Kinetochore components are required for central spindle assembly

Abstract

A critical structure poised to coordinate chromosome segregation with division plane specification is the central spindle that forms between separating chromosomes after anaphase onset^{1, 2}. The central spindle acts as a signaling center that concentrates proteins essential for division plane specification and contractile ring constriction³. However, the molecular mechanisms that control the initial stages of central spindle assembly remain elusive. Using Caenorhabditis elegans zygotes, we found that the microtubule bundling protein SPD-1^PRC1 and the motor ZEN-4^MKLP-1 are required for proper central spindle structure during its elongation^4-9. By contrast, we found that the kinetochore controls the initiation of central spindle assembly. Specifically, central spindle microtubule assembly is dependent upon kinetochore recruitment of the scaffold protein KNL-1, as well as downstream partners BUB-1, HCP-1/2^CENP-F, and CLS-2^CLASP; and is negatively regulated by kinetochore-associated protein phosphatase 1 (PP1) activity. This in turn promotes central spindle localization of CLS-2^CLASP and initial central spindle microtubule assembly through its microtubule polymerase activity. Together, our results reveal an unexpected role for a conserved kinetochore protein network in coupling two critical events of cell division: chromosome segregation and cytokinesis.

The central spindle coordinates accurate chromosome segregation with division plane specification during cytokinesis, by acting as a hub for the localization of key regulators of these processes. Little is known about the initial stages of central spindle assembly and in particular how de novo central spindle microtubules assemble. In HeLa cells, de novo microtubule generation required for central spindle assembly is mediated in part by the Augmin complex, which promotes non-centrosomal microtubule nucleation during anaphase ². Importantly, this protein complex is not present in every species and no ortholog of any subunit has been identified in C. elegans. While the conserved microtubule bundling protein SPD-1^PRC1 and the kinesin-6 motor ZEN-4^MKLP1 are clearly required for proper central spindle structure, neither of these proteins appear to have microtubule nucleating or assembly promoting activity ^10-12.

To study the initiation of central spindle assembly, we first developed an assay to quantify both central spindle microtubule density over time and overall central spindle mechanical integrity (see below) in the one-cell C. elegans embryo using both GFP-tagged TBB-2^β-Tubulin and AIR-2^AuroraB, a component of the chromosomal passenger complex that strongly localizes to the central spindle in anaphase ^{13, 14}. We combined semi-automated GFP∷TBB-2^β-Tubulin and GFP∷AIR-2^AuroraB intensity quantification over time with kymograph analysis of mCherry∷HistoneH2B to track sister chromatid dynamics over time (Fig. 1a, b, c and Supplementary Fig. 1a, b).

Figure 1. SPD-1^PRC1 and ZEN-4^MKLP1 are required for central spindle stabilization and bundling.

(a) Central spindle assembly in the indicated strains. Timings are relative to anaphase onset. (b) Kymographic assay for central spindle mechanical integrity analysis. (c) Functional assay for central spindle density analysis. (d) Schematics of SPD-1^PRC1 and ZEN-4^MKLP1 binding to overlapping microtubule plus-ends. (e) Kymographs for the indicated conditions. (f) Chromosome to chromosome distance after anaphase onset for the indicated conditions. The sample size (number of embryos analyzed) is provided in the figure and was generated by aggregation over 3 independent experiments. Error bars represent the SEM. (g) Percentage of one-cell embryos that displayed a breaking apart central spindle in the indicated conditions. (h) Central spindle assembly and phenotypes for the indicated conditions. The white arrows indicate the presence of remaining microtubules connected to each set of separating sister chromatids and that initially formed at the central spindle before it broke apart. (i) Quantification of central spindle intensity after spd-1^PRC1(RNAi) (GFP∷β-Tub, p<0.0001; GFP∷AIR-2, p<0.0001) and zen-4^MKLP1(RNAi) (GFP∷β-Tub, p<0.0001; GFP∷AIR-2, p<0.0001). The mean is shown for n=31, 11 and 21 embryos for control, spd-1(RNAi) and zen-4(RNAi), respectively. Data was aggregated over 3 independent experiments. One-way ANOVA was used to determine significance. (j) Central spindle assembly and phenotypes for the indicated conditions. (k) Quantification of central spindle intensity after gpr-1/2+spd-1^PRC1(RNAi) (GFP∷β-Tub, p=0.0750; GFP∷AIR-2, p=0.0103) and gpr-1/2+zen-4^MKLP1(RNAi) (GFP∷β-Tub, p=0.1578; GFP∷AIR-2, p=0.1047). The mean is shown for n=31, 20 and 16 embryos for gpr-1/2(RNAi), gpr-1/2+spd-1(RNAi) and gpr-1/2+zen-4(RNAi), respectively. Data was aggregated over 3 independent experiments. One-way ANOVA was used to determine significance. Error bars

Central spindle integrity has been attributed to the conserved microtubule bundling protein SPD-1^PRC1 and the kinesin-6 motor ZEN-4^MKLP1 (Fig. 1d). To determine if these proteins also play a role in the initiation of central spindle assembly, we monitored the assembly kinetics and mechanical integrity of the central spindle following RNAi mediated depletion. As expected, SPD-1^PRC1 or ZEN-4^MKLP1 depletion led to central spindle disruption and faster sister chromatid separation (Fig. 1e-i and Supplementary Video 1). Yet, in both SPD-1^PRC1 and ZEN-4^MKLP1 depleted embryos, a central spindle initially assembled but subsequently broke down into two half- or hemi-spindles connected to each set of rapidly separating chromatids (Fig. 1e, h). This rapid sister chromatid separation represents active cortical pulling forces acting on astral microtubules; forces that are normally counteracted by the central spindle, which slows down sister chromatid separation and limits the extent of segregation ¹⁵. Thus faster sister chromatid separation reveals reduced central spindle mechanical integrity. To further demonstrate this, we repeated these experiments in the presence of reduced cortical pulling forces upon GPR-1 and -2 depletion. GPR-1/2 are required for cortically anchoring the minus end-directed motor Dynein, which in turn generates strong pulling forces on spindle poles. Strikingly, central spindle integrity was almost fully rescued in SPD-1^PRC1 and ZEN-4^MKLP1 depleted embryos when cortical pulling forces are reduced (Fig. 1j, k, Supplementary Fig. 1c and Video 2). These results suggest that SPD-1^PRC1 and ZEN-4^MKLP1 are not required for promoting initial central spindle microtubule assembly but are essential for proper central spindle mechanical integrity during its elongation likely due to their role in microtubule crosslinking ^{1, 11, 12}.

To identify components involved in initial anaphase central spindle assembly, we analyzed the potential contribution of the kinetochore, which participates in pre-anaphase spindle formation in most systems and has been associated with microtubule polymerizing activity ¹⁶. To avoid the total failure of chromosome segregation that arises from full kinetochore inhibition following depletion of the core kinetochore scaffold protein KNL-1, we focused on the three sub-complexes/branches recruited downstream of KNL-1: RZZ/dynein (Rod, ZW10, Zwilch), Ndc80, and BUB-1 (Fig. 2a) ¹⁷. To probe the roles of the RZZ/dynein and Ndc80 sub-complexes, the two main microtubule-binding entities at the kinetochore, we targeted ZWL-1^ZWILCH and NDC-80, respectively. To probe the role of the BUB-1 branch, which is involved in the spindle assembly checkpoint signaling and recruits microtubule associated proteins that control kinetochore microtubule dynamics (e.g CLS-2^CLASP), we targeted BUB-1 itself ¹⁸. Inhibition of each sub-complex, through RNAi-mediated depletion of NDC-80, ZWL-1^ZWILCH, or BUB-1, led to chromosome mis-segregation events with frequent lagging chromosomes and anaphase chromatin bridges, as previously described (Supplementary Table 1) ^19-21. Interestingly, central spindle assembly was also dysregulated in all three RNAi conditions, demonstrating a post-anaphase role for kinetochores in central spindle formation and organization (Fig. 2b, c and Supplementary Video 3). ZWL-1^ZWILCH-depleted embryos had reduced GFP∷TBB-2^β-Tubulin and GFP∷AIR-2^AuroraB in the central spindle region but showed normal sister chromatid separation kinetics, suggesting that central spindle mechanical integrity was not significantly affected (Fig. 2c-e). Zygotes treated with ndc-80(RNAi) exhibited elevated GFP∷AIR-2^AuroraB concentration on a smaller central spindle, likely due to reduced sister chromatid segregation caused by defective chromosome-microtubule end-on attachment (Fig. 2b, c) ²². In contrast, the strongest phenotype was observed following BUB-1 depletion. First, bub-1(RNAi) zygotes had significantly faster and more extensive sister chromatid separation, suggesting that the mechanical integrity of the central spindle was highly impaired (Fig. 2d, e and Supplementary Table 1). Second, bub-1(RNAi) zygotes had lower levels of central spindle-associated GFP∷TBB-2^β-Tubulin and GFP∷AIR-2^AuroraB combined with disorganized and disrupted central spindles (Fig. 2c, f and Supplementary Table 1). Thus bub-1(RNAi) zygotes displayed both of the defects expected from compromised central spindle assembly.

Figure 2. A subset of kinetochore proteins are involved in central spindle assembly.

(a) Schematics of kinetochore sub-complexes analyzed. (b) Kymographs for the indicated conditions. (c) Chromosome to chromosome distance after anaphase onset for the indicated conditions. The sample size (number of embryos analyzed) is provided in the figure and was generated by aggregation over 3 independent experiments. Error bars represent the SEM. (d) Percentage of one-cell embryos that displayed a breaking apart central spindle in the indicated conditions. (e) Central spindle assembly and phenotypes for the indicated conditions. (f) Quantification of central spindle intensity after ndc-80(RNAi) (GFP∷β-Tub, p<0.0001; GFP∷AIR-2, p=0,7065), zwl-1 ^ZWILCH(RNAi) (GFP∷β-Tub, p<0.0001; GFP∷AIR-2, p=0.0417) and bub-1(RNAi) (GFP∷β-Tub, p<0.0001; GFP∷AIR-2, p<0.0001). The mean is shown for n=31, 13, 33 and 35 embryos for control, ndc-80(RNAi), zwl-1(RNAi) and bub-1(RNAi), respectively. Data was aggregated over 3 independent experiments. One-way ANOVA was used to determine significance. Error bars represent the SEM. Scale bars, 5 μm.

To further investigate the contribution of the BUB-1 branch to the initiation of central spindle assembly, we set out to determine whether proteins downstream of BUB-1 localize to the central spindle. During late prometaphase, BUB-1 kinase recruits two redundant CENP-F orthologs HCP-1 and -2 to the kinetochore, which in turn recruit the microtubule assembly regulator CLS-2^CLASP (Fig. 3a) ^{18, 23}. Interestingly, CLASP proteins have already been implicated in central spindle integrity in other systems ^{24, 25}. Immuno-localization of endogenous proteins and live imaging of GFP-tagged BUB-1, HCP-1^CENP-F and CLS-2^CLASP revealed that all three components were transiently localized to the central spindle region shortly after anaphase onset (Fig. 3b, c). HCP-1^CENP-F and CLS-2^CLASP were also detected in the spindle region throughout mitosis and CLS-2^CLASP concentrated at spindle poles. We further found that the central spindle localization of CLS-2^CLASP during anaphase, but not its spindle pole targeting, was dependent on the presence of BUB-1 (Fig. 3c, Supplementary Fig. 2a). We next tested the functional significance of HCP-1/2^CENP-F and CLS-2^CLASP localization on the central spindle. Full depletion of HCP-1/2^CENP-F or CLS-2^CLASP led to a failure in chromosome biorientation coupled to premature spindle pole elongation and sister chromatid co-segregation to the same spindle pole (Fig. Supplementary Fig. 2b, c, Video 4 and Supplementary Table 1) ¹⁸. In these embryos, a complete failure of central spindle assembly was observed (Supplementary Fig. 2c). Taken together, these results suggest that the BUB-1 kinetochore branch components relocalize to the central spindle during anaphase and are required for its initial assembly.

Figure 3. The BUB-1/HCP-1/2^CENPF/CLS-2^CLASP pathway is essential for initiating central spindle formation.

(a) Schematics of components of the BUB-1 kinetochore branch. (b) Fixed control embryos were stained to visualize DNA, α-Tubulin, BUB-1, HCP-1/2^CENP-F, CLS-2^CLASP and KNL-1 in anaphase. (c) Fluorescence intensity for indicated GFP-tagged proteins along the central spindle at anaphase onset (black dots), and 20 or 50 seconds after anaphase onset (respectively dark and light green dots). 0 μm corresponds to the position of chromosomes at anaphase onset. Black arrows indicate the spindle pole localization of CLS-2^CLASP. The sample size (number of embryos analyzed) is provided in the figure and was generated by aggregation over 2 independent experiments. Error bars represent the SEM. (d) Kymographs for the indicated conditions. (e) Chromosome to chromosome distance after anaphase onset for the indicated conditions. Chromosome to chromosome distance after anaphase onset for the indicated conditions. The sample size (number of embryos analyzed) is provided in the figure and was generated by aggregation over 3 independent experiments. Error bars represent the SEM. (f) Percentage of one-cell embryos that displayed a breaking apart central spindle in the indicated conditions. (g) Central spindle assembly and phenotypes for the indicated conditions. (h) Quantification of central spindle intensity after cls-2^CLASP(RNAi) (GFP∷β-Tub, p=<0.0001; GFP∷AIR-2, p<0.0001) and hcp-1/2^CENP-F(RNAi) (GFP∷β-Tub, p=<0.0001; GFP∷AIR-2, p<0.0001). The mean is shown for n=31, 17 and 13 embryos for control, cls-2(RNAi) and hcp-1/2 (RNAi), respectively. Data was aggregated over 3 independent experiments. One-way ANOVA was used to determine significance. Error bars represent the SEM. (i) Central spindle assembly and phenotypes for the indicated conditions. (j) Quantification of central spindle intensity after gpr-1/2+cls-2^CLASP(RNAi) (GFP∷β-Tub, p<0.0001; GFP∷AIR-2, p<0.0001). The mean is shown for n=31 and 17 embryos for gpr-1/2(RNAi) and gpr-1/2+cls-2(RNAi) respectively. Data was aggregated over 3 independent experiments. One-way ANOVA was used to determine significance. Error bars represent the SEM. Scale bars, 5 μm.

An alternative hypothesis is that failure of initial central spindle assembly could be due to the sister chromatid co-segregation to the same spindle pole observed in these embryos. Separation of sister chromatids to opposing spindle poles may be required for precise equatorial translocation of central spindle assembly factors between the chromosomes. To bypass any chromosome co-segregation defects, we performed a temporal dilution experiment by combining RNAi against HCP-1/2^CENP-F or CLS-2^CLASP with analysis of the first embryonic division after decreasing times following dsRNA injection. We identified a narrow temporal window during which sister chromatid alignment and co-segregation defects could be uncoupled from central spindle assembly in embryos with reduced levels of HCP-1/2^CENP-F or CLS-2^CLASP (Supplementary Fig. 2b, c and Video 4). Under these specific conditions, chromosomes aligned and segregated to opposite poles normally. However, the extent of sister chromatid and spindle pole segregation was still significantly greater than in control embryos, indicative of a loss of central spindle mechanical integrity (Fig. 3d-f, Supplementary Table 1 and Video 4 and 5). We also found microtubules and GFP∷AIR-2^AuroraB were absent from the central spindle region, indicative of a loss of central spindle assembly (Fig. 3g, h). These results not only demonstrated the crucial contribution of HCP-1/2^CENP-F and CLS-2^CLASP in central spindle microtubule assembly, they also highlight that this function requires higher protein levels than required for their role in chromosome alignment and separation to opposite spindle poles. Furthermore, in these embryos, and in contrast to SPD-1^PRC1 or ZEN-4^MKLP-1 depletion, the lack of a central spindle was evident immediately following anaphase onset and no hemi-spindle was visible (Fig. 1e, h vs 3d, g). Consistent with this, reducing cortical pulling forces by GPR-1/2 depletion in these embryos did not rescue central spindle assembly (Fig. 3i, j and Supplementary Video 2). Thus these results show that HCP-1/2^CENP-F and CLS-2^CLASP are essential for initiating central spindle microtubule assembly and suggests that their levels must be tightly controlled to avoid central spindle assembly defects.

We next tested if the initial metaphase kinetochore localization of these proteins is essential for the initiation of central spindle microtubule assembly during anaphase. Metaphase kinetochore targeting of BUB-1 requires the N-terminal half of KNL-1 via a series of eight phosphorylated MELT repeats ^26-28. Expression of an RNAi-resistant N-terminal truncation of KNL-1 (Δ85-505), that removes all eight MELTs, prevents kinetochore localization of BUB-1 when endogenous KNL-1 is absent (Fig. 4a) ²⁹. This truncation specifically disrupted the BUB-1 branch and did not lead to a complete disruption of the kinetochore, as GFP-tagged MIS-12 (Mis12 complex), KBP-4^Spc24 (Ndc80 complex) and CZW-1^ZW10 (RZZ complex) all localized to the kinetochores, while BUB-1, HCP-1^CENP-F and CLS-2^CLASP did not (Fig. 4b, c, Supplementary Fig. 3 and 4). This mutant thus uncouples the recruitment of the three major kinetochore sub-complexes/branches downstream of KNL-1. Mutant embryos expressing KNL-1 Δ85-505 exhibited the same central spindle assembly defects as seen following depletion of BUB-1: sister chromatid separation was faster than in controls, and GFP∷TBB-2^β-Tubulin and GFP∷AIR-2^AuroraB intensities were significantly decreased in the central spindle region (Fig. 4d-h). Additionally, the central spindle was completely disorganized and even broke apart (Fig. 4f, Supplementary Table 1 and Video 6). These results demonstrate that proper assembly of the kinetochore via KNL-1-dependent recruitment of BUB-1, HCP-1/2^CENP-F and CLS-2^CLASP during metaphase is required for the initiation of central spindle assembly after anaphase onset.

Figure 4. PP1-modulated KNL-1-dependent recruitment of BUB-1/HCP-1/2^CENP-F/CLS-2^CLASP controls central spindle assembly.

(a) Schematics of mutations engineered in KNL-1. (b) Fluorescence intensity of GFP-tagged CLS-2 along the central spindle at anaphase onset (black dots), and 20 or 50 seconds after anaphase onset (respectively dark and light green dots) in indicated KNL-1 mutants. All experiments were performed in absence of endogenous KNL-1. 0 μm corresponds to the position of chromosomes at anaphase onset. Black arrows indicate the spindle pole localization of CLS-2^CLASP. The sample size (number of embryos analyzed) is provided in the figure and was generated by aggregation over 2 independent experiments. Error bars represent the SEM. (c) Quantification of average CLS-2^CLASP fluorescence intensity centred on chromosome position at anaphase onset. Each value is normalized against the WT average intensity at 0 s. Error bars represent the SEM. (d) Kymographs for the indicated conditions. (e) Chromosome to chromosome distance after anaphase onset for the indicated conditions. The sample size (number of embryos analyzed) is provided in the figure and was generated by aggregation over 3 independent experiments. Error bars represent the SEM. (f) Percentage of one-cell embryos that displayed a breaking apart central spindle in the indicated conditions. (g) Central spindle assembly and phenotypes for the indicated conditions. (h) Quantification of central spindle intensity in KNL-1 A85-505 (GFP∷β-Tub, p<0.0001; GFP∷AIR-2, p=0.0002) and KNL-1 RRASA (GFP∷β-Tub, p<0.0001; GFP∷AIR-2, p=0.0003). The mean is shown for n=20, 22 and 21 embryos for KNL-1 WT, KNL-1 Δ85-505 and KNL-1 RRASA respectively. Data was aggregated over 3 independent experiments. One-way ANOVA was used to determine significance. (i) Central spindle assembly and phenotypes for the indicated conditions. (j) Chromosome to chromosome distance after anaphase onset for the indicated conditions. The sample size (number of embryos analyzed) is provided in the figure and was generated by aggregation over 3 independent experiments. Error bars represent the SEM. Scale bars, 5 μm.

As reducing the overall levels of HCP-1/2^CENP-F or CLS-2^CLASP or preventing their kinetochore targeting reduced central spindle microtubule assembly and organization, we set out to test if increasing the level of these proteins at the kinetochores would likewise increase central spindle robustness. To test this, we used a KNL-1 mutant expected to recruit excess BUB-1 to the kinetochores. At the very N-terminus of KNL-1, an RRVSF motif acts as a kinetochore-docking site for protein phosphatase 1 (PP1), which in turn dephosphorylates the eight MELT repeats required for BUB-1 kinetochore localization (Fig. 4a) ^{30, 31}. As expected, we found that expression of an RNAi-resistant KNL-1 RRASA that prevents PP1 docking following endogenous KNL-1 depletion resulted in a two-fold increase in the level of kinetochore BUB-1∷GFP compared to controls (Supplementary Fig. 4a-c). This increase in BUB-1 at the metaphase kinetochores led to persistent localization of GFP∷HCP-1^CENP-F and CLS-2^CLASP∷GFP on the central spindle compared to control embryos (Fig. 4b, c and Supplementary Fig. 4d-h). Correspondingly, the kinetics of sister chromatid separation were slower in embryos expressing the KNL-1 RRASA mutant and this correlated with a significant increase in GFP∷TBB-2^β-Tubulin and GFP∷AIR-2^AuroraB intensities in the central spindle region (Fig. 4d-h, Supplementary Table 1 and Video 6). Thus increasing the levels of BUB-1 on the kinetochores early in mitosis leads to persistent HCP-1^CENP-F and CLS-2^CLASP in the central spindle region, which initiates the assembly of a more robust central spindle in anaphase.

To test if this increase in central spindle intensity translated into improved mechanical integrity, we challenged the central spindle by enhancing astral microtubule-mediated cortical pulling forces. To do this, we depleted the kinesin-13 microtubule depolymerase KLP-7^MCAK, which leads to a dramatic increase in astral microtubule numbers and a corresponding increase in astral microtubule pulling forces ^{15, 32}. As expected, half of the central spindles that assembled in absence of KLP-7^MCAK broke apart and sister chromatids separated at a faster rate (Fig. 4f, i, j and Supplementary Video 7). Strikingly, both phenotypes were rescued in the KNL-1 RRASA mutant, consistent with these embryos having more robust central spindle assembly able to withstand abnormally elevated cortical pulling forces (Fig. 4i, j and Supplementary Video 7).

We next set out to determine how the kinetochore-dependent pathway promotes central spindle microtubule assembly. Secondary structure analysis of CLS-2^CLASP revealed three TOGL (Tumor Over-expressed Gene Like) domains characteristic of proteins with microtubule assembly-promoting activity and involved in microtubule or tubulin heterodimer binding (Fig. 5a) ³³. To test if CLS-2^CLASP TOGL domains are functional, we performed an in vitro microtubule assembly assay in the presence of increasing concentration of recombinant CLS-2^CLASP purified from insect cells (Supplementary Fig. 5a). Increasing the concentration of recombinant CLS-2^CLASP displayed all three characteristics expected for a protein with microtubule polymerase activity: 1) shortening of the initial microtubule nucleation lag phase, 2) acceleration of microtubule assembly and 3) an increase of the steady-state plateau (Fig. 5b). To test if CLS-2^CLASP activity is required for central spindle assembly, we generated an RNAi-resistant CLS-2^CLASP mutant (CLS-2^CLASP W57A-K177A-R224A named hereafter CLS-2^CLASP 3A) impaired in TOGL1 tubulin heterodimer binding (Fig. 5a). Mutating these three residues is sufficient to dramatically reduce microtubule polymerase activity of TOGL domain-containing proteins without disrupting overall protein folding ³⁴. Strikingly, expression of the CLS-2^CLASP 3A mutant following endogenous CLS-2^CLASP partial depletion severely disrupted central spindle assembly and mechanical integrity (Fig. 5c-f, Supplementary Fig. 2d and Video 8). Altogether, these results demonstrate that kinetochores play an active role in promoting the initial assembly of a normal central spindle via PP1-modulated KNL-1-dependent recruitment of BUB-1, HCP-1/2^CENP-F and CLS-2^CLASP in metaphase. This early phase is required for their subsequent translocation in the central spindle region at anaphase onset, which in turn promotes initial central spindle microtubule assembly through CLS-2^CLASP activity.

Figure 5. CLS-2^CLASP activity is required for central spindle assembly and a two-stage model of central spindle organization.

(a) Schematics of CLS-2^CLASP domain organization and engineered mutations. (b) Microtubule polymerization timecourses with indicated concentrations of CLS-2^CLASP∷6xHIS. Data shown represent one out of 3 independent experiments. (c) Central spindle assembly and phenotypes for the indicated conditions. (d) Quantification of central spindle intensity in CLS-2^CLASP WT and CLS-2^CLASP 3A (p<0.0001). The mean is shown for n=10 embryos in each condition. Data was aggregated over 3 independent experiments. A Student t-test was used to determine significance. (e) Percentage of one-cell embryos that displayed a breaking apart central spindle in the indicated conditions. (f) Chromosome to chromosome distance after anaphase onset for the indicated conditions. The sample size (number of embryos analyzed) is provided in the figure and was generated by aggregation over 3 independent experiments. Error bars represent the SEM. (g) GFP-tagged EBP-1^EB1 localization at the central spindle in the indicated conditions. (h) Quantification of EBP-1^EB1 intensity at the central spindle intensity in KNL-1 WT, KNL-1 Δ85-505 (p<0.0001), KNL-1 RRASA (p<0.0001) and cls-2^CLASP(RNAi) (p<0.0001). The mean is shown for n=13, 13, 10 and 10 embryos for KNL-1 WT, KNL-1 Δ85-505, KNL-1 RRASA and cls-2(RNAi) respectively. Data was aggregated over 3 independent experiments. One-way ANOVA was used to determine significance. (i) Quantification of average (CLS-2^CLASP and SPD-1^PRC1, between 10 and 80 sec after anaphase onset) or total (β-tubulin and EBP-1^EB1, between 8 and 80 sec after anaphase onset) fluorescence intensity at the central spindle over time. The mean is shown for n=13, 12, 13 and 11 embryos for GFP∷β-Tub, CLS-2∷GFP, SPD-1∷GFP and EBP-1∷GFP respectively. Data was aggregated over 2 independent experiments. Error bars represent the SEM. (j) A two-stage model of central spindle organization. BUB-1, HCP-1/2^CENP-F and CLS-2^CLASP are concentrated at the kinetochore during metaphase in a KNL-1-dependent manner opposed by PP1 activity. This concentration allows their timely translocation in the central spindle region at anaphase onset. Central spindle localized CLS-2^CLASP promotes microtubule formation via its microtubule assembly promoting activity. As CLS-2^CLASP leaves the central spindle region, SPD-1^PRC1 and ZEN-4^MKLP1 accumulate and, through their microtubule cross-linking activity, stabilize the central spindle during its elongation. Scale bars, 5 μm.

Here we have identified a new unexpected role for the kinetochore in driving central spindle assembly. We propose a model in which the central spindle is formed by a two-stage process (Fig. 5j). In the first stage, the newly identified kinetochore pathway promotes the initiation of de novo microtubule assembly between the segregating sister chromatids via CLS-2^CLASP activity. In the second stage, the microtubule bundling proteins SPD-1^PRC1 and ZEN-4^MKLP1 promote central spindle elongation and proper microtubule organization. In support of this model, GFP∷EBP-1^EB1 imaging revealed an initial burst of microtubule assembly in the central spindle region within seconds after anaphase onset, which was associated with peak levels of CLS-2^CLASP in the region (Fig. 5g-I, Supplementary Fig. 5b and Video 9 and 10). Accordingly, the GFP∷EBP-1^EB1 burst was abrogated by CLS-2^CLASP depletion or expression of KNL-1(Δ85-505), and increased following KNL-1 RRASA expression (Fig. 5g, h). We also found SPD-1^PRC1 was recruited to the central spindle later, after the initial structure was formed, consistent with a role in the second stage and not the first stage of central spindle assembly (Fig. 5i and Supplementary Fig. 5b). Moreover, we found that in C. elegans zygotes, CLS-2^CLASP recruitment to the central spindle was independent of SPD-1^PRC1 (Fig. 5c, d) ²⁵. Finally, we observed hemi-spindles connected to the segregating chromatids following SPD-1^PRC1 or ZEN-4^MKLP1 RNAi but not in BUB-1, HCP-1/2^CENP-F, or CLS-2^CLASP depleted embryos (Fig. 1e, h and 3d, g). Thus we have uncovered a novel role for a subset of kinetochore proteins in promoting and regulating the initiation of central spindle assembly, and potentially coupling the two most critical events of cell division: chromosome segregation and cytokinesis (Fig. 5j). Because kinetochore components and the central spindle are highly conserved throughout metazoans, the mechanisms elucidated here are likely to be conserved in vertebrates.