When unable to generate robust tension across sister kinetochore attachments, cells face prolonged mitotic duration and increased risk of chromosome missegregation, leading to aneuploidy or death. Dynamic microtubules generate robust tension with end-on attachments, whose coupling and regulation requires the function of specialized kinetochore components and MAPs, prominent examples of which are described above. Below we summarize recent advances in understanding how tension can act as a unifying factor, connecting events from the pericentromeric region to the outer kinetochore. Tension can alter kinetochore and centromere structure, regulate attachment strength, act as a modulator of Aurora B activity, and, in conjunction with the Spindle Assembly Checkpoint (SAC) promote timely anaphase onset (Figure 3). Tension across sister kinetochores is a fundamental quality of bipolar attachment that plays a key role in multiple mechanisms and thus unifies their individual functions to achieve accurate chromosome segregation.
Tension directly influences kinetochore and centromere structure
The tension produced across sister kinetochores during metaphase is a physiologically relevant force generated within the mitotic spindle (Chacón et al., 2014); (Ye et al., 2016). It has been a long-standing endeavour to elucidate how this tension status is sensed at kinetochores and centromeres. Notably, the structures of both kinetochores and centromeres are influenced by tension. Centromeres are the singular regions of DNA where kinetochore proteins assemble and they serve as a key connection point between replicated sister chromatids (for a review of centromere structure, see Lawrimore and Bloom (2022). It has long been thought that cohesion between sister chromatid DNA strands, provided by the cohesin complex, functions as the major mechanism for resistance to outward forces on their centromeres. One apparent limitation is that cohesin complexes can move relative to the associated DNA and, thus, allow centromeres to be pulled apart with relatively little resistance due to such DNA sliding. This challenge is met by ‘trapping’ the cohesin complexes between a pair of convergently oriented genes on either side of the centromere, thus limiting further DNA sliding and defining the boundaries of the pericentromeric region (Paldi et al., 2020). A notable property of DNA is that it is relatively floppy and easily extended, which does not generate much tension until it is largely extended (Bloom, 2008). Although alternative models have been proposed for the physical arrangement of the pericentromeric DNA between sister centromeres (Paldi et al., 2020), considering the length of DNA within the pericentromere, the bottlebrush model perhaps best accounts for this extensible property of DNA (Lawrimore et al., 2015). Briefly, the bottlebrush model posits that the pericentromeric DNA is organized by condensin and cohesin complexes into a looped loop structure with a central backbone and extending loops resembling the bottlebrushes used to clean test tubes (reviewed in Lawrimore and Bloom (2022). Notably, the bottlebrush model provides a mechanism by which the pericentromeric DNA adopts a stiff structure that allows for microtubule-dependent kinetochore movements to result in relatively greater tension. Centromeric stiffness increases during mitotic progression, with the greatest stiffness during metaphase (Harasymiw et al., 2019). This increased centromeric stiffness in metaphase, when tension is a vital signal for achieving bipolar attachments, passively increases tension in response to active microtubule-depolymerization forces, relative to centromeres with more extendable DNA. This centromeric stiffness creates a more sensitive mechanism to distinguish different kinetochore-microtubule configurations and ensure bipolar attachments are formed (Harasymiw et al., 2019). Altogether, the centromere and associated proteins are critical for proper tension generation and for enhancing the tension-responsive signaling that ensures bipolar kinetochore-microtubule attachments are established.
The kinetochore itself is a mechanically rigid structure compared to DNA. It has been proposed that specific proteins, such as Ndc80, are physically extended by microtubule pulling forces and thus provide a potential mechanism to report the kinetochore tension status (Suzuki et al., 2016). A related idea is that multiple kinetochore proteins or complexes change conformation or position relative to others in response to pulling forces, again serving as a physical indicator of tension (Joglekar et al., 2009); (Maresca and Salmon, 2009); (Uchida et al., 2009); (Wan et al., 2009). Indeed, a tension-dependent change in the shape of inner kinetochore proteins, particularly CENP-T, which undergoes elongation, has been observed in chicken DT40 cells (Suzuki et al., 2011). A FRET-based study using fluorescently labelled Ndc80 and kinetochore microtubules in human U2OS cells found that the number of Ndc80 molecules bound to kinetochore microtubules increased with tension (Yoo et al., 2018). A similar approach demonstrated that the Ndc80 clustering to individual microtubules is comparable at human and yeast kinetochores (Kukreja et al., 2020). Moreover, a distinct structural response has been observed between kinetochores that have lost tension versus those that lost attachment (Roscioli et al., 2020). KNL1 is shown to unravel at the loss of tension, while NDC80 jackknives due to microtubule detachment (Roscioli et al., 2020). This is unique to the other outer kinetochore proteins, which have high nematic order and do not undergo significant structural change in response to loss of tension or attachment (Roscioli et al., 2020). Altogether this evidence supports the idea that physical changes at the kinetochore serve as mechanical cues for tension-dependent processes in mitosis.
Tension regulates Aurora B and strengthens bipolar attachments
Aurora B (Ipl1) is needed to respond to tensionless kinetochores and is the major kinase involved in the associated error correction process (Chan and Botstein, 1993); (Biggins et al., 1999); (Cheeseman et al., 2002); (Tanaka et al., 2002). Error correction is the mechanism by which kinetochore-microtubule attachments experiencing insufficient tension are selectively destabilized, thus granting another chance to establish force-generating bipolar connections. It is mediated by Aurora B via phosphorylation of KMN proteins, such as the tail of Ndc80 (Biggins et al., 1999); (DeLuca et al., 2006); (DeLuca et al., 2011); (Welburn et al., 2010), as well as components of the Dam1/DASH complex (Tien et al., 2010); (Cheeseman et al., 2002). While Aurora B phosphorylates other spindle proteins, such as MAPs associated with the spindle midzone, the modification of kinetochore proteins facilitates error correction. Phosphorylation of the kinetochore components decreases their affinity for the microtubule, weakening the kinetochore-microtubule linkage and promoting detachment (Welburn et al., 2010); (Sarangapani et al., 2013). Conversely, the presence of sufficient tension avoids triggering error correction and, thus, selectively stabilizes bipolar microtubule-kinetochore attachments (McVey et al., 2021). In addition to avoiding Aurora B-mediated error correction, the kinetochore-microtubule linkage behaves like a catch bond in that its binding strength is enhanced under pulling forces (Akiyoshi et al., 2010). As a result, tension-generating bipolar attachments are inherently stabilized.
The mechanism by which Aurora B specifically targets tensionless kinetochore-microtubule attachments, yet avoids those under tension, remains elusive. There are several proposed models for Aurora B activity in response to insufficient tension at kinetochore attachments (for a review of models of Aurora B activity, see McVey et al. (2021). Two well-established classes of models are the spatial dependent (Liu et al., 2009) and the tension-sensitive activation models (Sandall et al., 2006); (Adams et al., 2000); (Bishop et al., 2005). In the spatial dependent models, once Aurora B is localized to the inner centromere, it is constitutively active and will phosphorylate any kinetochore substrate that comes within range. Thus, the components of sister kinetochores that lack bipolar pulling forces and the resultant tension to be sufficiently displaced away from the centralized Aurora B, will be modified, triggering detachment. Conversely, if microtubule-generated forces can pull sister kinetochores far enough apart, concomitantly producing tension, their components will be physically displaced out of the zone of Aurora B phosphorylation, which indirectly stabilizes the attachments. In the activation models, Aurora B activity is modulated by the tension status rather than the proximity of kinetochore substrates. Although Aurora B may encounter its kinetochore substrates, its kinase activity would be inhibited by higher tension or stimulated by low tension. In budding yeast, tension sensing and error correction by Aurora B can occur in the absence of its centromere localization, supporting the hypothesis that Aurora B activation is triggered by tension status (Campbell and Desai, 2013). Further work consistent with the activation model using purified yeast kinetochores suggests that tension-generating attachments can directly regulate Aurora B activity or oppose its outcome (de Regt et al., 2022). Another emerging model, supported by recent findings, is that Aurora B localization to the kinetochore itself, as opposed to the inner centromere, functions to phosphorylate substrates on low-tension kinetochores, while the kinase is evicted from kinetochores that establish tension (Reviewed in (Broad and DeLuca, 2020). This mechanism bears similarity to the activation model in some respects, particularly if Aurora B would be recruited back to kinetochores that subsequently lose tension. In addition to canonical models where Aurora B activity mainly promotes detachment, a recent study provides evidence that the tension status influences the downstream effect of Aurora B phosphorylation. Namely, under low tension Aurora B caused kinetochore microtubules to depolymerize without detachment, but under high tension, the microtubules detach, a difference which the authors propose may be relevant to correcting distinct attachment errors (Chen et al., 2021).
Tension and the spindle assembly checkpoint work together to promote timely metaphase-anaphase transition
Much work has been done to understand how microtubule-generated tension at kinetochores and the SAC work together or independently to promote accurate chromosome segregation. These questions have been difficult to approach experimentally due to the challenge of isolating the effect(s) of the tension status from kinetochore attachment. Very briefly, conditions that reduce tension can also inhibit attachment or induce error correction-mediated detachment, while those preventing attachment also preclude tension. In the presence of unattached kinetochore(s), the SAC will delay anaphase onset (for a full review of the SAC, see (Lara-Gonzalez et al., 2021). Phosphorylation of the kinetochore protein Spc105/KNL1 at unattached kinetochores by Mps1 promotes localization of the Bub (Bub1, Bub3) and Mad (Mad1, Mad2, and BubR1/Mad3) proteins (London et al., 2012); (London and Biggins, 2014). This kinetochore localization leads to catalytic formation of the Mitotic Checkpoint Complex (MCC), which inhibits the Anaphase Promoting Complex (APC/C) by sequestering Cdc20, a required activator of the APC (Lara-Gonzalez et al., 2021).
SAC signalling is necessary to provide sufficient time to establish proper kinetochore-microtubule attachments yet also allows for timely mitotic progression. Along these lines, experiments in budding yeast suggest Bub1 works with the well-established tension-sensitive protein, Sgo1, and the phosphatase PP2A to prevent premature SAC silencing (Jin et al., 2017). On the other hand, recent data indicates that Bub1 and Aurora B work cooperatively to maintain SAC signalling once initiated, even after Mps1 activity at kinetochores has diminished (Roy et al., 2022).
One outstanding mystery, owing to the fact that unattached kinetochores are inherently tensionless, is whether the tension status plays any role in the canonical SAC mechanism at unattached kinetochores. It has been postulated that tension plays a key role in signalling the establishment of proper attachments, and thus silencing the SAC. Although significant work has been directed at understanding how SAC signalling is extinguished once appropriate conditions are met, the mechanistic basis remains largely obscure.
An important step in SAC activation is Mps1-mediated phosphorylation, and thus, silencing the SAC could entail preventing Mps1 phosphorylation of Spc105/KNL1. One potential mechanism is that microtubules and Mps1 competitively bind Ndc80, with microtubule binding displacing Mps1, thus silencing the SAC (Hiruma et al., 2015); (Ji et al., 2015). This competitive binding model has been challenged by work done in Drosophila demonstrating that as microtubule attachments are established, Mps1 localization at kinetochores decreases, except for a small fraction that remains on kinetochores until anaphase onset (Moura et al., 2017). Follow up work revealed that Mps1 localization to syntelic, end-on attached kinetochores proceeded their detachment (Hayward et al., 2022). Mps1 transiently localizes to these kinetochores, which have high levels of Aurora B phosphorylation, to promote timely error correction (Hayward et al., 2022).
Another potential mechanism is that the SAC is silenced by tension-dependent kinetochore stretching, and the sustained deformation of kinetochore components at the attachment interface (Maresca and Salmon, 2009); (Uchida et al., 2009). Other work in budding yeast has posited a mechanical switch in the kinetochore upon end-on attachment that prevents SAC signaling from persisting, where by Mps1 is prevented from phosphorylating Spc105 (Aravamudhan et al., 2015). In this model, Dam1 acts as a barrier preventing Mps1 access to Spc105 (Aravamudhan et al., 2015).
In general, it is thought that extinguishing SAC signaling, regardless of mechanism, entails either eviction of Mps1 from the kinetochore or a decrease in Mps1 activity. In Drosophila, Mps1 activation is regulated by PP1-87B phosphatase activity that antagonizes the increased Mps1 activity resulting from T-loop autophosphorylation (Moura et al., 2017). In human cells, this T-loop autophosphorylation, as well as phosphorylation by Aurora B that impacts Mps1 kinetochore localization and activation, is antagonized by PP2A-B56 (Hayward et al., 2019). In both cases, a decrease in phosphatase activity leading to persistent Mps1 T-loop phosphorylation and Mps1 activity resulted in prolonged mitotic arrest (Moura et al., 2017); (Hayward et al., 2019). Altogether, phosphoregulation of Mps1, resulting in its activation or kinetochore localization, is mediated via PP1-87B and PP2A-B56 to promote SAC silencing and timely anaphase onset.
Recent work has generated conflicting evidence for whether tension is needed to silence the SAC or if microtubule attachment alone is sufficient. PP1 is a phosphatase that antagonizes Mps1 activity at the kinetochore via dephosphorylation of kinetochore substrates. In HeLa cells, intra-kinetochore stretching is diminished in monopolar spindles, leading to decreased PP1 recruitment to kinetochores (Uchida et al., 2021). At these PP1 deficient kinetochores, downstream SAC proteins remained, leading to delayed anaphase onset, irrespective of Mps1 localization (Uchida et al., 2021). This work suggests that tension is needed for kinetochore stretching to silence the SAC. On the other hand, evidence indicates that a lack of tension can, in certain cases, activate the SAC. In HAP1 cells with deficient Kif18A, a kinesin-8 motor, the resulting lack of tension activates the SAC (Janssen et al., 2018). At Mad1 positive kinetochores in these cells, tubulin/microtubule signal was equivalent to that in control cells, demonstrating that kinetochores were fully attached even though the SAC was active (Janssen et al., 2018). These findings indicate that insufficient tension can activate the SAC, regardless of attachment status (Janssen et al., 2018). In budding yeast, deletion of the conserved kinesin-5, Cin8, results in a delay in anaphase onset due to a lack of tension via Ndc80 (Suzuki et al., 2018). This reduced tension is associated with sustained error correction-mediated phosphorylation of Ndc80 due to disrupted kinetochore recruitment of PP1 to antagonize Ipl1 activity, resulting in detachment and SAC activation (Suzuki et al., 2018). Together these studies reveal that the tension generated by end-on attachments is a signal that potentially facilitates multiple mechanisms that activate or silence the SAC.
Contrary to tension being a central signal, other evidence suggests that attachment is the major signal needed to satisfy or silence the SAC. Cells expressing a non-phosphorylatable Ndc80 tail mutant (Hec1-9A) can establish stable end-on attachments. Notably, when these cells are induced to form monopolar spindles, they are unable to generate the normal tension associated with bipolar attachments yet still establish stable end-on connections (Etemad et al., 2015). Although the SAC appears functional in these cells, the low-tension attachments can sufficiently satisfy the checkpoint (Etemad et al., 2015). In another study, Hec1-9A and wild type cells were treated with low dose nocodazole to deplete poleward pulling forces (Tauchman et al., 2015). While nocodazole treatment arrested wild type cells, the Hec1-9A cells satisfied the SAC and entered anaphase, which was corroborated by a decrease in Mad1 positive kinetochores (Tauchman et al., 2015). Additional evidence in human cells show that low kinetochore-microtubule occupancy, or fewer microtubules bound to a kinetochore compared to the typical number at those “fully attached”, does not impact SAC silencing (Etemad et al., 2019). SAC proteins were almost undetectable at kinetochores, despite only about half the total possible microtubules bound to kinetochores (Etemad et al., 2019). This body of work suggests that microtubule-kinetochore attachment is sufficient to silence the SAC, regardless of maximum kinetochore occupancy or tension-generating status. These experiments represent significant technical advances that continue to take us closer to understanding how tension contributes to signalling at kinetochores. They also highlight the intricate relationship between attachment and tension at kinetochores, and the challenges in elucidating the role of each in SAC activation or silencing.
While it remains unclear how or if tension contributes to SAC signalling, the tension status can impact anaphase onset independently of SAC silencing. Work in budding yeast shows Bub1 and Bub3 function outside of their canonical SAC role to delay anaphase onset in response to low-tension but attached kinetochores (Proudfoot et al., 2019). Thus, a tension-sensitive mechanism may provide extra time for kinetochores to come under tension prior to anaphase or to trigger error correction to sample for a bipolar attachment (Proudfoot et al., 2019). Altering the level of tension generated in the spindle by deleting specific MAPs (Cin8, Kip1, Ase1) has been shown to produce a graded response, where detachment mediated by kinetochore protein phosphorylation via Aurora B is enhanced by decreasing centromeric tension (Mukherjee et al., 2019). This work reveals that tension-sensitive signalling mechanisms can be sensitive to the magnitude of forces experienced.
While phosphorylation has been a widely investigated post-translational modification that mediates signalling at centromeres and kinetochores, SUMOylation of proteins has recently been found to also play a role. The SUMOylation status of Sgo1, a well-established tension-sensitive protein, influences the timing of anaphase onset (Su et al., 2021). Sgo1 is recruited to the pericentromeric region during metaphase by phosphorylation of S121 on histone H2A (Fernius and Hardwick, 2007); (Yamagishi et al., 2010). This positioning allows Sgo1 to promote chromosome biorientation by facilitating Chromosome Passenger Complex (CPC) localization to the centromere and recruiting cohesin and condensin (Nerusheva et al., 2014); (Verzijlbergen et al., 2014); (Peplowska et al., 2014). Sgo1 SUMOylation is needed to keep sister chromatids in a stable bioriented state. Metaphase arrested cells harbouring a mutation in the coiled-coil region of Sgo1 (sgo1-4R) that reduces SUMOylation display cycles of biorientation, loss of tension, detachment, then reattachment to become bioriented again (Su et al., 2021). Even with multiple rounds of unnecessary error correction, these cells have significantly lower chromosome missegregation levels than those lacking Sgo1. Together the results suggest that SUMOylation of Sgo1 and the CPC component Bir1 work to decrease Sgo1 localization, and thus dampen Aurora B (Ipl1)-mediated error correction to selectively stabilize bioriented kinetochore-microtubule attachments and promote timely anaphase onset (Su et al., 2021). More work is needed to understand how the tension status is transmitted to these downstream effectors.