 1995). This correlates with the high protein concentrations of the nuclear pore proteins (0.1–0.7μM) that we have measured (Figure 7A). Annulate lamellae play a critical role as a reservoir for the formation of the 4000 of nuclei that form during the embryonic cell divisions leading up to the mid-blastula transition and the onset of transcription (Kessel et al., 1986; De Magistris and Antonin, 2018; Feldherr, 1974). During meiotic maturation, NEBD and vesiculation of annulate lamellae occur at the same time, in a progressive spatial manner, starting from the vegetal pole. Importantly, the nucleus and annulate lamellae do not re-form between the two meiotic divisions, reducing the risk that an S-phase can take place between MI and MII (Bement and Capco, 1990). A gene ontology analysis of our phosphoproteome highlights that the phosphorylation of nuclear pore components is enriched in Classes II, III, and IV, displaying the highest level of enrichment in Class III, hence characteristic of MI (Figure 7B). The nucleoporin Nup53 (also called Nup35) is a component of the soluble core region of the nuclear pore complex that is extensively phosphorylated in mitosis by Cdk1 and Plk1 (Linder et al., 2017). Phosphomimetic mutations of all Cdk1 and Plk1 sites in Nup53 slow down the reformation of nuclear pores in interphase and mutation to phospho-null amino acids impairs the nuclear breakdown during meiosis (Linder et al., 2017). Our phosphoproteome reveals that Nup53 phosphorylation is more dynamic than previously believed (Figure 7C). Indeed, six phospho-peptides are found to peak in MI (Class III), nine phospho-peptides are phosphorylated both in MI and MII (Class II), and two phospho-peptides continuously increase in phosphorylation during meiotic maturation (Class IV) (Figure 7C). We were able to calculate the phospho-occupancy of the S59 (S66 in human), showing that it increases from 30% in prophase to 90% in MI oocytes and stays constant until MII (Class II) (Figure 7D). Several other transmembrane nucleoporins, notably NDC1 and POM121C, undergo a dramatic increase in phosphorylation at the transition to MI. Altogether, these observations indicate that dissociation of the nuclear envelope and the annulate lamellae occurring in MI might require either a specific set of phosphorylations and/or a general level of phosphorylation that is higher than the level of phosphorylation required to maintain the nucleus and the annulate lamellae dissociated during the MI-MII transition through MII.

An intriguing feature of oocytes in most, if not all, metazoan species is the elimination of their centrioles. This allows the sperm to contribute its centriole and thereby initiates the division cycle and the cell cycle. There is longstanding literature that proves that the oocyte or egg has retained its ability to generate centrioles. However, in nature, the female centrioles are lost and the sperm centriole is the main contributor. In Xenopus, oocytes entering prophase I contain a typical centrosome, including a pair of centrioles and pericentriolar material (PCM) containing γ-tubulin and pericentrin. Centrioles disappear at the end of the pachytene stage of prophase I and PCM disperses in multiple foci in the cytoplasm that are no longer functional as microtubule nucleation sites (Gard et al., 1995). At fertilization, the male gamete delivers two centrioles but does not provide the PCM (Delattre and Gönczy, 2004). Therefore, in most species (with the exception of rodents), the complete centrosome of the one-cell stage embryo is a bi-parental inheritance, reconstituted from paternal centrioles and maternal PCM (Delattre and Gönczy, 2004). In Xenopus, oocytes store enough quiescent centrosomal building blocks for duplicating and assembling 1000–2000 centrosomes (Gard et al., 1990). This implies that oocytes should comprise an enormous reserve of centrosomal proteins, i.e., centriolar and PCM proteins, that sustain the reconstitution of a functional centrosome at fertilization and its subsequent rounds of duplication in the early embryo.

Xenopus prophase oocytes are inefficient in their ability to support microtubule polymerization (Jessus et al., 1984; Gard and Kirschner, 1987; Heidemann and Kirschner, 1975), although microtubules are formed at the cortical and nuclear envelope level, where γ-tubulin foci are also localized (Gard, 1994; Jessus et al., 1988). The ability to assemble microtubule asters is acquired at the time of NEBD, when a giant disk-shaped MTOC generates a large microtubular monoaster at the basal region of the disintegrating nucleus (Huchon et al., 1981; Gard, 1994). Therefore, MTOC and microtubule-associated proteins (MAPs) must be finely regulated by post-translational modifications at the time of NEBD. The transient microtubule monoaster transports the meiotic chromosomes to the animal cortex, where the first meiotic spindle is assembled (Huchon et al., 1981; Gard, 1992). Due to the absence of centrioles, meiotic spindles are barrel-shaped and are formed by centrosome-independent microtubule nucleation originating from the chromosomes and involving chromatin and the Ran GTPase (Gruss and Vernos, 2004), much like spindles of vascular plant cells. Moreover, the two successive spindles are differentially regulated. The MI spindle is a transient and dynamic structure, while the MII spindle is stable for hours, anchored to the plasma membrane until fertilization (Gard, 1992). Therefore, it is possible that MAPs are specifically and differentially regulated at MI and MII (Classes III and V). Finally, a major change in microtubule assembly occurs during the transition from meiosis to embryonic mitosis. The assembly of the first embryonic spindle integrates the meiotic chromosome-dependent pathway with the newly formed centrosomes (Cavazza et al., 2016). Thereafter, subsequent embryonic spindles depend solely on the microtubule nucleating activity of the centrosomes (Kirschner, 1986). All these critical transitions in the way of assembling a microtubular spindle depend on the proteins stored in the oocyte.

Many regulators of the microtubular spindle are highly regulated during meiotic divisions, both at the level of their accumulation (Figure 2B–C) and by post-translation modifications (Figure 9). At least five microtubule regulatory proteins undergo specific patterns of phosphorylation that deserve attention (Figure 9). Katanin (KATNB1), which is responsible for the majority of M-phase severing activity in Xenopus eggs and is activated by Cdk1 (McNally and Thomas, 1998), displays a bi-modal pattern of phosphorylation, increasing at the time of formation of the monoaster-MTOC and MI spindle, decreasing during MI-MII transition, and increasing again at MII (Figure 9). Kif11/Eg5 phosphorylation occurs at T936 and T937 starting NEBD (Figure 9). Both phosphosites were reported in a broad range of models, and the second site was shown to be phosphorylated by Cdk1 (Blangy et al., 1995; Cahu et al., 2008; Giet et al., 1999; Figure 9). Kif11/Eg5 controls the relative stability of bipolar versus monopolar organization of spindles in Xenopus egg extracts (Mitchison et al., 2005) but also centrosome disjunction and/or separation. Hence, Kif11 is a good candidate to be an important player regulating the original changes in microtubule dynamics and organization of the two successive spindles. Kif15, also known as XKlp2, a plus end-directed kinesin protein required for centrosome separation and maintenance of spindle bipolarity in Xenopus egg extracts (Boleti et al., 1996), is phosphorylated at S681 and S692 during meiotic maturation. S681 is rather phosphorylated at MI, while S692 is specific to MII (Figure 9), suggesting a potential distinct regulation of Kif15 on microtubule dynamics at MI and MII. In contrast to Kif11/Eg5 and Kif15, some kinesins from the kinesin-14 family, such as Kifc1, whose protein level increases during meiotic divisions (Figure 2C), as well as dyneins, are motor proteins that move to microtubules in the opposite direction to kinesins. Among the dynein family, the two paralog genes dynein light intermediate chain (DynC1LI1 and DynC1LI2) play important roles in mitosis, including positioning the spindle and focusing the MTs into poles. Our phosphoproteome reveals multiple phosphorylations of these proteins, starting in MI and remaining constant or progressively increasing until MII, as for S197 of DynC1LI1 (S207 in human) (Figure 9). This agrees with previous data showing that S197 is phosphorylated by Cdk1 (Dell et al., 2000; Addinall et al., 2001) and that DynC1LI becomes hyperphosphorylated at the time of NEBD and remains hyperphosphorylated throughout the rest of meiosis (Huang et al., 1999). Additionally, our phosphoproteome reveals that three sites (495, 498, and 501) are phosphorylated around NEBD and get dephosphorylated after MI (Class III), hence being correlated with the first meiotic division and not the second one, while six other sites (197, 396, 409, 446, 452, and 467) are phosphorylated both in MI and MII (Class II) (Figure 9). Dynein is required for the organization of the original microtubule/MTOC array that is organized at NEBD, together with XMAP215, XKCM1, and Numa (Becker et al., 2003). The formation of this original oocyte monoaster/MTOC could involve the dynein phosphorylations that transiently take place precisely at this period.

Strikingly, MI and MII spindles segregate chromosomes and chromatids in a very different way. In MI, the homologous chromosomes are segregated, whereas in MII the sister chromatids are displaced in the daughter cells, due to a specific organization of kinetochores and regulation of cohesins. In MI, arm cohesins are phosphorylated and degraded, whereas centromeric cohesins are protected from cleavage by Sugoshin (Sgo1)-PP2A. This allows the separation of chromosomes, but chromatids are still attached together. In MII, centromeric cohesins are fully degraded, allowing the segregation of sister chromatids, as during mitosis (Wassmann, 2022). How a fraction of cohesins is protected from cleavage in MI but not in MII is still not entirely clear since the phosphorylation of cohesins and their partners, which are not very abundant proteins, is difficult to detect with conventional shotgun phosphoproteomics. Notably, our phosphoproteomic studies highlight new sites differentially phosphorylated between MI and MII within these proteins. Sgo1 and two components of the cohesin complex, Pds5b and Wapl, are increasingly phosphorylated during meiotic divisions (Class IV) (Figure 9—figure supplement 1). Noteworthy, S1069 of Wapl is a putative Plk1 site (Grosstessner-Hain et al., 2011; Kettenbach et al., 2011). By highlighting new sites differentially phosphorylated between MI and MII, our phosphoproteomic studies provide clues to elucidate the still unknown mechanisms of the meiotic mechanism of cohesion protection-deprotection.

We compared our phosphoproteome to a dataset obtained during meiotic divisions in mouse oocytes (Sun et al., 2024). We identified 408 phosphorylation sites present in both datasets (see Methods) (Supplementary file 5). These sites correspond to 320 Xenopus proteins and 277 mouse proteins. Note that when multiple mouse proteins share the same phospho-peptide sequence, they are grouped together in the mouse dataset. However, alignment with Xenopus data allows us to resolve which specific mouse protein is phosphorylated, thereby enriching the mouse dataset by leveraging information from Xenopus measurements. We then assessed whether the changes in phosphorylation between prophase and metaphase II were consistent between Xenopus and mouse meiotic maturation (Figure 10A). Interestingly, the phosphorylation dynamics were significantly correlated between the two datasets (Pearson coefficient: 0.39, p<0.0001) (Figure 10A). Furthermore, the changes in phosphorylation observed in Xenopus oocytes (range of the Log₂ Fold Change (MII/Pro): –4.27 and 7.03) were substantially greater than those measured in the mouse dataset (range of the Log₂Fold Change (MII/Pro): –2.55 and 3.72), likely reflecting the higher dynamic range achieved by our workflow. Among the key phosphorylation events detected in both datasets are those involved in the activation of Cdk1 and the Mos/MAPK pathway. Notably, phosphorylation of Plk1 at T210 (mouse numbering, T201 in Xenopus), which is catalyzed by Aurora A and is critical for Plk1 activation (Macůrek et al., 2008), as well as phosphorylation of Gwl at S442 (mouse numbering, 467 in Xenopus), increased in both datasets (Figures 6B and 10A–B). Similarly, both datasets consistently detected phosphorylation of Erk2 at T183 (mouse numbering, T188 in Xenopus), which reflects MEK-dependent activation of this kinase (Payne et al., 1991; Figures 6B and 10A–B). Among the conserved sites whose phosphorylation increases during meiosis, several are known targets of Cdk1 and Erk/MAPK (Figure 10B), consistent with the activation of these kinases during meiotic maturation. We experimentally confirmed the phosphorylation of Fak1 at S913, Erk2 at T188, and Plk1 at T201 by western blot (Figure 10C). Moreover, by microinjecting the Cdk1-specific inhibitor Cip1 in oocytes in order to prevent Cdk1 activation, we demonstrated that Cdk1 is required for phosphorylation of all these substrates, whether they are direct targets of Cdk1 or Erk/MAPK (Figure 10C). This aligns with previous reports indicating that MAPK pathway activation in oocytes occurs downstream of the initial activation of Cdk1 (Santoni et al., 2024). Interestingly, AKTS1/PRAS40 phosphorylation decreases during meiotic maturation in both species (Figure 10A–B). We experimentally validated the dephosphorylation of AKTS1/PRAS40 at S208 and showed that this event occurs at NEBD and depends on Cdk1 activity (Figure 10C). AKTS1/PRAS40 is a direct inhibitor of mTORC1 by competing with 4E-BP1 and S6K1 for binding to raptor (Wang et al., 2007; Oshiro et al., 2007). AKTS1/PRAS40 is phosphorylated on S184 (mouse numbering, S208 in Xenopus) by mTORC1, which leads to its dissociation from mTOR and enhances its ability to phosphorylate its substrates (Oshiro et al., 2007). Interestingly, previous studies suggest that the translation of ribosomal proteins, which are targets of mTOR signaling via the TOP motif located in their 5′-UTR, decreases during meiotic maturation (Luong et al., 2020). Our findings support these observations and provide additional evidence for the downregulation of mTOR activity upon meiotic resumption.

The research with Xenopus laevis was performed under the oversight of the Harvard Medical Area Institutional Animal Care and Use Committee in accordance with HMS IACUC protocol IS00001365. Mature Xenopus laevis females were used after priming with pregnant mare serum gonadotropin (PMSG, 50 IU per frog). A sample of ovary tissue was surgically removed, rinsed in 1x Marc’s modified Ringer’s (MMR), and placed in OR2 buffer (83 mM NaCl, 2.5 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 1 mM Na₂HPO₄, 5 mM HEPES, pH 7.8) and oocytes were manually defolliculated. Prophase oocytes were selected and kept in dishes with agar beds to avoid sticking to the dish. For the Cip1 experiment, prophase oocytes were microinjected with 60 ng of GST-XenCip1, as previously described (Santoni et al., 2024). Maturation was induced by incubation at 18°C with 10µg/mL progesterone in OR2 buffer.

Oocytes were flash frozen with liquid nitrogen in bunches of 10 or 50 for protein and phospho-proteomics measurements, respectively. All samples were taken in triplicates, each biological replicate was performed with different frogs. The first sample was taken from untreated oocytes at the time of progesterone treatment, the second timepoint 2 hr after progesterone treatment, with hourly timepoints past that, up to 9 hr post-treatment.

For each condition, the oocytes were lysed at a ratio of 6μL buffer/embryo in the same buffer as (Presler et al., 2017) (0.25M sucrose, 1% replacement NP-40, 10mM EDTA, 25mM HEPES, 1mM TCEP, 10mM Combretastatin, 10μM Cytochalasin D, Roche Protease Inhibitor Mini EDTA free, and Phosstop inhibitor (Roche) at pH 7.2). The yolk was removed by spinning the samples at 4000g for 4min at 4°C. Both the supernatant and all lipids were transferred to a new tube. The concentration of HEPES was raised to 0.1M and SDS was added to 20%. After the addition of DTT to 5mM, the samples were denatured in a 60°C water bath for 20min. After the samples temperatures returned to room temperature, N-Ethylmaleimide (NEM) was added to 15mM, tubes were mixed thoroughly, and the reaction proceeded for 1hr. Then additional DTT was added to 5mM to quench the excess NEM. The samples were flash frozen and stored at –80°C until the methanol-chloroform precipitation (Wessel and Flügge, 1984). Protein was precipitated by methanol-chloroform precipitation. Due to the large volume of lysate (~1mL), the precipitation was performed in glass 15mL Corex tubes (No. 8441). The tubes were cleaned for a 2 hr chromic acid incubation and rinsed extensively in filtered water. Then they were autoclaved and before use for the precipitation, they were rinsed with HPLC H₂0. The centrifugations were done in a JA 50 30 Ti rotor in a Beckman Avanti J-30I centrifuge at 21,000g for either 6min or 15min. After the last supernatant was removed, the pellets were air dried for 15min and resuspended in 0.6mL of 6M guanidine, 50M EPPS pH 8.5 and transferred to a 2mL Eppendorf tube. The samples were incubated for 5min at 65°C and the protein amount was determined by a BCA Assay. The solution was diluted to 2M guanidine with 10mM EPPS pH 8.5. Lysyl endopeptidase (FUJIFILM WAKO (129–02541) resuspended with HPLC H₂0 to ~2mg/mL) was added to a concentration of 20ng/mL and the initial digestion proceeded for 12hr at room temperature with gentle shaking. The samples were further diluted to 0.5M guanidine with 10mM EPPS pH 8.5 and Trypsin protease (Sequencing Grade Modified Trypsin, Promega) was added to a final concentration of 10ng/mL along with an additional 20ng/mL of LysC. The samples were incubated in a 37°C room for 8hr on a nutator.

To carry out the phosphoproteomic analysis, we followed the conventional steps of digesting the proteins from each time sample with trypsin and specifically labeling them with isobaric tags (TMT-10 reagents). Tagged samples and the phospho-peptides were enriched on an IMAC column. Importantly, we chose to multiplex peptides before the phospho-peptide enrichment to improve data quality. There is a trade-off of decreased yield and, therefore, depth, as it is not economical to label more than a few milligrams of material. We used 2.5–4 mgs of TMT-labeled peptides per replicate, enriched with 5μm Titanium Dioxide microspheres (GL Sciences 5020–75000) and fractioned as previously described (Presler et al., 2017). A typical yield of 50–80μg of peptides eluted from the column, with a median phospho-peptide enrichment of ~80%.

We adapted the method that we used to phosphatase-treat digested peptides (Presler et al., 2017) for the larger samples used here for the parallel phospho-enrichment. We chose to phosphatase treat when proteases were present in the samples (but after the protease digestion sequence described above) to enable the dephosphorylated peptides to have the cleavage pattern expected if they were present initially in the dephosphorylated state. We used the heat-labile Alkaline Shrimp Phosphatase (Affymetrix Inc, USA). We used a 3k Amicon Centrifugal Filter to exchange the enzyme storage buffer to 10mM EPPS pH 8.5 instead of Tris-HCl pH 7.5. We also concentrated the phosphatase to between 3–4U/μL (confirmed by p-nitrophenyl phosphate assay). We added an estimated 833 units of phosphatase to two of the samples built to contain representative mixture of peptides (TMT 131N, channel 10: 100 oocytes collected at the latest timepoint, and TMT 131 C, channel 11: 50 oocytes collected at the first timepoint combined with 50 oocytes collected at the last timepoint). To samples that were not phosphatase-treated, we added the same volume of blank phosphatase buffer: 5mM EPPS pH 8, 50% glycerol. To all samples, we added MgCl₂ to 10mM and EDTA to 0.1mM (to chelate metals that might compete with the magnesium necessary for enzymatic activity). All the samples were incubated with gentle shaking at room temperature for 12hr. The phosphatase was then inactivated by incubation with gentle shaking in an air incubator at 65°C for 15min. Peptide fractionation and TMT-MS3 LC/MS were carried out as done previously (Presler et al., 2017).

To compute phospho-occupancy, we used the phospho_occupancy_matlab package, available as part of our code repository (https://github.com/elizabeth-van-itallie/phospho_occupancy_matlab; Van Itallie, 2024) implementing the methods previously described in Van Itallie et al., 2025. We grouped phospho-peptides by unique phosphosite identifiers. In order to rigorously reflect the measurement accuracy and to take advantage of many measurements of the same phosphosite (including on peptides with different missed-cleavages or oxidized methionines), we used BACIQ (Peshkin et al., 2019) to compute the estimated trend with confidence intervals. Then, we created phosphosite sets to connect phosphosites on residues that could be on the same peptide. We made these assignments based on whether the individual phosphosites were ever measured with another phosphosite as a composite phosphosite. In order to determine occupancy, we needed to measure the phosphosite as a different form (usually not-phosphorylated). To do this, we identified non-phosphorylated peptides measured from the same experimental conditions that include the residue(s) of the phosphosite. Again, multiple non-phosphorylated peptides can include the residues of interest and the same peptide sequences can be measured multiple times. Therefore, we again use BACIQ (Peshkin et al., 2019) to aggregate these measurements in a trend with confidence intervals.

The relative protein abundance data across subsequent hourly timepoints post-progesterone stimulation was clustered using the K-means clustering algorithm with cosine similarity distance. The number of clusters was selected to ensure proteins in each cluster are closely concordant and under-clustered for visualization purpose, resulting in 60 and 90 clusters for protein and phosphopeptides, respectively. The code to perform this clustering procedure is available as Source code 1 and was run using MATLAB (version R2022b). The K-means clustering algorithm converges to local minima only, therefore, we performed multiple clustering realization to identify the most biologically plausible clusters, as shown in panels A-B. Resulting clusters are ordered by cardinality, most populated clusters first.

To compare the dynamics of mouse and Xenopus phosphosites between two species, we used the phospho_matching package, available as part of our code repository (https://github.com/elizabeth-van-itallie/phospho_matching; Van Itallie, 2025) implementing the methods previously described (Van Itallie et al., 2025). We first matched the phosphosites using the protein sequence and phosphorylated residues, then filtered the list to only keep matched phosphosites measured in both datasets. For cross-species phosphomatching, we used the default parameters that are defined in the methods section in Van Itallie et al., 2025: BLOSUM90 substitution penalty matrix and blastp alignment results with E-value less than 1e-20.

Peptide-Spectra matches were performed as previously described (Sonnett et al., 2018). Assignment of MS2 spectra was performed using the SEQUEST (Eng et al., 1994) algorithm by searching the data against the appropriate proteome reference set acquired from Xenbase (Xenopus laevis assembly 9.1) along with common contaminants: human keratins and trypsin. The target-decoy strategy was used to construct a second database of reversed sequences that were used to estimate the false discovery rate on the peptide level. SEQUEST searches were performed using a 20 ppm precursor ion tolerance with the requirement that both N- and C-terminal peptide ends are consistent with the protease specificities of LysC and Trypsin. TMT (+229.162932 Da) was set as a static modification on N-termini and lysine peptides, and N-ethyl maleimide (+125.047679 Da) was set as a static modification on cysteine residues. Oxidation of methionine (+15.99492 Da) was set as a variable modification. A peptide level MS2 spectral assignment false discovery rate of 1% was obtained by applying the target-decoy strategy with linear discriminant analysis. Peptides of 7 amino acids length or longer were ranked by linear discriminant analysis score and were filtered to a ratio of 1% reverses/forwards +reverses. Peptides were assigned to proteins and a second filtering step to obtain a 1% FDR on the protein level was applied. Peptides that matched multiple proteins were assigned to the proteins with the most unique peptides. We only used Isolation Specificity >0.75 spectra.

The absolute protein concentration was estimated according to previously published methods (Wühr et al., 2014; Peshkin et al., 2019), based here on ion current prorated to the isobarically labeled fractions. The respective sample fractions were estimated using our previously published Bayesian approach (Peshkin et al., 2019), which integrates peptide signal and peptide-level measurement agreement into a maximum likelihood estimate of the true protein ratio and the associated confidence interval (Peshkin et al., 2019). See Appendix Supplementary Methods.

Oocyte homogenization and SDS-PAGE were performed as previously described in Santoni et al., 2024. The following antibodies were used: Karyopherin (1:2000, goat, Santa Cruz sc-1863), pT188-pY190-Erk2 (1:2000, mouse, Cell Signaling 9106), pT201-Plk1 (1:1000, Abcam Ab39068), pS910-FAK (1:5000, rabbit, Invitrogen 4459 G) and pS183-Akts1/PRAS40 (1:5000, rabbit, Cell Signaling 5936). The appropriate secondary HRP-coupled antibodies were used at 1:10,000 dilution (Jackson Immunoresearch).