Actin-driven chromosome clustering facilitates fast and complete chromosome capture in mammalian oocytes

Accurate chromosome segregation during meiosis is crucial for reproduction. Human and porcine oocytes transiently cluster their chromosomes before the onset of spindle assembly and subsequent chromosome segregation. The mechanism and function of chromosome clustering are unknown. Here we show that chromosome clustering is required to prevent chromosome losses in the long gap phase between nuclear envelope breakdown and the onset of spindle assembly, and to promote the rapid capture of all chromosomes by the acentrosomal spindle. The initial phase of chromosome clustering is driven by a dynamic network of Formin-2- and Spire-nucleated actin cables. The actin cables form in the disassembling nucleus and migrate towards the nuclear centre, moving the chromosomes centripetally by interacting with their arms and kinetochores as they migrate. A cage of stable microtubule loops drives the late stages of chromosome clustering. Together, our data establish a crucial role for chromosome clustering in accurate progression through meiosis.

Human eggs frequently carry an incorrect number of chromosomes 1 , a condition referred to as aneuploidy. Most aneuploid eggs cannot develop to term upon fertilization, making high levels of aneuploidy in eggs a leading cause of pregnancy loss and infertility in women [1][2][3] . Aneuploidy in eggs arises from chromosome segregation errors during oocyte meiosis 1,4 . Despite its fundamental importance for fertility and development, our understanding of human oocyte meiosis is limited.
Chromosomes are segregated by a microtubule spindle, which in human oocytes assembles in the absence of centrosomes 5,6 . Defects during spindle assembly were linked to high aneuploidy levels in human eggs 5,7 . However, the mechanisms that facilitate spindle assembly in human oocytes remain largely unclear.
The nucleation of spindle microtubules after nuclear envelope breakdown (NEBD) is markedly delayed in human oocytes compared with mitotic cells 8 or mouse oocytes 9 . Interestingly, chromosomes in human oocytes undergo rapid movements after NEBD and before the onset of spindle assembly 5 . Upon NEBD, the chromosomes migrate centripetally towards each other, forming a compact chromosome cluster 5 . The chromosome cluster remains intact until a microtubule aster forms within the cluster's centre 5 . The aster grows and transforms into a bipolar spindle 5 . The physiological relevance of chromosome clustering in human oocytes and its underlying mechanism are unknown.
Chromosomes in starfish and Xenopus oocytes are also transported before spindle assembly. The nuclei of these oocytes are over two times larger than those of human oocytes and their chromosomes must be moved to bring them within reaching distance of spindle microtubules 10,11 . In starfish oocytes, each chromosome becomes enclosed by a dense layer of actin 12 , and then linked to a filamentous actin (F-actin) network 13 that contracts towards the centrosomes 14 . Once the chromosomes are within reaching distance 14 , the centrosomes rapidly assemble Article https://doi.org/10.1038/s41556-022-01082-9 oocytes, they rapidly assemble a spindle after NEBD 26 and do not form a chromosome cluster 9,26 .
We report that chromosome clustering in human and porcine oocytes is predominantly driven by actin cables that interact with the chromosomes' kinetochores and arms. A cage of microtubule loops forms around the chromosomes in the late stages of NEBD and transports chromosomes that have not yet been clustered by the actin cables. Chromosome clustering promotes rapid and complete capture of all chromosomes by the assembling spindle. Chromosome clustering is therefore an essential stage of porcine and human oocyte development.

Human and porcine oocytes cluster their chromosomes
We first characterized chromosome clustering in human oocytes and observed two distinct chromatin configurations in the intact nucleus. The chromosomes were either distributed throughout the intact nucleus in multiple clusters before NEBD (43% of oocytes) ( Fig. 1a and Supplementary Video 1), or associated with the nucleolus and hence partially clustered before NEBD (57% of oocytes) (Extended Data Fig. 1a and Supplementary Video 1). In both cases, chromosomes moved towards each other rapidly and typically formed a cluster within 30 min after NEBD (Fig. 1b,c and Extended Data Fig. 1a). The clusters were of a the meiotic spindle 12,13 . In Xenopus oocytes, the chromosomes are transported by a transient microtubule structure that assembles from an acentriolar microtubule organizing centre on the cytoplasmic side of the nuclear envelope 15 . The transient microtubule array associates with condensed chromosomes, rapidly migrates towards the cell cortex and transforms into the first mitotic spindle 15 . Notably, starfish and Xenopus oocytes do not form chromosome clusters during meiosis I. Whether the mechanisms of chromosome movement in starfish and Xenopus oocytes are relevant to the chromosome clustering observed in human oocytes is unclear.
The actin cytoskeleton plays a role in positioning chromosomes during mitosis of human cells 16 . The actomyosin network in mitotic cells forms in the cytoplasm around the nucleus during prophase and contracts upon entry into mitosis, thereby reducing the nuclear volume and consequently the volume occupied by the chromosomes 16 . Similar to starfish and Xenopus oocytes, this facilitates the capture of chromosomes by microtubules upon NEBD 16 .
In this Article, we investigated the mechanism and function of chromosome clustering in human and porcine oocytes. We used the porcine model system, as oocytes and early embryos of pigs closely resemble those of humans 7,[17][18][19][20][21][22] . Additionally, pigs resemble humans genetically, physiologically and anatomically [23][24][25] . Although mouse oocytes are a commonly used mammalian model, unlike human Article https://doi.org/10.1038/s41556-022-01082-9 similar volume in different oocytes (Fig. 1b). Ultimately, 89% of human oocytes formed a compact chromosome cluster before the onset of spindle formation (Fig. 1a,d,e). Studies in human oocytes are limited by the scarcity of available material. We hence studied meiosis in live porcine oocytes, which closely resemble human oocytes 7,[17][18][19][20][21][22] . Spindle organization is similar in porcine and human oocytes 7 . Moreover, we found that spindle assembly was also similar in porcine and human oocytes. After the onset of NEBD, the chromosomes moved into a cluster (Fig. 2a, Extended Data  Fig. 1b). The time between NEBD and microtubule aster formation, however, was only about 1.6 h in porcine oocytes (Extended Data Fig. 1c) compared with ~5 h in human oocytes 5 . Like in human oocytes 5 , spindle assembly in porcine oocytes depended on the small GTPase Ran (Fig. 2b-d and Supplementary Video 3), whose inhibition led to a massive reduction in spindle volume (Fig. 2b) and a delay in the onset of microtubule spindle assembly (Fig. 2c) mechanisms in porcine and human oocytes are closely related and involve chromosome clustering. We then studied the chromosome clustering phase in porcine oocytes in more detail and found that the chromatin distribution in the nucleus of pig oocytes before NEBD was similar to that in human oocytes (Supplementary Video 2). In most porcine oocytes, the chromosomes were scattered throughout the nucleus before NEBD (Fig. 2e and Extended Data Fig. 2a), although they partially associated with the nucleolus in some oocytes (Extended Data Fig. 3a,b). The chromosomes moved rapidly towards each other during the first ~30 min post-NEBD in porcine oocytes, as in human oocytes, (Fig. 2e-g and Supplementary Video 2), with an average speed of 0.43 µm min −1 (Fig. 2g). Subsequently, the chromosomes slowed down, until a cluster had formed (Fig. 2g). Fifty-six per cent of porcine oocytes eventually formed a chromosome cluster that comprised all chromosomes ( Fig. 2e and Extended Data Fig. 2c). The remaining 44% formed a large chromosome cluster with one or a few chromatin foci located nearby (Extended Data Fig. 2c,d). The progression and degree of chromosome clustering were thus similar in porcine and human oocytes, although the overall speed of the chromosomes was slightly higher in porcine oocytes, particularly during the second slower phase of chromosome clustering.

Actin and microtubules drive chromosome clustering
Actin or microtubules move chromosomes into the proximity of the microtubule spindle in starfish 13 and Xenopus 15,27 oocytes, respectively. To test whether actin and microtubules play a role in chromosome clustering in porcine oocytes, we treated oocytes with either cytochalasin D (actin depolymerization drug) or nocodazole (microtubule depolymerization drug), or with both drugs simultaneously (Fig. 3a), and compared the speed (Fig. 3b) and the efficiency of chromosome clustering ( Fig. 3c) with that of control oocytes.
During the first 30 min post-NEBD, the chromosomes moved rapidly in both control oocytes, and oocytes treated with nocodazole ( Fig. 3b). However, the speed of clustering was significantly reduced in cells treated with cytochalasin D and with a combination of the drugs (Fig. 3b). Between 30 min and 50 min post-NEBD, the speed of chromosome clustering of oocytes treated with cytochalasin D closely resembled that of control oocytes, whereas in oocytes treated with nocodazole and with the combination of the drugs the speed of clustering was significantly reduced (Fig. 3b). These data suggest that the first phase of chromosome clustering, which lasts between 0 min and 30 min post-NEBD, is driven by the actin cytoskeleton. The second phase, starting from 30 min post-NEBD, is dependent on microtubules.
Next, we asked whether the drug treatments perturbed the efficiency of chromosome clustering. Treatment with either nocodazole or cytochalasin D alone led to a small increase in dispersed or uncaptured chromosomes after NEBD (Fig. 3c). Co-treatment with both drugs had a strong effect and blocked chromosome clustering in half of the oocytes (Fig. 3c). Residual chromosome clustering in oocytes co-depleted for F-actin and microtubules could be driven by the oocytes' nuclear lamina, which persisted after NEBD for a prolonged time in both porcine (Extended Data Fig. 3a) and human oocytes (Extended Data Fig. 3b), and coalesced together with the chromosomes into a cluster when actin and microtubules were absent in porcine oocytes (Extended Data Fig. 3a and Supplementary Video 4). Thus, actin-and microtubule-dependent chromosome clustering mechanisms act cooperatively to ensure efficient chromosome clustering.

Clustering promotes capture of chromosomes by microtubules
Next, we investigated the physiological function of chromosome clustering. Pharmacological inhibition of actin and microtubule polymerization led to an increase in the number of oocytes that failed to cluster their chromosomes (Fig. 3c), but the persistent presence of the drugs in this experiment precluded the analysis of effects on subsequent spindle assembly. To overcome this limitation, we transiently treated oocytes with cytochalasin D and nocodazole during the chromosome clustering phase, followed by washout of the drugs (Fig. 3d). At NEBD, chromosomes in both DMSO-treated control oocytes and oocytes co-treated with nocodazole and cytochalasin D were similarly distributed, demonstrating that depolymerization of F-actin and microtubules did not affect chromosome distribution (Extended Data Fig. 3c). Oocytes with dispersed chromosomes were washed into drug-free medium after NEBD, and imaged live as they progressed through meiosis (Fig. 3d).
All chromosomes were captured by the spindle in 100% of control oocytes (Fig. 3e,f). In contrast, 29% of oocytes transiently co-treated with nocodazole and cytochalasin D (Fig. 3e,f) showed defects during chromosome capture; chromosomes either failed to be captured altogether (6% of oocytes), leading to spindles that did not contain the full set of chromosomes (Fig. 3f); or chromosome capture was severely delayed (23% of oocytes) (Fig. 3g), and chromosomes remained outside of the spindle region for prolonged time periods. Thus, chromosome clustering promotes the rapid and complete capture of chromosomes by the assembling spindle.

Clustering is driven by a dynamic network of actin cables
Having established the importance of chromosome clustering for the formation of the first meiotic spindle, we next investigated the mechanisms that drive chromosome clustering.
Our experiments established that the first phase of chromosome clustering is dependent on F-actin. Additionally, in 57% of live porcine oocytes, chromosome clustering was completed before microtubule nucleation became apparent (Extended Data Fig. 2c). This suggests that chromosome clustering is predominantly driven by F-actin.
To visualize actin filaments as they drive chromosome clustering, we expressed the F-actin probe EGFP-UtrCH (enhanced green fluorescent protein fused to the calponin homology domain of human Utrophin) 28,29 in live porcine oocytes. Labelling of F-actin revealed the transient formation of interlinked actin cables in the nuclear region   that encapsulated the clustered chromosomes (Fig. 4a). Importantly, the actin cables were also detectable in fixed oocytes stained with phalloidin and are thus not an artefact caused by the EGFP-UtrCH probe (Extended Data Fig. 4a). Similar to porcine oocytes, fixed human oocytes displayed prominent actin cables encapsulating chromosomes shortly after NEBD ( Fig. 4b and Extended Data Fig. 4b).
High-temporal-resolution microscopy revealed that new actin cables appeared primarily in the periphery of the nuclear region, from where they migrated inwards towards the centre of the nucleus (Fig. 4c-e and Supplementary Video 5). The assembly and migration of actin cables repeated multiple times after NEBD (Supplementary Video 5). The actin cables did not form simultaneously along the entire periphery of the nuclear region, but rather sequentially in different regions of the nuclear periphery (Supplementary Video 5). The actin filaments were significantly more likely to move inwards than outwards or sideways (Fig. 4f). The chromosomes, too, moved towards the centre of the former nucleus, with their mean distance from the cell membrane remaining largely constant during clustering (Fig. 4g,h).
To confirm that dynamic actin filaments are required for efficient chromosome clustering, we treated porcine oocytes with the actin-stabilizing drug jasplakinolide. Jasplakinolide treatment significantly decreased the speed of chromosome movement during 0-30 min after NEBD (Extended Data Fig. 4c,d) and interfered with chromosome capture in some oocytes (Extended Data Fig. 4e,f). Together, these data establish that dynamic actin filaments are required for efficient chromosome clustering.
Non-muscle myosin II (nm-myosin II) is a main driver of subcellular actin-dependent motility. In human mitotic cells, nm-myosin II is required for chromosome positioning after NEBD 16 . However, inhibition of nm-myosin II did not have an effect on chromosome clustering (Extended Data Fig. 4g,h), suggesting that nm-myosin II is dispensable for chromosome movements after NEBD in porcine oocytes.

Actin cables interact with kinetochores during clustering
Next, we investigated how the chromosomes interact with the actin cables during clustering in porcine oocytes. Strikingly, up to 50% of kinetochores per oocyte interacted with actin filaments, with an average of 18% of kinetochores per oocyte, as evident from fixing oocytes at random stages of chromosome clustering (Fig. 5a,b and Extended Data Fig. 5a). We also observed close contact between actin and chromosome arms (Fig. 4a,b and Extended Data Fig. 4a,b). On average, 85% of chromatin foci had at least one actin cable on their surface in fixed porcine oocytes (Extended Data Fig. 5b).
Three-colour imaging of chromatin, kinetochores and actin revealed that kinetochores and actin cables moved together in the same direction while they interacted ( Fig. 5c and Supplementary Video 6). Kinetochore-actin interactions persisted for ~6-7 min in several cases, thus spanning a considerable fraction of the chromosome clustering period (Fig. 5c). Coordinated kinetochoreactin movements remained detectable in oocytes treated with the microtubule-depolymerizing drug nocodazole (Extended Data Fig. 5c and Supplementary Video 6), indicating that the interaction and movements are independent of microtubules. Additionally, we observed examples of chromosomes whose kinetochores re-oriented towards the nuclear centre during the actin-dependent phase of chromosome clustering, indicative of kinetochore-led chromosome movements ( Fig. 5d and Supplementary Video 7).
Next, we examined whether chromosome clustering requires kinetochore-actin interactions, or whether chromosome arm-actin interactions are sufficient. Disrupting kinetochores by Trim-Away 30,31 was not possible because this approach is inefficient for nuclear proteins 30 . Instead, we treated porcine oocytes with the DNA cleaving antibiotic zeocin to generate kinetochore-containing and kinetochore-free chromosome fragments, following a previously established experimental strategy 12 . Oocytes treated with zeocin progressed into anaphase (Extended Data Fig. 5d,e), showing that zeocin does not block oocyte maturation. To evaluate microtubule-independent movements, we co-treated oocytes with nocodazole (Extended Data Fig. 5f-h). Small kinetochore-free chromosome fragments (<30 µm 3 ) clustered more slowly than small kinetochore-containing chromosomes (<30 µm 3 ) in control oocytes ( Fig. 6a-c). In contrast, kinetochore-containing and kinetochore-free fragments that were larger than 30 µm 3 clustered with similar efficiency (Fig. 6d). This shows that larger chromosomes are efficiently moved by chromosome arm-actin interactions. Smaller, kinetochore-free fragments, however, cannot be clustered efficiently.
The small kinetochore-free chromosome fragments eventually coalesced with a prominent delay (Fig. 6a) that correlated with the compaction of the nuclear lamina (Extended Data Fig. 6a), consistent with our hypothesis that the compacting nuclear lamina can coalesce chromosomes that failed to be clustered by other mechanisms, as described above. We infer that kinetochore-actin interactions are required to accelerate the clustering of small chromosomes.

Actin-interacting kinetochores can bind microtubules
In starfish oocytes, the actin nucleation factor Arp2/3 assembles actin patches on chromosomes during chromosome migration, which sterically prevent the premature association of kinetochores with microtubules 12 . Similar actin patches do not exist in porcine oocytes (Figs. 4a,b and 5a and Extended Data Fig. 5a). Nevertheless, we considered that the actin cables that interact with kinetochores might also inhibit the association with microtubules. However, we found that the opposite was the case, the attachment of microtubules to kinetochores was lower in cells treated with cytochalasin D compared with DMSO-treated control oocytes (Fig. 6e,f). Additionally, co-staining of actin and microtubules in control oocytes revealed a fraction of kinetochores that was attached to microtubules while also interacting  with actin (Extended Data Fig. 6b,c). The fraction size corresponded roughly to what was expected for random overlap between actin-and microtubule-interacting kinetochores (Extended Data Fig. 6d). Thus, kinetochore-actin interactions do not block the formation of kinetochore-microtubule attachments in porcine oocytes.

Formin-2 and Spire mediate actin cable assembly
Next, we aimed to identify the F-actin assembly factors that drive the formation of actin cables in porcine oocytes. The cables were long and unbranched in nature (Figs. 4a-c and 5a,c), suggesting that their assembly is probably driven by formins rather than the Arp2/3 complex, which typically generates dense actin meshworks 12,32,33 . Indeed, treating porcine oocytes with the Arp2/3 inhibitor CK-666 had no effect on the speed of chromosome clustering between 0 min and 30 min post-NEBD (Extended Data Fig. 7a,b), indicating that Arp2/3 is dispensable for chromosome clustering.
Diaphanous-related formins assemble unbranched actin structures and rely on activation by Rho-GTPases 34,35 . To test their potential involvement in chromosome clustering, we inhibited Rho-GTPases with toxin B. We observed a full block of Rho-GTPase-dependent cytokinesis Article https://doi.org/10.1038/s41556-022-01082-9 during polar body extrusion (Extended Data Fig. 7c,d), indicating efficient inhibition of Rho-GTPase activity. However, the speed of chromosome clustering was not decreased (Extended Data Fig. 7e,f). Similarly, inhibition of the Rho-GTPases RhoA, RhoB and RhoC with exoenzyme C3 fully blocked cytokinesis during polar body extrusion (Extended Data Fig. 7g,h), but did not decrease the speed of chromosome clustering (Extended Data Fig. 7i,j). We infer that chromosome clustering is not driven by diaphanous-related formins.
Formin-2 is well established to assemble actin structures in mouse oocytes and zygotes 29,[36][37][38][39] . We found that Formin-2 was enriched in the nucleus of immunostained porcine oocytes (Extended Data Fig. 8a), making it an attractive candidate for the assembly of the actin cables. Formin-2 co-localized with actin filaments in the cytoplasm, suggesting that the antibody used for the immunofluorescence analysis was specific (Extended Data Fig. 8b). Endogenous expression of Formin-2 in porcine oocytes was further confirmed by western blot (Extended Data Fig. 8c).   Formin-2-dependent actin nucleation can be inhibited by expression of FH2-and KIND-domain constructs, which block the interaction between Formin-2 and Spire, an actin assembly factor that cooperates with Formin-2 to form actin filaments 36,[40][41][42] . Expression of FH2 and KIND in porcine oocytes significantly decreased the speed of chromosome clustering (Fig. 7a-d). Moreover, FH2 expression reduced the abundance of actin cables in the nuclear region (Fig. 7e,f). Interestingly, FH2 expression led to a partial block of cytokinesis during polar body extrusion, suggesting that Formin-2 and Spire are involved in polar body extrusion in porcine oocytes (Extended Data Fig. 8d,e). We conclude that the chromosome-interacting actin cables are assembled by Formin-2 and Spire.

Microtubule loops transport residual unclustered chromosomes
To investigate how microtubules transport the chromosomes that were not clustered by actin cables by 30 min post-NEBD, we performed high-resolution microscopy of fixed porcine oocytes. We observed a cage of microtubule loops that encapsulated the chromosome cluster 30 min post-NEBD ( Fig. 8a and Supplementary Video 8). Similar to kinetochore-associated microtubules, these microtubule loops were not depolymerized by briefly placing the oocytes on ice (Extended Data Fig. 9 and Supplementary Video 9), indicating that the microtubule loops are stable. We detected a similar cage of microtubule loops in human oocytes as well, as early as 30 min after NEBD ( Fig. 8b and Supplementary Video 8), suggesting that the formation of a microtubule cage is conserved between species.
The microtubule loops formed only during later stages of chromosome clustering (Fig. 8a,b). Thus, in the absence of F-actin, the microtubule loops transported the chromosomes with a notable delay (Figs. 3a and 8c). Live imaging of microtubules in cytochalasin D-treated porcine oocytes revealed that the chromosomes interacted with microtubule loops and clustered as the network contracted (Fig. 8c). These data establish that microtubule loops can move the chromosomes when F-actin-based clustering fails.

Discussion
Here we report that chromosome clustering before spindle assembly in human and porcine oocytes prevents chromosome losses and facilitates the rapid and complete capture of all chromosomes by the assembling spindle. Our data indicate that chromosome clustering is safeguarded by multiple, partially redundant mechanisms, underscoring its importance (Fig. 8d).
Actin and microtubules were reported to drive chromosome transport between NEBD and the onset of spindle assembly in other species and cell types, where chromosome clusters do not form. Importantly, all of these mechanisms are distinct. The actomyosin network that forms around the disassembling nuclear envelope in human mitotic cells 16 is located on the cytoplasmic side of the nuclear envelope 16 and requires nm-myosin II activity to move the chromosomes 16 . Despite these mechanistic differences, the network in mitotic cells was proposed to reduce the nuclear volume in order to facilitate chromosome capture, suggesting functional similarities.
Starfish oocytes, unlike human and porcine oocytes, rapidly assemble a spindle with centrosomes and do not form a chromosome cluster. However, they too assemble an actin network in the nuclear region upon NEBD 13 . The actin meshwork does not interact specifically with kinetochores and has different dimensions from the actin cables that cluster chromosomes in mammalian oocytes as described here [12][13][14]43 . Moreover, in starfish oocytes, the chromosomes are coated by patches of Arp2/3-nucleated actin that shields chromosomes from microtubules 12 . In contrast, the chromosomes are neither coated nor shielded by actin in mammalian oocytes. Nevertheless, actin is essential for accurate progression through meiosis in all three species.
In Xenopus oocytes, a ~200-µm-wide array of microtubules moves across the nucleus to gather the chromosomes 15,44,45 . The microtubule loops we observed in porcine and human oocytes are morphologically very different from the microtubule array described in Xenopus oocytes 15,27 . However, it remains to be investigated whether there are similarities in the molecular players that drive the assembly of these two structures.
Taking all evidence together, there appears to be a conserved challenge that oocytes of various species must cope with. Their chromosomes are typically located within large nuclei, and they often lack centrosomes for spindle assembly, instead employing a chromosomedriven, Ran-dependent spindle assembly pathway. Work in mouse and Xenopus extracts shows that chromosome-driven spindle assembly results in the formation of multiple spindles in the presence of spatially separated chromosomal foci 46-48 . Oocytes thus have a conserved need to gather the chromosomes into one place, to prevent chromosome losses and to assemble a single spindle.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions

Ethics approval
The use of immature unfertilized human oocytes in this study has been approved by the UK's National Research Ethics Service under the REC Reference 11/EE/0346 (IRAS Project ID 84952). The porcine oocytes used in this study were obtained from a local abattoir as a waste product of the slaughtering process, and as such do not require ethics approval for usage in Germany where the study was conducted.

Primary porcine oocytes
All porcine oocytes were collected from pre-pubertal animals of female sex (exact age unknown). Porcine ovaries were obtained from a local abattoir and transported to the laboratory in a thermo-flask. Cumulus-oocyte complexes (COCs) were aspirated from 2-6-mm-large antral follicles using a 17-gauge needle affixed to a 1 ml disposable syringe. The aspired follicular fluid was collected into a 50 ml Falcon tube containing 2 ml of M2 medium with 1 mM dbcAMP. COCs were allowed to sediment for 15 min, and then washed extensively through four 35 mm Petri dishes, each containing 2 ml of pre-warmed M2 medium with 1 mM dbcAMP. Only fully grown oocytes with a homogeneous cytoplasm and at least three to five complete layers of compact cumulus cells were selected for the experiments. COCs were denuded using a series of transfer pipettes with a tip of defined diameter (175 µm, 145 µm and 135 µm) attached to an EZ-Grip denudation pipettor. Oocytes were cultured in M16 (5% CO 2 ) or M2 medium at 38.5 °C. In most experiments, prophase arrest was maintained by supplementing the culture medium with dbcAMP. For experiments involving timed fixations and a few live cell imaging experiments specified below, cells were cultured for approximately 12 h in medium supplemented with 50 µM Butyrolactone I (Abcam, ab141520; 50 mM stock in ethanol). To induce the resumption of meiosis, oocytes were washed into dbcAMP/ Butyrolactone I-free M2 medium.

Primary human oocytes
All human oocytes obtained for this study were sourced from patients of female sex undergoing fertility treatments at Bourn Hall Clinic (Bourn, Cambridgeshire, UK) between 12 January 2017 and 24 July 2019 after having obtained fully informed consent. All donations were anonymous. Patients were not monetarily compensated for their donation. Ages of human oocyte donors from Fig. 1 and Extended Data Fig. 1a Table 2. Fifty-six germinal vesicle (GV) stage oocytes from 39 donors were obtained for immunofluorescence analysis. All donors underwent ovarian stimulation for intracytoplasmic sperm injection. Only oocytes that were immature and hence unsuitable for intracytoplasmic sperm injection were donated to the study. None of the oocytes used in this study were freeze-thawed. Oocytes were cultured in G-MOPS medium (Vitrolife, 10129) supplemented with 10% FBS (GIBCO, 16000044) under mineral oil (Merck, 8012-95-1) at 37 °C. Only oocytes that appeared morphologically healthy and underwent NEBD within 24 h upon retrieval were included in the study. NEBD of the GV was observed using a Primo Vision Evo+ timelapse camera (Vitrolife) installed inside the incubator. Thirty-nine oocytes were fixed at defined timepoints around NEBD. NEBD was scored by the disappearance of the GV observed using the Primo Vision Evo+ and confirmed under the stereomicroscope. Thirteen cells were fixed in the GV stage, 25 cells were fixed within 30 min of NEBD and 1 cell was fixed 1 h after NEBD. The remaining 17 oocytes had NEBD shortly before they were obtained from the IVF clinic. All 56 human oocytes were co-stained with an anti-α-tubulin antibody (MCA78G, Bio-Rad; 1:3,000) to visualize microtubules, an anti-LaminA/C antibody (Sigma-Aldrich, SAB4200236; 1:50) to visualize the nuclear lamina, 100 µM Hoechst 33342 (Molecular Probes; 20 mM stock) to stain DNA, and phalloidin conjugated with Alexa Fluor 488 or Rhodamine (Invitrogen; 1:20) to label F-actin. Twenty-seven oocytes had visible actin cables encapsulating the chromosomes, 21 oocytes had a microtubule meshwork encapsulating the chromosomes, four had already started to assemble a spindle, two had a bipolar spindle and two cells were not well fixed as evident from a poor preservation of the actin and microtubule cytoskeleton following immunofluorescence analysis. The images and graphs in Fig. 1 and Extended Data Fig. 1a were generated by analysing videos from 35 live human oocytes that were recorded as part of a previous study 5 . All cells included in the analysis underwent NEBD within 24 hours of retrieval, formed MI spindles and progressed through anaphase I.
Exoenzyme C3 (Cytoskeleton, CT03) was reconstituted in embryo tested water to obtain a 10 µM stock. Oocytes were microinjected with 10 pl of 2.4 µM exoenzyme C3, or the equivalent amount of bovine serum albumin (BSA, Sigma, A3311). Clostridium difficile toxin B (Sigma-Aldrich, SML1153) was dissolved in RNase free water to obtain a 1.5 µM stock. Oocytes were microinjected with 10 pl of 100 nM toxin B, or an equivalent amount of BSA.

Zeocin preparation
Zeocin powder (InvivoGen, ant-zn-1p) was dissolved in HEPES buffer at a stock concentration of 100 mg ml −1 (stock stored at 4 °C). To induce chromosome fragmentation, porcine oocytes were treated for 15 h with 500 µg ml −1 of zeocin diluted in M16 medium. To stop chromosome fragmentation, cells were transferred into zeocin-free M2 medium.

Cold-mediated microtubule depolymerization assays
Porcine oocytes were incubated on ice for 7 min and immediately fixed and processed for immunofluorescence microscopy as described below.

Immunofluorescence
To obtain oocytes at defined timepoints after NEBD, cells were monitored on a confocal microscope with transmission, or with a Primo Vision system. NEBD was determined as the moment when the nucleoli began to shrink.

Confocal microscopy
Confocal imaging was performed in ~20 µl of M2 medium (for live oocytes) or PBS with 1% polyvinylpyrrolidone (for fixed oocytes) under paraffin oil in a 35 mm dish with a #1.0 glass coverslip. Images were acquired with LSM800, LSM880 and LSM900 confocal laser scanning microscopes (Zeiss) equipped with an environmental incubator box and a 40× C-Apochromat 1.2 NA water-immersion objective. Live samples were imaged at 38.5 °C, and fixed at 25 °C. Automatic 3D tracking was implemented for time-lapse imaging with a temporal resolution between 1.5 min and 10 min using AutofocusScreen 57 or MyPiC 58 on the LSM880, or by using a customized sample tracking solution provided by ZEISS Microscopy on LSM800 and LSM900 microscopes. In all experiments, control and experimental groups were imaged together, under identical imaging conditions on the same microscope. For some images presented in the figures, noise was reduced with a Gaussian filter in ZEN (Zeiss). Airyscan images were acquired using the airyscan module on LSM800, LSM880 and LSM900 confocal laser scanning microscopes (Zeiss) and processed in ZEN (Zeiss) after acquisition. Care was taken that imaging conditions (laser power, pixel-dwell time and imaging frequency and detector gain) did not cause phototoxicity (for live imaging), photobleaching or saturation. The actin cortex was saturated in experiments where F-actin in the cytoplasm was imaged.

Automatic detection of NEBD in porcine oocytes on a confocal microscope
The resumption of meiosis in porcine oocytes is asynchronous. Thus, to perform high-resolution spatial and temporal imaging of F-actin after NEBD, we designed a macro that allowed us to automatically detect NEBD.
To determine the timing of NEBD, we took advantage of a fluorescently labelled (Alexa Fluor 647 NHS Ester (succinimidyl ester); Invitrogen, A20006) 40 kDa dextran (Molecular Probes, A20006), which localizes to the cytoplasm in GV oocytes but upon NEBD quickly distributes within the nucleus. We recorded coarse resolution images of the dextran in the early stages of meiosis. The acquired images were read and analysed using the Bioformats 59 package within a MATLAB Compiler Runtime Environment. To evaluate if NEBD had initiated, a sphere of 10 µm in diameter was placed at the centre of mass of the chromosome signal, and the average intensity of the dextran within this sphere was computed at every timepoint. A switch to high-resolution image acquisition was triggered when the intensity within the sphere became higher than a variable threshold. The threshold was determined for each oocyte at every timepoint individually, and was based on the median and the standard deviation of previous measurements. The script keeps the time interval for each imaged oocyte as the image acquisition interval set in the Zen software (Zeiss). The script was designed to work with the Zen Blue (Zeiss) image acquisition software.

Light sheet microscopy
Light sheet imaging was performed on an LS1 Live light sheet microscope system by Viventis. The illumination optics consist of two opposing Nikon CFI Plan Fluor air illumination objectives. We used a Nikon CFI75 Apochromat 25xC W NA 1.1 water immersion objective with switchable magnification between 18.75× and 37.5× (field of view 700 µm and 340 µm, respectively) as detection objective. The microscope had a fast motorized six-position filter wheel with emission filters in front of the camera. The following laser lines and emission filters were used: Scientific, NP0008) with 100 mM dithiothreitol was then added to the sample to a final 1× concentration, and the mixture was immediately snap-frozen in liquid nitrogen. Samples were thawed and frozen thrice more, followed by a 5 min incubation at 95 °C. Samples were resolved on a 10-well NuPAGE

General quantifications
Time-lapse videos of live porcine oocytes were analysed using Imaris 8.4.1 or Imaris 9.3.1 (Bitplane). NEBD in human and porcine oocytes was defined as the last time point before the nucleolus begins to visibly shrink/disappear.

Automatic creation of chromosome isosurfaces in Imaris
To consistently reconstitute the chromosome isosurfaces in Imaris, we used a MATLAB script published elsewhere 56 .

Automatic quantification of chromosome clustering speed
To calculate the speed of chromosome clustering, we first calculated the largest pairwise distance measured between all chromosomes for every timepoint. The change of the largest pairwise distance between the timepoints was then used to derive the speed of chromosome clustering.
To consistently and reliably quantify the largest pairwise distance between all chromosomes, we developed an in-house MATLAB script.
The script requires that isosurfaces are defined as vertices connected into polygons, and hence works only with versions older than Imaris 9. The script computes distances between all pairs of vertices via the 'pdist' MATLAB function. The largest distance is then selected and depicted in Imaris as a pair of spots. The largest distance is also added as a custom statistic to the isosurface object. For usage within Imaris, the script was compiled for MATLAB Compiler Runtime v95.

Automatic creation of a convex hull from 3D reconstituted chromosomes
To create a 3D convex hull, we used an in-house-developed MATLAB script. The script requires that isosurfaces are defined as vertices connected into polygons, and hence works only with versions older than Imaris 9. The script uses the MATLAB function 'convhulln' to identify the subset of vertices of a given isosurface making up the facets of the corresponding convex hull of this isosurface. For usage within Imaris, the script was compiled for MATLAB Compiler Runtime v93.

PIV analysis from high-temporal-resolution F-actin imaging after NEBD
To measure velocity vectors within the nuclear area, we first determined the boundaries of the nucleus based on EGFP-Lamin B1 signal. For this, the signal of the nuclear lamina was segmented using edge-aware local contrast enhancement followed by edge detection. The resulting fragmented and noisy segmentation was translated to polar coordinates and the outliers were removed. The centre of the polar coordinate system was defined by the centre of mass of the segmented nuclear lamina. Closing of the membrane was achieved by fitting a smoothing spline to the coordinates theta and rho of the remaining pixels.
The smoothing spline was expanded by 1.5 µm or 2.5 µm per direction to produce a segmented image of the nuclear lamina.
The analysis was performed in MATLAB using 'localcontrast', 'edge' (with Canny algorithm), 'rmoutliers' and 'fit' functions. The exact parameters for these functions had to be adjusted in a few cases of lower signal-to-noise ratio. To import the images into MATLAB R2018b, we used the toolbox Bioformats 59 .
To distinguish movement of the entire cell or nucleus from movement of actin, the images were stabilized with the option 'translation' using the Zeiss Zen function 'time alignment' on the signal of the nuclear lamina.
Actin velocities were detected using PIVlab 60 in MATLAB R2018b on the stabilized images. The Wiener2 denoise and low pass filters with 5 pixel window size were used for pre-processing. The particle image velocimetry (PIV) algorithm used was 'FFT window deformation'. The interrogation windows were reduced in three steps of 64, 32 and 24 pixels width. The correlation robustness was set to 'extreme'. All options for post-processing were disabled, which resulted in weak filtering of the velocity vectors.
The direction of the velocity vectors was set relative to the centre of the nucleus. The centre of the nucleus was determined on the basis of the centre of mass of the widened smoothing spline. The segmentation of the nucleus was used to identify which vectors were located inside the nucleus, but not to mask the nuclear membrane for the PIV analysis.
Circular statistics were computed using the Circular Statistics Toolbox 61 . Mean directions and the 95% confidence intervals are based on the velocity vectors with speeds exceeding two pixels per frame or 1.8 µm min −1 .

Semi-automatic analysis of chromosome clustering upon chromosome fragmentation
H2B-positive foci were detected, counted and analysed using the surface function in Imaris 9.6.0 (Bitplane). Foci were classified manually as kinetochore-containing or kinetochore-free on the basis of the presence or absence of the hCenpC signal, respectively. The total time required for complete chromosome clustering was calculated as the time between NEBD and the first timepoint when only one H2B-positive surface was detected. Cells were scored as having completed chromosome clustering only if all the H2B-foci formed a single surface during the duration of imaging. Individual H2B foci with and without a kinetochore were tracked manually to estimate the time required for their clustering. Only foci that could be reliably tracked were analysed. Chromatin surrounding the nucleolus was excluded from the analysis of individual H2B foci.

Co-localization analysis for Formin-2 and F-actin
The specificity of the anti Formin-2 antibody was tested using the statistic 'Shortest distance to Surface' computed using Imaris 9.7.2 and Imaris 9.3.0 by Bitplane.
For this purpose, spots were created once on the Formin-2 signal and once on the Formin-2 signal mirrored along the x axis. These spots were compared relative to a surface segmentation of the actin signal (Phalloidin). Before the analysis, a region of interest of 10 µm × 10 µm × 10 µm was manually cropped. The region of interest was placed so that it included neither the nucleus nor the very edge of the image. Around 100 spots were created per cell to improve comparability.
The surface on actin was created with a batch script previously published elsewhere 56 using Otsu's method for thresholding and ensuring the volume of the actin surface to be within 30 µm³ to 1,000 µm³.

Manual quantification of the mean fluorescence intensity of actin filaments after NEBD
Quantification of the mean fluorescent intensity of actin filaments was done in ImageJ on two-dimensional images. To quantify the mean Nature Cell Biology Article https://doi.org/10.1038/s41556-022-01082-9 fluorescence intensity of actin filaments after NEBD, a region of interest of consistent size was placed next to the chromosome signal. Background was subtracted on the basis of intensity measurements in a region outside of the cell.

Statistics and reproducibility
Average (mean), standard deviation and statistical significance were calculated in GraphPad Prism (8.4.3). Statistical tests selected for each dataset are indicated in the figure legends. All data were tested for normal distribution in GraphPad Prism (8.4.3) before selection of the statistical tests. All graphs were generated in GraphPad Prism (8.4.3), except for the graphs in Fig. 4d, which were generated in Origin 2021b, and in Fig. 4e, which were generated in MATLAB R2018b. All data from porcine oocytes are from at least two independent experiments/multiple biological replicates. The number of experimental replicates for each assay is specified in the figure legends. All replication attempts were successful. No statistical methods were used to pre-determine sample size. Due to the large size and opaque nature of porcine oocytes, cells that had nuclei positioned in the top third of the cell could not be imaged well and were hence excluded from the analysis. Cells that died during imaging were excluded from the analysis. For each independent experiment, porcine oocytes were collected from multiple ovaries/ animals and subsequently pooled together. The oocytes were then randomly split into a control group and an experimental group. The investigators were not blinded to allocation during experiments and outcome assessment.

Resource availability
Contact for reagent and resource sharing. Further information and requests for resources and reagents should be directed to and will be fulfilled by the corresponding author Melina Schuh (melina.schuh@ mpinat.mpg.de).

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
Source data are provided with this paper. All other data supporting the findings of this study are available from the corresponding author on reasonable request. The primary microscopy data were not uploaded to a data repository due to their large size, but are available from the corresponding author on reasonable request.