Introduction
The spindle checkpoint prevents aneuploidy, a hallmark of most aggressive tumors, by delaying the metaphase–anaphase transition until the attachment of all chromosomes to spindle microtubules has been achieved1, 2, 3. The first checkpoint proteins (including Mad1, Mad2, Bub1, Bub3, BubR1 and Mps1) were identified through yeast genetic screens, and later found to have conserved functions in higher eukaryotes1, 2. Bub1, BubR1, Mad1 and Mps1 are required to recruit Mad2 to unattached kinetochores, placing Mad2 at the bottom of the kinetochore-localized checkpoint pathway1, 2, 3, 4. Although the molecular function of most upstream checkpoint components remains unclear, recent studies indicate that kinetochore-bound Mad2–Mad1 complexes serve as templates for the structural activation of Mad2 and generation of diffusible checkpoint complexes that inhibit the activity of anaphase-promoting complex/cyclosome (APC/C)3.
The molecular dissection and modelling of spindle-checkpoint signalling requires a complete inventory of all checkpoint proteins. The full complement of spindle-checkpoint proteins in higher eukaryotes seems to exceed that of yeast, as recent studies have identified additional regulators of the spindle checkpoint (such as Zw10, Rod, and p31comet) that lack clear yeast orthologues5, 6. We set out to identify additional human spindle-checkpoint proteins by screening short hairpin RNA (shRNA) libraries that target human kinases and phosphatases using a high-throughput visual screening platform (Fig. 1a, b; and see Supplementary Information, Table S1 for detailed library information). A total of 2533 hairpins targeting 780 genes were screened. Using this screening protocol, control transfected cells exhibited efficient mitotic arrest (only 5–15% non-mitotic cells) 24 h after taxol treatment (Fig. 1c). In contrast, cells transfected with shRNAs targeting BubR1 bypassed the checkpoint arrest, as evident from the dramatic increase in non-mitotic cells to over 80% (Fig. 1c and see Supplementary Information, Table S2).
Figure 1: shRNA screen identifies new candidate spindle-checkpoint kinases.
(a) Schematic representation of high-throughput screening protocol. Arrayed plasmid-based shRNA vectors were cotransfected into HeLa cells with a dsRed-expressing vector in a 96-well format. Cells were treated with taxol (100 nM) 48 h after transfection and fixed for visual inspection 24 h after taxol addition. (b) Pie chart of shRNA library composition. The number of hairpins targeting each gene class is indicated. (c) Representative images from high-throughput screen illustrating the phenotypic difference between checkpoint-arrested (control shRNA) and checkpoint-bypass cells (BubR1 shRNA). Mitotic arrest is characterized by round cell shape and condensed chromatin (upper panels), whereas checkpoint-bypassing cells have a flattened morphology and multilobed nuclei (lower panels). (d, e) HeLa cells were transfected with the indicated siRNAs and treated as described in a. The percentage of non-mitotic cells (d) and cells with multi-lobed or interphase nuclei (e) was determined by visual inspection. The values represent averages from three independent experiments and error bars indicate s.d. (f) Representative images of HeLa cells transfected with siRNAs targeting luciferase (negative control) or TAO1 siRNA 3 illustrating the phenotypic differences between mitotic arrest (condensed chromosomes) and checkpoint bypass (multi-lobed nuclei). Cells were treated as described in a. (g) Evaluation of protein knockdown for different siRNAs targeting TAO1. HeLa cells were treated with the indicated siRNAs (numbers refer to oligonucleotide number), cell extracts collected 72 h after transfection and then probed with the indicated antibodies. (h) TAO1 siRNA 3 is specific for TAO1 and does not affect expression of its paralogues TAO2 or TAO3. HeLa cells were treated with TAO1 siRNA 3. Twenty-four hours later, cells were transfected with the indicated expression constructs (TAO1–HA, TAO2–HA or TAO3–HA). After siRNA transfection (48 h), cell extracts were collected and probed with the indicated antibodies. The scale bars in c and f represent 20 
m.
The screening of our kinase library identified all known spindle-checkpoint kinases (for example, BubR1, Mps1, Bub1 and AuroraB), thereby validating the screening approach (see Supplementary Information, Table S2). Importantly, the screen also identified several novel candidate spindle-checkpoint components in the kinase library (see Supplementary Information, Table S2), including TAO1, MARK1, SNRK1 and Nek2A (which was recently found to associate with Mad1 and proposed to control spindle-checkpoint function7). To validate these newly identified candidate spindle-checkpoint regulators, small interfering RNAs (siRNAs) were designed that target different regions within these genes. For each candidate gene examined, additional siRNAs were identified that caused checkpoint bypass (Fig. 1d–f and see Supplementary Information, Fig. S1a). Moreover, the phenotypic penetrance of TAO1 siRNAs correlated well with their efficiency in protein-level reduction (Fig. 1g). MARK1, however, may have scored because of off-target activities as there was no clear correlation between knockdown efficiency and phenotype penetrance (see Supplementary Information, Fig. S1b, c).
Subsequent investigations were focused on the protein kinase TAO1 because it caused the most penetrant checkpoint defect (Fig. 1d–e). TAO1 is thought to regulate p38 MAPK signalling8, which was previously implicated in spindle-checkpoint control9. However, consistent with two recent studies10, 11, no spindle-checkpoint defect was observed in p38-depleted cells (Fig. 1d, e). Human TAO1 has two paralogues, TAO2 and TAO3 (ref. 12). TAO1 depletion was specific as it did not affect the expression of TAO2 or TAO3 (Fig. 1h), or the levels of other checkpoint proteins (see Supplementary Information, Fig. S1d). To further validate the specificity of TAO1 RNAi, functional-rescue experiments were performed using an siRNA-resistant TAO1 expression construct. The spindle-checkpoint defects observed in TAO1-depleted cells were significantly rescued by expression of an siRNA-resistant wild-type TAO1 kinase (TAO1-siRes; Fig. 2a, b). In contrast, an siRNA-resistant kinase-dead TAO1 mutant (TAO1D151A) failed to restore spindle-checkpoint function (Fig. 2a–c; TAO1D151A-siRes), indicating that the kinase activity of TAO1 is required for its spindle-checkpoint function.
Figure 2: TAO1 kinase activity is elevated in mitosis and required for its checkpoint function.
(a, b) Transfection of siRNA-resistant wild-type TAO1 (TAO1-siRes) but not the kinase-dead mutant TAO1D151A (TAO1D151A-siRes) rescues the spindle-checkpoint defect of TAO1 siRNA treated cells . HeLa cells were transfected with the indicated siRNAs (a). Twenty-four hours later, the rescuing plasmids were cotransfected with H2B–GFP. After siRNA transfection (48 h), 100 nM taxol was added for an additional 24 h before fixation. GFP-positive cells were analysed for mitotic arrest and checkpoint bypass (multi-lobed nuclei). Western blotting shows similar expression levels of wild-type and catalytically inactive Myc–TAO1 (b). (c) In vitro kinase assay of wild-type (Myc–TAO1) and kinase-dead mutant TAO1D151A (Myc–TAO1D151A). HeLa cells were transiently transfected with the indicated vectors for 48 h. Myc-tagged proteins were immunoprecipitated and assayed for in vitro kinase activity, using MBP as a substrate. (d) TAO1 kinase activity is elevated in mitosis. Endogenous TAO1 protein was immunoprecipitated from log phase (Cyc), thymidine-arrested (Thym) or nocodazole-arrested (Noc) HeLa cell lysates, and assayed for in vitro kinase activity using MBP as a substrate. A fraction of the immunoprecipitate was blotted for TAO1 protein levels, and lysates were blotted for the mitotic marker protein phospho-histone H3 (p-H3). (e) Quantification of the total and relative kinase activity (TAO1 kinase activity:TAO1 protein levels). (f) TAO1 protein levels decrease as cells exit from mitosis. HeLa cells were released from a nocodazole arrest and immunoblotted for the indicated proteins. The error bars in a represent s.d. (n >100 from three independent experiments).
Full size image (46 KB)Given the requirement for the kinase activity of TAO1 in spindle-checkpoint function, we next explored whether its kinase activity is cell-cycle regulated. TAO1 kinase activity was elevated more than fourfold in mitotically arrested compared with S-phase-arrested cells (Fig. 2d, e and see Supplementary Information, Fig. S2a, b), resulting from both increased protein levels and an approximately twofold increase in relative kinase activity in mitotic cells (Fig. 2d, e). Similarly to other important regulators of mitosis (such as cyclin B13), TAO1 protein levels decreased as cells progressed through the metaphase–anaphase transition and exited from mitosis (Fig. 2f).
Recent studies revealed that the spindle-checkpoint defects in response to Aurora B inactivation are an indirect consequence of the role of Aurora B in correcting erroneous microtubule–kinetochore attachments14. As TAO1 has previously been implicated in regulating microtubule dynamics15, we wanted to exclude the possibility that the checkpoint function of TAO1 could be explained by a similar indirect mechanism. To this end, mitotic progression of H2B–GFP-expressing cells was monitored by live-cell imaging shortly after nocodazole treatment, which depolymerizes microtubules and thereby activates the spindle checkpoint. Both control and Aurora B-depleted cells arrested in mitosis after nocodazole treatment, whereas TAO1-depleted cells rapidly exited mitosis (Fig. 3a–c). These findings support the hypothesis that the checkpoint function of TAO1 is direct and not a secondary effect of altered microtubule dynamics or defects in microtubule–kinetochore attachment.
Figure 3: TAO1-depleted cells display precocious anaphase onset and chromosome missegregation.
(a–c) TAO1-depleted cells rapidly bypass nocodazole-induced checkpoint arrest. Time-lapse analysis of H2B–GFP-expressing HeLa cells treated with the indicated siRNAs after nocodazole treatment. Successive frames from representative live-cell movies are shown. Control cells (a) remain rounded with condensed chromosomes, whereas TAO1-depleted cells (b) initiate anaphase movements and cytokinetic ingression (white arrow). (c) Quantification of the percentage of cells that remained arrested in mitosis for longer than 3 h in response to nocodazole treatment. (d, e) TAO1 is required for proper mitotic progression in the absence of microtubule poisons. Time-lapse analysis of H2B–GFP-expressing HeLa cells treated with siRNAs against control or TAO1 siRNA 3. Successive frames taken every 3 min from a representative live-cell movie are shown. Note that control cells (d) delay the onset of anaphase in response to unaligned chromosomes (white arrow), whereas TAO1-depleted cells (e) initiate anaphase despite the presence of unaligned chromosomes (white arrows). White arrowheads indicated lagging chromosomes. (f) Quantification of the mitotic defects from live-cell movies in d and e. (g) Partial TAO1-depletion with TAO1 siRNA 1 causes mitotic arrest (no segregation), whereas more complete depletion of TAO1 protein levels using TAO1 siRNA 3 (see Fig. 1g) causes defective segregation because of impaired spindle-checkpoint function (see Fig. 1e). The scale bars in a, b, d and e represent 10 m. The error bars in c, f and g represent s.e.m. (n >100 from three independent experiments).
We next explored whether TAO1 is also required for mitotic progression during an unperturbed cell cycle. Control-depleted cells delay the onset of anaphase in response to unaligned chromosomes (Fig. 3d). In contrast, many TAO1-depleted cells initiated anaphase in the presence of unaligned chromosomes and displayed segregation errors in anaphase (Fig. 3e). Quantification of mitotic defects in TAO1-depleted cells revealed frequent failure in the formation of the metaphase plate (fivefold increase), the presence of unaligned and lagging chromosomes during anaphase (fivefold and 17-fold increase, respectively), and cytokinesis defects (sixfold increase; Fig. 3f). These findings indicate that TAO1, similarly to other checkpoint proteins16, 17, is required for proper mitotic progression, even in an unperturbed cell cycle.
In live-cell analyses, partial depletion of TAO1 using siRNA 1 caused cells to arrest in mitosis with no segregation (Fig. 3g), whereas a greater depletion using TAO1 siRNA 3 (Fig. 1g) caused checkpoint failure and defective segregation (Figs. 1e and 3g). This seemingly paradoxical situation has previously been observed for the spindle checkpoint kinase Bub1 and has been attributed to its dual functions in regulating chromosome congression and spindle-checkpoint signalling18, 19. Partial depletion of Bub1 is thought to impair chromosome congression, but not checkpoint function, thus leading to mitotic arrest19, 20. Consistent with the hypothesis that TAO1 may also regulate aspects of chromosome congression, many TAO1-depleted cells initiated anaphase before the formation of a clear metaphase plate (Fig. 3e, f). To exclude the possibility that this apparent congression defect is an indirect consequence of premature anaphase initiation, anaphase onset was delayed by treating cells with the proteasome inhibitor MG132, which prevents the initiation of anaphase by stabilizing APC substrates. After 1 h MG132 treatment, 10% of Mad2- or control-depleted mitotic cells displayed incomplete chromosome congression, whereas more than 60% of TAO1-depleted cells exhibited uncongressed chromosomes (Fig. 4a). In addition, broken spindle structures with fewer microtubules were frequently observed in TAO1-depleted cells (Fig. 4b). Using complementation experiments with siRNA-resistant wild-type TAO1 and kinase-dead constructs, the kinase activity of TAO1 was shown to be required for proper chromosome congression (Fig. 4c). Taken together, these findings demonstrate that the kinase activity of TAO1 is required for efficient chromosome congression.
Figure 4: TAO1 kinase regulates chromosome congression in an Aurora-B independent manner.
(a) TAO1 is required for chromosome congression. HeLa cells were treated with the indicated siRNAs for 48 h, treated with MG132 for 1 h to block cells at the metaphase–anaphase transition, fixed and stained for chromosomes using DAPI for microscopic analysis. The graph shows the percentage of mitotic cells with uncongressed chromosomes (chromosomes that failed to align on metaphase plate). (b) Immunofluorescence microscopy images illustrating congression defects and abnormal spindle structures in TAO1-depleted cells. Cells were treated as described in a and probed with CREST antiserum as kinetochore marker (red), 
-tubulin antibodies (green) and DAPI for DNA staining (blue). (c) Expression of siRNA-resistant wild-type TAO1 kinase, but not its kinase dead mutant, rescues congression defects in TAO1-depleted cells. HeLa cells were treated with the indicated siRNAs, and 24 h later were transfected with His–DsRed vector (transfection marker) in the presence or absence of the indicated siRNA-resistant TAO1 vectors. The graph shows the percentage mitotic cells with uncongressed chromosomes after MG132 treatment (1 h). (d, e) Co-inactivation of TAO1 and Aurora B leads to increased congression defects. Still images of congression plates (d) indicating the extent of misaligned chromosomes in mitotic cells treated with the indicated siRNA and/or chemical inhibitors following MG132 treatment (1 h). Percentage of cells with misaligned chromosomes (e; 0, 1–4 and >4 chromosomes positioned away from the spindle equator). (f) TAO1 inhibition does not affect phosphorylation of the Aurora B substrate CENPA. Immunoflourescence microscopy images of control or TAO1 siRNA-treated cells stained using anti-phospho-CENPA antibody. The scale bars in b, d and f represent 10 m. The error bars in a and e represent s.e.m. (n >100 from three independent experiments) The error bars in c represent s.e.m. (n >35 from three independent experiments).
Previous studies of dual-function checkpoint kinases established that Bub1 and Aurora B function in parallel pathways19, whereas BubR1 was proposed to function upstream of Aurora B21. We therefore examined the epistatic relationship between TAO1 and Aurora B. The simultaneous inhibition of TAO1 and Aurora B function (using either siRNAs or the chemical inhibitor VX-680; ref. 22) led to more pronounced congression defects than their individual depletions (Fig. 4d, e and see Supplementary Information, Fig. S2c). Moreover, in contrast with BubR1 depletion21, the depletion of TAO1 did not affect the in vivo phosphorylation of the previously described Aurora B kinase substrate CENPA23 (Fig. 4f). These findings are consistent with the interpretation that TAO1 and Aurora B function in independent pathways.
To explore how TAO1 depletion compromised checkpoint function, we investigated the kinetochore enrichment of other spindle checkpoint proteins in TAO1-depleted cells. The kinetochore localization of Bub1, BubR1, Mad1, Mps1 and Cdc20 was unperturbed in TAO1-depleted cells (Fig. 5a, c, see Supplementary Information, Fig. S3 and data not shown). In contrast, Mad2 was no longer enriched on kinetochores in the majority of prometaphase cells — even after brief exposure to the spindle poison, nocodazole, to increase the number of unattached kinetochores (Fig. 5b, c). These findings suggest that TAO1 regulates the spindle checkpoint, at least in part, by promoting the enrichment of Mad2 at unattached kinetochores. Investigation of the subcellular localization of TAO1 in mitotic cells indicated that TAO1 localizes to speckles throughout the cell, some of which may colocalize with kinetochores (see Supplementary Information, Fig. S4). Although consistent with a weak and transient kinetochore interaction, the low frequency of costaining prevents a definitive conclusion that Tao1 localizes to and controls the checkpoint at the kinetochore.
Figure 5: TAO1 recruits Mad2 to unattached kinetochores and interacts with BubR1.
(a, b) Immunofluorescence microscopy images of prometaphase cells (after 1 h exposure to 100 ng ml-1 nocodazole) treated with the indicated siRNAs and probed with Mad2 (a) or Mad1 (b) antibody (green). CREST antiserum (red) is used as kinetochore marker and DNA stained with DAPI (blue). Insets show the staining of one kinetochore pair at higher magnification. (c) Quantification of Mad2, Mad1 and BubR1 levels on prometaphase kinetochores. The error bars represent s.d. and n >150 for each data point. (d) TAO1 regulates the kinetochore-dependent spindle checkpoint, but not mitotic timing pathway. The timing of anaphase onset was determined from live-cell movies taken 48 h after treatment with the indicated siRNAs. The graph shows the cumulative frequency of cells that have entered anaphase at the indicated times after nuclear envelope breakdown (NBD, defined as time zero). More than one hundred cells were analysed for each siRNA depletion. (e) Exogenously expressed TAO1 (wild-type and kinase-dead mutant) associates with BubR1. HeLa cells were transiently transfected with the indicated Myc-tagged expression constructs. After transfection (24 h), cells were treated for 8 h with nocodazole before harvesting. Anti-Myc immunoprecipitates were probed for with anti-Myc and BubR1 antibodies (rabbit anti-BubR1). Myc–Cdc20 and Myc–USP44 serve as a positive and negative control, respectively. (f) Endogenous TAO1 associates with BubR1 in vivo. Extracts were prepared from asynchronously growing (cyc) or nocodazole-arrested (noc, released from thymidine block into 100 ng ml-1 nocodazole for 12 h) HeLa cells. Endogenous TAO1 was immunoprecipitated with polyclonal anti-TAO1 antibodies and the immunoprecipitates were probed for TAO1 and BubR1 (rabbit anti-BubR1). Mock immunoprecipitations using control rabbit IgG were used as negative control. (g) Reciprocal coimmunoprecipitation between BubR1 and TAO1 in nocodazole-arrested HeLa cells. Additional controls and lysates are shown in the Supplementary Information, Fig. S5. The scale bars in a and b represent 10 m.
The inability to recruit Mad2 to kinetochores should prevent template formation at kinetochores and the structural activation of Mad2 (ref. 3), potentially explaining checkpoint failure. Checkpoint failure falls into two distinct categories: failure to regulate the kinetochore-mediated 'checkpoint pathway' that monitors microtubule attachment; and failure to regulate global 'mitotic timing' and restrain anaphase onset during early mitosis24. Proteins involved in these two categories can be distinguished by the extent to which their depletions accelerate the timing of anaphase onset in cells24. Cells with reduced TAO1 function exhibited anaphase timing (24 
2 min) similarly to Mad1-depleted cells (22 2 min), but in contrast with Mad2 (16 2 min), did not markedly accelerate the onset of anaphase (Fig. 5d). These findings indicate that the disruption of the Mad2–Mad1 template by the depletion of TAO1 impairs checkpoint signalling, but does not affect Mad2-dependent mitotic-timing during early mitosis.
The only two checkpoint proteins whose depletion is known to cause a failure to recruit Mad2 to kinetochores, despite normal localization of Mad1, are TAO1 and BubR1 (ref. 24 and this study), although the functional consequences of BubR1 depletion on Mad2 localization may vary between different cell types and organisms18, 25. Therefore, we tested whether TAO1 may work together with BubR1 to control Mad2 recruitment to unattached kinetochores. Consistent with this hypothesis, immunoprecipitation experiments showed that TAO1 associates with BubR1 in vivo, and that BubR1–TAO1 complex levels were higher in checkpoint-arrested (by addition of nocodazole) than in asynchronous cells (Fig. 5e–g and see Supplementary Fig. S5a–c). These findings suggest that TAO1 and BubR1 may function within a common pathway to regulate spindle-checkpoint signalling and/or microtubule attachment.
Here, we performed a genetic screen that identified novel candidate spindle-checkpoint kinases and identified a critical role for TAO1 in spindle-checkpoint signalling. Our findings provide evidence for a new regulatory step in checkpoint signalling that precedes the amplification step postulated by the structural activation of Mad2 model3. According to this model, a diffusible APC/C-inhibitory Mad2 species is generated in a catalytic fashion using Mad2 bound to Mad1 (Mad2–Mad1) on kinetochores as template. Failure to recruit Mad2 to kinetochores would essentially stall template formation, thus preventing any further generation of the diffusible APC-inhibitory Mad2–Cdc20 complexes and causing spindle-checkpoint failure. Consistent with this model, we find that specific disruption of the Mad1–Mad2 template on kinetochores by TAO1 depletion leads to a checkpoint defect. We propose that TAO1 regulates the kinetochore recruitment of Mad2 as a seeding step to form the Mad2–Mad1 template on unattached kinetochores.
How TAO1 controls Mad1–Mad2 association is unclear. However, the observations that cells with reduced BubR1 and TAO1 function impair the recruitment of Mad2, but not that of Mad1, to unattached kinetochores (this study and refs 24, 25) raises the possibility that BubR1 and TAO1 may work together in the same pathway to actively recruit Mad2 to kinetochores. This is supported by the observation that TAO1 and BubR1 associate in a complex that is enriched in checkpoint-activated cells. Although we do not know how TAO1 and BubR1 ensure the seeding step and integrate it with the state of microtubule occupancy, our finding that the kinase activity of TAO1 is elevated in mitosis and is required for its spindle-checkpoint function, together with previous observations that both Mad2 and Mad1 are phosphorylated during mitosis in human cells26, raise the possibility that BubR1, TAO1 or a TAO1-regulated kinase regulates Mad2 kinetochore-localization through direct phosphorylation of Mad2 and/or Mad1.
We show that TAO1 has dual mitotic functions, regulating both chromosome congression and checkpoint signalling. Mechanistically, TAO1-mediated steps in segregation seem to be distinct from Aurora B kinase, which is involved in error correction. The spindle-checkpoint kinases Bub1 and BubR1 are also required for proper chromosome congression18, 19, 21, 27, but it is currently unclear whether these proteins regulate congression independently of their spindle-checkpoint function or whether these two processes are coordinately regulated to facilitate error correction. The latter scenario would be similar to the DNA damage-response pathway, which not only halts cell-cycle progression, but also activates cellular pathways that promote repair of the damaged DNA28. In this light, it is interesting to note that TAO1 has been shown to regulate microtubule dynamics through phosphorylation-dependent activation of microtubule-associated regulatory kinases (MARKs)15. As kinetochore attachment is thought to occur by a search-and-capture mechanism, which depends on the dynamic instability of microtubules29, it is tempting to speculate that TAO1 facilitates kinetochore–microtubule attachment by regulating microtubule dynamics. How the activity of TAO1 is regulated and integrated with classic checkpoint-signalling pathways will be the focus of future studies.
Methods
Construction of kinase and phosphatase shRNA library.
To generate the kinase and phosphatase shRNA library, shRNA clones targeting these gene classes were identified in a sequence-validated genome-wide miR30-based shRNA collection30 and rearrayed. The number of hairpins targeting each gene class is shown in Fig. 1b and oligonucleotide sequences for each hairpin in the library are shown in the Supplementary Information, Table S1. Transfection-quality plasmid DNA was isolated using 96 Turbo Kits (Qiagen, Germantown, MD) on a BioRobot (Qiagen) or Biomek FX (Beckman, Fullerton, CA) platform.
shRNA screen and hit validation.
HeLa cells were plated at 2,000 cells per well (96-well plate) and 24 h later cotransfected with arrayed shRNA-expressing pSM2 plasmids (100–150 ng) and 10 ng dsRed-expressing vector using Exgen500 (Fermentas, Ontario, Canada). After 48 h, taxol (100 nM) was added and cells fixed with 4% formaldehyde 24 h after taxol treatment. Hoechst 33342 (1 g ml-1) was added to stain the DNA. Cells were scored visually on an inverted microscope (Axiovert 200; Zeiss, Thornwood, NY). After the primary screen (for primary screen results see Supplementary Information, Table S3, potential positive shRNA-plasmids were reisolated, sequenced and reprobed in the taxol assay. shRNA-plasmids that scored positive in at least three independent confirmation experiments are listed in Fig. 1d.
The sequences of siRNA oligonucleotides (Dharmacon, Lafayette, CO) used for secondary hit validation are listed in the Supplementary Information, Table S4. Cells were transfected with siRNAs (100 nM) using Oligofectamine (Invitrogen, Carlsbad, CA) according to the manufacturers protocol. TAO1 siRNA 3 was used for the detailed study of its loss-of-function phenotypes unless otherwise stated in the figure legends. For control siRNA treatment, siRNAs targeting luciferase or Lamin A were used.
Plasmids constructs.
The coding sequences for TAO1 were PCR-cloned into Gateway-compatible entry vectors and transferred into Myc-tagged (amino terminus) expression vectors (gift from J. Jin, Harvard Medical School, Boston, MA) using an LR recombination kit (Invitrogen). TAO1–HA, TAO2–HA and TAO3–HA were previously described12. The kinase-dead TAO1D151A was constructed by site-directed mutagenesis using a Quickchange kit (Stratagene, La Jolla, CA). The siRNA-resistant TAO1 constructs was generated by introducing four silent base-pair changes into the target mRNA sequence using Quickchange mutagenesis kit (Stratagene).
Cell culture and synchronization.
HeLa cells were grown in DMEM supplemented with 10% FBS and antibiotics. For the nocodazole release experiment, cells were treated for 24 h with thymidine (2.5 mM) and then released for 14 h into fresh medium containing nocodazole (100 ng ml-1). For inhibiting AuroraB activity, cells were treated with VX680 (2.5 m, a generous gift from N. Gray, Harvard Medical School, Boston, MA, and synthesized according to published procedures) before the experiment. Mitotic cells were isolated by mitotic shake-off and released from the mitotic arrest by washing three times with PBS and plating into fresh medium.
Immunoblotting and immunoprecipitation.
Whole-cell extracts were prepared by cell lysis in SDS sample buffer, resolved by SDS–PAGE and transferred to nitrocellulose membranes and probed with the indicated antibodies. Rabbit anti-Cdc27 (sc-5618; Santa Cruz Biotechnology, Santa Cruz, CA), mouse anti-cyclin B1 (sc-245; Santa Cruz), rabbit anti-GFP (ab290; Abcam, Cambridge, UK), rabbit anti-GAPDH (Santa Cruz), rabbit anti-BubR1 (kind gift from H. Yu, Southwestern Medical School, Dallas, TX), rabbit anti-TAO1 (Bethyl Laboratories, Montgomery, TX) were used.
For immunoprecipitation reactions, cells were lysed in CHAPS lysis buffer (50 mM Tris–HCl at pH 7.5, 150 mM NaCl, 0.3% CHAPS, phosphatase-inhibitor cocktail (Calbiochem, San Diego, CA) and EDTA-free complete protease inhibitor mix (Roche, Basel, Switzerland). The specified antibodies were added to cleared lysates, incubated for 2 h on ice, followed by 3 h rotation with protein G–Sepharose beads (Pierce, Rockford, IL) at 4 °C. The beads were then washed five times with CHAPS lysis buffer. Proteins were eluted off the beads by boiling in SDS sample buffer, separated by SDS–PAGE and blotted with the indicated antibodies.
TAO1 kinase assay.
For in vitro kinase assays, cells were lysed in lysis buffer (50 mM Tris–HCl at pH 7.2, 150 mM NaCl, 0.5% NP40, phosphatase-inhibitor cocktail and EDTA-free complete protease inhibitor mix). Lysates were precleared for 30 min at 4 °C with protein G–Sepharose beads. The specified antibodies were added to cleared lysates, incubated for 1 h on ice, followed by 3 h rotation with protein G–Sepharose beads (Pierce) at 4 °C. The beads were then washed four times with lysis buffer, and twice with kinase buffer (50 mM Tris at pH 7.5, 10 mM MgCl2 and 1 mM DTT). The kinase assays were performed in 20 l reactions, containing 50 mM cold ATP, 3 Ci 32P-ATP, 1 mM DTT and 5 g maltose-binding protein (MBP) in kinase buffer. The reactions were incubated at 30 °C for 30 min, quenched with SDS sample buffer and analysed by SDS–PAGE followed by autoradiography.
Immunofluorescence microscopy.
Cells were fixed, permeabilized and blocked as previously described31. A deltavision RT deconvolution microscope equipped with a 100
objective was used for image acquisition. For immunostaining, rabbit anti-Mad1 polyclonal antibodies24, rabbit anti-Mad2 polyclonal antibodies(Covance, Austin, TX), anti-BubR1 and anti-Bub1 antibodies (gift from S.Taylor, University of Manchester, UK), anti-Cdc20 monoclonal antibodies (Santa Cruz), CREST antisera (gift from W. Earnshaw, University of Edinburgh, UK), phospho-CENPA antibodies (gift from D. Allis, Rockefeller University, NY) and cross-adsorbed secondary antibodies from Molecular Probes (Carlsbad, CA) were used. For quantification of kinetochore signals, CREST staining was used as reference as previously described31. At least 150 kinetochores (>10 cells) were quantified for each staining using IMARIS. DAPI (4 ng ml-1) was used to stain chromosomes in fixed cells.
Live-cell time-lapse imaging and analysis.
Cells were imaged 48 h after transfection in T 0.15 mm dishes (Bioptechs, Butler, PA) in CO2-independent medium (Gibco, Carlsbad, CA) at 37 °C. Exposures (0.1 s) were acquired every 3 min for 8 h using a 20 NA 0.75 objective on an Applied Precision Deltavision RT microscope equipped with a Mercury 100 W lamp, GFP-long pass filter (Chroma, Rockingham, VT), phase-contrast filters and Coolsnap HQ camera. The timing of anaphase onset was determined as previously described24.
Note: Supplementary Information is available on the Nature Cell Biology website.
