Aurora A, MCAK, and Kif18b promote Eg5-independent spindle formation
- 1.8k Downloads
Inhibition of the microtubule (MT) motor protein Eg5 results in a mitotic arrest due to the formation of monopolar spindles, making Eg5 an attractive target for anti-cancer therapies. However, Eg5-independent pathways for bipolar spindle formation exist, which might promote resistance to treatment with Eg5 inhibitors. To identify essential components for Eg5-independent bipolar spindle formation, we performed a genome-wide siRNA screen in Eg5-independent cells (EICs). We find that the kinase Aurora A and two kinesins, MCAK and Kif18b, are essential for bipolar spindle assembly in EICs and in cells with reduced Eg5 activity. Aurora A promotes bipolar spindle assembly by phosphorylating Kif15, hereby promoting Kif15 localization to the spindle. In turn, MCAK and Kif18b promote bipolar spindle assembly by destabilizing the astral MTs. One attractive way to interpret our data is that, in the absence of MCAK and Kif18b, excessive astral MTs generate inward pushing forces on centrosomes at the cortex that inhibit centrosome separation. Together, these data suggest a novel function for astral MTs in force generation on spindle poles and how proteins involved in regulating microtubule length can contribute to bipolar spindle assembly.
KeywordsEg5 Kif15 Aurora A Kif18B MCAK Spindle
The bipolar spindle is a microtubule (MT)-based structure required for successful chromosome segregation during mitosis. Assembly of the bipolar spindle requires tight regulation of a wide variety of microtubule-associated proteins (MAPs), including MT motors from the kinesin family of proteins (Walczak and Heald 2008). An essential and highly conserved protein for bipolar spindle assembly is kinesin-5 (Eg5 in humans). Eg5 forms a unique tetrameric configuration, hereby enabling it to crosslink and slide antiparallel MTs apart and thereby driving centrosome separation and bipolar spindle assembly (Kashina et al. 1996; Kapitein et al. 2005). Inhibition or depletion of Eg5 results in a mitotic arrest and subsequent cell death due to the formation of monopolar spindles in nearly all organisms tested (Sawin et al. 1992; Blangy et al. 1995; Mayer et al. 1999; Ferenz et al. 2010). Therefore, Eg5 is an attractive anti-mitotic target for cancer therapy (Rath and Kozielski 2012).
Recent studies reported the existence of redundant pathways, cooperating with Eg5 to drive centrosome separation and bipolar spindle assembly. In human cells, kinesin-12 (Kif15/Hklp2 in humans) was identified to cooperate with Eg5 in bipolar spindle assembly (Tanenbaum et al. 2009; Vanneste et al. 2009). Ectopic overexpression of Kif15 bypasses the requirement for Eg5 in bipolar spindle assembly (Tanenbaum et al. 2009). In addition, we and others have shown that human cells, treated with Eg5 inhibitors can easily acquire the ability to build a bipolar spindle in the absence of Eg5 activity, but become dependent on Kif15 for bipolar spindle formation (Raaijmakers et al. 2012; van Heesbeen et al. 2013; Sturgill and Ohi 2013; Ma et al. 2014; Sturgill et al. 2016).
To identify genes that are required for Eg5-independent bipolar spindle assembly, we performed a genome-wide small interfering RNA (siRNA) screen in HeLa and HeLa-derived Eg5-independent cells (EICs, (Raaijmakers et al. 2012). We searched for genes that specifically arrested EICs in mitosis, using a high content, fixed cell assay. We identified the mitotic kinase Aurora A and two kinesins that regulate MT dynamics, MCAK (Kif2C/kinesin-13) and Kif18b (kinesin-8), to be essential for bipolar spindle assembly in EICs. Our data reveals two novel mechanisms that are required for Eg5-independent bipolar spindle assembly and uncovers three potential targets for combination therapy with Eg5 inhibitors.
A genome-wide siRNA screen identifies three genes required for bipolar spindle assembly in Eg5-independent cells
MCAK, Kif18b, and Aurora A are essential for bipolar spindle assembly in EICs and in cells with reduced Eg5-activity
Both Eg5 and Aurora A inhibitors are promising anti-cancer drugs that are currently investigated in a number of clinical trials (Lens et al. 2010; Janssen and Medema 2011). The dramatic increase in monopolar spindles in cells treated with Aurora A siRNA and Eg5-inhibitors made us wonder if combined treatment of Eg5 inhibitors with Aurora A inhibitors might lead to a synergistic effect in blocking bipolar spindle formation and as a long-term consequence, in decreased cell proliferation. To test this, we treated both parental and EICs with the selective Aurora A inhibitor MLN8054 (Manfredi et al., 2007). Similar to siRNA treatment, concentrations up to 300 nM MLN8054 did not affect bipolar spindle formation in parental cells, but efficiently blocked bipolar spindle formation in EICs (Fig. 2d) and in parental cells with reduced Eg5 activity (Fig. 2e). In addition, long-term treatment with low doses of MLN8054 (50–100 nM) blocked proliferation in EICs, while similar doses did not affect EICs after removal of STLC (Fig. 2f). Furthermore, long-term treatment of parental HeLa cells with low doses of Eg5 and Aurora A inhibitors that efficiently blocked the formation of bipolar spindles (Fig. 2e) also fully blocked proliferation and induced apoptotic cell dead as shown by the increased levels of cleaved poly(ADP-ribose) polymerase (PARP) by caspase-3 (Supplementary Fig. S4). These data suggest that combining Eg5 and Aurora A inhibitors might have increased efficacy versus monotherapy. In addition, it might prevent the development of resistance to Eg5 inhibitors.
MCAK, Kif18b, and Aurora A are required for bipolar spindle maintenance in the absence of Eg5 activity
To test if the action of MCAK, Kif18b, and Aurora A is restricted to the assembly of the bipolar spindle during prometaphase, or if their function is also required for the maintenance of the metaphase bipolar spindle, we arrested parental HeLa cells in metaphase using the proteasome inhibitor MG132 and subsequently inhibited all Eg5 activity using a high dose of STLC (Tanenbaum et al. 2009; van Heesbeen et al. 2014). As we have shown previously, control cells maintain a bipolar spindle after treatment with STLC, due to the action of Kif15 (Tanenbaum et al. 2009). Similar to Kif15 depletion, MCAK, Aurora A, and to a lesser extent Kif18b depletion, results in collapse of the bipolar spindle upon Eg5 inhibition (Fig. 3c). Curiously, we find a difference in the number of bipolar spindles that collapsed in the MCAK-depleted cells versus the Kif18b-depleted cells. While MCAK and Kif18b act together to control astral MT dynamics (Tanenbaum et al. 2011a; Stout et al. 2011), previous studies showed that additional, Kif18b-independent roles for MCAK at kinetochores and spindle poles exist (Walczak et al. 2013) and likely contribute to the differences we observed in the number of cells in which we observed collapse of the bipolar spindle. These results indicate that the action of these proteins is not restricted to the assembly of the bipolar spindle, but is also required for maintenance of the metaphase bipolar spindle.
Excessive astral microtubule nucleation blocks centrosome separation and bipolar spindle assembly
Regulation of MT dynamics during mitosis is a tightly regulated process (Howard and Hyman 2007). The MT motors from the kinesin-13 and kinesin-8 family have both been shown to control MT depolymerization (Walczak et al. 2013). MCAK (kinesin-13) is a non-processive motor that diffuses along the MT lattice to reach the ends of MTs (Helenius et al. 2006) and contains an internal motor domain required for its MT depolymerizing activity (Hunter et al. 2003). Besides that, MCAK can also track the growing plus-ends of MTs through a direct interaction with EB1 (Moore et al. 2005; Lee et al. 2008). Previous studies implicated a role for kinesin-13 family members in mitosis, including the regulation of spindle bipolarity (Ganem and Compton 2004; Kollu et al. 2009) and positioning of the mitotic spindle (Rankin and Wordeman 2010). However, the mechanisms by which MCAK contributes to Eg5-independent bipolar spindle assembly are still unclear. In contrast to kinesin-13, kinesin-8 motors contain a N-terminal motor domain and have shown to be processive, plus-end-directed motors (Gupta et al. 2006; Varga et al. 2006; Mayr et al. 2007), that depolymerize MTs at the plus-ends in a length-dependent manner (Varga et al. 2006; Varga et al. 2009). Kif18b has also been shown to accumulate at MT plus-ends through a direct interaction with EB1 (Tanenbaum et al. 2011a; Stout et al. 2011). In addition, Kif18b interacts with MCAK, hereby promoting the plus-end accumulation of each other (Tanenbaum et al. 2011a). While MCAK regulates MT depolymerization at different locations in the cell, including kinetochores, centrosomes, and astral MTs (Andrews et al. 2004; Kline-Smith et al. 2004; Tanenbaum et al. 2011b), the localization of Kif18b is negatively regulated by Aurora kinases and has only been found at the plus-tips of astral MTs (Tanenbaum et al. 2011a). Taking into account that Kif18b localization is restricted to astral MTs and MCAK is a non-processive motor, it is unlikely that they act in sliding anti-parallel MTs. However, considering that MCAK and Kif18b both regulate astral MT depolymerization by forming a mitosis-specific complex (Tanenbaum et al. 2011a), we wondered if astral MT length control could influence bipolar spindle assembly.
Phosphorylation of S1169 by Aurora A is required to target Kif15 to the spindle
The role of Aurora A in centrosome separation and bipolar spindle formation is controversial. Early studies in Drosophila showed that mutations in Aurora A led to centrosome separation defects and monopolar spindle formation (Glover et al. 1995). Consistently, studies in mouse embryonic fibroblasts showed that Aurora A deletion led to the formation of monopolar spindles (Cowley et al. 2009). In contrast, Aurora A deletion in chicken DT40 cells led to the formation of small bipolar spindles (Hégarat et al. 2011) and studies in human cells observed a wide variety of phenotypes, including chromosome misalignments, multipolar spindles, and monopolar spindles (Marumoto et al. 2003; Hoar et al. 2007). The discrepancies observed between model systems could be explained by the methods used to inactivate or deplete Aurora A from cells or by different contributions of parallel pathways involved in centrosome separation and bipolar spindle assembly (Smith et al. 2011). Despite the fact that a wide variety of Aurora A substrates have been identified (Lens et al. 2010; Hochegger et al. 2013), clear downstream targets involved in centrosome separation and bipolar spindle assembly are poorly understood. Eg5 was described to be phosphorylated by Aurora A in Xenopus (Giet et al. 1999), but since we identified Aurora A in an Eg5-independent background, this cannot be its only target for its function in bipolar spindle assembly. Furthermore, centrosome maturation and MT nucleation might also indirectly affect centrosome separation, although these function are likely not affected in our system, since we did not observe major defects in prophase centrosome separation in both normal cells and EICs (Fig. 3a, b). At last, the fact that we observed a rapid bipolar spindle collapse in cells in which we blocked both Eg5 and Aurora A activity simultaneously (Fig. 3c) points towards a target that is also involved in the maintenance of the bipolar spindle.
Next, we tested if overexpression of the Kif15 mutants could bypass the requirement of Eg5 in cells where we depleted endogenous Kif15 (Tanenbaum et al. 2009). As expected, overexpression of the different constructs did not affect bipolar spindle assembly in the absence or presence of MLN8054 (Fig. 5c, black and orange bars). However, while full Eg5 inhibition efficiently blocked bipolar spindle formation in control-transfected cells (Fig. 5c, middle bar), overexpression of wild-type and S1169D-mutated Kif15 fully restored bipolar spindle formation in Eg5-inhibited cells (Fig. 5c, yellow bars). In contrast, expression of S1169A-mutated Kif15 did not restore bipolar spindle formation, indicating that this mutant is not capable to bypass the Eg5 requirement (Fig. 5c, yellow bars). Finally, when we combined Eg5 inhibition with partial Aurora A inhibition (500 nM MLN8054), we observed a pronounced decrease in the amount of bipolar spindles when we overexpressed wild-type Kif15 (Fig. 5c, red bars). Strikingly, the S1169D mutant was still partially active and about 25 % of the cells formed bipolar spindles upon combined inhibition of Eg5 and Aurora A (Fig. 5c, red bars). These results indicate that Aurora A directly regulates Kif15 by targeting it to the spindle during mitosis through phosphorylation on S1169.
Here, we performed a genome-wide siRNA screen in parental and EICs cells to identify novel factors involved in Eg5-independent bipolar spindle formation. Using our setup, we identified three genes required for bipolar spindle assembly in EICs. We show that the microtubule motors MCAK and Kif18b are required for bipolar spindle assembly in EICs and normal cells with reduced Eg5-activity. In contrast to Eg5, which directly drives bipolar spindle assembly by sliding antiparallel MTs apart (Kashina et al. 1996; Kapitein et al. 2005; Tanenbaum et al. 2009), we show evidence that the contribution of MCAK and Kif18b to bipolar spindle assembly is likely mediated by their function in regulating the length and number of astral MTs during mitosis. Although we cannot rule out additional contributions of MCAK and Kif18b in bipolar spindle assembly, we hypothesize that in the absence of either MCAK or Kif18b, the tight balance in astral MTs nucleation and depolymerization is lost and excessive astral MTs generate inward pushing forces on centrosomes when these MTs collide with the cortex (Fig. 5e). Under normal conditions, these forces are not sufficient to counteract outward forces, but when Eg5 activity is compromised, growing astral MTs generate forces on the cortex to counteract the remaining centrosome separation forces by Eg5 and Kif15. Our results indicate that MCAK and Kif18b have an equal and non-redundant contribution in regulating astral MT dynamics. However, we did observe a more rapid collapse of the preassembled bipolar spindle after MCAK depletion compared to Kif18b, suggesting possible Kif18b-independent involvement of MCAK. Furthermore, interfering with cortical tension could rescue bipolar spindle assembly for both MCAK and Kif18b only to a limited extend. We therefore cannot exclude that additional functions of MCAK and Kif18b contribute to bipolar spindle assembly. In line with that, MCAK was previously shown to be involved in the regulation of KT-MT turnover, and this function might also contribute to bipolar spindle assembly and maintenance (Andrews et al. 2004; Kline-Smith et al. 2004). In addition to MCAK and Kif18b, we identified Aurora A in our screen to be required for bipolar spindle assembly in EICs. Although Aurora A was previously identified to act synergistically lethal with Eg5 inhibitors, its downstream targets for controlling centrosome separation are poorly understood (Ma et al. 2014). We now show that Aurora A functions in bipolar spindle formation by controlling the localization and activity of Kif15. Spindle localization of Kif15 is decreased upon inhibition of Aurora A. While Kif15 function is under normal conditions redundant for bipolar spindle assembly, its function is essential for EICs and cells with reduced Eg5-activity (Tanenbaum et al. 2009; Raaijmakers et al. 2012; van Heesbeen et al. 2014), explaining the high sensitivity for Aurora A inhibition under this condition. Due to conflicting results from recent studies (Sturgill and Ohi 2013; Drechsler et al. 2014; Sturgill et al. 2014), it is currently unclear how Kif15 functions at the molecular level and whether Kif15 acts on antiparallel MTs in the spindle. Interestingly, both Aurora A and Kif15 require TPX2 for their function and depletion of TPX2 prevents spindle targeting of Kif15. How phosphorylation of S1169 contributes to spindle targeting of Kif15 is still unclear, but it might affect the interaction of TPX2 with the C-terminal leucine zipper (Tanenbaum et al. 2009), or affect the previously proposed non-motor MT-binding domain of Kif15 (Sturgill et al. 2014).
Although we find direct phosphorylation of Kif15 by Aurora A, it has likely more targets required for bipolar spindle assembly. We observed that cells expressing high levels of S1169A Kif15 were still able to form bipolar spindles. This could indicate that additional Aurora A phosphorylation sites on Kif15 are present. Aurora A also contributes to MT nucleation and KT-MT stability (Kinoshita et al. 2005; Ertych et al. 2014), which has been shown to contribute to bipolar spindle assembly and maintenance (Sturgill and Ohi 2013; van Heesbeen et al. 2014). Most likely, a combination of regulating MT dynamics and kinesins like Kif15, explains the synergistic effect we see after combined inhibition of Eg5 and Aurora A.
Both Eg5 and Aurora A inhibitors are currently being tested as potential anti-cancer drugs in clinical trials (Rath and Kozielski 2012; Malumbres and Pérez de Castro 2014). In order to enhance efficacy, we propose that combination therapy of Eg5 and Aurora A inhibitors might be beneficial because of three main reasons. First, the combined treatment shows a very strong synergistic effect in the formation of monopolar spindles, even when both proteins are only partially inhibited. Second, the development of resistance mechanisms for Eg5 inhibitors (Tanenbaum et al. 2009; Raaijmakers et al. 2012) will likely be prevented by combining the Eg5 and Aurora A inhibitors. And last, there are currently no Kif15 inhibitors available, which makes Aurora A inhibitors currently the most attractive candidate to increase the efficacy for Eg5 inhibitors. Taken together, we unraveled new mechanisms for bipolar spindle assembly that might have promising translational applications.
Screen setup, analysis, and normalization
The human ON-TARGETplus siRNA SMARTpool library (Dharmacon) was used for the primary screen. The siRNAs for the secondary screen were manually picked and re-tested. For the deconvolution screen, the four single siRNAs of the SMARTpool were tested separately. The primary screen was performed in duplicate, the secondary and deconvolution screen were screened in triplicate.
For the primary and secondary screen, siRNA libraries were aliquoted in a 384-well format using a Sciclone liquid handling robot (Caliper). Deconvolution screen was performed in a 96-well format. A final concentration of 20 nM siRNA per well was used. Per transfection, 0.075 μl RNAiMAX (Invitrogen) and 10 μl Opti-MEM (GIBCO) were added to the siRNA and incubated for about 20 min. One thousand five hundred cells diluted in 40 μl media were added to the wells after incubation of the transfection reagents, using a MultiDrop Combi bulk dispenser (Thermo).
After 48 h of culturing, the cells were fixed for 10 min using a final concentration of 4 % formaldehyde (3× formaldehyde in PBS was added to the wells). Fixation reagent was added using a Multidrop Combi bulk dispenser (Thermo), primary and secondary antibodies were added using the Sciclone (Caliper), and all washing step was performed in an AquaMax 2000 plate washer (MDC).
After staining of the wells, the mitotic index of the wells was analyzed using a Cellomics Arrayscan VTI (Thermo Scientific) using a 10× (0.50 NA) objective. Four images were acquired per well. Image analysis was performed using Cellomics Target Activation Bioapplication (Thermo Scientific). Cells were identified based on the DAPI staining and were scored to be mitotic if the phospho-Histone H3 signal reached a set threshold.
The raw mitotic index data was normalized using the CellHTS2 package (Boutros et al. 2006). For the primary screen, sample-based normalization was used. For the secondary and deconvolution screen, control-based normalization was used. After subtraction of the normalized mitotic indexes, the top genes (EICs specific) from the primary screen, that had a normalized difference in the mitotic index of 8, were selected for the secondary screen. Similar criteria were used for the secondary screen. For the deconvolution screen, a siRNA duplex was confirmed on-target when the increase in the normalized mitotic index was >2 times standard deviation of the negative control (siGAPDH) in all replicates.
Cell culture, transfection, and drug treatment
Cells were cultured in DMEM (GIBCO), supplemented with 6 % fetal calf serum, 100 U/ml penicillin and 100 μg/ml streptomycin. siRNAs were transfected using RNAiMax (Invitrogen) according to the manufactures guidelines. DNA was transfected using FuGENE 6 (Promega) according to the manufactures guidelines. The following siRNAs were used in this study: GAPDH OTP SMARTpool (Dharmacon), MCAK/Kif2C OTP SMARTpool (Dharmacon), Kif18b OTP SMARTpool (Dharmacon), Aurora A OTP SMARTpool (Dharmacon), Eg5/Kif11 OTP SMARTpool (Dharmacon), Kif15/HKlp2 OTP SMARTpool (Dharmacon) and custom siRNA (GAATGACTGATGAAGTCGA, Ambion, Tanenbaum et al. 2009), and Dynein heavy chain (Walczak et al. 1996). The following expression constructs were used in this study: mouse pTON-bEGFP-Kif15 (Tanenbaum et al. 2009). Phosphomutants of Kif15 were generated using site-directed mutagenesis. STLC (Sigma) was used at a concentration of 20 and 0.75 μM for EICs and parental cells, respectively. MLN8054 (Millenium Pharmaceuticals), MG132 (Sigma), nocodazole (Sigma), cytochalasin D (Sigma), and Y-27632 (Sigma) were all used at the indicated concentrations.
Cells were grown on 10-mm glass coverslips and pre-extracted for 60 s in PEM buffer (100 mM PIPES, 10 mM EGTA, 1 mM MgCl, and 0.1 % Triton X-100) followed by fixation in 4 % formaldehyde in PEM buffer with 0.3 % Triton X-100 for 10 min at room temperature. The following primary antibodies were used: α-tubulin antibody (Sigma) was used 1:10,000, phospho-H3 (Serine 10, Millipore) was used at 1:1500, γ-tubulin antibody (Abcam) was used 1:500. All antibodies were incubated overnight at 4 °C. Secondary antibodies (Alexa 488, 568, 647, Molecular Probes) were incubated for 1 h at room temperature. DAPI was added before mounting using ProLong Gold (Invitrogen). Images were acquired using a Deltavision deconvolution microscope (Applied Precision) with a 60× (NA 1.42) or a 100× (NA 1.40) oil objective, Softworx (Applied Precision), Fiji image software and Adobe Photoshop and Illustrator CS6.
Cells were plated on 8-well glass-bottom dished (LabTek). Cells were imaged using a Delatavision deconvolution microscope (Applied Precision) equipped with a heated chamber and cultured in L-15 CO2-independent medium (GIBCO). Images were acquired every 4 min using a 20× (NA 0.25) objective. Z-stacks were acquired with 2.5-μm intervals. Images were processed using Softworx (Applied Precision), Fiji image software and Adobe Photoshop and Illustrator CS6.
Cells were plated at a density at a density of 10,000 cells per well in a 48-well plate, treated as indicated and grown for about 7 days. Cells were fixed and stained using methanol and crystal violet.
Cells were counted and lysed using Laemmli buffer (120 mM Tris pH 6.8, 4 % SDS, 20 % glycerol). Protein levels were analyzed by western blot. The following antibodies were used: MCAK (Walczak et al. 1996) was used 1:1000, Kif18b (Tanenbaum et al. 2011a) was used 1:500, Aurora A (Cell Signaling) was used 1:1000, α-tubulin (Sigma) was used 1:10,000, Cdk4 (Santa Cruz) was used 1:2000, and cleaved PARP (Cell Signaling) was used 1:1000, Hsp90 (Santa Cruz) was used 1:2000.
Identification of phosphorylation sites
Five micrograms of recombinant mouse His-GFP-Kif15, purified from SF9 cells was incubated with 0.75 μg recombinant human His-Aurora A (Enzo Lifesciences) for 30 min in kinase buffer (50 mM Tris pH 7.5, 15 mM MgCl, 2 mM EGTA, 0.5 mM Vanadate, 1 mM DTT) in the presence of 60 μM ATP. Kinase assay using recombinant His-Aurora A and recombinant human histone-3 (NEB) served as a control. Phosphorylation sites on Kif15 were identified by mass-spectrometry with a nano-LC-LTQ-Orbitrap (Thermo Scientific).
We thank members of the Medema lab for helpful discussion. We thank Ana R.R. Maia and Andre Koch for critically reading the manuscript.
RGHPH, MET, DAE, and RHM designed the experiments. RGHPH and JAR carried out the experiments. RGHPH and JAR analyzed experiments. RGHPH and DAE analyzed screen data. CL and DL provided technical assistance. VAH and AJRH carried out and analyzed mass-spectrometry analysis. RGHPH and RHM wrote the manuscript.
Compliance with ethical standards
This article does not contain any studies with human participants or animals performed by any of the authors.
Conflict of interest
The authors declare that they have no conflict of interest.
This work was supported by the Netherlands Genomics Initiative of NWO and a ZonMW TOP project (40–00,812–98-10,021) to R.H.M.
- Helenius J, Brouhard G, Kalaidzidis Y, Diez S, Howard J (2006) The depolymerizing kinesin MCAK uses lattice diffusion to rapidly target microtubule ends. Nature 441(7089):115–9Google Scholar
- Hunter AW, Caplow M, Coy DL, Hancock WO, Diez S, Wordeman L, Howard J (2003) The kinesin-related protein MCAK is a microtubule depolymerase that forms an ATP-hydrolyzing complex at microtubule ends. Mol Cell 11(2):445–57Google Scholar
- Rath O, Kozielski F (2012) Kinesins and cancer. 1–13. doi: 10.1038/nrc3310
- Sturgill EG, Ohi R (2013) Kinesin-12 differentially affects spindle assembly depending on its microtubule substrate. Curr Biol 23(14):1280–1290 doi: 10.1016/j.cub.2013.05.043
- Walczak CE, Heald R (2008) Mechanisms of mitotic spindle assembly and function. In: International Review of Cytology. Elsevier, pp 111–158Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.