Aurora B phosphorylates spatially distinct targets to differentially regulate the kinetochore-microtubule interface

Abstract

Accurate chromosome segregation requires carefully regulated interactions between kinetochores and microtubules, but how plasticity is achieved to correct diverse attachment defects remains unclear. Here we demonstrate that Aurora B kinase phosphorylates three spatially distinct targets within the conserved outer kinetochore KNL1/Mis12 complex/Ndc80 complex (KMN) network, the key player in kinetochore-microtubule attachments. The combinatorial phosphorylation of the KMN network generates graded levels of microtubule-binding activity, with full phosphorylation severely compromising microtubule binding. Altering the phosphorylation state of each protein causes corresponding chromosome segregation defects. Importantly, the spatial distribution of these targets along the kinetochore axis leads to their differential phosphorylation in response to changes in tension and attachment state. In total, rather than generating exclusively binary changes in microtubule binding, our results suggest a mechanism for the tension-dependent fine-tuning of kinetochore-microtubule interactions.

Figure 1. Phosphorylation of the KMN network by Aurora B does not affect kinetochore assembly

(A) In vitro assembly properties of the C. elegans KMN and K_SDM_SDN_SD network. The KNL-1 and MIS-12 complex were co-purified and incubated with the NDC80 complex, followed by gel filtration chromatography of the sample. Elution profiles of the KMN network (top), the K_SDM_SDN_SD network (middle), and the NDC-80 complex alone (bottom) are shown. The KNL-3, KNL-1 and NDC-80 components were detected by Western blotting. Also see Figure S2. (B) Images of live mitotic cells from clonal human cell lines stably expressing moderate amounts of GFP-hDsn1, GFP-hDsn1_SA and GFP-hDsn1_SD. Each fusion protein localizes to kinetochores throughout mitosis. Scale bar, 10 μm. (C) Purification of GFP-hDsn1, GFP-hDsn1_SA and GFP-hDsn1_SD and their associated proteins using a one-step immunoprecipitation to directly isolate the tagged hDsn1 from stable clonal cell lines. Percent sequence coverage from the mass spectrometric analysis of these samples for each KMN network component and associated spindle checkpoint components is indicated. (D) Immunofluorescence images acquired using antibodies against Hec1 (hNdc80), hKNL1 and hDsn1 in HeLa cells under control conditions or after 3 hr treatment with nocodazole, taxol, monastrol, or ZM447439. Scale bar, 10 μm. Also see Figure S2. (E) Bar graph showing the average ratio of kinetochore fluorescence intensities (percent relative to control) for the indicated antibodies/conditions. Error bars represent standard error. The fluorescence intensity of 30–80 kinetochores per cell was measured for 5 to 10 cells per condition.

Figure 2. Combinatorial phosphorylation of the KMN network generates multiple levels of microtubule binding activity

(A) Left, diagrams and cartoon model showing the components and organization of the human and C. elegans KMN networks. (B) Top, gel showing microtubule co-sedimentation of 50 nM NDC-80 complex and NDC-80_SD complex. Bottom, graph showing the average percent protein bound relative to microtubule concentration for the wild type NDC-80 complex, NDC-80 complex phosphorylated by Ipl1 (data from (Cheeseman et al., 2006)), and the NDC-80_SD complex. (C) Co-sedimentation assays with the N-terminus of C. elegans KNL-1 and a KNL-1_SD and the full length KNL-1/MIS-12 complex and KNL-1_SD/MIS-12_SD complex. (D, E, and F) Graphs showing the microtubule binding activity for 50 nM input protein of the indicated KMN network mutants. The points were fitted with MATLAB using a modified Hill equation. Error bars represent the standard deviation. Representative gels for the data shown in these graphs and microtubule binding data for a K_SAM_SAN_SA mutant is included in Supplemental Fig. 3.

Figure 3. The phosphorylation of the KMN network has synergistic effects in vivo

(A and B) Growth curve analysis of cellular viability for the non-phosphorylatable (A) or phosphomimetic (B) KMN mutants in chicken DT40 cells. Tetracycline was added at time 0 to the culture to eliminate expression of the wild type protein. The number of cells not stained with trypan blue was counted. (C and D) Summary of phenotypes for complementation experiments in Ndc80, KNL1 or Dsn1 deficient DT40 cells expressing the non-phosphorylatable (C) or phosphomimetic (D) mutants. In cell lines that express Dsn1 in combination with KNL1 or Ndc80 phospho-mutants, the endogenous Dsn1 is also present. 500–1000 cells were counted for each cell line. (E) Chromosome morphology and mitotic spindle shown by DNA staining using DAPI (blue) and α-tubulin staining (green) in the Ndc80, KNL1 or Dsn1 deficient DT40 cells expressing the non-phosphorylatable or phosphomimetic mutants. White arrowheads show kinetochore attachment defects. (F) Outer kinetochore assembly state in the phospho-mutants shown by DNA staining using DAPI (blue) and Ndc80 staining (green). Scale bars: 10 μm.

Figure 4. Aurora B phosphorylation of the KMN network occurs under conditions where attachments must be corrected

(A) Western blots validating the phosphospecific hDsn1 and hKNL1 antibodies with recombinant protein using the indicated conditions. (B) Immunofluorescence images acquired using the indicated antibodies in untreated HeLa cells or in cells treated with nocodazole, taxol, monastrol or ZM447439 for 3 hrs. Numbers indicate the average percent quantitated fluorescence (+/− sem) relative to control treated metaphase cells. Images are scaled equivalently except for the nocodazole treated samples. Phospho-antibodies show spindle pole background staining that is not eliminated by ZM447439 treatment. (C) Ratio of fluorescence intensities for the images in (B) expressed as a percentage relative to controls. Asterix indicates that intensity in taxol treated cells for S60p hKNL1 was not determined due to extensive spindle background in this condition. (D) Immunofluorescence images acquired using the indicated antibodies in HeLa cells treated with 6 ng/ml Nocodazole to generate cells with both aligned (green boxes) and mis-aligned (red boxes) chromosomes. The checkpoint protein Bub1 localizes preferentially to unattached kinetochores or those that are not under tension. Phospho-Dsn1 shows a statistically significant increase (95% confidence interval in a T-test) on mis-aligned chromosomes (125 kinetochores analyzed) relative to aligned chromosomes (240 kinetochores analyzed). Scale bars, 10 μm.

Figure 5. Differential phosphorylation of the KMN network under distinct kinetochore attachment states modulates outer kinetochore microtubule binding activity

(A) Image and linescan showing position of mCherry-Hec1 relative to a Mis12 targeted FRET sensor. (B) Table showing the average relative position of an N-terminally tagged mCherry-Hec1 (+/− SEM) to the indicated FRET probe tested as illustrated in (A). At least 5 different cells were imaged for each probe with at least 10 kinetochore pairs per cell. These positions agree closely with those determined by (Wan et al., 2009). (C) Average FRET emission ratio (+/− SEM) for an artificial Aurora B substrate FRET probe (Liu et al., 2009) targeted to kinetochores using either the Ndc80/Hec1 N-terminus or C-terminus in the indicated conditions (n > 20 cells per condition). (D) Graph as in (C) for FRET probe targeted with hMis12 or the Ndc80/HEC1 N or C-terminus for time points following the wash out of 30 ng/ml nocodazole. For each time point, >20 kinetochores were measured in >20 cells.

Figure 6. Differential phosphorylation of the KMN network under distinct kinetochore attachment states modulates outer kinetochore microtubule binding activity

Cartoon depiction of the KMN network at kinetochores under attached (with tension between sister chromatids), intermediate (with partial tension), and unattached (with no tension) states. A gradient of Aurora B phosphorylation emanates from the inner centromere. When tension is present, the kinetochore is stretched relative to its resting state such that the Ndc80 complex is positioned distally to the inner centromere. When no tension is present, the Ndc80 complex is more proximal to the inner centromere, resulting in higher levels of phosphorylation. Values for the position of the indicated subunits within the kinetochore are from (Wan et al., 2009).