Anchoring cortical granules in the cortex ensures trafficking to the plasma membrane for post-fertilization exocytosis

Following fertilization, cortical granules exocytose ovastacin, a metalloendopeptidase that cleaves ZP2 in the zona pellucida surrounding mouse eggs to prevent additional sperm binding. Using high- and super-resolution imaging with ovastacinmCherry as a fluorescent marker, we characterize cortical granule dynamics at single granule resolution in transgenic mouse eggs. Newly-developed imaging protocols provide an unprecedented view of vesicular dynamics near the plasma membrane in mouse eggs. We discover that cortical granule anchoring in the cortex is dependent on maternal MATER and document that myosin IIA is required for biphasic trafficking to the plasma membrane. We observe local clearance of cortical actin during exocytosis and determine that pharmacologic or genetic disruption of trafficking to the plasma membrane impairs secretion of cortical granules and results in polyspermy. Thus, the regulation of cortical granule dynamics at the cortex-plasma membrane interface is critical for exocytosis and the post-fertilization block to sperm binding that ensures monospermic fertilization.

R egulated exocytosis is a fundamental cellular event that releases cargo molecules at the cell surface for specific physiological tasks including neurotransmission, immune response, and reproduction 1 . In mouse eggs, exocytosis of cortical granules (CGs) makes the extracellular zona pellucida impermeable to additional sperm in the post-fertilization block to polyspermy 2 . CGs are derived from the Golgi apparatus and migrate towards the cortex during oogenesis where they accumulate in a uniform layer that is displaced 0.4 to 0.6 μm inward from the plasma membrane 3 . Anchoring CGs at the periphery is correlated with the acquisition of exocytotic competence, defined as the ability to undergo exocytosis in response to increased intracellular calcium 4 . In mice, the transport of CGs to the cortex is dependent on actin 5 and the egg cytoplasmic actin network mediates longrange transport of vesicles 6 . A recent study carefully documented two pathways that translocate CGs from the center of the egg to the periphery during meiosis. One depends on myosin Va powered movement along actin filaments and the other relies on CG binding to Rab11a vesicles to hitchhike to the egg cortex. Defects in these pathways lead to polyspermy 7 .
However, our current understanding of CG dynamics at the cortex-plasma membrane interface in mouse eggs remains limited. In the absence of protocols to perform live-imaging near the plasma membrane, earlier studies relied on staining fixed CGs with the relatively non-specific LCA (lens culinaris agglutinin) lectin. In additional, mouse CGs are~200 nm which is near the diffraction limit of standard light microcopy 8 . However, total internal fluorescence microscopy (TIRFM) combined with superresolution structured illumination microscopy (SIM) can overcome these limitations and provide sub-diffractive, live-imaging of fluorescently labeled subcellular organelles [9][10][11] .
We have previously reported a subcortical maternal complex (SCMC) that forms during oogenesis and, when disrupted, delays or abrogates cleavage-stage embryogenesis [12][13][14][15][16] . Its biological function(s) during oogenesis have been less well investigated and its peripheral location suggests potential involvement in membrane-associated processes including vesicle trafficking and exocytosis. We also have reported on ovastacin, a zinc metalloendopeptidase that was identified as a pioneer marker of mammalian CGs. It is released during exocytosis and cleaves ZP2 in the extracellular zona pellucida to prevent sperm binding which provides a potent post-fertilization block to polyspermy 17,18 .
Here, we seek to gain insight into CG biology using multiple live-cell imaging modalities with enhanced spatiotemporal resolution and transgenic mice expressing fluorescently tagged ovastacin mCherry as a marker of CGs. We genetically document the involvement of the SCMC component MATER in anchoring CGs at the egg cortex and demonstrate a role for myosin IIA in CG trafficking and clearance of cortical actin prior to exocytosis. Perturbation of CG trafficking at the cortex-plasma membrane interface leads to polyspermy and adversely affects in vivo female fertility.

Results
Mouse CGs accumulate in the egg cortex. To perform livediffraction limited and super-resolution (instant TIRF-SIM) imaging of the cortical cytoskeleton and plasma membrane, the 8 μm thick zona pellucida was removed ( Supplementary Fig. 1a-d).
Live confocal microscopy of stimulated wild-type eggs in the presence of plasma membrane (CellMask) and actin-binding fluorogenic (SiR-actin) dyes documented an uneven plasma membrane topography and enrichment of filamentous actin (F-actin, referred to as actin hereafter) as a cytoskeletal scaffold. Single XZ optical sections spanning the membrane clusters showed individual microvilli-like structures and actin fibers that extended in the axial direction. 3D recordings further documented dynamic disappearance and reappearance of actin in a single lateral plane ( Supplementary Fig. 2a-f).
CGs initially were detected by confocal microscopy in fixed eggs from transgenic mice expressing ovastacin mCherry (Astl mCherry ) using fluorescently tagged lens culinaris agglutinin (LCA) or ovastacin mCherry (Fig. 1a). To determine whether CGs were enriched in the cortex prior to exocytosis, we imaged zonafree Astl mCherry eggs at single granule resolution. Live confocal microscopy showed accumulation of CGs in the egg cortex (Fig. 1b). Surprisingly, 3D confocal recordings that sampled the cortex of zona-free eggs every 5 s (Fig. 1c, inset; Supplementary Movies 1, 2) revealed that CGs not only co-localized and moved along actin (Fig. 1c, C2), but actin also polymerized and extended into the imaging plane towards a single CG (Fig. 1c, C1; arrowhead). Following attachment, actin retracted and pulled the CG towards an area, which accumulated a new membrane microdomain (Fig. 1c, C1; dashed area). This contrasted with a stationary CG (Fig. 1c, C1; arrow) which appeared near the membrane at the onset of imaging. The proportion of CGs that associated with actin was higher than those without contact suggesting that CGs interact with the cortical actin scaffold for movement (Fig. 1d). The co-localization and associated movement with cortical actin also were confirmed in fixed samples using super-resolution instant structured illumination microscopy iSIM (Fig. 1e, f).
To follow dynamic movement of granules deeper into the cytoplasm of stimulated mouse eggs, we performed rapid 4D iSIM imaging 19,20 . iSIM projections showed enrichment of CGs beneath the plasma membrane (Fig. 1g, inset; Supplementary Movie 3) and individual CGs could be resolved as they approached the plasma membrane ( Fig. 1g; circle in lower panels). Detailed analysis of vesicle movement documented more CGs in the cortex where they also resided for a longer time than CGs present deeper in the cytoplasm (Fig. 1h, i). These results indicate that mouse CGs interact with a dynamic actin cytoskeleton in the egg cortex prior to exocytosis.
MATER anchors CGs in the egg cortex. We used Astl mCherry female mice and antibodies to MATER (official gene name, Nrlp5) to confirm the peripheral location of the SCMC and CGs in fully grown oocytes and ovulated eggs (Fig. 2a). We then established Mater Null ; Astl mCherry mice in which CGs did not preferentially accumulate in the cortex of eggs but were scattered throughout the cytoplasm (Fig. 2b). This striking redistribution was quantified by live confocal microscopy of stimulated eggs with >2-fold increase in cytoplasmic fluorescence intensity of mCherry-labeled CGs in the Mater Null background (Fig. 2c, d).
Similar observations were obtained with super-resolution iSIM on fixed samples that documented a >4-fold decrease in the number of CGs in the cortex of Mater Null eggs (Fig. 2e, f). These findings were confirmed by electron microscopy in which CGs were nearly absent in the cortex, but present in the cytoplasm of Mater Null eggs, whereas the opposite distribution was observed in wild-type eggs (Fig. 2g, h). Thus, anchoring CGs in the egg cortex requires the presence of maternal MATER.
Biphasic CG trafficking to the plasma membrane. Instant TIRF-SIM images were acquired every 1 s over 50 time points in zonafree mouse eggs to observe the motion of 4686 CGs in zona-free Mater Het and Mater Null eggs at the plasma membrane ( Fig. 3a; insets; Supplementary Movies 4,5). The imaging rate was sufficient to track the movement and identification of three fluorescent subpopulations based on the duration of their tracks (Fig. 3b). We noted a strong bimodal distribution in Mater Het eggs with one peak representing short tracks (<8 s) and another peak for long tracks, in which granules were already present at the starting frame and still observed at the last frame of the experiment (49 s). We classified the duration of the tracks between the two peaks as medium. Although short-, medium-, and longduration granules were present in Mater Null eggs, the absolute number of CGs (total: 1511) at the plasma membrane was lower (Fig. 3b). This was anticipated based on our earlier observation that the number of granules anchored to the cortex was reduced in Mater Null eggs (Fig. 2). Our particle tracking analysis uncovered specific differences in the three subpopulations. As a percentage, Mater Null eggs not only contained more short and long tracks ( Fig. 3c; chi-squared test, p < 2.2e −16 ), but they were also substantially depleted in tracks of medium duration when visualized using a hexbin density plot (Fig. 3d, diagonal and bottom middle of plots). In addition, shortduration granules moved faster than their medium-or longduration counterparts in the absence of MATER ( Supplementary  Fig. 3a, b). We further interrogated the mean fluorescent track intensities across the three subpopulations and did not observe significant differences in the track intensities of short-, medium-, and long-duration granules in Mater Het eggs. Notably, track intensities were substantially higher in all three populations in stimulated Mater Null eggs ( Fig. 3e; Supplementary Fig. 4). Whereas the absence of MATER does not appear to affect trafficking to the plasma membrane from any subpopulation, our data indicates that the interaction of granules with the plasma membrane and the subsequent release of ovastacin is impaired.
Next, we analyzed the track intensity of only long-duration granules. We identified individual appearing granules in Mater Het eggs that moved into focus at the plasma membrane and became brighter (Fig. 3f) suggesting that sustained secretion requires arrival of new vesicles at the plasma membrane, which are rendered fusion-competent 21 . Strikingly, appearing vesicles were nearly absent in Mater Null eggs indicating that recruitment of new granules to the membrane was impaired ( Fig. 3g; Supplementary  Fig. 4). We also identified vanishing granules, where the fluorescent intensity decayed (Fig. 3f, g). Since instant TIRF-SIM does not indicate directionality, we cannot determine whether the disappearance of granule fluorescence was due to exocytosis or retreat into the cell interior. However, having established that the track intensity of long-duration granules was significantly higher in Mater Null eggs due to impaired interactions at the plasma membrane, we favor the model that most of the vanishing granules are undergoing exocytosis and noted fewer vanishing granules in Mater Null eggs ( Fig. 3g; Supplementary  Fig. 4). These differences suggest that exocytosis could be delayed in the absence of MATER. Depth-based pseudo coloring of cortical granules in wild-type and Mater Null eggs (lower panels, XY projections). Color-scale bar, color and frame number ranging from dark blue (top plane, Z = 8 μm) to white (bottom plane, Z = 0 μm). Representative iSIM projections are shown. Scale bar, 5 μm. f Comparison of cortical granules quantified in the cortex of wild-type (n = 10) and Mater Null eggs (n = 6) in e. ***p < 0.001 by two-tailed Student's t-test. g Transmission electron microscopy of wild-type and Mater Null eggs. Arrows, cortical granules; arrowhead, microvilli; ZP, zona pellucida. Scale bar, 2 μm. h Quantitative comparison of cortical granules in the cortex and cytoplasm of electron micrographs from wild-type (n = 11) and Mater Null eggs (n = 12). ***p < 0.001 by two-tailed Student's t-test. Error bar = s.e.m. Source data are provided as a Source Data file We also clustered tracks into a third category, which neither conformed to appearing nor vanishing profiles. This neither category contained a range of intensity profiles ( Fig. 3g; Supplementary Fig. 4), but the fluorescent intensity remained roughly constant or varied enough that no clear pattern was observed (Fig. 3f). Interestingly, this category of granules was predominantly observed in Mater Null eggs suggesting that it could further contribute to a delay in exocytosis. Thus, particle tracking at high spatiotemporal resolution with instant TIRF-SIM in combination with computational analysis showed that anchoring CGs critically affects their trafficking from the cortex to the plasma membrane in mouse eggs. Non-muscle myosin IIA associates with CGs. We determined that fluorescent myoIIA in transgenic mice 22 was present in the cortex of fixed, phalloidin-stained fully-grown oocytes and stimulated eggs ( Supplementary Fig. 5a). To investigate the possibility that myoIIA was recruited onto CGs, we generated mice carrying both the fluorescent reporter knock-in MyoIIA EGFP and Astl mCherry transgenes. Acquiring TIRF images every 5 s over 5 min was sufficient to follow CGs for an extended period Intriguingly, myoIIA EGFP associated with two populations of granules: appearing and stationary. As previously documented with instant TIRF-SIM, we noted appearing granules moving into the TIRF field over time and becoming brighter. In contrast, the brightness did not change in stationary granules, which were present from the first to the last frame of the experiment (Fig. 4a).
Tracking the motion of 5496 CGs and plotting their bimodal distribution over the 5-min imaging period showed that 1033 (~20%) were stationary (Fig. 4b, c). Confocal sections of fixed eggs not only showed accumulation of myoIIA EGFP and ovastacin mCherry in the cortex, but also diffraction-limited puncta of EGFP-labeled myoIIA, some of which appeared near ovastacin mCherry -labeled CGs ( Fig. 4d; inset; arrows). To spatially resolve the two signals, we turned to superresolution iSIM. Notably, myoIIA EGFP diffusively localized around EGFP-void areas 15 min post stimulation ( Fig. 4e; dashed area). However, we also observed a preferential enrichment of myoIIA EGFP as a set of puncta ( Fig. 4e; arrows). The spatial resolution of iSIM indicated that only single punctum marking EGPF-labeled myoIIA co-localized with mCherry-labeled CGs (Fig. 4e, f). The data must be interpreted cautiously, however, because at iSIM resolution, a single punctum could contain more than one molecule. Interestingly, we also detected the presence of CGs inside EGFP-void areas ( Fig. 4e; XZ plane; arrowhead). As the motor domain in the head portion of the myoIIA molecule contains a well-characterized actin-binding site 23,24 , we reasoned that myoIIA EGFP puncta were recruited to newly assembled actin. Immunostaining of myoIIA EGFP expressing eggs with phalloidin confirmed that single myoIIA EGFP puncta overlapped with actinrich foci ( Supplementary Fig. 5b, c). Altogether, these results suggest that non-muscle myoIIA associates with mouse CGs trafficking to the plasma membrane for exocytosis.
Clearance of cortical actin prior to exocytosis. Cortical actin presents a barrier that cells must first clear before subcellular vesicles can fuse with the plasma membrane 25 . To determine whether myoIIA EGFP -void areas correlated with the clearance of cortical actin, we analyzed fixed, ovastacin mCherry eggs after stimulation for 5 or 15 min using iSIM. Whereas a mostly dense cortical actin scaffold was present after 5 min of stimulation ( Fig. 5a), we observed prominent actin-free areas after 15 min ( Fig. 5a; inset, dashed area). Consistent with the presence of CGs in myoIIA EGFP -void areas (Fig. 4e), we also detected CGs inside actin-free areas ( Fig. 5a; inset, arrows) suggesting that local clearance of the cortical actomyosin network could provide access of CGs to the plasma membrane for exocytosis ( Fig. 5a; schematic). In contrast, a diffuse and dense cortical actin meshwork persisted in Mater Null eggs even after 15 min of stimulation indicating that the absence of MATER not only impairs CG anchoring at the egg cortex, but also cortical actin turnover associated with exocytosis in mouse eggs (Fig. 5b, c).
We hypothesized that myoIIA mediated contraction leads to actin clearance. Intrigued, however, by our earlier observation that actin clearance was impaired in the absence of MATER, we first performed immunoblot analysis in unstimulated wild-type and Mater Null eggs using an isoform-specific myoIIA antibody. Strikingly, immunoblots detected intact myoIIA protein in wildtype, but only small amounts in Mater Null eggs (Fig. 5d). Myosin V, implicated in CG trafficking 7 , was also decreased in Mater Null eggs indicating that more than one class of myosins is affected in the absence of MATER. To rule out the possibility that low protein expression levels in mutant eggs masked the detection of sub-diffractive myoIIA, we also imaged the cortex using iSIM and observed myoIIA puncta in stimulated Mater Null eggs ( Fig. 5b) indicating the presence of myoIIA, albeit at low levels.
Blebbistatin inhibits myoIIA motor ATPase activity, the small molecule inhibitor ML-7 targets the myosin light chain kinase (MLCK), which is responsible for regulatory light chain (RLC) phosphorylation and myoIIA filament assembly 26,27 . In contrast, actin dynamics are largely driven by Arp2/3-and formindependent actin nucleation, which are inhibited by CK666 and SMIFH2, respectively, and myoIIA-dependent contractile forces 28,29 . Stimulation of zona-free eggs in the presence of myoIIA and actin inhibitors for 15 min led to repression of actin clearance (Fig. 5e, f). Together, these results argue that cortical actin is locally cleared in mouse eggs prior to exocytosis. This process is not only dependent on maternal MATER, but also on myoIIA activity and initial actin nucleation.

Impaired trafficking delays exocytosis and causes polyspermy.
To determine the biological significance of impaired CG trafficking and actin clearance for the zona pellucida block to polyspermy, we performed in vitro fertilization of Mater Null and wild-type eggs. We noticed significantly more sperm in the perivitelline space of two-cell embryos (Fig. 6a, b; 5.4 ± 0.69 s.e.m.; Fig. 3 Biphasic cortical granule trafficking to the plasma membrane. a Live, instant TIRF-SIM imaging of cortical granules during exocytosis at the plasma membrane. Zona-free Mater Het (left) and Mater Null (right) eggs expressing ovastacin mCherry (red) were stimulated to exocytose and imaged in the presence of the membrane dye CellMask (green) at 1 s intervals (×50) (see corresponding Supplementary Movies 4,5). Two representative granules (inset) in circles that are color-coded for duration in seconds (color bar, lower left). Scale bar, 5 μm and 1 μm (inset). b Histograms comparing duration of cortical granule tracks in stimulated Mater Het (top; n = 3175) and Mater Null (bottom; n = 1511) eggs. Bimodal distribution with one peak representing short tracks (<8 s) and another peak for long tracks (49 s). Tracks between the two peaks were medium in duration. Student's t-test, p < 0.001) after fertilization of eggs from Mater Null females indicating a decrease in the zona block to penetration. This deprecation of the post-fertilization zona block was also observed in vivo (Fig. 6c) where more than 60% of fertilized zygotes collected from Mater Null females had one or more sperm in the perivitelline space (Fig. 6c, d; arrows; 1.82 ± 0.25 s.e.m.; chi-squared test, p < 0.001). The increased frequency of additional sperm in the perivitelline space was also detected in two-cell embryos derived from Mater Null females ( Supplementary Fig. 6a, b). More striking, the number of zygotes with three pronuclei collected from Mater Null females after in vivo mating increased to 15%. Consequently, one of the blastomeres in two-cell embryos from Mater Null females contained two nuclei instead of one ( Fig. 6e; Supplementary Fig. 6c). To follow the kinetics of CG exocytosis, we acquired confocal z-stacks of stimulated live whole mount eggs expressing ovastacin mCherry at 10-min intervals. The intensity of ovastacin mCherry fluorescence in the cortex of eggs heterozygous for the Mater null allele declined sharply 10-20 min after stimulation. In contrast, in eggs homozygous of the Mater null allele, there was an increase in cortical fluorescence 60 min after stimulation, presumably due to the continued translocation of CGs from the interior to the periphery of the egg  Fig. 7).
Due to maternally imposed embryonic lethality that precludes progression beyond 2-4 cell, Mater Null mice are infertile in vivo. 12 To determine the effect of myoIIA on in vivo fertility, MyoIIA Flox/Flox and egg-specific (Tg)Zp3-Cre mice were crossed to establish a mouse line (MyoIIA mNull/mNull ) with an eggspecific (maternal) conditional deletion of the gene (Fig. 6f;  Supplementary Fig. 6g). These female mice had smaller litters (3.3 ± 1.7) compared to MyoIIA Flox/Flox female mice (7.0 ± 2.4). After in vitro fertilization, significantly more sperm were observed in the perivitelline space of embryos derived from MyoIIA mNull/mNull females (Fig. 6g, h;   MyoIIA mNull/mNull mice. Taken together, these results document that MATER, a component of the SCMC, and non-muscle myosin IIA participates in CG trafficking and timely exocytosis to prevent polyspermy in mouse eggs.

Discussion
Three post-fertilization blocks to gamete interaction have evolved in mammals to prevent polyspermy that is embryonic lethal. In mice, the first two occur rapidly after fertilization and prevent additional sperm from fusing with the egg plasma membrane or penetrating the surrounding extracellular zona pellucida 18,30,31 . The third and definitive block prevents sperm from even binding to the surface of the zona matrix and depends on egg CG exocytosis and the resultant cleavage of ZP2. Using diffractionlimited and super-resolution imaging of individual CGs fluorescently stained with ovastacin mCherry in transgenic mice, we have investigated CG dynamics at the cortex-plasma membrane interface that prevents polyspermy in mouse eggs. Our data support a model in which MATER is present in the cortex underlying the egg's plasma membrane where it anchors CGs which are associated with non-muscle myosin IIA. During fertilization and egg activation, cortical actin is cleared and myoIIA is required for CGs to traffic to the plasma membrane prior to exocytosis (Fig. 7). The SCMC is formed in the subcortex of oocytes during oogenesis and is composed of at least five components including MATER and FLOPED [12][13][14][15][16] . While investigating a possible role in CG biology, we determined that MATER was required to position CGs in the periphery near the egg plasma membrane. The SCMC is symmetrically present in MII eggs 13 including a CG free region that is established by an actin cap 32 . Curiously, although MATER is not present in the subcortex in the absence of FLOPED 13 , we did not observe displacement of CGs in eggs derived from The number of sperm in the perivitelline space (PVS) for each genotype (9-36 two-cell embryos from three biologically independent samples). Box plot includes the mean (horizontal line) and data between the 25th and 75th percentile. Error bars indicate the 90th and 10th percentiles; outliers are indicated by dots. ***p < 0.001 by two-tailed Student's t-test. c In vivo fertilized Mater Het and Mater Null zygotes were fixed for staining with WGA-Alexa Fluor 633 (zona pellucida), phalloidin (F-actin) and DAPI (DNA). Arrows, supernumerary sperm in the PVS; pronuclei enclosed by dashed line. Scale bar, 20 μm. d Quantification of sperm in the PVS from c was determined for each genotype (13-60 embryos from three biological samples). ***p < 0.001 by chisquared test. e Quantification of pronuclei in embryos from c. **p < 0.01 by chi-squared test. f Ovulated eggs from wildtype (upper) and MyoIIA mNull (lower) were immunostained with WGA-Alexa Fluor 633 (magenta) and anti-MyoIIA (green). Representative confocal cross sections of z-projections are shown with a magnified view (inset) of the myoIIA region indicated by arrows. Scale bar, 20 μm. g After in vitro insemination, wild-type and MyoIIA mNull eggs were cultured to the 4-cell stage and fixed for staining with WGA-Alexa Fluor 633 (zona pellucida), phalloidin (F-actin) and DAPI (DNA). Arrows, supernumerary sperm in the perivitelline space; dashed lines, nuclei. Scale bar, 20 μm. h The number of sperm in the PVS for each genotype in g (15-18 embryos from two biologically independent samples). ***p < 0.001 by chi-squared test. Source data are provided as a Source Data file homozygous Floped Null female mice. Like the other SCMC members, MATER has motifs associated with protein-protein interaction, specifically 13 leucine-rich repeats 33 near its carboxyl terminus. Whether cortical anchoring of CGs by MATER is facilitated by direct or indirect interaction(s) with proteins present on the vesicular membrane is currently not known. However, MATER appears to be necessary, but not sufficient, for positioning CGs in the subcortex of mouse eggs.
In the present study, we document a dense actin meshwork underlying the egg plasma membrane where CGs are anchored. Pharmacological disruption of cortical actin in mouse eggs results in an increased appearance of granules near the plasma membrane prior to exocytosis which supports a model whereby cortical actin acts as a barrier to granule docking. A similar model has been proposed for bi-phasic insulin-granule exocytosis 34,35 in which F-actin is organized as a dense web beneath the plasma membrane to impede access of insulin-containing granules to the cell surface 36,37 . The first rapid phase involves membrane fusion of a limited pool of release-ready insulin granules already present at the plasma membrane 38,39 . Insulin granules that are deeper within the cell and trafficked to the cell periphery during the second phase secretion slowly replenish this pool 39 , similar to what is observed with CGs. In our model, long-duration vesicles, which are tethered to the plasma membrane, could be required for a slow and sustained release of cargo molecules, whereas short-duration vesicles likely interact transiently with the plasma membrane for a more rapid release as cortical actin is cleared.
Whereas biphasic trafficking is not affected in the absence of MATER, the interaction of long-duration granules is significantly impaired. Furthermore, we uncover an unexpected role for CGs of medium-duration. The strong decrease of medium-duration granules and the delay in exocytosis in Mater Null eggs suggests that this subpopulation of CGs is directly or indirectly required for one or more steps of secretion in mouse eggs, either by priming long-duration vesicles for the second phase of secretion by an unknown mechanism or coordinating timely clearance of cortical actin as well as trafficking of short-duration vesicles to the plasma membrane.
Alternatively, one or two myosin II isoforms could generate pulsatile contractile forces on the cortex, which would drive clearance of cortical actin during exocytosis [40][41][42] . Our finding that clearance of cortical actin is repressed in the presence of myosin inhibitors supports the concept that ATPase activity and activation of myoIIA are required to disassemble cortical actin. We provide genetic evidence in Mater Null eggs on the importance of myoIIA expression for actin clearance during exocytosis observing strongly reduced myoIIA levels, which could also impair pulsatile contractile forces. In addition to myosin II, myosin V was also reduced in Mater Null eggs. Importantly, myosin V has been demonstrated to associate with insulin-containing granules and CGs 7,43 , illustrating the requirement of the myosin motors behind granule movement to the cell periphery and secretion.
Unlike other secretory organelles, CGs are not renewed or maintained in the cytoplasm once released at the plasma membrane. Even though instant TIRF-SIM does not indicate directionality, the disappearance of single granule fluorescence appears to be an example of exocytosis rather their retreat into the cell interior and rejoining of the cytoplasmic pool. This conclusion is furthermore supported in Mater Null eggs in which exocytosis is delayed containing fewer disappearing, vanishing granules.
After fertilization, CGs release ovastacin, a metalloendopeptidase that cleaves ZP2 and precludes subsequent sperm binding to the zona pellucida 17 . As post-fertilization cleavage of ZP2 proceeds with slow kinetics (10-20 min) after in vitro stimulation, the coordination of actin clearance and exocytosis appears to be well adapted to the small size of mouse CGs and the regulated release of their contents. In the absence of MATER, supernumerary sperm accumulate in the perivitelline space between the egg plasma membrane and the inner aspect of the extracellular zona pellucida matrix. This is associated with delayed release of ovastacin and cleavage of ZP2, and is in accord with earlier studies in which genetic ablation of ovastacin prevents the normal post-fertilization block to sperm binding and penetration of the zona matrix 18 . The molecular basis of the postfertilization block to gamete fusion remains indeterminate, but is reported to be independent of CG exocytosis 30,44 and can be bypassed by intracellular sperm injection 31,45 . The relatively low level of polyspermy that we observe in Mater Null (15%) is more striking in MyoIIA mNull eggs which raises the possibility that the trafficking of CGs to the plasma membrane plays a role, either directly or indirectly, in the membrane block to gamete fusion.

Methods
Ethical compliance and source of reagents. Mice were maintained in compliance with the guidelines of the Animal Care and Use Committee of the National Institutes of Health under a Division of Intramural Research, NIDDK-approved animal study protocol. Unless otherwise noted, chemical and other reagents were obtained from Sigma-Aldrich. See Supplementary Table 2 for parameters of imaging experiments.
Mice and genotyping. Mater Null ( 12 ) and Floped Null ( 13 ) mice expressing ovastacin mCherry were generated by crossing the KO lines to transgenic Astl mCherry mice 32 . Transgenic Astl mCherry mice were also crossed to mice expressing EGFP-tagged non-muscle myosin IIA (MyoIIA EGFP ) 22,46 . Conditional Myosi-nIIA mNull (MyoIIA mNull ) mice were generated by crossing homozygous female MyosinIIA Flox/Flox mice to homozygous floxed males positive for Tg (Zp3-Cre) 47 Fig. 7 Model of cortical granule dynamics at the cortex-plasma membrane interface to prevent polyspermy in mouse eggs. MATER anchors cortical granules in the cortex underlying the egg's plasma membrane where they associate with non-muscle myosin IIA, a motor protein. During fertilization and egg activation, cortical actin is cleared and cortical granules traffic to the plasma membrane. Fusion of cortical granules at the plasma membrane releases their contents, including zinc and the metalloendopeptidase ovastacin, into the extracellular space. These cortical granule contents prevent sperm penetration of the zona matrix and cleave the zona pellucida protein ZP2 to ensure monospermic fertilization EmeraldAmp GT CR Master Mix (Takara Bio USA) and gene specific primers (Supplementary Table 1) were used to amplify specific DNA fragments. PCR was performed with annealing temperatures of 58°C and 30 cycles (for Ovastacin mCherry ), 58°C and 28 cycles (for Mater), 55°C and 30 cycles (for Floped), 58°C and 40 cycles (for MyosinIIA EGFP ), 60°C and 30 cycles (for Cd9), 58°C and 35 cycles (for MyosinIIA Flox ), 51.7°C and 35 cycles (for Zp3-Cre).
For in vitro fertilization, eggs were incubated in human tubal fluid medium (Zenith Biotech) for 6 h at 37°C with fresh motile sperm (5 × 10 5 ml −1 ) which had been capacitated for 2 h at 37°C. Zygotes and two-cell embryos were also obtained from females hormonally induced to ovulate followed by in vivo mating with males proven to be fertile. The presence of a copulation plug on the following day was used to confirm successful mating. Pronuclear-staged zygotes and two-cell embryos were isolated from oviducts and collected in M2 medium 49 .
Fixed and live confocal microscopy. Images were acquired with a Zeiss LSM 780 confocal microscope equipped with a Stable Z heating system (Bioptechs) and a 40x C-Apochromat 1.2 NA water-immersion objective lens. For live imaging, zonafree eggs were mounted on poly-L-lysine coated Bioptechs Delta-T culture dishes with pre-warmed M2 or CaCl 2 -free M16 medium at 37°C in 5% CO 2 . Typically, cells were first imaged by taking 6-8 μm z-stacks with 0.2 μm axial steps near the coverslip followed by time lapse acquisition in a single focal plane close to the coverslip every 5 s for up to 5 min. Images were then deconvolved using Huygens Essential Software version 4.0 (Scientific Volume Imaging). A theoretical pointspread function was based on microscope parameters, model confocal parameters established by Huygens and the Classic Maximum Likelihood Estimation algorithm was used to restore images. Sampling intervals were set manually to the actual experimental values, together with refractive indices and excitation/emission wavelengths.
Total internal reflection fluorescence microscopy (TIRFM). For TIRF imaging of CGs, zona-free eggs were mounted as described for confocal microscopy. Oocytes were excited using a 100 mW 488-nm laser introduced at the appropriate incident angle through the TIRF-slider (Zeiss) and ×63 objective (Zeiss, 1.46 NA Oil, alpha Plan-Apochromat). Images were acquired with a digital Evolve 512 EMCCD camera (Photometrics, Tucson, AZ, USA) controlled with Zeiss Zen 2 software. The size of each pixel was 254 × 254 nm. To observe CG motioñ 250 nm near the plasma membrane in mouse eggs, cells were imaged every 10 s for 5 min. Images were taken with an exposure time of 200 ms using 4% maximal laser output. Under these experimental conditions, no reduction in fluorescence from photobleaching was observed.
Instant structured illumination microscopy (iSIM). Images were acquired with a VT-iSIM Multi-Point Super Resolution Imaging System (VisiTech International) connected to an Olympus microscope body via a regular c-mount that housed objectives (Olympus UPLSAPO 60XS, 1.3 NA Silicone Oil, for live and fixed sample imaging, or 100 × 1.49 NA Oil objective, for fixed zona-free mouse eggs), an automated XY stage (Applied Scientific Instrumentation) and a custom-built heating System to maintain 37°C for live-imaging experiments. The imaging pixel size for the ×60 and ×100 objective was 107 nm and 64 nm, respectively. For excitation, 488 nm, 561 nm and 642 nm lasers were used. Lasers were combined with an automated 3-position dichroic changer. For live-imaging, zona-free eggs were mounted on poly-L-lysine coated 35 mm glass bottom microwell dishes (MatTek Corporation). Image acquisition was controlled with Meta-Morph software. Images were acquired with an exposure time of 250 ms using 25-40% maximal laser output. The illumination intensity at the back aperture was measured at 1.23 mW, 0.42 mW and 0.26 mW for 488 nm, 561, nm and 642 nm excitation, respectively. Z-stacks were captured with 0.2 μm axial steps (30-35 z-sections) for live samples and 0.1 μm axial steps (80-90 z-sections) for fixed samples. Images were subjected to GPU deconvolution 52 using an ImageJ plugin with a calculated point spread function, destriping and backgroundsubtraction (Microvolution LLC). Although 3D imaging techniques including confocal microscopy subject the mouse eggs to more dose than do TIRF techniques, we did not observe substantial bleaching or phototoxicity in our live experiments, suggesting that these factors did not influence our study. In addition, brightfield inspection of the morphology did not suggest any light-induced effects.
Live instant TIRF-SIM. Instant TIRF-SIM experiments were conducted on a laboratory assembled instant SIM 19 that had been modified to provide TIRF illumination 20 . Briefly, the main changes to the original instant SIM were: (1) placement of an annular mask in an upstream plane conjugated to the back focal plane of the objective, thereby removing low-angle, subcritical illumination and ensuring TIRF; and (2) using an ultrahigh numerical aperture (NA) objective lens (Olympus, APON100xHOTTIRF, 1.7 NA) with matched, high-index coverslips (Olympus, 9-U992) to facilitate TIRF. The imaging pixel size was 33.4 nm. 488-and 561-nm lasers were used for illumination, and either notch (Semrock, NF03-488E-25 and NF03-561E-25 notch filters for imaging green and red emission) or bandpass (Semrock, FF03-525 525/50, for imaging green emission) filters were employed to filter fluorescence. Zona-free eggs were mounted on poly-L-lysine coated highindex coverslips and imaged within the evanescent penetration depth of~120 nm. Exposure times were typically 40-50 ms. Illumination laser power before the objective was measured at 1.0 mW, implying an average intensity (given the 58 μm × 52 µm field of view) of~33 W/cm 2 . Temperature was maintained at 37°C using an incubation chamber (Warner Instruments, H301-MINI). The raw images acquired by instant TIRF-SIM were post-processed, including background subtraction and deconvolution, to achieve the full twofold resolution gain relative to conventional widefield fluorescence imaging. Background was usually estimated by averaging 100 images acquired without illumination. For deconvolution, we used the Richardson-Lucy algorithm with an experimental point spread function derived by registering and then averaging twenty 100 nm yellow-green beads. Deconvolution was implemented in MATLAB 2017a (Mathworks). The number of iterations was set to 10 for all images.
To validate these protocols, we imaged zona-free eggs from Cd9 Null female mice. CD9 is a tetraspanin protein that localizes to microvilli of membranes 53 and Cd9 Null eggs have an altered membrane due to varied length, thickness and density of their microvilli 54 . Projected confocal images detected a >1.5-fold reduction in the fluorescent intensity of Cd9 Null compared to wild-type eggs ( Supplementary  Fig. 2g, h). This observation suggests an altered cell surface that adversely affects the uptake of membrane dye and was confirmed at higher spatial resolution using instant TIRF-SIM ( Supplementary Fig. 2i).
Live-cell F-actin and plasma membrane labeling. F-actin was labeled with 1 μM of the live-cell Spirochrome probe SiR-actin (Cytoskeleton) which is based on natural actin-binding jasplakinolide. The plasma membrane was labeled with CellMask Green using the manufacturer's recommended 1x working solution (ThermoFischer Scientific). Zona-free eggs were mounted on poly-L-lysine coated coverslips) with pre-warmed M2 or CaCl 2 -free M16 medium supplemented with SiR-actin, CellMask Green and incubated at 37°C in 5% CO 2 for 30 min prior to imaging. Neither CG exocytosis or fertilization was inhibited under these experimental conditions even with 5 μm SiR-actin ( Supplementary Fig. 2j, k).
Electron microscopy. Super-ovulated eggs from wild-type and Mater Null females were fixed in 1% glutaraldehyde in PBS buffer (pH 7.4) and incubated at 4°C for 2 h. After extensive washing in PBS buffer, the eggs were embedded in 2% agarose. The samples were then dehydrated through a graded series of ethanol solutions and processed for embedding in LR-White resin. Ultrathin sections were obtained with an ultramicrotome (Microm International GmbH) and mounted on formvar coated nickel grids. For lectin-cytochemistry, grids were incubated with WGA-HRP as previously described 17 . Ultrathin sections were counterstained with uranyl acetate followed by lead citrate and imaged in a Phillips Tecnai 12 transmission electron microscope. The percentage of WGA-positive CGs was calculated over twelve ultrathin sections of five eggs from wild-type and Mater Null females. Image analysis was performed using Software Mip4 Advanced (Microm Image Processing Software, Digital Image Systems).
In vitro cleavage assay. The post-fertilization cleavage of ZP2 (120 kD) results in two fragments (30 and 90 kD) rendering the zona matrix unreceptive to sperm binding. Ovulated and parthenogenetically activated eggs as well as two-cell embryos with intact zonae pellucidae from wildtype, Mater Null , and Floped Null mice were resolved by SDS-PAGE and cleavage was analyzed by immunoblotting with monoclonal antibody m2c.2 to the C-terminus of mouse ZP2 55 . After stripping membranes with Restore Western Blot Stripping Buffer (Thermo Fisher Scientific), the membranes were reprobed with α-MATER (1:1000), α-FLOPED (1:1000) or αtubulin (1:1000) antibody. Uncropped images of immunoblots are shown in Supplementary Fig. 8.
Image analysis. Line profiles were generated in Fiji 56 and measurements were exported to Excel for visualization. Kymographs analyzing the fluorescent signal over time were prepared using the Multi Kymograph tool in Fiji. To detect and measure cortical actin clearances, deconvolved images were imported to Fiji and normalized using the percentile threshold with "dark background" unchecked. Clearances were detected using the analyze particle function with a clearance area of 1-40 μm 2 with the assumption of uniform circularity set at 0.2-1.00. Clearance measurements (number, area) were exported to Excel for statistical analysis and the frequency of clearances was determined per μm 2 per given cell.
To follow CGs coming from TIRFM, iSIM and instant TIRF-SIM datasets, we performed semi-automated tracking using the plugin TrackMate Fiji Plugin 57 (http://fiji.sc/TrackMate). For particle detection, the difference of gaussian (DoG) detector was used with estimated blob diameter of 0.6 μm, and an initial quality threshold of 120. The particles were further filtered using auto thresholds and linked with a simple linear assignment problem (LAP) linker. Linking and gap-closing distances were set to 1.0 μm and the maximum frame gap was set to 2.
We developed a Python library for working with TrackMate output data and performing analysis. Briefly, TrackMate data were imported and tracks assigned unique identifiers. For each track, derived values were calculated for each timepoint such as displacement and percent of max intensity. Aggregate values were then calculated per track, such as R 2 , slope, and p-value of linear regression of percent max intensity; bounding ellipses; duration; total displacement. Thresholds for duration classes (short, medium, long) were empirically determined based on histograms of duration. To identify tracks as approaching or vanishing, we required the linear regression of percent-of-max-intensity to have R 2 > 0.6 (empirically determined based on histograms), p < 0.05, and either a positive (for approaching) or negative (for vanishing) slope. This algorithm can be generalized for analyses in other systems.
Statistical analysis. Unpaired two-tailed Student's t test was used for comparison of two samples with normal distribution. N-1 chi-squared test was used for comparison of percentages. p < 0.05 was considered significant. Statistical analyses and graphing were performed using Excel and SigmaPlot 12.3 (Systat Software). All bar and boxplot graphs presented in this study indicate the mean value and the standard error of the mean (s.e.m.).
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The source underlying Figs. 1d, 1i, 2d, 2f, 2h, 4c, 5c, 5f, 6b, 6d, 6e, 6h, and Supplementary Figs. 2h, 6 and 7 are provided as a Source Data File. Other data that support the findings of this study are available from the corresponding author on request.

Code availability
A zip file of the Python library, the Jupyter notebook, the raw data and a full environment specification of the installed packages used for analysis of TrackMate output data can be found at https://doi.org/10.5281/zenodo.2605576.