The Aurora-B-mediated phosphorylation of SHCBP1 regulates cytokinetic furrow ingression

The central spindle is an array of antiparallel and interdigitating microtubules that forms between two segregating chromosomes and is essential for cleavage furrowing (Glotzer, 2009). Centralspindlin is a central spindle-localized protein complex that is composed of the plus-end-directed motor protein MKLP1/KIF23 and the Rho family GTPase-activating protein MgcRacGAP/CYK-4/RacGAP1 (Mishima et al., 2002; White and Glotzer, 2012). Accumulating evidence has demonstrated that centralspindlin plays a pivotal role in the activation of RhoA at the cell equator (Yüce et al., 2005; Nishimura and Yonemura, 2006; Petronczki, et al., 2007; Miller and Bement, 2009). Once MgcRacGAP is phosphorylated by PLK1, ECT2 is recruited to the central spindle to activate RhoA (Burkard et al., 2009; Wolfe et al., 2009). Although this model clearly shows that MgcRacGAP is essential for RhoA activation, the functional role of the GAP domain remains elusive (Canman et al., 2008; Jordan and Canman, 2012; Loria et al., 2012). A recent study showed that MgcRacGAP inactivates Rac1 to promote furrowing in mammalian cells (Bastos et al., 2012); however, it remains unknown whether there are any mechanisms that regulate the GAP activity during cytokinesis.

Centralspindlin has multiple roles and associates with several proteins during cytokinesis. RacGAP50C, a Drosophila melanogaster ortholog of MgcRacGAP, interacts with anillin to link the central spindle with the actomyosin contractile ring to allow furrowing (D'Avino et al., 2008; Gregory et al., 2008). During the final stage of cytokinesis, centralspindlin associates with FIP3, CEP55 and ARF6 to recruit recycling endosomes to the midbody to complete membrane abscission (Simon et al., 2008; Zhao et al., 2006; Joseph et al., 2012). A recent study identified a novel partner of centralspindlin named Nessun Dorma, which was shown to be essential for the cytokinesis of germ-line cells in D. melanogaster (Montembault et al., 2010). In this study, we show that AurB phosphorylates SHCBP1, a human ortholog of Nessun Dorma, to promote the inactivation of Rac1 by MgcRacGAP and induce furrow ingression.

To gain insight into the functional role of SHCBP1, we examined whether SHCBP1 was phosphorylated during mitosis. Nocodazole-released HeLa cells were lysed and affinity precipitated using an anti-SHCBP1 antibody, and the immunoprecipitates were subjected to silver staining after gel electrophoresis. The SHCBP1 band was excised, and the phosphorylation status was determined by mass spectrometry analysis. In this analysis, we found that Ser634 of SHCBP1 was phosphorylated during mitosis. Ser634 is located in the consensus sequence of the Aurora B (AurB) phosphorylation site and is conserved among vertebrates, but there are no clear corresponding serine residues in D. melanogaster and Caenorhabditis elegans (Fig. 1A). We performed an in vitro kinase assay to determine whether AurB directly phosphorylates Ser634. Wild-type SHCBP1 (WT-SHCBP1) was phosphorylated by AurB, and the substitution of Ser634 with alanine (to generate S634A-SHCBP1) abolished this phosphorylation (Fig. 1B). To confirm the phosphorylation of Ser634 in cells, we generated an antibody that detects phosphorylated Ser634. The pS634 antibody detected recombinant WT-SHCBP1 phosphorylated by AurB, but not S634A-SHCBP1 (Fig. 1C). The antibody did not detect phosphatase-treated SHCBP1 (Fig. 1D). Because the antibody detected various phosphorylated proteins in immunoblots of whole-cell extracts (supplementary material Fig. S1A), we immunoprecipitated SHCBP1 with the anti-SHCBP1 antibody and then used the pS634 antibody to detect the phosphorylation of the endogenous protein. The addition of Aurora kinase inhibitors eliminated mitotic Ser634 phosphorylation (Fig. 1E). Ser634 was phosphorylated during mitosis and dephosphorylated once the cells exited mitosis (Fig. 1F). These results show that Ser634 of SHCBP1 is phosphorylated by AurB during mitosis. We next determined if AurB and SHCBP1 are colocalized. As previously reported, AurB localized to the central spindle during anaphase and to flanking regions of the midbody during cytokinesis (Fig. 1G). SHCBP1 was diffusely localized between two separating chromosomes during early anaphase and started to accumulate at the central spindle during late anaphase and at the midbody during cytokinesis. This specific localization pattern was not observed in cells transfected with SHCBP1 siRNA (supplementary material Fig. S1B). Although AurB and SHCBP1 were colocalized at the central spindle during late anaphase, they exhibited different localizations during early anaphase and cytokinesis.

To determine the physiological role of Ser634 phosphorylation, we exogenously expressed GFP-tagged WT-SHCBP1, S634A-SHCBP1 and S634D-SHCBP1, a phospho-mimic mutant, in HeLa cells. The expression of WT-SHCBP1 and S634D-SHCBP1 resulted in ∼20% and 40% of multinuclear cells, respectively. Surprisingly, nearly 70% of S634A-SHCBP1 cells became multinuclear (Fig. 2A). We monitored the division of cells expressing each protein by time-lapse microscopy and observed three types of cytokinesis defects: an initiation defect, a furrowing defect and an abscission defect (Fig. 2B). An initiation defect indicates that the cells did not initiate furrowing, and a furrowing defect indicates that the cells underwent furrow regression. Cells that completed furrowing and formed the midbody but eventually fused back together to form binuclear cells were categorized as having an abscission defect. Analysis using time-lapse microscopy revealed that cells with an initiation defect or a furrowing defect were significantly more frequent among S634A-SHCBP1 cells than among WT-SHCBP1 cells and S634D-SHCBP1 cells (Fig. 2C). Interestingly, S634D-SHCBP1 expression induced an abscission defect, which may be the reason for the increased percentage of multinuclear cells in S634D-SHCBP1 cells compared to WT-SHCBP1 cells in Fig. 2A.

We next examined the localization of RhoA and anillin, which are essential for furrowing, in the presence of exogenously expressed SHCBP1 protein. In most of the GFP-expressing cells, RhoA was localized to the equatorial cortex during anaphase (Fig. 2D). Although the expression of WT-SHCBP1 suppressed RhoA localization to some extent, more than 80% of S634A-SHCBP1-expressing cells exhibited reduced RhoA accumulation at the cortex (Fig. 2D). In addition, the expression of S634A-SHCBP1 significantly suppressed anillin localization to the contractile ring (Fig. 2E). These results indicate that the AurB-mediated phosphorylation of Ser634 of SHCBP1 plays a pivotal role in the regulation of furrow ingression.

In the immunofluorescence analysis, we noticed that GFP–WT-SHCBP1 and GFP–S634A-SHCBP1 exhibited rather different localization patterns during early anaphase. Approximately 20% of early anaphase cells showed localization of GFP–WT-SHCBP1 or GFP–S634D-SHCBP1 at the central spindle (Fig. 3A). Interestingly, more than 60% of early anaphase cells showed clear localization of GFP–S634A-SHCBP1 at the central spindle. All the proteins accumulated similarly at the midbody (Fig. 3A). The difference in localization during early anaphase did not result from the higher expression of GFP–S634A-SHCBP1, as the transfection efficiencies were almost identical, and the expression levels of GFP–WT-SHCBP1 and GFP–S634A-SHCBP1 were equal (supplementary material Fig. S2A). We also performed time-lapse analysis and observed the premature localization of GFP–S634A-SHCBP1 at the central spindle (Fig. 3B,C). Our attempts to examine the localization of phosphorylated SHCBP1 using an anti-pS634 antibody were unsuccessful because the antibody reacted with additional phosphorylated proteins, and the signal at the central spindle was not reduced by SHCBP1 siRNA transfection (supplementary material Fig. S2B).

The localization of SHCBP1 is dependent on its association with centralspindlin (Montembault et al., 2010). We next examined whether the phosphorylation of SHCBP1 affects this protein's association with one component of centralspindlin, MgcRacGAP. We expressed GFP–WT-SHCBP1 or GFP–S634A-SHCBP1 in HeLa cells, incubated the cells with or without nocodazole and performed immunoprecipitation with an anti-GFP antibody to obtain phosphorylated and non-phosphorylated SHCBP1 bound to agarose beads. The beads were incubated with recombinant GST–MgcRacGAP, and then GST–MgcRacGAP bound to SHCBP1 was precipitated and subjected to immunoblot analysis. As shown in Fig. 3D, GFP–WT-SHCBP1 from non-treated cells associated with GST–MgcRacGAP to a greater extent than GFP–WT-SHCBP1 from nocodazole-arrested cells did. By contrast, GFP–S634A-SHCBP1 from nocodazole-arrested cells and untreated cells associated similarly with GST–MgcRacGAP. We also examined the association of GFP–S634D-SHCBP1 and GST–MgcRacGAP. Compared with GFP–S634A-SHCBP1, GFP–S634D-SHCBP1 showed reduced affinity toward GST–MgcRacGAP (Fig. 3E). We next examined the interaction of endogenous SHCBP1 and MgcRacGAP. HeLa cells incubated with or without nocodazole were lysed, immunoprecipitated with an anti-SHCBP1 antibody and subjected to immunoblot analysis. The association of SHCBP1 with MgcRacGAP was reduced in nocodazole-arrested cells (Fig. 3F). These results indicate that the AurB-mediated phosphorylation of Ser634 reduces the association of SHCBP1 with MgcRacGAP and prevents the premature localization of SHCBP1 to the central spindle. Although Ser634 is not located in the region that binds to MgcRacGAP (Montembault et al., 2010), the phosphorylation may induce structural changes that hide the binding region for MgcRacGAP.

The inactivation of Rac1 by MgcRacGAP is essential for the completion of cytokinesis in mammalian cells (Bastos et al., 2012). We performed an in vitro GAP assay to determine whether SHCBP1 has any effects on the MgcRacGAP-mediated inactivation of Rac1. Consistent with previous reports (Touré et al., 1998; Jantsch-Plunger et al., 2000; Kawashima et al., 2000), MgcRacGAP exhibited GAP activity toward Rac1 and Cdc42. WT-SHCBP1 suppressed the MgcRacGAP-stimulated GTP hydrolysis activity of Rac1 in a concentration-dependent manner (Fig. 4A). Δ1–270-SHCBP1, which does not associate with MgcRacGAP (supplementary material Fig. S3A), and S634D-SHCBP1 did not affect the GAP activity (Fig. 4B). In addition, Δ1–443 MKLP1 (deleted of amino acids 1–443), lacking the kinesin motor domain that is dispensable for the association with MgcRacGAP, did not inhibit the GAP activity (supplementary material Fig. S3B). These results show that SHCBP1 can suppress the MgcRacGAP-mediated inactivation of Rac1.

We next examined whether Rac1 inactivation can rescue the cytokinesis defect induced by S634A-SHCBP1. HeLa cells that constitutively expressed wild-type or dominant-negative Rac1 were established by retrovirus infection. The cells were transfected with a plasmid encoding GFP, GFP–WT-SHCBP1 or GFP–S634A-SHCBP1, and cell division was monitored by time-lapse microscopy. The inhibition of Rac1 activity significantly rescued the initiation defect and the furrowing defect induced by GFP–S634A-SHCBP1 expression (Fig. 4C). The furrowing defect induced by GFP–WT-SHCBP1 expression was also rescued by Rac1 inhibition (Fig. 4C). These results suggest that S634A-SHCBP1 inhibits furrowing by activating Rac1. Interestingly, the abscission defect was promoted by Rac1 inhibition in the presence of either GFP–WT-SHCBP1 or GFP–S634A-SHCBP1. Although further analysis is needed, SHCBP1 may play some role in abscission in combination with Rac1.

In this study, we identified a novel regulatory mechanism for cytokinetic furrow ingression mediated by the phosphorylation of SHCBP1 by AurB. An in vitro kinase assay and immunoblot analysis revealed that Ser634 of SHCBP1 is phosphorylated by AurB during mitosis. An AurB phosphorylation site mutant, S634A-SHCBP1, was prematurely recruited to the central spindle during anaphase and induced a furrowing defect. These results indicate that AurB-mediated phosphorylation negatively regulates SHCBP1 localization to the central spindle to promote furrow ingression. Rac1 is inactive at the cell equator (Yoshizaki et al., 2003), and the inactivation of Rac1 by MgcRacGAP is required for the localization of active RhoA (Bastos et al., 2012). We found that SHCBP1 suppressed the MgcRacGAP-mediated inactivation of Rac1 in vitro. The inactivation of Rac1 rescued the initiation and furrowing defect induced by S634A-SHCBP1. Taken together, our results indicate that phosphorylation by AurB negatively regulates SHCBP1 localization and promotes the inactivation of Rac1 by MgcRacGAP to induce the ingression of the cleavage furrow (Fig. 4D).

HeLa cells were propagated in DMEM supplemented with 10% FBS. To generate an anti-SHCBP1 antibody, the N-terminus of SHCBP1 (aa 1–316) with a 6xHis tag was injected into a rabbit, and the antibody produced was purified using HiTrap NHS-activated HP columns (GE Healthcare BioScience, Uppsala, Sweden) coupled to recombinant 6xHis-SHCBP1 (aa 1–316). An anti-pSer634 antibody was generated by injecting a peptide [KKKRLS(P)ELGIC] into a rabbit (MBL, Nagoya, Japan). Other antibodies were purchased from the following manufacturers: anti-MgcRacGAP antibody, Bethyl Laboratories (Montgomery, TX); anti-α-tubulin and anti-β-actin antibodies, Sigma; anti-RhoA and anti-anillin antibodies, Santa Cruz (Santa Cruz, CA); and anti-AurB antibody, BD Biosciences (San Jose, CA). Anti-GFP antibodies were obtained from NeuroMab (Davis, CA) and MBL.

Cells were fixed with ice-cold methanol/acetone (1∶1) and blocked with PBS containing 7% FBS for 30 minutes. The cells were incubated with the primary antibody for 1 hour, incubated with Alexa-Fluor-488- or -594-labeled secondary antibodies (Invitrogen) for 1 hour and then analyzed. For the detection of RhoA localization, cells were fixed with 10% TCA (Yonemura et al., 2004). The dilutions of antibodies used were: anti-SHCBP1 (1∶100), anti-MgcRacGAP (1∶100), anti-α-tubulin (1∶500), anti-RhoA (1∶50), anti-anillin (1∶100), anti-GFP (1∶200) and anti-AurB (1∶100). Images were taken using an FV1000 confocal microscope (1.4 NA 100× oil immersion objective) or a BX60 fluorescence microscope (Olympus, Tokyo Japan).

Cells cultured on glass-bottom dishes were transfected with plasmids (1 µg) using Lipofectamine 2000 (Invitrogen). Starting 12 hours later, the cells were monitored using a time-lapse microscope system (LCV110, Olympus) equipped with Retiga Exi camera (QImaging, Tokyo, Japan) for 24 hours. Images were analyzed using the MetaMorph Imaging System (Universal Imaging, Silicon Valley, CA).

GST–MgcRacGAP (full length), GST–WT-SHCBP1 (full length), GST–S634D-SHCBP1 (full length) and GST–Δ1–270-SHCBP1 (deletion of aa 1–270) were purified from E. coli. An in vitro GAP assay was performed using the RhoGAP ASSAY Biochem Kit (Cytoskeleton, Denver, CO).

Three independent experiments were performed, and the results were compared using Student's t-test. The data are represented as the means ± s.d.