A steady-state metaphase spindle maintains constant length, although the microtubules undergo intensive dynamics. Tubulin dimers are incorporated at plus ends of spindle microtubules while they are removed from the minus ends, resulting in poleward movement. Such microtubule flux is regulated by the microtubule rescue factors CLASPs at kinetochores and depolymerizing protein Kif2a at the poles, along with other regulators of microtubule dynamics. How microtubule polymerization and depolymerization are coordinated remains unclear. Here we show that TPX2, a microtubule-bundling protein and activator of Aurora A, plays an important role. TPX2 was phosphorylated by Aurora A during mitosis. Its phospho-null mutant caused short metaphase spindles coupled with low microtubule flux rate. Interestingly, phosphorylation of TPX2 regulated its interaction with CLASP1 but not Kif2a. The effect of its mutant in shortening the spindle could be rescued by codepletion of CLASP1 and Kif2a that abolished microtubule flux. Together we propose that Aurora A–dependent TPX2 phosphorylation controls mitotic spindle length through regulating microtubule flux.
Proper formation and function of a mitotic spindle guarantees an even distribution of chromosomes between daughter cells. A metaphase spindle maintains certain length, although polymerization constantly happens at the plus ends of the spindle microtubules (MTs) and depolymerization at the minus ends (Kwok and Kapoor, 2007). This balance is achieved mainly via CLASPs and a Kinesin-13 family protein Kif2a (Maiato et al., 2003, 2005; Gaetz and Kapoor, 2004; Ganem and Compton, 2004; Laycock et al., 2006; Maffini et al., 2009), along with a few other regulators of MT dynamics (Goshima et al., 2005; Buster et al., 2007; Kwok and Kapoor, 2007). TPX2 was characterized as an important MT-associated protein involved in the chromosome-dependent spindle assembly (Gruss et al., 2001, 2002; Schatz et al., 2003; Tsai et al., 2003). During mitosis, high concentrations of RanGTP in the vicinity of chromosomes dissociate TPX2 from the inhibition of importin α/β (Gruss et al., 2001; Tsai et al., 2003; Clarke and Zhang, 2008). Free TPX2 binds and activates Aurora A, an important kinase for centrosome function and spindle assembly (Kufer et al., 2002; Eyers et al., 2003; Tsai et al., 2003; Eyers and Maller, 2004; Fu et al., 2007). TPX2 possesses a strong ability to nucleate MTs (Schatz et al., 2003; Brunet et al., 2004). It, together with RanGTP, was found to stimulate the MT-dependent MT nucleation in Xenopus laevis egg extracts (Petry et al., 2013). Despite the efforts to explore TPX2 functions in multiple systems, elucidating its function has been challenging, as both depletion and overexpression result in similar multiple defects, including spindle collapse and multipolarity (Garrett et al., 2002; Gruss et al., 2002; Bird and Hyman, 2008; Eckerdt et al., 2008).
In this work, we reveal that TPX2 takes a unique role in coordinating MT polymerization/depolymerization to maintain metaphase spindle length. We find that TPX2 is phosphorylated by Aurora A and is essential for normal MT flux on the metaphase spindle. The effect of the phospho-null TPX2 in shortening the spindle could be rescued by removing both CLASP1 and Kif2a that abolishes the MT flux. Thus TPX2 phosphorylation regulates mitotic spindle length through the MT flux.
Results and discussion
Phosphorylation of TPX2 by Aurora A is required for normal spindle length
To address how TPX2 regulates spindle assembly, we collected asynchronous and mitotic HeLa cell lysates to investigate its posttranslational modifications. We observed robust elevation in the expression level of endogenous TPX2 in mitosis and a slight upward band shift on the gel indicating possible modifications (Fig. 1 A). The band shift was abolished when the mitotic lysate was pretreated by λ-phosphatase, suggesting that TPX2 was phosphorylated. We confirmed this by analyzing equivalent aliquots of the samples on a gel incorporating Mn2+-phos-tag to bind phospho-proteins and reduce their mobility. It showed that most of TPX2 was phosphorylated during mitosis (Fig. 1 A).
We then looked for the kinase responsible for TPX2 phosphorylation. Previous reports have shown that TPX2 is phosphorylated by Aurora A in vitro (Kufer et al., 2002; Eyers et al., 2003; Eyers and Maller, 2004) and in a computational screen (Sardon et al., 2010). We therefore arrested HeLa cells in mitosis, treated them with Aurora A inhibitor MLN8237, and found that nonphosphorylated TPX2 accumulated (Fig. 1 B). This showed that phosphorylation of TPX2 in vivo depends on the kinase activity of Aurora A. Nevertheless, the phosphorylation of TPX2 was not completely abolished, indicating that some other kinases are involved.
In searching the potential Aurora A phosphorylation sites for the consensus sequence [K/R]X[S/T][AFILMV] (Ferrari et al., 2005; Ohashi et al., 2006), five serine/threonine residues were identified in human TPX2, of which only two are conserved from human to Xenopus (Fig. 1 C, arrows). Double and single mutations were made to confirm that TPX2-2A (S121/125A) could hardly be phosphorylated by Aurora A in vitro (Fig. 1 D), and both serine residues contributed to the proper phosphorylation (Fig. 1 E). This is in accord with a phosphoproteomics study that identified S121 and S125 as potential phosphorylation sites for Aurora A in mitotic HeLa cells (Kettenbach et al., 2011).
To determine the biological importance of TPX2 phosphorylation, we expressed the GFP-tagged TPX2-WT or -2A in HeLa cells. While the exogenous TPX2 caused a small amount of multipolar and short bipolar spindles, TPX2-2A showed a remarkable dominant-negative effect in generating short bipolar spindles (Fig. 1, F and G). The mean distance between spindle poles was reduced to 8.4 µm compared with the normal length of 12.3 µm (Fig. 1 H). We then established isogenic stable HeLa cell lines using the Flp-In T-REx system, which allows tetracycline-inducible expression of GFP-tagged TPX2-WT, -2A, and -2D (all containing synonymous mutations resistant to siRNA suppression) from a specific genomic location. Quantification of native GFP fluorescence in single mitotic cells and its proportion on spindle MTs showed that the expression levels of GFP-TPX2-WT, -2A, and -2D were comparable and that they were recruited to the spindle similarly (Fig. S1, A and B). TPX2 was depleted from these cells (Fig. S1 C) while the exogenous proteins were induced to express. The cells were synchronized in prometaphase by thymidine followed by Eg5 inhibitor, STLC, and then released into medium containing MG132 to allow full establishment of the bipolar spindles. We observed four categories of spindle phenotypes after TPX2 knockdown, as described previously (Gruss et al., 2002; Kufer et al., 2002; Bird and Hyman, 2008): first and most dominantly, spindle poles collapsed and two asters formed without organizing into bipolar structures; second, prometaphase-like spindles with two poles not fully separated and chromosomes not properly aligned; third, short bipolar spindles with chromosomes aligned at the equator; and fourth, multipolar spindles. Expression of GFP-TPX2-WT, -2A, or -2D could rescue most of the phenotype for collapsed poles and restore bipolarity (Fig. S1 D). However, bipolar spindles in the 2A-expressing cells were shorter compared with control, WT, or 2D-expressing cells (Fig. 1, I and J; and Videos 1–3). We also observed a small increase in the mitotic index when cells were depleted of endogenous TPX2 and expressing 2A (7.9%, compared with 5.3% in control, 5.2% in cells expressing WT, and 5.5% in 2D). This could be largely accounted for by the increase of cell population in prometaphase (3.8%, compared with 1.8% in control, 1.9% in cells expressing WT, and 2.2% in 2D), indicating that TPX2-2A causes a slightly prolonged early mitosis. Together our results show that phosphorylation of TPX2 by Aurora A is required for normal spindle length.
TPX2 phosphorylation rescues the short spindle phenomenon caused by Aurora A depletion
We then asked whether TPX2 functions downstream of Aurora A, or rather affects Aurora A in a feedback loop. Immunoprecipitation showed that GFP-TPX2-WT and -2A had similar affinity to Aurora A (Fig. S1 E), indicating that the interaction is independent of TPX2 phosphorylation. Then we depleted endogenous Aurora A and transiently expressed different TPX2 constructs (Fig. 2 A). Loss of Aurora A caused a 47% decrease in the mitotic index consistent with a previous report that Aurora A RNAi cells enter mitosis less efficiently (Hirota et al., 2003). We observed in 40% of metaphase cells a decrease in the spindle length upon Aurora A depletion (Fig. 2, B and C; Bird and Hyman, 2008). Expression of exogenous TPX2-WT or -2A did not rescue the phenomenon; however, TPX2-2D, a phospho-mimic mutant, restored the spindle length (Fig. 2, B and C). Together our data suggest that TPX2 is the main Aurora A target that regulates the spindle length.
We asked how TPX2 affects the spindle length. Recent reports showed that TPX2 interacts with Eg5 and facilitates the centrosome separation in early mitosis (Eckerdt et al., 2008; Ma et al., 2011; Gable et al., 2012). We found that TPX2-2A and -2D showed no preference in binding Eg5, although both were 41% less strong than WT (Fig. S1, F and G). The spindle localization of Eg5 was not affected by expression of TPX2 or its mutants (Fig. S1, H and I). These led us to propose that TPX2 phosphorylation regulates spindle length through another pathway. To ask which exact stage of spindle assembly is affected, we took advantage of an 11–amino acid peptide derived from the TAT protein transduction domain that can direct the efficient uptake of fusion proteins across the cell membrane within minutes (Liu et al., 2009). To test whether TPX2 affects the metaphase spindle independent of its role on centrosome separation, we synchronized the U2OS cells using double thymidine block and the Eg5 inhibitor STLC to induce monopolar spindle formation. These cells were then released into medium containing MG132 to allow full establishment of bipolar spindles before incubation with recombinant TAT-RFP-TPX2-WT, -2A, or -2D (Fig. 2, D and E). Compared with the normal spindle length of 12.2 µm, TPX2-2A decreased the distance between spindle poles to 10.1 µm, whereas WT and 2D barely had any effect (Fig. 2, F and G). These data show that TPX2 phosphorylation is necessary for maintaining the spindle length after the spindle reaches its steady state.
Phospho-null TPX2 inhibits MT flux on metaphase spindle
MT flux has been implicated as a major event in maintaining spindle length at later stages of mitosis (Laycock et al., 2006; Buster et al., 2007; Maffini et al., 2009). Based on our observations above, we tested whether TPX2 affects the rate of tubulin dimers moving along the spindle MTs. We cotransfected U2OS cells with photoactivatable (PA) GFP-tagged α-tubulin and different mCherry-TPX2 constructs. In a time-lapse manner, we activated a rectangular region of PAGFP–α-tubulin near the chromosomes and tracked its movement (Fig. 3, A–C; and Videos 4–6). In the background of endogenous TPX2, when expressing either mCherry-tagged TPX2-WT or -2D, the tubulin dimers showed clear poleward flux at the rate of 1.0 ± 0.2 µm/min, whereas when expressing 2A, the velocity of MT flux was greatly reduced to 0.16 ± 0.12 µm/min (Fig. 3, D–F). Together, these show that phosphorylation of TPX2 is necessary for the MT flux on metaphase spindle.
TPX2 phosphorylation regulates its interaction with CLASP1 but not Kif2a
We next asked how TPX2 might affect MT flux. We therefore expressed GFP-tagged CLASP1, CLASP2, or Kif2a in HeLa cells and pulled down endogenous TPX2 from mitotic lysates. We found that TPX2 specifically interacted with GFP-CLASP1 and -Kif2a (Fig. 4 A). In a parallel experiment, we pulled down endogenous CLASP1 and Kif2a from mitotic HeLa lysate and found that CLASP1 and Kif2a formed a complex with endogenous TPX2 (Fig. 4 B). Immunoprecipitation of either one of the three proteins always pulled down the other two, suggesting that they might be in a complex (Fig. S1 L). To test if these interactions reflect direct binding, we purified recombinant GST-tagged CLASP1 and Kif2a, mixed with recombinant His-GFP-tagged TPX2, and performed in vitro GST pull-down. This revealed that CLASP1 and Kif2a directly bound to TPX2 (Fig. S1 K).
We then examined whether TPX2 phosphorylation affects these interactions using three approaches. First, we transfected cells with GFP-tagged TPX2-WT, -2A, or -2D and pulled down endogenous CLASP1. This showed that CLASP1 bound to less TPX2-2A than -WT and -2D (Fig. 4 C). We also cotransfected cells with GFP-Kif2a and Flag-tagged TPX2-WT, -2A, or -2D, and performed immunoprecipitation using Flag antibody. This showed that Kif2a bound to TPX2-WT, -2A, and -2D with similar affinity (Fig. 4 D). In a second approach, we transfected cells with GFP-CLASP1 or -Kif2a, treated them with MLN8237 for 30 min, and performed immunoprecipitation using GFP antibody. This showed that CLASP1 pulled down less and that Kif2a pulled down a similar amount of TPX2 in the presence of MLN8237 (Fig. S1 J). Finally, we performed in vitro GST pulldown where we mixed GST-tagged CLASP1 or Kif2a with His-GFP–tagged TPX2-WT, -2A, or -2D. This revealed that there was less affinity of CLASP1 to TPX2-2A (Fig. S1 K); Kif2a appeared to bind more TPX2-2A than -WT and -2D (Fig. S1 K), in contrast to our immunoprecipitation experiment that failed to reveal any difference (Fig. 4 D and Fig. S1 J). One reason for the discrepancy could be that the in vitro system did not fully represent what happened in the cell lysate. Nevertheless, these data show that the interaction between TPX2 and the MT flux factor CLASP1 is Aurora A phosphorylation–dependent and that abolishing TPX2 phosphorylation reduces its affinity to CLASP1.
TPX2 phosphorylation regulates spindle length through MT flux
The above result that phosphorylation of TPX2 increased its affinity to CLASP1 but not Kif2a led us to hypothesize that TPX2 phosphorylation might be important to maintain the balance between MT polymerization and depolymerization. To test this, we co-depleted CLASP1 and Kif2a while transiently expressing TPX2-2A in HeLa cells. Previous reports have shown that depletion of CLASP1 or Kif2a results in short/monopolar spindles in human cells; but if both are depleted, the spindle remains normal length (Ganem and Compton, 2004; Maffini et al., 2009). This functional cooperation is also seen in Drosophila S2 cells (Laycock et al., 2006; Buster et al., 2007). We confirmed that upon double knockdown of CLASP1 and Kif2a, there was no obvious MT flux on the metaphase spindle, as previously reported (Fig. S2, A and B; Buster et al., 2007; Maffini et al., 2009). Under this condition, TPX2-2A could no longer show the dominant-negative effect of generating short spindles (Fig. 4, E and F). This result suggests that TPX2 phosphorylation is required for normal spindle length through regulating MT flux. Expression of TPX2-2A greatly restored the spindle bipolarity upon Kif2a depletion (Fig. S2 C), which is reminiscent of CLASP1 knockdown (Ganem and Compton, 2004; Maffini et al., 2009). This suggests that TPX2 phosphorylation positively regulates the function of CLASP1.
Phosphorylation of TPX2 by Aurora A is important for spindle organization in Xenopus extracts
To address the conservation of the above Aurora A–TPX2 pathway, we used similar strategies in mitotic Xenopus egg extracts. First, two serine residues in Xenopus TPX2 sequence (xTPX2, S90 and S94) corresponding to the S121/125 of human TPX2 were mutated. Another serine (S48), which is not conserved in human TPX2, was also included based on a previous report (Fig. 1 C; Eyers and Maller, 2004). The incorporation of labeled 32P into xTPX2-3A was greatly reduced (Fig. S3, A and B), and deletion of the xTPX2-pEg2 interacting domain (aa 1–39 of xTPX2) did not affect the phosphorylation of xTPX2 in vitro (Fig. S3, A and C). The GFP-tagged recombinant xTPX2-WT, -3A, and -3D were subsequently subjected to functional analysis in mitotic Xenopus egg extracts that had been depleted of pEg2 or xTPX2 (Fig. 5, A and B). In the mock-depleted extract, we observed bipolar spindles and a small proportion of other structures including asters and abnormal spindles with fewer MTs or unfocused poles. Depletion of xTPX2 caused a significant decrease in the percentage of bipolar spindles, which could be rescued by adding back recombinant xTPX2-WT and -3D but not -3A (Fig. 5, C and D). Similarly, the effects of pEg2 depletion could only be rescued by pEg2 itself or xTPX2-3D, but not xTPX2-WT or -3A (Fig. 5, C and D). These results indicate that the Aurora A–dependent phosphorylation of TPX2 is a conserved mechanism to regulate bipolar spindle assembly.
Our study reveals a novel function of TPX2 in scaling the spindle in later stage of mitosis when the spindle reaches a steady state (Fig. S2 D). This function of TPX2 is under the regulation of Aurora A phosphorylation on two conserved serine residues, the role of which has not been studied. We found that TPX2 interacts with MT flux factors CLASP1 and Kif2a. The affinity of TPX2 to CLASP1 is phospho-dependent, yet this is not the case for the TPX2–Kif2a interaction. It is therefore tempting to speculate that the phosphorylation of TPX2 regulates the activity of CLASP1 via a not-yet-known mechanism, and this could affect the activity of CLASP1 in incorporating tubulin dimers at the plus end of the spindle MTs. Both depletion of CLASP1 and expression of TPX-2A restore the spindle bipolarity upon Kif2a knockdown, suggesting that TPX2 phosphorylation positively regulates CLASP1. TPX2 might have an effect on Kif2a since they interact directly, but this is likely not to be affected by its phosphorylation state. Therefore, loss of phosphorylation in the mutant TPX2 may destroy the balance between MT polymerization and depolymerization, resulting in slow MT flux and short spindles. The N terminus of CLASP1 causes multiple dominant-negative defects in mitotic Xenopus extracts including hollow, enlarged, and multipolar spindles (Patel et al., 2012), which are reminiscent of our observations after xTPX2 depletion. Kif2a was shown to drive the spindle scaling during Xenopus embryogenesis through the fluctuation of its MT-destabilizing activity (Wilbur and Heald, 2013). These observations could possibly reflect the same pathway. Of course we could not rule out the possibility that other factors affecting MT flux might also be regulated by TPX2 phosphorylation. For instance, Kinesin-13 family members MCAK (Ganem et al., 2005) and KLP59C, Kinesin-8 KLP67A, and MT polymerization factor Dis1/Msps/XMAP215/TOG (Buster et al., 2007), as well as dynein, which can promote the poleward streaming of CLASP in Drosophila (Reis et al., 2009). Therefore, future work needs to study how phosphorylation of TPX2 regulates CLASP1 and whether it has impacts on other factors listed above, in order to understand how TPX2 regulates MT flux and the metaphase spindle length.
Our study and others have suggested that the role of TPX2 in scaling the spindle might be conserved during evolution. In C. elegans, TPXL-1 RNAi shows a strong effect on reducing the spindle length; the protein amounts at the centrosome and its gradient decay length along MTs are proportional to the spindle length (Greenan et al., 2010). In Xenopus tropicalis egg extracts where chromosome-induced spindles are smaller than those in Xenopus laevis, TPX2 is threefold more abundant (Helmke and Heald, 2014). It would be interesting to test whether the phosphorylation of TPX2 by Aurora A also plays similar role in these organisms. Nevertheless, Drosophila might be an exception because TPX2 homologue has almost no effect on mitosis in S2 cells (Goshima, 2011). Collectively, our study shows that TPX2 phosphorylation by Aurora A is responsible for maintaining metaphase spindle length by regulating the MT flux in human cells. This process is important for the later chromosome separation and the faithful cell division.
Materials and methods
Molecular cloning, protein purification, and antibody preparation
GFP-CLASP1 and -CLASP2 constructs were provided by H. Maiato (Institute for Molecular Cell Biology, Porto, Portugal), and GFP-Kif2a construct was provided by G. Fang (Genentech, San Francisco, CA) and C.-Y. Jang (Sookmyung Women’s University, Seoul, South Korea). Human Aurora A and TPX2 were cloned by RT-PCR from HeLa cell lysate as described previously (Fu et al., 2009). Xenopus pEg2 and xTPX2 were cloned from a cDNA library of Xenopus oocytes (Takara Bio Inc.). Mutants and truncates of TPX2/xTPX2 were generated by PCR with specific primers. For overexpression in HeLa cells, TPX2 cDNAs were cloned into pEGFP-C1 (6084-1; Takara Bio Inc.) or p3×FLAG-CMV vector (E4026; Sigma-Aldrich) using BamHI–SalI on the cDNAs and BglII–SalI on the vectors. To obtain recombinant proteins, TPX2 cDNAs were inserted into pET28a (69864-3; EMD Millipore) using SalI–HindIII, xTPX2 using EcoRI–SalI, and pEg2 using BamHI–XhoI, and expressed in E. coli BL21 (DE3 pLys). Based on the need of the individual experiment, either GFP was fused to the N terminus of the TPX2 cDNAs to increase protein stability (Brunet et al., 2004), or TAT-RFP was introduced to enable instant absorption by the cells. All proteins were affinity purified using Ni-NTA agarose (for His-tagged proteins; QIAGEN) or Glutathione Sepharose 4B beads (for GST-tagged proteins; GE Healthcare), according to the manufacturers’ protocol. Antibodies against GFP, Aurora-A, and TPX2 were raised in both rabbits and mice using His-tagged full-length proteins as described previously (Fu et al., 2009); antibodies against pEg2 and xTPX2 were raised in rabbits using His-tagged full-length proteins, and Kif2a and CLASP1 in mice, using GST-tagged Kif2a 1–95 aa and CLASP1 1239–1538 aa, respectively. Other antibodies were as follows: mouse anti-flag (F1804; Sigma-Aldrich), rabbit anti–γ-tubulin (T3320; Sigma-Aldrich), mouse anti–α-tubulin (T9026; Sigma-Aldrich), rabbit anti-pericentrin (ab4448; Abcam); rabbit anti-actin (sc-130656; Santa Cruz Biotechnology, Inc.), mouse anti-Eg5 (611187; BD), mouse anti–phospho-histone H3 (Ser-10; 05-806, EMD Millipore), rabbit anti-TPX2 (12245; Cell Signaling Technology), and rabbit anti-Kif2a (HPA004716; Sigma-Aldrich), Alexa Fluor 488 donkey anti–mouse, Alexa Fluor 546 donkey anti–mouse, and Alexa Fluor 546 donkey anti–rabbit (all from Invitrogen). Goat anti–rabbit/mouse IgG HRP was from Jackson ImmunoResearch Laboratories, Inc.
Cell synchronization, siRNA and plasmid transfections, and cell line establishment
HeLa or U2OS cells were cultured in DMEM supplemented with 10% bovine calf serum, 100 U/ml penicillin, and 100 µg/ml streptomycin (all from Hyclone; GE Healthcare) at 37°C and 5% CO2. To be synchronized in mitosis for immunoprecipitation, HeLa cells were treated with double thymidine, released for 8 h, and supplemented with 50 ng/ml nocodazole for another 6 h. To be synchronized in metaphase for immunofluorescence (Fig. S1 D), HeLa cells were treated with thymidine for 24 h, released for 3 h, and incubated with 5 µM STLC for 8 h to induce monopolar spindle formation. These cells were then released into fresh medium containing 20 µM MG132 for 2 h to allow full establishment of bipolar spindles. To deliver siRNA and plasmid DNA, cells grown on coverslips were transfected using Lipofectamine 2000 (Invitrogen). The following sequences were used: Aurora A, 5′-UUCUUAGACUGUAUGGUUA-3′ (GE Healthcare); TPX2, 5′-CCAGUUGACAACACUUACU-3′; CLASP1, 5′-GCCAUUAUGCCAACUAUCU-3′ (Mimori-Kiyosue et al., 2005); Kif2a, 5′-GGCAAAGAGAUUGACCUGG-3′ (Ganem and Compton, 2004); and negative control, 5′-UUCUCCGAACGUGUCACGU-3′. For making TPX2 RNAi-resistant mutants, 5′-CCGGTCGATAATACGTATTACAAAGAGGCA-3′ and 5′-TGCCTCTTTGTAATACGTATTATCGACCGG-3′ were used.
To establish tetracycline-inducible cell lines for expressing GFP-TPX2-WT, -2A, and -2D, Flp-In T-REx host HeLa cell lines (Shrestha et al., 2014) were cotransfected with pOG44 and a Gateway vector (Torres et al., 2009) containing GFP-TPX2-WT, -2A, or -2D (all containing synonymous mutations resistant to siRNA suppression) under the control of CMV/TetO2 promoter and FRT recombinant site. Homologous recombination between FRT sites in the Gateway vector and on the host cell chromosome, catalyzed by the Flp recombinase expressed from pOG44, generated the derivative cell lines into which the gene had been stably integrated into the host cell chromosome. These cells were then selected and maintained using the puromycin resistance (Life Technologies) acquired as a consequence of the recombination event. Next, single colonies were sorted using FACS and tested for the expression levels upon tetracycline induction. The ones that expressed comparable amount of GFP-TPX2-WT, -2A, and -2D were used for experiments. To avoid leaking expression of the exogenous proteins, cells were cultured in DMEM supplemented with 10% qualified FBS (Life Technologies).
TAT fusion protein treatment
U2OS cells were synchronized with double thymidine, released for 8 h, and incubated with 5 µM STLC for 14 h to induce monopolar spindle formation. These cells were then released into fresh medium containing 20 µM MG132 for 1.5 h to allow for full establishment of bipolar spindles. Finally, 5 nM of recombinant protein TAT-RFP-TPX2-WT, -2A, or -2D was added to the medium and the cells were incubated for 1 h before fixation.
λ-PPase treatment and phos-tag gel electrophoresis
Cells were collected from a 35-mm dish and rinsed twice with ice-cold HBS consisting of 20 mM Hepes, pH 7.4, and 150 mM NaCl. They were subsequently lysed on ice for 15 min with RIPA buffer, containing 1% NP-40, 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5 mM EDTA, 0.5% sodium deoxycholate, 0.1% SDS, and complete EDTA-free protease inhibitors (Roche). Lysates were centrifuged at 20,000 g for 15 min, and supernatants were supplemented with 10× PMP buffer, 10× MnCl2, and 400 U λ-PPase (New England Biolabs, Inc.). The reactions were kept at 30°C for 30 min before being quenched with SDS sample buffer. Phos-tag gel electrophoresis was performed according to the manufacturer’s protocol (Wako Pure Chemical Industries). In brief, 8% separating gel was prepared with 25 µM Phos-tag Acrylamide AAL-107 and 50 µM MnCl2. We then mixed thoroughly, allowed the gel to polymerize for 1 h, and prepared the normal stacking gel. Samples were run at low voltage around 60 V until the Bromophenol blue reached the bottom. The gel was then soaked in transfer buffer containing 1 mmol/liter EDTA for 10 min and transfer buffer without EDTA for another 10 min before transferring to PVDF membrane for antibody blotting.
Protein–A Sepharose beads (GE Healthcare) were washed three times with PBS, diluted into 500 µl, and incubated with indicated antibody in each experiment for 1 h at 4°C. Beads were then washed three times with lysis buffer, consisting of 0.5% NP-40, 20 mM Tris-HCl, pH 7.4, 500 mM NaCl, 0.5 mM EGTA, 10 mM NaF, 2 mM sodium vanadate, 10 mM β-glycerophosphate, 10 µg/ml leupeptin, 0.7 µg/ml pepstatin A, 1.5 µg/ml aprotinin, and 1 mM PMSF. For immunoprecipitation, mitotic HeLa cells transfected with the indicated constructs were shaken off the culture dish and lysed on ice for 30 min with lysis buffer. Lysates were centrifuged at 16,000 g for 15 min, and supernatants were incubated with beads for 1 h at 4°C on the rotator. After six washes with lysis buffer, the bound proteins were analyzed by Western blotting.
In vitro binding assay
Recombinant GST/GST-Kif2a/GST-CLASP1 and GFP/GFP-TPX2/GFP-2A/GFP-2D proteins were diluted to a final concentration of 1 mg/ml in PBS. 10 µg of GST-tagged and 30 µg of GFP-tagged proteins were mixed for each reaction in 500 µl PBS/0.1% Triton X-100 and rotated for 0.5 h at 4°C. Next, 20 µl of prewashed Glutathione Sepharose 4B beads (GE Healthcare) was added and the whole mixture was rotated for another 1.5 h at 4°C. Finally, the beads were spun down, washed four times with PBS/0.1% Triton X-100, boiled in 2× sample buffer, and analyzed by Western blotting analysis.
Immunofluorescence microscopy, live cell imaging, and statistical analysis
Cells grown on coverslips were fixed with precooled methanol for 6 min on ice. After rehydration in TBS, the cells were incubated with the primary antibody overnight at 4°C. Then they were washed three times in TBS and incubated with the secondary antibody for 45 min at room temperature. Coverslips were mounted onto slides by using Mowiol containing 1 µg/ml DAPI and analyzed under a microscope (Axiovert 200M; Carl Zeiss) equipped with a 63×/1.4 NA oil objective lens. Images were captured with a camera (MRM CCD; Carl Zeiss) and Axiovision image acquisition software. For live cell imaging, U2OS cells were cotransfected with photoactivatable GFP-tagged α-tubulin and different mutants of mCherry-TPX2 (Videos 4–6). Before imaging, cells were changed to CO2-independent medium (18045-088; Life Technologies) supplemented with 200 mM l-glutamine and 10% FBS. Images were captured at 37°C under a confocal microscope (LSM 510 Meta; Carl Zeiss; 100×/1.4 NA oil objective lens, an AxioCam digital microscope camera, and LSM510 software). Bipolar spindles were selected according to the red channel (mCherry-TPX2) and the 405 nm laser was used to activate a rectangular region of PAGFP–α-tubulin signal near the chromosomes (time point = 0). The movement of the tubulin dimers was then tracked using 488/514 nm lasers every 15 s. Alternatively, HeLa cells cultured in DMEM supplemented with 10% FBS were tracked every 3 min with a microscope (ECLIPSE Ti; Nikon) at 37°C and in a 5% CO2 environment (Videos 1–3). Images were captured with a 60×/1.4 NA oil objective lens, an EM CCD (Hamamatsu Photonics, Inc.), and Volocity software. To determine the spindle length, the distances between two spindle poles were measured by AxioVision LE software. The mean spindle length was determined from at least three independent experiments and graphed with the SEM using GraphPad Prism 5 software (GraphPad Software). Statistical significance was also analyzed for p-values using GraphPad Prism 5. To measure the fluorescence intensity of TPX2 or Eg5 in a single cell or on a spindle, ImageJ was applied to select the region of the cell or spindle and the total fluorescence was calculated by the size of the area times the mean intensity that had been deducted from the background intensity. To determine the velocity of MT flux, spindles were first rotated to the horizontal position according to the red channel (mCherry-TPX2). Each frame from the same movie was then adjusted using ImageJ plugin StackReg to minimize the movement of the whole spindle. Next, the ImageJ macro Spindle Microtubule Frap/Flip Analyzer (http://cmci.embl.de/downloads/spindlefanalyzer; Ma et al., 2010) was applied. The macro uses the red channel to create a mask that specifies the area of the spindle, and then this specified area is measured for the integrated intensity profile of the green channel (projected in the y axis, and measured along the x axis). The profile was plotted in Excel after smoothed via B-spline and the x coordinate of the maximum intensity was exported to represent the position of the fluxing tubulin dimers. The velocity of MT flux was defined by the ratio between the distance relative to the equator and the time, and based on our observation it’s mostly linear within the first 60 s. For this reason we always evaluated the mean velocity at time point = 60 s for the statistical data in Fig. 3 F.
To test the phosphorylation of different TPX2 mutants, 1 µg of recombinant protein was incubated with 1 µg Aurora A (or pEg2) kinase for 20 min at 30°C. Kinase assays were performed in 50 mM Tris-HCl, pH 7.4, 50 mM NaCl, 10 mM MgCl2, 1 mM DTT, and 100 µM ATP containing 3,000 Ci/mmol γ-[32P]ATP. Reactions were quenched with SDS sample buffer and analyzed by SDS-PAGE and autoradiography.
Xenopus egg extract experiments
Xenopus CSF extracts were prepared as described previously (Fu et al., 2010). In brief, eggs were washed sequentially with XB (10 mM Hepes, 0.1 mM CaCl2, 100 mM KCl, 1 mM MgCl2, and 50 mM sucrose, pH 7.8) and CSF-XB buffer (10 mM Hepes, 2 mM MgCl2, 0.1 mM CaCl2, 100 mM KCl, 5 mM EGTA, and 50 mM sucrose, pH 7.8), and centrifuged in CSF-XB plus cytochalasin B (100 µg/ml) at 13,000 rpm for 10 min (TLS55 rotor; Beckman Coulter) to obtain the supernatant. To deplete pEg2 or xTPX2, protein A–conjugated Dynal beads were coupled with xTPX2 or pEg2 antibody, incubated with CSF on ice for 40 min, and removed by a magnet. We then repeated the process to make sure of full depletion. To analyze the rescue effects by different recombinant proteins, demembranated Xenopus sperms (600 nuclei/µl), Rhodamine-labeled tubulin, and these proteins were added to the extracts to allow spindle assembly for 45 min at 21°C. The samples were then fixed with fixation buffer (Marc’s modified ringers [MMR] plus 4% paraformaldehyde and 50% glycerol) and analyzed under a microscope (Axiovert 200M; Carl Zeiss).
Online supplemental material
Fig. S1 shows the binding capacity of TPX2 phospho mutants to different interacting partners. Fig. S2 supports the data presented in Fig. 4 (E and F) and shows that TPX2 regulates metaphase spindle length through affecting MT flux. Fig. S3 shows that Xenopus TPX2 is phosphorylated by Aurora A. Video 1 shows spindle formation in the background of GFP-TPX2. Video 2 shows spindle formation in the background of GFP-TPX2-2A. Video 3 shows spindle formation in the background of GFP-TPX2-2D. Video 4 shows MT flux on metaphase spindles in the background of GFP-TPX2. Video 5 shows MT flux on metaphase spindles in the background of GFP-TPX2-2A. Video 6 shows MT flux on metaphase spindles in the background of GFP-TPX2-2D.
We thank Dr. Jing Wu (Tsinghua University, Beijing, China) for the data analysis of MT flux.
This work was supported by funds from the National Natural Science Foundation of China (91313302, 31071188, 31030044, 90913021, and 31371365) and the State Key Basic Research and Development Plan (2010CB833705 and 2014CB138402).
The authors declare no competing financial interests.
J. Fu and M. Bian contributed equally to this paper.