The kinetochore–microtubule interface at a glance

Cell division is a fundamental process that is carried out by all organisms. During mitosis, the genetic material of each cell must be evenly distributed between both resulting daughter cells. The loss or gain of even a single chromosome can result in catastrophic consequences for the organism (Schukken and Foijer, 2018). To ensure the accurate distribution of the DNA in eukaryotes, a large macromolecular complex termed the kinetochore assembles onto the centromere of each chromosome during mitosis. The architecture of the kinetochore will not be extensively discussed in this Cell Science at a Glance article, but has been the topic of several recent review articles (Hara and Fukagawa, 2018; Hinshaw and Harrison, 2018; Joglekar and Kukreja, 2017; Musacchio and Desai, 2017; Nagpal and Fukagawa, 2016; Pesenti et al., 2016). Instead, we will focus on the interface between the kinetochore and microtubules, dynamic polymers of tubulin heterodimers. This kinetochore–microtubule interface is critically important, as depolymerization of kinetochore-associated microtubules ultimately provides the driving force for chromosome segregation (Musacchio and Desai, 2017).

Successfully harnessing the force released by microtubule depolymerization and transducing that force to the DNA requires a sufficiently stable kinetochore–microtubule attachment. The generation of a stable microtubule attachment is not a trivial task. Within the vast expanse of the cytoplasm, the kinetochore must locate and bind to microtubules. If even a single kinetochore lacks microtubule attachments, the spindle assembly checkpoint will prevent anaphase onset (Rieder et al., 1995). Furthermore, the kinetochores of each sister chromatid must bind to microtubules that emanate from opposing poles of the bipolar spindle to achieve a state termed bi-orientation. Kinetochore–microtubule interactions are therefore initially highly dynamic to facilitate correction of erroneous attachments. Once bi-orientation is achieved, the kinetochore–microtubule attachment must be stabilized for force transduction, yet also remain sufficiently dynamic so as to maintain its association even as the microtubule polymerizes and depolymerizes. In this article and accompanying poster, we discuss the features and molecular players that have key roles in overcoming these challenges in order to facilitate the formation of robust kinetochore–microtubule interactions. We focus on the kinetochore-localized proteins that bind microtubules, as well as proteins that alter the dynamics and organization of kinetochore-bound microtubules.

Accurate chromosome segregation depends on the generation of end-on kinetochore–microtubule interactions where the plus-end of the microtubule is embedded within the kinetochore. However, some kinetochores will initially associate with the side of the microtubule, rather than the end (Barisic et al., 2014; Kapoor et al., 2006; Magidson et al., 2011; Tanaka et al., 2005). These lateral associations are mediated by one of two kinetochore-localized, microtubule-based motors – cytoplasmic dynein and centromere-associated protein E (CENP-E). As a minus-end-directed motor, dynein transports chromosomes towards the spindle pole (Li et al., 2007; Vorozhko et al., 2008; Yang et al., 2007) and thereby towards a region of high microtubule density. Chromosome congression is then promoted by the plus-end directed activity of CENP-E (McEwen et al., 2001) (see poster). During mitosis, microtubule plus-ends are located at both the cell cortex and equator (Prosser and Pelletier, 2017). The directionality of CENP-E-driven chromosome transport is guided by tubulin detyrosination (Barisic et al., 2015), a modification that is found to be enriched in the equator-oriented microtubules of the mitotic spindle and depleted in the cortical microtubules (Gundersen and Bulinski, 1986). In this way, the combined actions of dynein and CENP-E help to ensure timely chromosome alignment, and also promote the formation of end-on attachments by ensuring incorporation of the chromosomes into the spindle (Itoh et al., 2018) (see poster).

To generate the initial lateral microtubule interactions, kinetochores expand their reach by forming a structure that is termed the fibrous corona (Jokelainen, 1967; Magidson et al., 2015; McEwen et al., 1993; Rieder, 1982). A subset of outer kinetochore proteins, including dynein (Wordeman et al., 1991), CENP-E (Cooke et al., 1997), CENP-F (Rattner et al., 1993; Zhu et al., 1995) and the Rod–ZW10–Zwilch (RZZ) complex (Rod is also known as kinetochore-associated protein 1; KNTC1) (Basto et al., 2004; Starr et al., 1998), form an extended crescent-shaped structure that surrounds the kinetochore in the absence of microtubules (Dong et al., 2007; Echeverri et al., 1996; Hoffman et al., 2001; Thrower et al., 1996), thereby creating a large platform to capture microtubules. In Xenoups oocytes, an even larger expansion that includes more kinetochore proteins has also been recently described (Wynne and Funabiki, 2015).

The eventual conversion from a lateral to an end-on attachment is regulated by the counteracting functions of Aurora B kinase and protein phosphatase 2A (PP2A), and facilitated by the complex between Astrin (also known as SPAG5) and small kinetochore-associated protein (SKAP; also known as KNSTRN) (Shrestha et al., 2017) (Astrin–SKAP complex; see poster). Although geometrically distinct from the final goal of end-on microtubule attachments, the early establishment of kinetochore–microtubule interactions sets the stage for the subsequent mitotic events and helps facilitate chromosome alignment at the metaphase plate.

Mature end-on microtubule attachments are required for chromosome bi-orientation and satisfaction of the spindle assembly checkpoint (Kuhn and Dumont, 2017). The central player in the formation of these stable attachments is the Ndc80 complex, a component of the kinetochore scaffold 1 (KNL1)/Mis12/Ndc80 (KMN) network. The KMN network serves as the key link between the microtubules and the DNA (see poster), as it also binds the DNA-associated inner kinetochore proteins CENP-C and CENP-T (Gascoigne et al., 2011; Huis In ‘t Veld et al., 2016; Kim and Yu, 2015; Malvezzi et al., 2013; Nishino et al., 2013; Przewloka et al., 2011; Rago et al., 2015; Schleiffer et al., 2012; Screpanti et al., 2011). CENP-T further acts as a platform to expand the microtubule-binding capacity of the kinetochore, as a single CENP-T protein can recruit two Ndc80 complexes in addition to an entire complement of the KMN network (Huis In‘t Veld et al., 2016; Pekgoz Altunkaya et al., 2016; Rago et al., 2015). The precise stoichiometry of Ndc80 and the other components of the kinetochore–microtubule interface are likely to be central to the structure, organization and functionality of these interactions. However, for simplicity, the accompanying poster primarily focuses on the core concepts, rather than attempting to define the relative and absolute molecular numbers.

To ensure proper microtubule interactions, the KMN network is regulated by a variety of mechanisms. For example, Ndc80 binding to microtubules is negatively regulated by the RZZ complex in early mitosis, and that inhibition is relieved by recruitment of dynein to the kinetochore (Cheerambathur et al., 2013). The RZZ complex and dynein show dynamic localization to kinetochores, with RZZ localization being highest at nuclear envelope breakdown, and dynein localization being highest later in prometaphase (Itoh et al., 2018). In this way, RZZ-mediated inhibition of Ndc80 likely prevents the formation of stable end-on attachments early in mitosis when there is a high frequency of incorrect kinetochore–microtubule interactions (Cheerambathur et al., 2013), such as syntelic attachments – where both sister kinetochores are attached to the same pole or merotelic attachments – where a single kinetochore attaches to microtubules from both spindle poles (see poster).

In addition to regulation by RZZ, Ndc80 is also negatively regulated by phosphorylation by mitotic kinases, including Aurora A and Aurora B (Cheeseman et al., 2002; Chmatal et al., 2015; DeLuca et al., 2006; Shrestha et al., 2017; Ye et al., 2015). Aurora-mediated phosphorylation decreases the affinity of Ndc80 for microtubules (Cheeseman et al., 2006), thereby allowing for correction of aberrant kinetochore–microtubule interactions. Consistent with this, Ndc80 phosphorylation is highest in prometaphase and decreases in metaphase through the combined actions of two phosphatases, PP1 and the PP2A-B56 holoenzyme (Liu et al., 2010; Posch et al., 2010; Schleicher et al., 2017). However, recent work has also suggested that a subset of phosphorylation sites are maintained by Aurora A throughout mitosis to ensure proper microtubule dynamics (DeLuca et al., 2017). Thus, the proper segregation of the DNA relies on precise control of the microtubule-binding activity of the KMN network.

During mitosis, the kinetochore needs to not only establish end-on microtubule attachments, but also maintain those attachments while the microtubule grows and shrinks. Indeed, associations with dynamic microtubules contribute to multiple aspects of mitosis. For example, in addition to motor-driven chromosome congression, as discussed above, depolymerization-coupled pulling on kinetochores also contributes to chromosome congression and relies on the ability of the kinetochore to maintain its association with a depolymerizing microtubule (Auckland and McAinsh, 2015). Additionally, chromosomes undergo oscillations during metaphase (Jaqaman et al., 2010; Skibbens et al., 1993), thereby requiring one sister chromatid to associate with depolymerizing microtubules while the other sister chromatid associates with elongating microtubules. Finally, chromosome segregation during anaphase is driven by the association of kinetochores with depolymerizing microtubules (Musacchio and Desai, 2017).

Although critically required for the formation of stable microtubule attachments, the role of Ndc80 in microtubule tip tracking is less clear. In vitro, Ndc80 complexes are unable to track depolymerizing microtubules (Schmidt et al., 2012) unless the complex is artificially oligomerized (McIntosh et al., 2008; Powers et al., 2009; Volkov et al., 2018). It is unclear how well this oligomerization mimics the organization of the ∼14 Ndc80 complexes that bind to each kinetochore microtubule in human cells (Suzuki et al., 2015), but it suggests that Ndc80 contributes to the associations with dynamic microtubules. Indeed, the phosphorylation state of the Ndc80 complex tunes the association of kinetochores with elongating microtubules in vivo (Long et al., 2017).

In fungi, the ring-like Dam1 complex facilitates processive microtubule interactions by binding both Ndc80 and the microtubule, and sliding along the microtubule as protofilaments peel away during depolymerization (Grishchuk et al., 2008; Lampert et al., 2010; Miranda et al., 2005; Tien et al., 2010; Westermann et al., 2005). Interestingly, this mechanism is not widely conserved, as metazoans do not contain homologs of the Dam1 complex (van Hooff et al., 2017). Instead, in species lacking the Dam1 complex, the spindle and kinetochore-associated protein 1 (Ska1) complex likely serves as a functional analog (Gaitanos et al., 2009; Welburn et al., 2009), although the Ska1 complex does not form a ring-like structure (Monda et al., 2017; Schmidt et al., 2012; Welburn et al., 2009).

Ska1 is recruited to the kinetochore by the Ndc80 complex (Cheerambathur et al., 2017; Janczyk et al., 2017), and the phosphatases PP1 and PP2A also promote accumulation of Ska1 at aligned kinetochores (Sivakumar and Gorbsky, 2017) (see poster). Indeed, the levels of kinetochore-localized Ska1 increase throughout congression, thereby enhancing the ability of mature kinetochore–microtubule attachments to sustain load-bearing forces (Auckland et al., 2017). In vitro, Ska1 autonomously tracks both depolymerizing and elongating microtubules (Monda et al., 2017; Schmidt et al., 2012), supporting the model that Ska1 contributes to the ability of the kinetochore to generally associate with dynamic microtubules (Helgeson et al., 2018).

In addition to Ska1, other factors may contribute to dynamic microtubule interactions. For example, CENP-F also tracks depolymerizing microtubules in vitro (Volkov et al., 2015). Taken together, the integrated microtubule-binding activities of Ska1, the Ndc80 complex and perhaps other proteins create the dynamic interface required for persistent association with microtubules (see poster).

Cycles of microtubule polymerization and depolymerization generate forces that can be transmitted through the kinetochore to drive chromosome movement. Withstanding and harnessing that force requires the kinetochore–microtubule attachment to be sufficiently robust and stable. Although the Ndc80 complex is critical for the formation of microtubule attachments (Cheeseman et al., 2006; DeLuca et al., 2006), other microtubule-binding proteins play key roles in stabilizing and strengthening those microtubule interactions. First, the microtubule-bound Ska1 complex strengthens Ndc80-mediated microtubule interactions and is capable of bearing load (Cheerambathur et al., 2017; Helgeson et al., 2018), in addition to its role in tracking dynamic microtubules (Monda et al., 2017; Schmidt et al., 2012). Second, the Astrin–SKAP complex specifically localizes to microtubule-attached and bi-oriented kinetochores, where it binds to microtubules synergistically with Ndc80 (Kern et al., 2017) (see poster). Thus, the activities of Ndc80, Ska1, Astrin–SKAP and potentially other proteins, create a stable kinetochore–microtubule interface that is capable of withstanding the significant force exerted by the mitotic spindle.

In most organisms, each kinetochore binds a bundle of multiple microtubules (∼17 in human cells; McEwen et al., 2001; Wendell et al., 1993), collectively referred to as a kinetochore fiber (denoted k-fiber). In contrast to other populations of mitotic microtubules, k-fibers are uniquely stable, as evidenced by their persistence after cold treatment (Salmon and Begg, 1980). This enhanced stability is the result of their plus-ends being embedded within kinetochores, and the cross-linking and bundling of adjacent microtubules. The complex comprising transforming complex acidic coiled-coil-containing protein 3 (TACC3), colonic and hepatic tumor overexpressed gene protein (ch-TOG; also known as CKAP5) and clathrin, stabilized by phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit α (PI3K-C2α; encoded by PIK3C2A) (Gulluni et al., 2017), crosslinks microtubules and contributes to the organization of the mitotic spindle (Booth et al., 2011; Nixon et al., 2017). K-fiber stabilization is also achieved through the microtubule-bundling activity of hepatoma up-regulated protein (HURP; also known as DLGAP5) (Koffa et al., 2006; Silljé et al., 2006; Wong and Fang, 2006) (see poster). Together, the stabilization and organization of k-fibers by these proteins ultimately allows for the generation of sufficient force to drive accurate chromosome segregation.

Chromosome congression, bi-orientation and segregation depend on precise regulation of the stability and dynamics of the microtubules in the mitotic spindle. This balance is achieved though the combined actions of numerous proteins that both positively and negatively regulate microtubule growth through diverse mechanisms. Regulators that localize at or near kinetochores include the kinesin-like protein Kif18A (Mayr et al., 2007), mitotic centromere-associated kinesin (MCAK; also known as KIF2C) (Wordeman and Mitchison, 1995), ch-TOG (Gergely et al., 2003), cytoplasmic linker protein 170 (CLIP-170; also known as CLIP1) (Dujardin et al., 1998), and the CLIP-associating protein (CLASP) (Maiato et al., 2003) and end-binding (EB or MAPRE) (Juwana et al., 1999) families of microtubule-binding proteins (see poster). Metaphase chromosome oscillations are regulated by the kinetochore-localized kinesin Kif18A, which suppresses microtubule dynamics (Du et al., 2010; Stumpff et al., 2008, 2012). At anaphase onset, Kif18A and other factors must be dephosphorylated to allow for a switch from oscillations to robust poleward movement of the separated sister chromatids (Su et al., 2016). MCAK is a kinesin-13 family member (Lawrence et al., 2004) and a microtubule depolymerase (Desai et al., 1999; Hunter et al., 2003). By promoting microtubule depolymerization at kinetochores that are not yet bi-oriented and stably associated with the microtubule, MCAK facilitates the correction of erroneous kinetochore–microtubule attachments and thereby increase the likelihood of achieving bi-orientation (Kline-Smith et al., 2004). The activity of MCAK is counteracted by ch-TOG, a highly processive microtubule polymerase (Brouhard et al., 2008). CLIP-170 and CLASPs also promote microtubule growth, with CLIP-170 promoting the transition from microtubule depolymerization to microtubule polymerization (Komarova et al., 2002) and CLASPs promoting the incorporation of tubulin subunits into k-fibers (Maiato et al., 2005). The EB family of proteins, comprising EB1, EB2 and EB3 (also known as MAPRE1, MAPRE2 and MAPRE3, respectively) in mammals (Su and Qi, 2001), are autonomous plus-end-tracking proteins (Bieling et al., 2007) that recruit a variety of other proteins that contain cytoskeleton-associated proteins (CAP)-Gly domains or SxIP motifs (Kumar and Wittmann, 2012) to affect microtubule growth or organization (Browning et al., 2003; Komarova et al., 2005; Mimori-Kiyosue et al., 2005; Niethammer et al., 2007; Su et al., 1995) (see poster).

The functions of these regulators of microtubule dynamics are also subject to regulatory control themselves. For example, MCAK is phosphorylated by numerous mitotic kinases (Andrews et al., 2004; Lan et al., 2004; Ohi et al., 2004; Sanhaji et al., 2010; Zhang et al., 2007, 2011). Recent work has demonstrated that CDK1-mediated phosphorylation of a single threonine residue in the MCAK motor domain is sufficient to block the ability of MCAK to distinguish the end of the microtubule from the lattice, and thereby reduce microtubule depolymerization (Belsham and Friel, 2017). Additionally, MCAK undergoes significant structural rearrangements upon binding a microtubule that allow for optimal depolymerase activity (Burns et al., 2014; Ems-McClung et al., 2013; Talapatra et al., 2015). Collectively, these diverse microtubule-associated proteins facilitate cell division by ensuring a dynamic mitotic spindle.

During every cell division, numerous kinetochore-localized microtubule-binding proteins must act to ensure the accurate segregation of the DNA to the daughter cells. A stable, bipolar spindle must be built, the kinetochore of each sister chromatid must attach to microtubules emanating from one of the spindle poles, and each kinetochore must be capable of maintaining its microtubule attachment, despite elongation and depolymerization of the microtubule. Recent work has defined numerous molecular players that are involved in each of these key aspects of mitosis. Given the complexity of the kinetochore, it is perhaps unsurprising that many of these proteins have opposing activities: dynein and CENP-E transport chromosomes in opposite directions, Astrin–SKAP must stabilize the kinetochore–microtubule attachments yet also allow for dynamic interactions and Ska1-mediated tip-tracking, and MCAK destabilizes microtubules, whereas ch-TOG, CLIP-170 and CLASPs promote microtubule growth. A key goal for future studies will therefore be to uncover how the diverse activities of these molecules are integrated to ensure faithful chromosome segregation.

We thank Leah Bury, Nolan Maier, Lily McKay and Gunter Sissoko for critically reading and providing comments on this manuscript. We apologize to the many authors whose work we could not include due to space restrictions.