PKCɛ switches Aurora B specificity to exit the abscission checkpoint

Abstract

The 'NoCut', or Aurora B abscission checkpoint can be activated if DNA is retained in the cleavage furrow after completion of anaphase. Checkpoint failure leads to incomplete abscission and a binucleate outcome. These phenotypes are also observed after loss of PKCɛ in transformed cell models. Here we show that PKCɛ directly modulates the Aurora B-dependent abscission checkpoint by phosphorylating Aurora B at S227. This phosphorylation invokes a switch in Aurora B specificity, with increased phosphorylation of a subset of target substrates, including the CPC subunit Borealin. This switch is essential for abscission checkpoint exit. Preventing the phosphorylation of Borealin leads to abscission failure, as does expression of a non-phosphorylatable Aurora B S227A mutant. Further, depletion of the ESCRT-III component and Aurora B substrate CHMP4C enables abscission, bypassing the PKCɛ-Aurora B exit pathway. Thus, we demonstrate that PKCɛ signals through Aurora B to exit the abscission checkpoint and complete cell division.

Figure 1. PKCɛ specifically phosphorylates Aurora B S227 at the midbody during cytokinesis.

(a) A midbody protein-biased peptide array of PKCɛ putative substrates was assayed for ³²P-ATP incorporation and normalized to the pseudosubstrate peptide as a positive control. Data for all peptides analysed in Supplementary Data 1. (b) In vitro kinase assay of PKCɛ and Aurora B with and without the PKC inhibitor bis-indolylmeleimide I (BIM) (1 μm). (c) Protein sequence alignment of human Aurora kinase family members and Aurora B of various species. S227 (human Aurora B sequence) is highlighted in red, T232 is boxed in red. (d) Proximity ligation assay between Aurora B and PKCɛ in DLD1 GFP-Aurora B WT cell line. The interaction between CPC members Aurora B and INCENP served as a positive control for this assay. Primary antibodies against the endogenous proteins were used for the detection of Aurora B, PKCɛ and INCENP. DAPI (blue), proximity ligation assay (PLA) (red), GFP-Aurora B (green). Scale bar, 10 μm. (e) Confocal imaging shows Aurora B pS227 (green) staining at the cytokinesis midbody (tubulin—red) in DLD1 cells (white arrows and inset) and not in the presence of the PKCɛ selective inhibitor BLU577 (500 nM) for 30 min. Cells were scored for the presence or absence of Aurora B pS227 at the midbody (statistics analysed by Student's t-test; ****P<0.0001; error bars represent mean±s.e.m.). A minimum of 30 high-resolution, single-cell images per condition from 12 experiments in two different cell lines were acquired, a representative image is shown here. Scale bar, 10 μm.

Figure 2. Aurora B S227 phosphorylation is required for successful completion of cytokinesis.

(a,b) Cells where PKCɛ is inhibited and/or Aurora B cannot be S227 phosphorylated do not undergo successful abscission. (a) The number of binucleate cells was assessed after 24 h induction of GFP-Aurora B WT (4.07%±1.45), GFP-Aurora B S227A (12.03%±2.27)±BLU577 (500 nM) (WT 9.97%±2.01 versus S227A 14.5%±2.94) expression in the DLD1 cell lines. Graph represents the mean (±s.e.m.) of three independent experiments of >500 scored cells per condition. Student's t-test, not significant (NS)=P>0.05, *=P≤0.05, **=P≤0.01, ***=P≤0.001. (b) Wide-field time-lapse microscopy of DLD1 GFP-Aurora B cell lines. Cells were scored for the outcome of cytokinesis where failed cytokinesis is a binucleate cell and successful cytokinesis resulted in two daughter cells. Cells were induced for GFP-Aurora B expression for 16 h before image acquisition. Graph represents the mean (±s.e.m.) of three independent experiments where a minimum of 100 cells were scored per condition. Two-way analysis of variance (ANOVA), ***=P≤0.001. (c,d) There is an increase in cells with DNA trapped in the cytokinesis furrow if PKCɛ is inhibited and Aurora B cannot be phosphorylated on S227. (c) Confocal images of DLD1 GFP-Aurora B (green) cell lines stained for Lap2β (red) to identify DNA (DAPI—blue) bridging during cytokinesis. Scale bar, 10 μm. (d) DLD1 GFP-Aurora B cell lines, which stably express RFP-Lap2β were assessed for the presence of a Lap2β-positive bridge as they underwent cytokinesis (WT 6.25%±12.5 versus S227A 43.125%±19.2; left panel) and the outcome of cytokinesis (binucleate cells: WT 6.25%±12.5 versus S227A 80.9%±21.5; right panel). Graph represents the mean (±s.e.m.) of three independent experiments where a minimum of 100 cells scored per condition. Two-way ANOVA, *=P≤0.05, ***=P≤0.001.

Figure 3. Aurora B phosphorylated on S227/T232 has unique set of substrates including Borealin S165.

(a) A peptide array of known Aurora B substrates compared the phosphorylation of peptides by either doubly phosphorylated (WT) Aurora B to those phosphorylated by singly phosphorylated (S227A) Aurora B. Graph shows ratio of ATP incorporation into peptides phosphorylated by Aurora B WT and S227A recombinant protein. Data for all peptides analysed in Supplementary Data 2. (b) In vitro kinase assay to detect Borealin Ser165 phosphorylation by recombinant Aurora B(WT) or (S227A) protein. (c,d) DLD1 cells were induced to express GFP-Borealin WT or GFP-Borealin S165A for 16 h before analysis. (c) Representative confocal images of GFP-Borealin cells in each phase of mitosis and cytokinesis. Scale bar, 10 μm. (d) DLD1 cells induced to express GFP-Borealin were scored for the outcome of cytokinesis, successful cytokinesis resulting in two daughter cells or failed cytokinesis resulting in a binucleate cell. Graph is the mean (±s.e.m.) of three independent experiments where more than 100 cells per condition were scored. Two-way analysis of variance, **=P≤0.01.

Figure 4. PKCɛ and Aurora B influence CHMP4C localization during cytokinesis.

(a) DLD1 GFP-Aurora B cell lines were transiently transfected with three separate CHMP4C siRNA, imaged using live-cell time-lapse microscopy and scored for outcome of cytokinesis; successful cytokinesis resulting in two daughter cells or failed cytokinesis resulting in a binucleate cell. Graph represents the mean (±s.e.m.) of three independent experiments; a minimum of 50 cells were counted per experiment. Two-way analysis of variance (ANOVA), **=P≤0.01, ***=P≤0.001. (b) HeLa cells were transfected with PKCɛ siRNA, a pool of three CHMP4C siRNA or PKCɛ siRNA and the pool of CHMP4C siRNA, imaged using live-cell time-lapse micrsocopy and scored for outcome of cytokinesis as per the criteria above. Graph represents the mean (±s.e.m.) of three independent experiments; a minimum of 50 cells were counted per experiment. Two way ANOVA, not significant (NS)=P>0.05, **=P≤0.01. (c) DLD1 GFP-Aurora B WT and S227A cell lines (green) were transiently transfected with HA-CHMP4C (red) to look for midbody localization during cytokinesis. Cells were scored for the presence or absence of HA-CHMP4C at the midbody. A minimum of 12 high-resolution, single-cell images per condition from four experiments were acquired; a representative image is shown here. Scale bar, 10 μm. Student's t-test, *=P≤0.05. (d) DLD1 GFP-Borealin WT and S165A cell lines (green) were transiently transfected with HA-CHMP4C (red) to look for midbody localization during cytokinesis. Cells were scored for the presence or absence of HA-CHMP4C at the midbody. A minimum of five high-resolution, single-cell images per condition from two experiments were acquired; a representative image is shown here. Scale bar, 10 μm. (e) Working model: chromatin trapped in the cytokinesis furrow engages the Aurora B-dependent abscission checkpoint. We propose that the association through T232 only phosphorylated Aurora B and Borealin, CHMP4C is maintained in a S210 phosphorylated, closed, inactive conformer distal to the midbody. On bridge resolution, PKCɛ phosphorylates Aurora B on S227, which in turn phosphorylates Borealin S165, allowing for CHMP4C to assume an open, active conformation to polymerize with other ESCRT-III components and facilitate successful abscission.