Antagonism between the dynein and Ndc80 complexes at kinetochores controls the stability of kinetochore-microtubule attachments during mitosis

Chromosome alignment and segregation during mitosis depends critically on kinetochoremicrotubule (kMT) attachments that are mediated by the function of the molecular motor cytoplasmic dynein, and the kinetochore microtubule (MT) binding complex, Ndc80. The RZZ (Rod-ZW10-Zwilch) complex is central to this coordination as it has an important role in dynein recruitment and has recently been reported to have a key function in the regulation of stable kMT attachment formation in C. elegans. However, the mechanism by which kMT attachments are controlled by the coordinated function of these protein complexes to drive chromosome motility during early mitosis is still unclear. In this manuscript, we provide evidence to show that Ndc80 and dynein directly antagonize each other’s MT-binding. We also find that severe chromosome alignment defects induced by depletion of dynein, or the dynein adapter spindly, are rescued by codepletion of the RZZ component, Rod, in human cells. Interestingly, the rescue of chromosome alignments defects was independent of Rod function in activation of the spindle assembly checkpoint and was accompanied by a remarkable restoration of stable kMT attachments. Furthermore, rescue of chromosome alignment was critically dependent on the plus-end-directed motility of CENP-E, as cells codepleted of CENP-E along with Rod and dynein were unable to establish stable kMT attachments or align their chromosomes properly. Taken together, our findings support the idea that the dynein motor may control the function of the Ndc80 complex in stabilizing kMT attachments either directly by interfering with Ndc80-MT binding, and/or indirectly by modulating the Rod-mediated inhibition of Ndc80.


Introduction
Faithful chromosome segregation during mitosis requires proper chromosome congression, which relies on multiple mechanisms that ultimately lead to chromosome bi-orientation, an arrangement where the chromosomes are connected to microtubules from both spindle poles (1). In the classical 'search and capture' model, chromosomes move toward the spindle equator as a result of their biorientation (2). After nuclear envelope breakdown (NEB), chromosomes have been known to congress by two main mechanisms: microtubule polymerization/depolymerization-based motion (3) and motor-dependent transport along microtubules achieved by the coordinated activities of Dynein, CENP-E and Chromokinesins (4)(5)(6). The peripheral chromosomes are first transported by dynein to a microtubule-dense region near the spindle pole, from where they move towards the spindle equator along pre-existing spindle microtubules with the help of the CENP-E kinetochore motor (7)(8)(9).
The RZZ (Rod-ZW10-Zwilch) complex has been reported to be a key player in the spindle assembly checkpoint (SAC) activation as it is required to recruit SAC proteins, Mad1 and Mad2 to kinetochores in both Drosophila and humans (10)(11)(12)(13)(14)(15). More importantly, it has also been shown that the RZZ complex is important to recruit dynein to kinetochores through its direct association with the dynein adaptor protein, Spindly (16)(17)(18)(19)(20)(21). However, it is clear that there are also RZZindependent mechanisms (such as the CENP-F-NudE pathway) contributing to this function (22,23). The dynein motor has been shown to be involved in rapid movement of mono-oriented chromosomes towards the spindle poles via dynamic lateral interaction between kinetochores and astral microtubules during early prometaphase, thus contributing to chromosome alignment

Results and Discussion
Evidence for coordination between the dynein and Ndc80 kinetochore modules for proper chromosome alignment in humans It is well-known that the attachments between kinetochores and kMTs in early mitosis are dynamic in nature to favor kinetochore MT-motor dependent chromosome motility that drives chromosome congression, and aid in attachment error correction (35). It is also established that the Ndc80 complex at kinetochores form strong attachments with spindle MTs to stabilize kMT attachments during chromosome alignment and bi-orientation at the spindle equator in metaphase (36,37), and that purified Ndc80 binds to microtubules with high affinity in vitro (30,38,39). Consistent with this, we find that relatively low concentrations of GFP-tagged Hec1/Nuf2 dimer of the Ndc80 complex (1-5 nM) bound readily to Dylight405-labeled MTs immobilized on coverslips in in vitro TIRF microscopy assays (Fig. 1A). In addition, it is known that the Ndc80 complex and the dynein motor share similar binding sites on MTs (40,41), which points to the notion that Ndc80-mediated kMT attachments may be mutually exclusive with dynein-based attachments during mitosis. We tested this directly by carrying out TIRF-M assays using GFP-tagged Hec1/Nuf2 dimer and TMRlabelled Dynein-Dynactin-BicD2 (DDB) (42). We found that the presence of higher concentrations of the Hec1/Nuf2 dimer (20 nM), strongly inhibited the binding of the DDB complex to MTs, supporting this prediction (Fig. 1, B and C). Consistent with this finding, it has been observed in Xenopus cells that the velocity of dynein-based poleward movement was substantially enhanced in the absence of the Ndc80 complex (26).
Our data supports the idea that the MT-binding activity of the Ndc80 complex could be directly affected by higher local concentrations of dynein at prometaphase kinetochores. We surmise that the initial capture and dynein-dependent poleward motility of kinetochores could thus be a natural bias for attachments that are dynamic in nature and a mechanism that prevents Ndc80-mediated stable attachments during early mitosis. While our data supports the hypothesis that dynein and Ndc80 directly affect each other's MT-binding, it is also possible that there are mechanisms similar to that observed in C. elegans involving the components of the dynein module, including Spindly and Rod, that might modulate the function of Ndc80 in humans, and we sought to test this possibility next.

Defects in chromosome alignment induced by Spindly or dynein depletion are rescued by codepletion of Rod
As Spindly has been shown to relieve the inhibition of Ndc80 by Rod to enable the formation of Ndc80-mediated stable kMT attachments in worms (17,27), we tested whether Rod is functionally related to Spindly in humans similar to that of the observation in worms. We sought to assess the phenotype of chromosome alignment in cells where the function of Rod and/or the dynein anchor, Spindly, was disrupted by RNAi-mediated knockdown of both the proteins. Efficient depletion of the target proteins was validated by immunoblotting as well as immunostaining analyses (Fig. 1, D and E; Supplemental Fig. S1A). We found that mitotic cells depleted of Rod (Rod siRNA ) exhibit no apparent defect in chromosome alignment at the metaphase plate (Fig. 1E, Supplemental Fig.   S1A). As observed in worms (17,27), the severe chromosome misalignment produced by Spindly depletion (Spindly siRNA ) was rescued by the codepletion of Rod (Spindly/Rod siRNA ) (Fig. 1, E and F; Supplemental Fig. S1A). The frequency of mitotic cells with misaligned chromosomes was significantly lower after Spindly/Rod siRNA , compared to that of Spindly siRNA (Fig. 1F). These observations suggest that the modulation of Rod function by Spindly to aid in the formation of Ndc80-mediated stable kMT attachments is conserved from worms to humans; but the molecular mechanism of how this inhibition is accomplished is poorly understood.
Since dynein employs Spindly as an adaptor to bind to the RZZ complex and get recruited to kinetochores (17)(18)(19), we then tested if chromosome alignment defects after dynein siRNA are also rescued by codepletion of Rod. To effectively deplete dynein, we designed a new siRNA targeting the 3'UTR sequence of the DYNC1H1 gene. Efficient depletion of the target proteins was validated by immunoblotting as well as immunostaining analyses (Fig. 2, A-C; Supplemental Fig. S1A). As expected (21,24,43), depletion of dynein caused a significant increase in the percentage of mitotic cells with misaligned chromosomes in HeLa cells (Fig. 2, C and D). On the other hand, the frequency of cells with misaligned chromosomes was significantly reduced after dynein/Rod siRNA compared to that of dynein siRNA , and was almost similar to that of control siRNA (Fig. 2, C and D; Supplemental Fig. S1A). As a positive control, depletion of Ndc80 using siRNA-mediated knockdown of the Hec1 subunit (Ndc80 siRNA ) expectedly led to severe chromosome alignment defects (44,45) (Fig. 2D; Supplemental Fig. S2, A and B). We further confirmed the rescue of chromosome alignment defects after dynein/Rod siRNA by live imaging. We found that ~80% of control cells could align their chromosomes at the metaphase plate within 30 min of the NEB (nuclear envelope breakdown) whereas ~75 % dynein siRNA cells were not able to do so even 120 min after the NEB. On the other hand, ~60 % of dynein/Rod siRNA cells could align their chromosomes with only a mild delay compared to that of control siRNA cells (Fig. 2, E and F; Supplemental movies 1-4). As expected, we also observed severe chromosome alignment defects by live cell imaging after Ndc80 siRNA ( Fig. 2F; Supplemental Fig. S2C; Supplemental movies 5 and 6). Thus, our results suggest that the defect in chromosome alignment resulting from dynein depletion was surprisingly rescued by codepletion of Rod.
The existing paradigm demonstrating a critical role for dynein in the rapid poleward movement of chromosomes during early mitosis to drive chromosome alignment originates from studies in large mitotic cells such as newt pneumocytes where the chromosomes are separated by relatively large distances (several 10s of µms) from the spindle poles (46). However, in smaller mitotic cells such as HeLa where the chromosomes are separated only by smaller distances for the spindle poles (usually within 5-10 µms), we surmise that the disengagement of dynein/spindly from Rod and Ndc80 is a major cause of chromosome misalignment after dynein-or spindly-depletion, due to which Rod is able to impart a sustained inhibition of Ndc80 function. It is also possible that the polar ejection forces on chromosome arms produced by the MT plus-end-directed chromokinesin motors drive the chromosomes away from the spindle poles during early mitosis in the absence of dynein or spindly to hinder proper chromosome alignment (4,47) (also see in Fig. 3).

The restoration of proper chromosome alignment in Rod-depleted cells is independent of its function in the spindle assembly checkpoint activation
The RZZ complex is important to recruit the spindle assembly checkpoint (SAC) proteins, Mad1 and Mad2 to kinetochores in both Drosophila and humans (10)(11)(12)(13)(14)(15). To scrutinize the possibility that the rescue of chromosome alignment after dynein/Rod siRNA was not due to aberrant checkpoint silencing, we tested the recruitment of SAC protein Mad1 to kinetochores of mitotic cells in prometaphase. We found that detectable levels of Mad1 was still present at kinetochores in prometaphase cells depleted of Rod, dynein, or of both Rod and dynein (Supplemental Fig. S1B).
When we analyzed mitotic progression by live imaging, we found no apparent defect in chromosome alignment at the metaphase plate after dynein/Rod siRNA similar to that of Rod siRNA (Fig. 2, E and F). Moreover, in both cases, live cells that were not treated with MG132 neither underwent premature anaphase onset during mitosis nor exhibited micro-nuclei formation in interphase (data not depicted), implying that the function of Rod in chromosome alignment could be independent of its role in SAC activation during mitosis in human cells. These data also suggest that dynein/Rod siRNA cells can align their chromosomes properly without impairing checkpoint activation.

The rescue of chromosome alignment defects after dynein and Rod codepletion is accompanied by restoration of attachment and stability of kMT
Proper chromosome alignment at the metaphase plate is accompanied by the stabilization of kMTs, which are MT bundles that extend from the bi-oriented kinetochores to spindle poles (48)(49)(50). Our immunostaining data showed that kMTs resistant to cold treatment were markedly reduced in mitotic cells after dynein siRNA as compared to that of control siRNA (Fig. 3, A-C). Surprisingly, we found that dynein/Rod siRNA cells were able to form robust kMTs to a similar extent as was observed after control siRNA or Rod siRNA (Fig. 3, B and C). As a positive control, we observed a severe lack of cold-stable kMTs after Ndc80 siRNA (Fig. 3, B and C). These data suggest that Rod's inhibition of Ndc80-mediated kMT attachments (27) is promoted in the absence of the counter-activity of dynein on Rod; as a consequence kMT attachments are unstable in dynein-depleted cells.
Analyses of kMT attachments in cells treated with 0.2mM CaCl2 followed by immunostaining showed that dynein siRNA caused misaligned chromosomes with syntelic, monotelic and unattached kMT attachments, and reduced the average distance between sister kinetochore pairs (k-k distance) at the spindle equator of metaphase plate (Fig. 3, D-F). In the absence of the poleward-directed motor activity of dynein, we surmise that the misaligned chromosomes are driven to form syntelic and monotelic kMT attachments due to the activity of plus-end-directed chromokinesin motors (4,47). We found that the defects in kMT attachments and k-k distances resulting from dynein depletion were rescued by codepletion of Rod. The number of defective kMT attachments was significantly lower after dynein/Rod siRNA compared to that of dynein siRNA , and was similar to that of control siRNA (Fig. 3, D and E). The average k-k distances after dynein/Rod siRNA was restored to that of control siRNA (Fig. 3F), possibly due to the re-establishment of stable kMT attachments that we observed. As a positive control, Ndc80 siRNA cells exhibited severely defective kMT attachments and abnormal inter-kinetochore stretch (44,51) (Fig. 3, D-E). These results suggested that codepletion of Rod rescued the defects in chromosome alignment resulting from dynein depletion by restoring the robustness of kMTs, the stability of kMT attachments, and average interkinetochore distances.
We hypothesize that during early mitosis, in the absence of the kinetochore dynein module, Ndc80 plays a role in the formation of initial kMT attachments, which are subsequently stabilized after chromosome alignment at the metaphase plate driven by CENP-E and/or chromokinesins. To support this prediction, we tested the status of kMT attachment of misaligned chromosomes in cells depleted of dynein or Ndc80. Close inspection of misaligned chromosomes showed that kinetochores were attached to MTs with syntelic or monotelic orientation after dynein siRNA while those in Ndc80 siRNA cells remained unattached (Fig. 3D), suggesting that Ndc80 might have an unexplored, yet important role in initial kMT capture. These observations could also explain how chromosomes are still captured by spindle MTs after dynein/Rod siRNA . These results support the idea that a key function of Rod is to inhibit Ndc80 because kMT attachments are rescued when Rod is codepleted with either spindly (17,27,52) or dynein (this study).
Together, these data suggest that Rod is a negative regulator of stable kMT attachments and serves not only to perturb the function of the Ndc80 complex (27) but also that it's inhibitory function is controlled by Spindly and dynein, the mechanism for which is yet unclear. The above data also suggest that chromosome alignment can be achieved at the spindle equator in the absence of the dynein module for human kinetochores, when normal chromosome alignment could possibly occur either by the activity of residual CENP-F/NudE-recruited dynein (22,23) and/or by the activity of the plus-end directed kinetochore motor, CENP-E.

The chromosome alignment in cells codepleted of dynein and Rod is dependent on the motor activity of CENP-E
As normal chromosome alignment persists after dynein/Rod siRNA , we sought to investigate if the plus-end directed kinetochore motor CENP-E, which is involved in transporting unattached sister kinetochores along pre-existing microtubule bundles to the metaphase plate (4,8,53,54), was involved in chromosome congression by simultaneously perturbing CENP-E function using siRNA-mediated knockdown in dynein/Rod siRNA cells. Efficient depletion of the target proteins was validated by immunoblotting as well as immunostaining analyses (Fig. 4, A and C; Supplemental Fig. S3, A and B). Immunostaining data showed that the frequency of cells with severe chromosome misalignment (more than 5 chromosomes) was significantly higher after dynein/Rod/CENP-E siRNA (~54%) as compared to that after control siRNA (~5%), dynein/Rod siRNA (~16%), or CENP-E siRNA (~ 21%) ( The lack of rescue of chromosome alignment after dynein/Rod/CENP-E siRNA prompted us to test whether stable kMT attachments were formed normally in these cells. We observed a substantial decrease in the intensity of kMTs at the spindle equator after dynein/Rod/CENP-E siRNA similar to that of Ndc80 siRNA (Fig. 4, E and F), and in contrast to what was previously observed after dynein/Rod siRNA or CENP-E siRNA (Fig. 3, B and C; 4, E and F). We believe that the severe chromosome misalignment produced after dynein/Rod/CENP-E siRNA prevents proper kinetochore biorientation, due to which the kMTs retain their cold sensitive nature. Moreover, the average inter-kinetochore distance was significantly reduced after dynein/Rod/CENP-E siRNA in contrast to that of control siRNA or CENP-E siRNA , and similar to that of Ndc80 siRNA (Fig. 4G). These observations suggest a biased mechanism for the rescue of chromosome alignment after dynein/Rod siRNA that is mediated by the plus-end directed motility of CENP-E.

Summary and conclusions
Overall, this study establishes a functional relationship between the kinetochore dynein module and Ndc80 module for proper chromosome alignment in humans. Rod is a key recruiter of the dynein module because it is involved in recruiting both dynein and Spindly to kinetochores (16,18,19,52,56). Our experimental results show that depletion of Rod alone has no apparent effect on chromosome alignment (Fig. 1, E and F). We found that severe defects of chromosome alignment after Spindly siRNA were rescued by codepletion of Rod in humans similar to the observation in worms (17,27). Surprisingly, we also found that severe defects in chromosome alignment after dynein siRNA were rescued by codepletion of Rod. As Spindly, a dynein adapter, counteracts the inhibitory role of Rod in the formation of Ndc80-mediated stable end-on kMT attachments (17,27), the analogous relationship between dynein and Rod; similar to that of Spindly and Rod, led us to propose that dynein could also directly counteract the inhibitory role of Rod on stable kMT attachments during early mitosis. Therefore, the phenotype of either Spindly siRNA or dynein siRNA represent the outcome of the disengagement of Rod from these factors, thus leading to severe chromosome alignment defects. Taken together these observations lead us to conclude that the role of dynein module in chromosome alignment depends on the function of Rod. It is not clear at this point whether it is the Rod/Spindly-dependent or CENP-F/NudE-dependent kinetochore dynein population, that is critical for the dynein-module mediated control of Ndc80 function in early mitosis.
Under normal condition after the NEB, kinetochores that are initially attached to spindle MTs by "search and capture" mechanism get rapidly transported poleward along MTs primarily by minus-end-directed motor force of dynein, and consequently chromosomes congress to the spindle equator by the activity of CENP-E motor (4). From recent studies (17,27 and this study), we propose a refined model for controlling kMT attachments during early mitosis in human cells, where Spindly inhibits Rod directly and/or through the dynein motor. Initial kMT attachments can still be formed by Ndc80 but are not stabilized due to antagonistic activity between the dynein module and Ndc80 (Fig. 5, details in legends). However, the precise mechanism for how Spindly/dynein interferes with Rod function and how Rod inhibits Ndc80 is stillunclear. The initial kMT attachments allow dynein-mediated transport of chromosomes to the spindle pole in early mitosis during which both CENP-E activity and end-on attachment formation are expected not to be favored (7). Therefore, in cells codepleted of dynein and Rod during early mitosis, initial kMT attachments can be formed by Ndc80 and chromosomes can be transported poleward by shrinkage of peripheral MT bundles, after which they get congressed at the metaphase plate mediated by CENP-E activity. Our current data leads us to propose the existence of two possibly interconnected mechanisms to control the stability of kMT attachments by the Ndc80 complex. Dynein might inhibit Rod to aid in Ndc80-mediated initial kMT attachments, while at the same time, or in an independent manner, directly inhibit the binding of the Ndc80 complex to MTs to prevent premature kMT attachment stabilization during in early mitosis.
Thus, it is becoming clear that stable kMT attachments are regulated not only by components within the dynein module but also by a direct interplay between the dynein module and the Ndc80 complex to prevent premature kMT stabilization. Thus, our results reveal a further layer of the elaborate network involving these distinct modules that are under tight spatiotemporal regulation to control kMT attachments and drive proper chromosome alignment at the metaphase plate. It will be important to continue studying the relationship between these modules with each other and with the protein complexes at the plus-ends of MTs in an effort to understand how they might coordinate the events involved in the formation and maturation of kMT attachments during chromosome alignment and segregation.

Cell culture, transfections and drug treatments
HeLa cells were grown in Dulbecco's modified Eagle medium (DMEM, Life Technologies) supplemented with 10% fetal bovine serum at 37 ˚C in humidified atmosphere with 5% CO2. For  and subjected to immunodetection using appropriate primary antibodies. Blocking and antibody incubations were performed in 5% non-fat dry milk. Proteins were visualized using horseradish peroxidase-conjugated secondary antibodies diluted at 1:2,000 (Amersham) and the ECL system, according to the manufacturer's instructions (Thermo Scientific).

Statistical analysis
Mann-Whitney U-test was used for comparison of dispersion, and a two-sided t-test was used for comparison of average. The statistical analyses were done with Prism software (GraphPad).
Samples for analysis in each data set were acquired in the same experiment, and all samples were calculated at the same time for each data set.