Introduction

Microtubules are semi-rigid polymers that form a crucial component of the cell cytoskeleton alongside actin and intermediate filaments. In coordination with molecular motors and microtubule-associated proteins (MAPs), microtubules perform essential functions, including cell shaping [1,2], intracellular trafficking [3], and cell migration [4]. Microtubules also play major roles in cell division; for example, spindle microtubules assemble into the mitotic spindle and segregate chromosomes, while astral microtubules help position the spindle [5–11]. Along a structural line, each microtubule comprises protofilaments that are linear chains of tubulin dimers added or removed dynamically. Microtubule dynamic instability, characterised by stochastic transitions between growing and shrinking phases, is regulated in vivo by numerous MAPs [12–14]. The mechanical properties of microtubules, particularly their flexural rigidity—i.e., their resistance to bending—have received comparatively less attention. Notably, microtubules exhibit the highest flexural rigidity among cytoskeletal filaments [15]. While in vitro reconstituted microtubules are bent at the scale of a few millimetres [15–18], in vivo observations revealed that microtubules appear bent at the micrometre scale [19–21]. On one hand, high microtubule rigidity is necessary for efficient intracellular trafficking [22], for generating pushing forces against the cell periphery that contribute to spindle or nucleus positioning [23–30], or for exerting pushing forces on chromosome arms to produce the polar ejection force with the assistance of chromokinesins [31,32]. On the other hand, microtubules must also be flexible enough to bend, for example when growing through the crowded cytoplasm [33]. This highlights the need to regulate microtubule flexural rigidity in vivo to support its various biological functions [21].

Research across various biological systems has identified three major protein families involved in regulating microtubule rigidity: the Doublecortin family (including DCX and DCLK1), the MAP2/MAP4/Tau superfamily, and the SPD-1/Ase1/PRC1/MAP65 family [16,18,34–36]. While the latter family plays crucial roles in non-neuronal contexts, the first two have been primarily studied in neuronal systems. For instance, the simultaneous knockdown of DCX and DCLK1 in rat neurons led to more curved microtubules at the growing axon tip [37]. Doublecortin-family proteins associate with microtubules through their doublecortin domains. Notably, these proteins bind between the microtubule protofilaments at the vertices of four tubulin dimers, contributing to the lateral coupling of adjacent protofilaments [38,39]. This binding likely restricts protofilament sliding, thereby increasing microtubule rigidity [40,41]. Along a similar line, Tau-deficient axons have been shown to exhibit curled microtubules [42]. Tau family proteins bind longitudinally along the outer ridges of microtubule protofilaments. It would strengthen the longitudinal contacts between tubulin heterodimers within a protofilament, straightening the protofilaments [34,43,44]. In contrast to Doublecortin and Tau proteins, MAP65/Ase1, by binding to individual microtubules and microtubule bundles, decreases their flexural rigidity [36]. Upon cross-linking antiparallel microtubules, proteins of this family favour the formation and stabilisation of microtubule bundles [36,45–47]. Importantly, unlike the SPD-1/Ase1/PRC1/MAP65 family, both the Doublecortin and Tau families can also regulate microtubule dynamics, making it challenging to disentangle their roles in mechanical versus dynamic regulation. Some in vivo studies indicate that these proteins stabilise microtubules and/or promote their growth [42,48]. They may reduce their dynamics by decreasing catastrophe frequency, depolymerisation and nucleation rates [48–50]. While the impacts of controlling microtubule rigidity have primarily been studied in neurons, microtubule-rigidity contribution during cell division also warrants investigations.

Previous studies on cell division involving the targeting of proteins from the Doublecortin, Tau, and PRC1 families revealed several defects. These included mitotic spindle mispositioning, abnormalities in spindle formation and integrity, and impaired cell cycle progression [51–60]. During neurogenesis, altered expression of Doublecortin-family proteins led to incorrect cell-fate determination for neural progenitors [51,52]. Additionally, the deregulation of Tau (whether through mutation or overexpression) resulted in mitotic defects, such as issues with spindle formation, aneuploidy and chromosome misalignment [61,62]. Proteins from the PRC1 family were also suggested to regulate mitotic spindle mechanics [45,47]. Thus, targeting proteins from these three families can lead to a wide range of meiotic and mitotic disturbances, although the underlying mechanisms remain unclear. Proposed explanations include perturbations in microtubule stability and in molecular motor activity or recruitment [51–56,63]. Whether the underlying mechanism could involve microtubule rigidity is yet to be investigated.

In the Caenorhabditis elegans zygote, the pronuclei-centrosome complex (NCC) and the mitotic spindle exhibit stereotyped movements during cell division [64]. It makes this system a well-established model for studying asymmetric divisions. In these divisions, the correct positioning and orientation of the spindle are critical for cell fate determination and, more broadly, for the organism’s development [9]. In further detail, the NCC forms at the zygote posterior pole, migrates anteriorly, and the centrosomes align with the anteroposterior axis during prophase. The mitotic spindle assembles near the cell centre after the nuclear envelop breakdown (NEBD), remains centred during metaphase, and is then positioned posteriorly, mainly during anaphase [65]. During this posterior displacement, the spindle undergoes transverse oscillations driven by strong cortical pulling forces [66]. While not essential for asymmetric division, these oscillations have been instrumental in studying spindle positioning forces [67–72]. Astral microtubules, which emanate from the centrosomes, play a central role in spindle movements. These microtubules can transmit or generate three types of forces [5,7,73]: first, cortical pulling forces mediated by dynein motors anchored at the cell cortex interacting with shrinking microtubules [67,74,75]; second, cytoplasmic pulling forces that are reaction forces generated by organelle movement along microtubules, driven by dynein motors [76]; and third, cortical pushing forces generated by microtubules that grow against the cell cortex [25,27,77]. Various models have addressed the roles of these forces. Kimura et al. proposed that cytoplasmic pulling forces contribute to NCC centration [76], while others have suggested that these forces do not aid in maintaining spindle position at the cell centre during metaphase [25] or in driving anaphase spindle oscillations [68]. Cortical pulling forces, which are stronger at the posterior, are well-established drivers of posterior spindle displacement and anaphase oscillations [67–69,75,78,79]. However, a restoring force must exist during spindle-pole oscillations, notably to stabilise the spindle position against these strong cortical pulling forces and permit oscillations. This centring mechanism could arise from the embryo’s geometry [68,75] or microtubule dynamics [67]. Alternative mechanisms include cytoplasmic pulling forces or cortical pushing forces. The latter may result from astral microtubules pushing against the cortex or from microtubule buckling [27,66,67]. We and others suggested that microtubule buckling—bending under compressive load—contributes to centring the spindle, in particular for spindle maintenance at the cell centre during metaphase [25,27,77,80]. It positions the flexural rigidity of microtubules as a candidate key factor in spindle choreography. We envisioned that sufficiently high microtubule rigidity is necessary to enable stronger cortical pushing forces. These forces may be required to counterbalance elevated cortical pulling forces during anaphase.

In C. elegans zygote, mutations in zyg-8^dclk1, the sole Doublecortin family member, have been shown to cause improper spindle positioning during anaphase, as observed through immunofluorescence and DIC microscopy [54,81]. ZYG-8 contains a kinase domain at the C-terminus and two doublecortin (DC) domains at the N-terminus, the latter enabling its binding to microtubules. Studies of the human homolog DCLK1 revealed that its kinase domain auto-phosphorylates the DC domains. Besides, hyperphosphorylation of these domains reduces DCLK1’s binding affinity to microtubules [82]. In C. elegans zygote, mutations in either domain of zyg-8 led to spindle orientation defects [54]. It has been proposed that ZYG-8 promotes microtubule assembly, which is required for correct spindle positioning [54,81]. Interestingly, Srayko et al. found that ZYG-8 contributes to microtubule growth, although to a lesser extent compared to the well-established polymerase ZYG-9^XMAP215 [48]. Besides, ZYG-8 was shown to limit astral microtubule nucleation, although the perturbations were less pronounced than those observed with spd-2^CEP192(RNAi) [48]. Could these moderate effects on microtubule dynamics caused by depleting ZYG-8 be sufficient to explain the spindle mis-positioning seen in zyg-8 mutants during anaphase? To address this question and better understand ZYG-8’s role in spindle positioning, we used microscopy techniques that allow dynamic studies in live embryos, along with advanced image analysis. We explored the potential roles of ZYG-8 in regulating both microtubule dynamics and flexural rigidity. We applied three complementary genetic perturbations: RNAi-mediated depletion of ZYG-8, overexpression, and the thermosensitive zyg-8(or484ts) mutant that disrupts microtubule binding [81]. Upon zyg-8 targeting, we analysed anaphase spindle oscillations and microtubule cortical contacts. We found that the observed changes could not be explained solely by altered microtubule dynamics. This prompted us to investigate whether ZYG-8 also regulates microtubule rigidity during zygote division. To assess changes in microtubule rigidity, we notably measured the distributions of microtubule local curvatures. We also examined spindle positioning forces using three biophysical assays. Our results demonstrated that ZYG-8 is needed to maintain the correct spindle orientation during anaphase, primarily by permitting high centring forces. Importantly, these centring forces depend on microtubule flexural rigidity, and their elevated levels specifically counterbalance the increased cortical pulling forces observed during anaphase. Overall, our study highlights the need for sufficient microtubule rigidity for proper mitotic spindle positioning and orientation during C. elegans zygote division.

Results

1. 1. ZYG-8 affects spindle-pole oscillations beyond microtubule growth and nucleation

In C. elegans zygotes, mutations in zyg-8^dclk1 disrupted spindle positioning, particularly during anaphase, as shown by DIC microscopy recordings [54,81]. These mutations were proposed to perturb astral microtubule stability, as microtubules stained with anti-tubulin antibodies appeared shorter. Immunofluorescence revealed that ZYG-8 localises at least to the spindle and its poles [54]. To investigate its association with astral microtubules, we generated three strains using CRISPR-Cas9 genome editing: one expressing fluorescently tagged ZYG-8 (mNeonGreen::ZYG-8) from the endogenous locus; a second with zyg-8 tagged at the locus with three OLLAS epitope tags; and a third carrying an integrated transgene overexpressing zyg-8 (pPie-1::mNG::3*OLLAS::ZYG-8) in addition to the endogenous copy (S1 Table). During the first embryonic division, in both endogenously expressed and overexpressed zyg-8 strains, ZYG-8 showed significant co-localisation with astral and spindle microtubules (S1 Fig, Method M5). This localisation was consistent with its previously proposed role in promoting microtubule assembly [54].

Previous studies in C. elegans zygotes focussing on ZYG-8 have either examined its role in spindle positioning using zyg-8 mutants or investigated its effects on microtubule dynamics through zyg-8(RNAi) treatment [48,54,81]. We sought to complement these studies by introducing three distinct genetic perturbations targeting zyg-8. First, we used the thermosensitive mutant zyg-8(or484ts), which carries a mutation in the first doublecortin domain [54,83]. This mutant has been reported to prevent ZYG-8’s ability to bind microtubules when maintained at the restrictive temperature of 25°C for more than 12 hours [81]. Second, we used the strain described above that overexpresses ZYG-8. Western blot analysis of worm lysates revealed increased OLLAS levels, indicating elevated ZYG-8 expression, as the OLLAS tag was inserted into the overexpression transgene (S2A Fig, Method M6). Besides, mNG::ZYG-8 fluorescence on both spindle and astral microtubules in zygotes increased approximately threefold (S2C, S2D and S2G Fig, Method M6). Third, we reduced ZYG-8 expression using RNAi. Western blot analysis of worm lysates from the strain carrying OLLAS-tagged zyg-8 showed decreased OLLAS levels upon zyg-8(RNAi), indicating reduced ZYG-8 expression (S2B Fig, Method M6). Besides, we observed an approximately fivefold decrease in mNG::ZYG-8 fluorescence on zygotic microtubules upon zyg-8(RNAi) (S2E, S2F and S2H Fig, Method M6). These genetic perturbations—loss of ZYG-8 from microtubules, overexpression, and partial depletion—provided the tools to investigate how different ZYG-8 levels affect mitotic spindle choreography.

Since ZYG-8 has been proposed to play a key role in spindle positioning, particularly during anaphase, we focussed on spindle pole oscillations occurring at this stage [66]. These oscillations have been widely used in previous studies to assess perturbations in spindle-positioning forces [67–69,71,72]. We quantified both the amplitude and frequency of the maximal oscillation at each spindle pole in strains expressing GFP::TBB-2 (zyg-8(RNAi)) or EBP-2::mKate2 (zyg-8 mutants) or mCherry::tubulin (zyg-8 overexpression) (Method M7). Following zyg-8(RNAi), we observed a large increase in oscillation amplitudes (Fig 1A–1C) and a major decrease in frequencies (Fig 1A, 1B and 1D). In zyg-8(or484ts) mutants at the restrictive temperature, oscillation amplitudes were further elevated (Figs 1C and S3). In mutants, oscillation frequencies at the anterior centrosome were further decreased compared to RNAi treatment, while those at the posterior centrosome were more variable than in other conditions – likely due to spindle misorientation – and therefore not significantly affected (Fig 1D). The stronger phenotype in mutant embryos compared to zyg-8(RNAi)-treated embryos was consistent with the partial depletion of ZYG-8 by RNAi (S2B and S2H Fig). These results suggested that ZYG-8 acts to limit spindle oscillation amplitudes during anaphase. In embryos overexpressing zyg-8, maximal amplitudes remained unchanged (Fig 1C), but oscillation frequencies increased significantly (Fig 1D), indicating that oscillation frequency depends on ZYG-8 levels.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. ZYG-8 limits spindle-pole oscillation amplitudes and increases their frequencies during anaphase.

(A-B) Exemplar posterior centrosome positions along the transverse axis for (A) a control RNAi embryo and (B) a zyg-8(RNAi)-treated embryo. Maximal amplitude is highlighted in green, while its frequency is annotated in blue. We applied a moving-average filter over a window size of 5 s to smooth the centrosome positions. Exemplar images used for centrosome analysis with a GFP::TBB-2 fluorescent labelling. (C) Maximal oscillation amplitudes and (D) their frequencies for the (cross) anterior and (circle) posterior centrosomes during anaphase. We tracked the centrosomes and analysed their positions in: (red) N = 16 zyg-8(RNAi)-treated embryos and (black) N = 10 control RNAi embryos expressing GFP::TBB-2; (light green) N = 21 zyg-8(or484ts) mutants and (light blue) N = 17 untreated embryos, both at the restrictive temperature (Rest. T°) expressing EBP-2::mKate2; (purple) N = 19 zyg-8 overexpressing embryos and (grey) N = 17 untreated embryos expressing mCherry::tubulin (Method M7). N represents the total number of embryos analysed across all replicates. The brown squares and error bars correspond to the means and SD. *** indicates significant differences (p ≤ 1x10^-4) and ns denotes non-significant differences (p > 0.05) (Method M15). Exemplar posterior centrosome positions along the transverse axis, along with representative images for a zyg-8(or484ts) mutant and an untreated embryo, both at the restrictive temperature, are shown in S3 Fig.

https://doi.org/10.1371/journal.pgen.1012196.g001

We and others have shown that the characteristics of spindle-pole oscillations were sensitive to microtubule dynamics [25,72]. Besides, a previous study has investigated microtubule growth and nucleation perturbations in C. elegans zygotes across a broad range of MAP depletions, including ZYG-8 [48]. This study found that ZYG-8 promotes microtubule growth and limits microtubule nucleation, although its effects were weaker than those of the polymerase ZYG-9^XMAP215 and the nucleator SPD-2^CEP192. To further explore the functions of ZYG-8, we quantified both parameters in our three genetic perturbations. We used an EB-labelling of microtubule plus-ends to assess whether the astral microtubule growth rate was altered (Method M8, Fig 2A). We detected no significant changes in growth rate in either zyg-8(RNAi)-treated embryos (Fig 2B), for which depletion was partial (S2B Fig), or in those overexpressing zyg-8 (Fig 2C). This suggested that the oscillation amplitude variations in these two perturbative conditions could have an alternative origin. In zyg-8(or484ts) mutants at the restrictive temperature, we observed a significant 15% reduction in microtubule growth rate (Fig 2D), consistent with previous results using zyg-8(RNAi) by injection [48]. To test whether this 15% reduction in microtubule growth could account for the exaggerated spindle-pole oscillations observed in zyg-8 mutants, we performed a partial RNAi depletion of the microtubule polymerase ZYG-9^XMAP215 [48,84,85]. This treatment shortened the mitotic spindle (S4A Fig), confirming efficient depletion [84], and reduced the microtubule growth rate by 16%, comparable to the effect in zyg-8 mutants (S4B Fig). However, spindle-pole oscillation amplitudes and frequencies were either unaffected or slightly reduced, and did not phenocopy the zyg-8 mutant phenotype (S4C and S4D Fig). These results suggested that the limited contribution of ZYG-8 in promoting microtubule growth was not sufficient on its own to account for its critical function in regulating spindle dynamics during anaphase.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. In zyg-8 mutants, microtubule growth rate is reduced by 15%.

(A) Microtubule growth rate assay by tracking comets in embryos whose microtubule plus-ends were fluorescently labelled with EBP-2::mKate2 (Method M8). Example of a control RNAi embryo processed with Kalman denoising, a preprocessing step performed prior to quantification, visualised with the Fire LUT. Scale bar represents 10 µm. (B-D) Growth rates of astral microtubules measured at (cross) anterior and (circle) posterior centrosomes: (B) (red) N_c = 27 comets from anterior centrosome and N_c = 23 comets from posterior centrosome in 10 zyg-8(RNAi)-treated embryos, and (black) N_c = 26 comets from anterior centrosome and N_c = 24 comets from posterior centrosome in 8 control RNAi embryos; (C) (purple) N_c = 25 comets from anterior centrosome and N_c = 25 comets from posterior centrosome in 10 zyg-8 overexpressing embryos and (black) N_c = 25 comets from anterior centrosome and N_c = 25 comets from posterior centrosome in 10 untreated embryos; (D) (light green) N_c = 37 comets from microtubules emanating from the anterior centrosome and N_c = 13 comets from posterior centrosome in 10 zyg-8(or484ts) mutants, and (light blue) N_c = 31 comets from anterior-centrosome and N_c = 19 posterior-centrosome comets in 13 untreated embryos, both at the restrictive temperature (Rest. T°). N_c represents the total number of comets analysed across all replicates, separately for anterior and posterior centrosomes, with about 5 comets analysed per embryo. The brown squares and error bars correspond to the means and SD. * and *** indicate significant differences with 1x10^-3 < p ≤ 1x10^-2 and p ≤ 1x10^-4, respectively. ns denotes non-significant differences (p > 0.05) (Method M15).

https://doi.org/10.1371/journal.pgen.1012196.g002

Alternatively, ZYG-8 has been proposed to limit microtubule nucleation in the C. elegans zygote [48]. To test this, we estimated the nucleation rate of astral microtubules under our three genetic perturbation conditions. Specifically, we counted comet-like structures representing EBP-2::mKate2 labelled plus-ends of growing microtubules emanating from centrosomes (Method M8, Fig 3A). We observed a significant increase in the nucleation rate in zyg-8(or484ts) mutants and a decrease upon zyg-8 overexpression, whereas zyg-8(RNAi) had no significant effect (Fig 3B–3D). Since centrosome diameter correlates with nucleation capacity [86–88], we also measured centrosome size using mCherry::TBG-1 (Method M9, Fig 3E and 3F). We found statistically significant differences in centrosome diameter, with a 12% decrease upon zyg-8 overexpression (Fig 3G) and a 6% increase following zyg-8(RNAi) treatment (Fig 3H). These results collectively indicated that ZYG-8 functions to restrict microtubule nucleation. Next, we tested whether altered nucleation alone could explain the spindle-pole oscillation phenotype observed upon zyg-8 perturbation. To this end, we partially depleted SPD-2^CEP192 by RNAi to reduce microtubule nucleation and assessed the effects on spindle pole oscillations [89]. spd-2(RNAi) caused a 21% reduction in centrosome diameter (S5A and S5B Fig), an effect comparable to—but slightly more pronounced than—that of zyg-8 overexpression (Fig 3E and 3G). While zyg-8 overexpression increases oscillation frequency by 20% (Fig 1D), spd-2(RNAi) decreased frequency by 11% (S5D Fig). Neither condition significantly altered oscillation amplitude (Figs 1C and S5C). Together, these findings suggested that ZYG-8’s role in limiting microtubule nucleation alone was not sufficient to account for the anaphase spindle oscillation phenotypes observed upon zyg-8 targeting.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. ZYG-8 limits microtubule nucleation.

(A) Microtubule nucleation rate assay by counting comets crossing a half-circle region of interest positioned 9 µm from the centrosome in embryos whose microtubule plus-ends were fluorescently labelled using EBP-2::mKate2 (Method M8). Example of a control RNAi embryo processed with Candle denoising, a preprocessing step performed prior to comet counting. Scale bar represents 10 µm. (B-D) Nucleation rates of the astral microtubules with N indicating the number of analysed centrosomes per condition: (B) (light green) N_e = 8 zyg-8(or484ts) mutants and (light blue) N_e = 9 untreated embryos, both at the restricted temperature (Rest. T°); (C) (purple) N_e = 10 zyg-8 overexpressing embryos and (black) N_e = 10 untreated embryos; (D) (red) N_e = 10 zyg-8(RNAi)-treated embryos and (black) N_e = 8 control RNAi embryos. (E-F) Regions centred on the centrosomes (40 x 40 pixels; 6.4 x 6.4 µm) from exemplar microscopy images showing centrosomes fluorescently labelled with mCherry::TBG-1: (E) representative images of untreated and zyg-8 overexpressing embryos; (F) representative images of control RNAi and zyg-8(RNAi)-treated embryos. Scale bars represent 2 µm. (G-H) Diameters during metaphase of (cross) anterior and (circle) posterior centrosomes (Method M9): (G) (purple) N_e = 21 zyg-8 overexpressing embryos and (black) N_e = 9 untreated embryos; (H) (red) N_e = 27 zyg-8(RNAi)-treated embryos and (black) N_e = 18 control RNAi embryos. N_e represents the total number of embryos analysed across all replicates. The brown squares and error bars correspond to the means and SD. ◊, *, and ** indicate significant differences with 1x10^-2 < p ≤ 5x10^-2, 1x10^-3 < p ≤ 1x10^-2, and 1x10^-4 < p ≤ 1x10^-3, respectively. ns denotes non-significant differences (p > 0.05) (Method M15).

https://doi.org/10.1371/journal.pgen.1012196.g003

1. 2. Cortical microtubule dynamics do not support a role for ZYG-8 in microtubule stabilisation

Previous works in human cells and in vitro have proposed that members of the Doublecortin family stabilise microtubules by preventing catastrophe and/or inhibiting depolymerisation [49,90–92]. We investigated whether a similar regulatory mechanism operates in the C. elegans zygote. Inspired by a previous study in nematodes on the role of CLS-2^CLASP in microtubule stability [93], we imaged, at the cell cortex, embryos expressing GFP::TBB-2 (zyg-8(RNAi) treatment) or mCherry::tubulin (zyg-8 overexpression) and examined both the number and duration of astral microtubule contacts there (Fig 4A and 4B and S1–S4 Movies). Since microtubule catastrophes are rare in C. elegans zygote cytoplasm, most astral microtubules grow until they reach the cortex [72,94]. If ZYG-8 stabilises microtubules by limiting depolymerisation or preventing catastrophe in the cytoplasm, zyg-8(RNAi) would be expected to reduce the number of cortical contacts. However, we observed the opposite: zyg-8(RNAi) increased the density of cortical contacts by an average of 50%, while zyg-8 overexpression led to a 13% reduction (Fig 4C and 4D). These results argued against a cytoplasmic stabilisation role for ZYG-8. Next, to explore whether ZYG-8 might stabilise microtubules at the cell cortex by reducing catastrophe frequency, we analysed the duration of microtubule cortical contacts. If ZYG-8 acted to stabilise microtubules at the cortex, zyg-8(RNAi) would be expected to increase the number of short-duration contacts. In contrast, zyg-8(RNAi) resulted in a 10–15% reduction in the short-duration contacts (Fig 4E and 4F). Conversely, zyg-8 overexpression led to a 7% increase in short-duration contacts (Fig 4G and 4H). Altogether, these cortical measurements did not support a role for ZYG-8 in stabilising microtubules within the cytoplasm or at the cortex.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Perturbations in microtubule cortical dynamics upon targeting zyg-8 do not support a role for ZYG-8 in stabilising microtubules.

(A-B) Embryos whose microtubules are labelled by GFP::TBB-2 and imaged at the cortical plane (Method M4): examples of (A) a control RNAi embryo (S1 Movie), and (B) a zyg-8(RNAi)-treated embryo (S2 Movie) at anaphase onset. Green spots highlight the detected contacts (Method M10). Scale bars represent 10 µm. (C-D) Embryo-averaged density of astral microtubule contacts at the cortex plane along mitosis. We tracked and analysed the cortical contacts of microtubules labelled: (C) by GFP::TBB-2 in (red) N = 14 zyg-8(RNAi)-treated embryos and (black) N = 17 control RNAi ones; (D) by mCherry::tubulin in (purple) N = 8 zyg-8 overexpressing embryos and (black) N = 8 untreated ones. (E) Embryo-averaged distributions of cortical lifetimes of astral microtubules during metaphase (Kolmogorov–Smirnov test: p = 6x10^-4, Methods M10 and M15). Embryos are the same as in panel C. (F) Relative counts of contacts with durations ≤ 0.4 s (from distributions presented in panel E), highlighting differences in brief contact durations. (G) Embryo-averaged distributions of cortical lifetimes of astral microtubules during metaphase (Kolmogorov–Smirnov test: p = 0.18, Methods M10 and M15). Embryos are the same as in panel D. (H) Relative counts of contacts with durations ≤ 0.4 s (from distributions in panel G), highlighting variations in short-lived durations. Representative movies of each of the four present conditions are provided as S1-S4 Movies. Microtubule contacts were tracked using u-track (Method M10). N represents the total number of embryos analysed across all replicates. The brown squares and error bars correspond to the medians and the quartiles. ◊ and *** indicate significant differences with 1x10^-2 < p ≤ 5x10^-2 and p ≤ 1x10^-4, respectively (Method M15). This figure does not include data for zyg-8(or484ts) mutant, as severe spindle misorientation precluded reliable quantification for this analysis.

https://doi.org/10.1371/journal.pgen.1012196.g004

To further test this finding, we compared cortical microtubule dynamics and spindle pole oscillations in zyg-8(RNAi)-treated embryos with those following depletion of CLS-2^CLASP—a known microtubule stabiliser that prevents catastrophe and promotes rescue [93,95]. We validated the penetrance of the RNAi treatment by observing accelerated mitotic spindle elongation, which suggested midzone integrity perturbations [97,98] (S6A Fig). At the cell cortex, in cls-2(RNAi)-treated embryos, we found a reduced density of microtubule contacts during anaphase (S6F Fig), and a decreased proportion of long-duration contacts (S6G Fig). These results aligned with CLS-2’s known role in catastrophe prevention, but differed from zyg-8(RNAi) phenotypes, further suggesting that ZYG-8 does not primarily prevent microtubule catastrophe in C. elegans zygote. Notably, cls-2(RNAi) caused an increase in oscillation amplitude, although weakly significant, with no significant variation in oscillation frequency (S6B and S6C Fig). These phenotypes were distinct from zyg-8(RNAi). This suggested that altered microtubule catastrophe alone induces only minor perturbations in spindle pole oscillations and, therefore, could not explain the pronounced oscillation phenotypes observed in zyg-8(RNAi) embryos.

We then compared zyg-8(RNAi) phenotypes with those following hypomorphic depletion of KLP-7^MCAK, which destabilises microtubules by promoting depolymerisation [96]. We confirmed klp-7(RNAi) treatment penetrance by observing slightly increased spindle elongation (suggesting only a mild alteration of midzone integrity) and longer metaphase spindle, as expected [99] (S7A Fig). At the cell cortex, KLP-7 depletion increased the microtubule contact density and the proportion of long-duration contacts (S7F and S7G Fig). These findings reflected impaired depolymerisation activity. If ZYG-8 stabilised microtubules by preventing depolymerisation, we would have observed cortical microtubule phenotypes opposite to those of KLP-7 depletion. This was not the case. Regarding spindle pole oscillations, KLP-7 depletion increased amplitudes at the posterior centrosome and decreased frequencies at both centrosomes (S7B and S7C Fig)—phenotypes not opposite to zyg-8(RNAi). Thus, a putative role of ZYG-8 in preventing microtubule depolymerisation could not account for the observed phenotypes.

Overall, our data indicated that the zyg-8-associated perturbations in spindle-pole oscillations and cortical microtubule dynamics could not be fully attributed to alterations in microtubule dynamics alone. Interestingly, proteins of the Doublecortin family have been proposed to regulate microtubule rigidity in neurons [37,100]. We asked whether the observed perturbations in spindle-pole oscillations and cortical microtubule behaviour could additionally result from altered microtubule rigidity upon zyg-8 targeting.

1. 3. ZYG-8 would control microtubule flexural rigidity

Direct measurements of microtubule rigidity in C. elegans embryos are not feasible, as standard techniques (e.g., thermal fluctuations, optical trap/tweezers, or calibrated flow) require isolated microtubules [15,17,34,101–103]. As a quantitative alternative, we employed a set of complementary metrics that collectively capture different aspects of microtubule bending behaviour (Method M11). Such an approach has been previously used as a proxy for rigidity under the assumption that the pulling forces acting on the microtubule network are similar between conditions [20,104,105]. Notably, microtubules in axonal growth cones were more often strongly curved upon depletion of the Doublecortin-family proteins DCX and DCLK1 [37]. First, we analysed local curvature distributions: a higher proportion of strongly curved segments reflects reduced rigidity (S8A Fig). To complement this population-level analysis with a more filament-specific readout, we extracted the 95^th percentile of curvature values along individual microtubules, as a robust measure of maximum bending per microtubule (S8B Fig). Finally, we incorporated filament tortuosity—a global shape descriptor independent of curvature computation and less sensitive to sampling resolution. Defined as the ratio between curvilinear and end-to-end distances, tortuosity has also been used previously [35] (S8C Fig).

We first asked whether astral microtubules displayed increased local curvature in zyg-8(or484ts) mutants or in zyg-8(RNAi)-treated embryos. We imaged live embryos expressing GFP::TBB-2 using a super-resolved confocal microscope (Nikon with Nsparc detector, Method M4). We observed regions of high local curvatures along astral microtubules, which appeared more frequently in the mutants (Fig 5A and S5 and S6 Movies) than in untreated embryos (Fig 5B and S7 Movie), both at the restrictive temperature. Besides, astral microtubules in the mutants were long and extended to the cell periphery. Consistently, astral microtubules exhibited increased bending in live zyg-8(RNAi)-treated embryos (S9A Fig and S8 Movie) compared to control RNAi embryos (S9B Fig and S9 Movie). In contrast, zyg-8 overexpression in live embryos did not significantly alter microtubule curvature amplitudes or frequencies (S10 Fig and S10 and S11 Movies). The fast motion of astral microtubules required high image acquisition rate to avoid motion blur, which led, unfortunately, to images with low signal-to-noise ratio complicating microtubule curvature quantification. In zyg-8(or484ts) mutants, significant out-of-plane movements hindered microtubule visualisation along their full length, calling for acquisitions on multiple z-sections (S5 and S6 Movies). As a result, we opted for investigating curvature phenotypes in fixed samples.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. In zyg-8(or484ts) mutants, astral microtubules are highly bent during the exaggerated spindle pole oscillations.

Time-lapse confocal super-resolved images along anaphase of (A) an exemplar zyg-8(or484ts) mutant and (B) an exemplar untreated embryo, both at the restrictive temperature. Microtubules were labelled by GFP::TBB-2. Time is indicated from first image in mm:ss:ms. Scale bars represent 10 µm. These time-lapse images are sourced from the S5 and S7 Movies. Red arrows highlight bent microtubules. Similar images for zyg-8(RNAi) and zyg-8 overexpression are shown in S9 and S10 Figs, with associated movies provided as S8–S11 Movies.

https://doi.org/10.1371/journal.pgen.1012196.g005

We imaged microtubules stained for α-tubulin using a confocal super-resolution microscopy (Methods M3 and M4). In zyg-8(RNAi)-treated fixed embryos, we again observed more bent astral microtubules (S11A and S11B Fig). This phenotype was even more pronounced in zyg-8(or484ts) fixed mutants at the restrictive temperature, where microtubule curvatures were noticeably increased (S11C and S11D Fig). We also detected fragmented microtubules in the cytoplasm of some fixed mutant embryos (S11D Fig). Interestingly, such fragmentation may result from excessive curvatures, as previous studies have linked microtubule breakage to strong bending [104,106,107]. Fragmentation could also contribute to the appearance of shorter astral microtubules. This phenotype—very-short astral-microtubule—was observed only in a subset of zyg-8(or484ts) fixed mutants, resembling the initial images of stained microtubules in this mutant [54,81].

To go beyond visual inspection, we quantified the distribution of microtubule curvatures. Accurate segmentation was essential for this analysis but proved challenging due to faint and uneven microtubule labelling. To overcome this, we assembled a five-step image processing pipeline enabling the segmentation of astral microtubules and computation of local curvatures on a pixel-by-pixel basis using a three-point method (S8D –S8J Fig, Method M11). By averaging curvature distributions across conditions, we observed a slight decrease in the proportion of low local curvatures following zyg-8(RNAi) treatment (Fig 6A and 6B). In zyg-8(or484ts) mutants at the restrictive temperature, the distribution shifted significantly further toward higher curvature values compared to controls, showing a stronger effect than RNAi (Fig 6C and 6D). Notably, some local curvature values reached up to 0.8 µm^-1, consistent with curvature thresholds associated with microtubule breakage [106,107]. Focussing on the highest curvatures, we observed increases of 13% in metaphase and 12% in anaphase following zyg-8(RNAi). These increases were further amplified in zyg-8(or484ts) mutants, reaching 33% and 42% in metaphase and anaphase, respectively (Fig 6E and 6F). Lastly, microtubule tortuosity was significantly elevated following zyg-8(RNAi) treatment, and increased even further in zyg-8(or484ts) mutants at the restrictive temperature (Fig 6G and 6H). Together, these analyses demonstrated that ZYG-8 modulates microtubule curvature and tortuosity, supporting its role in regulating microtubule flexural rigidity.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Astral microtubules appear more curved and tortuous in zyg-8(RNAi)-treated embryos, and even more so in zyg-8(or484ts) mutants.

Characterisation of microtubule shapes in fixed embryos analysed using the image-processing pipeline depicted in S8 Fig. (A-B) Embryo-averaged distributions of astral microtubule local curvatures of (red) zyg-8(RNAi)-treated embryos and (black) control RNAi embryos: (A1) during metaphase with (red) N = 5 z-projected stacks and (black) N = 6 z-projected stacks (Kolmogorov–Smirnov test: p = 0.36, Method M15); and (B1) during anaphase with (red) N = 8 z-projected stacks and (black) N = 7 z-projected stacks (Kolmogorov–Smirnov test: p = 0.36, Method M15). (A2, B2) Focus on the proportions of low local curvatures (≤ 0.01 µm^-1) for the same embryos as in panels A1 & B1. (C-D) Embryo-averaged distributions of astral microtubule local curvatures of (light green) zyg-8(or484ts) mutants and (light blue) untreated embryos, both at the restrictive temperature: (C1) during metaphase with (light green) N = 8 z-projected stacks and (light blue) N = 9 z-projected stacks (Kolmogorov–Smirnov test: p = 0.03, Method M15); and (D1) during anaphase with (light green) N = 6 z-projected stacks and (light blue) N = 6 z-projected stacks (Kolmogorov–Smirnov test: p = 0.014, Method M15). (C2, D2) Focus on the proportions of low local curvatures (≤ 0.01 µm^-1) for the same embryos as in panels C1 & D1. (E-F) Medians per z-projected stack of the highest curvatures (measured as 95^th percentile curvature): (E) during metaphase (data are the same as in panels A & C); and (F) during anaphase (same data as in panels B & D). (G-H) Medians per z-projected stack of microtubule tortuosity: (G) during metaphase, using the same embryos as in panels A, C & E; and (H) during anaphase, using the same embryos as in panels B, D & F. Embryos were fixed and immuno-stained for α-tubulin; microtubules were segmented; and local curvature and tortuosity were computed (Methods M3, M4, M11). N represents the total number of z-projected stacks analysed across all replicates. The brown squares and error bars correspond to the medians and the quartiles. ◊, *, **, and *** indicate significant differences with 1x10^-2 < p ≤ 5x10^-2, 1x10^-3 < p ≤ 1x10^-2, 1x10^-4 < p ≤ 1x10^-3, and p ≤ 1x10^-4, respectively. ns denotes non-significant differences (p > 0.05) (Method M15). This figure does not include data for zyg-8 overexpressing embryos, as live imaging (S10 Fig) demonstrated no significant differences in the astral microtubule curvatures.

https://doi.org/10.1371/journal.pgen.1012196.g006

We foresaw that reduced rigidity would lead to increased bending, resulting in weaker pushing forces against the cortex. This, in turn, would delay force-dependent catastrophes and increase the cortical lifetimes of astral microtubules [108,109] (Fig 7A). We used the established agent-centred simulation Cytosim [110] to test this hypothesis (S2 Table, Method M12). We conducted three sets of simulations in which only the rigidity of astral microtubules was varied, using values of 2, 10, and 25 pN.µm² (S12–S14 Movies and S12A Fig). These values were based on in vitro measurements of microtubule rigidity [15–17,101,103]. We found that these in silico changes in rigidity affected both microtubule curvature and tortuosity. Specifically, the highest curvatures and tortuosity increased as rigidity decreased (S12B and S12C Fig). Importantly, microtubules with lower rigidity exhibited longer cortical lifetimes, supporting our initial prediction (S12D Fig). Returning to in vivo experiments, we measured the cortical contact durations of astral microtubules in zyg-8(RNAi)-treated embryos expressing GFP::TBB-2 and found an increased proportion of microtubules with prolonged contacts compared to control RNAi embryos (Fig 4E). This observation was consistent with reduced microtubule rigidity in zyg-8(RNAi)-treated embryos.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Depletion of ZYG-8 significantly weakens the cortical pushing events.

(A) Diagram illustrating the correlation between microtubule rigidity and cortical lifetime: (orange) softened microtubules lead to increased pushing-event lifetimes compared to (green) rigid microtubules, due to weaker pushing force (red arrow) delaying catastrophe. Note that rescue of depolymerising astral microtubules is rare in C. elegans zygote cytoplasm, supporting continued depolymerisation. (B, F) Lifetimes and (C, G) densities of the cortical pulling events, and (D, H) lifetimes and (E, I) densities of the cortical pushing events during anaphase. We tracked and analysed the microtubule contacts at the cell cortex (B-E) using a GFP::TBB-2 fluorescent labelling in (red) N = 14 zyg-8(RNAi)-treated embryos and (black) N = 17 control RNAi embryos (same embryos as for Fig 4C, 4E and 4F), or (F-I) using a YFP::TBA-2 fluorescent labelling in (blue) N = 8 gpb-1(RNAi)-treated embryos and (black) N = 9 control RNAi embryos (Method M10). N represents the total number of embryos analysed across all replicates. ◊ and * indicate significant differences with 1x10^-2 < p ≤ 5x10^-2 and 1x10^-3 < p ≤ 1x10^-2, respectively. ns denotes non-significant differences (p > 0.05) (Method M15). This figure does not include data for zyg-8(or484ts) mutant, as severe spindle misorientation precluded reliable quantification. Additionally, data for zyg-8 overexpressing embryos are not included, as this figure focuses on elucidating the origin of increased spindle-pole oscillation amplitudes—a phenotype not observed in this condition (Fig 1C).

https://doi.org/10.1371/journal.pgen.1012196.g007

Finally, we investigated spindle pole oscillations in light of ZYG-8’s role in microtubule rigidity. We previously proposed an antagonistic-motors model to explain these oscillations, in which the oscillation frequency is in particular proportional to the square root of the centring rigidity [66]. This centring rigidity has been associated with the pushing force generated by microtubules growing against the cortex [25,77]. Consequently, lower microtubule rigidity—leading to reduced pushing force—was expected to lower centring rigidity, and thus decrease oscillation frequency. Indeed, we observed a reduction in oscillation frequency in zyg-8(RNAi)-treated embryos (Fig 1D). Conversely, this frequency was increased in embryos overexpressing zyg-8 (Fig 1D).

Taken together, these three experiments supported a model in which ZYG-8 contributes to microtubule stiffening.

1. 4. ZYG-8 ensures spindle centring mostly by controlling microtubule rigidity

Although the centring force generated by microtubules pushing against the cell cortex—critically dependent on microtubule rigidity—was mostly invoked in metaphase [25,27,80], it is also thought to act as a restoring force for spindle rocking in anaphase [66,69]. Indeed, we previously proposed that, when a spindle pole moves toward one side of the cortex, the centring spring opposes this displacement [66]. Consequently, a reduction in microtubule rigidity would decrease the stiffness of this centring spring, allowing greater spindle pole displacement and thereby increasing oscillation amplitudes (Fig 8G). In zyg-8(RNAi) condition or in zyg-8(or484ts) mutants, we observed a large increase in the amplitudes of spindle pole oscillations (Fig 1C). We therefore considered two possible and non-mutually exclusive explanations for this increase: enhanced cortical pulling or reduced cortical pushing.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. In zyg-8(or484ts) mutants, overly soft microtubules result in a misoriented and mispositioned spindle at the anaphase end.

(A-B) Time-lapse series of (A) an exemplar untreated embryo and (B) an exemplar zyg-8(or484ts) mutant along anaphase, both at the restrictive temperature (Rest. T°) and with EBP-2::mKate2 labelling. The green arrows indicate the spindle’s final positions along the anteroposterior (AP) axis and the dotted yellow lines indicate the final orientations. The red double-headed arrows underline the minimal centrosome-to-cortex distances. The scale bars represent 10 µm. (C-D) Spindle (C) positions along the AP axis and (D) angles for (light green) N = 20 zyg-8(or484ts) mutants and (light blue) N = 17 untreated embryos, both at the restrictive temperature: (circle) at -60 s from anaphase onset, (square) at anaphase onset, and (cross) at anaphase end. (E-F) Spindle final (E) positions along the AP axis and (F) angles for (black) N = 16 zyg-8(RNAi)-treated embryos and (red) N = 10 control RNAi embryos. We tracked and analysed the centrosomes of embryos with (C-D) EBP-2::mKate2 labelling and (E-F) GFP::TBB-2 labelling (Method M7). N represents the total number of embryos analysed across all replicates. The brown squares and error bars correspond to the means and SD. * and *** indicate significant differences with 1x10^-3 < p ≤ 1x10^-2 and p ≤ 1x10^-4, respectively. ns denotes non-significant differences (p > 0.05) (Method M15). (G) Schematics illustrating how a large reduction in microtubule flexural rigidity, which reduces cortical pushing force, permits centrosome proximity to the cell cortex. This, in turn, causes defects in spindle positioning and orientation at the end of anaphase. Data for zyg-8 overexpressing embryos are not included in this figure, as it focuses on the consequences of reduced microtubule rigidity—a phenotype not observed under this condition (S10 Fig).

https://doi.org/10.1371/journal.pgen.1012196.g008

To distinguish between these possibilities, we first measured the most posterior position of the posterior centrosome, projected along the anteroposterior (AP) axis, during anaphase following zyg-8(RNAi) treatment. We found no significant difference compared to controls (80.7 ± 1.4% of AP axis for zyg-8(RNAi) (N = 16), versus 79.8 ± 1.0% for control RNAi embryos (N = 10), Student’s t-test: p = 0.07). This suggested that posterior displacement, driven by cortical pulling forces, remained largely unaltered. Next, we applied the DiLiPop analysis, a biophysical method that enables to assess cortical pulling and pushing events during anaphase. Our previous work has shown that two distinct populations of astral microtubules reside at the cell cortex, disentangled by statistical analysis of cortical contact durations [80] (Method M10). By genetically perturbing either pulling or pushing cortical forces, we previously associated the short-lived population with pulling events and the long-lived population with pushing [80]. Upon zyg-8(RNAi), we observed a 10% rise in the lifetime of the pulling events (Fig 7B) and a 3% reduction in their density (Fig 7C). Since these measurements reflect the number of events (i.e., the number of microtubule cortical tracks, not the instantaneous density of cortical contacts), longer-lasting pulling events (increased lifetimes) reduce the number of distinct pulling events that can occur within the same time window (reduced density). This occurs because all available cortical force generators are engaged in pulling during anaphase [66,72,75], explaining the simultaneous observation of increased pulling-event lifetimes and decreased pulling-event density in zyg-8(RNAi) embryos. However, these modest changes in pulling events could not account for the strong increase in oscillation amplitudes. For comparison, when we depleted GBP-1 to enhance cortical pulling force [111–113], we observed a significant increase in pulling-event density, as expected, while the lifetimes of both pulling and pushing events remained unaffected (Fig 7F–7I). This phenotype differed from that of zyg-8(RNAi), leading us to conclude that zyg-8(RNAi) had at most a minor effect on cortical pulling forces. Concurrently, we found that pushing microtubules resided at the cortex significantly longer upon zyg-8(RNAi), with a 37% increase in lifetimes (Fig 7D). This was consistent with their sensitivity to microtubule rigidity (Fig 7A). Besides, the density of pushing events increased by 22% (Fig 7E), partially due to an enhanced microtubule nucleation capacity (Fig 3F). Despite the rise in pushing event density, we proposed a global reduction in effective cortical pushing force, since a bent microtubule has a limited pushing force (Euler force). Consequently, the larger oscillation amplitudes observed following zyg-8(RNAi) treatment likely resulted from reduced pushing force, while cortical pulling forces appeared to be largely unaffected.

To further investigate whether cortical pulling forces remain unaffected when targeting zyg-8, we tracked cortical dynein using a CRISPR strain expressing GFP::DHC-1 [114] (Method M13), a biophysical assay previously employed [114,115]. It is important to note that only cortical dynein molecules incorporated into at least the GPR-1/2-LIN-5-dynein trimeric complex actively generate pulling forces on astral microtubules [74,116]. In control RNAi embryos, we identified two distinct dynein populations at the cell cortex: a short-lived population with a lifetime of 0.66 s, and a long-lived population with a lifetime of 1.66 s (S3 Table). The lifetime of the predominant short-lived population aligns with previously reported measurements [115], and falls within the range of cortical pulling event lifetimes (Fig 7B and 7F), suggesting that this population likely includes active pulling events. Upon zyg-8(RNAi), our analysis revealed similar dynein dynamics (S3 Table). The increased dynein density aligns with the elevated microtubule contact density (Fig 4C), while the prolonged dynein lifetimes mirror the increased microtubule cortical contact durations. Crucially, the magnitude of these changes in dynein dynamics was comparable to those observed for microtubule characteristics, suggesting no additional regulatory mechanism. Given that not all cortical dynein is engaged in pulling [114,115] and that cortical force generators are saturated (i.e., actively pulling) during anaphase in untreated embryos [66,72,75], it is unlikely that ZYG-8 depletion substantially increases cortical pulling forces.

To examine the role of ZYG-8 in controlling microtubule rigidity—and, in turn, the centring force—we sought an alternative characterisation of spindle positioning forces at the cellular level. To do this, we used a biophysical assay previously developed by our laboratory focusing on the spindle positional micro-movements acquired at high frame rate. We extracted the mechanical fingerprint of these micro-movements along the transverse axis during anaphase. By fitting a second-order model to the experimental spectrum, we derived three mechanical parameters that characterise spindle position mechanics: the diffusion coefficient D and two corner frequencies f_c and f₀ [25] (Method M14). Our analysis revealed that ZYG-8 depletion caused a large 79% reduction in the centring-to-damping corner frequency f_c, compared to the control (Table 1). We also observed a 24% decrease in the diffusion coefficient D, which scales inversely with damping, and a 35% increase in the damping-to-inertia frequency f₀ (Table 1). We interpreted the decrease in f_c as primarily reflecting a centring force reduction due to softer microtubules [25]. The moderate decrease in D likely resulted from an increased number of microtubules contacting the cortex. The rise in f₀ could be partly attributed to the reduced D, and appeared inconsistent with any notable increase in inertia, suggesting that the pulling forces were only mildly or not significantly affected. Overall, the mechanical fingerprint of spindle positioning following zyg-8(RNAi) treatment was qualitatively consistent with a reduction in the centring force, mainly due to softer microtubules. This reduction likely contributed to the increased amplitude of spindle pole oscillations.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Targeting zyg-8 or ptl-1 reduces the centring frequency.

https://doi.org/10.1371/journal.pgen.1012196.t001

In zyg-8(or484ts) mutants at the restrictive temperature, we observed larger amplitudes of spindle-pole oscillations compared to those seen with zyg-8(RNAi) treatment (Figs 1C and S3). We hypothesised that this phenotype was due to a severely weakened centring force caused by excessively soft microtubules. This was supported by the increased proportion and magnitude of microtubule curvatures observed in these mutants (Figs 5 and 6). To test this hypothesis, we analysed spindle positional micro-movements in zyg-8(or484ts) mutants. We found a very large and significant reduction in the centring-to-damping corner frequency f_c, while the other two mechanical parameters, D and f₀, were not significantly altered (Table 1). These results suggested that in zyg-8(or484ts) mutants, the pronounced reduction in f_c, indicative of weakened centring forces, accounted for the exaggerated spindle pole oscillations. This reduction was likely due to decreased microtubule rigidity, consistent with the observed abnormalities in microtubule shapes.

Altogether, our findings demonstrated that ZYG-8 regulates the cortical pushing force responsible for spindle centring, primarily by modulating microtubule flexural rigidity. This pushing force acted as a restoring force during anaphase spindle rocking. While cortical pulling forces remain the dominant drivers of spindle posterior displacement, our work highlighted the important, though often underappreciated, contribution of pushing forces to overall spindle choreography.

1. 5. Very soft microtubules disrupt final spindle positioning and orientation

Having identified an additional role for ZYG-8 in controlling microtubule flexural rigidity, we next investigated the importance of this function for successful asymmetric mitosis. In zyg-8(or484ts) mutants at the restrictive temperature, the enhanced spindle-pole oscillation amplitudes – greater than those observed in zyg-8(RNAi) – resulted in a significantly reduced centrosome-to-cortex distance (S3B Fig). We thus examined the consequences of this reduced distance in the course of anaphase. In untreated embryos, spindle oscillations gradually built up and then died down, ultimately leaving the spindle properly aligned along the anteroposterior (AP) axis of the embryo (Fig 8A). In contrast, in zyg-8(or484ts) mutants, the closer proximity of the centrosome to the cell cortex pulled the entire spindle further toward the cell periphery. This appeared to prevent the spindle from re-centring during the oscillation die-down (Fig 8B). The reduction in cortical pushing disrupted the force balance, favouring excessive net pulling. As a result, the spindle failed to return to the central position by the end of anaphase, leading to both mispositioning and misorientation (Fig 8B–8D). Notably, embryonic lethality of zyg-8(or484ts) mutants was elevated (> 99%) at the restrictive temperature. Based on these observations, we proposed that the spindle misalignment seen in zyg-8(or484ts) mutants arose from impaired pushing forces, which disturbed the balance between pulling and pushing forces required for proper spindle positioning. Notably, spindle mispositioning and misorientation in zyg-8(or484ts) mutants were restricted to anaphase, as we observed no significant differences in spindle position or orientation between mutants and untreated embryos at either mid-metaphase or anaphase onset (Fig 8C and 8D). These observations suggested that the anaphase-specific increase in cortical pulling forces may contribute to the observed defects.

We thus hypothesised that reducing cortical pulling force could rescue the final spindle orientation in zyg-8(or484ts) mutants, by preventing an excessive imbalance between cortical pulling and pushing forces. To test this, we depleted GPR-1/2^LGN by RNAi, a component of the cortical force generator that pulls on astral microtubules. In non-mutated embryos treated with gpr-1/2(RNAi) at the restrictive temperature, we observed no spindle pole oscillations, and the most-posterior position of the spindle was significantly shifted toward the cell centre, consistent with a decrease in cortical pulling forces (S13A Fig). Similarly, in zyg-8(or484ts) mutants treated with gpr-1/2(RNAi) at the restrictive temperature, oscillations were absent and the most-posterior position was also shifted closer to the cell centre (S13A Fig). Strikingly, depleting GPR-1/2 in zyg-8(or484ts) mutants rescued the spindle misorientation defects (S13B Fig). This finding indicated that low microtubule rigidity did not affect final spindle orientation when cortical pulling forces were reduced. These results highlighted the critical importance of maintaining a proper balance between cortical pulling and pushing forces for accurate spindle positioning and successful cell division.

Importantly, we quantified spindle final position and orientation across all experimental conditions in our study. The selective perturbation of microtubule growth via zyg-9(RNAi)—which induces changes in growth rates comparable to those observed in zyg-8(or484ts) mutants—did not affect spindle final position or orientation (S4E and S4F Fig). Moreover, specific impairments in microtubule nucleation (SPD-2 depletion; S5E and S5F Fig), microtubule stability (CLS-2 depletion; S6D and S6E Fig), or microtubule depolymerisation (KLP-7 depletion; S7D and S7E Fig) similarly failed to perturb spindle final position or orientation. Together, these observations further supported our conclusion that reduced microtubule rigidity, rather than altered microtubule dynamics, is the determining factor of spindle misorientation in zyg-8(or484ts) mutants.

In conclusion, sufficiently rigid microtubules were required to generate the centring force necessary during anaphase, when cortical pulling forces are at their peak. This centring force prevented the centrosome from being displaced too far from the centre, ensuring proper final orientation of the spindle (Fig 8G). Reliable spindle positioning at the end of mitosis is crucial for accurate cleavage furrow placement during cytokinesis, a process that determines the proper partitioning of cell fate determinants [120].

Discussion

Based on three distinct, complementary genetic perturbations of zyg-8, our work highlights a dual role for ZYG-8^DCLK1 during mitosis, involving the control of both microtubule rigidity and dynamics. We first confirmed previous findings by Srayko et al., showing that ZYG-8 modestly promotes microtubule growth while limiting the nucleation of astral microtubules in C. elegans zygote [48] (Figs 2 and 3). However, the modest changes observed in astral microtubule dynamics in the cytoplasm upon zyg-8 targeting, as well as its potential role in stabilising microtubules, are not sufficient to fully explain the phenotypes related to spindle pole oscillations during anaphase, spindle final position and orientation, or the altered behaviour of astral microtubule contacts at the cell cortex. If ZYG-8’s main function were to stabilise microtubules by reducing their catastrophe rate at the cell cortex, zyg-8(RNAi) treatment would be expected to decrease the cortical lifetimes of astral microtubules and reduce their contact counts. This phenotype would resemble what we observed with cls-2(RNAi) (S6F and S6G Fig). In contrast, we found the opposite effect: zyg-8(RNAi) increases both astral microtubule contact counts and durations (Fig 4C and 4E). Upon klp-7(RNAi), which reduces microtubule depolymerisation, we found an increase in both microtubule contact counts and durations (S7F and S7G Fig). If ZYG-8 primarily acted to prevent microtubule depolymerization, the phenotypes would have differed markedly from those in zyg-8(RNAi)—yet they remained similar. Similarly, if ZYG-8 primarily promoted microtubule growth, zyg-8(RNAi) should result in fewer astral microtubules reaching the cortex and shorter contact durations, as previously seen with zyg-9(RNAi) [80]. Instead, in our current study, we observed the opposite phenotypes (Fig 4C and 4E–4F). Furthermore, if ZYG-8 limited microtubule catastrophe within the cytoplasm, then zyg-8(RNAi) would be expected to decrease the number of astral microtubule contacts at the cell cortex, as more microtubules would depolymerise before reaching it. This, again, contradicts our observations (Fig 4E). By contrast, the increased number of astral microtubule contacts upon zyg-8(RNAi) aligns with a role for ZYG-8 in limiting microtubule nucleation. However, this effect alone cannot fully account for the observed variations in microtubule cortical-contact number. While our comparisons using depletions of various MAP provide valuable insights into the role of microtubule dynamics in spindle oscillations and cortical microtubule contact phenotypes, we acknowledge some limitations. First, some perturbations introduced additional confounding effects, e.g., cls-2(RNAi) affecting midzone integrity, complicating direct comparisons with zyg-8 perturbations. Second, cumulative effects of multiple perturbations may contribute to the phenotypes observed upon zyg-8 targeting. Nonetheless, taken collectively, our set of experiments suggests that alterations in microtubule dynamics alone cannot fully recapitulate the spindle-pole oscillation variations, spindle final position defects, and cortical microtubule contact phenotypes observed upon zyg-8 targeting. This supports our conclusion that ZYG-8 plays an additional role beyond simply regulating microtubule dynamics.

Given that Doublecortin family proteins have been proposed to regulate microtubule rigidity, we investigated whether ZYG-8 performs a similar function in C. elegans zygotes. Initial evidences for ZYG-8’s role in rigidifying microtubules came from quantitative measurements of microtubule curvature. We observed increased tortuosity and curvature—both in magnitude and frequency—in zyg-8(RNAi) embryos and zyg-8(or484ts) mutants (Fig 6). The gradual differences in these curvature and tortuosity phenotypes between the two conditions aligned with ZYG-8’s binding level to microtubules (S2 Fig) [81]. Furthermore, the consistency of the curvature perturbations across multiple experimental replicates (fixed and live embryos) provided strong support for ZYG-8’s role in controlling microtubule rigidity (Figs 5, 6, and S9). A second line of evidence came from the agreement between our simulation predictions and experimental observations regarding the relationship between microtubule cortical lifetime and rigidity (Figs 4E, 4G, and S12). Our DiLiPop analysis, which enables monitoring cortical microtubule force-generating events, provided a third line of evidence. It revealed that the reduction in the cortical lifetimes is associated specifically to the pushing events: less rigid microtubules bend more when growing against the cortex, thereby reducing pushing force, delaying catastrophe onset, and prolonging cortical contact lifetimes (Fig 7A and 7D). As a fourth argument, the altered frequency of spindle pole oscillations upon zyg-8 targeting is consistent with a role in controlling microtubule rigidity [66,77] (Fig 1D). Increased cortical pulling forces upon zyg-8(RNAi) could be thought to alternatively explain the observed increase in microtubule bending. However, such a view contradicts our DiLiPop experiments (Fig 7B and 7C). Additionally, cortical dynein tracking revealed that changes in dynein behaviour were insufficient to account for a large increase in pulling forces (S3 Table). While we did observe increased dynein density—consistent with more microtubules reaching the cortex upon ZYG-8 depletion—this did not translate into increased pulling events, as GPR-1/2 association in the force-generating complex and dynein engagements remain the limiting factors. The increase in cortical dynein lifetimes aligned with the prolonged residency times of cortical microtubules, without suggesting an additional contribution. We conclude that ZYG-8 primarily acts by increasing microtubule rigidity rather than by regulating cortical pulling force generation.

Our data suggest that, while ZYG-8 contributes to microtubule dynamics, its primary role in regulating spindle positioning and orientation is mediated through modulation of microtubule rigidity, particularly under conditions of elevated cortical pulling forces (Fig 8A–8D). This mechanical function likely represents the dominant mechanism accounting for the observed phenotypic effects, with alterations in microtubule dynamics playing a secondary role. Microtubule rigidity is more substantially reduced in zyg-8(or484ts) mutants compared to zyg-8(RNAi) embryos (Figs 5, 6, and S9). This correlates with a greater decrease in centring forces (as measured by positional fluctuation assays) (Table 1). These findings align with the gradual phenotypes observed in spindle-pole oscillations, and spindle final position or orientation (Figs 1 and 8). Why ZYG-8 plays a greater role in regulating microtubule rigidity rather than microtubule dynamics. This is likely because ZYG-8 and EB compete for the same microtubule binding site at the corner of four tubulin dimers [39,121,122]. However, EB binds preferentially to the GTP cap at growing ends [121], while Doublecortin-family proteins bind in vitro to the whole lattice [123]. In vivo, DCX was excluded from microtubule ends [124]. An exclusion of ZYG-8 from microtubule ends in C. elegans zygote could account for differences in the efficiency of its dual role in regulating microtubule dynamics and rigidity. This may be particularly relevant in specific cellular contexts where proteins essential for microtubule growth or nucleation (e.g., γ-tubulin ring complexes) are absent or present at low levels. In neurons, proteins from the Doublecortin family are highly expressed and regulate microtubule rigidity [37,100]. These proteins may also contribute to microtubule nucleation and/or growth, consistent with their in vitro capabilities [38,91,92,125]. For instance, collateral branching in hippocampal neurons was significantly delayed following DCX depletion, possibly reflecting impaired microtubule dynamics [126]. Overall, ZYG-8^DCLK1 performs multiple microtubule-related functions in a context-dependent manner. We propose that the control of microtubule flexural rigidity represents a novel function during cell division and possibly beyond. While our data strongly implicate microtubule rigidity in spindle misorientation in zyg-8(or484ts) mutants, we cannot rule out the possibility of additional contributions from mild perturbations in microtubule dynamics.

Looking ahead, it will be important to explore the mechanisms by which ZYG-8 rigidifies the microtubules. Several mechanisms have been proposed, primarily in the context of neurons. Here, we draw on the body of research on the DCX protein, which shares the same two microtubule-binding domains as DCLK1^ZYG-8. Besides, these two doublecortin domains are highly conserved in ZYG-8 [54]. (1) Microtubules can be rigidified by hindering the sliding of protofilaments through the binding of MAPs across these protofilaments. DCX binding on the microtubules is compatible with this mechanism since it reinforces lateral and longitudinal dimer couplings [38]. (2) Rigidification can also occur by bundling multiple microtubules together. Previous studies revealed such a mechanism for the Doublecortin family in vitro and in vivo [57,90,91,100,125,127,128]. In particular, zyg-8(or484ts) mutants displayed a perturbed bundle organisation of microtubules in touch receptor neurons [100]. However, in the C. elegans zygote, electron microscopy failed to reveal microtubule bundles in the spindle or astral microtubules close to the poles [129]. Therefore, we disfavour such a mechanism. (3) Microtubule rigidity is also regulated through post-translational modifications (PTM) [130–134]. Notably, a study showed that DCX expression increases tubulin polyglutamylation in neurons [135]. However, the ttll-4(-);ttll-5(-);ttll-11(-) triple mutants, which prevented tubulin glutamylation, did not show any defect in embryonic viability. It suggests that microtubule glutamylation may not significantly affect the zygote cell division [136]. (4) It was previously shown that damages along microtubule lattice soften microtubules [137,138]. Thus, self-repair of local damages may also help restore the rigidity or stabilise the curved region [139]. Interestingly, the number of microtubule defects could notably correlate with the microtubule growth rate [140,141]. The latter is particularly high in nematodes [142]. If a high occurrence of holes exists in the C. elegans embryo, ZYG-8 could contribute to healing as the Doublecortin family also acts as tubulin oligomer chaperones [92]. However, to date, there has been no evidence of microtubule lattice defects in C. elegans zygote. For example, embryos with EBP-2 fluorescent labelling did not show stationary fluorescent spots along the astral microtubules [48,67]. The various mechanisms proposed above, in which ZYG-8 could play a key role in regulating microtubule flexural rigidity, are not mutually exclusive; several of them may be combined. ZYG-8 rigidifying the microtubule by coupling the protofilaments appears to us the most plausible mechanism currently.

To explore whether other microtubule rigidity regulators contribute to spindle positioning in C. elegans embryos, we examined two candidates: PTL-1, the sole nematode Tau homolog [117,118], and SPD-1^PRC1 [36]. PTL-1 depletion resulted in modest increases in oscillation amplitudes (34% anterior, 13% posterior, S14A Fig) and slight frequency decreases (9% anterior, 11% posterior, S14B Fig), alongside a 60% reduction in centring-to-damping corner frequency (Table 1). While these alterations resembled zyg-8(RNAi) phenotypes, their reduced magnitude likely reflects PTL-1’s limited zygotic expression [119]. Similarly to zyg-8(RNAi) treatment, we observed no defects in spindle final position or orientation following ptl-1(RNAi) (S14C and S14D Fig). Conversely, SPD-1 depletion produced no detectable changes in spindle oscillations (S14E and S14F Fig) or spindle final position and orientation (S14G and S14H Fig). This is consistent with SPD-1’s spindle midzone localisation and absence from astral microtubules [59,143,144]. These comparative analyses highlight the unique role of ZYG-8 among microtubule rigidity regulators in controlling external forces on the spindle during C. elegans zygote division.

Focusing on mitotic spindle positioning in the C. elegans zygote, the current study both complements and departs from previous investigations. Firstly, we show that controlling microtubule flexural rigidity participates in the spindle choreography. In particular, this control helps prevent the spindle poles from approaching the cell periphery too closely during anaphase (Fig 8A, 8B, and 8G). Consistently, we observe increased amplitudes of spindle pole oscillations following ZYG-8 or PTL-1 depletion (Figs 1C and S14A). This control through microtubule rigidity is complementary to the well-established regulation involving microtubule dynamics [25,67,72,80]. Secondly, our study underlines that cortical pushing force contributes to spindle positioning during anaphase, although in a manner distinct from that of cortical pulling forces. Cortical pulling forces are dominant, and their asymmetry, controlled by PAR polarity, ensures spindle posterior displacement [74,78,145–148]. To date, cortical pushing forces have been proposed to contribute to spindle centring during metaphase [25,27,29]. During anaphase, although restoring forces are known to be necessary for spindle oscillations, the role of cortical pushing force as a restoring force was not demonstrated [66]. Besides, it is often assumed that cortical pushing force plays a minimal role during anaphase, due to the high magnitude of cortical pulling forces and the dampening effect of microtubule buckling [68,79,149]. However, we report that weakening the cortical pushing forces, as observed in zyg-8(or484ts) mutants, hinders spindle alignment along the AP axis (Fig 8B). This force imbalance compromises the final position and orientation of the spindle (Fig 8C–8D). Interestingly, decreasing cortical pulling force by depleting GPR-1/2 in zyg-8(or484ts) mutants rescues the spindle misorientation phenotype (S13 Fig). Together, these findings highlight the importance of sufficiently strong cortical pushing forces during anaphase, which contribute to proper force balance and ensure accurate spindle orientation at the end of mitosis. Thirdly, we show that in zyg-8(or484ts) mutants, astral microtubules are still contacting the opposite cortex (Fig 5A and S5 and S6 Movies). This rules out a model in which spindle misorientation arises from a failure of microtubules to reach the cortex, as previously suggested [54]. Overall, the astral microtubule network functions like a spring, restricting the spindle transverse motion away from the centre in C. elegans zygotes [25,27,77]. From this perspective, microtubule flexural rigidity emerges as a key physical parameter controlling the magnitude of centring forces. This mechanical contribution is essential for generating cortical pushing forces strong enough to counteract the intense cortical pulling forces that act during anaphase.

We foresee that the essential role for pushing extends beyond the nematode’s first division and may be relevant to other asymmetric divisions and maybe further. It is well established that the pulling forces dominate spindle orientation. It is accounted for by the canonical model based on the molecular machinery composed of the cortical Gα-LGN^GPR-NuMA^LIN-5 complex and the cytoplasmic dynein motor [10,150]. However, less attention has been paid to the contribution of the pushing forces to spindle orientation [151]. The dependence of spindle orientation on cell shape, which is often neglected, suggests that the pushing force could play a significant role in this process [152–156]. In other systems, some studies found that the spindle did not align with the dynein-enriched cortical zones [157–160]. Furthermore, spindle orientation and centring can occur through cortical-dynein-independent mechanisms, such as those involving MARK2/Par1 kinase [29,161]. Consequently, the traditional canonical model alone cannot account for spindle orientation in various contexts. Consistent with these observations, we propose that the cortical pushing force contributes to this phenomenon and that the flexural rigidity of microtubules must be sufficiently high. Notably, the orientation of the mitotic spindle defines the direction in which a cell divides. Therefore, many biological processes, ranging from germline stem cell division to epithelial tissue homeostasis and regeneration, require correct mitotic spindle orientation. Defects in spindle orientation have been associated with various diseases, such as organogenesis and morphogenesis failures, or cancer [162–169].

By uncovering ZYG-8’s dual role in C. elegans mitosis, we can now question its contribution to zygote meiosis, especially through controlling microtubule flexural rigidity. A recent study highlighted that ZYG-8 promotes meiotic spindle stability and suggested that it may regulate microtubule dynamics and motor-driven forces [56]. We wonder whether some of the perturbations that the authors observed upon zyg-8 targeting could be attributed to the deregulation of microtubule rigidity. Notably, they found that most zyg-8(or484ts) meiotic spindles were significantly bent at the restrictive temperature. Additionally, some images of the mutant showed microtubules exhibiting high local curvatures. Similar perturbations were also observed following ZYG-8 depletion using the Auxin-inducible degradation system. In light of these observations, it is plausible that ZYG-8 contributes to the assembly or/and stability of meiotic spindles in the zygote by stiffening microtubules.

In conclusion, our findings regarding the pivotal role of ZYG-8^DCLK1 in spindle positioning and orientation, primarily through rigidifying microtubules during mitosis, offer not only a deeper understanding of the fundamental mechanisms governing cell division but also highlight significant implications for cancer therapy. Given that accurate spindle positioning is essential for maintaining the balance between cell proliferation and differentiation, our results may have broader relevance for understanding how disruptions in these processes contribute to tumour initiation and progression [170–172]. The frequent deregulation of DCLK1 in various solid tumours (e.g., those from the colon, pancreas, kidney and breast) suggests its involvement in a common process such as cell division [173–176]. Additionally, the proteins of the Tau family are also found to be deregulated or mutated in cancers [177–179]. While our study focuses on fundamental mechanisms, these observations raise the possibility that microtubule rigidity could represent an underexplored aspect of cancer cell biology.

Materials and methods

M4- Fluorescence microscopic imaging conditions

The various microscopy setups used are presented in Table 2. Embryos were imaged at 23 °C, except otherwise stated. Images were stored using the Omero software [183].

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Four microscopy setups used in the present study.

https://doi.org/10.1371/journal.pgen.1012196.t002

M10- Landing assay to study astral microtubule cortical contacts

The entire microtubule lattices were fluorescently labelled using GFP::TBB-2, YFP::TBA-2, or mCherry::tubulin to view them in all their states. We used the same approach as the one previously implemented in [80] and available on Zenodo (doi: 10.5281/zenodo.4552485); we combined a Kalman filtering [188] (gain set to 0.5 and initial noise to 0.05) and u-track [192] whose parameters are listed in Table 3. S1–S4 Movies show representative examples of microtubule contact tracking under the four conditions we studied. After tracking the microtubule contacts, we analysed them using a two-step approach to examine distinct aspects of microtubule-cortical interactions.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. U-track parameters used in the landing assay (* for GFP::TBB-2 labelling; ** for mCherry::tubulin labelling; *** for GFP::DHC-1 labelling).

https://doi.org/10.1371/journal.pgen.1012196.t003

1. Basic quantification. We quantified the number of microtubule cortical contacts per frame, averaged them over a 10-second window, and normalised them by the embryo area. This provided an instantaneous count of microtubule contacts throughout mitosis for each embryo, which we then averaged across embryos within each experimental condition (e.g., Fig 4C). We also computed the duration histograms of astral microtubule contacts per embryo and averaged these histograms across embryos for each condition (e.g., Fig 4E). This analysis assessed whether ZYG-8 contributes to microtubule stabilisation by preventing, e.g., cytoplasmic depolymerisation or cortical catastrophe events.
2. DiLiPop (Distinct Lifetime Population) analysis. This statistical analysis of the microtubule cortical contact durations reveals the short-lived and long-lived populations of astral microtubules at the cortex associated with pulling and pushing events, respectively [80]. This analysis evaluated ZYG-8’s role in cortical force generation. By globally fitting the embryo-set duration distributions of a given condition, we obtained the characteristic lifetimes and densities of the cortical pulling and pushing events (e.g., Fig 7B-7E). Error bars of the lifetimes and densities were obtained by bootstrapping as explained in [80]. These densities of events are distinct from instantaneous contact density and thus independent of the cortical contact durations (except in the case of anaphase pulling events due to all force generators being involved in pulling). We noticed that the lifetimes of pulling and pushing events in the present study differed from those in our previous work [80]. This discrepancy likely stemmed from differences in microtubule labelling (GFP::TBB-2 vs. YFP::TBA-2) and variations in microscope sensitivity. Therefore, we consistently compared results from experiments conducted within the same time frame (within two months).

M11- Mapping astral microtubules and characterising their local curvatures and their tortuosity

We stained microtubules using an anti-α-tubulin antibody, clone DM1A raised in mouse, Alexa Fluor 488 conjugate (Sigma-Aldrich 16–232), diluted at 1/500 in PBS-Tween (0.2%) with BSA (1%). It targets several ɑ-tubulin isoforms in various species. Antibody incubation was performed overnight at 4 °C. Slides were then rinsed for 10 minutes in PBS-Tween (0.2%), twice, in a light-free environment. We could then visualise microtubules all along their length. We identified the z-section(s) as the more appropriate for astral microtubule segmentation. For most embryos, we found a single z-section per embryo for which both centrosome asters were visible. For some embryos having the two centrosomes on distinct z-sections, we considered two planes of interest.

Five-step image processing pipeline to map the local curvatures of the astral microtubules (S8D Fig). (1) We used the planes above and below the plane of interest to perform an extended depth of field projecting the 3 z-sections into a single image (S8E Fig) [193]. We thus better visualise microtubules all along their length. (2) We applied the Orientation Filter Transform (OFT, number of orientations = 20; radius = 5 pixels) since it is effective in enhancing filamentous pattern against noises [20,194] (S8F Fig). (3) We then used the interactive machine-learning tool, Ilastik, to segment the astral microtubules (S8G Fig). The dedicated paragraph below provides more detailed information about our segmentation. (4) We generated the skeleton of the filament traces from the binary segmentation, taking advantage of SOAX (S8H Fig). Further information about the use of SOAX is available in a dedicated paragraph below. (5) We finally measured the local curvatures of astral microtubules pixel-by-pixel from the skeleton coordinates, enabling mapping of the local curvatures (S8I Fig). For a set of three neighbouring points sampling the curve, the local curvature corresponds to the inverse of the radius of the circle going through the mid-point and best fitting the two others. Local curvatures were computed along microtubules according to the three-point method [195] using a code retrieved on Matlab Central [196]. In a nutshell, microtubules are viewed as parametrised curves; at each point, the x and y coordinates are represented as second-order polynomials of the curvilinear abscissa passing through the point and its two neighbours. The curvature is then computed with the Frenet-Serret formula.

Ilastik: We initially trained Ilastik using the conditions zyg-8(RNAi) and its control RNAi during metaphase. This involved approximately 10 manual annotations of microtubules across three embryos for each condition to cover various intensities, contrasts and microtubule shapes. We then used this trained model, referred to as #1M (M for metaphase, Table 4), to segment, in a semi-supervised way, the astral microtubules in metaphase images. Next, we trained the model to segment anaphase images of zyg-8(RNAi)-treated embryos and their controls. Indeed, denser microtubule networks in anaphase required additional annotations. This led to the upgraded model #1A (Table 4) that we used to segment anaphase images. We then analysed zyg-8(or484ts) mutants and untreated embryos in metaphase. The model was further improved by manually annotating three mutants, resulting in model #2M (Table 4), which we used to segment images of both mutants and untreated embryos in metaphase. We applied a similar process for segmenting anaphase images of mutants and untreated embryos, enhancing model #2M with a few annotations of anaphase mutants to get the model #2A (Table 4). The table below summarised the Ilastik models used to produce the data displayed in the different figures.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Summary of the different Ilastik models used for segmenting microtubules.

https://doi.org/10.1371/journal.pgen.1012196.t004

SOAX: We extracted the centrelines from the thick filaments utilising SOAX mostly with standard parameters. We made the few specific adjustments highlighted in Table 5. These five adjusted parameters were crucial for accurately delineating the curved regions of the microtubules. The settings used ensured that the centrelines faithfully represented the shapes of the microtubules, as confirmed through visual validation. Additionally, manual corrections could be applied at this stage, especially when two microtubules were mistakenly merged into one.

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. SOAX parameters used for microtubule skeletonization.

https://doi.org/10.1371/journal.pgen.1012196.t005

Microtubule shape quantification: We analysed only the astral microtubules that were at least 4 µm long after segmentation, as shorter ones often represented fragments. This approach was appropriate for comparing conditions without seeking absolute quantification. We conducted three complementary approaches.

We computed the 95^th percentile curvature for each microtubule to capture faithfully the highest curvatures (S8B Fig) and then calculated the median of these values across all microtubules in each z-projected stack. Each stack exhibited 15–74 microtubules. We finally compared the set of medians between two conditions (Fig 6E and 6F).

1. For each z-projected stack, we calculated the normalised histogram of local curvatures with a bin size of 0.05 µm^-1 (S8A and S8J Fig) and averaged these histograms per condition (Fig 6A–6D). The curvatures, expressed in per micrometre, range from 0.025 to 1 µm^-1, corresponding to curvature radii from 40 µm to 1 µm, respectively. 40 µm could be interpreted as mostly straight astral microtubules.
2. We measured the tortuosity for each microtubule, defined as the curvilinear distance divided by the end-to-end distance (S8C Fig). Straight microtubules have tortuosity values near one, while bent microtubules show higher values. Then, we calculated the medians of tortuosity values across all microtubules in each z-projected stack, and compared the set of medians between two conditions (Fig 6G and 6H).

Supporting information

Acknowledgments

Strains TH27 and TH30 were kindly offered by Prof Anthony A. Hyman. We thank Dr Grégoire Michaux for supplying the feeding clone library and for technical support. Strains MCP347, MCP361 & MCP554 were generated by SEGiCel (SFR Santé Lyon Est CNRS UAR 3453, Lyon, France) with the support of CNRS and IBiSA. We acknowledge funding for the microscopes used in this study: the spinning disc microscope by the CNRS, Rennes Metropole (AIS 16C0400, H. Bouvrais) and Region Bretagne (AniDyn-MT grant, J. Pécréaux); the Zeiss Airyscan 2 confocal microscope via INCA (ITMO cancer, Grant for the acquisition of equipment for cancer research), Rennes Métropole (AIS), EU FEDER (CPER B2S) and Biosit; and the Nikon Nsparc confocal microscope through grants from IT Cancer of INSERM (Framework of the 2021-2030 Cancer Control strategy, N°24CQ009-00), Agence Nationale de la Recherche (ANR-23-CE13-0003, GroBluRe, G. Michaux), and Rennes Métropole (AIS 23C0774, H. Boukhatmi). Some Caenorhabditis elegans strains were provided by the Caenorhabditis Genetics Center (CGC), which is funded by National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440; University of Minnesota). We thank S. Dutertre and X. Pinson of the Microscopy Rennes Imaging Center (MRic, BIOSIT, Biogenouest) for their assistance. MRic is part of the national infrastructure France-BioImaging supported by the French National Research Agency (ANR-10-INBS-04). We thank the Pécréaux lab for their support, and the Gibeaux, Giet, Huet, Michaux, and Tramier labs for helpful discussions. We also thank L. Chesneau for assistance in designing the CRISPR strains, and A. Pacquelet, B. Lacroix and C. Heligon for their careful reading of the manuscript.