Plastin and spectrin cooperate to stabilize the actomyosin cortex during cytokinesis.

Cytokinesis, the process that partitions the mother cell into two daughter cells, requires the assembly and constriction of an equatorial actomyosin network. Different types of non-motor F-actin crosslinkers localize to the network, but their functional contribution remains poorly understood. Here, we describe a synergy between the small rigid crosslinker plastin and the large flexible crosslinker spectrin in the C. elegans one-cell embryo. In contrast to single inhibitions, co-inhibition of plastin and the βH-spectrin (SMA-1) results in cytokinesis failure due to progressive disorganization and eventual collapse of the equatorial actomyosin network. Cortical localization dynamics of non-muscle myosin II in co-inhibited embryos mimic those observed after drug-induced F-actin depolymerization, suggesting that the combined action of plastin and spectrin stabilizes F-actin in the contractile ring. An in silico model predicts that spectrin is more efficient than plastin at stabilizing the ring and that ring formation is relatively insensitive to βH-spectrin length, which is confirmed in vivo with a sma-1 mutant that lacks 11 of its 29 spectrin repeats. Our findings provide the first evidence that spectrin contributes to cytokinesis and highlight the importance of crosslinker interplay for actomyosin network integrity.

Introduction

The actin cytoskeleton is necessary for a wide range of cellular processes that require force generation and cell-shape changes, including cell division, cell motility, and tissue morphogenesis. Generation and propagation of contractile force requires myosin II motors and non-motor crosslinkers (crosslinkers hereafter). Bipolar myosin II filaments slide actin filaments past one another at the expense of ATP, while crosslinkers maintain F-actin network connectivity during filament sliding.

Depending on their size, structural flexibility, and concentration, crosslinkers bridge F-actin into networks of different complexities and mechanical properties. Smaller crosslinkers, such as α-actinin and plastin, tend to form tight parallel actin bundles, while larger crosslinkers, such as filamin and spectrins, organize F-actin into loosely crosslinked meshworks.3, 4, 5, 6 Crosslinkers may additionally differ in their subcellular localization and response to tension and may have other functions apart from F-actin crosslinking. How F-actin networks organize and behave in the presence of multiple crosslinkers remains poorly explored.

During cytokinesis, animal cells assemble a contractile ring at the cell equator beneath the plasma membrane, and subsequent ring constriction ultimately leads to separation of the two daughter cells. The contractile ring consists of formin-nucleated non-branched F-actin that aligns circumferentially in a tight band. Besides non-muscle myosin II (hereafter myosin), other ring components include crosslinkers such as α-actinin, fimbrin/plastin (hereafter plastin), anillin, and septins. While myosin has long been known to be essential for cytokinesis, the role of crosslinkers remains unclear. When depleted individually, only cortexillins in D. discoideum and anillin in D. melanogaster and mammalian cultured cells have been shown to cause significant cytokinesis failure.8, 9, 10 However, cortexillins are not well conserved between species, and anillin is a multifunctional protein that interacts with several contractile ring components, so its other functions also contribute to cytokinesis. The only known example of functional redundancy among crosslinkers during cytokinesis is between fission yeast α-actinin and fimbrin.

In the present study, we describe a synergy between plastin and spectrin, two crosslinkers that are highly conserved through evolution. Plastin and spectrin belong to the calponin homology (CH) domain superfamily of crosslinkers (like filamin, α-actinin, and dystrophin), whose actin-binding domains consist of a double calponin-like sequence.

Plastin is a small globular protein composed of two EF hands and two adjacent actin-binding domains (ABDs). Plastin binds branched and non-branched F-actin and assembles F-actin into tightly packed bundles in parallel or antiparallel orientation in vitro.,, Plastin has been described to localize in cytokinetic rings, neurons, intestinal microvilli, hair cell stereocilia, and fibroblast filopodia, and to contribute to cytokinesis, invasion by pathogenic bacteria, endocytosis, and epidermal morphogenesis.13, 14, 15, 16

Spectrins are large and flexible crosslinkers that have not been previously implicated in cytokinesis. The functional form of spectrin consists of two antiparallel heterodimers of β-spectrin and α-spectrin. The resulting tetramers are elongated rods with ABDs at either end. Spectrins also contain a pleckstrin homology (PH) domain and interact with plasma membrane-bound proteins, thus linking the actomyosin cytoskeleton to the plasma membrane. There are two types of β-spectrin, referred to as conventional β-spectrin and heavy β-spectrin (βH), which differ in the number of spectrin repeats and subcellular localization., Tetramers of conventional β-spectrin protect the cortex from mechanical stresses in erythrocytes and axons, where they form regular polygonal or longitudinal F-actin networks, respectively., They are also required for ciliogenesis and the ability to sense and respond to external mechanical stimuli in C. elegans., βH-spectrin was shown to have mechanosensitive properties and to accumulate in regions of higher tension in D. melanogaster muscle cells. Furthermore, βH-spectrin is involved in embryonic development and tissue morphogenesis in D. melanogaster and C. elegans.,

C. elegans contains one ortholog each for plastin (plst-1), α-spectrin (spc-1), conventional β-spectrin (unc-70), and βH-spectrin (sma-1).,,, This contrasts with vertebrates, which express three plastins, two α-spectrins, four β-spectrins, and one βH-spectrin in a tissue-specific manner.,, The fact that these and other crosslinkers are represented by a single gene in C. elegans facilitates functional studies aimed at uncovering synergies among crosslinkers. Here, we find that, in contrast to single inhibitions, co-inhibition of plastin and βH-spectrin results in penetrant cytokinesis failure in the C. elegans one-cell embryo. Combining engineered βH-spectrin mutants with analysis of myosin and F-actin dynamics, we demonstrate that plastin and βH/α-spectrin jointly organize and stabilize cortical F-actin at the cell equator, which is critical for contractile ring assembly. Our results emphasize the importance of understanding how different types of crosslinkers cooperate to organize actomyosin networks in vivo.

Results

Multiple crosslinkers localize to the equatorial cell cortex and the tip of the ingressing cleavage furrow during cytokinesis

To assess localization dynamics of crosslinkers in one-cell C. elegans embryos undergoing their first cytokinesis, we used fluorescent versions of plastin (PLST-1::GFP), anillin (GFP::ANI-1), and septin (UNC-59::GFP) (Table S1). Cortical imaging allowed visualization of contractile ring assembly at the cell equator, and central plane imaging during furrow ingression offered a side view of the constricting contractile ring (Figure 1A). PLST-1::GFP formed linear bundles at the equatorial cortex and a meshwork with distinct puncta in the surrounding cortex. This localization pattern was similar to that of F-actin (LifeAct::GFP), as reported previously. By contrast, GFP::ANI-1 and UNC-59::GFP localized to cortical patches that were enriched in the equatorial region, similar to myosin (NMY-2::GFP). UNC-59::GFP was more prominent on the anterior cortex than GFP::ANI-1 and NMY-2::GFP. All probes were strongly enriched at the tip of the ingressing furrow during contractile ring constriction. These results illustrate that cytokinesis occurs in the presence of multiple crosslinkers, consistent with the idea that different types of crosslinkers jointly organize the actomyosin network.

Figure 1

Cytokinesis completes when F-actin crosslinkers are depleted individually

(A) Images of the cortex (left) and central plane (right) in C. elegans one-cell embryos expressing fluorescent markers of actin, plastin, anillin, septins, and myosin at initial equatorial deformation and 50% furrow ingression, respectively. Scale bars, 10 μm.

(B) Schematic illustrating cytokinesis completion or failure during the first embryonic division.

(C) Percentage of one-cell embryos that complete/fail cytokinesis after RNAi-mediated depletion of crosslinkers. N indicates the number of embryos analyzed.

(D) Total cytokinesis time (mean ± 95% CI), defined as the interval between anaphase onset and the contractile ring reaching a diameter of 5 μm (see STAR Methods for details), in embryos depleted of crosslinkers. N indicates the number of embryos analyzed. Statistical significance was determined using one-way ANOVA followed by Dunnett’s multiple comparison test: ∗∗∗∗p ≤ 0.0001.

See also Figure S1 and Tables S1 and S2.

A survey of nine conserved crosslinkers identifies plastin/PLST-1 as the only crosslinker whose depletion slows cytokinesis

To analyze crosslinker function during cytokinesis, we first identified C. elegans crosslinker orthologs that are highly conserved across species and have been demonstrated to crosslink F-actin in vitro:,27, 28, 29, 30, 31, 32 α-actinin (atn-1), plastin (plst-1), filamin (fln-1), β-spectrin (unc-70), βH-spectrin (sma-1), septins (unc-59 and unc-61), anillin (ani-1), and coronin (cor-1). Cytokinesis completed successfully and with normal timing when ATN-1, FLN-1, SPC-1, UNC-59, UNC-61, ANI-1, and COR-1 were individually depleted by RNAi (Figures 1B–1D; note that spc-1(RNAi) also inhibits UNC-70 and SMA-1, the other spectrin complex subunits). Depletion of each crosslinker was validated as described in Figure S1. Depletion of PLST-1 slowed cytokinesis (278 ± 54 s versus 217 ± 29 s in controls), and 6% of plst-1(RNAi) embryos failed to complete furrowing (Figures 1C and 1D). Ring constriction rate (the slope of the linear region between ∼75% and 10% furrow ingression) was not affected after plst-1(RNAi) (0.19 ± 0.04 μm/s versus 0.19 ± 0.02 μm/s in controls), confirming that PLST-1 acts primarily at the early stages of cytokinesis (ring assembly and furrow initiation) (Figure 2D)., We conclude that among the nine conserved crosslinkers examined, only PLST-1 depletion impacts cytokinesis in the C. elegans one-cell embryo.

Figure 2

Inhibiting PLST-1 together with SPC-1 or SMA-1, but not UNC-70, results in cytokinesis failure

(A) Percentage of cytokinesis completion/failure in one-cell embryos after crosslinker co-inhibition.

(B) Schematic illustrating the two types of spectrin tetramers: α-spectrin (SPC-1) pairs with either conventional β-spectrin (UNC-70) or βH-spectrin (SMA-1). CH, calponin homology domain; SH3, Src homology domain; PH, pleckstrin homology domain; SR, spectrin repeat.

(C) Representative kymographs of the equatorial region in one-cell embryos. The first frame corresponds to anaphase onset. Scale bar, 10 μm.

(D) Individual curves of contractile ring diameter over time after anaphase onset. Gray zone indicates the region corresponding to 75%–10% furrow ingression, which was used to calculate ring constriction rate. N indicates the number of embryos analyzed.

See also Video S1 and Tables S1 and S2.

Co-inhibition of PLST-1 and βH-spectrin/SMA-1 results in cytokinesis failure

To address potential cooperation between PLST-1 and other crosslinkers during cytokinesis, we combined the plst-1(prt89) mutant, which carries a premature stop in its open reading frame, with RNAi-mediated depletion of ANI-1, FLN-1, UNC-59, UNC-61, COR-1, or SPC-1 (Figure 2A). Additionally, we combined plst-1(prt89) with an atn-1 mutant that lacks the actin-binding domain (atn-1ΔABD). We found that spc-1(RNAi) in plst-1(prt89) embryos was the only condition that significantly increased cytokinesis failure (86% failure in plst-1(prt89);spc-1(RNAi) versus 15% in plst-1(prt89); Figures 2A and 2C; Video S1). Since the α-spectrin SPC-1 can form tetramers with either the conventional β-spectrin UNC-70 or the βH-spectrin SMA-1 (Figure 2B), we next performed unc-70(RNAi) or sma-1(RNAi) in plst-1(prt89) embryos. Similar to plst-1(prt89);spc-1(RNAi), plst-1(prt89);sma-1(RNAi) caused cytokinesis failure in 87% of embryos, which were unable to form a cytokinetic furrow (Figures 2A–2D; Video S1). Consistent with the RNAi results, we were unable to generate viable double mutants carrying plst-1(prt89) and the null mutant sma-1(ru18). Homozygous sma-1(ru18);plst-1(prt89) animals that reached adulthood laid on average 2 embryos. 95% of the laid progeny died during embryogenesis and 5% arrested at the L1 stage. In contrast to sma-1(RNAi);plst-1(prt89), unc-70(RNAi);plst-1(prt89) had no further adverse effects on cytokinesis beyond those observed in plst-1(prt89) alone (Figures 2A and 2C), which is in agreement with the minimal expression of UNC-70 in the early embryo., We conclude that SPC-1/SMA-1 α/βH-spectrin tetramers become essential for cytokinesis in the absence of PLST-1.

Video S1. Co-inhibition of PLST-1 and SMA-1 results in cytokinesis failure, related to Figure 2

Time-lapse series of the central plane in dividing one-cell embryos expressing NMY-2::GFP. First time point corresponds to anaphase onset (0 s) and frames are 10 s apart. Frame rate is 6 frames per second and time stamp is in seconds. Scale bar, 10 μm.

PLST-1 and SMA-1 have distinct localization profiles at the cell cortex

To assess SMA-1 localization during cytokinesis, we generated a fluorescent version using the split GFP(spGFP) approach, in which the 11-stranded β-barrel structure of GFP is reconstituted from two separately expressed fragments: GFP1-10 (β strands 1–10) and GFP11 (β strand 11). We inserted 6 tandem copies of GFP11 downstream of the sma-1 open reading frame by genome editing and crossed animals expressing endogenous SMA-1::6xGFP11 with animals expressing transgenic GFP1-10 under the control of a germline promoter. Imaging of one-cell embryos revealed that the SMA-1::spGFP signal was significantly dimmer than that of PLST-1::GFP and formed small cortical dots that became enriched at the cell equator simultaneously with PLST-1::GFP (Figures 3A and S7A). While PLST-1::GFP localized to bundles similar to LifeAct::RFP (Figure 3A and signal overlap in Figure 3B), SMA-1::spGFP dots decorated those bundles but did not overlap with PLST-1::mCherry nor NMY-2::mKate2 (Figure 3B). The expression level and localization pattern of endogenous SPC-1::GFP was comparable to that of SMA-1::spGFP, and immunofluorescence with affinity-purified anti-SMA-1 antibody confirmed SMA-1’s dotty distribution and equatorial accumulation during cytokinesis (Figures 3C and 3D). We conclude that SMA-1 is expressed at low levels in the one-cell embryo, that PLST-1 and SMA-1 have distinct localizations within the actomyosin network, and that, consistent with a previous report in mouse embryonic fibroblasts in interphase, spectrins and actomyosin distinctly organize in space.

Figure 3

PLST-1 and SMA-1 show distinct profiles of localization on the cell cortex

(A) Images of the cortex in one-cell embryos expressing SMA-1::spGFP (splitGFP) or PLST-1::GFP. Numbers indicate the time in seconds after anaphase onset. Scale bars, 10 μm.

(B) Images of the cortex in one-cell embryos co-expressing either the two fluorescent crosslinkers or one of the fluorescent crosslinkers and fluorescent LifeAct or NMY-2. Scale bar, 10 μm. Blow-ups of the equatorial region are shown on the right. Scale bars, 2 μm.

(C) Image of the cortex in a one-cell embryo expressing SPC-1::GFP. Scale bar, 10 μm.

(D) Images of the cortex and central plane of a fixed one-cell embryo stained with anti-SMA-1 and anti-α-tubulin antibodies. Scale bars, 10 μm.

See also Figure S7A and Tables S1 and S2.

Synergy between SMA-1 and PLST-1 does not require SMA-1’s SH3 and PH domains

PLST-1 is a small protein whose only known function is F-actin crosslinking, whereas the much larger SMA-1 contains three functionally distinct domains: an N-terminal ABD, an Src homology-3 (SH3) domain, thought to be involved in signal transduction that is inserted between spectrin repeats in the N-terminal half, and a C-terminal PH domain, which mediates interactions with the plasma membrane. The remainder of the protein consists of spectrin repeats, which confer an elongated shape and flexibility, promote spectrin dimerization and tetramerization, and may mediate additional protein-protein interactions (Figures 2B and 4A). To define the contribution of SMA-1 domains to the observed synergy between SMA-1 and PLST-1, we used genome editing to delete the regions encoding the SH3 domain (ΔSH3) and the PH domain (ΔPH) from the sma-1 locus (Figure 4A). sma-1ΔSH3 and sma-1ΔPH animals were viable, grew to normal adult size, and were fully fertile. This contrasts with sma-1(ru18) animals, which are small and have a reduced brood size. Immunoblots with anti-SMA-1 antibodies showed that the expression levels of SMA-1ΔSH3 and SMA-1ΔPH were similar to wild-type SMA-1, while no detectable SMA-1 was expressed in the sma-1(ru18) mutant (Figure 4B).

Figure 4

Synergy between SMA-1 and PLST-1 does not require SMA-1-mediated F-actin network anchoring to the plasma membrane or SMA-1 SH3 domain-mediated signaling

(A) Schematic of wild-type SMA-1, SMA-1ΔSH3, and SMA-1ΔPH. CH, calponin homology domain; SH3, Src homology domain; PH, pleckstrin homology domain; SR, spectrin repeat.

(B) Immunoblots of adult lysate with an affinity-purified antibody raised against the actin-binding domain (ABD) of SMA-1. α-tubulin serves as the loading control.

(C and D) Percentage of cytokinesis completion/failure in one-cell embryos. N indicates the number of embryos analyzed.

(E) Images of the cortex in embryos expressing NMY-2::GFP. Numbers indicate the time in seconds after anaphase onset. Scale bars, 10 μm.

See also Video S2 and Tables S1 and S2.

We next crossed the sma-1 domain mutants with the plst-1(prt89) mutant and obtained homozygous viable double mutants. Imaging of one-cell embryos expressing NMY-2::GFP showed that the percentage of cytokinesis failure remained low in plst-1(prt89);sma-1ΔSH3 and plst-1(prt89);sma-1ΔPH (19% and 0%, respectively, versus 8% in plst-1(prt89); Figures 4C and 4D). Furthermore, cortical LifeAct::GFP and NMY-2::mCherry localization patterns in the single and double mutants were similar to those observed in controls and the plst-1(prt89) mutant, respectively (Figure 4E; Video S2). SMA-1 establishes interactions with the plasma membrane directly through its PH domain and indirectly through other membrane-localized proteins, namely, ankyrin, ezrin-radixin-moesin, and protein 4.1. We therefore co-depleted ankyrin (UNC-44), ezrin-radixin-moesin (ERM-1), and protein 4.1 (FRM-1) by RNAi in plst-1(prt89);sma-1ΔPH embryos. The efficiency of the triple RNAi was validated by confirming: signal disappearance in a strain that expresses ERM-1::GFP, delayed hatching due to FRM-1 depletion, and decreased brood size due to UNC-44 depletion.35, 36, 37 Cytokinesis successfully completed in 96% of triple-depleted plst-1(prt89);sma-1ΔPH embryos (Figure 4D). We conclude that neither the SH3 nor the PH domain of SMA-1 is required for the synergistic action between SMA-1 and PLST-1 during cytokinesis, which completes even when 3 additional membrane anchoring proteins are co-depleted in sma-1ΔPH embryos. Although we cannot rule out the possibility that the SH3 domain may play a role in linking SMA-1 to the plasma membrane, these results suggest that cytokinesis failure in the absence of PLST-1 and SMA-1 is unlikely due to problems in anchoring the F-actin cortex to the plasma membrane.

Video S2. SMA-1 mutant versions lacking the PH domain, the SH3 domain, or 11 of its 29 spectrin repeats do not aggravate the cytokinesis phenotype in plst-1(prt89) embryos, related to Figures 4 and 7

Time-lapse series of the cell cortex in dividing one-cell embryos co-expressing LifeAct::GFP and NMY-2::mCherry. Cortical images are maximum intensity projections of seven z sections 0.5 μm apart acquired every 5 s from anaphase onset (0 s). Frame rate is 5 frames per second and time stamp is in seconds. Scale bar, 10 μm.

Co-inhibition of PLST-1 and SMA-1 results in the collapse of equatorial F-actin bundles

We next asked whether PLST-1 and SMA-1 synergize through their F-actin crosslinking activities. To generate a SMA-1 mutant defective in actin-binding, we sought to delete its CH domains (CH1 and CH2). Although numerous attempts failed to produce precise full deletions, we succeeded in obtaining multiple partial deletions. We chose two for further characterization: SMA-1ΔABD#1, which only retains the last 14 residues of CH2, and SMA-1ΔABD#2, which retains the first 17 residues of CH1 and the entire CH2 (Figure S2A). Both mutants phenocopied the sma-1(ru18) mutant: adult animals were small and had a reduced brood size, and embryos completed cytokinesis successfully. Similar to genetic crosses with sma-1(ru18), homozygous sma-1ΔABD#1;plst-1(prt89) animals were unable to propagate, but in this case they laid more embryos (16 on average). These embryos did not hatch (91%) or arrested at the L1 stage (9%). Immunoblots with the antibody raised against the SMA-1 PH domain showed that protein levels were reduced in the two sma-1ΔABD mutants (Figure S2B), which complicated the interpretation of the phenotype. We therefore asked whether partial RNAi-mediated depletion of SMA-1 in plst-1(prt89) embryos, to levels comparable to those in sma-1ΔABD mutants, was per se sufficient to induce cytokinesis failure (Figure S2C). In contrast to penetrant sma-1(RNAi), partial sma-1(RNAi) did not prevent cytokinesis in plst-1(prt89) one-cell embryos (Figure S2D). It is therefore likely that SMA-1’s inability to bind actin contributes to the embryonic lethality in plst-1(prt89);sma-1ΔABD mutants.

To examine how SMA-1 contributes to cortical F-actin dynamics during cytokinesis, we imaged LifeAct::GFP. In one-cell control embryos, approximately 70 s after anaphase onset, and coincident with the beginning of equatorial deformation, circumferentially aligned F-actin bundles could be observed in the division plane (Figure 5A). The localization pattern of F-actin in sma-1(RNAi) embryos was indistinguishable from that in controls (Figures 5A and S5A). In plst-1(prt89) embryos, F-actin bundles took longer to align circumferentially (Figures 5A and 5C). In plst-1(prt89);sma-1(RNAi) embryos, equatorial F-actin bundles initially attempted to align circumferentially, similar to F-actin bundles in plst-1(prt89) embryos, but became increasingly disorganized around the time of equatorial deformation (Figures 5B and 5C). At this time, bright actin clusters started to appear in the equatorial region, and their number increased over time (Figures 5A, 5B, 6C, S5A). These bright F-actin clusters were never observed in control or sma-1(RNAi) embryos and were only rarely observed in plst-1(prt89) embryos (Figures 5A and 6C). Strikingly, equatorial F-actin bundles had completely disappeared by 312 ± 64 s after anaphase onset, such that only F-actin clusters remained. Eventually, the equatorial deformation regressed 382 ± 82 s after anaphase onset (Figure 5A; Video S3). We conclude that co-inhibition of PLST-1 and SMA-1 severely disrupts the F-actin network at the cell equator, resulting in cytokinesis failure.

Figure 5

Co-inhibition of PLST-1 and SMA-1 results in collapse of F-actin bundles at the cell equator

(A) Images of the cortex in one-cell embryos expressing LifeAct::GFP. Numbers indicate the time in seconds after anaphase onset. Scale bar, 10 μm.

(B) Images of the equatorial region in a plst-1(prt89);sma-1(RNAi) embryo expressing LifeAct::GFP. The first frame corresponds to the time at which a distinct F-actin equatorial band has formed. Scale bars, 10 μm.

(C) Deviation of F-actin bundles from vertical alignment at the equatorial cortex versus time after anaphase onset (mean ± SEM).

(D) Images of the central plane in plst-1(prt89);sma-1(RNAi) embryos expressing LifeAct::GFP and NMY-2::mCherry. Numbers indicate the time in seconds after anaphase onset. Blue arrowheads point to membrane/cortex invaginations. Scale bar, 10 μm.

(E) Number of invaginations in the one-cell embryo during successive 25 s intervals after anaphase onset. For each time point the total number of invaginations observed in 10 embryos are plotted.

(F) Number of cortical invaginations in the one-cell embryo (mean ± 95% CI; 10 embryos) in a 400 s interval starting at anaphase onset. Statistical significance was determined using one-way ANOVA followed by Bonferroni’s multiple comparison: ∗∗∗∗p ≤ 0.0001; ns, not significant (p > 0.05).

See also Figures S2 and S5, Video S3, and Table S1.

Figure 6

Co-inhibition of PLST-1 and SMA-1 leads to coalescence of equatorial myosin into large puncta, which behave identically to puncta that form after Latrunculin A treatment

(A) Images of the cortex in one-cell embryos expressing NMY-2::mCherry. Yellow arrows and arrowheads point to myosin patches and myosin puncta, respectively. Numbers indicate the time in seconds after anaphase onset. Scale bar, 10 μm.

(B) Images of the cortical equatorial region in a plst-1(prt89);sma-1(RNAi) embryo co-expressing NMY-2::mCherry and LifeAct::GFP. Numbers indicate the time in seconds after the initial enrichment of NMY-2::mCherry at the cell equator. A blow-up of the 210 s time point (outlined by a dashed line) is shown on the right. Arrowheads point to F-actin clusters and myosin puncta that co-localize. Scale bars, 10 μm.

(C) Quantification of the number of myosin puncta and F-actin clusters (mean ± 95% CI) over time after anaphase onset. N indicates the number of embryos analyzed.

(D) Images of the cortex in one-cell embryos co-expressing LifeAct::GFP and NMY-2::mCherry after acute treatment with 10 μM Latrunculin A. Numbers indicate the time in seconds after anaphase onset, when Latrunculin A was added. Scale bar, 10 μm.

(E) Cortical images of plst-1(prt89);sma-1(RNAi) and Latrunculin A-treated one-cell embryos expressing NMY-2::mCherry showing fusion of myosin puncta. Scale bar, 5 μm.

(F) Quantification of NMY-2::GFP fluorescence recovery after photobleaching (mean ± 95% CI). Values were normalized to the fluorescence before photobleaching. Results from photobleaching of the regions in the schematics on the right and of the regions shown in Figures S6A and S6B were combined. Dashed lines are the single exponential fitting curves from which the half time of recovery and mobile fraction were extrapolated. N indicates the number of bleaching events analyzed (one per embryo).

(G and H) Quantification of the mobile NMY-2::GFP fraction (G) and the NMY-2::GFP half time of recovery (H) (mean ± 95% CI). Statistical significance was determined using one-way ANOVA followed by Bonferroni’s multiple comparison test: ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001; ns, not significant (p > 0.05).

See also Figures S5 and S6, Video S3, and Table S1.

Video S3. PLST-1 and SMA-1 cooperate to stabilize the actomyosin cortex during cytokinesis, related to Figures 5 and 6

Time-lapse series of the cell cortex in dividing one-cell embryos co-expressing LifeAct::GFP and NMY-2::mCherry. Cortical images are maximum intensity projections of seven z sections 0.5 μm apart acquired every 5 s from anaphase onset (0 s). Frame rate is 5 frames per second and time stamp is in seconds. Scale bar, 10 μm.

When astral microtubules of the mitotic spindle pull on a weakened actomyosin cortex, plasma membrane and cortical invaginations form. Consistent with F-actin network collapse and consequent weakening of the cortex, we observed numerous invaginations when imaging the central plane in plst-1(prt89);sma-1(RNAi) embryos (Figures 5D–5F). These invaginations were observed from anaphase onset and became more frequent ∼300 s later (Figure 5E), just prior to F-actin bundle collapse (Figure 6C). plst-1(prt89) embryos also presented invaginations, but these were less pronounced and much less frequent, and their number decreased as cytokinesis progressed. Mitotic spindle morphology and positioning was identical in control, plst-1(prt89), and plst-1(prt89);sma-1(RNAi) embryos expressing GFP::β-tubulin (Figure S3), suggesting that the plasma membrane invaginations in plst-1(prt89);sma-1(RNAi) embryos were not caused by increased astral microtubule density or altered centrosome proximity to the cell cortex.

Directed F-actin cortical flows are a hallmark of early cytokinesis.,, To assess whether perturbed F-actin cortical flows could contribute to cytokinesis failure in plst-1(prt89);sma-1(RNAi) embryos, we analyzed flows by Particle Image Velocimetry from anaphase onset in embryos expressing LifeAct::GFP (Figure S4). Posterior-anterior directed (X-component) flows in control embryos, which were particularly prominent on the posterior side, initiated before equatorial deformation and peaked shortly thereafter (Figures S4A and S4B). As described previously,, F-actin cortical flows in plst-1(prt89) embryos were erratic and decreased in velocity. The flow profile in plst-1(prt89);sma-1(RNAi) embryos was not further aggravated relative to that in plst-1(prt89) embryos. We also analyzed the Y-component flow profile and found it to be similar between plst-1(prt89) and plst-1(prt89);sma-1(RNAi) embryos (Figure S4C). These results imply that the failure to form a contractile ring in plst-1(prt89);sma-1(RNAi) embryos is not attributable to impaired F-actin flows.

F-actin bundle collapse after co-inhibition of PLST-1 and SMA-1 coincides with the coalescence of myosin into large puncta

To further assess the status of the actomyosin network, we imaged the cortex of embryos co-expressing LifeAct::GFP and NMY-2::mCherry. Initially, the patchy equatorial NMY-2::mCherry distribution was similar in control and plst-1(prt89);sma-1(RNAi) embryos. However, NMY-2::mCherry patches in plst-1(prt89);sma-1(RNAi) embryos quickly gave way to large, bright puncta (Figures 6A–6C and S5A). The appearance of NMY-2::mCherry puncta (∼115 s after anaphase onset) coincided with the appearance of F-actin clusters, and they frequently co-localized (Figures 6B and 6C). The number of bright NMY-2::mCherry puncta increased as F-actin bundles disappeared from the cell equator and reached a maximum 320 s after anaphase onset, shortly after the complete collapse of F-actin bundles (Figures 5A, 5B, and 6C). The remaining faint patches of NMY-2::mCherry disappeared along with the F-actin bundles. At this time (382 ± 82 s after anaphase onset) all cortical NMY-2::mCherry was in the form of bright puncta, and the cortex relaxed. The number of NMY-2::mCherry puncta decreased thereafter and all puncta eventually disappeared around 550 s after anaphase onset. We conclude that the disintegration of F-actin bundles at the cell equator of plst-1(prt89);sma-1(RNAi) embryos coincides with a re-distribution of myosin from faint patches to bright puncta.

Myosin localization dynamics after co-inhibition of PLST-1 and SMA-1 resemble those observed after drug-induced F-actin depolymerization

Myosin puncta are known to form at the cell surface in embryos treated with Latrunculin A, a drug that inhibits F-actin polymerization by sequestering actin monomers. To assess the similarity between the myosin puncta in plst-1(prt89);sma-1(RNAi) embryos and those observed after Latrunculin A treatment, we permeabilized one-cell embryos co-expressing Lifeact::GFP and NMY-2:mCherry and acutely treated them with 10 μM Latrunculin A at anaphase onset. As previously described, Latrunculin A treatment resulted in depolymerization of the cortical F-actin network and failure of cytokinesis (Figure 6D). Concomitant with F-actin depolymerization, NMY-2::mCherry accumulated in bright puncta at the cell surface, especially at the cell equator where the contractile ring would be assembled. These NMY-2::mCherry puncta were undistinguishable in brightness, size, and shape from those we observed in plst-1(prt89);sma-1(RNAi) embryos, and in both conditions puncta occasionally fused with one another (Figure 6E).

We further examined the dynamics of myosin puncta by photobleaching regions of two different sizes at the cortical equator during contractile ring assembly in embryos expressing NMY-2::GFP (Figures S6A and S6B). Quantification of the NMY-2::GFP signal over time post-bleaching revealed a fast and full recovery in control embryos, with a half time of recovery of 23.3 ± 3.7 s. In contrast, recovery of the NMY-2::GFP signal was significantly slower in both Latrunculin A-treated and plst-1(prt89);sma-1(RNAi) embryos, with a half time of recovery of 44.6 ± 5.4 s and 41.7 ± 5.5 s, respectively. Signal recovery occurs mostly from the cytoplasm in plst-1(prt89);sma-1(RNAi) and Latrunculin A-treated embryos, as actomyosin cortical flows are essentially absent. In controls, it is possible that cortical flows contribute to signal recovery. We note, however, that the recovering signal at the cell equator does reach a plateau, which is not what would be expected if the bi-directional flows converging on the equator were driving recovery (in this case the signal should keep increasing). The mobile NMY-2::GFP fraction also decreased slightly in either condition (0.94 ± 0.05 and 0.93 ± 0.05 in Latrunculin A-treated and plst-1(prt89);sma-1(RNAi) embryos, respectively; Figures 6F–6H). Thus, Latrunculin A treatment and plst-1(prt89);sma-1(RNAi) have the same effect on cortical myosin turnover.

Collectively, our results support the idea that myosin puncta observed in the absence of PLST-1 and SMA-1 form as a consequence of F-actin bundle collapse at the cell equator. We conclude that the combined action of plastin and βH-spectrin is critical for maintenance of the cortical F-actin network that forms the contractile ring.

In silico simulations recapitulate contractile ring formation in the presence of a small and rigid and a large and flexible crosslinker

To gain insight into how plastin and βH-spectrin cooperate to stabilize the contractile ring, we developed an object/agent-based computational model for contractile ring formation using Cytosim. In our simulations, F-actin, myosin bipolar filaments, plastin, and βH-spectrin were individually represented in a 2D space that mimics the equatorial cortex of the C. elegans one-cell embryo (Figures 7A and 7B; Video S4; STAR Methods). F-actin was modeled as either static or dynamic; myosin filaments were modeled as rigid rods with heads that can independently bind and walk along F-actin and dwell at the barbed-end of filaments before unbinding; each crosslinker consisted of two heads that bind to different actin filaments, and a connecting backbone, which is short and rigid for plastin (12 nm), and long and flexible for βH-spectrin (estimated to be about 250 nm when stretched).

Figure 7

In silico simulations recapitulate contractile ring formation in the presence of a small and rigid and a large and flexible crosslinker

(A) The 4 cytoskeletal elements of the simulations (not to scale).

(B) Time series of ring formation for the reference simulation with all elements (white, actin fibers; green, myosin; red, plastin; blue, βH/α-spectrin tetramer). A blow-up of the marked region is shown to highlight vertical F-actin alignment.

(C) Success rate for different scenarios.

(D) Ring metric over time for the scenario with all elements for different F-actin dynamics. The inset shows the success rate for each scenario as a function of different F-actin dynamics.

(E) Success rate for higher (2×) and lower (0.5×) amounts of each crosslinker. Corresponding results with no crosslinker variation are shown with black borders for reference.

(F) Time of ring formation ± SD as a function of changes in the amount of each crosslinker (measured for the successful cases). Top: changes in timing relative to simulations with all elements present. Bottom: changes in timing relative to simulations where the other crosslinker is absent.

(G) Ring metric evolution for different lengths of βH-spectrin with all elements present. The inset shows the time of ring formation ± SD as a function of βH-spectrin length.

(H) Schematic of wild-type SMA-1 and SMA-1Δ11SR.

(I) Percentage of cytokinesis completion/failure in one-cell embryos. N indicates the number of embryos analyzed.

(J) Images of the cortex in embryos expressing LifeAct::GFP. Numbers indicate the time in seconds after anaphase onset. Scale bar, 10 μm.

See also Figure S7, Videos S2 and S4, and Tables S1–S3.

Video S4. Simulations with all four elements, no βH-spectrin, no plastin, and neither βH-spectrin nor plastin, related to Figure 7

Time-lapse of typical simulations of ring formation with (left to right): all elements present, no spectrin present, two cases with no plastin present, and neither plastin nor spectrin present. In the first 3 panels a stable ring is assembled, while in the last two panels the rings break into asters. The successful and failed rings with no plastin present (third and fourth panels) have identical amounts of actin, myosin and spectrin. Colors as in Figure 7B (white - actin fibers, green - myosin, red - plastin, and cyan - βH/α-spectrin tetramer). Every frame is 1 s of simulated time over a total of 100 s (20 s to 120 s after anaphase).

To test whether the combination of these 4 cytoskeletal elements could recreate contractile ring formation, we defined the metric M for successful ring assembly based on the distribution of the motors along the ring network (Figure S7B). M varied between one, when motors move toward the middle of the simulated space horizontally and uniformly spread across the vertical parallel to the actin filaments, and zero, when the ring network ruptures because of connectivity loss, and the motors contract the network into clusters (Figure S7B; STAR Methods). By comparing our metric to videos of simulations, we determined that a ring has successfully formed if M exceeds a threshold value of 0.5. Using this threshold, we calibrated our simulations to the four scenarios: all 4 elements, no βH-spectrin, no plastin, and neither βH-spectrin nor plastin (no crosslinkers). Similar to our experimental results (Figure 2A), the success rate of ring formation was 100%, 98%, 92%, and 15% in the four scenarios, respectively (Figures 7C and S7E–S7H). In the 4 elements scenario, the calibrated system also recapitulated the time interval of contractile ring formation (average 55 s, range 20 s to 75 s; Figure 7D), which is similar to the interval between equatorial actin enrichment (∼20 s after anaphase onset, Video S3) and equatorial deformation, when F-actin bundles are already aligned (Figure 5A). The calibration was possible when the number of plastin connectors was at least one order of magnitude higher than the number of βH-spectrins (Table S3), which is consistent with the higher signal of PLST-1::GFP relative to that of SMA-1::spGFP or SPC-1::GFP (Figure S7A). Changes in the polymerization and depolymerization rates of actin filament ends had little impact on timing of ring assembly or success rate in the 4-element scenario (Figure 7D).

Doubling the amount of plastin had adverse effects on the success rates, while halving its amount only affected simulations with no βH-spectrins. Changes in the amounts of βH-spectrins did not affect the success rate when all elements are present, but halving it did affect the success rate in the absence of plastin (Figure 7E). An excess of either crosslinker in the 4-element scenario slowed down ring formation times, which is in agreement with previous reports that relate increased network connectivity to decreased contractility., Interestingly, increasing plastin in the absence of spectrin or vice versa also led to an increase in ring formation time, indicating that one crosslinker does not compensate for the other. Decreasing the amount of either crosslinker to 75% or 50% sped up ring formation time (Figure 7F). The likely reason for this is that, as long as there is some level of F-actin connectivity, and in the absence of anchoring to a plasma membrane and of other actomyosin regulators that could provide drag, it is easier for motors to align the ring network.

The main difference between SMA-1 and UNC-70 structure is the number of spectrin repeats that confers different lengths to the two molecules. To test whether the length of SMA-1 is important for its function during cytokinesis, we simulated ring assembly in the presence of βH-spectrin that ranged from half to double its estimated length of 250 nm and concluded that within this range βH-spectrin length had negligible effects on success rate and timing of ring formation (Figure 7G; Video S4). This is in agreement with results we obtained in vivo with a truncated version of SMA-1 that lacks 11 of its 29 spectrin repeats (SMA-1Δ11SR; Figure 7H). SMA-1Δ11SR is approximately the same length as UNC-70, which contains 17 spectrin repeats. Animals expressing SMA-1Δ11SR were viable, cytokinesis in one-cell embryos was normal, and expression of SMA-1Δ11SR in the plst-1(prt89) background neither aggravated the time required for ring formation nor the success rate of cytokinesis (Figures 7I and 7J). We conclude that the difference in length between the two β-spectrins is not a critical parameter for SMA-1’s cooperation with plastin.

Discussion

Studies in vitro and theoretical approaches suggest that F-actin crosslinking is critical for tension propagation and contractility of actomyosin networks,,, yet the role of F-actin crosslinkers in vivo remains poorly understood. We determined the contributions of nine conserved crosslinkers to embryonic cytokinesis in C. elegans and found that none of them are essential when inhibited individually. Inhibition of the plastin homolog PLST-1, which has previously been implicated in cytokinesis, delays contractile ring assembly but does not affect the speed of ring closure., In the sensitized plastin inhibition background, the βH-spectrin SMA-1, but none of the other crosslinkers we examined, becomes essential for cytokinesis. Our findings indicate that interplay between plastin and βH/α-spectrin occurs via their actin-binding activities and does not involve SMA-1’s plasma membrane anchoring function. Thus, two types of F-actin crosslinker with distinct properties collaborate in actomyosin network organization during cytokinesis. The only previous hint that spectrins may have a role in animal cell division comes from work in D. melanogaster ovaries, where α-spectrin-deficient cysts were found to have fewer cells. Of note, cell division in bacteria involves the spectrin-like protein EzrA, which may contribute to the condensation of the tubulin FtsZ (tubulin-homolog) filaments that compose the cytokinetic Z-ring. However, EzrA is a small protein of 65 kDa that only contains five spectrin-like repeats and lacks an ABD, implying a different function from that of SMA-1 that we report here.

The small (9–12 nm) and rigid plastin, whose main structural feature are two ABDs in tandem, organizes non-branched F-actin in tight, parallel bundles.,,, By contrast, conventional spectrin tetramers are large and flexible (an estimated 80 nm when relaxed and 200 nm when extended)., β-spectrins contain up to 30 spectrin repeats linked by flexible loops. Unfolding of the spectrin repeats under shear stress allows for spectrin extension, which provides red blood cells with the ability to deform when crossing blood vessels. In axons, spectrins provide protection from mechanical stress, caused, for example, by animal movement in C. elegans. In C. elegans, spectrins are also required for F-actin bundle organization and reinforcement during the cycles of stretching and contraction in the spermatheca, where oocytes enter to be fertilized before being expelled into the uterus, and spectrins help stabilize F-actin bundles in epidermal cells, allowing for their remodeling when muscle cells contract during embryo elongation. Whether βH-spectrin is under tension during cytokinesis and whether it offers mechanical protection to the cell equator remains to be investigated.

PLST-1 is clearly bundling equatorial F-actin, as its signal and that of LifeAct overlap. In contrast, SMA-1 appears to fill the spaces between F-actin bundles, which supports the idea that the two crosslinkers make distinct contributions to the stability of the contractile ring. Super-resolution microscopy will be required to better understand SMA-1’s distribution within the actomyosin network. In our in silico model, successful ring formation requires that plastin is one order of magnitude more abundant than spectrin, and, while either crosslinker is sufficient for actomyosin ring formation, plastin must still be more abundant than spectrin in these scenarios. This requirement probably reflects the size difference between plastin and spectrin: the compact plastin can only crosslink actin filaments that are already in close proximity, whereas the approximately 20 times longer βH/α-spectrin can crosslink filaments at much greater distance. Variations in the absolute amounts of βH/α-spectrin had negligible effects in our simulations, but doubling plastin adversely affected ring formation success rate in the absence of βH/α-spectrin. This is because an excess of plastin stiffens the network and stalls the motors, whereas too much spectrin, because of its length and flexibility, still allows the network to be remodeled.

An interesting open question is why the βH-spectrin SMA-1 is used for cytokinesis and not the conventional β-spectrin UNC-70, whose expression in the early embryo appears to be negligible (GFP-tagged endogenous expression is not detectable on the cortex, and only a feeble signal is detected by immunofluorescence at the cell-cell boundary of the 2-cell embryo). The most obvious difference between βH-spectrin and conventional β-spectrin is length, but our simulations and in vivo results with truncated SMA-1, which contains only one more spectrin repeat than UNC-70, argue against this being an important factor. Another feature that distinguishes the two β-spectrins is the ability to respond to mechanical stress: βH-spectrin is mechanosensing in micropipette aspiration assays in S2 cells, whereas conventional β-spectrin is not. βH-spectrin’s mechanosensing behavior explains its recruitment to the fusion site of myoblasts in Drosophila in response to the transient increase in protrusive forces by the attacking cell. At that site, βH/α-spectrin keeps cell-adhesion molecules fenced and serves as a sieve to focus the protrusions of the attacking cell, facilitating the fusion event. In the context of cytokinesis, mechanosensing could potentially allow the βH-spectrin SMA-1 to detect tension at the cell equator that builds due to myosin recruitment, thereby facilitating local spectrin accumulation. If mechanosensing by βH-spectrin is important for cytokinesis, UNC-70 may not be able to functionally replace SMA-1. Unfortunately, we have not been able to test this, as efforts to force transgenic UNC-70 expression in the one-cell embryo have been unsuccessful, presumably due to the well-documented potency of silencing mechanisms that operate in the C. elegans germline.

Our analysis of myosin and F-actin dynamics shows that the equatorial actomyosin network progressively disintegrates in the absence of βH-spectrin and plastin. It is known that myosin can remodel isotropic F-actin meshworks into asters in vitro because myosin filaments dwell at F-actin plus ends, which results in plus-end clustering., Addition of α-actinin leads to the stabilization of asters, which become larger, interconnected, and able to move toward one another until they merge. In SMA-1 and PLST-1 co-inhibited embryos, it is possible that equatorial F-actin orientation becomes randomized and that the end-dwelling behavior of NMY-2 drives its coalescence into clusters of F-actin plus-ends. Lack of interconnectivity between these clusters would then prevent network contraction. In addition, F-actin itself is likely to become destabilized in the absence of βH-spectrin and plastin, since we show that the bright cortical myosin puncta resemble those that form after Latrunculin A treatment. In this view, the roles of βH-spectrin and plastin during cytokinesis go beyond mere crosslinking. Several crosslinkers have been reported to stabilize F-actin in vitro by decreasing its depolymerization rate and, in some instances, by concomitantly increasing the elongation rate or decreasing access to the severing protein cofilin.53, 54, 55 There is also in vivo evidence for F-actin stabilization by crosslinkers: plastin stabilizes actin patches in yeast, espin stabilizes and lengthens F-actin bundles in microvilli of epithelial cultured cells and in stereocilia in hair cells, and fascin and forked stabilize bristles in D. melanogaster.,, During contractile ring assembly in cytokinesis, when formin-mediated F-actin elongation occurs concomitantly with alignment and compaction of F-actin bundles, a dual ability to crosslink and stabilize F-actin would appear to be advantageous. Following contractile ring assembly, however, contractile ring constriction requires F-actin depolymerization., This raises the question of whether the F-actin crosslinking and stabilization activities of SMA-1 and PLST-1 might be regulated to facilitate F-actin depolymerization during ring constriction. In the future, it will be important to investigate how the cooperation between plastin and spectrin impacts F-actin network dynamics in vitro, including in the presence/absence of cofilin, which would allow further refinement of our in silico model.

Our findings highlight the importance of exploring how interplay between crosslinkers of variable size, flexibility, and mechanical properties contribute to actomyosin network contractility and stability. For cytokinesis, we envision the following sequence of events: equatorial plastin promotes myosin motor-driven F-actin alignment through F-actin bundling as well as stabilization, the latter possibly involving inhibition of F-actin depolymerization or restriction of cofilin access to F-actin. As tension builds at the cell equator because of continuous myosin recruitment, βH-spectrin accumulates locally due to its mechanosensing properties and helps protect the circumferential F-actin bundle array from mechanical stress. It will be interesting to test this model using additional engineered plastin and spectrin mutants and to investigate potential roles of the plastin/βH-spectrin pair and other crosslinker combinations in actomyosin-dependent processes beyond cytokinesis.

STAR★Methods

Key resources table

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact Ana X Carvalho (anacarvalho@ibmc.up.pt).

Materials availability

Key strains, plasmids and antibodies are available upon request through the Lead Contact.

Experimental model and subject details

C. elegans strains

The C. elegans strains used in this study are listed in the Key resources table. Strains were maintained at 20°C on nematode growth medium (NGM, for 1 L: 3 g NaCl, 17 g agar, and 2.5 g peptone, plus 1 mL of 1 M CaCl2, 1 mL 5 mg/mL cholesterol in ethanol, 1 mL of 1 M MgSO4 and 25 mL of 1 M KH2PO4 pH 6 after autoclaving at 110°C for 30 minutes and cooling down to 55°C) plates seeded with Escherichia coli OP50.

Method details

RNA interference

RNAi was performed either by feeding animals with E. coli HT115 bacteria expressing double stranded RNA (dsRNA) or by injecting animals with dsRNA.

To produce dsRNA for injection, primers with tails containing T3 and T7 RNA polymerase promoters were used to amplify the desired regions from N2 cDNA (Table S2). The PCR products were transcribed with T3 or T7 RNA polymerases (MEGAscript Transcription Kit; ThermoFischer Scientific). Transcription reactions were cleaned up (NucleoSpin RNA Clean-up, Macherey-Nagel) and diluted with 3 × soaking buffer (32.7 mM Na2HPO4, 16.5 mM KH2PO4, 6.3 mM NaCl, and 14.1 mM NH4Cl). Injection of dsRNA at ∼1 μg/μl was performed in L4-stage animals of strains GCP113 or GCP570, and the animals were incubated at 20°C for 48-54 hours prior to dissection for imaging. RNAi by injection was used to deplete ATN-1, PLST-1, FLN-1, UNC-70, UNC-59, ANI-1, and ERM-1. In the case of fln-1(RNAi), dsRNAs fln-1#1 and fln-1#2 were mixed together to cover all isoforms.

For RNAi by feeding, we used the RNAi feeding vector L4440 transformed into E. coli HT115. L4440 vectors for RNAi of cor-1, frm-1, unc-44 or perm-1 were obtained from the Ahringer library (Source Bioscience) and sequenced to confirm the gene target. L4440 for sma-1 RNAi was generated by inserting a sma-1 fragment amplified from N2 cDNA (primers 5′-CCCCCGATATCGACACCCTTTCAAATG-3′ and 5′-CCCCCGATATCCCTTCTGTTCCATCTCGTTTC-3′) into the EcoRV restriction enzyme site. The bacteria were grown overnight at 37°C in 50 mL Luria broth medium containing 12.5 μg/ml tetracycline and 100 μg/ml ampicillin. The culture was centrifuged for 10 minutes at 2,500 g. The cell pellet was resuspended in 2.5 mL Luria broth medium containing 12.5 μg/ml tetracycline, 100 μg/ml ampicillin, and 1 mM IPTG. 75 μl were used to inoculate NGM plates to which 100 μl of a 1:1:1 mix of 100 mg/ml ampicillin, 1 M IPTG, and 5 mg/ml tetracycline had been added 2 hours prior. The expression of double-stranded RNA was induced overnight at room temperature and in the dark. 25-30 L4-stage animals of strains GCP22, GCP113, GCP570, GCP646, GCP927 or GCP1170 were added to the plates and incubated at 20°C for 48-54 hours prior to dissection for imaging.

Generation of mutants using CRISPR/Cas9

Guide RNAs and repair templates used to edit the C. elegans genome, as well as diagnostic primers used to screen for positive editing events, are listed in Table S2. Modified regions were confirmed by sequencing. To remove potential off-target mutations, modified animals were subjected to six rounds of outcrossing with wild-type N2 animals. Other modifications (e.g., fluorescent markers) were subsequently introduced by mating.

For generation of sma-1ΔPH, sma-1ΔABD#1 and atn-1ΔABD, we used a plasmid-based approach, in which a single guide RNA (sgRNA) and the Cas9 protein are co-expressed from pDD162. Guide sequences were chosen using the online tools https://zlab.bio/guide-design-resources and http://crispor.tefor.net/. Single-stranded repair templates (Integrated DNA Technologies) included 50-60 bp homology regions flanking the Cas9 cleavage site and, when necessary, silent mutations to avoid repair template recognition by Cas9. For sma-1ΔABD#1, the intended modification was the removal of the entire ABD. However, the last 14 residues of the CH2 were retained. The sequence of the modified region is as follows: 5′-cagGTTCGTGTTCCAACAAGTGGTGCACCACCAGTTCGCGCAGATGCCAATGGAACAGATCAAGACGAGTTCAATAATGAGACACTGTACTTTGAAAGATCACGCATTCGGACATCAATCATTACATATGTTTCCTTGTACTATCATCATTTTGCAAAGCAAAAGACTGAGATGACAGGAGCTCGACGTATTGCTAATgtaagtgtttataagttcatg-3′ (bases in lowercase indicate intron regions).

For sma-1ΔSH3, sma-1ΔABD#2, and sma-1Δ11SR, we injected a ribonucleoprotein (RNP) mix consisting of crRNA/tracrRNA (CRISPR/Cas9 design tool ht https://www.idtdna.com/site/order/designtool/index/CRISPR_PREDESIGN; Integrated DNA Technologies) and purified recombinant Cas9 protein. The repair template for sma-1ΔABD#2 was partially single-stranded (assembled by overlap extension PCR) and consisted of 120-bp homology regions flanking the Cas9 cleavage site, silent mutations to avoid repair template recognition by Cas9, and the sma-1 intron between exons 3 and 4. For sma-1ΔSH3 and sma-1Δ11SR, the single-stranded repair template included 50-bp homology regions flanking the Cas9 cleavage site. For sma-1ΔABD#2, the intended modification was the removal of the CH1 domain (part of exon 3 and exon 4) leaving the intron between exons 3 and 4 but not the intron between exons 4 and 5. However, sequencing of the modified region revealed that the first 17 residues of CH1 were retained (retains part of exon 3, entire intron between exons 3 and 4, and part of exon 5): 5′-aaagttttaaagatccgaaaaaaaactgagaacgtttagaaatatttaacagttcaattagcatcaagttcataatgattttttcagGTTCGTGTTCCAACAAGTGGTGCACCACCAGTTCGCGCAGATGCCAATGGAACAGATCAAGACGAGTTCAATAATGAGACACTGTACTTTGAAAGATCACGCATTCGGACACTGCAGGACGAACGTGTGCACATTCAGAAGAAGACGTTCACAAAATGGTGCAACTCCTTTCTTAATCGG- intron3/4 as in sma-1 transcript R31.1a, Wormbase version WS279 - GATGAGGAAGAGCGTGGAGAGCGAAAACATGCCAAAGATGCCTTGCTGTTGTGGTGTCAGAGAAAAACAGCTGGATATCCAAATGTTCGCATCGAAAACTTCACTACAAGTTGG-3′. For sma-1Δ11SR the intended modification was the removal of 13 spectrin-repeats, however only 11 spectrin repeats were removed (mid spectrin repeat #14 to mid spectrin repeat #25). The obtained sequence is as follows: 5′-CGTGACGTTGACGAGTTTGAGCAATGGATGGCAGACAAAATGGCCAACATGCCACGCTCACACGAACTGTACGAGATTGATCGCTTGAAGAAGCGTGCGGATGGACTTCTAGCCAGAGAGCATCACGATGCAATGTCGATTGCTGCGAAACAACGCAAACTGGAAGCATTGTTCGGAGACCTCTGCAGAGAATGTGCCAGACGACGAACTCAAATTGTTGATGCTTCCAAATATCACAAGTTTGTTAGACAAGCCGATGATTTGTCAGACTGGCTACGTGAGAAAGAAAGGTCCGCATCAGCTGAGGATTACGGTCAAGATTTGGAGGATTGCCAACAAATTATTGAGCAGTTCGAATCTACAGTTCGTGAATTGGCAGCTGCTGGAGAGCGTGTTGCTGCTGTCCAACGTGCACAAGAAGATCTTCTACGAAGTGGACATCCATATGGAGCATCAATCACTGCAAAAGGCGCTGATGTTCAGAGACTATGGACTCATGTCAATGAAGTGGCTAATGAGAGAAAGCAAGCGTTGAATGGAGCTCGTCAAGTTCATCGATTCGACCAAGAAGCCGATCAGATATTGAATTGGTTACAAGATAAGGAAGCGACTGGAGTAGCAATGGAACAGGAAGATCTTTCGAGAGCTGATCTTGCTTCCGTGAA-3′ (14 base pairs in italics replaced the region removed from mid spectrin repeat #14 to mid spectrin repeat #25, and did not cause a frameshift).

In all cases repair templates and pDD162 or RNPs were injected into gonads of young adult N2 hermaphrodites. To facilitate the identification of successfully injected animals, injection mixes also contained a repair template and pJA58 (a gift from Andrew Fire, Addgene plasmid # 59933; http://addgene.org/59933; RRID: Addgene_59933) or crRNA to generate the R92C mutation in dpy-10, which causes a dominant roller phenotype in mutated F1 progeny.

Tagging of endogenous SMA-1 and PLST-1

Guide RNAs and repair templates used, as well as diagnostic primers used to screen for positive editing events, are listed in Table S2. All modified regions were confirmed by sequencing.

To generate an endogenous fluorescent version of SMA-1, we used the self-complementing split GFP system. We inserted 6 copies of GFP11 in tandem downstream of the sma-1 open reading frame using plasmid-based CRISPR/Cas9-mediated genome editing, as described above. The plasmid template used for homology recombination was pAC571, which was built using Gibson assembly. The fragments used for pAC571 assembly were as follows: pBluescript backbone, amplified with primers 5′-CATGTGAGCAAAAGGCCAGC-3′ and 5′-ATCGCCCTTCCCAACAGTTG-3′; 1 kb left homology region in the sma-1 open reading frame, amplified from N2 genomic DNA (gDNA) with primers 5′-TCAGGCTGCGCAACTGTTGGGAAGGGCGATAGCTTTCTTCCCTCCGTC-3′ and 5′ CTTTGAATGTTTGGATCCACGTTTAAACAAGGATCCGAATCCTTTTCGTTCTG-3′ (bases in bold indicate silent mutations introduced to avoid repair template recognition by Cas9); 6xGFP11, amplified from pDC592 with primers 5′-CTTGTTTAAACGTGGATCCAAACATTCAAAGGGAGGGAGGGCCGGCTCTG-3′ and 5′-GGTGATTCCAGCAGCGTTAACGTACTCAT-3′ (bases in bold indicate silent mutations introduced to avoid repair template recognition by Cas9; bases in italics indicate part of the linker region); and 1 kb right homology region in the sma-1 3′UTR, amplified from N2 gDNA with primers 5′-ATGAGTACGTTAACGCTGCTGGAATCACCTAAATACGTCACCACACGCTGATCTTC-3′ and 5′-CTGGCCTTTTGCTGGCCTTTTGCTCACATGCCATTTTGTTGGATTGCTC-3′ (bases in bold indicate silent mutations introduced to avoid repair template recognition by Cas9). Animals expressing SMA-1::6xGFP11 were crossed with animals expressing GFP1-10 from the mex-5 promoter (GCP794 strain). GCP794 was obtained by subjecting the strain OD4016 (kindly provided by Arshad Desai) to six rounds of outcrossing with N2 animals.

To generate an endogenous mCherry-tagged version of PLST-1 (PLST-1::mCherry), we generated a partially single-stranded repair template (hybrid template), as in Dokshin et al., 2018. The repair template consisted of 120-145-bp homology regions flanking the Cas9 cleavage site, silent mutations to avoid repair template recognition by Cas9, and a 10 amino acid linker followed by the mCherry tag. The long fragment of the repair template was amplified with the primers: 5′ aaaatcgtccaaaaccccaaattgttttcagGTGAAGCCAAAAATGGTGTTGACAGTGTTTGCATGTCTAATGGCTCGTGACTATCTTCCCGATATGAAGCAAGGAGCAGCTTCTGCTCCGATCGTCCCAATGATCAATGGAAATGAGGACCCTTGGAGGGTACCGGTAGAAAAAATGGTCTCAAAGGGTGAAGAAGATA-3′ (145 bp of homology region in the plst-1 open reading frame; bases in bold indicate silent mutations introduced to avoid repair template recognition by Cas9; bases in italics correspond to the 10 amino acid linker, bases in lowercase indicate introns) and 5′ aatttttgtgagaaaaaagatggaaaaacgggaaaatagagatatatatatggaggaaaatttgtatttccgcgattttttttcggcgggaaattcaagaaattcgaaatttttgcagaaCTACTTATACAATTCATCCATGCCACC-3′ (120 bp of homology region in the plst-1 3′ UTR; underlined bases in lowercase indicate 3′UTR region). The short fragment of the repair template was amplified with the following primers: 5′ GATCGTCCCAATGATCAATGGAAATGAGGACCCTTGGAGGGTACCGGTAGAAAAAATGGTCTCAAAGGGTGAAGAAGATA-3′ (bases in bold indicate silent mutations introduced to avoid repair template recognition by Cas9; bases in italic indicate the 10 amino acid linker) and 5′ CTACTTATACAATTCATCCATGCCACC-3′. Animals expressing PLST-1::mCherry were subjected to six rounds of outcrossing with N2 animals.

Live imaging

Gravid hermaphrodites were dissected and one-cell embryos were mounted in a drop of M9 (86 mM NaCl, 42 mM Na2HPO4, 22 mM KH2PO4 and 1 mM MgSO4) on 2% agarose pads overlaid with a coverslip. Live imaging was performed at 20°C.

Images in Figures 1, 3D, 4, 5, 6, 7, S1B, S1E, and S3–S6 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 60x oil-immersion Plan-Apochromat objective (N.A. 1.4), extra 1.5x magnification, and solid-state lasers of 488 nm (50 mW) and 561 nm (50 mW). For image acquisition, an electron multiplication back-thinned charge coupled device camera (iXon Ultra 897; Andor Technology) was used. Acquisition parameters, shutters and focus were controlled by Andor iQ3 software. Images in Figures 3 and S7A were acquired on a similar system equipped with a confocal scanner unit (CSU-22; Yokogawa Electric Corporation), an iXon EM+ DU-897 (Andor Technology), mounted on an inverted microscope (IX81; Olympus) equipped with a 100x oil-immersion UPLSAPO objective (N.A. 1.4), extra 2x magnification, and solid-state lasers of 488 nm (50 mW) and 561 nm (50 mW).

For cortical imaging, 7 × 0.5 μm z stacks were collected every 5 s in embryos of strains GCP21, GCP22, GCP113, GCP570, GCP809, GCP831, GCP832, GCP838, GCP927, GCP991, GCP1014, GCP1036, GCP1163, GCP1164, GCP1169, GCP1170, GCP1176, GCP1187, GCP1207, GCP1208, RZB217, OD26, OD130 and GOU2936 (Figures 1A, 3, 4E, 5A, 5B, 6A, 6B, 6D, 6E, 7J, S5A, S6, and S7A). An image at the center of the embryo was also acquired to determine the timing of anaphase onset. For central plane imaging in Figures 1A, 2C, 5D, S1B, S1F, and S3A, 6 × 1 μm z stacks were collected every 10 s in embryos of strains GCP21, GCP22, GCP113, GCP570, GCP646, GCP831, GCP832, GCP927, GCP991, OD26, OD130 and GOU3103. For fluorescence recovery after photobleaching (FRAP) experiments (Figures 6F–6H and S6), a FRAPPA photobleaching module (Andor Technology) placed between the spinning disk head and the microscope was used. Photobleaching was performed by 3 sweeps of a 405 nm laser with 100% power and 40 μs dwell time.

For measurement of total cytokinesis time, contractile ring diameter decrease, ring constriction rate and percentage of cytokinesis completion/failure in Figures 1C, 1D, 2A, 2D, 4C, 4D, 7I, and S2D, images were acquired with an epifluorescence microscope (Zeiss Axio Observer Z1) equipped with a 63x Plan-Apochromat objective (N.A. 1.4), a Colibri.2 LED light source, and an Orca Flash 4.0 camera (Hamamatsu). Acquisition parameters, shutters, and focus were controlled by Zen 2.3 software (Zeiss). 7 × 1 μm z stacks were collected every 10 s in embryos of strains GCP113, GCP570, GCP809, GCP838, GCP1014 and GCP1036.

Image processing

Image processing and analysis was performed using Fiji (ImageJ; National Institutes of Health) or MATLAB (MathWorks). Z stacks taken on the cell cortex were projected using the maximum intensity projection tool. The equatorial region of the central plane was selected to make the kymographs displayed in Figures 2C and S1F, the equatorial region of the maximum projection of cortical planes was selected to make the kymographs displayed in Figures 5B, 6B, and S5A, and a 150 × 5 pixel rectangle positioned at the center of cortical plane was selected to make the kymographs in Figure S4D, using the Make Montage tool. Images in each figure panel are scaled equally, except for Figures 1A and 3. Graph plotting, curve fitting and linear regressions were performed with Prism 8.0 (GraphPad Software) or MagicPlotPro 2.8.2.

Latrunculin A treatment

Acute treatment with Latrunculin A in Figures 6D–6H and S6 was performed in perm-1(RNAi) permeabilized embryos. 25–30 L4 animals were placed on a plate containing 0.005 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) and E. coli HT115 bacteria expressing dsRNA against perm-1. After incubation for 14–18 hours at 20°C, adult hermaphrodites of strains GCP22 or GCP113 were dissected and permeabilized embryos were filmed in meiosis medium (25 mM HEPES, pH 7.4, 0.5 mg/ml inulin, 20% heat-inactivated fetal bovine serum, and 60% Leibowitz-15 medium) without compression. 10 μM Latrunculin A (Sigma) was added at anaphase onset. At the end of each video, medium containing 33 μM FM4-64 (Molecular Probes) was added to the imaging chamber to confirm that the imaged embryo was permeable.

Generation of antibodies against SMA-1

Two affinity-purified rabbit polyclonal antibodies against the actin-binding domain (SMA-1 ABD in Figures 4B and S1A) and pleckstrin homology domain (SMA-1 PH in Figures S2B and S2C) of SMA-1 were generated.To produce GST-tagged ABD, cDNA encoding amino acid residues 195-519 of SMA-1 (residue numbers as in isoform a, from transcript R31.1a, Wormbase version WS279) was cloned into the EcoRI and NotI restriction sites of pGEX-6P1 using primers 5′-CCCGAATTCATTTCCGGAGACAAACTCGG-3′ and 5′-CCCGCGGCCGCTTCCTGTCTTTGAAGCTCGGC. To produce GST-tagged PH domain, cDNA encoding residues 3689-3898 of SMA-1 was cloned into the BamHI and NotI restriction sites of pGEX-6P1 using primers 5′-CCCGGATCCAGAAAGACTCAGGAAATCTCTC-3′ and 5′-CCCGCGGCCGCTTAGTTGTACGCATACGACTTGAG-3′. GST::ABD and GST::PH were expressed in E. coli BL21 and Rosetta, respectively, purified and injected into rabbits at the in-house Animal Facility. Sera were affinity purified on a HiTrap N-hydroxysuccinimide column (GE Healthcare) against covalently coupled ABD and PH (TEV-cleaved with prescission protease in cleavage buffer: 50 mM HEPES pH 7.6, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.01% Tween 20, at 4°C overnight), aliquoted in 10% glycerol, snap-frozen and stored at −80°C.

Protein extracts and immunoblotting

For the immunoblots in Figures 4B, S1A, S2B, and S2C, protein extracts were prepared from 100 adult stage animals of strains N2, GCP808, AZ30, GCP986 and GCP564. Animals were collected in M9 medium with 0.1% Triton X-100 and washed three times in the same medium. The pellet of worms was resuspended in 100 μl of SDS-PAGE sample buffer (250 mM Tris pH 6.8, 30% v/v glycerol, 8% w/v SDS, 200 mM DTT and 0.04% w/v bromophenol blue) and one-third of the volume of quartz sand (Sigma) was added. Tubes were subject to three 5-minute cycles of boiling at 95°C and vortexing, after which the quartz sand was pelleted and the supernatant recovered. Protein samples (20-30 μl, equivalent to 20-30 animals) were resolved by SDS-PAGE (8% acrylamide gel) and transferred to a 0.2-μm nitrocellulose membrane (GE Healthcare). Membranes were blocked with 5% non-fat dry milk in TBST (20 mM Tris, 140 mM NaCl, and 0.1% Tween, pH 7.6) and probed at 4°C overnight with 1 μg/ml anti-SMA-1 ABD antibody or 1 μg/ml anti-SMA-1 PH antibody and 1:5000 anti-α-tubulin antibody (DM1-α, Sigma). Membranes were washed three times with TBST, incubated with HRP-conjugated secondary antibodies (mouse anti-rabbit 1:5000 or goat anti-mouse 1:5000; Jackson ImmunoResearch) for 1 hour at room temperature, and washed again three times with TBST. Proteins were visualized by chemiluminescence using Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific) and imaged in a ChemiDoc XRS+ System with Image Lab Software (Bio-Rad).

Immunofluorescence

For the immunofluorescence in Figure 3D, twelve gravid adult hermaphrodites were dissected in 5 μl of M9 buffer on a poly-L-lysine coated slide. A 13 mm2 round coverslip was placed over the M9, and slides were placed into a container with liquid nitrogen. After flicking away the coverslip, samples were fixed for 20 minutes in −20°C methanol. Samples were re-hydrated in PBS (137 mM NaCl, 2.7 mM KCl, 8.1 mM Na2HPO4, 1.47 mM KH2PO4), twice for 5 minutes, blocked with AbDil (PBS, 2% BSA, 0.1% Triton X-100) in a humid chamber for 30 minutes at room temperature, and incubated in AbDil with rabbit anti-SMA-1 PH antibody (1:1000) and anti-α-tubulin antibody (1:1000, DM1-α, Sigma), overnight at 4°C. After washing four times for 5 minutes in PBS, samples were incubated in AbDil with Alexa Fluor 488 anti-rabbit IgG (1:300; Jackson ImmunoResearch) and Alexa Fluor 594 goat anti-mouse IgG (1:300; Jackson ImmunoResearch) for 1 hour at room temperature. Samples were washed four times for 5 minutes in PBS and mounted in Prolong Gold (Invitrogen).

Computational simulations

We simulated the formation of cytokinetic rings using the Open Source engine Cytosim, which is capable of simulating the thermal environment of the cell as well as the forces generated by the interaction of biological fibers and binding proteins. We identified 4 essential components to simulate the assembly and maintenance of ring structures: F-actin, myosin motor minifilaments, and the crosslinkers plastin and βH-spectrin. Parameters that characterize each component are listed below (corresponding references are in Table S3).

Actin fibers were modeled as a series of geometrically polarized rigid segments connected by flexible hinges, with an overall rigidity 0.075 pN μm2. Initially segments were 0.1 μm long forming a nongrowing fiber of total length 1 μm, with a barbed and a pointed end. Actin dynamics are very complex and mostly mediated by auxiliary proteins such as formin, ADF/cofilin, capping proteins, and breakage events. We chose to model actin dynamics in simple terms, where each end of a fiber can stochastically switch between a growth state and a shrink state at fixed rates, and ends grow and shrink with pre-defined speeds. We prevented fibers from disappearing by defining a minimum length of 0.01 μm where they remain inactive until a rescue event. When the depolymerizing end encounters a crosslinker, it has a 90% chance of continuing depolymerizing (and unbinding the crosslinker) and a 10% chance of stopping depolymerizing. Values were adjusted so that average fiber length remains roughly constant.

The contraction of actin fibers into a ring is driven by molecular motors designed to model non-muscle myosin II. The backbone of the motor is a 0.15 μm long rigid solid. There are 4 binding heads on each end, equally spaced by 0.015 μm starting from the ends of the solid for a total of 8 binding domains. Each head can bind to any actin fiber within a range of 0.01 μm at a rate of 10 Hz and while bound can unbind at a rate of 0.3 Hz. While bound to an actin fiber, heads walk toward the barbed end of the fiber with a specified maximum speed which can be suppressed by the amount of drag force on the motor with a stall force of 4 pN. Maximum speed was chosen to match the proper time frame measured experimentally. Binding heads follow normal unbinding rules when reaching the end of fibers as opposed to immediately falling off. If the barbed end of an actin fiber attempts to grow or shrink while a motor head is attached, the motor head will remain attached to the end. Allowing for immediate myosin unbinding from depolymerizing barbed ends does not affect timing and success rate when all components are present, but slows down ring formation for all perturbed cases and leads to expulsion of most actin filaments from the ring midline (Figure S7D).

Plastin is a crosslinker consisting of two binding heads connected by a 0.012 μm long spring with stiffness 250 pN/μm. If both heads are unbound then either head can bind any actin fiber within 0.01 μm at a binding rate of 10 Hz. While one head is bound, the second head can only bind actin segments closely aligned with the first segment. The maximum allowed angular difference between actin segments is 10 degrees. Bound heads unbind at a rate of 0.05 Hz. While attached to dynamic fibers, binding heads have a 10% chance of rescuing a depolymerizing fiber.

βH-spectrin has two heads like plastin but has an extended flexible body of 0.25 μm and no angular restriction while binding actin fibers. The backbone is modeled as a series of rigid segments of 0.05 μm in length, and a rigidity 3 orders of magnitude lower than that of our actin filaments (7.5∗10−5 pN μm2). At each end of the βH-spectrin fiber there is a binding head with identical binding range, binding rates and rescue behavior as plastin.

We modeled cytokinetic ring assembly similarly to the work of Bidone et al. We randomly placed all 4 components in a 10 μm by 4 μm rectangular space with periodic boundary conditions along the top and bottom edges. Increasing the “circumference” of our space by values higher than 4 μm did not change our results (Figure S7C). The initial horizontal range of our components is about 3 μm since we are concerned with the assembly of the ring itself and not the flux of material from the sides. Our space had background viscosity and temperature typical to those in cells of C. elegans embryos.

To quantify ring formation timing and success rate we created a metric M measuring the spatial distribution of myosin motors. This metric is defined by three calculations. We first created separate histograms of the horizontal and vertical positions of the center of masses of the motors. The exact binning method was unimportant, but the horizontal center bin should coincide with the ring midline since an ideal ring will concentrate all motors’ horizontal position in the exact center. We then found the standard deviation of the bin counts in each direction and normalized by dividing by the largest possible deviation, representing a distribution with all motors in one bin. Since a ring should be localized in the horizontal but uniform in the vertical, we defined our metric as:M ranges from 0 to 1 (Figure S7B). By comparing our metric to videos of simulations, we determined that a ring had successfully assembled if M exceeded a threshold value of 0.5.

To reproduce experimental results, we simulated four different configurations of components. All configurations had actin fibers and myosin motors since those are the minimal components necessary to have a contractile network. We then simulated 100 trials with (1) no crosslinkers, (2) plastin only, (3) βH-spectrin only, and (4) both plastin and βH-spectrin. We determined a final ring success rate by the percentage of successful trials (M > 0.5).

Quantification and statistical analysis

Image quantifications

The total cytokinesis time in Figure 1 was obtained by determining the time elapsed between anaphase onset and the contractile ring reaching a diameter of 5 μm, measured in the z-plane with the widest gap between the two tips of the ingressing cleavage furrow. Anaphase onset was defined by direct visualization of DNA, labeled with H2B::mCherry. Ring constriction rate was the slope of the linear region between ∼75% and 10% furrow ingression.

To quantify the number of actin clusters and myosin puncta on the equatorial cortex in Figure 6C, an intensity threshold was applied (1500 gray values for NMY-2::mCherry and 2500 gray values for LifeAct::GFP) and myosin structures larger than 0.4 μm and with a circularity of 0.5-1 and actin structures larger than 0.3 μm and circularity of 0.3-1 were selected using the Analyze Particles tool (Fiji).

The percentage of cytokinesis success/failure in Figures 1C, 2A, 4C, 4D, 7I, and S2D was assessed by counting the number of one-cell embryos in which furrowing completed/did not complete and dividing it by the total number of embryos filmed.

Deviation of F-actin bundles from vertical alignment at the equatorial cortex in Figure 5C was quantified in maximum projections of cortical planes in embryos expressing LifeAct::GFP (strains GCP22 and GCP927). The Directionality plugin for ImageJ was used in a 30 × 70 pixel box placed over the furrow region, using the local gradient orientation method. The average directionality (in degrees) was determined for each time point and the deviation from vertical alignment (in degrees) was calculated by subtracting the value of the obtained angles from 90°. Absolute numbers were plotted.

Spindle behavior in Figure S3B was determined by recording the distances between the plasma membrane and the anterior/posterior centrosomes, as well as between the two centrosomes over time after anaphase onset. All curves were aligned to the center of the centrosome-centrosome distance. Values were normalized to the length of the embryo at metaphase, which was the distance between the outermost points of the embryo along the anterior-posterior axis at metaphase.

Actin cortical flows

Actin cortical flows were determined in maximum projections of cortical planes in embryos expressing LifeAct::GFP [strains GCP22 and GCP927 with and without sma-1(RNAi)]. Flows were analyzed between anaphase onset and the end of cytokinetic furrow ingression. Magnitude and direction of cortical flows were quantified using a Particle Image Velocimetry (PIV) MATLAB script (see Key resources table). A 216 × 128 pixel (38.5 × 22.8 μm) region was applied to all embryos to exclude cortical curved peripheries. Two iterations were performed to calculate maximum particle displacement: a larger interrogation window of 64 × 64 pixels was used in the first iteration and a smaller interrogation window of 32 × 32 pixels was used in the second iteration. Consecutive interrogation windows overlapped by 50%. After flow field generation, the 216 × 128 pixel region was divided in two adjacent anterior and posterior regions: anterior cortex (1st-148th pixel; 0.0–26.3 μm) and posterior cortex (149th-216th pixel; 26.3–38.4 μm). Each velocity vector was resolved into a Vx component along the anterior-posterior axis and a Vy component along the dorsal-ventral axis. The mean Vx and Vy in the anterior and posterior cortex were estimated for each time point and plotted as a function of time to trace the cortical flow profiles in Figure S4.

FRAP

Quantification of NMY-2::GFP fluorescence signal in FRAP experiments in Figure 6F was performed in three different regions of interest (ROIs): ROI 1 – 10 × 13 pixel box in Figure S6A or 30 × 140 pixel box in Figure S6B, corresponding to the bleached region; ROI 2 - 300 × 200 pixel oval that covers most of the cortex and is used to correct for overall photobleaching of the sample resulting from acquisition and photobleaching sweeps; and ROI 3 – 20 × 20 pixel box, corresponding to the camera background measured outside the embryo. Mean fluorescence intensity in the three ROIs was determined over time, pre- and post-photobleaching. After background subtraction and photobleaching correction, the myosin signal in the photobleached region was normalized to the average signal in that region before photobleaching and plotted over time. Curve fitting with a one-phase association exponential equation was performed to extrapolate the half time of recovery and mobile faction, using Prism 8.0 (GraphPad Software):where y0 is the fluorescence intensity right after photobleaching, Plateau is the fluorescence at infinite times, and K is the rate constant. The values of Plateau and K were obtained after fitting the curves with observed data to one phase association equation.

The half time of recovery was the time when the fluorescence intensity (y) equaled:The mobile fraction, Xm, was calculated using the formula:where ybleached denotes the average fluorescence intensity of 3 time points before photobleaching.

Statistical analysis

Statistical analysis was performed with Prism 8.0 software (GraphPad). Error bars represent 95% confidence interval of the SEM (95% CI), SEM (Figures 6C, S4B, and S4C), or SD (Figures 7F, 7G, and S7C). Statistical significance tests were performed using one-way ANOVA followed by Dunnett or Bonferroni’s multiple comparison test, as indicated in the figure legends.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse monoclonal anti-α-tubulin	Sigma-Aldrich	Cat#T6199; RRID: AB_477583
Rabbit polyclonal anti-SMA-1 ABD	This paper	N/A
Rabbit polyclonal anti-SMA-1 PH	This paper	N/A
Goat anti-rabbit IgG secondary HRP-linked	Jackson ImmunoResearch	Cat#111-035-003; RRID: AB_2313567
Goat anti-mouse IgG secondary HRP-linked	Jackson ImmunoResearch	Cat#115-035-003; RRID: AB_10015289
Donkey anti-rabbit IgG secondary labeled with Alexa Fluor 488	Jackson ImmunoResearch	Cat#711-545-152; RRID: AB_2313584
Donkey anti-mouse IgG secondary labeled with Alexa Fluor 594	Jackson ImmunoResearch	Cat#715-585-150; RRID: AB_2340854
Bacterial and virus strains
E. coli OP50 C. elegans food source	Caenorhabditis Genetics Center	OP50
E. coli HT115 RNAi bacteria (targeting sma-1)	This study	R31.1 (ORF-ID)
E. coli HT115 RNAi bacteria (targeting frm-1)	Source Bioscience (Ahringer library)	ZK270.2 (ORF-ID)
E. coli HT115 RNAi bacteria (targeting unc-44)	Source Bioscience (Ahringer library)	B0350.2 (ORF-ID)
E. coli HT115 RNAi bacteria (targeting perm-1)	Source Bioscience (Ahringer library)	T01H3.4 (ORF-ID)
Chemicals, peptides, and recombinant proteins
Recombinant Cas9 protein	In-house	N/A
Prescission protease	In-house	N/A
Isopropyl b-D-1-thiogalactopyranoside (IPTG)	NZYTech	Cat#MB026
Latrunculin A	Sigma-Aldrich	Cat#L5163
FM4-64	Invitrogen	Cat#T13320
Pierce ECL Western Blotting Substrate	Thermo Fisher Scientific	Cat#32106
Nitrocellulose membrane	Amersham	Cat#GE10600001
Triton X-100	BDH Prolabo	Cat#437002A
Tween	BDH Prolabo	Cat#437082Q
Deposited data
Code to run the simulation is available at https://gitlab.com/JulioMBelmonte/cytokinesis2021	N/A	N/A
Experimental models: Organisms/strains
C. elegans N2 Bristol	Caenorhabditis Genetics Center (CGC)	N2
ItIs157 [pAC16;pie-1/LifeAct::GFP; unc-119 (+)]; ltIs37 [pAA64; pie-1/mCherry::his-58; unc-119 (+)]; unc-119(ed3) III (?)	Our lab	GCP21
ItIs157 [pAC16;pie-1/LifeAct::GFP; unc-119 (+)]; unc-119(ed3) III (?); prtSi2[pAC71;Pnmy-2:nmy-2reencoded::mCherrry::StrepTagII::3′UTR-nmy-2; cb-unc-119(+)]II	Our lab	GCP22
nmy-2(cp13[nmy-2::GFP+LoxP]) I; unc-119(ed3) III (?); ltIs37 [pAA64; pie-1/mCherry::his-58; unc-119 (+)]	Our lab	GCP113
plst-1(prt89) IV	Our lab	GCP564
plst-1(prt89) IV; nmy-2(cp13[nmy-2::GFP+LoxP]) I; unc-119(ed3) III (?); ltIs37 [pAA64; pie-1/mCherry::his-58; unc-119 (+)]	This study	GCP570
atn-1(prt87)V; plst-1(prt89) IV; nmy-2(cp13[nmy-2::GFP +LoxP]) I; unc-119(ed3) III (?); ltIs37 [pAA64; pie-1/mCherry::his-58; unc-119 (+)]	This study	GCP602
plst-1(prt89) IV; ijmSi8[pJD362; Pmex-5::gfp::tbb-2; mCherry::his-11; cb-unc-119(+)] II; unc-119(ed3) III (?)	This study	GCP646
ItSi1188[pDC600; Pmex-5::GFP1-10::3′UTR-tbb-2::gpd-2/3 operon linker::mCherry-his-15::3′UTR-unc-54; cb-unc-119(+)] II; unc-119(ed3) III (this is OD4016 after 6 outcrosses)	This study	GCP794
sma-1ΔPH(prt133)V	This study	GCP808
sma-1ΔPH(prt133) V; nmy-2(cp13[nmy-2::GFP+LoxP]) I; unc-119(ed3) III (?); ltIs37 [pAA64; Ppie-1::mCherry::his-58; unc-119 (+)]	This study	GCP809
plst-1(msn190[plst-1::GFP]) IV; nmy-2(cp52[nmy-2::mKate2+ LoxP unc-119(+) LoxP]I; Itls37 [pAA64; pie-1/mCherry::his-58 +unc-119(+)]; unc-119(ed3) III (?)	This study	GCP831
[plst-1(msn190[plst-1::GFP])] IV; Itls37 [pAA64; pie-1/mCherry::his-58 +unc-119(+); unc-119(ed3) III (?)	This study	GCP832
plst-1(prt89) IV; sma-1ΔPH(prt133) V; nmy-2(cp13[nmy-2::GFP +LoxP]) I; unc-119(ed3) III (?); ltIs37 [pAA64; pie-1mCherry::his-58; unc-119 (+)] IV	This study	GCP838
sma-1ΔABD#1(prt156) V	This study	GCP901
plst-1(prt89) IV; ItIs157 [pAC16; pie-1/LifeAct::GFP; unc-119 (+)]; prtSi2[pAC71;Pnmy-2::nmy-2reencoded::mCherry::StrepTagII::3′UTR-nmy-2; cb-unc-119(+)] II; unc-119(ed3) III (?)	Our lab	GCP927
sma-1ΔSH3(prt163) V	This study	GCP986
sma-1::6xGFP11(prt159) V; ItSi1188 [pDC600; Pmex-5::GFP1-10::3′UTR-tbb-2::gpd-2/3 operon linker::mCherry-his-15::3′UTR-unc-54; cb-unc-119(+)] II; unc-119(ed3) III (?); nmy-2(cp52[nmy-2::mKate2+LoxP unc-119(+) LoxP] I	This study	GCP991
plst-1(prt89) IV; sma-1ΔSH3(prt163) V; nmy-2(cp13[nmy-2::GFP+LoxP]) I; unc-119(ed3) III (?); ltIs37 [pAA64; pie-1/mCherry::his-58; unc-119 (+)] IV	This study	GCP1014
sma-1ΔSH3(prt163) V; unc-119(ed3) III (?); nmy-2(cp13[nmy-2::GFP+LoxP]) I; ltIs37 [pAA64; pie-1/mCherry::his-58; unc-119 (+)] IV	This study	GCP1036
sma-1Δ11SR(prt192) V	This study	GCP1102
sma-1ΔABD#2(prt185) V	This study	GCP1138
ItIs157 [pAC16;pie-1/LifeAct::GFP; unc-119 (+)]; unc-119(ed3) III (?); prtSi2[pAC71;Pnmy-2:nmy-2reencoded::mCherrry::StrepTagII::3′UTR-nmy-2; cb-unc-119(+)] II; sma-1ΔSH3 (prt163) V	This study	GCP1163
plst-1(prt89) IV; ItIs157 [pAC16; pie-1/LifeAct::GFP; unc-119 (+)]; prtSi2[pAC71;Pnmy-2::nmy-2reencoded::mCherry::StrepTagII::3′UTR-nmy-2; cb-unc-119(+)] II; unc-119(ed3) III (?); sma-1ΔSH3 (prt163) V	This study	GCP1164
ItIs157 [pAC16;pie-1/LifeAct::GFP; unc-119 (+)]; unc-119(ed3) III (?); prtSi2[pAC71;Pnmy-2:nmy-2reencoded::mCherrry::StrepTagII::3′UTR-nmy-2; cb-unc-119(+)] II; sma-1ΔPH(prt133) V	This study	GCP1169
plst-1(prt89) IV; ItIs157 [pAC16; pie-1/LifeAct::GFP; unc-119 (+)]; prtSi2[pAC71;Pnmy-2::nmy-2reencoded::mCherry::StrepTagII::3′UTR-nmy-2; cb-unc-119(+)] II; unc-119(ed3) III (?); sma-1ΔPH(prt133) V	This study	GCP1170
ItIs157 [pAC16;pie-1/LifeAct::GFP; unc-119 (+)]; unc-119(ed3) III (?); prtSi2[pAC71;Pnmy-2:nmy-2reencoded::mCherrry::StrepTagII::3′UTR-nmy-2; cb-unc-119(+)] II; sma-1 ΔSR11(prt192)V	This study	GCP1176
plst-1(prt89) IV; ItIs157 [pAC16; pie-1/LifeAct::GFP; unc-119 (+)]; prtSi2[pAC71;Pnmy-2::nmy-2reencoded::mCherry::StrepTagII::3′UTR-nmy-2; cb-unc-119(+)] II; unc-119(ed3)III (?); sma-1 ΔSR11 (prt192) V	This study	GCP1187
sma-1::splitGFP (prt159) V; ItSi1188[ pDC600; Pmex-5::sGFPS1-10::tbb-2_3′UTR::gpd-2/3 operon linker::mCherry-his-15::unc-54-3′UTR; cb-unc-119(+)] II; unc-119(ed3) III (?); plst-1[prt198(plst-1::linker::mCherry)] IV	This study	GCP1207
sma-1::splitGFP(prt159) V; ItSi1188[ pDC600; Pmex-5::sGFPS1-10::tbb-2_3′UTR::gpd-2/3 operon linker::mCherry-his-15::unc-54-3′UTR; cb-unc-119(+)] II; unc-119(ed3) III (?); zbIs2(pie-1::lifeAct::RFP)	This study	GCP1208
plst-1[prt198(plst-1::mCherry)] IV	This study	GCP1209
sma-1(ru18) V	CGC	AZ30
ijmSi8[pJD362; Pmex-5::GFP::tbb-2; mCherry::his-11; cb-unc-119(+)] II; unc-119(ed3) III	Dumont Lab	JDU21
unc-119(ed3) III; ltIs20 [pASM10; pie-1/GFP::unc-59; unc-119 (+)]	OD Lab	OD26
unc-119(ed3) III; ltIs28 [pASM14; pie-1/GFP-TEV-StrepTagII::ANI-1; unc-119 (+)]; ltIs37 [pAA64; pie-1/mCherry::his-58; unc-119 (+)]	OD Lab	OD130
ItSi1188 [pDC600; Pmex-5::GFP1-10::3′UTR-tbb-2::gpd-2/3 operon linker::mCherry::his-15::3′UTR-unc-54; cb-unc-119(+)] II; unc-119(ed3) III	OD Lab	OD4016
spc-1(cas815[spc-1::GFP]) X	Ou Lab	GOU2936
unc-70(cas962[GFP::unc-70]) V	Ou Lab	GOU3103
plst-1(msn190[plst-1::GFP]) IV	Zaidel-Bar Lab	RZB213
plst-1(msn190[plst-1::gfp]) IV; zbIs2(pie-1::LifeAct::RFP)	Zaidel-Bar Lab	RZB217
erm-1(mib15[erm-1::eGFP]) I	Boxem Lab	BOX213
Oligonucleotides
See Table S2	N/A	N/A
Recombinant DNA
Expression vector L4440	Bioscience/ Ahringer library	L4440
Plasmid pJA58	Fire Lab/ Addgene	CAS 59933
Plasmid pAC571	This study	pAC571
Plasmid pDC592	Cheerambathur Lab	pDC592
Plasmid pDD162	Goldstein Lab	pDD162
Expression vector for sma-1 ABD - pGEX-6P1_GST_TEV_sma-1_ABD (see STAR Methods for details)	This study	pAC543
Expression vector for sma-1 PH - pGEX-6P1_GST_TEV_sma-1_PH (see STAR Methods for details)	This study	pAC553
Software and algorithms
ImageJ (Fiji)	Schindelin et al.60	https://imagej.net/software/fiji/; RRID: SCR_003070
GraphPad Prism	GraphPad Software	http://www.graphpad.com/; RRID: SCR_002798
Cytosim	Nédélec & Foethke61	https://gitlab.com/JulioMBelmonte/cytosim
MagicPlotPro	Magicplot Systems, LLC	https://magicplot.com
MATLAB / Particle Image Velocimetry (PIV)	MathWorks	https://openresearchsoftware.metajnl.com/articles/10.5334/jors.bl/; RRID:SCR_001622