Haspin inhibitors reveal centromeric functions of Aurora B in chromosome segregation

Abstract

Haspin phosphorylates histone H3 at threonine-3 (H3T3ph), providing a docking site for the Aurora B complex at centromeres. Aurora B functions to correct improper kinetochore-microtubule attachments and alert the spindle checkpoint to the presence of misaligned chromosomes. We show that Haspin inhibitors decreased H3T3ph, resulting in loss of centromeric Aurora B and reduced phosphorylation of centromere and kinetochore Aurora B substrates. Consequently, metaphase chromosome alignment and spindle checkpoint signaling were compromised. These effects were phenocopied by microinjection of anti-H3T3ph antibodies. Retargeting Aurora B to centromeres partially restored checkpoint signaling and Aurora B-dependent phosphorylation at centromeres and kinetochores, bypassing the need for Haspin activity. Haspin inhibitors did not obviously affect phosphorylation of histone H3 at serine-10 (H3S10ph) by Aurora B on chromosome arms but, in Aurora B reactivation assays, recovery of H3S10ph was delayed. Haspin inhibitors did not block Aurora B localization to the spindle midzone in anaphase or Aurora B function in cytokinesis. Thus, Haspin inhibitors reveal centromeric roles of Aurora B in chromosome movement and spindle checkpoint signaling.

Figure 1.

Haspin inhibitors reduce H3T3ph and displace Aurora B from centromeres but not the central spindle. (A) HeLa cells were released from double thymidine block and, after 7 h, 5 µM nocodazole was added for 6 h. Mitotic cells collected by “shake-off” were replated in 5 µM nocodazole, 20 µM MG132, and Haspin inhibitors for 1 h. Immunoblots of cell lysates are shown. (B) U2OS cells were released from thymidine block and, after 7 h, 33 nM nocodazole was added. After another 4 h, 0.33 µM nocodazole, 20 µM MG132, and kinase inhibitors were added for 1 h before fixation and immunofluorescence microscopy. To visualize residual H3T3ph, two different red channel exposure times are shown. (C) The ratio of centromere to chromosome arm Aurora B intensity was determined for cells treated as in B (15 centromeres/cell; n = 8 or 9 cells). Means + SD are shown (error bars); ***, P < 0.001 vs. DMSO. (D) Asynchronous U2OS cells were treated with Haspin inhibitors for 2 h. Immunofluorescence microscopy of anaphase cells is shown. Bars, 5 µm.

Figure 2.

Haspin inhibitors influence the maintenance of Aurora B activity toward centromeric targets. (A) Nocodazole-arrested U2OS cells were obtained as in Fig. 1 B, then kinase inhibitors and 20 µM MG132 were added for 1 h in the presence (MCAK and Aurora B) or absence (CENP-AS7ph and H3S10ph) of 0.33 µM nocodazole. Mitotic Aurora B localization, the centromeric staining intensity of MCAK and CENP-AS7ph, and H3S10ph intensity on chromosomes were classified for at least 100 cells in each condition in one experiment by immunofluorescence microscopy. Similar results were obtained in duplicate experiments. (B and C) Example images of cells treated as in A. Bars, 5 µm. (D) The intensities of CENP-AS7ph and H3S10ph on mitotic chromosomes in each condition were quantified (n = 6–14 cells). Results are expressed as a ratio to centromeric autoantigen staining intensity at the same centromeres. Means + SD are shown (error bars); ***, P < 0.001 vs. DMSO.

Figure 3.

Artificial retargeting of Aurora B restores MCAK and CENP-AS7ph at centromeres of Haspin-inhibited cells. (A and B) HeLa cells were transfected with CENP-B–EGFP or EGFP–CENP-B–INCENP plasmids between double thymidine treatments. 7 h after release from G1/S, 30 nM nocodazole was added to accumulate mitotic cells. Then, 3.5 h later, medium containing 20 µM MG132 with or without 10 µM Haspin inhibitors or 50 nM Hesperadin, in the presence (A) or absence (B) of 0.33 µM nocodazole, was added for 75 min. Approximately 100 mitotic cells in each condition from one experiment were classified according to the intensity of centromeric MCAK (A) or CENP-AS7ph (B) by immunofluorescence microscopy. Similar results were obtained in a second experiment. Bars, 5 µm.

Figure 4.

Haspin inhibitors delay Aurora B activation. (A) Treatment scheme for Aurora B reactivation assays. (B) Immunofluorescence microscopy of CENP-AS7ph and H3S10ph in HeLa cells treated as in A. Bar, 5 µm. (C) Approximately 100 mitotic cells in each condition from one experiment were classified according to the intensity of CENP-AS7ph or H3S10ph staining. Similar results were obtained in a second experiment using the Aurora B inhibitor ZM447439.

Figure 5.

Haspin inhibitors compromise KT-MT attachment correction. (A) HeLa cells were released from thymidine treatment and, after 7 h, 100 µM monastrol was added for 3 h to accumulate cells in mitosis with incorrect KT-MT attachments. Then 50 nM Hesperadin was added to inhibit Aurora B, together with 20 µM MG132. After 1.5 h, monastrol and Hesperadin were removed by washing into fresh medium containing Haspin inhibitors or controls in the continued presence of MG132. Approximately 200 cells were classified in each condition by fluorescence microscopy. Means ± SD are shown (error bars), n = 3. (B) Immunofluorescence microscopy of Dsn1-S109ph during Aurora B reactivation in the presence or absence of Haspin inhibitors in cells treated as in Fig. 4 A. (C) Approximately 100 mitotic cells in each condition from one experiment as in B were classified according to phosphorylated Dsn1 (Dsn1-S109ph) or total Dsn1 (Dsn1) staining intensity at kinetochores. Similar results were obtained in a duplicate experiment. (D) HeLa cells were transfected with plasmids encoding CENP-B–EGFP or EGFP–CENP-B–INCENP between and after double thymidine treatments, then treated as in B. (E) The intensity of Dsn1-S109ph staining in EGFP-positive cells from D was classified in one experiment as in C. Similar results were obtained in a duplicate experiment. Bars, 5 µm.

Figure 6.

Haspin inhibitors compromise chromosome alignment. (A) U2OS cells expressing Histone-H2B-mRFP and γ-tubulin–GFP were exposed to vehicle alone (DMSO), and mitotic progression was followed by live confocal fluorescence microscopy. Maximum intensity projections of H2B-mRFP fluorescence from selected frames are shown. Video 1 shows complete data including γ-tubulin–GFP fluorescence. (B and C) As above, for a cell treated with 10 µM LDN-211898 (B) or 10 µM 5-iodotubercidin (C). Arrowheads indicate misaligned or lagging chromosomes. Asterisks indicate the first frame in which cytokinetic furrowing was observed. Also see Videos 2 and 3. Bars, 10 µm. (D) Mitotic defects enumerated from live imaging movies. See Fig. S4 C for further details.

Figure 7.

Haspin inhibitors compromise the spindle checkpoint response to 5 µM nocodazole. (A) HeLa cells were synchronized by thymidine treatment and, 7 h after release, 5 µM nocodazole was added for 7.5 h. Mitotic cells were harvested by mitotic “shake-off” and replated in the continued presence of 5 µM nocodazole together with 5-iodotubercidin and the Aurora B inhibitor ZM447439. After 13.5 h, total cell lysates were analyzed by immunoblotting. (B) HeLa cells were transfected with EGFP–CENP-B or CENP-B–INCENP–EGFP plasmids between and after double thymidine treatments, and then treated essentially as in A, except that cells were replated on coverslips coated with poly-d-lysine. Mitotic indices were determined from ∼100 cells in each condition by DNA staining and fluorescence microscopy (n = 3). Means + SD are shown (error bars); ***, P < 0.001; ND, not detected. (C and D) As for A, but using LDN-192960 (C) or LDN-211898 (D). Black lines indicate that intervening lanes have been spliced out. (E–H) U2OS cells expressing Histone-H2B-mRFP and γ-tubulin–GFP were synchronized and treated with 5 µM nocodazole essentially as described for HeLa cells in A. Mitotic cells were replated in imaging dishes and kinase inhibitors were added in the continued presence of 5 µM nocodazole. Each symbol represents the time at which a cell began to exit from mitosis or die in mitosis, as determined from live imaging series collected over 15 h.

Figure 8.

Haspin inhibitors delay BubR1 recruitment to kinetochores. (A) Immunofluorescence microscopy of BubR1 during Aurora B reactivation in the presence or absence of Haspin inhibitors in HeLa cells treated as in Fig. 4 A, but using 5 µM nocodazole throughout. Bar, 5 µm. (B) Approximately 100 mitotic cells in each condition from A were classified according to BubR1 intensity at kinetochores. Means ± SD are shown (error bars), n = 3. (C) HeLa cells were transfected with plasmids encoding EGFP–CENP-B or CENP-B–INCENP–EGFP between and after double thymidine treatments, and then treated essentially as in A. Bars, 5 µm. (D) For cells in C, the intensity of kinetochore BubR1 was classified in at least 100 cells per condition in one experiment. Similar results were obtained in a duplicate experiment.

Figure 9.

Microinjection of mitotic cells with anti-H3T3ph compromises the spindle checkpoint response in combination with Aurora B inhibition. (A) LLC-PK expressing EGFP-topoisomerase IIα arrested in mitosis with 0.17 µM (left) or 3.3 µM nocodazole (right) were injected with anti-H3T3ph solution containing Dextran Texas red, and time-lapse phase contrast and fluorescence images were collected every 15 min. Note the nuclear reformation at the last time point in each case. Times are in hours:minutes. Bar, 10 µm. (B–E) After treatment with the nocodazole and ZM447439 combinations indicated, exit from mitosis or death in mitosis was enumerated from live imaging series collected over 15 h as in A.