Coregulation of NDC80 Complex Subunits Determines the Fidelity of the Spindle-Assembly Checkpoint and Mitosis

This article is featured in Selected Articles from This Issue, p. 421

Kinetochores facilitate the tethering of spindle microtubules to pairs of sister chromatids during mitosis (1). Dysregulation of kinetochore factors is increasingly being recognized as a major source of errors in chromosome segregation and genome instability (2). Recent advancements have shed light on the central role of the KMN network in establishing and maintaining microtubule attachment to kinetochores (3, 4). The ten-subunit KMN network consists of three complexes: a two-subunit KNL1 complex (KNL1C; composed of KNL1 and ZWINT), a four-subunit MIS12 complex (MIS12C; composed of MIS12, PMF1, NSL1, and DSN1), and a four-subunit NDC80 complex (NDC80C; composed of SPC24, SPC25, NDC80, and NUF2). The NDC80C within the KMN network consists of two dimers, NDC80–NUF2 and SPC24–SPC25, assembled end-to-end (3, 5). Although the N-terminal calponin homology (CH) domains of NDC80 and NUF2 mediate the interaction with plus-ends of spindle microtubules, the C-termini of SPC24 and SPC25 tether NDC80C to the kinetochore (3, 6–11).

MIS12C connects NDC80C and KNL1C to kinetochores through an interaction with CENP-C (9, 12–19). When kinetochores are not attached by microtubules, NDC80C is phosphorylated by centromeric Aurora B (AURKB) to facilitate the recruitment of MPS1 (20, 21). The direct competition between MPS1 and microtubules for NDC80C binding provides a mechanism for the cell to detect unattached kinetochores. MPS1 then phosphorylates multiple MELT motifs on KNL1, providing docking sites for the BUB1–BUB3 complex (22–29). Phosphorylation of BUB1–BUB3 by MPS1 further recruits the MAD1–MAD2 complex and other MCC components including BUBR1–BUB3 and CDC20 to initiate the activation of the spindle-assembly checkpoint (SAC; refs. 30–36).

Loss-of-function studies indicated that loss of NDC80C prevents kinetochore–microtubule (KT–MT) attachment and chromosome alignment. RNAi of NDC80 and SPC25 in human cell lines results in a failure of chromosomes to align at spindle equators (37, 38). Likewise, downregulation of NUF2 through RNAi reduces the ability of kinetochores to form stable attachments to spindle microtubules, resulting in the activation of the SAC and prometaphase arrest (39). Similar effects of loss-of-function in the components of NDC80C were observed in yeast (40, 41), Xenopus (42), chicken DT40 cells (43), and mouse oocytes (44), indicating the highly conserved nature of NDC80C's functions. In agreement with the dual role of NDC80C in KT–MT attachment and SAC activation, RNAi and antibody injection experiments revealed that depleting NDC80 or NUF2 causes the reduction of outer kinetochore proteins including dynein/dynactin and SAC components MPS1, MAD1, MAD2, ZW10, and ROD (20, 21, 39, 42, 45–48).

Precisely how the expression of NDC80C subunits is controlled remains to be characterized. Several studies suggest that NDC80C is regulated during the cell cycle in a proteasome-dependent manner. In budding yeast, Ndc80p is targeted for degradation during meiotic prophase by the ubiquitin ligase APC^Ama1 upon Aurora B (AURKB)-dependent phosphorylation (49–51). Similarly, proteasome-dependent degradation of NDC80 is hypothesized to occur upon mitotic exit in mammalian cells (52).

Here we investigated the effects of the loss of SPC24 in NDC80C regulation, KT–MT interaction, and SAC activation. Previous knockdown studies using siRNAs showed that SPC24 depletion represses cell growth and promotes apoptosis (53, 54). Shortcomings of tactics involving RNAi include incomplete knockdown, lack of specificity, and relatively slow responses also prevent unequivocal conclusions to be made. Using a more powerful tool of combining CRISPR-Cas9–mediated knockout (KO) with conditional rescue, we showed that acute depletion of SPC24 induced impairments in KT–MT attachment followed by antagonistic defects in SAC maintenance. Moreover, we found that depletion of SPC24 led to a decrease in the expression of other subunits of the NDC80C, suggesting strong posttranslational coregulation of the complex. Expression of the SPC24–SPC25 subcomplex alone was unable to restore the activation of the SAC.

pCMV(CAT)T7-SB100 expressing Sleeping Beauty transposase was a gift from Zsuzsanna Izsvak (Addgene #34879). CRISPR-Cas9 plasmids were generated by annealing the indicated pairs of oligonucleotides followed by ligation into BbsI-cut pX330 (a gift from Feng Zhang; obtained from Addgene; Addgene#42230): SPC24 (5′-CACCGCGACATAGAGGAGGTGAGCC-3′ and 5′-AAACGGCTCACCTCCTCTATGTCGC-3′); SPC25 (5′-CACCGGACACCTCCTGTCAGATGG-3′ and 5′-AAACCCATCTGACAGGAGGTGTCC-3′); NDC80 (5′-CACCGTGTCAGGAAGTTCGTATGAG-3′ and 5′-AAACCTCATACGAACTTCCTGACAC-3′); NUF2 (5′-CACCGAAGTCATTCACCCGGCAGAT-3′ and 5′-AAACATCTGCCGGGTGAATGACTTC-3′).

The cDNA of SPC24 was obtained from amplifying HeLa cDNA (prepared with reverse transcription using random hexamers) with oligonucleotides 5′-AATTTGCGCGGGTTGGAGCCTG-3′ and 5′-TGAAATGGATCTGACCACGG-3′. The PCR product (using 5′-GGATGTATAAGGCCTCCGTCATGGCCG-3′ and 5′-CGGGTACCGAGCTCGAATTCTACCACTCGGTGT-3′) was digested with NcoI and EcoRI and ligated into NcoI-EcoRI-cut pUHD-SB-mAID/Hyg (55) using Seamless Ligation Cloning Extract (SLiCE) cloning method (56) to generate mAID-SPC24 in pUHD-SB-mAID/Hyg. CRISPR-resistant silent mutations were introduced into SPC24 with a double PCR method using primers: 5′-AAACATCTGCCGGGTGAATGACTTC-3′ and 5′-AGCCCTTGTGACACCTCCTCT-3′; 5′-AGCCCTTGTGACACCTCCTCT-3′ and 5′-CATGTCTATCGATCTTATCATGTCTG-3′. The PCR product (template: CRISPR-resistant mAID-SPC24 in pUHD-SB-mAID/Hyg; primers: 5′-GGATGTATAAGGCCTCCGTCATGGCCG-3′ and 5′-CGGGTACCGAGCTCGAATTCTACCACTCGGTGT-3′) was cloned into NcoI-EcoRI-cut pUHD-SB-mAID-2AU/Hyg (57) using SLiCE cloning to obtain mAID-SPC24 in pUHD-SB-mAID-2AU/Hyg.

Inducible turn-on SPC24 constructs were generated by inserting FLAG-SPC24 (full-length, NΔ69, or CΔ132) into Drosophila/Bombyx ecdysone receptor (DBEcR)–destabilization domain (DD) system (58). The inserts were generated by a double PCR procedure. The first PCR products (for full-length and CΔ132) containing the FLAG-tag were generated using FLAG-3C-cyclin B1(NΔ88)-DD in pDBEcR-pIND(SP1)-DD-W/Bla (58) as a template and the forward–reverse primer pairs: 5′-TTCTCTAGGCACCGGT-3′ and 5′-GCGGAAGGCGGCCACCATGGGCCCCT-3′. The PCR products containing the SPC24 fragment were generated using mAID-SPC24 in pUHD-SB-mAID/Hyg as a template and the forward–reverse primer pairs: 5′-AGGGGCCCATGGTGGCCGCCTTCCGC-3′ and 5′-TCCACCTGCACTCCGAACCACTCGGTGTCCACCA-3′. The NΔ69 insert was generated similarly except the reverse primer 5′-CACCTGCTCCTTCACCATGGGCCCC-3′ and forward primer 5′-GGGGCCCATGGTGAAGGAGCAGGTG-3′ were used in amplifying the FLAG-tag and SPC24, respectively. The second PCR was performed using forward primer (5′-AGAAGAACTCACACACAGCTAGCCACAATG-3′ and reverse primers 5′-TCCACCTGCACTCCGAACCACTCGGTGTCCACCA-3′ (for full-length and NΔ69) or 5′-TCCACCTGCACTCCGAAGACTGTCGTGTCCTCG-3′ (for CΔ132) to create FLAG-SPC24 with overhangs. The FLAG-SPC24 fragments were cloned into NheI-EcoRI-cut pDBEcR-pIND(SP1)-DD/Bla (58) using SLiCE cloning method.

Inducible turn-on NDC80 construct were generated by inserting FLAG-NDC80 (full-length) into DBEcR–DD system (58). The inserts were generated by a double PCR procedure. The first PCR containing the FLAG-tag was generated using FLAG-3C-cyclin B1(NΔ88)-DD in pDBEcR-pIND(SP1)-DD-W/Bla (58) as a template and the forward–reverse primer pairs: 5′-TTCTCTAGGCACCGGT-3′ and 5′-GAACTGCGCTTCATCACCATGGGCCCC-3′. The PCR products containing the NDC80 fragment were generated using the cDNA of NDC80 as a template and the forward and reverse primer pairs: 5′-GGGGCCCATGGTGATGAAGCGCAGTTC-3′ and 5′-TCCACCTGCACTCCGAATTCTTCAGAAGACT-3′. The cDNA of NDC80 was obtained from amplifying RPE1 cDNA (cDNA (prepared with reverse transcription using random hexamers) with oligonucleotides 5′-AAATTCGAACGGCTTTGG-3′ and 5′-TACACTTTACTGAGACAATT-3′. The second PCR was performed using forward primer (5′-AGAAGAACTCACACACAGCTAGCCACAATG-3′ and reverse primer 5′-TCCACCTGCACTCCGAATTCTTCAGAAGACT-3′ to create FLAG-NDC80 with overhangs. The FLAG-NDC80 fragment was cloned into NheI-EcoRI-cut pDBEcR-pIND(SP1)-DD/Bla (58) using the SLiCE cloning method.

Inducible turn-on NDC80 construct with C-terminal deletion (CΔ197) was generated with site-directed mutagenesis using mutagenic forward–reverse primer pairs: 5′-TTCGGAGTGCAGGTGGAAA-3′ and 5′-AGCCTCGGGATTAAACTTAATTTCAA-3′.

Stealth siRNA targeting CDC27 (CCACAUUGGAGUAGUUCAACAUGCA) was obtained from Thermo Fisher Scientific, and siRNA targeting SPC24 (GAGCCUUCUCAAUGCGAAGTT) was obtained from GenePharma. Transfection of siRNA (10 nmol/L) was performed using Lipofectamine RNAiMAX (Thermo Fisher Scientific), following the manufacturer's instructions.

The following cell lines were obtained from the indicated sources: A375, A549, H1299, HT29, MCF10A, Ramos, RPE1, and THLE-3 (American Type Culture Collection), IMR-90 (Coriell Cell Repositories), U2OS (Clontech), MCF7 (a gift from Yong Xie, Hong Kong University of Science and Technology), Hep3B (a gift from Nathalie Wong, Chinese University of Hong Kong, and HCT116 (a gift from Bert Vogelstein, The Johns Hopkins University).

The HeLa cell line used to create ^mAIDSPC24^KO was a clone expressing tTA tetracycline transactivator, the F-box protein AFB2, and histone H2B-Clover (58). ^mAIDSPC24^KO cells were established by transfecting HeLa cells with SPC24 CRISPR-Cas9 in pX330, mAID-SPC24 in pUHD-SB-mAID-2AU/Hyg, pCMV(CAT)T7-SB100, and a plasmid expressing blasticidin-resistant gene. Transfected cells were enriched by culturing in a medium containing blasticidin for 48 hours, before being selected in a medium containing hygromycin B for two weeks. Single-cell–derived ^mAIDSPC24^KO colonies were obtained through limiting dilution in 96-well plates.

The ^mAIDSPC24^KO cell line used for immunostaining was generated by transfecting a HeLa cell line expressing tTa tetracycline transactivator (59) with SPC24 CRISPR-Cas9 in pX330, mAID-SPC24 in pUHD-SB-mAID/Hyg, pSBbi-TIR1/Pur (55), and pCMV(CAT)T7-SB100.

To generate SPC24 and NDC80 turn-on cell lines, ^mAIDSPC24^KO cells were transfected with pCMV(CAT)T7-SB100 and FLAG-SPC24 (full length, NΔ69, or CΔ132) or FLAG-NDC80 in pDBEcR-pIND(SP1)-DD/Bla, respectively. Transfected cells were selected with a medium containing blasticidin for two weeks to obtain a mixed population.

HeLa cells were propagated in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) calf serum and 50 U/mL of penicillin–streptomycin (Thermo Fisher Scientific). Other cell lines were propagated according to the suppliers’ instructions. Unless specifically stated, cells were treated with the following reagents at the indicated final concentrations: Barasertib (Selleck Chemicals; 25 nmol/L), BI 2536 (Selleck Chemicals; 2.5 nmol/L), blasticidin (Thermo Fisher Scientific; 3.75 μg/mL for transient selection; 2.5 μg/mL for stable selection), cycloheximide (Sigma-Aldrich; 10 μg/mL), doxycycline (Dox; Sigma-Aldrich; 2 μg/mL), hygromycin B (Thermo Fisher Scientific; 0.25 mg/mL), indole-3-acetic acid (IAA; Sigma-Aldrich; 50 μg/mL), MG132 (Sigma-Aldrich; 10 μmol/L), NOC (Sigma-Aldrich; 100 ng/mL), Ponasterone A (Santa Cruz Biotechnology; 5 μmol/L), PTX (Sigma-Aldrich; 125 ng/mL), puromycin (Sigma-Aldrich; 0.75 μg/mL for transient selection; 0.3 μg/mL for stable selection in HeLa, 3 μg/mL for transient selection in RPE-1 cells), RO3306 (Santa Cruz Biotechnology; 10 μmol/L), SB743921 (Selleck Chemicals; 10 nmol/L), Shield-1 (0.5 μmol/L; AOBIOUS), thymidine (Santa Cruz Biotechnology; 2 mmol/L), VX-689 (MK-5108; Selleck Chemicals; 250 nmol/L), and Z-VAD-FMK caspase inhibitor (Enzo Life Sciences; 10 μmol/L).

Transfection was carried out using a calcium phosphate precipitation method (60). Synchronization using double thymidine and NOC shake-off was conducted following previously described protocols (61). Colony formation assays were performed by seeding ^mAIDSPC24^KO cells at a density of either 400 or 800 cells per 60-mm plate, followed by treatment with buffer or DI. After 14 days, colonies were fixed with methanol:acetic acid (2:1) and stained with 2% (w/v) crystal violet. Cell-free extracts were prepared as previously described (58).

Cells were seeded onto 24-well cell culture plates and placed into an automated microscopy system equipped with a temperature, humidity, and CO₂ control chamber (Zeiss Celldiscoverer 7). Images were captured every 10 minutes for up to 24 hours, including channels for bright field and histone H2B-Clover. Data acquisition was carried out using Zeiss ZEN 2.3 (blue edition), and subsequent analysis was performed using ImageJ (NIH). After mitosis, one of the daughter cells was randomly selected and continued to be tracked.

Flow cytometry analysis after propidium iodide staining was performed as previously described (62). In brief, cells were trypsinized, washed with PBS, and fixed with ice-cold 80% ethanol. The cells were then stained with a solution containing 40 μg/mL of propidium iodide and 40 μg/mL of RNaseA at 37°C for 30 minutes. The DNA content of 10,000 cells was analyzed using FACSAria III flow cytometer (BD Biosciences).

Total RNA extraction, reverse transcription PCR, and real-time PCR were performed as previously described (63). Primers against SPC25 were: 5′-AGTACGGACACCTCCTGTCAG-3′ and 5′-TCTCAACCATTCGTTCTTCTTCC-3′. The expression of SPC25 mRNA was normalized to that of actin. Fold change of the sample normalized to control was calculated by the 2^–ΔΔCt method.

The following antibodies were obtained from the indicated sources: β-actin (A5316; Sigma-Aldrich), APC4 (ab72149; Abcam), APC11 (14090; Cell Signaling Technology), phospho-AURKA^Thr288/AURKB^Thr232/AURKC^Thr198 (2914; Cell Signaling Technology), CDC20 (sc-5296; Santa Cruz Biotechnology), CDC27 (610455; BD Biosciences), CREST (90C-CS1058; Fitzgerald Industries), cyclin B1 (sc-245; Santa Cruz Biotechnology), cyclin E (HE12, sc-247; Santa Cruz Biotechnology), phosphorylated histone H3^Ser10 (sc-8656R; Santa Cruz Biotechnology), FLAG (F3165; Sigma-Aldrich), KNL1 (ab70537; Abcam); MAD2 (17D10, sc-47747; Santa Cruz Biotechnology), MAD2 (raised against bacterially expressed GST-MAD2; ref. 64), NDC80 (sc-81283; Santa Cruz Biotechnology), NUF2 (sc-271251; Santa Cruz Biotechnology), cleaved PARP1 (552597; BD Biosciences), pericentrin (ab220784; Abcam), PTTG1 (DCS-280, sc-56207; Santa Cruz Biotechnology), SPC24 (A16601; ABClonal), SPC25 (HPA047144; Atlas Antibodies), acetylated α-tubulin (sc-23950; Santa Cruz Biotechnology), and α-tubulin (Alexa-Fluor-488–conjugated, 5063; Cell Signaling Technology). Immunoblotting was performed as previously described (63). The positions of molecular size standards (in kDa) are indicated in the Figures. Band intensity of NDC80C subunits was quantified with Image Lab software (version 5.2.1 build 11, Bio-Rad Laboratories) using serially diluted samples of the untreated ^mAIDSPC24^KO cell line as standard curves. Relative band intensity was calculated after normalization with actin signals. Immunoprecipitation of FLAG-tagged proteins was performed using anti-DYKDDDDK Tag (L5) Affinity Gel (651503; BioLegend). Preparation of polyclonal antibodies against MAD2 and immunoprecipitation procedures were as previously described (64) except that protein A/G PLUS-Agarose (sc-2003; Santa Cruz Biotechnology) was used.

Cells for immunofluorescence microscopy were prepared as previously described (65). Cells were fixed with ice-cold methanol at −20°C for 10 minutes before permeabilization with 0.4% Triton X-100 in PBS and blocking with 2% BSA in PBS for 30 minutes. For kinetochore staining, cells were preextracted with PHEM buffer (100 mmol/L PIPES; 20 mmol/L HEPES pH 6.9, 5 mmol/L EGTA, 2 mmol/L MgCl₂, 0.2% Triton X-100) for 45 seconds before fixed with 3.7% paraformaldehyde for 10 minutes at 25°C. Centrosomes, kinetochores, and centromeres were labeled using antibodies against pericentrin, KNL1, and CREST, respectively (added sequentially each for 1 hour at 25°C). Secondary antibodies Alexa-Fluor-568 goat anti-rabbit IgG and Alexa-Fluor-647 goat anti-human IgG (Thermo Fisher Scientific) were added sequentially, each for 1 hour at 25°C. Mitotic spindles were labeled using Alexa-Fluor-488–conjugated α-tubulin overnight at 4°C. Alternatively, stable microtubules were labeled using antibodies against acetylated α-tubulin, and then by Alexa-Fluor-488 goat anti-mouse IgG (Thermo Fisher Scientific). The cells were washed three times with 0.1% Triton X-100 in PBS (for 5 minutes each time) between each immunologic labeling. Nuclei were counterstained using Hoechst 33258 (200 ng/mL) for 10 minutes. Cold-stable k-fibers were stabilized by adding 500 μL of 100 mmol/L HEPES pH 7.2 dropwise to the cells and incubating on ice for 10 minutes before fixation with 3.7% paraformaldehyde in PHEM buffer for 10 minutes at 25°C. For MAD2 and kinetochore staining, cells were preextracted with PHEM buffer and fixed with 3.7% paraformaldehyde for 10 minutes at 25°C.MAD2 and kinetochores were labeled using antibodies against MAD2 and CREST respectively, added sequentially each for 2 hours at 25°C. Secondary antibodies Alexa-Fluor-647 goat anti-rabbit IgG and Alexa-Fluor-488 goat anti-human IgG (Thermo Fisher Scientific) were added sequentially, each for 1 hour at 25°C. FLAG was labeled using an antibody against FLAG for 2 hours at 25°C. Secondary antibody Alexa-Fluor-633 goat anti-mouse IgG was then added for 1 hour at 25°C. CREST was then labeled with primary and secondary antibodies as described above.

Data acquisition was carried out with a Celldiscoverer 7 fluorescence microscope or LSM980 confocal microscope equipped with AiryScan 2 (Zeiss) using Zen 2.3 (blue edition) software. Z-stack images were captured across a sample thickness of 12 μm (step size: 0.4 μm). Representative images were generated using maximal projection of raw images. Images were analyzed with ImageJ (NIH). Three-dimensional and two-dimensional distances between centrosomes were determined using Z-stack images and corresponding maximal projection, respectively. Interkinetochore distance was defined as the three-dimensional distance between paired KNL1 foci. The abundance of stable microtubules was quantified by the average pixel intensity of acetylated α-tubulin within the spindle apparatus. Chromosome distribution was measured by the area of Hoechst 33258 fluorescence signal above threshold. Spindle angles of mitotic cells were calculated as follows: |${{\cos }^{ - 1}}( {\frac{{2{\rm{D}} - {\rm{intercentrosomal\ distance}}}}{{3{\rm{D}} - {\rm{intercentrosomal\ distance}}}}} )$|⁠. MAD2 and CREST colocalization was assessed by quantifying the signals of CREST and MAD2 foci (within CREST foci) by measuring their integrated pixel intensities using CellProfiler (Version 4.2.1; ref. 66).

Box-and-whisker plots were generated using RStudio (version 1.2.5019) and Prism (version 9.5.1(528); GraphPad Software, LLC). The center lines represent the medians, the box limits indicate the interquartile range, and the whiskers extend to the most extreme data points that were no more than 1.5 times the interquartile range from the 25th and 75th percentiles. Statistical significance was determined using the Mann–Whitney test. Superplots were generated using Prism (67). The mean of three experiments was used to calculate the average (horizontal bar) and standard error of the mean (error bar). Statistical significance was determined using individual cells instead of the mean of each experiment.

All primary data are available upon request.

SPC24 is an integral component of NDC80C (Fig. 1A). Although gene disruption represents the most direct approach in gene silencing, it is generally irreversible and cannot be used for essential genes such as components of NDC80C. To achieve more precise and tighter silencing of SPC24, we generated a conditional cell line based on a dual transcription–degron system we designed previously (refs. 55, 63; Fig. 1B). Concurrent with the disruption of SPC24 with CRISPR-Cas9, a mini auxin-induced degron (mAID)-tagged SPC24 under the control of a Tet-Off promoter was delivered to the genome using Sleeping Beauty transposase. While the transcription of the ^mAIDSPC24 could be turned off using doxycycline (Dox), preexisting ^mAIDSPC24 could be targeted for proteolysis with indole-3-acetic acid (IAA). Cells lacking endogenous SPC24 and expressing ^mAIDSPC24 (designated as ^mAIDSPC24^KO herein) were able to degrade ^mAIDSPC24 in response to Dox and IAA (DI), in effect producing an SPC24-deficient environment. Single-colony–derived cell lines expressing different levels of ^mAIDSPC24 were isolated (Fig. 1C). Flow cytometry analysis revealed that SPC24-deficient cells were arrested with G₂–M DNA contents (Fig. 1D).

Using a ^mAIDSPC24^KO clone that expressed ^mAIDSPC24 at a similar level as endogenous SPC24, we found that ^mAIDSPC24 could be degraded rapidly to less than 1% after exposure to DI for 6 hours (Fig. 1E; also see later in Fig. 4B). Higher concentrations of IAA further accelerated the loss of ^mAIDSPC24 to beyond detection limit of the antibodies at 6 hours (Supplementary Fig. S1A and S1B). The destruction of ^mAIDSPC24 was accompanied by the accumulation of histone H3^Ser10 phosphorylation and cleaved PARP1, indicative of mitotic blockage and apoptosis, respectively (Fig. 1E). Consistent with these results, flow cytometry analysis revealed a progressive increase in G₂–M cells and sub-G₁ apoptotic cells. Finally, clonogenic survival analysis indicated that long-term survival was abolished after the loss of SPC24 (Fig. 1F).

These results demonstrate that our system of combining CRISPR-Cas9–mediated KO and degron-controlled SPC24 can trigger a robust SPC24 deficiency and mitotic arrest.

To determine if depletion of SPC24 yielded a protracted mitotic arrest or a more transient mitotic delay, we first synchronized ^mAIDSPC24^KO cells with a double thymidine block procedure before releasing them into either normal or DI-containing medium. Fig. 2A shows that cells containing ^mAIDSPC24 were able to enter and exit mitosis normally, as indicated by the transient nature of histone H3^Ser10 and Aurora kinase phosphorylation, cyclin B1 accumulation, as well as the change of DNA contents from 4N to 2N. By contrast, SPC24-deficient cells were blocked in a 4N state containing cyclin B1 and phosphorylated histone H3^Ser10 and Aurora kinases.

We next performed live-cell imaging analysis on ^mAIDSPC24^KO cells and validated at the single-cell level that SPC24-depleted cells underwent protracted mitosis (Fig. 2B). However, cells that entered mitosis relatively early after DI treatment were associated with a prolonged mitotic arrest followed by cell death (Fig. 2B; a representative example shown in Fig. 2C). By contrast, cells that entered mitosis after a relatively long period following DI treatment were characterized by a shorter mitotic arrest, premature sister chromatid separation, and multipolar cell division followed by cytokinesis failure (a representative example is shown in Fig. 2C). An explanation of these observations is that although partial depletion of ^mAIDSPC24 (for cells that entered mitosis early) was sufficient to induce mitotic blockage, more complete depletion resulted in defective SAC and premature mitotic exit. To corroborate this observation, we synchronized ^mAIDSPC24^KO cells using a double thymidine block and added DI at various times before or after release (Fig. 2D). Addition of DI at the time or at 3 hours after the release from thymidine resulted in prolonged mitotic arrest and subsequent cell death. By contrast, the addition of DI before the thymidine release resulted in shorter mitotic arrest with cytokinesis failure or mitotic slippage.

Consistent with the above, MAD2 recruitment to the kinetochores was increased after 8 hours of DI treatment, followed by a reduction upon further incubation (Fig. 2E). Similarly, the formation of MAD2–CDC20 complexes was induced upon initial silencing of SPC24, but decreased after prolonged incubation with DI (Fig. 2F). In the absence of SPC24, the SAC could not be maintained even in the presence of microtubule inhibitors paclitaxel (PTX; Fig. 2B) or nocodazole (NOC; Fig. 2F). The decrease in mitotic markers, including histone H3^Ser10 phosphorylation, cyclin B1, and securin (PTTG1), further verified the impairment of SAC after prolonged or complete removal of SPC24. Live-cell imaging also confirmed that PTX-induced mitotic arrest was shortened in the absence of SPC24 (Fig. 2B).

SPC24-deficient cells were unable to form proper spindle and metaphase plate (Fig. 3A). They also contained a reduced number of stable acetylated tubulin-containing microtubules (Supplementary Fig. S2A) stable kinetochore-binding microtubules (k-fibers; Supplementary Fig. S2B), and displayed defective polar ejection in monopolar spindles induced by KIF11 inhibition (Supplementary Fig. S2C). Accordingly, the interkinetochore distance was decreased, suggesting a reduction of interkinetochore tension (Fig. 3B). The defective KT–MT stability in SPC24-deficient cells was also reflected by the significant increase in intercentrosomal distance and spindle angle (Fig. 3C).

Taken together, these results demonstrate that SAC is activated transiently by SPC24 silencing, but cannot be sustained upon complete depletion of SPC24.

We observed that concurrent with the depletion of ^mAIDSPC24, the expression of other subunits of the NDC80C (SPC25, NDC80, and NUF2) was also reduced (Fig. 4A; see also Figs. 1E and 2A). Clones of ^mAIDSPC24^KO with different ^mAIDSPC24 expressions also displayed codepletion of NDC80C subunits, excluding the possibility of clonal effects (Supplementary Fig. S3A). Quantification of band intensities indicated that the loss of other NDC80C subunits occurred at a similar kinetics as the destruction of ^mAIDSPC24 (Fig. 4B). Moreover, turning off ^mAIDSPC24 to different extents using serial dilutions of Dox (Fig. 4C) or DI (Fig. 4D) induced proportional reduction in other NDC80C subunits. By subjecting synchronized ^mAIDSPC24^KO cells to varying durations of DI treatment, we found that although the destruction of ^mAIDSPC24 was sufficient to induce a mitotic arrest, the reduction of the entire NDC80C after prolonged DI treatment correlated with a higher proportion of cells undergoing mitotic slippage and cytokinesis failure (Fig. 4E).

Collectively, these data indicate that the expression of the entire NDC80C is reduced after SPC24 is depleted.

One trivial explanation for the codepletion of NDC80C subunits is that the closely associated subunits were inadvertently cotargeted by the AID system. To exclude this possibility, we demonstrated that the expression of other NDC80C subunits was also reduced after SPC24 was targeted with CRISPR-Cas9 without the use of the AID system (Fig. 5A). Moreover, the entire NDC80C was downregulated when ^mAIDSPC24 was turned off either through promoter control (Fig. 5B) or with siRNA (Supplementary Fig. S4A). These findings were consistent across several normal and cancer cell lines with different levels of NDC80C, including RPE1, HCT116, and H1299, indicating that the effects were not only limited to HeLa cells (Supplementary Fig. S4A). Together, these results indicated that the codepletion of NDC80C subunits was not owing to cotargeting by the AID system.

We next transfected CRISPR-Cas9 targeting SPC25 into ^mAIDSPC24^KO cells and found that depletion of SPC25 diminished the expression of ^mAIDSPC24, NDC80, and NUF2 (Fig. 5C). Furthermore, silencing any one of the four subunits of NDC80C resulted in downregulation of other subunits of the complex, indicating that the coregulation of NDC80C subunits is not specific to SPC24 (Fig. 5D).

Given the observed coregulation of NDC80C subunits, we predicted that cell lines containing low expression of SPC24 may also exhibit reduced expression of other subunits. In agreement with this, an analysis of normal and cancer cell lines from different tissue origins revealed that cell lines with relatively low SPC24 expression also tended to possess a decreased expression of SPC25, NDC80, and NUF2 (Supplementary Fig. S4B).

We investigated the potential mechanism underlying the downregulation of other NDC80C subunits upon depletion of SPC24 first by examining the transcription of SPC25. Figure 5E shows that SPC25 mRNA was not affected by DI treatment in ^mAIDSPC24^KO cells, indicating that the reduction of SPC25 in the absence of SPC24 is likely due to posttranslational regulation. We next used cycloheximide to block protein translation and found that the NDC80C subunits were relatively stable proteins. However, their half-lives were significantly shortened upon treatment with DI (Supplementary Fig. S5A), indicating an increased turnover of NDC80C proteins when SPC24 was depleted.

Very little is known about the turnover mechanism of NDC80C. The major ubiquitin ligase during mitosis, APC/C, is implicated in the meiotic degradation of Ndc80p in budding yeast (49–51). To determine if the turnover of NDC80C in human cell lines is regulated by APC/C, we downregulated one of APC/C's components, CDC27, with siRNA and examined the degradation of NDC80C. Supplementary Fig. S5B shows that depletion of CDC27 prevented the degradation of the APC/C substrate cyclin B1 when cells exited mitosis. By contrast, NDC80C subunits were degraded after SPC24 degradation both in the presence or absence of CDC27, and both during mitotic block and after mitotic exit. These results suggested that the degradation of NDC80C probably does not depend on APC/C in human cells.

As silencing of SPC24 resulted in mitotic arrest, it could be argued that the alteration of NDC80C subunits was caused by cell-cycle effects. This is unlikely an explanation as the expression of different NDC80C subunits remained relatively constant in cells synchronously released from double thymidine block into the cell cycle (Supplementary Fig. S6A). Similarly, no change in the expression of NDC80C was observed in cells released from mitosis into G₁ (Supplementary Fig. S6B). Finally, using the CDK1 inhibitor RO3306 to prevent SPC24-depleted cells from entering mitosis demonstrated that mitotic arrest is not required for the codepletion of the NDC80C subunits (Supplementary Fig. S6C).

As the loss of SPC24 also promoted mitotic cell death, it is conceivable that apoptosis is involved in the co-downregulation of NDC80C. To exclude this possibility, apoptosis was suppressed using the pan-caspase inhibitor Z-VAD-FMK. Supplementary Fig. S6D shows that Z-VAD-FMK reduced PARP1 cleavage induced by SPC24 depletion but not the decrease of NDC80C, indicating that mitotic cell death is not required for the coregulation of the NCD80C subunits.

Overall, these data indicate that subunits of the NDC80C are co-downregulated at the posttranslational level independently on cell-cycle arrest and cell death.

Given that the downregulation of any one subunit of NDC80C automatically resulted in a reduction of other subunits, we sought to uncouple the regulation of the two NDC80C subcomplexes by using SPC24 mutants that can bind SPC25 but not NDC80–NUF2. The SPC24 was transcriptionally controlled with an inducible promoter based on a hybrid DBEcR, which can transactivate a modified ecdysone promoter in the presence of the ecdysone agonist Ponasterone A. The SPC24 is also tagged with a mutated FK506-binding protein-12 (FKBP12)-derived DD to target it to proteolysis, which could be stabilized with the small-molecule Shield-1. The addition of Ponasterone A and Shield-1 together (PS herein) increases the transcription and protein stability of SPC24^DD, respectively (58). Incorporating this inducible system into ^mAIDSPC24^KO cells enabled a rapid switch between wild-type and mutant SPC24. After silencing ^mAIDSPC24, the stability of all NDC80C subunits could be restored by introducing SPC24^DD in a dose-dependent manner (Fig. 6A). In marked contrast, overexpression of an N-terminally truncated SPC24^DD (NΔ69; ref. 68) stabilized SPC25, but not NDC80 and NUF2. The NDC80–NUF2 subcomplex could not be stabilized even when NΔ69 was turned on for up to 48 hours, indicating that it was not due to a lag in synthesis (Supplementary Fig. S7A).

We found that although full-length SPC24^DD could coimmunoprecipitate all NDC80C subunits, SPC24(NΔ69) only formed a complex with SPC25 (Fig. 6B). As a further control, we also introduced a C-terminally truncated mutant of SPC24 (CΔ132) into the SPC24-deficient environment. None of the NDC80C subunits was coimmunoprecipitated with CΔ132, confirming that the C-terminal region of SPC24 is essential for NDC80C formation (68). Consistent with this result, none of the NDC80C subunits was stabilized by CΔ132.

We also analyzed the effects of SPC24(NΔ69) on mitosis using live-cell imaging (Fig. 6C; Supplementary Fig. S7B). As described above, turning off ^mAIDSPC24 resulted in mitotic arrest followed by mitotic slippage. Turning on full-length SPC24^DD, but not NΔ69, restored relatively normal mitosis, suggesting that the SPC24–SPC25 subcomplex cannot prevent mitotic arrest in the absence of NDC80–NUF2. This was also reflected in the ability of full-length SPC24, but not NΔ69, to increase overall cell survival (Fig. 6D).

Cells expressing SPC24(NΔ69) underwent mitotic slippage or cytokinesis failure similarly as in SPC24-deficient cells (Supplementary Fig. S7B), suggesting that the SPC24–SPC25 subcomplex alone was insufficient to maintain SAC activation. In agreement with this, although APC/C targets accumulated in SPC24-containing cells treated with NOC, they were degraded in cells containing SPC24(NΔ69; Fig. 6E). Furthermore, kinetochore localization of MAD2 in SPC24-deficient cells was restored by full-length but not SPC24(NΔ69; Fig. 6F). Unlike full-length SPC24^DD, NΔ69 was not associated with kinetochores (Supplementary Fig. S7C).

Using a similar approach, we investigated whether the NDC80–NUF2 subcomplex could be stabilized in the absence of SPC24–SPC25 by expressing NDC80^DD in the ^mAIDSPC24^KO background. In the presence of SPC24, NDC80^DD was able to coimmunoprecipitate all other NDC80 subunits. In the absence of SPC24, however, NDC80^DD is specifically associated with and stabilized NUF2 only (Fig. 6G).

Collectively, these results indicate that although the downregulation of one subunit of the NDC80C destabilizes the entire complex, the SPC24–SPC25 and NDC80–NUF2 subcomplexes can be individually stabilized using ectopically expressed subunits.

Given that NDC80C downregulation produces potentially conflicting signals from defective KT–MT attachments and reduced SAC activation, we next investigated if the sensitivity to drugs that activate or disrupt the SAC can be potentiated by NDC80C downregulation. To achieve partial downregulation of ^mAIDSPC24, a relatively low concentration of Dox (without IAA) was used to induce a reduction of the four NDC80C subunits (Supplementary Fig. S8A). This led to a slight elevation of the G₂–M population (Fig. 7A) and both mitotic duration and cell death (Fig. 7B). The cells were then exposed to low concentrations of NOC or PTX, which inhibit microtubule polymerization or depolymerization, respectively. The sublethal concentrations of the two microtubule inhibitors caused a mitotic delay (Fig. 7A and B). Importantly, NOC and PTX-induced mitotic block and apoptosis in SPC24-compromised cells, as indicated by an increase in the G₂–M and sub-G₁ populations, respectively (Fig 7A). Live-cell imaging further confirmed the presence of mitotic defects at the single-cell level (Fig 7B). Notably, NOC and PTX-induced more rapid and extensive mitotic cell death in SPC24-downregulated cells compared with control cells (Fig. 7C; Supplementary Fig. S8B).

The mitotic kinases Aurora kinase A (AURKA) and polo-like kinase 1 (PLK1) are frequently overexpressed in human malignancies. Similar to microtubule inhibitors, pharmacologic inhibitors targeting these kinases can induce mitotic arrest and cell death (69, 70). We found that SPC24-depleted cells were more sensitive to G₂–M arrest and apoptosis induced by VX-680 (AURKAi) or BI 2536 (PLK1i; both used at sublethal concentrations that did not cause mitotic blockage on their own; Fig. 7D). Live-cell imaging further confirmed the prolonged duration of mitosis (Fig. 7E) and reduced cell survival (Fig. 7F) in SPC24-depleted cells following treatment with AURKAi and PLK1i.

These results suggest that the downregulation of NDC80C enhances mitotic arrest rather than promoting mitotic slippage in response to SAC-activating drugs. Considering that decreased NDC80C levels also reduce SAC activation, we investigated whether SPC24 depletion could promote mitotic slippage induced by inhibition of AURKB. AURKB phosphorylates multiple components of the KMN network. Phosphorylation of DSN1 enhances the binding of MIS12C to CENP-C (4, 71–73). When kinetochores lack attachment to microtubules, AURKB phosphorylates NDC80C to facilitate the recruitment of MPS1 for SAC activation (3, 7, 74, 75). Furthermore, AURKB-mediated phosphorylation of NDC80 reduces the binding of microtubules to NDC80C, allowing dynamic interaction between microtubules and kinetochores (3, 76, 77). Consistent with previous findings (78), treatment with a relatively low concentration of the AURKB inhibitor Barasertib (AURKBi) resulted in an increased frequency of mitotic slippage (Fig. 7G). Notably, SPC24 downregulation doubled the frequency of mitotic slippage induced by AURKBi (from 34% to 70%), indicating that the effects of AURKBi can be enhanced following NDC80C downregulation.

Collectively, these data indicate that corepression of NDC80C subunits enhances the cytotoxicity of mitotic inhibitors, both in terms of mitotic arrest and mitotic slippage.

A starting point of our study involved the development of a genetic tool as an alternative to traditional chemical methods that disrupt KT–MT attachment, which can have off-target effects on nonmitotic processes (79). In this study, we used CRISPR-Cas9 to KO the endogenous SPC24 gene, while simultaneously rescuing the cells with ^mAIDSPC24 under the control of an inducible protomer. This approach allowed rapid depletion of SPC24 within a few hours (Fig. 1E; Supplementary Fig. S1B) and enabled us to synchronize ^mAIDSPC24^KO cells using a double thymidine block followed by release into DI-containing medium to obtain mitotic cells (Fig. 2A). It is noteworthy that the concentrations of DI used in our experiments did not affect long-term cell survival in both cancer (HeLa) and normal (hTERT-immortalized RPE1) cell lines (58).

Multiple lines of evidence suggest that SPC24-deficient cells were delayed in the early stages of mitosis. These include the enrichment of cells with G₂–M DNA contents (Fig. 1D and E), accumulation of various mitotic markers (Fig. 2A), and prolonged mitosis (Fig. 2B). Immunostaining revealed abnormalities in spindle and chromosomal alignment (Fig. 3A), along with increased intercentrosomal distance and spindle angle (Fig. 3C), reduced stable microtubules (Supplementary Fig. S2), and decreased interkinetochore tension (Fig. 3B). The increase in multipolar division upon SPC24 depletion (Fig. 2C) may be attributed to the role of NDC80C in clustering extra centrosomes (80).

It should be noted that the mitotic arrest induced by SPC24 deficiency was not as stringent as that imposed by microtubule inhibitors. A significant proportion of SPC24-deficient cells were able to exit mitotic arrest, displaying characteristics of both mitotic slippage and premature sister chromatid separation (Fig. 2B and C). This is consistent with the established role of NDC80C in the recruitment of SAC components to the kinetochores (20, 21, 46–48). Consequently, the loss of SPC24 compromised the mitotic arrest induced by PTX and NOC (Fig. 2B, F). The compromised SAC was reflected in the reduction of MAD2 at the kinetochores (Fig. 2E) and in the soluble MAD2–CDC20 complexes (Fig. 2F). This was accompanied by the degradation of APC/C targets, including cyclin B1 and PTTG1 (Fig. 2F). Notably, defective SAC activation was especially pronounced in cells with prolonged SPC24 depletion (Fig. 2D), suggesting that whereas partial depletion of NDC80C was sufficient to hinder KT–MT attachment, more complete depletion was required to abolish the SAC. Because the depletion of SPC24 inevitably resulted in the downregulation of other subunits (Fig. 4A and B), it is not possible to unequivocally demonstrate that the loss of SPC24 alone is sufficient to trigger the mitotic arrest.

The precise mechanism underlying the coregulation among NDC80C subunits has yet to be fully elucidated. However, we have ruled out the possibility that it is a consequence of codegradation mediated by the AID system (Fig. 5A and B) or a decrease in transcription (Fig. 5E). Furthermore, the codownregulation of NDC80C subunits was not specific to SPC24, as depletion of any individual subunit resulted in a reduction in the other subunits of the complex (Fig. 5D). Whether the stability of other NDC80C-associated proteins is affected by NDC80C depletion requires further investigation. However, our data suggest that the expression of KNL1 at the kinetochores was unaffected by SPC24 depletion (Fig. 3B).

A plausible explanation for the instability of NDC80C is that its expression is cell-cycle regulated and is affected by the mitotic arrest induced by SPC24 depletion. In budding yeast, it has been shown that Ndc80p undergoes degradation during meiotic divisions through a mechanism involving AURKB and the ubiquitin ligase APC^Ama1 (49–51). Similarly, partial degradation of NDC80 during mitotic exit has been reported in HCT116 cells (52). Nonetheless, we found little evidence of NDC80C degradation during mitotic exit in NOC-synchronized HeLa cells (Supplementary Fig. S6B). Another possibility is that the stability of NDC80C may depend on its binding to kinetochores. We think this is unlikely an explanation because, although NDC80C is typically not localized to kinetochores during interphase (43), it is expressed at comparable levels between interphase and mitosis (Supplementary Fig. S6A and S6B). Furthermore, it is unlikely that the stability of NDC80C is affected by binding to microtubules, as the abolition of NDC80C–microtubule interaction by NOC did not impact NDC80C stability (Supplementary Fig. S6B).

The structure of NDC80C comprises two dimers, NDC80–NUF2 and SPC24–SPC25 (ref. 6; Fig. 7H). We speculate that the entire tetramer is normally required for maintaining structural integrity and stability. As coiled-coil structures are known to provide stabilization effects (81), a reduction in one subunit of the NDC80–NUF2 or SPC24–SPC25 dimer could potentially accelerate the turnover of the remaining subunit. However, the effects of the downregulation of one NDC80C dimer on the other dimer are somewhat unexpected. It is noteworthy that although downregulation of SPC24–SPC25 strongly destabilized NDC80–NUF2, the converse downregulation of NDC80–NUF2 had a weaker effect on SPC24–SPC25 (Fig. 5D). One possibility is that exposure of the free C-termini of the NDC80–NUF2 dimer is more destabilizing than exposure of the free N-termini of the SPC24–SPC25 complex. Notably, when SPC24(NΔ69) was overexpressed, it could form a stable complex with SPC25 independently of NDC80–NUF2 (Fig. 6A and B). Similarly, overexpressed NDC80 could form a stable complex with NUF2 independently of SPC24–SPC25 (Fig. 6G). However, the SPC24–SPC25 subcomplex alone was insufficient to maintain SAC activation in the absence of NDC80–NUF2 (Fig. 7H). This finding is consistent with the role of the N-terminal domains of NDC80 and NUF2 in recruiting MPS1 (20, 21).

Our results also highlight a potential caveat concerning results obtained from targeting individual subunits within protein complexes. The coregulation of NDC80C subunits suggests that a deficiency in one subunit, whether due to variations in gene copy-number or epigenetic regulation, may determine the expression of other subunits. This observation is consistent with the correlation in the expression between NDC80C subunits across different cell lines (Supplementary Fig. S4B). Interestingly, we observed that ectopic expression of SPC24 in HeLa cells does not lead to an increase in the expression of other NDC80C subunits, suggesting that the normal expression of the NDC80C is already maximized (our unpublished data).

An implication of the dual role of NDC80C in KT–MT attachment and SAC activation is its potential as a target or prognostic marker for antimitotic therapies. For example, SPC24 has been found to be overexpressed in various cancers, including liver, lung, breast, and thyroid cancer (53, 54, 82, 83). In our study, we observed that the downregulation of NDC80C enhanced mitotic arrest and cell death induced by sublethal concentrations of NOC or PTX (Fig. 7A-C). The synergistic effect was likely due to a reduction in the frequency of spindle–chromosome attachment caused by diminished KT–MT interaction (SPC24 depletion) and altered microtubule dynamics (NOC/PTX). The mechanisms underlying the synergism between SPC24 depletion and inhibitors of PLK1 and AURKA are likely to be multifaceted and distinct from those observed with microtubule inhibitors. PLK1 plays an essential role in stable KT–MT attachment (84–87), whereas AURKA can directly phosphorylate NDC80 and regulate metaphase KT–MT dynamics (88). Moreover, AURKA is involved in the activation of PLK1 itself (89). Hence, the disruption of KT–MT interaction by PLK1i and AURKAi may be further accentuated by the reduction in the frequency of KT–MT interaction caused by NDC80C downregulation (Fig. 7D–F).

Although NDC80C is involved in both KT–MT attachment and SAC activation, its downregulation enhances mitotic arrest rather than promoting mitotic slippage in response to SAC-activating drugs. On the other hand, we found that downregulation of SPC24 increased the mitotic slippage induced by AURKBi (Fig. 7G). This finding agrees with the observed synergism between partial depletion of NUF2 and an AURKB inhibitor in promoting mitotic slippage (90). AURKB phosphorylates NDC80C to facilitate the recruitment of MPS1 for SAC activation (3, 7, 74, 75). The combined downregulation of NDC80C and AURKBi may act synergistically in promoting mitotic slippage by reducing the recruitment of MPS1. Furthermore, the extended duration of mitosis in NDC80C-depleted cells may provide a larger time window for AURKBi to induce mitotic slippage. Overall, downregulation of NDC80C can either enhance mitotic arrest or promote mitotic slippage in response to treatments that activate or inactivate the SAC, respectively.

R.Y.C. Poon reports grants from Research Grants Council and Innovation and Technology Commission during the conduct of the study. No disclosures were reported by the other authors.

S. Kim: Conceptualization, investigation, methodology, writing–original draft, writing–review and editing. T.T.Y. Lau: Investigation, methodology, writing–original draft, writing–review and editing. M.K. Liao: Investigation, methodology. H.T. Ma: Conceptualization, methodology, writing–original draft, writing–review and editing. R.Y.C. Poon: Conceptualization, supervision, funding acquisition, writing–original draft, writing–review and editing.

R.Y.C. Poon was a recipient of the Croucher Foundation Senior Research Fellowship. We thank Wing Man Yuen for technical assistance. This work was supported in part by grants from the Research Grants Council (16102919, 16103222, and N_HKUST636/20) and the Innovation and Technology Commission (ITCPD/17-9) to R.Y.C. Poon.