Inhibitor-3 ensures bipolar mitotic spindle attachment by limiting association of SDS22 with kinetochore-bound protein phosphatase-1

Abstract

Faithful chromosome segregation during mitosis is tightly regulated by opposing activities of Aurora B kinase and protein phosphatase-1 (PP1). PP1 function at kinetochores has been linked to SDS22, but the exact localization of SDS22 and how it affects PP1 are controversial. Here, we confirm that SDS22 is required for PP1 activity, but show that SDS22 does not normally localize to kinetochores. Instead, SDS22 is kept in solution by formation of a ternary complex with PP1 and inhibitor-3 (I3). Depletion of I3 does not affect the amount of PP1 at kinetochores but causes quantitative association of SDS22 with PP1 on KNL1 at the kinetochore. Such accumulation of SDS22 at kinetochores interferes with PP1 activity and inhibits Aurora B threonine-232 dephosphorylation, which leads to increased Aurora B activity in metaphase and persistence in anaphase accompanied with segregation defects. We propose a model in which I3 regulates an SDS22-mediated PP1 activation step in solution that precedes SDS22 dissociation and transfer of PP1 to kinetochores, and which is required for PP1 to efficiently antagonize Aurora B.

Keywords: Aurora B; chromosome segregation; kinetochore; mitosis; protein phosphatase‐1.

Figure 1. Inhibitor-3 (I3) and SDS22 are required for proper chromosome alignment and progression through mitosis

1. Efficiency of siRNA-mediated depletion of indicated PP1-interacting proteins shown by Western blot. p97 was probed as a loading control. Luciferase siRNA (siLuc) was used as a negative control.

2. Increased chromosome misalignment in SDS22- or I3-depleted HeLa cells. Cells were treated with indicated siRNAs, fixed and DAPI-stained. Depletion of the RVXF-motif protein NIPP1 served as a control. Representative confocal images. Scale bar, 5 μm. Percentage of metaphase cells with misaligned chromosomes was quantified. Error bars indicate s.d. of three independent experiments with 60 cells per condition. P-values were calculated using a Mann–Whitney U-test (***P < 0.001; **P ≤ 0.01; *P ≤ 0.05). See Supplementary Fig S1A for rescue experiments.

3. SDS22 or I3 depletion causes delay of anaphase onset. The timing between nuclear envelope breakdown (NEBD) and anaphase onset (AO) was determined in live cells expressing H2B-RFP and IBB-GFP after treatment with indicated siRNAs (see also Supplementary Fig S1B for image sequences of example movies). NEBD was monitored as a nuclear efflux of IBB-GFP, and AO was monitored as the start of separation of the sister chromatids using H2B-RFP. Percentages of cells committing anaphase in the indicated time windows are plotted. n ≥ 133 cells per condition, summed up from three independent experiments.

Source data are available online for this figure.

Figure 2. Depletion of SDS22 and I3 affect PP1 activity at the kinetochore

A Increased Aurora B autophosphorylation upon SDS22 or I3 depletion. Representative images of HeLa cells in metaphase depleted with indicated siRNAs. Cells were stained with Aurora B-specific (AurB), phospho-T232 (phospho-AurB) antibodies and DAPI. Note non-specific staining of the centrosome with pT232 antibodies. Scale bar, 5 μm. B, C Quantification of phospho-AurB (B) or the ratio of phospho-AurB versus total AurB (C) signal intensity on chromatin. Box blots show median, lower and upper quartiles (line and box), 10^th and 90^th percentiles (whiskers) and outliers (•). ***P < 0.001. Data from three independent experiments with 50 cells per condition. D Increased Dsn1 phosphorylation at S109 upon SDS22 or I3 depletion. Representative images of HeLa cells in metaphase treated with indicated siRNAs and stained with Dsn1 phospho-S109-specific antibodies and CREST serum. Note non-specific staining of the centrosome with S109 antibodies. Scale bar, 5 μm. E Quantification of the ratio of Dsn1 pS109 versus CREST signal intensity at kinetochores. Box blots show median, lower and upper quartiles (line and box), 10^th and 90^th percentiles (whiskers) and outliers (•). ***P < 0.001. Data from three independent experiments with 50 cells per condition. F Increased BubR1 recruitment at metaphase kinetochores upon SDS22 or I3 depletion. Representative images of HeLa cells in metaphase treated with indicated siRNAs and stained with BubRI and CREST antibodies. Also note for comparison the absence of BubR1 signal in AurB-depleted cells in spite of severe misalignment. Scale bar, 10 μm. G Quantification of the ratio of BubR1 versus CREST signal intensity on kinetochores of aligned metaphase plates. Box blots show median, lower and upper quartiles (line and box), 10^th and 90^th percentiles (whiskers) and outliers (•). **P < 0.01, ***P < 0.001. Data from three independent experiments with a total of 42 cells per condition. AurB-depleted cells were not quantified due to absence of metaphase plates.

Figure 3. Depletion of SDS22 and I3 does not affect the amount of PP1γ at the kinetochore

1. PP1γ localization is not affected by SDS22 or I3 depletion. Stable GFP-PP1γ-expressing HeLa cells were depleted of indicated proteins, fixed and permeabilized, and DAPI-stained. Representative confocal images. Scale bar, 10 μm.

2. Quantification of GFP-PP1γ intensity at kinetochores normalized to cytoplasmic signal intensity. The graph shows the mean and standard deviation of three independent experiments, in which the intensities of at least 3 kinetochores/cell in at least 10 metaphase cells per condition were measured. n.s., P > 0.05.

Figure 4. Depletion of I3 induces re-localization of SDS22 to kinetochores

1. Neither SDS22 nor I3 colocalize with PP1γ at metaphase kinetochores under control conditions. Stable HeLa cell lines expressing GFP fusions of indicated proteins were fixed and DAPI-stained. SDS22-GFP is expressed under its own promoter in BAC transgenic cells. For localization images of GFP-I3-expressing cells in all mitotic stages see also Supplementary Fig S2. Representative confocal images. Scale bar, 10 μm.

2. Expression levels of GFP fusions relative to endogenous proteins in Western blots stained with specific antibodies. Actin served as a loading control. Asterisks denote non-specific bands.

3. SDS22 re-localization upon I3 depletion. SDS22-GFP cells were treated with indicated siRNAs and processed as in (A). Representative images. Note the SDS22-GFP signal at kinetochores in I3, but not control-depleted cells. Scale bar, 10 μm.

4. Depletion efficacy for individual siRNAs as determined by Western blot.

5. Quantification of (C). Error bars indicate s.d. of three independent experiments with 25 cells per condition. ***P < 0.001.

Source data are available online for this figure.

Figure 5. I3 limits association of SDS22 with KNL1-bound PP1

1. SDS22 localizes to the outer kinetochore upon I3 depletion (see also Supplementary Fig S4). SDS22-GFP cells were treated with indicated siRNAs, fixed and stained with CREST serum. Scale bar, 10 μm. Transient expression of siRNA-resistant GFP-I3 restores the phenotype (Supplementary Fig S3).

2. SDS22 localization to the kinetochore depends on KNL1. SDS22-GFP cells were depleted of I3 alone, or I3 and KNL1, with indicated siRNA combinations. siLuc served as a control and to adjust siRNA concentrations. The KNL1 knockdown efficiency was analyzed by visualizing PP1 localization (Supplementary Fig S5). Scale bar, 10 μm.

3. Depletion efficiency of I3 was confirmed by Western blot.

4. Quantification of (B). Error bars indicate s.d. of 3 independent experiments with 25 cells per condition. ***P < 0.001.

5. I3 depletion stimulates SDS22 association with KNL1-bound PP1. Stable cell lines expressing GFP or a soluble GFP-KNL1 variant were treated with indicated siRNAs. GFP or GFP-KNL1 was affinity-isolated and associated proteins analyzed by Western blot with indicated antibodies. Right panel, quantification of Western blot signals normalized to siLuc. Error bars represent s.d. with n = 3. Note the increase of SDS22 association with PP1γ-KNL1 specifically upon I3 depletion, while PP1γ binding to KNL1 is unaffected.

Source data are available online for this figure.

Figure 6. Forced localization of SDS22 to kinetochores by overexpression inhibits PP1-mediated dephosphorylation of Aurora B

A SDS22 recruitment to kinetochores depends on PP1 binding. Stable GFP-PP1γ cells were transiently transfected with mCherry fusions of SDS22 wild-type (wt) or either of the two PP1-binding-deficient SDS22 variants E192A or W302A, fixed and DAPI-stained. Note also that PP1γ localization is not affected by SDS22 overexpression (see also Supplementary Fig S7 for I3 overexpression). Representative confocal images. Scale bar, 10 μm. B Western blot of overexpression levels probed with dsRED (upper panel) or SDS22 (lower panel) antibodies. C–F Overexpression of SDS22 wild-type (wt), but not PP1-binding-deficient mutants increases Aurora B phosphorylation and BubR1 levels at kinetochores. Cells were transiently transfected with indicated constructs as in (A). Fixed cells were stained with DAPI and either Aurora B-specific and phospho-T232 (phospho-Aurora B) antibodies (C) or CREST and BubR1-specific antibodies (E). Representative confocal images. Scale bars, 5 μm. Quantification of (C) and (E) is shown in (D) and (F), respectively. Box plot as in Fig 2B. Data from three independent experiments with 25 cells per condition. ***P < 0.001. Source data are available online for this figure.

Figure 7. Binding of SDS22 to PP1 inhibits dephosphorylation of Aurora B in vitro

A PP1 specifically dephosphorylates Aurora B pT232. Recombinant, autophosphorylated Aurora B (left panel) or in vitro phosphorylated GST-Survivin (right panel) was incubated either alone or with purified PP1 at indicated concentrations. Lambda phosphatase (λPP) was used as a positive control. Phosphorylation at T232 (Aurora B) and T34 (Survivin) was detected by Western blot with phospho-specific antibodies and equal loading confirmed by Ponceau S staining. Note PP1-mediated dephosphorylation of Aurora B, but not of Survivin. B, C SDS22 inhibits PP1-mediated dephosphorylation of Aurora B at T232 in a dose-dependent manner. Phosphorylated GST-Aurora B was incubated with purified PP1 without or with the indicated concentration of purified SDS22 and phosphorylation monitored as in (A). Equal loading was monitored with GST antibodies. Western blot signals from three independent experiments were quantified (C) and normalized to maximum and minimum values. Error bars represent s.e.m. Source data are available online for this figure.

Figure 8. Depletion of SDS22 or I3 interferes with inactivation of Aurora B at centromeres and chromosome segregation during anaphase

A, B I3 depletion induces SDS22 localization to kinetochores also in anaphase. Stable GFP-SDS22 cells were depleted and processed as in Fig 4C. Scale bar, 10 μm. The percentage of cells with SDS22 on anaphase kinetochores was determined (B). Error bars indicate s.d. of three independent experiments with 20 cells per condition. ***P < 0.001. C HeLa cells were depleted of indicated proteins and processed for immunodetection of Aurora B or pT232 as in Fig 2A. Note persisting phospho-Aurora B on anaphase chromosomes in SDS22- or I3-depleted cells. Scale bar, 5 μm. D, E Quantification of signal intensities in (C). Box plot as in Fig 2B. Data from three independent experiments with 50 cells per condition. ***P < 0.001. F, G Representative images of anaphase cells treated with indicated siRNAs and quantification of cells with anaphase defects. Error bars indicate s.d. of 3 independent experiments with 50 cells per condition. *P < 0.05; **P < 0.01.

Figure 9. Model figure for the role of SDS22 and I3 in the regulation of kinetochore-bound PP1

PP1 dynamically exchanges between a KNL1-bound state at the kinetochore and a SDS22-I3-bound state in solution. SDS22 activates PP1, for example, by contributing to the incorporation of metals at the active site, but keeps it inactive while bound. I3 prevents SDS22-PP1 from binding to KNL1 and may facilitate SDS22 dissociation from PP1. Lack of I3 causes (inactive) SDS22-PP1 to bind KNL1, which then fails to antagonize Aurora B. PP1* denotes active PP1.