During mitosis, the cell sequentially constructs two microtubule-based spindles to ensure faithful segregation of chromosomes. A bipolar spindle first pulls apart the sister chromatids, then a central spindle further separates them away. Although the assembly of the first spindle is well described, the assembly of the second remains poorly understood. We report here that the inhibition of Aurora A leads to an absence of the central spindle resulting from a lack of nucleation of microtubules in the midzone. In the absence of Aurora A, the HURP (also known as DLGAP5) and NEDD1 proteins that are involved in nucleation of microtubules fail to concentrate in the midzone. HURP is an effector of RanGTP, whereas NEDD1 serves as an anchor for the γ-tubulin ring complex (γTURC). Interestingly, Aurora A phosphorylates HURP and NEDD1 during assembly of the initial bipolar spindle. We show here that the expression of a NEDD1 isoform mimicking phosphorylation by Aurora A is sufficient to restore microtubule nucleation in the midzone under conditions of Aurora A inhibition. These results reveal a new control mechanism of microtubule nucleation by Aurora A during assembly of the central spindle.
Achievement of faithful segregation of chromosomes relies on the coordination of a complex set of events. It begins with anaphase A and B, then continues with telophase and ends with the final abscission step that allows the physical separation of daughter cells. From onset of anaphase to abscission, the events controlling chromosome segregation and physical separation of the two daughter cells must be tightly coordinated (Mendoza and Barral, 2008; Fededa and Gerlich, 2012). Any mistakes during the coordination of these events can lead to severe aneuploidy and chromosomal instability. These two defects are now largely recognized to be responsible for tumorigenesis (Ganem et al., 2007), but are also believed to participate in cancer cell resistance to chemotherapeutic treatments (Puig et al., 2008). During anaphase A, the movement of chromosomes towards the spindle poles relies mainly on the depolymerization of microtubules (MTs) in the bipolar spindle. In anaphase B, by contrast, the separation of chromosomes continues as a result of polymerization of MTs in a central spindle assembled between the two sets of chromosomes already separated (Scholey et al., 2016). In addition, this anaphase central spindle is also involved in preventing mis-segregation of mis-attached chromosomes (Puig et al., 2008; Scholey et al., 2016) and serves as a scaffold to recruit structural and regulatory proteins required for the establishment of the actomyosin ring (Akhshi et al., 2014). The central spindle arises from the rapid growth of MTs that mainly emerge from the inter-chromosomal region during early anaphase (Uehara and Goshima, 2010) and are then stabilized and bundled to form an array of antiparallel MTs. The establishment of the central spindle requires MT-associated proteins that participate in nucleation, stabilization and bundling, as well as signalling molecules that regulate the different processes required for the assembly. Only a few molecular effectors have been identified to date, such as the RanGTP effector HURP (also known as DLGAP5), TACC3 and CLASP proteins (Lioutas and Vernos, 2013; Inoue et al., 2004; Maton et al., 2015). After this phase of polymerization, MTs are then amplified using pre-existing MTs as templates in an augmin complex-dependent manner (Uehara and Goshima, 2010) and concomitantly stabilized and bundled to form the robust central spindle through recruitment of the centralspindlin complex (Uehara and Goshima, 2010; Mishima et al., 2002; Mishima et al., 2004). The size of the central spindle is regulated by Aurora B-dependent phosphorylation of Kif2A (Uehara et al., 2013) and Kif4A (Nunes Bastos et al., 2013; Hu et al., 2011). All these proteins are also involved in the process of MT polymerization and/or stabilization during prometaphase (Koffa et al., 2006; Wong and Fang, 2006; Maton et al., 2015; Zhang et al., 2018; Ganem and Compton, 2004; Wandke et al., 2012). The structural similarities between bipolar and central spindle assembly organization have led to the assumption that these two spindles are assembled and regulated according to similar principles. However, recent work has shed light on differences between the mechanisms controlling bipolar spindle and central spindle assemblies (Hu et al., 2011).
Aurora A kinase is one of the major regulators of mitotic spindle assembly. While the prometaphase functions of Aurora A are now quite well understood, its post-metaphase role has only recently been clearly demonstrated by Hégarat et al. (2011), who used a conditional knock-out of Aurora A and pharmacological inhibition of Aurora B to show that these two kinases probably cooperate to control MT depolymerization at the end of anaphase. Additionally, Lioutas and Vernos (2013) have used the Aurora A kinase inhibitor MLN8237 to show that Aurora A participates in nucleation and stabilization of central spindle MTs at least in part by phosphorylating TACC3. Moreover, we have previously used a chemical genetics approach to demonstrate that Aurora A is involved in central spindle stabilization through P150Glued (also known as DCTN1) phosphorylation (Reboutier et al., 2013) and in spindle assembly checkpoint maintenance (Courthéoux et al., 2018). Whether Aurora A regulates MT nucleation at the midzone through phosphorylation of targets other than TACC3 during central spindle assembly remained an open question that we address in the present work. We show that inhibition of Aurora A during MT regrowth in anaphase leads to an absence of MTs in the midzone, demonstrating that the kinase is indeed required for central spindle MT nucleation. We also show that upon Aurora A inhibition, one of its prometaphase substrates, the RanGTP effector HURP, failed to relocate from kinetochore (KT)-MTs in prometaphase to central spindle MTs during anaphase. Finally, we show that phosphorylation by Aurora A, of the γ-tubulin ring complex (γTURC) adaptor protein NEDD1 on Ser405, another prometaphase substrate of Aurora A, is required for MT nucleation in the midzone. These findings shed new light on the molecular mechanisms underlying the initial steps of central spindle MT nucleation, and reveal a role of Aurora A in the process of central spindle assembly.
Aurora A activity is required for midzone MT nucleation during central spindle assembly
Aurora A has been implicated in MT nucleation from both centrosomes and chromosomes during prometaphase (Pinyol et al., 2013; Sardon et al., 2008; Reboutier et al., 2012; Terada et al., 2003; Scrofani et al., 2015). Here, we analysed its function in regulating MT nucleation during central spindle assembly in anaphase. Specific and rapid inhibition of Aurora A activity in early anaphase was achieved using two methods: a previously validated chemical genetics system and the Aurora A inhibitor MLN8237 (Reboutier et al., 2013; Courthéoux et al., 2018). Briefly, in the first approach, we used two U2OS cell lines, WT-AurA cells (negative control) that stably express a GFP-tagged wild-type form of Aurora A, and AS-AurA cells that stably express a GFP-tagged mutated form of Aurora A that can be specifically inhibited by the ATP analogue 1-naphthyl PP1 (1-NA-PP1). Both transgenic Aurora A coding sequences contain silent mutations that make them resistant to RNA interference (siRNA). As a consequence, following endogenous Aurora A downregulation by siRNA treatment, these two cell lines solely express exogenous WT-AurA or AS-AurA proteins but at levels similar to the endogenous one (Fig. 1A). In parallel, we also used U2OS cells that do not express any exogenous Aurora A (Ctl-AurA) in which we inhibited the kinase with its inhibitor MLN8237 (250 nM) (MLN-AurA) for a short time window (5 min) (Lioutas and Vernos, 2013; Cheeseman et al., 2011; Zeng et al., 2010; Asteriti et al., 2014).
To address whether Aurora A is involved in the nucleation of MTs during anaphase, we carried out a depolymerization and regrowth assay and followed the growth of MTs using live-cell microscopy in cell lines expressing GFP–tubulin. Briefly, WT-AurA and AS-AurA cells were first depleted of endogenous Aurora A with siRNA treatment (Fig. 1B). All cell lines were then synchronized in anaphase by first arresting them in early G2 phase with the CDK1 inhibitor RO3306 (Vassilev, 2006; Coldwell et al., 2013; Ma and Poon, 2011; Tsai and Zheng, 2005) and then by releasing the cells for 35 min. As soon as they entered anaphase (defined by the beginning of chromosome separation), WT-AurA and AS-AurA cells were treated with 10 µM 1-NA-PP1 and normal U2OS cells with 250 nM MLN8237 (MLN-AurA) for 5 min. At these concentrations 1-NA-PP1 and MLN8237 show high specificity for Aurora A (Reboutier et al., 2013; Lioutas and Vernos, 2013). Cells were then maintained for 20 min in cold medium containing 1-NA-PP1 or MLN8237 with 10 ng/ml nocodazole to depolymerize MTs. Nocodazole was then removed by washing cells with warm medium containing 1-NA-PP1 or MLN8237, and the cells were imaged using a spinning-disc confocal microscope. WT-AurA cells showed progressive nucleation of MTs both at the spindle poles and in the midzone (100%, n=13, Fig. 1B, first row) (as for Ctl-AurA, not shown). In comparison, 90% of AS-AurA and MLN-AurA cells displayed a stronger rate of MT growth at the spindle poles and a dramatic decrease in MT polymerization in the midzone (Fig. 1B, second and third rows), with even a total absence of midzone MTs in 33% of AS-AurA cells (n=30).
To quantify the number of MTs in the midzone, we repeated the experiments, then fixed the cells at different time points after the removal of nocodazole (t=0, t=30 s, 2 min, or 5 min), visualized MTs by immunofluorescence and manually counted them (Fig. 1C,D). At t=0, we observe that while there were no polar MTs in the WT-AurA cells, the Aurora A-inhibited AS-AurA cells still presented MTs emerging from spindle poles (Fig. 1C). In the midzone, however, the average number of MTs was significantly lower after inhibition of Aurora A (AS-AurA and MLN-AurA) than in cells containing active Aurora A (WT-AurA) (Fig. 1D). We decided to inhibit Aurora A with MLN8732 in the rest of the study because we observed similar inhibition of central spindle assembly in AS-AurA or MLN-AurA cells, as already reported (Reboutier et al., 2013; Lioutas and Vernos, 2013). Interestingly, when Aurora A was left active for the first 5 min of anaphase and inhibited only during the depolymerization and regrowth phases, we did not observe any defect in MT regrowth (Fig. 2A,B; Movies 1, 2 and 3). This probably means that Aurora A kinase activity is required for MT nucleation during the very beginning of anaphase. Inhibition of Aurora A kinase activity during anaphase is accompanied by decreased nucleation of MT in the midzone, indicating that Aurora A is critical for MT nucleation during central spindle assembly. However, inhibition of Aurora A kinase activity in anaphase simultaneously induces increased MT nucleation from centrosomes, such as would occur if the factors necessary for the assembly of the central spindle remained localized to the centrosomes instead of migrating in the midzone. Aurora A might thus be required to relocalize MT nucleation factors from the bipolar spindle to the midzone. To investigate this, we decided to follow two of these factors, HURP and NEDD1.
HURP localization is affected by Aurora A inhibition in anaphase
HURP is involved in bipolar spindle assembly, KT-MT stabilization and interkinetochore tension prior to anaphase (Koffa et al., 2006; Silljé et al., 2006; Wong et al., 2008). HURP is phosphorylated by Aurora A, which increases its affinity for MTs as well as its stability. (Yu et al., 2005; Wong et al., 2008). This prompted us to investigate the behaviour of HURP under Aurora A inhibition. We first decided to analyse Aurora A and HURP colocalization during mitosis. Using Aurora A–GFP-expressing U2OS cells, we observed a partial colocalization between Aurora A and HURP in prometaphase (Fig. 3A). As already reported, HURP and Aurora A localize along KT-MTs and accumulate at MT ends in the vicinity of kinetochore (Fig. 3A, enlarged middle panel and graphic) (Courthéoux et al., 2018; Wu et al., 2013). During anaphase A, HURP and Aurora A remain at KT-MTs but also appear at the new MTs in the central spindle (Fig. 3B, first row enlarged view, blue arrows). During anaphase B, KT-MTs are shorter, the HURP signal appears to decrease at KT-MTs but accumulates with Aurora A to freshly nucleated MTs at the central spindle (Fig. 3B, second row enlarged view, blue arrows). We then asked whether inhibition of Aurora A with MLN8237 would affect HURP localization. In anaphase A, HURP was found to still localize to central spindle MTs but, interestingly, no longer colocalize with Aurora A (Fig. 3B, third row enlarged view). In anaphase B, however, HURP was no longer found at the midzone (Fig. 3B, bottom row enlarged view, C,D); only rare central spindle MTs remained decorated with HURP (Fig. 3B, bottom row enlarged view, blue arrow). Our data indicate that Aurora A kinase activity is required for the presence of HURP in the midzone, although this may be indirectly due to the absence of microtubules. We then decided to focus on a MT nucleation factor.
NEDD1 is required for microtubule nucleation at the central spindle during anaphase and its localization is affected by Aurora A inhibition
NEDD1 is the anchoring protein of the γTURC that targets this complex to centrosomes, chromosomes and pre-existing MTs to promote MT polymerization and growth during bipolar spindle assembly in prometaphase (Haren et al., 2006; Lüders et al., 2006; Zhang et al., 2009). Although NEDD1 appears to be a key component of several MT assembly pathways, a role for NEDD1 during central spindle assembly has not been described. Therefore, we decided to check if NEDD1 was involved in the midzone MT nucleation process. As previously described for Aurora A (Marumoto et al., 2005; Reboutier et al., 2013), NEDD1 was also present as dots on the central spindle during anaphase A (Fig. 4B, first row). A reduction in the number of NEDD1 dots at the central spindle was observed after NEDD1 siRNA treatment (Fig. 4A) indicating that the dots we followed corresponded to NEDD1 protein (Fig. 4B, second row). The use of super resolution revealed that many NEDD1 dots on the central spindle also contained Aurora A, suggesting that the proteins could organize into clusters (Fig. 4C, bottom row, blue arrows).
Next, we examined the effect of NEDD1 depletion on central spindle assembly (Fig. 4D). When significant depletion of NEDD1 was achieved, about two thirds of the cells presented serious mitotic spindle defects and did not progress through prophase. Yet, one third of the cells, which probably had lower levels of NEDD1 depletion, entered anaphase. Among these cells, 95% (n=20 in three independent experiments) displayed a strong central spindle assembly defect never observed in control conditions. In particular, NEDD1-depleted cells had an overall weaker α-tubulin signal in the midzone compared to control cells (Fig. 4D, compare the first with the second row), meaning that NEDD1-depleted cells had nucleated fewer MTs in the midzone. Furthermore, the fluorescence intensity peaks in the midzone, characteristic of bundled MTs in control conditions, were absent when NEDD1 was depleted (Fig. 4D).
NEDD1 phosphorylation on Ser405 by Aurora A is involved in microtubule nucleation at the central spindle during anaphase
We tested whether Aurora A is involved in NEDD1 phosphorylation during anaphase and whether this event is required for the nucleation of MTs in the midzone. In order to obtain a large majority of cells synchronized in late mitotic phases, particularly in anaphase, we used a pharmacological approach developed by Hu and colleagues (2008). This approach involves blocking cells in monopolar prometaphase with S-trityl-L-cysteine [STLC, a kinesin-5 (also known as KIF11) inhibitor], and then forcing them to progress through and exit mitosis using purvalanol A (a CDK1 inhibitor) (Hu et al., 2011, 2008; Özlü et al., 2010). To evaluate the phosphorylation state of NEDD1 during anaphase, we blocked WT-AurA or AS-AurA cells depleted for endogenous Aurora A in monopolar mitosis with STLC. We then treated cells with purvalanol A, and after 5 min added 1-NA-PP1 to ensure that the cells had received the inhibitor while exiting mitosis (Fig. 5A). Interestingly, while WT-AurA cells treated with 1-NA-PP1 exhibited a robust spindle composed of parallel MTs (Fig. 5B, upper row), AS-AurA cells treated with 1-NA-PP1 presented a diffused and disorganized MT network (Fig. 5B, lower row). Phosphorylation of NEDD1 was analysed by western blot in extracts prepared from control cells or from Aurora A-inhibited cells synchronized in monopolar mitotic exit (cells were lysed at t=15 min) (Fig. 5C). In control conditions (WT-AurA+1-NA-PP1), NEDD1 appears as two bands, with the upper band being the most abundant form (Fig. 5C, first lane). Treatment of this extract with lambda protein phosphatase (λPP) resulted in the disappearance of the upper band and an increase in the intensity of the lower band, indicating that the upper band corresponds to the phosphorylated form of NEDD1 and the lower band to the unphosphorylated form (Fig. 5C, third lane). In comparison, when Aurora A was inhibited (AS-AurA+1-NA-PP1), two bands with similar intensities were detected (Fig. 5C, second lane), suggesting that Aurora A participates in NEDD1 phosphorylation during anaphase. To further investigate the role of Aurora A at the central spindle, we asked whether NEDD1 and Aurora A could colocalize at MT nucleation sites after recovery from cold shock. Cells stably expressing Aurora A–GFP were immunostained for NEDD1 and tubulin 15 s after recovery, and super-resolution images reveal a colocalization of NEDD1, Aurora A and tubulin at the midzone when MTs start to repolymerize (Fig. 6A, blue arrows). Because NEDD1 is directly phosphorylated on Ser405 by Aurora A in prometaphase to regulate MT nucleation from chromosome during bipolar spindle assembly (Pinyol et al., 2013), we decided to focus on this phosphorylation event in anaphase. We expressed various FLAG-tagged phosphorylation mutants of NEDD1 in cells depleted for endogenous NEDD1 (Fig. 6B). The level of FLAG–NEDD1 protein expressed was checked by western blot; as the wild-type form of NEDD1 (WT-NEDD1) runs faster than the FLAG-tagged variants on SDS-PAGE, both isoforms can be seen on lane 3 of Fig. 6B. We then counted the number of microtubule fibres present in the midzone. Interestingly, depletion of NEDD1 or expression of an isoform that cannot be phosphorylated on Ser405 by Aurora A (S405A-NEDD1, SA-NEDD1) leads to a major decrease in central MT nucleation after cold shock recovery (Fig. 6C, compare results set 1 with sets 2 and 5). This defect is rescued by expression of either WT-NEDD1 or the phosphomimetic mutant S405D-NEDD1 (SD-NEDD1) (Fig. 6C, result sets 4 and 6; Table S1). To confirm that Aurora A is responsible for the phosphorylation of NEDD1 on Ser405 that is required for MT nucleation at the midzone, we inhibited Aurora A using the analogue-sensitive allele and expressed NEDD1 mutants in a cold shock and recovery experiment. Endogenous Aurora A was depleted by siRNA and different FLAG-tagged NEDD1 variants were expressed (WT, SA and SD) (Fig. 7A, result sets 3, 4 and 5). Cells were immunostained with anti-tubulin and anti-FLAG antibodies, and MTs were counted in the midzone at different time points after recovery (Fig. 7B, blue arrows). Only expression of the NEDD1 phosphomimetic mutant (SD) partially rescued Aurora A inhibition, suggesting that NEDD1 phosphorylation on Ser405 by Aurora A is necessary for MT nucleation at the central spindle (Fig. 7C, result set 5; Table S2). Additionally, we analysed the localization of WT-NEDD1 and phospho-mutants S405A-NEDD1 and S405D-NEDD1 by expressing FLAG-tagged versions of the protein and quantifying their levels in the midzone in the presence of active Aurora A or in the presence of MLN8237. In the presence of active Aurora A, the same levels of WT-NEDD1 and S405D-NEDD1 were present in the mid-zone, while the amount of S405A-NEDD1 was much smaller. More importantly, in the presence of MLN8237, the level of WT-NEDD1 decreased in the midzone to reach a level equivalent to that of S405A-NEDD1. Also, the presence of S405D-NEDD1 in the midzone was insensitive to MLN8237 (Fig. 8A,B). This clearly demonstrates that Aurora A-dependent NEDD1 phosphorylation on Ser405 is necessary to localize NEDD1 in the midzone in order to nucleate MTs for central spindle assembly.
We have previously shown that Aurora A is involved in central spindle assembly (Reboutier et al., 2013). Here, we aimed to better understand the function of Aurora A during this event and identify the molecular mechanism(s) that contribute to this function. To this end, we used two complementary approaches. Firstly, a chemical genetics strategy we developed previously (Reboutier et al., 2013; Courthéoux et al., 2018) and secondly, pharmacological inhibition of Aurora A with the MLN8237 inhibitor. Using this drug, we confirmed results previously obtained with the chemical genetics strategy showing that inhibition of Aurora A during early anaphase caused the disappearance of the central spindle (Reboutier et al., 2013). We also confirmed results from Lioutas and Vernos (2013) showing that inhibition of Aurora A in early anaphase triggers a strong decrease in the nucleation rate of MTs within the midzone. Furthermore, our data indicate that inhibition of Aurora A triggers the stabilization of polar MTs and increases their growth. Similar results have been previously obtained by Floyd and colleagues (2008), who showed that stabilization of Aurora A after Cdh1 depletion results in brighter tubulin signal at the spindle poles during mitotic exit. An increase in the stability and/or growth of polar MTs may explain central spindle perturbation. However, these results suggest that factors normally contributing to midzone MT nucleation are probably retained at the centrosomes when Aurora A expression is perturbed. Interestingly, our results indicate that Aurora A activity is required within a short window of time at the entry to anaphase. Inhibition of Aurora A more than 5 min after anaphase onset results in no observable effect on midzone MT nucleation or central spindle assembly. Furthermore, our results show that the earlier Aurora A is inhibited, the stronger the effect on the central spindle assembly. This probably means that Aurora A is not required for central spindle assembly when midzone MT nucleation is already at an advanced stage. Overall, these results indicate that the central spindle defect related to inhibition of Aurora A can be explained by the contribution of the kinase to the initial nucleation of midzone MTs.
Our next goal was to identify Aurora A targets implicated in the nucleation of midzone MTs. Since the RanGTP pathway has been shown to be involved in central spindle assembly (Uehara and Goshima, 2010), we chose to focus our attention on two prometaphase substrates of Aurora A, and components of the RanGTP pathway, HURP and NEDD1. The interaction of HURP with MTs is regulated by Aurora A through phosphorylation of its C-terminal domain (Wong et al., 2008). Here, we observed an absence of HURP in the midzone during anaphase under Aurora A inhibition, suggesting that the kinase regulates HURP also in anaphase. However, multiple reasons might explain the absence of HURP in the midzone. It could be a consequence of the absence of MT in this area, since the location of HURP depends on the presence of MTs (Koffa et al., 2006). It might also result from an absence of HURP phosphorylation by Aurora A, leading to its degradation (Wong et al., 2008).
But how does MT nucleation starts in the midzone? Could Aurora A participate in early central spindle assembly events? We decided to ask whether the γTURC anchoring protein NEDD1, required for MT nucleation during prometaphase, would play a similar function in anaphase (Haren et al., 2006). To investigate the suggested role of NEDD1 in the early steps of central spindle assembly, we first examined NEDD1 localization and the consequences of its depletion. NEDD1 localized to the central spindle MTs with Aurora A, and in addition, Aurora A localized to NEDD1–tubulin dots after recovery from cold shock. Finally, NEDD1 depletion triggered strong defects in central spindle assembly, with fewer MTs bundled.
NEDD1 functions are highly regulated by mitotic kinases such as CDK1, PLK1, Aurora A and Nek9. Its phosphorylation regulates the various functions of NEDD1 and is required, for example, for the binding of NEDD1 to the γTURC and for its targeting to centrosomes, spindle and chromosomes (Haren et al., 2006; Lüders et al., 2006; Zhang et al., 2009; Manning et al., 2010; Johmura et al., 2011; Gomez-Ferreria et al., 2012; Sdelci et al., 2012; Pinyol et al., 2013). NEDD1 is also phosphorylated by Aurora A on Ser405 during pre-anaphase stages. This event is critical for RanGTP aster formation and chromatin-driven MT assembly in Xenopus egg extracts (Scrofani et al., 2015). Despite this complex phosphorylation profile during the cell cycle, NEDD1 appears predominantly as two bands in western blots, the upper being the phosphorylated form (Lüders et al., 2006; Zhang et al., 2009; Johmura et al., 2011; Sdelci et al., 2012; Gomez-Ferreria et al., 2012; Pinyol et al., 2013). In anaphase, we detected those two bands of NEDD1, the upper one being the strongest. Inhibition of Aurora A resulted in a decrease in intensity of the upper band concomitant with an increase in the lower one, indicating that Aurora A also participates in NEDD1 phosphorylation after metaphase. Interestingly, the S405A-NEDD1 mutant protein does not localize to the midzone, whereas S405D-NEDD1 (which mimics constitutive phosphorylation) localizes to the midzone in an Aurora A-independent manner. Furthermore, in depolymerization and regrowth assays during anaphase, the S405D-NEDD1 mutant partially rescued the midzone MT nucleation defect induced by Aurora A inhibition. The fact that this rescue is only partial suggests that the kinase phosphorylates multiple substrates to control anaphase, and that NEDD1 is one of them. Indeed, Aurora A phosphorylates p150Glued and TACC3 during anaphase; a lack of phosphorylation of these substrates also affects central spindle formation (Reboutier et al., 2013; Lioutas and Vernos, 2013). How those Aurora A substrates cooperate to control central spindle assembly remains to be found. For instance, does TACC3 activate Aurora A in anaphase as already reported during meiosis (Pascreau et al., 2005)? Does p150Glued help to transport NEDD1 with dynein? We are currently working towards addressing these questions.
As a whole, the data presented here indicate that NEDD1 phosphorylation on Ser405 by Aurora A is crucial for midzone MT nucleation during the initial steps of the central spindle assembly. Our work provides new insight into the regulatory mechanisms of MT nucleation during the initial steps of central spindle assembly. Specifically, we show for the first time that, once the mitotic spindle is correctly built and the checkpoint satisfied, central spindle assembly requires the prompt redistribution of NEDD1, and possibly other MT nucleation factors, from centrosomes to the midzone. Moreover, we show that this event requires Aurora A-dependent phosphorylation of NEDD1 on Ser405. This initial nucleation step is then followed by the processes of MT amplification and stabilization to ultimately build a robust and functional central spindle (Uehara and Goshima, 2010).
MATERIALS AND METHODS
DNA constructs and siRNA sequences
The DNA constructs allowing stable expression of WT-AurA or AS-AurA in U2OS cells have previously been described in Reboutier and colleagues (2013). The DNA constructs allowing the expression of the WT, S405A or S405D FLAG-tagged forms of NEDD1 have previously been described in Pinyol and colleagues (2013). Aurora A and NEDD1 were depleted from cells using siRNA oligonucleotides with the sequences 5′-AUGCCCUGUCUUACUGUCA-3′ and 5'-GGGCAAAAGCAGACAUGUG-3′, respectively. For siRNA negative control experiments, cells were transfected with a siRNA oligonucleotide targeting luciferase with the sequence 5′-CGUACGCGGAAUACUUCGA-3′.
Cell culture, synchronization and transfection
Human U2OS and HeLa cells were respectively grown in McCoy's and DMEM media with penicillin and streptomycin (Invitrogen) and 10% fetal calf serum (PAA). For siRNA transfection in co-transfection experiments, cells were transfected in culture medium using Jetprime (Polyplus Transfection) according to the manufacturer's instructions. The medium was changed after 24 h. For anaphase enrichment, cells were arrested in G2 with 10 μM RO3306 (Calbiochem) for 16 h and then washed three times for 30 s with the fresh pre-warmed medium. The first cells enter anaphase approximately 35 min after the first wash. For monopolar mitotic exit synchronization, we followed the protocol published by Hu and colleagues (2008). Briefly, cells were treated with 2 µM STLC (Tocris) for 24 h and then released into mitotic exit by treatment with 30 µM purvalanol A (Tocris). Cells reached monopolar mitotic exit 15 min after purvalanol A addition. 1-NA-PP1 was purchased from Tocris.
Microtubule regrowth quantification
After anaphase enrichment, the first anaphase was visualized using DIC and cells were then maintained in cold medium containing 1-NA-PP1 or MLN8237 plus 10 ng/ml nocodazole for 20 min to depolymerize MTs. Nocodazole was then removed by washing cells with warm medium containing 1-NA-PP1 or MLN8237, and the cells were immunostained as described below. We quantified the number of MTs in the midzone by repeating the experiments in which we fixed the cells at different time points after the removal of nocodazole (t=0, t=30 s, 2 min, or 5 min) visualized MTs by immunofluorescence (IF) and manually counted them.
For experiments presented in Figs 1C, 2A, 6C and 7C, cells were plated onto glass coverslips, fixed with methanol at −20°C for 10 min, and then washed three times in PBS and saturated with PBS-BSA 1% for 1 h at room temperature. For the other experiments, cells were plated onto glass coverslips, permeabilized in PBS Triton X-100 0.1% for 30 s and then fixed with 4% PFA in 10 mM MES buffer, 138 mM KCl, 3 mM MgCl2, 0.1 g/ml sucrose and 2 mM EGTA, pH 6.1. Cells were then washed three times in TGS (Tris glycin-SDS) buffer and saturated with PBS-BSA 1% for 1 h at room temperature. Antibodies in PBS-BSA 1% were added to the cells: rat anti-α-tubulin (1:1000; clone YL1/2, Millipore), rabbit anti-HURP polyclonal (1:500; HPA005546, Sigma-Aldrich), mouse anti-NEDD1 (1:200; clone MO5, Abnova), and mouse anti-FLAG (1:1000; clone M2, Sigma-Aldrich). The cells were incubated with antibodies overnight at 4°C, and then washed three times with PBS and incubated in the dark with secondary antibodies (Alexa Fluor 488 anti-mouse or Alexa Fluor 555 anti-rabbit; 1:1000, Invitrogen) for 1 h at room temperature. After three final washes with PBS, the coverslips were mounted with ProLong Gold (Invitrogen) with 1 µg/ml DAPI (Sigma-Aldrich). Imaging was performed using Olympus IX-71 microscope, with a 60× oil immersion objective lens, controlled by Delta Vision SoftWorx (Applied Precision). Image stacks were deconvolved and quick-projected, except when specified in the text. For midzone MT quantifications, we manually counted each α-tubulin labelled MT located in the midzone. Three independent experiments with 20 fixed cells per replicate were counted. For midzone tubulin fluorescence measurement, the mean fluorescence in a cropped region corresponding to the midzone was measured after background subtraction using ImageJ. The ImageJ software was used for NEDD1 dot quantification. The midzones were cropped, then we used the ‘Threshold function’ to automatically create a mask showing NEDD1 dot as clusters of black pixels (using the Otsu algorithm), and the numbers of clusters were counted using the ‘Analyze particles’ function.
Super resolution microscopy
Images in Figs 3A,B, 4C and 6A were acquired using Airyscan Zeiss microscope, using full Airyscan function for the four channels. Cells stably expressing Aurora A–GFP were fixed and immunostained using a modification of the ‘immunofluorescence’ protocol described above. GFP Nanobooster was used as a secondary antibody (1:500; gba488-100, Chromotek) to enhance GFP signal.
Cells were grown in LabTek I chambered coverglass (Nunc). Before microscopy, the medium was changed to CO2-independent medium supplemented with 10% FBS and 200 mM L-glutamine (Invitrogen). Time-lapse images were acquired using a Plan Apo 60×/1.4 NA objective on an Eclipse Ti-E microscope (Nikon) equipped with a spinning disk (CSU-X1, Yokogawa), a thermostatic chamber (Life Imaging Service), a Z Piezo stage (Marzhauser), and a charge-coupled device camera (CoolSNAP HQ2, Roper Scientific). MetaMorph Software (Universal Imaging) was used to collect the data. Frames were recorded every 30 s. Images are maximum projections of 20 z-planes acquired at 0.6 μm steps. Time-lapse data were processed using MetaMorph. Anaphase B onset corresponding to the beginning of central spindle assembly was determined as the moment when the first MTs of the central spindle were visible.
A paired t-test of the mean was performed using IGOR (Wavemetrics) and P-values are presented in Tables S1 and S2.
Western blot analysis
Cells were directly lysed in Laemmli buffer and proteins were resolved by SDS-PAGE (7%/13% discontinuous gradient for Aurora A and FLAG-tagged NEDD1, and 10% Anderson gels for NEDD1 phosphorylation) (Anderson et al., 1973) and transferred onto nitrocellulose membranes. The membranes were blocked for 1 h at room temperature with TBST 4% milk, and incubated overnight with the following antibodies: mouse anti-Aurora A (1:100; clone 35C1; Cremet et al., 2003), mouse anti-FLAG (1:5000; clone M2, Sigma-Aldrich), and mouse anti-NEDD1 (1:2000; clone MO5, Abnova). Membranes were incubated with secondary antibodies coupled to HRP (Jackson ImmunoResearch) for 1 h and antibody binding was detected by enhanced chemiluminescence (Pico or Dura, Pierce). Proteins extracts were treated with lambda protein phosphatase according to the supplier's protocol (New England Biolabs).
The authors dedicate this manuscript to the memory of Jean-Yves Cremet, who passed away April 7th, 2014. Microscopy work was performed on the IBiSA platform at Microscopy Rennes Imaging Center, with financial support from the Structure Fedérative de Recherche Biosit.
Author contributions metadata
Conceptualization: T.C., D.R.; Methodology: T.C., D.R.; Software: T.C., D.R.; Validation: T.C., D.R.; Formal analysis: T.C., D.R.; Investigation: T.C., D.R.; Resources: T.C., D.R., T.V., J.-Y.C., C.B., I.V., C.P.; Data curation: T.C., D.R.; Writing - review & editing: T.C., D.R., C.P.; Supervision: C.P.; Project administration: C.P.; Funding acquisition: T.C., C.P.
This research was funded by Ligue Nationale Contre le Cancer (grant number LNCC label 2014-2017), Agence Nationale de la Recherche (grant number Aurora), Centre National de la Recherche Scientifique (grant number annuel budget), Université de Rennes 1 (grant number annuel budget), and Région Bretagne/ Fédération Hospitalo-Universitaire to T.C. (grant number SAD 2018-2018).
The authors declare no competing or financial interests.