DDA3 and Mdp3 modulate Kif2a recruitment onto the mitotic spindle to control minus-end spindle dynamics

Chromosome movement, such as chromosome congression to the mitotic equator and synchronous segregation to opposite sides of the cell during mitosis, is essential for chromosome integrity (Cheeseman and Desai, 2008; Tanaka et al., 2005). The dynamic properties of plus-end microtubules (MTs) are spatially and temporally regulated by different plus-end-tracking proteins (+TIPs), which act primarily as MT-stabilizing factors and link MT ends to cellular structures, such as the cell cortex or kinetochores (Akhmanova and Steinmetz, 2008; Howard and Hyman, 2003; Perez et al., 1999; Schuyler and Pellman, 2001). MT minus-end depolymerization also contributes to MT turnover, whereas MT minus ends in the cell do not grow (Dammermann et al., 2003) because they are capped by specific factors, such as the γ-tubulin ring complex (γ-TURC) (Kollman et al., 2011). α-tubulin–β-tubulin dimers translocate toward the poles through a process termed poleward MT flux, which is based on the depolymerization of the minus ends of spindle MTs at the poles and the net polymerization of their plus ends in the metaphase spindle, resulting in a poleward stream of α-tubulin–β-tubulin (Ganem and Compton, 2006; Kwok and Kapoor, 2007). Thus far, the only known minus end-stabilizing factors are members of the calmodulin-regulated spectrin-associated protein (CAMSAP, also known as patronin) family, which are recruited to growing minus ends to protect them against depolymerizing enzymes (Baines et al., 2009; Goodwin and Vale, 2010; Meng et al., 2008).

Recently, DDA3 (also known as PSRC1) was identified as a new class of spindle dynamics regulator that acts as a MT-destabilizing protein by interacting with the minus-end MT depolymerase Kif2a and the plus-end MT depolymerase MCAK (also known as Kif2C) (Jang and Fang, 2011; Jang et al., 2008). Whereas recombinant DDA3 protein and unphosphorylated DDA3 stabilize MTs in interphase, exhibiting both MT-polymerizing and -bundling activities (Hsieh et al., 2012; Jang et al., 2011), it destabilizes spindle MTs during mitosis by recruiting Kif2a (Jang et al., 2008). Given that DDA3 is hyper-phosphorylated by mitotic kinases, such as Cdk1 and Aurora A during mitosis (Jang et al., 2010, 2011), the function of DDA3 might be converted from an MT stabilizer into an MT destabilizer by phosphorylation at mitotic entry. In addition, DDA3 regulates astral spindle formation and spindle positioning by targeting Cep290 to the centrosome (Song et al., 2015). To understand the detailed regulation mechanism of minus-end spindle dynamics by DDA3, we investigated additional components of the DDA3 complex, with a focus on the MT minus ends. Our results revealed that microtubule-associated protein 7 domain-containing 3 (Mdp3; also known as MAP7D3) forms a mitotic complex with DDA3 and regulates mitotic processes through (1) the inhibition of DDA3-mediated Kif2a recruitment and, concomitantly, (2) fine tuning the turnover of spindle MTs and poleward flux.

To investigate the molecular mechanism of DDA3-mediated regulation of spindle dynamics, we have previously purified the DDA3 complex from mitotic cells expressing DDA3 C-terminally tagged with the S-peptide and GFP (DDA3–S–GFP), and components of the DDA3 complex were analyzed by mass spectrometry (Jang et al., 2008). Interestingly, Mdp3 was identified as a DDA3-associated protein (Fig. S1A). Although Mdp3 has been identified as a spindle-associated protein through proteomic analysis (Sauer et al., 2005; Sun et al., 2011), its mitotic function has not been characterized. Hence, we asked whether endogenous Mdp3 and DDA3 form a complex during the cell cycle. As expected, DDA3 was detected in the Mdp3 immunoprecipitates in early mitosis but not in interphase (Fig. 1A). Reciprocal immunoprecipitations also showed that DDA3 and Mdp3 co-precipitated in early mitosis (Fig. 1B). Gel filtration analysis of mitotic cell extracts demonstrated the presence of Mdp3 and DDA3, as well as Kif2a, in the same molecular mass fractions (Fig. 1C). Strikingly, depletion of DDA3 or Mdp3 increased the level of Mdp3 or DDA3 in the complex, respectively. In agreement with previous findings (Sun et al., 2011), endogenous Mdp3 was localized to the mitotic spindle from prometaphase to anaphase (Fig. 1D) and colocalized with DDA3 at the spindle during metaphase (Fig. S1B). These data indicate that Mdp3 is a new component of the DDA3 complex during mitosis.

To determine the function of Mdp3 in the DDA3 complex, we depleted endogenous Mdp3 protein by using small interfering RNAs (siRNAs) and found a twofold increase in the mitotic index as measured by flow cytometry (Fig. 2A,B). Immunostaining of Mdp3-depleted cells revealed that the loss of Mdp3 localization to the spindle resulted in abnormal spindle structure and chromosome congression defects in metaphase cells (Fig. 2C). In mitotic cells, we observed a 10% increase in prometaphase cells and more than a tenfold increase in metaphase cells with unaligned chromosomes upon depletion of Mdp3 (Fig. 2D,E). Further analysis revealed a sixfold increase in anaphase cells with lagging chromosomes compared with control cells (Fig. 2F). Consistent with these data, quantitative analysis of mitotic progression by time-lapse imaging indicated that the depletion of Mdp3 prolonged the duration of prometaphase [from nuclear envelope breakdown (NEB) to the initial formation of the metaphase plate] and metaphase (from the initial formation of the metaphase plate to anaphase onset; Fig. 2G,H and Movies 1–3). Remarkably, Mdp3-depleted cells also showed chromosome segregation defects, such as lagging chromosomes in anaphase and chromosome bridges in telophase and cytokinesis (Fig. 2G; Movie 3). To investigate whether Mdp3 is involved in chromosome congression, we sought to arrest cells at metaphase by adding the proteasome inhibitor MG132. Notably, most of metaphase cells depleted of Mdp3 showed severe chromosome congression defects (Fig. 2I,J). Whereas re-expression of GFP–DDA3 or GFP–Mdp3 rescued the mitotic defects caused by depletion of the endogenous proteins, GFP–DDA3 expression in Mdp3-depleted cells or GFP–Mdp3 expression in DDA3-depleted-cells was insufficient to overcome chromosome alignment defects (Fig. 2K; Fig. S1C). Furthermore, the exogenous expression of DDA3 and Mdp3 had no reciprocal effect on their respective localizations (Fig. S1C). These results suggest that Mdp3 and DDA3 play different roles in the same complex with respect to chromosome congression.

To delineate the role of Mdp3 in the DDA3 complex, we asked whether the chromosome congression defect caused by depletion of Mdp3 results from impaired spindle dynamics, as occurs in DDA3-depleted cells. The lack of congression of unaligned chromosomes in Mdp3-depleted cells resulted from the absence of MT attachment, as evidenced by Mad2 and BubR1 remaining on the kinetochores of unaligned chromosomes (Fig. 3A,B). However, the depletion of Mdp3 did not abolish the tension across sister kinetochores of aligned chromosomes (Fig. 3C) and decreased the interpolar distance (Fig. 3D), suggesting that depletion of Mdp3 increases spindle dynamics in addition to disruption of spindle attachment. Consistent with this, spindle MT immunofluorescence intensity decreased by 50% in Mdp3-depleted cells compared with control cells (Fig. 3E). As expected, the spindle MT fluorescence intensity significantly increased (by 100%) in cells overexpressing Myc–Mdp3 compared with control cells (Fig. 3F,G). We consistently found that spindle MTs in Mdp3-depleted cells were more sensitive to limited nocodazole treatment and repolymerized more slowly than in control cells (Fig. 3H,I), indicating that Mdp3 is involved in spindle MT polymerization and/or stabilization.

As DDA3 regulates spindle dynamics by recruiting Kif2a, and depletion of Mdp3 destabilizes spindle MTs, we reasoned that the decrease in spindle stability and consequent chromosome alignment defects in Mdp3-depleted cells could be explained by an overloading of Kif2a onto the spindle. To explore this idea further, we confirmed the interaction between Mdp3 and Kif2a during mitosis by immunoprecipitation using an anti-Mdp3 antibody (Fig. 4A). Consistent with the results shown in Fig. 1C, depletion of DDA3 or Mdp3 increased the interaction between Mdp3 and Kif2a, or DDA3 and Kif2a (Fig. 4B,C), respectively. As a result, the level of Kif2a on the spindle (Fig. 4D,E) and around the spindle pole increased upon Mdp3 depletion (Fig. 4F). Furthermore, Kif2a recruitment was diminished in cells overexpressing Myc–Mdp3 compared with control cells (Fig. 4G–I). Because the levels of spindle MTs and Kif2a in DDA3 and Mdp3 double-depleted cells were similar to those in DDA3-depleted cells (Fig. 4J,K; Fig. S2A,B), we conclude that Mdp3 acts as a negative regulator against DDA3 for Kif2a recruitment. Although the increase in tubulin levels in DDA3-depleted cells impaired chromosome alignment (Jang et al., 2008), chromosome congression was not perturbed by the increase in tubulin levels in Mdp3-overexpressing cells (Fig. 2K). Because, in the same manner, the decrease in tubulin levels in DDA3-overexpressing cells also did not disrupt chromosome congression (Fig. 2K), we reasoned that endogenous DDA3 in Mdp3-overexpressing cells or Mdp3 in DDA3-overexpressing cells might compensate for overexpression effects of Mdp3 or DDA3 on chromosome congression, respectively. To analyze the effect of Mdp3 on spindle dynamics, the turnover rates of α-tubulin–β-tubulin heterodimers on the mitotic spindle were measured using a fluorescence loss in photobleaching (FLIP) experiment. In contrast to cells in which DDA3 was depleted (Jang et al., 2008), the half-life of GFP–α-tubulin was decreased in Mdp3-depleted cells (Fig. S2C–E; Movies 4 and 5). To analyze the effect of Mdp3 and DDA3 on minus-end MT dynamics, the recovery rate of GFP signals in the bleached region was measured using a fluorescence recovery after photobleaching (FRAP) experiment. Although depletion of Mdp3 increased the recovery rate by facilitating MT flux, DDA3 depletion decreased the recovery rate compare to control cells (Fig. 4L; Fig. S3A, Movies 6–8). To measure the velocity of poleward flux directly, the region of interest (ROI) in the spindle was bleached continuously while time-lapse images were captured every 1 s to record the expansion of bleached region. As expected, depletion of Mdp3 increased the speed of poleward flux (Fig. 4M; Fig. S3B, Movies 9–11), suggesting that Mdp3 negatively regulates minus-end MT dynamics and MT flux by inhibiting Kif2a recruitment.

MT minus ends are clustered at a high density in the globular centrosomal regions and are capped with γ-TURC, and for this reason, the molecular machinery of spindle dynamics at the minus end remains relatively uncharacterized compared with the plus end (Kollman et al., 2011). The data presented here suggest a new regulatory mechanism for MT minus-end dynamics by demonstrating that the DDA3 complex modulates the level of Kif2a at the minus end of MTs to orchestrate minus-end MT dynamics and poleward MT flux. In future studies, it will be important to investigate the molecular linker for the cross-talk between plus- and minus-end MT dynamics. Elucidation of the physiological function and the regulatory circuit of DDA3 complex in mitosis will open the door to investigate its role and regulatory mechanism in chromosome integrity and tumor development.

The full-length Mdp3 gene was purchased from OriGene (UAS) and subcloned into the pCS2+ vector with an N-terminal GFP or Myc tag.

The following antibodies were used for western blotting: anti-HSP70 (SC-66048, Santa Cruz Biotechnology; 1:10,000), anti-p38-MAPK (SC-535, Santa Cruz Biotechnology; 1:2000), anti-Mdp3 (HPA035598, Sigma; 1:1000), anti-cyclin B1 (SC-594, Santa Cruz Biotechnology; 1:2000), anti-Myc (SC-789, Santa Cruz; 1:2000), and anti-Kif2a (NB500-180, Novus Biologicals; 1:2000).

HeLa cells were cultured in Dulbecco's modified Eagle's medium (DMEM, WelGENE Inc.) supplemented with 10% fetal bovine serum (FBS, Invitrogen), penicillin (100 units/ml) and 100 µg/ml streptomycin (Invitrogen) and tested for contamination. The siRNA sequences targeting Mdp3 were 5′-CUACAUUCGUCUACUGAUA-3′ (siMdp3-A) and 5′-UAUGAAGAGUCUGGUAAUA-3′ (denoted siMdp3-B) (Thermo Scientific). The sequence targeting DDA3 was 5′-AAGCAAGACTTCAGTAGCATT-3′ (denoted siDDA3) (Thermo Scientific). The control siRNA (siGL2, denoted siControl) was 5′-CGTACGCGGAATACTTCGATT-3′. siRNAs were transfected into cells using DharmaFect 1 (Dharmacon). The DNA transfection was performed using Lipofectamine 2000 (Invitrogen), according to the manufacturer's protocol.

Immunostaining was performed with standard protocols (Jang et al., 2008). Images were obtained using a 1.4 NA plan-Apo 100× oil immersion lens (Carl Zeiss) and deconvolved using AutoDeblur v9.1 and AutoVisualizer v9.1 (AutoQuant Imaging). All of the images are maximum projections from deconvolved z-stacks of 0.2-µm thick focal planes. The images for quantification were acquired under a constant exposure in each channel for all cells and measured fluorescence intensity using AutoVisualizer v9.1. Background levels from ROIs outside of the spindle and with no internal reference were subtracted from the results. Primary antibodies used were against Mad2 (PA5-21594, Thermo Scientific Pierce Antibodies; 1:100), BubR1 (LS-C2771, LifeSpanBioSciences; 1:100), CREST (15-235-0001, Antibodies Incorporated; 1:100), Hec1 (GTX70268, Gentex; 1:100), CENP-E (sc-22790, Santa Cruz Biotechnology; 1:100), Mdp3 (HPA035598, Sigma; 1:100), Kif2a (NB500-180, Novus Biologicals; 1:100), and β-tubulin (E7 monoclonal antibody, AB_2315513, Developmental Studies Hybridoma Bank; 1:100).

For time-lapse microscopy, HeLa cells stably expressing GFP–histone-H2B were cultured in Leibovitz's L-15 medium (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen) and 2 mM L-glutamine (Invitrogen). The images were acquired every 3 min for 5 h with AxioVision 4.8.2 (Carl Zeiss).

For FLIP, FRAP, and MT flux, HeLa cells stably expressing GFP–α-tubulin were placed in a sealed growth chamber heated to 37°C. GFP–α-tubulin in the cytoplasm or in the spindle was photobleached with a laser, and images were acquired with ZEN software (Carl Zeiss) under a LSM 700 confocal microscope (Carl Zeiss) with a 40× lens.

The DDA3 complex was purified as described previously (Cheeseman et al., 2004; Jang et al., 2008). For gel filtration, the liquid chromatography analysis of mitotic cell extracts was performed with Vision workstation (Applied biosystems) at the Korea Basic Science Institute (Seoul). Sample was separated on a Yarra SEC-3000 (7.8×300 mm, Phenomenex, Torrance).

The data were analyzed using Student's t-test. A P-value <0.05 (two-tailed) was considered statistically significant.