Cortical recruitment of centralspindlin and RhoA effectors during meiosis I of Caenorhabditiselegans primary spermatocytes

Cytokinesis is a critical step in the creation of two daughter cells from a single progenitor cell. While cytokinesis must follow nuclear division, the mechanisms that coordinate these two steps are not completely understood. Moreover, there is evidence that these mechanisms may differ among cell types, even in a single organism. Although meiosis, the processes by which recombinant haploid genomes arise during gametogenesis, has been extensively studied in a number of organisms, including the nematode Caenorhabditis elegans, the division of the meiotic cells during spermatogenesis is far less well characterized.

In C. elegans, spermatogenesis occurs during the L4 larval stage in males and hermaphrodites. C. elegans syncytial pachytene spermatocytes (equivalent to the spermatogonia in vertebrates) are attached to the syncytial cytoplasmic core, known as the rachis (L'Hernault, 2006; Ward et al., 1981). When animals reach the L4 larval stage, meiosis begins, and rachis-attached primary spermatocytes undergo meiotic recombination after which primary spermatocytes bud off from the rachis and enter pro-metaphase of meiosis I (Shakes et al., 2009). The 4N primary spermatocyte divides into two 2N secondary spermatocytes, each of which further divides into two haploid spermatids through myosin-II-mediated incomplete cytokinesis. The resulting spermatids are connected by an anucleate residual body, and the spermatids subsequently detach from the residual body through a myosin-VI-mediated process and, ultimately, differentiate into spermatozoa (Hu et al., 2019).

An actomyosin-based contractile ring is the apparatus that mediates the cell membrane ingression that occurs at the cell equator during the final step in mitosis and meiosis. The timing of contractile ring assembly is tightly coordinated with chromosome segregation, and this involves precise temporal control of cell cycle regulators. Chromosome segregation is initiated when the anaphase-promoting complex (APC) ubiquitylates cyclin B1 (CYB-1 in C. elegans) and separase-inhibitor securin (IFY-1 in C. elegans) and targets them for proteasomal destruction (Peters, 2002). Once cyclin B1 and securin are degraded, separase (SEP-1 in C. elegans) cleaves the cohesin rings that link the two sister chromatids, following which the chromosomes segregate to their respective poles (Cheeseman and Desai, 2008; Rago and Cheeseman, 2013; Santaguida and Musacchio, 2009; Danlasky et al., 2020). On the other hand, cyclin B1 degradation also inactivates Cdk1, and this allows centralspindlin to associate with microtubules and to concentrate on the central spindle (Mishima et al., 2004). Cdk1 inactivation also promotes the association of Ect2, a Rho guanine-nucleotide-exchange factor (RhoGEF), with centralspindlin and the cell membrane (Yüce et al., 2005; Su et al., 2011). Activated Ect2 then promotes activation of the small GTPase RhoA (also known as RHO-1 in C. elegans) to indirectly activate non-muscle myosin II (NMY-2 in C. elegans) (Matsumura, 2005) and directly activate formin-mediated F-actin assembly for contractile ring formation (Otomo et al., 2005).

The central spindle, which is composed of overlapping, antiparallel, non-kinetochore microtubules, forms between the segregating chromosomes during anaphase. The interdigitated plus ends of the central-spindle microtubules associate with both centralspindlin and the chromosomal passenger complex (CPC) (Glotzer, 2009). Centralspindlin is a heterotetrameric complex composed of two molecules each of a Rho GTPase-activating protein (RhoGAP) (MgcRacGAP, also known as RACGAP1, in mammals; CYK-4 in C. elegans), and the kinesin-6-family member MKLP1 (KIF23 in mammals, ZEN-4 in C. elegans) (Mishima and Glotzer, 2003; Pavicic-Kaltenbrunner et al., 2007); this complex is required for organizing the central spindle (Jantsch-Plunger et al., 2000) and activating RhoA for furrow induction (Basant and Glotzer, 2018). Oligomerization of centralspindlin promotes the processivity of MKLP1 on microtubules and consequently the accumulation of centralspindlin at the central spindle (Hutterer et al., 2009). Before anaphase onset, in human cells, 14-3-3 proteins bind to phosphorylated S710 on MKLP1 to inhibit centralspindlin oligomerization. However, upon anaphase onset, the serine/threonine protein kinase component of the CPC, Aurora B kinase (AURKB in human, AIR-2 in C. elegans), phosphorylates MKLP1 at S708 to prevent the interaction of 14-3-3 and MKLP1, and, consequently, allows centralspindlin oligomerization (Guse et al., 2005; Douglas et al., 2010). Moreover, in C. elegans zygotes, depletion of PAR-5, the 14-3-3 protein expressed in the germline (Wang and Shakes, 1997), can induce ectopic membrane recruitment of centralspindlin (Basant et al., 2015), and a point mutation at S682 on ZEN-4, the conserved binding site for 14-3-3 in MKLP1, recapitulates many consequences of par-5 depletion (Basant et al., 2015). Thus, centralspindlin oligomerization is inhibited by PAR-5, but this can be overridden by AIR-2 phosphorylation of ZEN-4 (Basant et al., 2015; Douglas et al., 2010).

Several events of meiosis have been studied in spermatocytes, but the cellular division of primary spermatocytes into secondary spermatocytes has not been comprehensively characterized. Therefore, we investigated this process by performing high-resolution time-lapse confocal microscopy, and we examined the spatiotemporal regulation of contractile ring assembly in C. elegans primary spermatocytes by using probes for the contractile ring, spindle midzone, DNA and nuclear pores, as well as for cell cycle regulators. We show that cyclin B degradation significantly precedes securin degradation. Centralspindlin and RhoA effectors were recruited to the equatorial cortex of dividing primary spermatocytes for contractile ring assembly before securin degradation and successful homologous chromosome segregation, and this process could occur in spermatocytes in which the activity of the APC was significantly reduced. Furthermore, we show that primary spermatocytes require the PAR-5–ZEN-4 interaction for proper initiation of cellular division. Our findings help enhance current understanding of the division of primary spermatocytes in sperm production.

To investigate the cellular division of primary spermatocytes, we imaged primary spermatocytes in L4 hermaphrodites expressing non-muscle myosin II (NMY-2::GFP) and mCherry-tagged histone H2B (mCherry::H2B), focusing on the primary spermatocytes situated at the proximal end of the rachis (Fig. 1A). Time zero was defined as one frame before the moment at which homologous chromosomes exhibited clear segregation in anaphase I (Fig. 1B). Chromosome congression at the metaphase plate was apparent for ∼2.5 min prior to anaphase onset. Before entering metaphase of meiosis I, as assessed by nuclear envelope breakdown (NEBD), NMY-2::GFP was mostly localized at the rachis and the cortex of primary spermatocytes. However, ∼7 min before (−420 s) detectable segregation of homologous chromosomes, NMY-2::GFP began to gradually accumulate at the equatorial cortex, and after another ∼3–4 min (−240 to −180 s), a clear NMY-2::GFP-labeled contractile ring was apparent (Fig. 1C). Kymograph analysis and quantitation of cortical intensity demonstrated that cortical accumulation of NMY-2::GFP began before homologous chromosomes segregated and continued following anaphase as the membrane ingressed (Fig. 1C,D; Fig. S1A, Movie 1). Similar behavior of NMY-2::GFP accumulation in primary spermatocytes was observed in the germline of C. elegans males (Fig. S1B). The actomyosin scaffold protein ANI-1 also gradually accumulated at the equatorial cortex, beginning at −360 s (Fig. 1E,F; Fig. S1A). During meiosis I, contractile rings began to slightly constrict before detectable homologous chromosome segregation, but this accelerated upon anaphase (Fig. 1D,F).

To test whether the NMY-2::GFP that accumulates at the equatorial cortex originates from the rachis, we photobleached rachis-localized NMY-2::GFP and then examined the localization of NMY-2::GFP in the primary spermatocytes (Fig. 1G). Extensive photobleaching of rachis-localized NMY-2::GFP did not prevent accumulation of labeled myosin II at the equatorial cortex of primary spermatocytes (Fig. 1H,I; Movie 2); NMY-2:GFP at the rachis did not recover significantly during the imaging period. Thus, the contractile ring components at the rachis are stably associated with that site and do not make a major contribution to the contractile ring.

The aforementioned observations suggest that contractile ring components accumulate at the equatorial region prior to the clear separation of homologous chromosomes. This previously unreported order of events in primary spermatocytes could result from delayed chromosome segregation after the metaphase-to-anaphase transition or from accumulation of contractile ring components before the metaphase-to-anaphase transition. To distinguish between these possibilities, we examined cell cycle progression by following the behavior of various cell cycle markers. We examined a nuclear pore component, NPP-2, to detect the integrity of the nucleus, and monitored the levels of securin (IFY-1), separase (SEP-1) and cyclin B (CYB-1). Using NPP-2::GFP and mCherry::H2B, we measured the time interval between NEBD and segregation and detected the morphology of homologous chromosomes in primary spermatocytes. At approximately −520 s (range: −570 to −480 s) NEBD occured (Fig. 2A,B; Movie 3), and the homologous chromosomes started to congress at the metaphase plate. At time zero, the homologous chromosomes began to segregate to opposite poles during anaphase I (Fig. 1B, Fig. 2A).

Next, we examined the timing of anaphase I events in primary spermatocytes by investigating the distribution and intensity of securin (IFY-1::GFP) and separase (SEP-1::GFP). IFY-1::GFP accumulated on the chromatin abruptly at −540 s upon NEBD. More importantly, the IFY-1::GFP intensity dropped sharply at the onset of detectable homologous chromosome separation, and the signal fully disappeared at 60 s, a time point at which homologous chromosomes were significantly segregated (Fig. 2C,D; Movie 4). Separase, SEP-1::GFP, accumulated on the chromatin more gradually than IFY-1::GFP and it briefly accumulated on the spindle midzone immediately following the onset of homologous chromosome segregation, before dropping below the detection limit at 90 s (Fig. 2E,F; Movie 4). These results suggest a conventional order of events during meiosis. Lastly, we measured the dynamics of CYB-1::YFP in meiosis I. Before NEBD, CYB-1::YFP accumulated in the nucleus of primary spermatocytes. Upon NEBD, CYB-1::YFP colocalized with the chromatin and its levels decreased gradually throughout meiosis I, and the signal was undetectable by −90 s (Fig. 2G,H; Movie 5). Thus, two APC substrates, CYB-1 and IFY-1, exhibited distinct temporal patterns of destabilization, with IFY-1 disappearing later and more suddenly than CYB-1. Nevertheless, the localization patterns of IFY-1::GFP and SEP-1::GFP suggest that the metaphase-to-anaphase transition occurs just before detectable separation of homologous chromosomes, and that contractile ring assembly significantly precedes these events.

We next investigated whether the APC regulates the timing of contractile ring assembly. Depletion of either MAT-1 or EMB-30 by RNAi did not inhibit segregation of homologous chromosomes (Fig. S2A–C), suggesting that the proteins were incompletely depleted. Nevertheless, in these spermatocytes, ANI-1::GFP was robustly recruited to the equatorial cortex before the onset of anaphase I. This contractile ring component was recruited more robustly than in control spermatocytes, consistent with a delay in anaphase onset. To more fully inactivate the APC, we used RNAi to deplete EMB-30 in mat-1(ax144) mutant animals. In spermatocytes in which APC was depleted to levels sufficient to cause chromosome segregation failure and M-phase arrest (∼30% of the imaged spermatocytes), contractile ring assembly and constriction were still observed (Fig. 2I,J; Movie 6). Thus, APC activity might be dispensable for contractile ring assembly, though we cannot rule out the possibility that residual low levels of APC activity might suffice to trigger its assembly.

To further characterize cell division during meiosis I, we imaged dividing spermatocytes expressing SPD-1::GFP and mCherry::H2B to determine when the spindle midzone assembles; SPD-1 is a microtubule-bundling protein that is required for spindle midzone stability (Verbrugghe and White, 2004). Before NEBD, SPD-1::GFP accumulated in the nuclei of primary spermatocytes (Fig. 3A,B), whereas following NEBD, SPD-1::GFP intensity dropped sharply and then faintly accumulated around the centrosomes at around −120 s (Fig. 3A). Subsequently, as homologous chromosomes began their poleward movement, SPD-1::GFP localized to the spindle midzone and its intensity increased rapidly, peaking at ∼90–120 s, after which its intensity began to decline rapidly, and the signal disappeared by ∼240 s (Fig. 3C; Movie 7). These results indicate that the spindle midzone assembles after homologous chromosomes exhibit detectable segregation and that a stable midzone structure is absent during mid-to-late anaphase I. This agrees with recently published work (Fabig et al., 2020).

Next, we investigated the pattern of centralspindlin accumulation during meiosis I in primary spermatocytes. GFP::ZEN-4 and GFP::CYK-4 were observed in the nucleus during interphase. After NEBD, centralspindlin localized around the chromatin, and colocalized with the aligned chromosomes around metaphase I (Fig. 4A,B). Unexpectedly, GFP::CYK-4 and GFP::ZEN-4 were detected at the cortex during metaphase I and continued to accumulate equatorially (Fig. 4C,D; Movie 8). Cortical localization of CYK-4 and ZEN-4 was interdependent in primary spermatocytes (Fig. S3). As the chromosomes visibly separated, centralspindlin transiently localized at the midzone, but became almost undetectable by ∼90 s. As centralspindlin dissociated from the midzone, it prominently accumulated at the equatorial cortex (Fig. 4A,B,E).

To ascertain whether ZEN-4 and CYK-4 are required for NMY-2 recruitment at the equatorial cortex, we imaged and analyzed the dynamics of cortical NMY-2::GFP after RNAi-mediated depletion of either ZEN-4 or CYK-4. The penetrance of the RNAi treatment was examined by quantifying the levels of their corresponding fluorescent markers in the RNAi experiments (Fig. S4). As compared with the localization in control cells (Fig. 1C), the cortical localization of NMY-2::GFP was markedly delayed in both zen-4(RNAi) primary spermatocytes and cyk-4(RNAi) primary spermatocytes (Fig. 4F,G,I); residual recruitment of NMY-2::GFP might result from incomplete target depletion by RNAi. Moreover, NMY-2::GFP accumulation was significantly delayed in primary spermatocytes of cyk-4(or749) mutant animals at the restrictive temperature (Fig. 4H,I), and the rate of contractile ring constriction was significantly reduced in cyk-4(or749) (Fig. 4J). Conversely, NOP-1 depletion did not markedly alter the cortical recruitment of NMY-2 relative to that in control cells (Fig. S2D,E). These results suggest that centralspindlin promotes NMY-2 cortical localization and contractile ring assembly during meiosis I.

Because cortical localization of centralspindlin is inhibited by PAR-5 in C. elegans zygotes (Basant et al., 2015), we investigated whether PAR-5 also regulates the localization of centralspindlin in primary spermatocytes. First, we examined the subcellular localization of PAR-5 in primary spermatocytes. PAR-5::GFP was mostly detected in the nucleus before NEBD, and after NEBD, it remained in the vicinity of the chromosomes and its intensity decreased gradually. At anaphase I, PAR-5::GFP localized to the spindle midzone and then dissociated following homologous chromosome segregation (Fig. S5A,B).

We next determined the subcellular localization of ZEN-4 in spermatocytes depleted of PAR-5. The intensity of ZEN-4 at the rachis did not differ significantly between control and par-5(RNAi) primary spermatocytes (Fig. 5A,B,D; Movie 9), but the nuclear-localized or chromatin-associated pool of ZEN-4 was markedly diminished in par-5(RNAi) (Fig. 5A,B,E). Nevertheless, GFP::ZEN-4 was detected at the equatorial cortex, albeit in reduced amounts and with a delay relative to the localization in controls (Fig. 5F). These results suggest that PAR-5, directly or indirectly, stabilizes ZEN-4 during meiosis I.

Centralspindlin oligomerization and cortical localization are regulated by antagonistic interactions between AIR-2 and PAR-5 (Basant et al., 2015; Douglas et al., 2010) (Fig. 5G). In addition, PAR-5 may have multiple substrates in spermatocytes, therefore we mutated the PAR-5-binding phosphosite in ZEN-4 to examine how PAR-5 regulates centralspindlin. We hypothesized that abolishing ZEN-4 S682 phosphorylation would mimic PAR-5 depletion. To test this, using the CRISPR/Cas9 system, we generated a ZEN-4(S682A) GFP reporter and compared the intensity of this marker at the rachis and in the nucleus with that of the wild-type tagged protein (Fig. 5C; Movie 10). Whereas the intensity of GFP::ZEN-4 and GFP::ZEN-4(S682A) did not differ markedly at the rachis (Fig. 5D), the nuclear intensity of GFP::ZEN-4(S682A) was significantly weaker than that of control (Fig. 5E). Furthermore, as observed in PAR-5-depleted spermatocytes, GFP::ZEN-4(S682A) accumulated at the equatorial cortex at reduced levels, and its recruitment was considerably slower relative to that in the control, both before and after homologous chromosome segregation (Fig. 5F). However, the nuclear intensity of GFP::ZEN-4 in par-5(RNAi) was statistically stronger than that of GFP::ZEN-4(S682A); this is likely due to the homozygous CRISPR mutant allele having a stronger effect on centralspindlin, as compared to the effect of possibly incomplete depletion of PAR-5 by RNAi.

Next, to test whether ZEN-4(S682A) regulates the cortical recruitment of NMY-2, we examined NMY-2::GFP and mCherry::H2B in the zen-4(S682A) background. Similar levels of NMY-2::GFP accumulated at the rachis in zen-4(S682A) mutant animals and controls (Fig. S5C), but cortical recruitment of NMY-2 was reduced and contractile ring constriction was delayed in zen-4(S682A) mutant animals (Fig. 5H,I). These findings suggest that PAR-5 binding to ZEN-4 at S682 stabilizes centralspindlin and thereby indirectly promotes the post-NEBD accumulation of centralspindlin at the cortex; this, in turn, results in timely cortical myosin accumulation, consistent with our other evidence indicating that myosin accumulation is dependent on centralspindlin.

The order in which primary spermatocytes enter mitosis is known to be spatially organized at the proximal end of the gonad in young L4 worms (Fig. 1A); however, the mechanism that regulates this order remains unknown. Thus, we examined the division of the first four primary spermatocytes: the primary spermatocyte located at the most proximal end divided first (cell 1), followed by the cell adjacent to it in the same row (cell 2), which was followed by each of the remaining primary spermatocytes in a proximal-to-distal direction (Fig. 6A). However, in both par-5(RNAi) and zen-4(S682A) mutant animals, this order was not followed, and the primary spermatocytes instead divided in an apparently random order (Fig. 6B–D; Movie 11). To determine whether these regulatory mechanisms affect the outcome of spermatogenesis, we measured sperm numbers in the spermatheca in control, par-5(RNAi) and zen-4(S682A) mutant animals. Sperm nuclei were labeled by SPE-11::mCherry (Fig. S6A). Whereas the number of sperm stored per spermatheca was 103.8±4.1 (mean±s.e.m.) in control, the number was reduced to 70.8±6.1 and 59.2±4.6 in par-5(RNAi) and zen-4(S682A) mutant animals, respectively (Fig. S6B). These results indicate that the PAR-5–ZEN-4 interaction is required for proper primary spermatocyte division sequence and sperm production; however, the underlying mechanism is unknown.

During mitotic divisions, non-muscle myosin II (NMY-2), anillin (ANI-1) and actin filaments are functionally important contractile ring components that are recruited to the equatorial cortex during anaphase (Miller, 2011). During anaphase, contractile ring assembly at the equatorial cortex is regulated by a combination of aster-mediated inhibition of contractile proteins at the polar cortex and midzone activation of contractile proteins at the site of furrowing (Pintard and Bowerman, 2019). The sequential order of chromosome segregation and contractile ring assembly guarantees the separation of the two sets of chromosomes into two daughter cells. However, our results demonstrate that the contractile ring assembles at the division plane and ingresses slowly in primary spermatocytes before visible homologous chromosome segregation. Pre-anaphase contractile ring assembly can be induced by optogenetic activation of RhoA in cultured mammalian cells during metaphase, suggesting that RhoA activation, and not its downstream effectors, is subject to cell cycle control in mitotic cells (Wagner and Glotzer, 2016).

Homologous chromosomes begin to separate at the onset of anaphase I, when separase proteolytically cleaves the cohesin that binds the chromosomes together (Siomos et al., 2001). Cohesin activation is regulated by a combination of securin destruction and dephosphorylation of Cdk1 sites on separase (Dawson et al., 2018). In primary spermatocytes, the visible separation of homologous chromosomes correlated with the degradation of securin (IFY-1) and was preceded by the accumulation of separase (SEP-1) on the homologous chromosomes. The spindle midzone (labeled by SPD-1::GFP) assembled between the visibly separated homologous chromosomes. However, despite the sudden destruction of securin just before chromosome segregation, cyclin B degraded gradually before homologous chromosome segregation and before centralspindlin and myosin were recruited to the cortex. Interestingly, while impairment of APC function strongly delayed chromosome segregation, it did not delay contractile ring assembly. The simplest interpretation of these results is that APC activation occurs just prior to the segregation of homologs, but another process drives cyclin B destabilization and contractile ring assembly. There are precedents for APC-independent destruction of cyclins during meiotic metaphase in C. elegans. ZYG-11–CUL-2, a distinct E3 ubiquitin ligase, contributes to cyclin destabilization during oocyte meiosis (Liu et al., 2004; Sonneville and Gonczy, 2004). The involvement of this ubiquitin ligase in cyclin B degradation in meiotic primary spermatocytes warrants further investigation.

In C. elegans embryonic blastomeres, centralspindlin localizes at the spindle midzone during anaphase, is required for the integrity of the spindle midzone, and remains at the midbody until abscission (Hutterer et al., 2009; White and Glotzer, 2012). Centralspindlin can be detected on ingressing furrows in these blastomeres and is essential for promoting ingression in embryos deficient in NOP-1; however, centralspindlin accumulates at the cortex at levels that are only just above the limit of detection (Basant et al., 2015; Tse et al., 2012).

A drastically different pattern of centralspindlin accumulation was observed in primary spermatocytes. In these cells, centralspindlin accumulated at the cortex during metaphase I, and it was recruited to the central spindle during early anaphase I, but soon (∼90 s) after homologous chromosomes separated, centralspindlin dissociated from the central spindle. At this moment, the central spindle (labeled by SPD-1::GFP) was still present between the segregated homologous chromosomes. However, centralspindlin accumulated robustly on the equatorial cortex and was required for furrow ingression in spermatocytes. Thus, although centralspindlin can be detected at both the midzone and on the equatorial membrane during anaphase in both blastomeres and primary spermatocytes, the relative accumulation of centralspindlin at these two cellular locations differs markedly between the two cell types.

During mitosis, cortical accumulation of centralspindlin is regulated by its oligomerization, which is mediated by a small region at the C-terminus of the coiled-coil region of ZEN-4. Centralspindlin oligomerization is prevented by PAR-5/14-3-3 binding to ZEN-4 phospho-S682, and this binding is relieved by Aurora B kinase phosphorylation of ZEN-4 at S680 (Basant et al., 2015; Douglas et al., 2010). This phosphoregulation also appears to regulate centralspindlin in spermatocytes, but it has distinct consequences. Depletion of PAR-5 or mutation of the S682 site to alanine (to preclude PAR-5 binding) resulted in a reduction in both the nuclear and cortical pools of centralspindlin, but not on the rachis. These results suggest that PAR-5 may stabilize cytoplasmic ZEN-4. PAR-5/14-3-3-family proteins have been reported to regulate the stability of numerous proteins. Stabilization by 14-3-3-family members has been reported for Cdt2 (Havens and Walter, 2011), leucine rich repeat protein kinase 2 (LRRK2) (Zhao et al., 2015) and phosphatidylinositol 4-kinase IIIβ (PI4KB) (Chalupska et al., 2017), whereas a 14-3-3 family member promotes the degradation of MDMX, a homolog of MDM2 (LeBron et al., 2006).

Unlike germ cells in the mitotic zone of the gonad in C. elegans (Lan et al., 2019), primary spermatocytes are oriented regularly at the proximal end of the gonad in late-L4 worms, and their division entry is stereotyped (Fig. 6). The first primary spermatocyte in the proximal end starts NEBD and undergoes meiosis I, and then is followed by its neighboring cells, from the proximal end to the distal end. This ordered division of primary spermatocytes was disrupted in PAR-5-depleted animals or in animals with the nonphosphorylatable ZEN-4 variant S682A. Depletion of PAR-5 more severely impacted the order of division than mutation of the PAR-5 binding site of ZEN-4, suggesting much of the effect is independent of centralspindlin. Indeed, PAR-5 is required for cell cycle regulation and the DNA damage checkpoint in the C. elegans germline (Aristizabal-Corrales et al., 2012). It is possible that mutation of the PAR-5 binding site on ZEN-4 indirectly impacts cell cycle control by increasing the effective pool of PAR-5. If this speculation is correct, it would imply that meiotic progression is highly sensitive to PAR-5 levels in the C. elegans germline.

All C. elegans strains were maintained at 16°C or 22°C on NGM plates (3 g NaCl, 2.5 g peptone, 17 g agar in 975 ml H₂O) seeded with Escherichia coli strain OP50. The strains used in this study are listed in Table S1.

ANA071 [adels [pMD191; mex-5::spd-1::GFP] II; itls37 [pAA64; pie-1::mCherry HIS-58] IV; itls38 [pAA1; pie-1::GFP::PH]] was a kind gift from Professor Marie Delattre (French National Centre for Scientific Research, Paris, France). An N-terminal GFP insertion at an endogenous site in ani-1, zen-4, zen-4(S682A) and npp-2 was generated using the CRISPR/Cas9 system, as described previously (Dickinson et al., 2015); an N-terminal GFP insertion at an endogenous site in cyk-4 was generated by SunyBiotech Company [strain name and genotype: PHX1060, cyk-4(syb1060)]. The Cas9–sgRNA constructs were obtained by inserting the target sequence into pDD162 (Dickinson et al., 2015), and the homologous-repair template plasmids were constructed through the insertion of the two homology arms into pDD282 (Dickinson et al., 2015) digested with ClaI and SpeI. The primers used in this study are listed in Table S2.

We followed the RNAi feeding methods described previously (Kamath and Ahringer, 2003). RNAi plasmids were obtained from an RNAi library (Source Bioscience). Bacteria were cultured and selected for ampicillin resistance in LB for 12 h at 37°C and were seeded onto NGM plates containing 100 μg/ml ampicillin and 1 mM IPTG. L1-stage worms were fed on these RNAi-feeding plates until microscopic imaging.

Late-L4 worms were anesthetized with 0.5% tetramisole, and young adult worms were dissected to free embryos. The samples were mounted on a 3–5% agarose pad with a coverslip and then sealed with petroleum jelly. In time-lapse imaging, GFP was visualized using 488 nm excitation and 525–550 nm emission filter sets, whereas RFP and mCherry were visualized using 561 nm excitation and 607–636 nm emission filter sets. Images in all channels were captured using 100 ms exposure time and 30 s intervals on an inverted spinning-disk confocal microscope (Olympus), using a Yokogawa CSU-X1 system, equipped with a 60×/1.4 NA objective and an EMCCD camera. All movies were acquired using MetaMorph software. Multi-z sections spanning 1 μm were captured, and a maximum projection was calculated. All movies were captured at ∼30–60 min before anaphase onset of the first spermatocyte in the proximal end, and the cell division of the most proximal five spermatocytes was recorded for statistical analysis.

Photobleaching experiments were performed on a spinning-disk confocal microscope equipped with a Mosaic system (Andor Technologies). The region of interest (ROI) was set using the oval selection tool, and photobleaching was performed for 2000 ms using 20% laser power. For comparison, the first proximal cell division was imaged before photobleaching, and then the nuclear region of the following cell was photobleached and the cortical dynamics were captured for subsequent analysis.

To calculate sperm number, L1-stage hermaphrodites expressing SPE-11::mCherry were placed on NGM plates seeded with corresponding RNAi bacteria and grown for ∼40 h. Subsequently, spermatheca of the worms that had completed spermatogenesis but contained no fertilized embryos were photographed and used for calculating the sperm number. In worms expressing zen-4(S682A);mCherry::H2B, sperm number was calculated using histone as the sperm marker. The projection images were used, and individual sections were also examined to ensure that no sperm was missed. For brood size measurement, three early L4-stage hermaphrodites were placed on one 3.5 cm NGM plate containing RNAi bacteria for 24 h and then transferred onto new plates daily. The total embryos and the hatched F1 generation were counted for quantification.

For fluorescence intensity analysis, the time of anaphase onset was set as the reference time of 0 s. The z-planes containing the entire cortical region were projected into a single image by using the batch z-projection process function in ImageJ (NIH, Bethesda, MD). To quantify protein dynamics, fluorescence intensity was measured using ImageJ in the following regions (set using the freehand selection tool of ImageJ): the cell cortex region that formed the contractile ring and was away from the rachis (F_Cortex), the background region (F_Background), the rachis region (F_Rachis) and the nucleus region (F_Nucleus). To eliminate any discrepancy between different worms or experimental microenvironments, the cortical intensity was normalized against the cortical intensity before the appearance of contractile ring components (F_{Cortex-background}). The F_{Cortex-background} was the average value of the cortical intensity of several time points before the assembly of the contractile ring. Thereafter, the data were normalized using this formula: F_Normalized=(F_Cortex/F_Background)/F_{Cortex-background}. The fluorescence intensities of the most proximal five nuclei were calculated and normalized to F_Background for quantification. Lastly, the normalized intensity was plotted against time. Statistical significance of differences between distinct treatments or samples was determined by applying the unpaired two-tailed Student's t-test by using GraphPad Prism software.

We thank SUSTech Core Research Facilities for providing equipment used in this study.