Reduction of cortical pulling at mitotic entry facilitates aster centration

Asters are intracellular structures of microtubules (MTs) that are radially organized around an MT-organizing center (MTOC), generally a centrosome (Bornens, 2012), or that are self-organized in presence of molecular motors (Mitchison and Field, 2021; Nédélec et al., 1997). The cell might contain a single aster, as in the case of the sperm aster, or two asters following centrosome duplication, which localize at each pole of the mitotic cell. Asters are involved in many essential functions in the cell including intracellular trafficking and organization (Hamaguchi et al., 1986), polarity establishment (Wodarz, 2002), guidance of nuclei (Reinsch and Gönczy, 1998) and determination of the cleavage axis during cell division (Devore et al., 1989; Strome, 1993). Although we have known for more than a century that the protoplasmic mass of the cell becomes equally segregated to the two daughter cells following cell division (Hertwig, 1884), the precise details of the mechanism of migration and centration of the mitotic apparatus is still not fully resolved.

Elegant experiments in amphibians, echinoderms and Caenorhabditis elegans zygotes, as well as in yeast and in vitro studies, have reported three possible mechanisms of how MTs can generate forces to displace nuclei and whole asters or MTOCs. One mechanism is based on cytoplasmic pulling (Li and Jiang, 2018; Minc and Piel, 2012; Minc et al., 2011; Pelletier et al., 2020; Wühr et al., 2009), one on cortical pulling (Grill et al., 2001; Kotak and Gönczy, 2013; Redemann et al., 2010) and one on MT pushing (Garzon-Coral et al., 2016; Laan et al., 2012; Meaders and Burgess, 2020; Tran et al., 2001). During centration of the sperm aster, cytoplasmic pulling exerts forces on the aster by taking advantage of the transport of organelles that travel on the MTs (Kimura and Kimura, 2011; Tanimoto et al., 2016). A cargo moving towards the MTs minus-end at the centrosome experiences an opposing drag force from the cytoplasm (Longoria and Shubeita, 2013; Palenzuela et al., 2020). Cytoplasmic pulling force thus scales with MT length as more cargoes cover longer MTs (De Simone et al., 2018; Kimura and Onami, 2005). Cortical pulling occurs when MTs contact a minus-end-directed molecular motor localized on the plasma membrane (Laan et al., 2012) often via interaction with cortical LGN/Pins/GPR1-2 and NuMA/Mud/Lin-5 (names for the equivalent three proteins in mammals and in Drosophila) complexes (di Pietro et al., 2016). Such minus-end-directed motor activity, supported by the rigidity of the actomyosin cortex, can pull on MTs, bringing the whole aster and mitotic apparatus (notably during anaphase of unequal cell division) towards the cell membrane (Grill and Hyman, 2005; Grill et al., 2001; Kiyomitsu, 2019). Finally, owing to the inherent dynamic instability of MTs (Burbank and Mitchison, 2006) and their polymerization against the membrane and its actomyosin-rich cortex (Rosenblatt et al., 2004), MTs can exert a pushing force in the opposite direction to MT growth (Sulerud et al., 2020) and this leads to displacement of asters away from the cell surface (Meaders and Burgess, 2020). Such force generation might be limited however by MT buckling when the MTs exceed ∼20 µm in length (Dogterom et al., 2005), although this would be more difficult to interpret if short branched MTs were also present (Field and Mitchison, 2018; Petry et al., 2013).

Striking examples of sperm aster or mitotic apparatus centration occur following fertilization and have been studied in a number of species. In most of these species, following fertilization, a sperm aster forms from the sperm-derived centriole at the site of sperm–egg fusion and thus near the plasma membrane and, as a consequence, far from the center of the fertilized egg (Ishihara et al., 2014). Interestingly however, the cell cycle phase during which aster migration occurs towards the center of the fertilized egg varies depending on species and is independent of the timing of fertilization. For example, sperm aster centration occurs during the first interphase in sea urchin immediately after fertilization (Minc et al., 2011) and in mouse oocytes MTOC centration occurs during interphase following sperm-induced exit of meiosis II (Scheffler et al., 2021), whereas it occurs during first prophase following fertilization at meiosis I in C. elegans (De Simone et al., 2018; Gönczy et al., 1999). In C. elegans, the sperm aster remains close to the cortex during interphase following fertilization. In human oocytes (Asch et al., 1995) as well as those of other primates (Hewitson and Schatten, 2002; Simerly et al., 2019), even though fertilization occurs at metaphase (Meta) II as in the mouse, centration occurs at entry into mitosis when the sperm aster has duplicated to become the first mitotic apparatus. It is currently unknown how centration of these sperm asters is (1) prevented during interphase and (2) triggered at mitotic entry.

To understand the mechanism of sperm aster and mitotic apparatus centration, we analyzed the fertilized eggs of the ascidian Phallusia mammillata, because ascidians form a sister group to the vertebrates (Delsuc et al., 2006). As in mammals, fertilization occurs during meiotic metaphase in the ascidian (Meta I for ascidians, Meta II for primates and humans). We discovered that the sperm aster in ascidians behaves in a similar manner to those in primates and humans: following fertilization the sperm aster grows yet remains cortical and does not migrate to the cell center during all of the first interphase and eventually migrates as a duplicated mitotic apparatus at mitotic entry. By combining experiments and numerical simulations, we investigated how the asymmetrically positioned sperm aster is prevented from migrating towards the center of the zygote during interphase and in addition how centration is eventually triggered at mitotic entry. We found that cortical pulling is switched off at mitotic entry facilitating aster centration at mitotic entry.

To monitor sperm aster and then mitotic apparatus migration while observing cell cycle stages during the first cell cycle, oocytes of Phallusia were co-injected with mRNAs encoding the MT-binding protein Ensconsin (Ens::3GFP) and Tomato-tagged histone H2B (H2B::Tom). These oocytes were fertilized and imaged from 10 min post fertilization (mpf) until cytokinesis. We observed that the aster changes shape, size and position throughout the cell cycle as detailed in the next paragraph (Fig. 1A; Movies 1 and 2). These changes were associated with cell cycle stages based on recognizable events: polar body (PB) extrusions, pronuclei (PN) formation, female PN migration, nuclear envelope breakdown (NEB), prometaphase, metaphase, anaphase and cytokinesis (Fig. 1B).

During the completion of meiosis, the sperm aster grows in the vegetal hemisphere, on the opposite side of the fertilized egg from the meiotic spindle which defines the animal pole (Roegiers et al., 1995). During meiosis I, the aster is in close proximity to the egg cortex (Dumollard and Sardet, 2001) whereas at meiosis II (7 to 20 mpf), the aster is spherical and located a few micrometers (5–10 µm) from the egg cortex. Upon entering interphase, a highly asymmetric sperm aster formed near the cortex and it did not migrate to the cell center (Fig. 1A,B). At this time, MTs grew and extended into the interior of the zygote (to capture the female PN located on the opposite side of the zygote at the animal pole) creating a highly asymmetric sperm aster with long MTs towards the cell center and animal pole, and shorter MTs towards the vegetal pole and nearest cortex (Fig. 1A; Movies 1 and 2). The centrosome then duplicated, and the two resulting centrosomes positioned on each side of the DNA at entry into mitosis where the mitotic apparatus centered (Fig. 1A, ‘NEB and prometaphase’). To summarize, the sperm aster becomes asymmetric during interphase yet remains cortical and does not migrate until entry into first mitosis. For clarity we refer to the globality of the aster and subsequent mitotic apparatus and spindle migration as ‘aster migration’.

Aster migration was quantified by measuring the distance between the cell center and the male DNA from 28 different embryos at several cell cycle stages (Fig. 1B; Fig. S1A). Given that there is no appreciable central migration of the sperm aster during all of meiosis and also because there are zygote changes shape due to ongoing Ca²⁺ oscillations (McDougall and Sardet, 1995), we focused on sperm aster location during interphase and first mitosis (Fig. 1B). In the embryos pooled in the dataset, the DNA was labeled either by expressing H2B::Tomato or H2B::Venus or by staining with Hoechst 33342 in live zygotes. During interphase the distance between the male DNA and the cell center did not change significantly (Fig. 1B), again indicating that the aster did not significantly center during interphase. Surprisingly, the main movement of centration occurred after PN fusion at mitosis onset. The distance of the DNA from the cell center decreased between the events of PN fusion and NEB and further declined between NEB and prometaphase or metaphase (Fig. 1A,B). Thus, the sperm aster remained close to the cortex during interphase and finally the aster started centering at mitosis entry (Fig. 1A,B). This temporal sequence of aster migration was confirmed by the quantification of fixed samples from batches of zygotes sampled every 10 min and immunostained to label MTs and DNA (Fig. S1A,B).

The main movement of migration started at mitosis entry even though the sperm aster was highly asymmetric with long MTs oriented into the interior of the zygote and shorter MTs oriented towards the vegetal cortex (Fig. 2A). Therefore, we investigated the causal role of mitosis entry on the aster migration by delaying entry into first mitosis by injection of a cyclin-dependent kinase (CDK) inhibitor (p21) coupled with GFP. The presence of p21 perturbs CDK activity (Levasseur et al., 2007). The injection of p21 protein did not disrupt the formation of the cortical interphase aster with long MTs extending into the zygote interior, and p21::GFP accumulated in the nuclei (Fig. 2A, right panel). We again monitored DNA position as a read out for aster migration rather than aster geometric center given that the aster shape changes considerably and also duplicates at mitosis entry complicating the measurements.

In zygotes with an unaltered cell cycle (Fig. 2B, top panel), the pronuclei formed, the female PN (orange arrow) migrated and reached the male PN (yellow arrowhead), then the mitotic apparatus migrated towards the cell center during mitosis (t=17 min post PN formation). In the p21-injected zygotes with prolonged interphase (Fig. 2B, bottom panel), the male and female pronuclei formed and the female PN migrated toward the male PN until it reached the centrosome (Fig. 2B, t=16 min) or the male PN. However, the aster and the male PN never migrated, and the aster remained asymmetric and uncentered (Fig. 2B, t=24 min). The injection of p21 was considered successful when interphase (here approximated by the time between PN formation and NEB) lasted more than twice the mean duration in control zygotes or greater than 24 min (Fig. 2C).

Quantification of the aster migration in the two conditions revealed that in control zygotes the distance between the cell center and the male PN decreased from 45 µm to 30 µm (mean±s.e.m.) from interphase to mitotic entry (Fig. 2D). In contrast, for the same timing at interphase in p21-injected zygotes the distance remained at 45 µm (mean±s.e.m.), showing an absence of aster migration. Thus, prolonging interphase prevents DNA centration and aster migration.

Given that the aster specifically centers at mitotic entry, we wondered what could cause the absence of migration in the interphase of the highly asymmetric sperm aster. The three mechanisms able to move the aster are cytoplasmic pulling (Li and Jiang, 2018; Minc et al., 2011; Wühr et al., 2009), cortical pulling (Grill et al., 2001; Kotak and Gönczy, 2013; Redemann et al., 2010) and cortical pushing (Garzon-Coral et al., 2016; Laan et al., 2012; Meaders and Burgess, 2020). Given that the aster is highly asymmetric with long MTs directed towards the interior of the zygote during interphase (Figs 1A, 2B, 3A), one might expect that aster migration would be powered by cytoplasmic pulling to displace the sperm aster away from the cortex during interphase (Minc et al., 2011). Thus, we first tested whether the absence of migration during interphase was caused by a lack of cytoplasmic pulling. We therefore examined whether these long interphasic MTs accumulated or transported organelles towards the aster center, which would indicate minus-end-directed transport and cytoplasmic pulling (Kimura and Kimura, 2011).

To determine whether there were any measurable differences in the quantity of cytoplasmic pulling forces during interphase versus mitosis, we first sought to identify an appropriate organelle to assess the presence of cytoplasmic pulling. We monitored the dynamic distribution of a number of organelles during sperm aster migration: the female PN, the endoplasmic reticulum (ER), endocytosed vesicles, yolk granules and mitochondria (Fig. 3A). Yolk granules and mitochondria were both excluded from the sperm aster center (Fig. 3A). ER and endocytosed vesicles both accumulated at the aster center (Fig. 3A). Even though the abundant ER is likely the organelle that provides the majority of cytoplasmic pulling forces it was not suited to assess cytoplasmic pulling activity because it is not a small discrete object and therefore its movement was difficult to track (Fig. 3A). We were particularly interested in exploiting the accumulation of endocytosed cell mask vesicles at the aster center as a proxy for overall cytoplasmic pulling activity (Fig. 3A). This is because these vesicles are discrete objects that are present during both interphase and mitosis, and their dynamics could be quantified during interphase and mitosis. Finally, even though the female PN migrates to the aster center during interphase (Fig. 3A), the female PN undergoes NEB at mitotic entry and was therefore unsuitable to probe cytoplasmic pulling during mitosis.

We thus measured vesicle movement towards and away from the sperm aster in zygotes during interphase and mitosis (Movie 3) as a proxy for cytoplasmic pulling activity. An automated non-biased analysis of the whole zygote showed that there was no significant difference in the number or speed of vesicles moving towards the center of the sperm aster during interphase and mitosis (Fig. 3B). The speed measured is the mean speed of the vesicles on their total trajectories, and the total transport represents the sum of the distance traveled by all vesicles towards the aster center during a 3 min period.

To confirm these automated tracking data, we also manually tracked vesicle trajectories in a region of interest (ROI) around the aster at several cell cycle stages and at high-speed (1 image/s) in one confocal plane and made time projections to visualize vesicle trajectories over the course of 1 min (Fig. 3Ci). We counted the number of vesicles manually in this ROI during interphase, NEB, prometaphase and anaphase (Fig. 3Cii). As found above (Fig. 3B), the overwhelming majority of vesicles that had a persistent track moved towards the aster center (Fig. 3Cii), and also, the number of vesicles moving towards the aster center (minus direction) did not change significantly when comparing interphase to mitosis (Fig. 3Cii). Although we could not detect a significant change in centrosome-directed endosome transport in our experiments, we cannot exclude the possibility that a significant difference would be detected with different cargo or different methods.

To conclude, using endocytosed vesicles as a proxy for organelle accumulation at the aster center, we have demonstrated that there is retrograde transport and, hence, active cytoplasmic pulling forces during interphase, and that the vesicle transport is not significantly different in interphase and mitosis. Provided that this unchanged motor activity can be extended to the other organelles, this indicates that another mechanism must account for the differences in aster migration behavior between interphase and mitosis, and especially, for the lack of aster migration in interphase.

The asymmetric configuration of the sperm aster in interphase (Fig. 2A) together with the accumulation of organelles at the aster center (Fig. 3) should favor an interphasic aster migration by cytoplasmic pulling in the direction of the longer MTs (Kimura and Kimura, 2011; Tanimoto et al., 2016). Given there is no aster migration despite the presence of cytoplasmic pulling, we hypothesized that another mechanism must prevent aster migration. A candidate mechanism to counteract the centering forces of cytoplasmic pulling is cortical pulling, which is able to cause an aster to become off-center (Laan et al., 2012). Cortical pulling occurs when minus-end-directed motors bound to the cortex interact with MTs (Laan et al., 2012). The transport of the plasma membrane to the centrosome is prevented by the thick actin cortex, which provides a support that resists membrane deformation. Hence, with an intact cortex, instead of bringing the plasma membrane towards the MT minus-end, the molecular motors pull the centrosome towards the plasma membrane (Fig. 4A). We analyzed whether such cortical pulling forces would thus restrain the aster from migrating to the cell center.

We examined cortical pulling over the zygote cell cycle using a membrane invagination assay (see schematic in Fig. 4A and Godard et al., 2021; Redemann et al., 2010; Rodriguez-Garcia et al., 2018). We used two pharmaceutical compounds to disrupt the actin cortex: cytochalasin B or latrunculin B. Both perturbed the cortical resistance necessary to prevent membrane deformation. Also of note is that the migration speed of cytoplasmic organelles including even large organelles, such as the female PN, is not changed significantly following actin cytoskeleton disruption [control speed=0.24±0.06 µm/min (n=8, mean±s.d.) versus speed following latrunculin=0.21±0.04 µm/min (n=6, mean±s.d.)]. By labeling the plasma membrane with a fluorescent dye (Cell Mask Orange), invaginations of the plasma membrane towards the centrosome could be observed at sites of cortical pulling (Fig. 4B). Depolymerization of MTs with nocodazole completely inhibited such invaginations (Fig. 4B, lower panel, and Godard et al., 2021). Latrunculin also completely disrupted the actin cortex (Fig. S2). In zygotes treated with latrunculin, invaginations were never observed at meiosis II (6 min before PN formation) whereas an average of 20±4 (mean±s.e.m.) invaginations was counted during interphase that decreased to 3±2 (mean±s.e.m.) invaginations at mitosis (1 min after NEB) (Fig. 4C; Movie 4). We determined the mean number of invaginations at different phases of the cell cycle, and also quantified the number of invaginations in two ROIs at the front (green ROI) and rear (yellow ROI) of the aster (Fig. 4D). More invaginations occurred behind the sperm aster (vegetal, yellow) during interphase as compared to the front end of the sperm aster (green and Fig. 4D). In contrast, at mitotic entry the density of cortical pulling sites is higher at the front end than at the rear end of the aster suggesting that cortical pulling stops at the rear or close end of the aster before it stops at the front end of the aster (at the other side of the zygote). These observations demonstrate that cortical pulling exerted by cortical motors on MTs is correlated with the cell cycle, with an increased activity during interphase compared to that seen in meiosis and mitosis. Also consistent with strong cortical pulling during interphase, the protein NuMA (which can bind dynein) is localized at the cortex during interphase, but in the cytoplasm during mitosis (Fig. S3).

To test the causality between the cell cycle phases and the presence of active cortical pullers, we performed the membrane invagination assay on two-cell stage embryos in which one cell was either blocked in mitosis or stalled in interphase while the sister cell kept cycling and served as a non-injected control cell (Fig. 4E). To manipulate the cell cycle at the two-cell stage we injected either p21 protein, to delay entry into mitosis, or a truncated non-destructible form of cyclin B protein (Δ90-cycB), to prevent exit from metaphase (Fig. 4E; Levasseur and McDougall, 2000). As for the zygotes, when the actin cytoskeleton has been disrupted with cytochalasin, the non-injected cell displayed invaginations in interphase whereas none were visible in mitosis (Fig. 4E). We found that in p21-injected cells membrane invaginations were detected all throughout the prolonged interphase (Fig. 4E, top row; Movie 5). By contrast, in Δ90-cycB-injected cells, no invaginations were visible during the prolonged metaphase (Fig. 4E, bottom row; Movie 6). We conclude that cortical pulling is cell cycle dependent and is upregulated during interphase and downregulated at mitosis (particularly starting at prophase and through prometaphase and metaphase when CDK1 activity is elevated).

Cortical pulling is active in interphase and therefore could be responsible for preventing aster migration before mitotic entry. If this is the case, we can expect an early centration before entry into mitosis when the cortex is disrupted. The disruption of the actin cortex was performed by adding latrunculin (Fig. S2). We found that when latrunculin was added during meiosis, the sperm aster was on average 60 µm away from the cell center (and thus closer to the cortex; Fig. S4) and remained so for more than 10 min, but that centration started earlier (during interphase) and persisted during early mitosis (Fig. 5A, orange). In contrast, the aster in control DMSO-treated zygotes (blue in Fig. 5A) was on average 40 µm away from the center and remained in position until mitotic entry, when the aster moved more abruptly toward the cell center. In both latrunculin and DMSO conditions, the aster migrated ∼15 µm (total distance) between PN formation and NEB (Fig. 5B), suggesting that although an intact cortex was not required for centration, the overall timing of centration and in particular the switchlike behavior of centration were perturbed by disrupting the cortex. These finding were also consistent with the results from a simulation lacking an intact cortex (Fig. S5). In addition, we suggest that the closer proximity of the sperm aster to the plasma membrane might be caused by the lack of efficient cortical pushing, which occurs during meiosis (Fig. S4).

Based on all the results, we propose the following mechanism for aster migration. The sperm aster grows in meiosis and moves off the cortex by pushing, but because it is relatively small and isotropic, in meiosis, cytoplasmic pulling should not result in centration. In interphase, cortical pulling brings the aster back to the cortex, despite the constant presence of cytoplasmic pulling. Finally, entry into mitosis triggers the migration of the spindle away from the cortex: cortical pulling stops at mitotic entry, MTs become more dynamic and the mitotic apparatus moves by cytoplasmic pulling. We designed a 2D agent-based stochastic computer model of aster migration based on the software Cytosim (Foethke et al., 2009) (Fig. 6A; Movie 7). The parameters of model simulations were set according to past studies on different species (Table S1) and adjusted to our data to match the timing of cell cycle and the cortical pulling activity. We verified the equivalence of the simulated and observed cytoplasmic pulling by comparing a simulation to latrunculin data (Fig. S5A,B, first row). Once an average reference simulation was established, we could remove specific forces to test the contribution of cortical pulling in the global scenario of aster migration (Fig. 6) or to explore other situations (Fig. S5). With a permanent cortical pulling, the aster was quickly brought to the cell cortex in meiosis and did not seem to leave the cortex during the cell cycle (Fig. 6A,B, second row). Thus, in this condition the migration did not fit the control profile of aster migration as the aster was always further from the cell center than in the control (Fig. 6B, second row).

In the complete absence of cortical pulling throughout the cell cycle (Fig. 6, third row) the aster position was slightly more centered in interphase, and the overall aster centration hence reached the same distance from the cell center as in the average reference simulation (Fig. 6, first and third rows). Indeed, in the absence of cortical pulling the aster centered as much as it did in the control dataset, however the centration began earlier, progressed at a constant velocity and did not show the static phase prior to aster migration (Fig. 6B, third row). The comparisons of simulations outputs to the data supported the scenario drawn from our results where aster centration is prevented by cortical pulling activity during interphase, and the inactivation of cortical pulling in mitosis permits aster movement. Only relying on simulations, we pushed the exploration of our model on questions still experimentally unanswered. These theoretical explorations suggested that cytoplasmic pulling is the main contributor to aster centration in mitosis as opposed to the pushing mechanism (Figs S5A and S4B, bottom row).

Finally, we used the simulation to examine a scenario where cortical pulling is non-uniformly inactivated at mitotic entry, that is if cortical pulling is first turned off near the aster before being completely inactivated (Fig. 7A). In this case, the centration curve of the simulated aster fitted the experimental curve better than when the cortical pulling was implemented as uniformly turning off (Fig. 7A and 6A,B). This simulation thus supports the data showing that there is a difference in the density of invaginations, and hence in cortical pulling, at the front and the rear of the aster at mitotic entry (Fig. 4D).

In addition to such an asymmetric loss of cortical pulling at mitotic entry, a careful analysis of timelapse movies (Movie 8) indicated that the zygote displayed a subtle and transient deformation at mitosis entry (Fig. 7B). This was characterized by a transient loss of cortical actin initiated in the vegetal hemisphere (Fig. 7B; Movie 8). Given the proposal that cortical tension could influence cortical pulling (Kozlowski et al., 2007) and also because cortical actin transiently falls at mitosis entry, we directly measured cortical tension with the micropipette aspiration technique, focusing on mitotic entry (Fig. 7C). We found that the cortical tension fell transiently at mitosis entry from ∼300 pN/µm at interphase to ∼100 pN/µm at NEB, then began to rise again during mitosis (Fig. 7C). The implication of this finding is that the loss of cortical tension we find also correlates with the loss of cortical pulling (see Discussion) rather than increased cortical pulling as modeled previously (Kozlowski et al., 2007).

A central question in the field of cell biology is how a cell divides into two equal sized daughter cells, which relies on positioning of the mitotic apparatus to the cell center. The mechanism(s) involved in centering are still not fully resolved although they often depend on MT-based cortical pulling, cytoplasmic pulling or cortical pushing depending on the species, cell type and cell cycle stage. Centration has been studied extensively in Xenopus, sea urchin and C. elegans zygotes. In the large oocytes of Xenopus (∼1 mm) cytoplasmic pulling forces provide the force for aster centration. For example, by microinjecting a dominant-negative fragment of the dynactin complex (p150–CC1) sperm aster centration was blocked (Wühr et al., 2010). Also, given that MTs of the sperm aster are too short to reach the cortex on the opposite side of the zygote, the primary mechanism for centration in Xenopus is cytoplasmic pulling in the direction of the longest MTs (Wühr et al., 2010). Sea urchins have smaller zygotes (∼100 µm) and even though MTs of the sperm aster are long enough to reach the opposite cortex, cortical pulling has not been reported to be involved and instead the two mechanisms reported to power sperm aster centration in sea urchin are based on cytoplasmic pulling (Tanimoto et al., 2016) and cortical pushing (Meaders et al., 2020). In C. elegans, centration occurs during prophase in a dynein-dependent manner (Gönczy et al., 1999). In order to distinguish between cortical and cytoplasmic dynein, factors that recruit dynein to the cortex have been depleted. RNAi knockdown of goa-1 and/or gpa-16 to deplete cortical dynein leads to a higher velocity of sperm aster centration, suggesting that cytoplasmic dynein is the primary force-generating mechanism for centration (De Simone et al., 2018). In addition, given that depletion of cortical dynein increased the velocity of centration, cortical pulling forces were suggested to counteract the cytoplasmic forces that power centration of the sperm aster (De Simone et al., 2018).

The examples of sperm aster centration in Xenopus and sea urchin occur during interphase, whereas in C. elegans, primates (Asch et al., 1995; Hewitson and Schatten, 2002; Simerly et al., 2019) and ascidian (here), centration occurs at mitotic entry. Moreover, the position and geometry of the ascidian sperm aster, with long MTs extending into the zygote interior and short MTs directed towards the proximal vegetal cortex, represents a configuration that should favor cytoplasmic pulling to displace the sperm aster towards the zygote center. We were therefore curious about what prevented sperm aster centration until mitotic entry. Cortical pulling responsiveness to cell cycle has been reported previously in C. elegans one-cell stage embryos (Bouvrais et al., 2021; McCarthy Campbell et al., 2009; Redemann et al., 2010). An increase in posterior cortical pulling forces displaces the mitotic apparatus towards the posterior cortex during anaphase (Grill et al., 2003; McCarthy Campbell et al., 2009; Redemann et al., 2010). Indeed, the number of short-lived MT plus-ends engaged in cortical pulling increased at the posterior pole of C. elegans one-cell stage embryos at anaphase onset (Bouvrais et al., 2021). This increase in cortical pulling has been linked with the fall in CDK1 activity – reducing the function of the proteasome, the anaphase-promoting complex (APC) or Cdc20 all delayed spindle displacement towards the cortex, whereas inactivating CDK1 in prometaphase caused premature spindle displacement (McCarthy Campbell et al., 2009). Although these findings indicate that cortical pulling increases at anaphase (Keshri et al., 2020; Kotak et al., 2013), one key additional point might be that cortical pulling is less prominent when CDK1 activity is elevated. Here, in the ascidian, we have noted a similar phenomenon, except that cortical pulling is elevated during interphase and reduced at mitotic entry when CDK1 activity increases.

In the ascidian P. mammillata, the sperm aster forms in the vegetal hemisphere of the zygote during meiosis II (Roegiers et al., 1995). Here, we demonstrate that during meiosis II, the sperm aster grows and remains roughly spherical as it moves slowly away from the proximal vegetal cortex (Fig. 1). At this stage the sperm aster remains relatively small (approximately one-third of the zygote diameter) and is located in the zygote vegetal hemisphere. Such a vegetal location and relatively small size ensures that sperm astral MTs do not reach the animal pole and thus interfere with the segregation of the meiotic chromosomes, which in smaller C. elegans zygotes is accomplished by preventing sperm aster growth during meiosis II (McNally et al., 2012). At entry into interphase, astral MTs extend throughout the zygote and capture the female PN located at the animal pole (Figs 1 and 2A). The female PN then migrates towards the center of the sperm aster during a short interphase (∼10 min) while the cortically located and highly asymmetric sperm aster remains in position near the vegetal pole (Figs 1 and 2A). Just prior to NEB, during prophase, the sperm aster begins abruptly to migrate accompanied by NEB and formation of a bipolar mitotic apparatus (Fig. 1).

What triggers the switch to induce migration at entry into mitosis? First, we delayed entry into mitosis to determine whether a causal relationship existed between entry into mitosis and sperm aster migration. Delaying entry into mitosis with the CDK inhibitor p21 prevented sperm aster migration (Fig. 2). Next, we teased apart the relative contributions of cortical pushing, cytoplasmic pulling and cortical pulling to determine which of these three mechanisms displayed a cell cycle-dependent change at mitotic entry that could explain how mitotic entry triggered sperm aster migration. Even though cortical pushing is active in meiosis (Fig. S4), its contribution to spindle migration at mitosis entry seems minor (Fig. S5). We therefore focused on cytoplasmic pulling and cortical pulling. Cytoplasmic pulling can be visualized through the movement of three different endomembrane structures towards the center of the sperm aster – the female PN, endoplasmic reticulum (ER) and vesicles (Fig. 3). Other major organelles, such as the yolk and mitochondria are not accumulated at the aster center (Fig. 3). The major organelle likely involved in cytoplasmic pulling is the ER, as suggested in Xenopus and sea urchin (Cheng and Ferrell, 2019; Mukherjee et al., 2020), but given that the ER is not a discrete organelle, it is not suited for quantitative measurements. We therefore chose to exploit the Cell Mask-labeled vesicles, which are small (∼1 µm) objects that can be analyzed quantitatively as a proxy for overall cytoplasmic pulling. We quantified the movement of the cytoplasmic vesicles to determine whether there was a measurable difference in cytoplasmic pulling between interphase and mitotic entry (Fig. 3). The data demonstrated that vesicle transport was unchanged between interphase and mitotic entry, suggesting a constant cytoplasmic pulling. We then sought to determine whether cortical pulling was more prominent during interphase or mitosis. To do so we exploited the membrane invagination assay following weakening of the cortex (Godard et al., 2021; Redemann et al., 2010; Rodriguez-Garcia et al., 2018). Interestingly, we noted that cortical pulling was greater during interphase than mitotic metaphase (Fig. 4). In particular, we noted more cortical pulling in the vegetal hemisphere of the zygote close to the cortical sperm aster (Fig. 4D). This is likely because of the radial organization of the aster so that the abundant MTs nearest the cortex elicited more cortical pulling sites (Fig. 4D). To directly test the correlation between mitosis and the absence of cortical pulling, we devised a two-cell stage assay to determine whether cortical pulling was a feature of interphase and switched off at mitotic entry. By either delaying mitotic entry with p21 or blocking exit from metaphase with Δ90-cycB in one sister cell (Fig. 4), we demonstrated that cortical pulling is elevated during interphase and switched off at mitotic entry when CDK1 activity is elevated. This observation could explain the switch-like behavior in migration, and it suggests that the increase in CDK1 activity switches off cortical pulling thus liberating the sperm aster from its cortical tethers and facilitating centration. Moreover, these data develop further the findings from C. elegans where cortical pulling was shown to be decreased when CDK activity was elevated (McCarthy Campbell et al., 2009).

By using the software Cytosim, we tested whether cortical pulling could prevent aster migration mediated by cytoplasmic pulling. Simulations showed that indeed, cortical pulling could prevent migration caused by long-MT-mediated cytoplasmic pulling. This supports the idea, drawn by the data, that switching off cortical pulling at mitotic entry is necessary for sperm aster migration. Furthermore, the simulation indicated that a total absence of cytoplasmic pulling prevented aster migration in mitosis (Fig. S5). Finally, given that cortical tension can influence cortical pulling (Kozlowski et al., 2007), we measured cortical tension at entry into mitosis. We found that cortical tension transiently falls at entry into mitosis, and this fall is accompanied by less cortical pulling forces near the aster (Fig. 4). Interestingly, this result stands out from what was previously modeled in C. elegans where a fall of cortical tension increased the cortical pulling (Kozlowski et al., 2007). We thus speculate that local changes in the cortex nearest the aster (coupled with changes in MT dynamics) lead to localized decoupling of the sperm aster first nearest the proximal vegetal cortex. We propose that centration then occurs via cytoplasmic pulling plus a contribution of the remaining cortical pulling from the animal hemisphere before all MTs become shortened at mitotic entry. This wavelike behavior of the cortex could be a consequence of the wavelike entry into mitosis that has been observed in oocytes and zygotes of Xenopus and starfish (Bischof et al., 2017; Pérez-Mongiovi et al., 1998) where CDK1 activity increases as a wave beginning at the sperm aster (Pérez-Mongiovi et al., 1998).

Overall, these findings demonstrate that aster migration in the ascidian occurs at mitotic entry, which causes the loss of cortical pulling first locally near the aster while cytoplasmic pulling remains active. Given that the sperm aster in primates also remains cortical during interphase and migrates towards the zygote center at mitosis entry (Asch et al., 1995; Hewitson and Schatten, 2002; Simerly et al., 2019) as in the ascidian zygote, it would be interesting to examine the relationship between mitotic entry and sperm aster and mitotic apparatus migration in primate zygotes to determine whether cortical pulling is also switched off at mitotic entry.

Phallusia mammillata adult animals were collected in Roscoff or Sète and kept at 16°C in the aquaria of the Centre de Ressources Biologiques (CRB) of the Institut de la Mer à Villefranche (IMEV) which is an EMBRC-France certified service (see https://www.embrc-france.fr/fr/nos-services/fourniture-de-ressources-biologiques/organismes-modeles/ascidie-phallusia-mammillata). The gametes were collected by puncturing separately the oviduct and the sperm duct. The sperm was kept dry at 4°C and could be used for fertilization up to 1 week after collection. Oocytes were used the day of collection after undergoing dechorionation by incubation in 0.1–0.2% trypsin in micro-filtered natural sea water (MSFW; Nalgene 0.2 µm filters, cat. no. Z370584) at 19°C for 90 min, and subsequent washes in MSFW supplemented with 5 mM tris(hydroxymethyl)methylamino propanesulfonic acid (TAPS; pH 8.2). All the subsequent manipulations of live embryos were performed in MSFW with 5 mM TAPS, using pipette tips, dishes, slides and coverslips coated with 0.1% gelatin and 0.1% formaldehyde (Sardet et al., 2011). For fertilization, a small volume of activated sperm (∼5 µl) was added to the oocytes in a 5 ml Petri dish. To activate the sperm, 6 µl of dry sperm was incubated for 20 min at 19°C in 500 µl of MSFW pH 9.2. Time post fertilization was measured starting when the oocyte first showed a shape change. For fixation of fertilized cultures, the time of fertilization was determined when ∼30% of the oocytes showed simultaneously the first deformation.

Synthetic mRNAs for microinjection were prepared using the mMESSAGE mMACHINE T3 kit (Ambion), from plasmids containing the gene of interest (EB3::3GFP, Ensconsin::GFP, PH::GFP, Histone H2b::Rfp1, iMyo::Scarlet, NuMA::Venus; McDougall et al., 2015) flanked by a T3 promoter and a polyA tail. mRNA yield was estimated by spectrophotometry. The mRNAs were stored at high concentration (>10 µg/µl) in 1 μl aliquots at −80°C, then thawed and diluted in distilled water for use. The mRNAs were microinjected into dechorionated oocytes that had been transferred to small glass wedges mounted onto 400 µl Perspex mounting chambers designed for horizontal microinjection (see detailed protocols in McDougall et al., 2014). mRNAs were injected at a pipette concentration of 5–6 μg/μl (injection volume is ∼1–2% volume of the egg) using a high-pressure system (Narishige IM300). mRNA-injected oocytes were left for 5 h or overnight before fertilization and subsequent confocal imaging.

LifeAct::Venus protein was injected to label actin, Atto565-tubulin to label MTs and p21::mCherry protein (p21 is the cyclin dependent kinase inhibitor 1a) was injected to arrest the cell cycle in interphase (Dumollard et al., 2011). The human p21 protein fused with mCherry was cloned in a pET11a vector [see Dumollard et al. (2011) for details] with 6 His-tag, and purified with a silica-based resin column (MACHEREY-NAGEL, protino Ni-IDA). Aliquots were flash frozen in liquid nitrogen and stored at −80°C. The protein was injected at a concentration of 30 mg/ml.

The construction and synthesis of the human Δ90 cyclin B1::GFP plasmid has been described previously (Levasseur and McDougall, 2000). The Δ90cyclin B::GFP fusion protein was stored at −80°C and was injected at a final concentration of ∼11 mg/ml.

The microinjection system described above for mRNAs injection was also used for protein injection. Protein-injected oocytes were left for 45 min at 18°C before fertilization and subsequent confocal imaging. The mouse anti-tubulin DM1A antibody at 1:500 (cat. no. MABT205, Sigma-Aldrich) was used to label MTs. Briefly, following three washes in PBS with 0.1% Triton X-100, zygotes were incubated with specific fluorescently labeled secondary antibodies at room temperature for 1–2 h (goat anti-mouse-IgG secondary antibodies diluted 1:200; Jackson Labs, cat. no. 115-095-003). Following two further washes in PBS with 0.1% Triton X-100, embryos were incubated for 10 min in PBS with 0.1% Triton X-100 containing Hoechst 33342 (5 µg/ml; Sigma-Aldrich, cat. no. 15533), washed twice and then mounted in Citifluor AF1 (Science Services, München, Germany) for imaging.

All imaging experiments were performed at 19°C using a Leica TCS SP8 inverted microscope fitted with Hybrid detectors and a 40×/1.1 NA water objective lens. To image aster migration, each fertilized egg or two-cell stage embryo was scanned by 4D live imaging of a whole embryo (xyzt) with a frame rate of at least 1 z-stack every 210 s. The imaging parameters were adapted to each fertilized egg: z step was between 0.5 and 2 µm and time step between 1 min and 3.5 min. For fixed samples, z-stacks were acquired at step size of 0.5 µm.

Eggs were fertilized in MSFW and transferred after observing the first deformation into a solution of MSFW with 5 mM TAPS containing 5 µM of latrunculin B (Sigma-Aldrich) diluted from a 10 mM stock solution (in DMSO). The embryos were then mounted on a slide in the latrunculin SW solution. The zygote plasma membrane was visualized either by microinjection of the PH::Tomato mRNA (as detailed in McDougall et al., 2015), or by addition of the membrane dye Cell Mask Orange (Thermo Fisher Scientific, Invitrogen, cat. no. C10045) at 1:1000. For controls, zygotes labeled with PH::Tomato or Cell Mask Orange were treated with DMSO at a dilution of 1:1000.

In the case of the embryos injected at the two-cell stage, the fertilized eggs were left to develop at 16°C in MFSW until they started cleaving. Then they were mounted in the injection chamber and injected with p21 or Δ90-cyclinB proteins as soon as the division finished. When two or three embryos were injected, they were immediately transferred in MFSW with 5 µM of cytochalasin B (Sigma-Aldrich) diluted from a 10 mM stock solution in DMSO, and mounted on a slide in this solution for imaging.

To image membrane invaginations in zygotes or two-cell stage embryos, 25-µm-thick stacks of confocal images (dz=2.5 µm) were acquired every 10 s in Cell Mask Orange-stained zygotes at ∼2–4 min after latrunculin incubation. The number of membrane invaginations was counted manually by counting invaginations present at 2–5 µm from the plasma membrane.

Membrane invaginations were imaged in two-cell embryos at a frequency of 1 z-stack every 30 s to 1 stack every 2 min. The presence or absence of invagination was scored manually on a z-projection image and cell cycle state of the cells were defined as the 15 min following NEB, and according to nuclei presence.

The quantification of the distance between the center of the zygote and the DNA throughout the cell cycle was performed in three steps, using Fiji (Schindelin et al., 2012).

First, the center of mass of zygotes were determined at each time point. To do so, a Gaussian blur (sigma=2) was applied to each xyz stack of the timelapse movie. Stacks were then made binary with the ‘Triangle’ method of the ‘Auto Threshold’ function. The Fiji plugin ‘3D Roi Manager’ (Ollion et al., 2015) created objects from the binary stacks, and output their center of mass. This method was verified by comparing the center of mass of the 3D zygote to the centroid of the 2D equatorial section with widest diameter.

Secondly, using the Fiji ‘Point tool’ and the ‘measure’ function, the xyz coordinates of the DNA label were obtained at each time point. When DNA was not labeled (in p21-injected embryos), or weak (at NEB), DNA position was approximated to be at an equal distance between the two spindle poles, or at the center of the aster when the spindle was not yet formed. The DNA label was chosen over the MT label to measure the aster migration because centrosome duplication occurs before the centration of the spindle, therefore while the spindle as a whole centers, each spindle pole starts centering and then diverges from the cell center to center the DNA.

Finally, the distance between the DNA coordinates and the center of mass was computed using the 3D Pythagorean theorem: d=√((x2−x1)²+(y2−y1)²+(z2−z1)²).

To image vesicle trafficking, eggs were fertilized, washed once, and immediately transferred to a GF-coated slide/coverslip in a MFSW containing Cell Mask Orange (1:1000). Then, 2D images acquired every second for 3 min. The imaged plane was selected to contain the center of one aster.

To measure relative movements of the vesicles, we combined three approaches. (1) For vesicles, we used the Fiji tool TrackMate with LoG particle detection and simple LAP tracker (Tinevez et al., 2017). (2) For aster localization we wrote (using MATLAB) a manual periodic tracking, with interpolation for intermediate time frames (available upon request). (3) For cell contour, we developed another MATLAB algorithm based on threshold optimization to extract the cell contour (available upon request). We combined information from those tools to quantify movement.

Cell Mask-labeled vesicles were also tracked manually by a senior researcher who was aware of experimental conditions in a ROI containing the sperm aster.

Glass pipettes with an internal diameter of 20 µm (Biomedical Instruments) were filled with SW and inserted into a pipette holder and connected to a Microfluidic Flow Control System Pump (Fluiwell, Fluigent) as previously (Godard et al., 2020). The pipette was then positioned on the zygote surface and a negative pressure (ΔP) applied causing a deformation inside the pipette until the critical pressure (P_c) had been reached (hemisphere of aspirated cortex). Then, ΔP was set at 3 P_c. ΔP was controlled and adjusted (if required) in order to prevent over-aspiration above the critical pressure. Images were captured every 30 s during the whole process to make a movie. For each frame of this movie, the surface tension (Tsurface) was calculated using the Young-Laplace equation Tsurface=ΔP /2(1/R_p-1/R_c), where R_p is the radius of the cell deformation in the micropipette, and R_c the radius of the zygote.

Agent-based simulations were performed in 2D using a custom version of the software Cytosim (https://cytosim.org/; Foethke et al., 2009) with the parameters provided in Table S1. Cytosim is a stochastic simulation engine that includes constituting elements of the cytoskeleton. It has been previously used to study spindle and centrosome dynamics and position (Khetan and Athale, 2016; Lacroix et al., 2018; Letort et al., 2016).

MTs are modeled as worm-like chains, characterized by a bending stiffness (see Table S1) and inextensibility. MTs can (de)polymerize and their instability is modeled by a stochastic alternation of growing and shrinking phases. The plus-end polymerizes until a catastrophe happens, starting a shrinking period. The centrosome is modeled as a bead, from which MTs are nucleated. MT minus-ends are anchored to the centrosome while their plus-end are directed outward from the aster.

A pushing force is generated by polymerization at the plus-end of the MTs pushing against the edge of the cell. When MTs push strongly on the cortex, polymerization is slowed down and they have a higher chance to undergo a catastrophe. Previous work (Letort et al., 2016) suggests that pushing cannot center the aster if MTs can glide along the cell membrane. In that work, gliding was prevented by pinning the tips of MTs to the point where they first reached the edge of the cell. Pinning and cortical pulling were not activated at the same time though, as pinning would make cortical pulling inefficient. To prevent MT gliding along the cell cortex, we did not model control cells as a proper circle, but as a crenelated polygon, representing the actomyosin cortex. Once a tip enters an alcove, it cannot slide anymore, as if it were stuck by intertwined actin filaments. This allowed to combine MT pushing and cortical pulling.

If cortical pulling is implemented by usual dynein, it often makes the aster spin around the cell. This is due to the fact that MTs tend to align with the edge of the cell, as more and more cortical dynein becomes attached to the MTs. The pulling force becomes higher and higher and the aster spins around. In vitro experiments suggest that dynein placed in front of a rigid barrier can control the dynamics of MTs (Laan et al., 2012). Such dyneins trigger a catastrophe when they bind to the end of a MT and regulate its shrinking speed, thereby generating pulling forces. We modified Cytosim to implement such a behavior: cortical dynein works like classical molecular motors except that they can only bind to a MT end, and they ‘chew’ the MT as they move forward. Like cytoplasmic dyneins, they have a stall force, and the load comes from a spring linking the dynein to its original position. However, unlike cytoplasmic dyneins, cortical dyneins cannot move backwards as it would imply the MTs polymerized again.

The custom version of Cytosim with this interaction implemented is available at https://gitlab.com/gslndlb/cytosim, in the branch dynein_chew. All configuration files used are in the cym folder.

Statistical tests and graphics were performed using the libraries rstatix, tidyverse, dplyr, ggpubr and ggplot2 from R software (R Studio, 2020) as well as Microsoft Excel (2013). Tests are provided in the figure legends. Diagrams were created with BioRender.com

We thank members of the Turlier and McDougall groups for technical advice and discussion. We are grateful to the Imaging Platform (PIM) and animal facility (CRB) of Institut de la Mer de Villefranche (IMEV), which is supported by EMBRC-France, whose French state funds are managed by the ANR within the Investments of the Future program under reference ANR-10-INBS-0, for continuous support.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.262037.reviewer-comments.pdf