Anillin and the microtubule bundler PRC1 maintain myosin in the contractile ring to ensure completion of cytokinesis

In animal cells, cytokinesis involves the assembly and constriction of a cortical actomyosin contractile ring that forms during anaphase at the equator of the cell and is anchored to the plasma membrane. Contractile ring assembly depends on the activation of the RhoA GTPase that in turn activates a formin to nucleate non-branched actin filaments and non-muscle myosin II to arrange the actin filaments and generate the force for ring constriction (Pollard and O'Shaughnessy, 2019). Other components of the contractile ring include membrane-associated septin filaments and the scaffolding protein anillin that binds actin, myosin and septins. The plasma membrane is pulled behind the contractile ring as it constricts, generating a cleavage furrow that separates the two sister cells until only a narrow cytoplasmic bridge connects them. Abscission terminates cytokinesis by completing the physical separation of sister cells.

Contractile ring assembly requires signals from the anaphase spindle, which consists of midzone and astral microtubules. Midzone microtubules form between the two sets of segregating chromosomes as anti-parallel bundles with plus ends interdigitating at the center and minus ends facing the spindle poles. Formation of stable midzone bundles, referred to as the central spindle, requires the microtubule-bundler PRC1 as well as the centralspindlin and chromosomal passenger protein complexes (Pollard and O'Shaughnessy, 2019; Basant and Glotzer, 2018; Carmena et al., 2012). At late stages of cytokinesis, the contractile ring compacts midzone and astral microtubules, which fill the narrow cytoplasmic bridge connecting the sister cells, and transitions into a midbody ring once constriction stops. The compacted central spindle matures into a midbody central core (where PRC1 and centralspindlin localize) and the midbody flank (where the chromosomal passenger complex localizes), and together with the midbody ring make up the midbody (Hu et al., 2012; Green et al., 2012). The midbody orchestrates the final step of cytokinesis by recruiting the components of the abscission machinery, such as endosomal sorting complexes required for transport (ESCRTs) in cultured human cells (Yang et al., 2008; Elia et al., 2011; Schiel et al., 2012).

Inhibition of PRC1, the centralspindlin complex or the chromosomal passenger protein complex prevents central spindle formation and results in cytokinesis failure in a variety of cells (Jiang et al., 1998; Mollinari et al., 2002; Vernì et al., 2004). In the Caenorhabditis elegans early embryo, the importance of central spindle formation for cytokinesis has been called into question. With the notable exception of the EMS cell of the four-cell embryo, the PRC1 homolog SPD-1 has been reported to be dispensable for successful cytokinesis (Verbrugghe and White, 2004; Green et al., 2013; Lewellyn et al., 2010). Furthermore, it was recently shown that a functional midbody still forms in C. elegans zygotes lacking the CENP-F-like proteins HCP-1/2, which are required for central spindle formation (Hirsch et al., 2022). The early embryo may also use alternative modes for abscission: the midbody ring alone has been proposed to be able to scaffold the abscission machinery, and abscission appears to be able to occur independently of ESCRTs and midzone microtubules (Green et al., 2013; König et al., 2017). Given that SPD-1 has been shown to be essential for cytokinesis in the EMS cell (Verbrugghe and White, 2004), we investigated the role of SPD-1-mediated microtubule bundling in more detail using live fluorescence imaging.

We show that acute SPD-1 inactivation using the previously described temperature-sensitive mutant spd-1(oj5) (Verbrugghe and White, 2004), but not RNAi-mediated depletion of SPD-1, results in frequent failure to complete the last stages of furrow ingression in the one-cell embryo and in all cells of the four-cell embryo, and we confirm that the EMS cell is particularly sensitive to SPD-1 inhibition. Detailed characterization of the inhibition phenotype reveals that SPD-1-mediated microtubule bundling is required to maintain the tight back-to-back configuration of the cleavage furrow and that, in the absence of midzone microtubule bundles, anillin (ANI-1) becomes essential to maintain myosin in the contractile ring specifically during the later stages of cytokinesis. Thus, microtubule bundling by SPD-1 contributes to the success of cytokinesis in all cells of the early embryo. Our results suggest that the requirements for contractile ring constriction change as the furrow approaches the central spindle, and that continued constriction and completion of cytokinesis is ensured by signals emanating from the central spindle/midbody and by anillin, which together re-enforce myosin localization in the contractile ring.

To study the requirement of SPD-1 for cytokinesis in the C. elegans early embryo, we took advantage of the temperature-sensitive mutant spd-1(oj5), which permits rapid inactivation by upshifting the temperature from 16°C to 26°C (Verbrugghe and White, 2004). For comparison, we depleted SPD-1 by RNAi (Fig. 1A). After anaphase onset, SPD-1 localized to the bundled microtubules of the midzone and persisted at the midbody (Fig. 1B). Penetrant inhibition of SPD-1 was expected to result in two separated half spindles with no overlapping microtubules in the way of the ingressing cytokinetic furrow (Fig. 1B′).

When SPD-1 was depleted by RNAi at 20°C or 26°C in embryos expressing fluorescent myosin (NMY-2::GFP), a marker for the plasma membrane [mCherry::PH(PLC1δ1)] and a marker for chromosomes (mCherry::HIS-58), the majority of EMS cells completed cytokinesis, and the rate of contractile ring constriction was unaffected (Fig. 1C,D,F). In contrast, and in agreement with a previous study (Verbrugghe and White, 2004), 78% of spd-1(oj5) EMS cells failed cytokinesis when upshifted to 26°C during prophase (Fig. 1C,F). Cytokinesis typically failed after complete furrow ingression, which occurred at the same rate as in controls (Fig. 1E). Inactivation of spd-1(oj5) at the beginning of furrowing or at 50% of furrow ingression resulted in a similar phenotype, which reveals that SPD-1 activity is required during the last stages of cytokinesis (Fig. 1C). At 26°C spd-1(oj5) also caused cytokinesis failure in the zygote and the other cells of the four-cell embryo (ABa, ABp, P2), but the failure rate was not as high as in the EMS cell (41% in the zygote or P0, 53% in the ABa, 40% in the ABp, and 30% in the P2 cell; Fig. 1G). We conclude that SPD-1 is required for cytokinesis in the C. elegans early embryo and that the EMS cell is particularly sensitive to SPD-1 inactivation.

To characterize the impact of SPD-1 inhibition on cytokinesis in more detail, we focused on the EMS cell in embryos co-expressing mCherry::HIS-58 and mCherry::PH(PLC1δ1). In control cells, a tight cleavage furrow forms and sister cells are juxtaposed with back-to-back plasma membranes until the end of furrow ingression. Different temperatures changed the duration of cytokinesis but not the tight furrowing (Fig. S1; Fig. 2A,A′). In contrast, SPD-1 inactivation at 26°C or spd-1(RNAi) at 20°C resulted in broadening of the cleavage furrow tip and in the creation of an elongated intercellular bridge between sister cells at the last stages of constriction (Fig. 2A,A′). In spd-1(RNAi) cells, the intercellular bridge elongated and thinned, and the bridge then shortened until the sister cells became juxtaposed. At this point, sister cells resembled those in control embryos (Fig. 2A,C). In the majority of spd-1(oj5) EMS cells at 26°C, the intercellular bridge thinned but did not seal, as after bridge shortening the cleavage furrow regressed and cytokinesis failed (424+50 s after anaphase onset, n=10). The phenotype of spd-1(oj5) at the semi-restrictive temperature of 22°C was similar to that of spd-1(RNAi) (Figs 1C, 2A-C; Fig. S2), suggesting that spd-1(RNAi) results in partial inhibition of SPD-1. The formation of an elongated intercellular bridge between sister cells was also observed during cytokinesis of the ABa, ABp and P2 cell, the one-cell embryo (P0), and cells of older embryos (16-32-cell stage) (Fig. 2D). These results reveal that SPD-1 is required to maintain the ingressing cleavage furrow in a tight back-to-back configuration.

To understand why partial SPD-1 inhibition allows cytokinesis to complete while penetrant inhibition does not, we examined the state of midzone microtubules. Bundled midzone microtubules were abundant in control embryos, were substantially decreased at the semi-restrictive temperature, and were undetectable at the restrictive temperature (Fig. 3A). For spd-1(RNAi) and corresponding control EMS cells, imaging could be performed under compression, which improves image quality because the spindle is situated closer to the coverslip (Fig. 3B). As in spd-1(oj5) at the semi-restrictive temperature, residual microtubule bundles were detected at the midzone after spd-1(RNAi) (Fig. 3B). In agreement with this, residual Aurora B (AIR-2::GFP) and centralspindlin (CYK-4::GFP or CYK-4::mNeonGreen) signal was detected in the intercellular bridge in spd-1(oj5) at 22°C and after spd-1(RNAi). At the restrictive temperature, neither AIR-2::GFP nor CYK-4::GFP were detected in the midzone of spd-1(oj5) cells. We conclude that cytokinesis completion after spd-1(RNAi) or in spd-1(oj5) at 22°C correlates with the presence of a residual central spindle (Fig. 3A-C).

We next examined how SPD-1 inhibition affects contractile ring components. During cytokinesis in control EMS cells, the contractile ring folds back onto itself at the tip of the cleavage furrow (Movie 1). Myosin (NMY-2::GFP) and anillin (ANI-1::GFP) localized in the contractile ring throughout constriction and persisted at the midbody. A fraction of myosin and anillin was shed from the midbody 582±28 s after anaphase onset (n=12), but most of the signal persisted at the midbody even after the midbody was released 1046±162 s after anaphase onset (n=7; Fig. S3A-B′). In contrast, myosin and anillin became dispersed along the intercellular bridge and spread into the lateral sides of the furrow in spd-1(oj5) cells at 22°C and 26°C (Fig. 4A-C). When the bridge reached its maximum length, the continuous myosin and anillin signal became fragmented and decreased over time. Residual myosin and anillin concentrated at the center of the bridge in a midbody-like structure in spd-1(oj5) at 22°C, which persisted during sister cell juxtaposition (Fig. 4A; Movie 2). The presence of residual Aurora B and centralspindlin at the center of the bridge after spd-1(RNAi) and spd-1(oj5) at 22°C supports the idea that partial inhibition of SPD-1 allows for formation of a mini-midbody (Fig. 4D; Fig. S3C). In agreement with this, residual ESCRT-I (TSG-101::GFP) was observed in the mini-midbody after spd-1(RNAi). In spd-1(oj5) at 26°C, no mini-midbody was observed, and myosin and anillin completely disappeared from the bridge shortly before furrow regression (Fig. 4B; Movie 3; Fig. S3C). These results suggest that the residual microtubule bundles observed in spd-1(RNAi) and spd-1(oj5) at 22°C support the formation of a mini-midbody, which in turn allows bridge sealing and completion of cytokinesis.

Anillin is involved in coordinating the transition from contractile ring to midbody ring during late cytokinesis in Drosophila melanogaster S2 cells (Kechad et al., 2012). In the C. elegans early embryo, however, the anillin ANI-1 is not essential for cytokinesis (Maddox et al., 2007). C. elegans also expresses two additional anillin paralogs (ANI-2 and ANI-3), but these have not been implicated in embryogenesis (Maddox et al., 2005). We next investigated how co-inhibition of ANI-1 and SPD-1 impacts cytokinesis in EMS cells. ANI-1 depletion on its own resulted in symmetric furrow closure (Fig. S4A,B), a slower second half of contractile ring constriction, and an even slower final phase of constriction (Fig. S4C). Central spindle microtubules were unaffected (Fig. S4D) and cytokinesis completed in all cells (n=20).

Strikingly, ANI-1 depletion in spd-1(oj5) at 22°C severely aggravated cytokinesis failure in EMS cells, with 88% of furrows not completing ingression (Fig. 5A; Movie 4). The presence of residual microtubule bundles at the midzone suggested that ANI-1 depletion in spd-1(oj5) at 22°C did not further compromise central spindle formation beyond the defects observed in spd-1(oj5) (Fig. 5B). Although an intercellular bridge formed in the double inhibition, it did not elongate or thin out as much as the intercellular bridge in spd-1(oj5) cells (Fig. 5C), reaching a maximum length of 3.5±0.5 µm [versus 5.6±0.5 µm in spd-1(oj5) cells at 22°C; n=10 and n=33, respectively; P=0.0001 (two-tailed Mann–Whitney test)]. After reaching a maximum of 97% ingression, furrow ingression in the double inhibition stalled until regression ensued (Fig. 5D; Movie 4). Mini-midbodies were not observed at any point (Fig. 5C). At 26°C, all ani-1(RNAi);spd-1(oj5) EMS cells failed cytokinesis. Also, 87% already stopped furrowing at 70±6% ingression (n=20) and, after a period of pause, their furrows regressed (Fig. 6A,B; Movie 5). We conclude that depletion of ANI-1 in SPD-1-inhibited cells aggravates cytokinesis defects without aggravating defects in central spindle formation.

To understand why co-inhibition of ANI-1 and SPD-1 aggravates cytokinesis defects relative to the single inhibitions, we examined myosin localization. In contrast to single inhibitions, depletion of ANI-1 in spd-1(oj5) cells at 22°C or 26°C caused myosin to progressively disappear from the contractile ring until it was barely detectable immediately before furrow regression (Figs 5C,C′,E, 6C-D; Movie 5). Furthermore, acute inactivation of myosin during the second half of furrow ingression in otherwise normal EMS cells, using the temperature sensitive mutant nmy-2(ne3409), resulted in abrupt stalling of ingression followed by furrow regression (Fig. 7A,A′). These results suggest that ANI-1 depletion in spd-1(oj5) cells aggravates cytokinesis defects due to a failure to maintain myosin in the contractile ring.

We found that 70% ingression, which is when furrowing stops on average in ani-1(RNAi);spd-1(oj5) cells at 26°C, corresponds to the point at which the contractile ring approaches midzone microtubules in control embryos: the diameter of the contractile ring when it first approached the central spindle, measured in embryos co-expressing NMY-2::mKate2 and a microtubule marker (GFP::TBB-2), was 5.5±0.2 µm, which corresponds to 69±1% ingression (n=13). Similar values were obtained when measuring the diameter of the contractile ring in embryos co-expressing SPD-1::GFP and NMY-2:.mKate2 (5.5±0.5 µm corresponding to 70±3% ingression) (Fig. 7B). Analysis of instantaneous contractile ring constriction rate showed that 72% ingression also corresponds to the stage when furrow ingression abruptly starts to slow down in control embryos (Fig. 7C).

These observations suggest that the central spindle slows the advance of the contractile ring. Together with the finding that co-inhibition of SPD-1 and ANI-1 lead to ring regression at a point when the cleavage furrow would normally encounter the spindle midzone, our results suggest that anillin in the contractile ring and the central spindle (through SPD-1 or other central spindle components) act jointly to induce changes in the contractile ring at this time point. These changes involve the maintenance of myosin in the contractile ring, thereby ensuring the continuation and completion of furrow ingression (Fig. 7D).

Our analysis of the temperature-sensitive spd-1(oj5) allele using a device for fast and reliable temperature control clarifies that penetrant inactivation of SPD-1, i.e. spd-1(oj5) at 26°C, results in frequent cytokinesis failure in all cells of the C. elegans early embryo. The consistent cytokinesis success that we observed after RNAi-mediated depletion of SPD-1 correlated with residual microtubule bundles traversing the furrow region, which indicates that RNAi is not sufficiently penetrant to reveal the requirement of SPD-1 for cytokinesis. This idea is further supported by the observations that the phenotype of spd-1(RNAi) resembled that of spd-1(oj5) at the semi-restrictive temperature (22°C). The EMS cell of the four-cell embryo was the most sensitive to SPD-1 inhibition, which is in agreement with the study that initially characterized the spd-1(oj5) mutant (Verbrugghe and White, 2004). The reason why EMS is more sensitive to SPD-1 inhibition is not clear and contrasts with the finding that EMS is more resistant to cytokinesis failure than ABa and ABp following formin (CYK-1) inactivation or Latrunculin A treatment (Davies et al., 2018). Protection against cytokinesis failure after actin perturbation in the EMS cell was found to be mediated by extrinsic regulation through Wnt/Src signaling and direct contact with the neighboring P2 cell. Our observations indicate that Wnt signaling originating from the P2 cell is unaffected after SPD-1 inactivation because spindle positioning, division axis orientation and the extent of EMS-P2 contact appeared to be normal (Walston and Hardin, 2006; Zhang et al., 2008; Sugioka and Bowerman, 2018). In addition, we observed the same sensitivity to SPD-1 inactivation when inactivation was acutely inflicted during furrow ingression, past the time when Wnt signaling is thought to be crucial. It remains, however, possible that the Wnt signaling pathway does have a function during late cytokinesis and that this function is perturbed when the central spindle and the midbody cannot form after acute SPD-1 inactivation. In fact, Wnt and Src effectors have been localized to the midbody in tissue culture cells (Fumoto et al., 2012; Kikuchi et al., 2010; Kaplan et al., 2004; Kasahara et al., 2007; Yu et al., 2021). Alternatively, the increased sensitivity of EMS to SPD-1 inactivation may be related to intrinsic cell-lineage-specific characteristics. In this latter case EMS descendants should also display increased sensitivity to SPD-1 inactivation, which remains to be determined.

Embryonic cytokinesis typically occurs with a back-to-back configuration of the cleavage furrow, which results in tightly juxtaposed sister cells. Intercellular bridges that separate sister cells during the last stages of cleavage furrow ingression are therefore minimal in normal embryos. Our findings reveal that full or partial inhibition of SPD-1 results in long intercellular bridges that start to form at ∼90% furrow ingression, which suggests that interactions between the tip of the furrow and the compacting spindle midzone are required to maintain a tight furrow during the last stages of cytokinesis. As elongated bridges form in several cells with distinct shapes, identities and number of cell-cell interactions, it is unlikely that they arise from pulling forces acting on the EMS cell poles. An attractive possibility is that SPD-1 inhibition compromises the adhesion between the two sides of the furrow. How midzone microtubules, SPD-1 or other midzone components would contribute will require further investigation. Of note, we found that partial inhibition of SPD-1 leads to elongation of the intercellular bridge, but the increased distance between the sister cells is transient as these eventually juxtapose. This suggests that any defects in establishing cell-cell adhesions is only temporary in this condition.

In the SPD-1-inhibited EMS, we found that contractile ring components are tightly localized to the tip of the ingressing furrow but become subsequently broadly distributed along the entire length of the intercellular bridge and beyond. If SPD-1 is penetrantly inhibited, bridge thinning is followed by myosin/anillin signal fragmentation until no signal is detectable, which precedes furrow regression. When SPD-1 is partially inhibited, which allows cytokinesis to complete, some myosin and anillin remain enriched at a mini-midbody within the intercellular bridge. Thus, our data suggest that the presence of a residual mini-midbody is crucial to ensure complete sealing during abscission, which is a prerequisite for successful completion of cytokinesis. However, it has previously been reported that abscission does not require central spindle microtubules and that the midbody ring is sufficient to scaffold the abscission machinery in the C. elegans one-cell embryo (P0) (Green et al., 2013; König et al., 2017). To test the requirement for microtubules, both Green et al. and Konig et al. used spd-1(RNAi), which our results suggest may not completely prevent microtubule bundle formation at the midzone. Whether the contrasting conclusions regarding the requirement of a midbody for abscission reflect inherent differences between P0 and EMS will require further investigation. We show that the formation of an elongated intercellular bridge after SPD-1 inhibition is not EMS-specific but mini-midbody formation could only be examined in EMS, where the position and orientation of the bridge facilitated this type of analysis. Electron tomographic reconstruction of late cytokinesis in the one- and four-cell embryo after full SPD-1 inactivation using spd-1(oj5) would be required to gain further insight into this issue.

We previously showed that myosin motor activity is required for cytokinesis in the C. elegans one-cell embryo (Osório et al., 2019). Using fast inactivation of a myosin temperature sensitive mutant, we now show that myosin is required for continuous furrow ingression: the moment myosin is inactivated, furrow ingression stalls and is followed by furrow regression. It is known that RhoA activation is required to recruit active myosin to the cell equator when furrowing initiates (Pollard and O'Shaughnessy, 2019). Our data reveal that this initial myosin loading step is not sufficient to complete cytokinesis. ANI-1 and SPD-1 act redundantly in the EMS cell to maintain myosin in the contractile ring during the second half of furrow ingression. Although cytokinesis completes when SPD-1 is partially inactivated or when ANI-1 is depleted by RNAi, cytokinesis fails when the two perturbations are combined: furrow ingression advances to final stages (97% ingression) but myosin progressively disappears from the intercellular bridge, which loses the ability to keep thinning and elongating, and eventually the furrow regresses. When ANI-1 is depleted in the background of penetrant SPD-1 inactivation, myosin disappears from the contractile ring earlier (70% ingression). Our data therefore supports the idea that myosin-mediated contractility is required until the end of cytokinesis, and that anillin in the contractile ring and SPD-1-bundled midzone microtubules jointly ensure that myosin is maintained/activated at the tip of the furrow during late cytokinesis. As myosin levels at the cleavage furrow have been shown to be a good proxy for RhoA activation (Zhang and Glotzer, 2015), it is likely that myosin's disappearance from the furrow region reflects RhoA inactivation. Indeed, both ANI-1 and midzone microtubules participate in RhoA activation. Anillin stabilizes active RhoA at the membrane, increasing its residency time to engage with effectors (Budnar et al., 2019) and is required for stable RhoA signal (Reyes et al., 2014). The centralspindlin component CYK-4, which localizes to midzone microtubule bundles, is required for ECT-2-mediated RhoA activation at the cleavage furrow, and its inactivation leads to a substantial decrease in active RhoA and myosin levels in the contractile ring and to furrow regression (Burkard et al., 2009; Wolfe et al., 2009; Gómez-Cavazos et al., 2020; Basant et al., 2015; Lewellyn et al., 2011; Canman et al., 2008; Tse et al., 2012; Zhang and Glotzer, 2015; Loria et al., 2012). As both anillin and SPD-1 interact with CYK-4 (Lee et al., 2017; Ban et al., 2004; Fu et al., 2009; D'Avino et al., 2008; Gregory et al., 2008), it is conceivable that co-inhibition of ANI-1 and SPD-1 directly perturbs CYK-4 activity (i.e. independently of the role of SPD-1 in central spindle formation), preventing it from activating RhoA during the second half of constriction. Unfortunately, it was not feasible to monitor RhoA activity using the biosensor that is available for C. elegans, because this sensor consists of a C-terminal ANI-1 fragment (Tse et al., 2012) and cannot be used to address RhoA behavior when the effects of anillin depletion are being investigated. Another possible explanation for the disappearance of myosin from the furrow region is that the contractile ring loses attachment to the membrane, as both ANI-1 and CYK-4 contain elements that associate directly with the membrane at the cleavage furrow (Sun et al., 2015; Lekomtsev et al., 2012).

It has been proposed that the contractile ring transitions into a midbody ring through molecular changes that occur at a late stage, when the contractile ring is almost fully closed (∼1 µm in Drosophila melanogaster S2 cells; Kechad et al., 2012). Anillin is thought to be required to template midbody ring assembly and to ensure stable anchoring of the midbody ring to the plasma membrane (Kechad et al., 2012). Our results raise the possibility that anillin-dependent molecular changes already occur when the contractile ring comes into contact with the spindle midzone. Indeed, the constriction rate decreases at 72% ingression in control EMS cells when the furrow encounters the midzone, and ANI-1 depletion considerably prolongs the last phase of furrow ingression, even when midzone microtubules are reduced and therefore represent less of an obstacle (Carvalho et al., 2009). The fact that depletion of ANI-1 and complete removal of midzone microtubule bundling stalls constriction at the point when the constricting ring would normally encounter the spindle midzone indicates that contact between the two structures ensures continuation of ingression. As the spindle midzone has different dimensions in tissue culture cells versus embryonic cells, the timing of contractile ring maturation during furrow ingression may vary between cell types.

Strains used in this study are listed in Table S1. Strains carrying the temperature sensitive alleles spd-1(oj5) or nmy-2(ne3409) were maintained at 16°C and the others at 16°C or 20°C on nematode growth medium (NGM) plates seeded with OP50 Escherichia coli.

RNAi was performed by feeding hermaphrodites with HT115 E. coli bacteria expressing double-stranded RNA (dsRNA) of interest from the L4440 vector (plasmid #1654, Addgene). To deplete SPD-1 by feeding RNAi, a 1590 bp region of the spd-1 locus was amplified from N2 genomic DNA with the primers 5′-CCCGGATCCATGTCCCGAAGGCACAGC-3′ and 5′-CCCGGATCCTCACAAAAACTGATTTCG, digested with BamHI restriction enzyme and cloned into L4440 in the BglII restriction enzyme site. The final plasmid was sequenced and transformed into HT115 E. coli. These bacteria were used to prepare RNAi feeding plates as previously reported (Silva et al., 2016). L4 animals were placed in the RNAi plates and incubated at 20°C for 44-48 h or at 26°C for 32 h before dissection for imaging. To deplete ANI-1, L4440 vectors carrying part of the sequence of ani-1 were obtained from the Ahringer library (Kamath et al., 2003; distributed by Source BioScience) and sequenced to confirm the gene target. L4 animals were fed with bacteria expressing double-stranded RNA against ani-1 and incubated at 16°C for 61 h-64 h before being dissected in cold M9 medium.

One-cell, four-cell or ∼16-cell C. elegans embryos were dissected from adult hermaphrodites and filmed. SPD-1 RNAi-depleted embryos were either imaged under compression on 2% agarose pads overlaid with a coverslip in a room acclimatized to 20°C, or under no compression in a drop of M9 medium (86 mM NaCl, 42 mM Na₂HPO₄, 22 mM KH₂PO₄ and 1 mM MgSO₄.7H₂O) placed on the CherryTemp chip set to 16°C, 22°C or 26°C. spd-1(oj5) and nmy2(ne3409) animals were filmed under no compression in a drop of cold M9 medium placed on the CherryTemp chip set to 16°C, 22°C or 26°C. The temperature in the CherryTemp chip was controlled by dedicated software (CherryBiotech). In spd-1(oj5) EMS cells, temperature was upshifted when chromosome condensation started, when the equatorial cortex started to deform at the beginning of furrow ingression (shallow deformation), or half way through cleavage furrow ingression, as indicated in the figures. In ABa, ABp and P2 cells, temperature was upshifted at the time of contractile ring assembly in the ABa cell; in P0 cells, temperature was upshifted at nuclear envelope breakdown. Assessment of cytokinesis failure or success was carried out in the strains GCP380 and GCP691, in movies that covered the entire process of furrowing until the following cell division, when both sister cells entered anaphase (Figs 1C,G, 5A, 6A).

All images were acquired on a spinning disk confocal system (Andor Revolution XD Confocal System; Andor Technology) with a confocal scanner unit (CSU-X1; Yokogawa Electric Corporation) mounted on an inverted microscope (Ti-E, Nikon) equipped with a 60× oil-immersion Plan-Apochromat objective (N.A. 1.4) and solid-state lasers of 488 nm (50 mW) and 561 nm (50 mW). An electron multiplication back-thinned charge-coupled device camera (iXon Ultra 897; Andor Technology) with 1×1 binning was used. Acquisition parameters, shutters and focus were controlled by Andor iQ3 software. Images were acquired in sets of ten z-planes 0.5- or 1-μm apart, every 10, 20 or 30 s.

All measurements and image processing were carried out using Fiji (ImageJ; Schindelin et al., 2012). Z-stacks were projected using the maximum intensity projection tool. Graph plotting and statistical analyses were performed with Prism 9.5.0 (GraphPad Software). All error bars represent the 95% confidence interval of the mean. Statistical significance was determined using a two-tailed Mann–Whitney test.

The contractile ring diameter was determined in EMS cells expressing NMY-2::GFP and mCherry::PH (Figs 1, 5D, 6B; Fig. S4C) or GFP::PH (Fig. 7A) by manually tracing a straight line between the two tips of the cleavage furrow on the z-plane, where this was the widest for each time point, and plotting against time after anaphase onset (first point when two masses of segregated chromatids were observed immediately after metaphase, as judged by chromosome labeling using the histone marker mCherry::HIS-58 or negative myosin labeling, which is cytoplasmic and absent from chromatin). Data from multiple rings were temporally aligned and averaged. The values were normalized to the diameter of the embryo measured at the cell equator at anaphase onset. Percent ingression was considered to be 0% before furrow ingression initiated and 100% when it completed.

For the graph of instantaneous constriction rate in Fig. 7C, the rate of ring constriction was calculated for pairs of consecutive time points by dividing the difference in diameter by the time interval. Individual rate measurements from all imaged embryos were pooled and the mean rate for the data points falling in overlapping 2-µm intervals was plotted against the contractile ring diameter at the center of each interval. Segmental linear regression (GraphPad) was used to determine the point of abrupt deceleration, which was considered to be the intersection of the line segment during which the constriction rate is decreasing only slightly and the line segment during which the constriction rate starts decreasing significantly.

In Fig. 7B, the point at which the contractile ring approached the central spindle was determined in EMS cells expressing NMY-2::mKate2 and GFP::TBB-2 or NMY-2::mKate2 and SPD-1::GFP, by manually tracing a straight line between the two tips of the cleavage furrow (as judged by NMY-2::mKate2 signal) on the z-plane where this was the widest, at the time point when the signal of NMY-2 and TBB-2 or SPD-1 first overlapped. The midzone length was determined by manually tracing a straight line between the two edges of the central spindle (as judged by TBB-2 or SPD-1 signal) at the time point just before it started being compacted by the advance of the cleavage furrow. In Fig. S4B, a line of 1.8 µm width was drawn over the two sides of the cleavage furrow at a point of 70% ingression in a control and an ani-1(RNAi) EMS cell expressing NMY-2::GFP, and the mean GFP fluorescence along the line was quantified.

In Fig. 2C, the period of bridge elongation corresponds to the interval of time between the point at which the intercellular bridge started to form (frame in which the furrow tip started to broaden) and the time at which it reached its maximum length (as judged by mCherry::PH signal); the period of bridge shortening corresponds to the interval of time between the point at which the intercellular bridge was the longest to the point at which the two ends of the bridge joined together. The point of sister cell juxtaposition corresponded to the time point at which the two sister cells became completely juxtaposed, as in control embryos. The end of furrowing was considered to be the point at which the distance between the two sides of the furrow was minimal. The point of furrow regression was considered to be the time frame when the two sides of the furrow separated. All reference points were determined in a minimum of 10 examples and the values shown correspond to the mean. In Fig. S3B, midbody shedding corresponded to the point at which some signal of anillin or myosin was released from the midbody, and midbody release was the time when the entire midbody separated from the back-to-back plasma membranes separating the sister cells.

We thank Julien Dumont, Julie Canman, Karen Oegema, Thomas Müller-Reichert, Michael Glotzer and Bob Goldstein for providing strains. Some strains were provided by the Caenorhabditis Genetics Center (CGC), which is funded by National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440).