Molecular mechanisms of kinesin-14 motors in spindle assembly and chromosome segregation Free

One of the most fundamental activities of life is the reproduction and the propagation of cells, which depends on accurate cell division. During cell division, the genetic material of the cell is duplicated and faithfully segregated into two daughter cells. This process occurs in virtually all cells, from prokaryotes to higher eukaryotes but excluding many terminally differentiated cell types (Brust-Mascher and Scholey, 2011; Holland and Cleveland, 2009). In eukaryotic cells, division follows mitosis, which includes spindle formation and nuclear envelope breakdown in higher eukaryotes− except yeast, which undergo closed mitosis (Güttinger et al., 2009). Once the interactions between microtubules within spindle and chromosomes are established, the chromosomes congress at the metaphase equatorial plate and, ultimately, move towards the opposite spindle poles (Güttinger et al., 2009; Pavin and Tolić, 2016; Tanaka and Desai, 2008). The genomic stability of the eukaryotic cell relies on error-free segregation of chromosomes during mitosis and meiosis (Aguilera and Gómez-González, 2008). Accurate segregation of chromosomes is fulfilled by two complementary mechanisms: chromosome-to-pole movements and elongation of the mitotic spindles (Brust-Mascher and Scholey, 2011; Brust-Mascher et al., 2015). Finally, the nuclear envelope is reformed around chromatin and the cytoplasm is partitioned to form two daughter cells during cytokinesis (Schellhaus et al., 2016).

Eukaryotes have evolved a specialized microtubule cytoskeleton for chromosome segregation that is distinct from bacterial chromosome segregation (Nogales et al., 1998; Dye and Shapiro, 2007; Scholey et al., 2003). In eukaryotes, specialized tubulin proteins assemble into polar microtubules (Nogales, 2000). Here, ensembles of cytoskeletal proteins at microtubule minus- (Wiese and Zheng, 2006) and plus- (Akhmanova and Steinmetz, 2008) ends together with multiple microtubule-associated proteins (MAPs) (Marx et al., 2006) and motile molecular motors (Lawrence et al., 2004; Vicente and Wordeman, 2015) act cooperatively to regulate mitosis with high fidelity (see Box 1). Molecular motor kinesins and dynein exert complex roles to regulate cytoskeleton dynamics, spindle morphogenesis and chromosome movement (Schliwa and Woehlke, 2003; Vicente and Wordeman, 2015). The discovery of microtubule-associated kinesin motors (Vale et al., 1985a,b) and their roles as crucial regulators of microtubule organization and chromosome segregation during the cell cycle (Gatlin and Bloom, 2010; Scholey et al., 1985; Wordeman, 2010) was the beginning of a new era in our understanding of the molecular mechanisms of cell division.

Generally speaking, the kinesin superfamily includes conventional kinesins as well as kinesin-like proteins that were discovered later and include the so-called N-terminal motor and C-terminal kinesin motor proteins (Vale and Fletterick, 1997); however, the morphological complexity of kinesins is much greater (Cross and McAinsh, 2014; Hirokawa et al., 2009; Lawrence et al., 2004; Wordeman, 2010). In contrast to plus-end-directed kinesins, kinesin-14 motors are specific minus-end-directed motors; they utilize the chemical energy of ATP hydrolysis to move along microtubules from their plus- to their minus-end (Friel and Howard, 2012; Schnitzer and Block, 1997; Shimizu et al., 1995). The kinesin-14 family is ubiquitous to all eukaryotes, with typically two to three members present that are different in their structure and function (Olmsted et al., 2015; Simeonov et al., 2009).

In this Commentary, we discuss the common and distinct roles of kinesin-14 family motors in model organisms. We focus on microtubule-associated functions, such as non-processive movement on microtubules and roles in microtubule crosslinking or sliding. We shed light on the molecular interactions of kinesin-14 members and their associated interaction factors, such as γ-tubulin, the plus-end microtubule protein EB1 (also known as MAPRE1 in humans) and the nuclear import machinery. We will also summarize the mutual antagonisms between kinesin-14 and kinesin-5 family members during spindle assembly, and discuss new roles of kinesin-14/-5 motors in microtubule nucleation. We, therefore, aim to provide a comprehensive summary of the cellular functions of kinesin-14 family members in various processes, such as microtubule nucleation, microtubule focusing, mitotic spindle organization and chromosome segregation during mitotic and meiotic events.

In 1929, Alfred Henry Sturtevant observed for the first time defective chromosome segregation in the claret non-disjunctional (ncd) mutant of Drosophila simulans (Sturtevant, 1929; Davis, 1969). Subsequently, this locus was cloned and found to encode a mitotic/meiotic kinesin in Drosophila (Endow et al., 1990; McDonald and Goldstein, 1990). The researchers revealed that Drosophila kinesin-14 Ncd is unusual as it moves from microtubule plus-ends towards the minus-ends in motility assays (McDonald et al., 1990; Walker et al., 1990). In budding yeast, another kinesin-14, Kar3, was independently identified as a motor crucial for nuclear fusion and meiosis by using genetic approaches (Meluh and Rose, 1990).

The idea of a standardized kinesin nomenclature (Lawrence et al., 2004) builds on other lineage studies of kinesins (Dagenbach and Endow, 2004; Kim and Endow, 2000; Lawrence et al., 2002; Miki et al., 2001; Moore and Endow, 1996) and categorizes the kinesin superfamily into 14 families with some outliers. The kinesin-14 family consists of two subfamilies kinesin14-A and kinesin14-B, although not all their members have so far been included in phylogenetic studies. Within these subfamilies further functional specification of kinesin-14 family members is evident beyond this simple division (Olmsted et al., 2015). In this Commentary, we focus on the best-studied kinesins, mostly members of the kinesin-14A family. To distinguish the different species, we refer to these kinesin-14 motors as Homo sapiens HSET (officially known as KIFC1), Mus musculus KIFC1, Xenopus laevis XCTK2 (officially known as Kifc1), Drosophila melanogaster Ncd, Arabidposis thaliana ATK5, Schizosaccharomyces pombe Pkl1 and Klp2, and Saccharomyces cerevisiae Kar3. Members of the kinesin-14B family (kinesin family members C2 and C3), such as H. sapiens KIFC2 and KIFC3, respectively; A. thaliana KatD (KIN14G) and the plant KCBP proteins (Lawrence et al., 2002; Wickstead and Gull, 2006) have been less-well studied. Moreover, a functional complexity of kinesins has been predicted in recent computational studies, supporting a kinesin diversity that reflects the complexity of the microtubule cytoskeleton (Wickstead et al., 2010).

The kinesin-14 proteins comprise three functional domains, an N-terminal tail domain, a central coiled-coil stalk domain and a C-terminal motor domain (Cross and McAinsh, 2014; Vale and Fletterick, 1997) (Fig. 1A). The globular motor domain contains the ‘catalytic core’ for ATP hydrolysis and the microtubule-binding site, both of which generate forces for mechanical movements and motor processivity (Friel and Howard, 2012; Hirokawa and Noda, 2008; Schnitzer and Block, 1997). The short and conserved neck region proximal to the catalytic domain functions as a mechanical transducer to regulate directionality (Case et al., 1997; Yamagishi et al., 2016). In Drosophila, the neck region at the proximal end of the coiled-coil stalk domain, adjacent to the motor, is essential for the regulation of lever-like rotations and step size (Hallen et al., 2011). Moreover, biochemical studies of Ncd stalk peptides (Ito et al., 2006) and additional studies (Makino et al., 2007) indicate the presence of sequence features that generate reversible and irreversible regions within the Ncd coiled-coil region. It has been proposed that a partial collapse of the Ncd stalk prevents braking that can occur when multiple Ncd motors are bound to a microtubule but without synchronized ATP hydrolysis and forcestroke generation (Makino et al., 2007). Several lines of evidence point to the tail domain as providing some of the most crucial functional cues to diversify kinesin-14 function. KIFC5A has an additional ATP-independent microtubule-binding domain (Zhang and Sperry, 2004) and Ncd has a positively charged sequence in its tail domain that tethers it to the C-terminal E-hook of tubulin (Furuta and Toyoshima, 2008). Mouse KIFC1 has tail domain sequences that target it to membrane-bound organelles (Zhang and Sperry, 2004); the tail domain of Pkl1 direct it to γ-tubulin (Olmsted et al., 2013, 2014). Additional studies in which chimeric kinesin-14 proteins of human HSET, Drosophila Ncd and fission yeast Pkl1 were generated also support this paradigm and confirm that tail domain sequences orchestrate function (Simeonov et al., 2009).

Kinesin motors are highly processive motors that convert the chemical energy of ATP hydrolysis into long-range directed motility traversing more than 50 steps. They undergo multiple catalytic cycles along microtubules before their detachment and release (Asbury et al., 2003; Yildiz et al., 2004). In a study of human kinesin, in which one of the heads had been mutated, alternating fast and slow stepping of the heterodimeric kinesin was observed in support of a ‘hand-over-hand’ model versus inchworm movement (Kaseda et al., 2003). In contrast, comparative analysis of the motility of homodimeric Ncd or the heterodimeric kinesin-14 motors Kar3−Vik1 and Kar3−Cik1 indicates that they are non-processive as single motors and only move 20−50 nm, which corresponds to approximately two to six 8-nm steps (Foster and Gilbert, 2000; Zhang et al., 2015). In the same study, a hold-and-release mechanism for kinesin-14 motors was proposed in contrast to the hand-over-hand model established for conventional kinesins (Zhang et al., 2015). To date, three main means used by kinesin-14 motors to generate weakly processive movements have been described (Fig. 1B). First, the non-motile microtubule-binding site at the tail domain increases the affinity to the microtubule lattices and, thus, increases both the dwell time and the ATP-independent diffusion steps of kinesin-14 motors on the microtubules. This is referred to as the ‘tail-dependent diffusion manner’ that allows gliding of the tail region along the microtubule due to the thermal fluctuation and the weak interactions that prevent the tail from sticking to and dissociating from the microtubules (Fink et al., 2009). Second, kinesin-14 motors can work in groups to stimulate their processivity along the microtubules. For example, two coupled Ncd motors can continually walk along the microtubule for more than 1 μm (Furuta and Toyoshima, 2008; Furuta et al., 2013). A single kinesin-14 Pkl1 is not highly processive and shows one-dimensional diffusion along the microtubules. However, several Pkl1 motors (approximately ten molecules) can work in a cooperative manner to move along the microtubules for several stepping cycles (Furuta et al., 2008) (Fig. 1B). Third, the interactions with partner proteins can also stimulate their processivity. For example, the budding yeast kinesin-14 Kar3 generates processive movement through its interactions with the non-motor proteins Cik1 or Vik1 (Manning et al., 1999; Mieck et al., 2015) as demonstrated by deletion of the non-catalytic domain of Cik1, resulting in increased diffusion of the Kar3−Cik1 heterodimer away from the microtubule. Velocity differences were also noticed dependent on the heterodimeric partner.

The kinesin-14 motors are found in dimeric form, and, in most species, the majority are homodimers (Fig. 1A), with the exception of several yeast species, for instance the heterodimers Kar3-Cik1 and Kar3-Vik1 in budding yeast (Barrett et al., 2000; Duan et al., 2012) and in Candida glabrata (Joshi et al., 2013).

Kinesin-14 motors display a slow velocity towards the microtubule minus-end compared with the plus-end-directed conventional kinesin (42.7±2.19 nm/s) (Endow and Waligora, 1998). For example, a single Ncd−Ncd dimer has a velocity of 15.2±0.3 nm/s (Endow and Waligora, 1998), single Pkl1−Pkl1 homodimer a velocity of 33±9 nm/s (Furuta et al., 2008) and Kar3−Cik a velocity of 77±23 nm/s (Mieck et al., 2015). The adjacent neck-motor junction is required for kinesin-14 Ncd minus-end movement (Endow and Waligora, 1998). The swinging motion of the C-terminal ‘neck mimic’ (especially the conserved AxxVNxT/C residues; in which x represents any amino acid) and the ATPase-dependent conformational changes of Ncd also yield a directional bias and movements towards the minus-end (Yamagishi et al., 2016). Kinesin-14 motors also utilize minus-end-directed forces and ATP-mediated powerstrokes to crosslink microtubules or to slide antiparallel microtubules (Fig. 1C). In addition, ADP release is slow in Ncd and, thus, is a rate-limiting step in its ATPase cycle (Foster and Gilbert, 2000).

According to the conventional powerstroke mechanism (i.e. one kinesin head results in turnover of one ATP molecule) for the motility of a single kinesin-14 motor (Endres et al., 2006; Gonzalez et al., 2013; Wendt et al., 2002), only one motor head needs to interact with microtubules and turnover of only one ATP molecule is necessary for the powerstroke, which combines both rotation and ATP turnover to slide the microtubules (Chen et al., 2012). However, a new ATP-promoted powerstroke hypothesis that implies two kinesin heads and turnover of two ATP molecules has been proposed for heterodimeric Kar3−Vik1 and Kar3−Cik1, as well as homodimeric Ncd−Ncd (Zhang et al., 2015). The interactions between both Ncd heads and microtubules depend on the nucleotide state but interactions between either Vik1 or Cik1 and microtubules are regulated by strain (Zhang et al., 2015). Furthermore, turnover of two ATP molecules is required for interaction with microtubules as well as force generation in the case of the kinesin-14 motors that bind to the adjacent microtubule lattices (Endres et al., 2006; Gonzalez et al., 2013; Kocik et al., 2009; Zhang et al., 2015).

For microtubule-organizing functions, such as crosslinking of parallel microtubules, sliding of antiparallel microtubules and focusing of the spindle pole, both the motor and tail domains of kinesin-14 motors are required (Fig. 1C) (Wendt et al., 2002). An exception to this is Pkl1, for which the tail domain was shown to be able to regulate nucleation without the motor domain in vitro and in vivo (Fig. 1B), and to do so in yeast and human cells (Olmsted et al., 2013, 2014). Microtubule-dependent D. melanogaster Ncd motors have been shown to localize to parallel microtubules where they exert their movement into the opposite direction, thus creating a ‘tug-of-war’ effect. This results in static crosslinking and bundling of parallel microtubules. However, Ncd can also promote efficient sliding of microtubules that have opposite polarity through the so-called ‘directional sliding’ mechanism (Fink et al., 2009) (Fig. 1C). Furthermore, in case of the X. laevis XCTK2, it has been demonstrated that, if the number of kinesins bound to microtubules is too high, gliding speed is significantly reduced, presumably owing to the non-synchronized state of multiple bound motors (Hentrich and Surrey, 2010). As mentioned previously, the presence of reversibly stable regions that can collapse in the coiled-coil domain (Hallen et al., 2008, 2011; Ito et al., 2006; Makino et al., 2007) might be able to partially compensate for any braking that results from unsynchronized power force generation.

During mitotic spindle assembly, a subset of kinesin-14 members appears to exert two functions: organization of microtubule minus-ends at poles and/or microtubule nucleation, both of which are mediated by direct interactions of kinesin-14 members with components of the microtubule organizing center (MTOC) and the γ-tubulin ring complex (γ-TuRC), although with different outcomes.

During spindle formation in HeLa cells, the γ-TuRC is initially recruited to the pole-distal regions and then moves towards microtubule minus-ends around the spindle poles. This process is mediated by cooperation between kinesin-14 HSET, kinesin-5 Eg5 (officially known as KIF11 in mammalian cells) and dynein (Lecland and Lüders, 2014). Both H. sapiens HSET and X. laevis XCTK2 affect spindle morphology. RNA interference (RNAi) of HSET results in broader spindles and less focused poles but without any changes in duration of mitosis or fidelity of chromosome segregation (Cai et al., 2009b). In mouse oocytes, the broader spindles and poles are more evident (Mountain et al., 1999) and, additionally, it has been shown that HSET can cluster centrosomes (Kwon et al., 2008).

In S. pombe, Pkl1 physically interacts with γ-TuRC through both the motor and the tail domain, and suppresses microtubule nucleation to influence spindle structure and function at γ-TuRC (Rodriguez et al., 2008, Olmsted et al., 2013, 2014). In its role at γ-TuRC to regulate nucleation, Pkl1 does not absolutely require its motor domain or stalk region, and the tail can be reduced to a 30 amino-acid-long region and still function (Olmsted et al., 2013, 2014). Genetic analysis supports this observation as synthetic interactions have been observed between Pkl1 and the γ-TuRC proteins Alp4 and Alp6 but not with other fission yeast mitotic motors, and there is no genetic interaction between the second fission yeast kinesin-14 Klp2 and any γ-TuRC proteins (Tange et al., 2004).

In D. melanogaster oocytes, minus-end-directed motility of Ncd is essential for focusing the spindle pole in meiosis I spindles that lack γ-tubulin. During meiosis,γ-tubulin interacts with Ncd to form the microtubule nucleating center at the spindle pole body (Endow and Komma, 1998).

The function of spindle pole organization and microtubule nucleation does not need to be distinct, and some kinesin-14 members, such as HSET, might possess both capabilities of spindle pole organization and microtubule nucleation (Walczak et al., 1997; Simeonov et al., 2009). It is worth noting that in meiotic cells and in somatic cells, spindles can nucleate around chromosomes in the presence of a Ran gradient and this may explain in part the evolutionary need for kinesin-14 members to have dual roles with γ-tubulin complexes at the different Ran-GTP concentrations (Khodjakov et al., 2000; Khodjakov and Rieder, 2001).

Surprisingly, minus-end-directed kinesins-14 are sometimes found at microtubule plus-ends (Ambrose et al., 2005; Braun et al., 2013; Maddox et al., 2003; Sproul et al., 2005); this is made possible through their association with plus-end-binding proteins. Microtubule plus-ends serve as sites crucial for microtubule capture, stabilization, spindle orientation (Akhmanova and Steinmetz, 2008; Howard and Hyman, 2003; Lee et al., 2000) and anchoring to specific cellular organelles (Vaughan et al., 2002) or to the cell cortex (Lansbergen and Akhmanova, 2006; Mimori-Kiyosue et al., 2005) (see Box 2).

In vitro microtubule-gliding assays demonstrated that H. sapiens HSET accumulates at the plus-ends of microtubules through direct interactions with EB1 (Braun et al., 2013) (Fig. 1B). Similarly, in the absence of EB1, D. melanogaster Ncd does not accumulate at either end of antiparallel microtubules (Fink et al., 2009). The tip-tracking property of Ncd is regulated by two concurrent mechanisms: first, the direct binding of Ncd to microtubule-bound EB1 and, second, the interactions of the Ncd tail domain with a composite site that is generated by EB1 and the microtubule surface (Szczęsna and Kasprzak, 2016). In S. pombe, kinesin-14 Klp2 is localized to the microtubule plus-ends through interaction of two of its SxIP motifs with Mal3 (a yeast EB1 homologue) (Mana-Capelli et al., 2012).

Taken together, these studies suggest that the EB1-dependent plus-end tracking, which occurs in diverse organisms, is a general mechanism for several members of the kinesin-14 family. An exception is S. cerevisiae Kar3−Vik1 that does not accumulate at microtubule plus-ends but, instead, is found along the microtubule lattice (Allingham et al., 2007; Maddox et al., 2000; Sproul et al., 2005).

A number of kinesins-14, including H. sapiens HSET (Cai et al., 2009b), R. norvegicus KIFC1 (Yang and Sperry, 2003), X. laevis XCTK2 (Cai et al., 2009b), D. melanogaster Ncd (Goshima and Vale, 2005) and S. pombe Klp2 (Troxell et al., 2001) contain a nuclear import signal (NLS) and show a nuclear localization pattern during interphase. This localization is achieved through interactions between their bipartite NLS at the tail domain with importin-α or -β and nuclear import via a Ran-GTP/GDP-mediated pathway (Cai et al., 2009b).

In Xenopus egg extracts, the small GTPase Ran-GTP mediates the interactions of importin-α or -β and XCTK2 through its bipartite NLS (Clarke and Zhang, 2008; Ems-McClung et al., 2004). When the physical Ran-GTP gradient gradually diminishes from the chromosomes to the spindle pole (Kalab et al., 2002), importin α/β competitively binds to XCTK2, reducing its spindle anchoring and increasing its turnover kinetics during spindle formation (Weaver et al., 2015).

In addition, S. pombe Pkl1 is only a mitotic motor, whereas Klp2 has a role in both the mitotic phase and interphase (Carazo-Salas and Nurse, 2006; Daga et al., 2006; Troxell et al., 2001). Moreover, as a heterodimer, the functions of S. cerevisiae Kar3 and its localization depends on its binding partner. During mitosis, Cik1 selectively dimerizes with Kar3 and is targeted either to the spindle microtubules (Hepperla et al., 2014) or to the microtubule plus-ends (Sproul et al., 2005), whereas, in interphase, Kar3−Cik1 localizes to the nucleus (Manning et al., 1999). By contrast, Vik1 is required for the localization of Kar3 at the poles of the mitotic spindle during mitosis (Manning et al., 1999).

In summary, the interaction between kinesin-14 members and microtubule minus-ends, microtubule plus-ends and partner proteins can also influence their subcellular localizations and cellular functions, which are crucial in fulfilling their functions and ensure the fidelity of cell division (Yount et al., 2015).

The bipolar spindle apparatus is maintained by antagonistic relationships. Accumulating evidence in diverse organisms reveals that on anti-parallel microtubules, a ‘tug-of-war’ takes place between kinesin-14 and kinesin-5 in order to crosslink and slide microtubules, and is a common mechanism to maintain correct spindle organization, as discussed in detail below (Brust-Mascher and Scholey, 2011; Mountain et al., 1999; Sharp et al., 2000) (Fig. 2). Whereas kinesin-14 motors are dimers, kinesin-5 forms a dumbbell-shaped homotetramer with two pairs of motor domains positioned at the opposite ends (Hentrich and Surrey, 2010; Kashina et al., 1996; Sawin et al., 1992). The kinesin-5 tetramer utilizes a combination of four motor-domain-binding and four nonmotor microtubule-binding sites for efficient microtubule crosslinking and sliding (Kapitein et al., 2008; Kaseda et al., 2009; Kwok et al., 2006; Weinger et al., 2011).

The antagonism between kinesin-14 and kinesin-5 also occurs on parallel spindle microtubules. First, kinesin-14 motors have the intrinsic ability to focus the minus-ends of microtubules into spindle poles, but kinesin-5 motors counteract focusing of the microtubule plus-ends at the poles (Hentrich and Surrey, 2010). Second, kinesin-14 motors generate inward pulling forces on spindles, whereas the kinesin-5 motors have been shown to generate outwards forces on spindles in vitro (Yukawa et al., 2015). Another mechanism by which kinesin-14 and kinesin-5 counterbalance their activities does not require the motor domains but is mediated primarily by the tail, i.e. the regulation of microtubule nucleation by γ-TuRC, as discussed below (Olmsted et al., 2013, 2014) (Fig. 2).

It is worth pointing out that only kinesin-14 is able to focus spindle poles, but not kinesin-5. In vitro self-organization experiments indicate that X. laevis XCTK2 is able to focus the minus-ends of parallel microtubules to asters without the need for additional factors but kinesin-5 Eg5 is unable to focus the microtubule minus-ends. Additionally, Eg5 antagonizes XCTK2 in focusing of the spindle pole, which slows down the accumulation of XCTK2. Together, this results in the formation of microtubule asters through an interconnected microtubule network and prevents pole formation by the plus-ends of parallel microtubules during spindle assembly (Hentrich and Surrey, 2010) (Fig. 2B).

A tug-of-war between kinesin-14 and kinesin-5 motors is a common mechanism to maintain correct spindle organization in diverse organisms. For example, inhibition of HSET leads to the disruption of microtubule aster assembly but when kinesin-5 Eg5 is also inhibited, microtubule aster formation, centrosome separation and spindle organization are all restored. Therefore, a balance between HSET and Eg5 is responsible for microtubule aster assembly both in vitro and in vivo (Mountain et al., 1999; Kim and Song, 2013) (Fig. 2A). Directional instability, whereby antiparallel microtubules undergo micrometer-range movements back and forth owing to the forces mediated by other kinesins, rather than a stable balance of forces, is exerted by XCTK2 and Eg5 on antiparallel microtubules due to the asymmetric roles of these antagonistic motors (Hentrich and Surrey, 2010) (Fig. 2B). This implies that other motors and spindle mechanisms can work together to form spindles when the kinesin-14–kinesin-5 pair is removed (Mountain et al., 1999). This notion is supported by an observation in fission yeast in which the kinesin-5 Cut7 that was thought to be essential (Hagan and Yanagida, 1999) is – together with Pkl1 – dispensable (Olmsted et al., 2013, 2014).

In fly, the classic ‘force-balance’ model for bipolar mitotic spindle formation suggests that spindles are maintained by the balance between outwards- and inward-sliding forces that are generated by kinesin-5 (KLP61F) and kinesin-14 motors (Ncd), respectively (Fink et al., 2009). The force-balance model gains supports from the findings that both Ncd and KLP61F can crosslink antiparallel microtubules and slide them to the opposite directions in competitive microtubule gliding experiments (Fink et al., 2009; Oladipo et al., 2007; Tao et al., 2006) (Fig. 2C).

In addition, an increase in the ratio between kinesin-14 and kinesin-5 motors can give rise to a scenario, whereby passive resistance slows down the velocity of the advantage motors and that of their relative movements to each other compared to prior, balanced steady-state. This has been raised in the so-called ‘motor protein friction’ model for competitive motility (Tao et al., 2006). Kinesin-14 and kinesin-5 family motors balance sliding but sliding is not a continuous process and results in episodes of unstable, oscillatory swings versus balanced braking and sliding by two motors (Tao et al., 2006). However, this motor protein friction model does not provide a good fit for the phenomenon of a balance point in the force-balance model (Fig. 2C).

More recently, a new, ‘fully stochastic force-balance’ model’ for prometaphase spindles in vitro and in vivo has been proposed (Civelekoglu-Scholey et al., 2010). This model incorporates microtubule-motor kinetics and dynamic stability of interpolar microtubules, and appears to fit the experimental and theoretical data better than the previous motor protein friction model. In this model, kinesin-14 Ncd and kinesin-5 KLP61F act synergistically, thereby providing opposing powerstrokes with load-dependent detachment, whereby the motors exert a constant force when they are loaded on microtubules and maintain the correct spacing between the poles. Furthermore, a specific ratio between Ncd and KLP61F is required for maintaining the steady-state of prometaphase spindles (Brust-Mascher et al., 2009; Civelekoglu-Scholey et al., 2010; Kwon et al., 2004; Sharp et al., 2000) (Fig. 2C).

In centrosome-controlled Drosophila embryo spindles, kinesin-5 KLP61F crosslinks and slides anti-parallel microtubules to mediate the poleward flux of interpolar and kinetochore microtubules; this is antagonized by kinesin-14 Ncd through mediating spindle collapses (Brust-Mascher et al., 2004; Brust-Mascher et al., 2009). In anastral (i.e. mitotic spindle formation without centrioles) X. laevis egg extracts, kinesin-5 Eg5 slides microtubules to poles and contributes to poleward flux and spindle length control (Miyamoto et al., 2004; Peterman and Scholey, 2009).

In S. pombe, Pkl1 has been shown to directly block microtubule nucleation at γ-TuRC, while the kinesin-5 Cut7 binds to γ-tubulin through its motor and BimC domains; this opposes the effect of Pkl1 in inhibiting microtubule nucleation. In this case, the microtubule-binding domain of Pkl1 is not required for the regulation of microtubule nucleation (Olmsted et al., 2014) (Fig. 2D).

Another study indicated that Pkl1 cooperates with the spindle-pole-associated protein mitotic spindle disanchored 1 (Msd1) to interact with γ-TuRC, before focusing the minus-ends of microtubules at the spindle pole body (Syrovatkina and Tran, 2015). Here, the deletion of pkl1 results in unfocused microtubule minus-ends at the spindle poles and in long microtubule protrusions. This can be rescued by the ablation of cut7 (Syrovatkina and Tran, 2015) (Fig. 2D). In the mitotic spindle pole body, Pkl1 is required for the localization of Msd1 and the mitosis-specific spindle pole body component Wdr8; it cooperatively interacts with these two proteins to mediate the spindle anchoring, where it counteracts the outward pushing forces that are generated by Cut7 (Yukawa et al., 2015). In addition, Pkl1 also functions as the spindle assembly checkpoint, which regulates chromosome bi-orientation by mediating the spindle pole assembly (Grishchuk et al., 2007). Collectively, there are at least three roles for kinesin-14 Pkl1 in yeast. One is the crosslinking and sliding of microtubules. Its second role is to mediate microtubule nucleation at the spindle pole in fission yeast and its third role is at the checkpoint.

In this section, we discuss the roles of kinesin-14 motors in centrosome clustering within wild-type cells, solid tumors and in hematologic malignancies, in the regulation of spindle length and morphology, as well as in chromosome alignment and segregation (Fig. 3).

In the oocytes of many species – including human, mouse, frog and fly – meiotic spindles organize in the absence of centrosomes (Dumont and Desai, 2012; Szollosi et al., 1972). In mouse oocytes, which lack conventional centrosomes, KIFC1 inhibition results in an aberrant barrel-shaped spindle during metaphase, as well as in broad spindle poles, reduced number of astral microtubules and, finally, arrest during metaphase II of meiosis (Mountain et al., 1999). In X. laevis egg extracts, which also lack centrosomes, XCTK2 is localized at mitotic spindles, crosslinks parallel microtubule and focuses the microtubule minus-ends to mediate mitotic spindle assembly in vitro (Walczak et al., 1997, 1998).

In Drosophila oocytes, the meiosis II spindle forms by reorganization of the meiosis I spindle fibers, rather than by its disassembly or breakdown. Here, Ncd stabilizes the newly nucleated microtubule minus-ends at the spindle poles and also accounts for recruitment of γ-tubulin to the spindle poles (Baumbach et al., 2015; Endow and Komma, 1997). γ-Tubulin and Ncd stabilize the association of the γ-tubulin-recruiting augmin complex with the polar regions of the spindles and facilitate microtubule organization and chromosome congression (i.e. the process of aligning chromosomes on the spindle) in oocytes (Colombié et al., 2013).

In different cell types, the length of the metaphase spindle is typically maintained at a characteristic constant length through sliding and microtubule−microtubule interactions, as well as through microtubule nucleation and dynamics. A constant spindle length is crucial for accurate chromosome-to-microtubule attachment and the fidelity of chromosome separation (Goshima and Scholey, 2010; Goshima et al., 2005b; Hepperla et al., 2014; Sharp et al., 1999; Syrovatkina et al., 2013).

In HeLa cells, ablation of HSET function leads to broader and shorter spindles, with an 18% decrease in distance between the poles compared to that of wild type. Mechanistically, HSET lengthens the spindle by the applying outward forces to slide the microtubules and modulates spindle organization by the crosslinking and sliding of mitotic spindles (Cai et al., 2009b) (Fig. 3A-C; Table 1).

However, other studies indicate that kinesins-14 generate a pull force on spindles. D. melanogaster Ncd is required for the maintenance of spacing between the duplicated centrosomes at prometaphase and metaphase through the pull forces, and also for the positioning of centrosomes at anaphase and telophase (Kwon et al., 2004; Sharp et al., 1999, 2000) (Table 1). Similarly, in A. thaliana, the mitotic spindles of atk5^−/− mutants exhibit greater distances in mean midzone length, splayed spindle poles, the spindle broadening phenotype and an increased duration during prometaphase (Ambrose and Cyr, 2007; Ambrose et al., 2005) (Table 1).

A recent study suggested a new model based on the deletion of Kar3, Cik1 or Vik1 in budding yeast, which results in shorter and severely misaligned spindles, and a dysfunctional spindle midzone at metaphase. Kar3−Cik1 is thought to improve the alignment efficiency of the interpolar microtubules in the midzone. This then allows the plus-end-directed kinesin-5 Cin8 to exert the outward forces on anti-parallel microtubules that are required for correct spindle bipolarity (Hepperla et al., 2014) (Table 1).

Human HSET has three possible roles in chromosome movement and dynamics during mitosis (Fig. 3A-C). The first is that HSET is not essential for the mitotic spindle assembly in cultured cells with centrosomes and that the role of kinesin-14 HSET is substituted by the centrosomes and NuMA (nuclear mitotic apparatus protein). The perturbation of HSET only causes a slightly prolonged prometaphase, which does not show obvious disruptions to characteristic chromosome dynamics, including those of chromosome movement toward spindle equator, congression, oscillation and separation (Gordon et al., 2001; Mountain et al., 1999).

The second possible role for HSET is in the regulation of chromosome alignment and segregation during mitosis (Kim and Song, 2013; Zhu et al., 2005). Depletion of HSET in HeLa epithelial cells or IMR-90 fibroblasts leads to several misaligned chromosomes at the spindle equator and, owing to the formation of multipolar spindles, results in lagging chromosomes, micronuclei, genomic instability in daughter cells (Kim and Song, 2013; Zhu et al., 2005) (Fig. 3B,C).

The third possible role for HSET is to facilitate the correct end-on attachment of kinetochores to K-fibers through a search-and-capture mechanism. This is fulfilled by interactions with regulatory subunits of the B56 family of protein phosphatase 2A (B56-PP2A; officially known as PTPA) or with kinetochore protein NDC80 homolog (NDC80) at the kinetochores in Hela cells under conditions where any end-on kinetochore attachments are lacking (Cai et al., 2009a; Xu et al., 2014).

During mitosis and meiosis in D. melanogaster, kinesin-14 Ncd is necessary for correct chromosome distribution and configuration, and Ncd mutants show separated or scattered chromosomes. During meiosis, nucleation occurs from Ncd-associated microtubule asters to generate spindle-associated bivalent chromosomes (Hatsumi and Endow, 1992; Matthies et al., 1996; Sköld et al., 2005).

Deletion of pkl1 in fission yeast only partially interferes with chromosome congression in the spindle midzone and abrogates the mitotic checkpoint that mediates chromosome bi-orientation (Grishchuk et al., 2007). Mutation of pkl1 or simultaneous mutation of pkl1 and kinesin-5 encoding cut7 lead to three main defects in chromosome segregation (1) unequal chromosome distribution, (2) lagging chromosomes and (3) chromosome loss (Olmsted et al., 2014). In the absence of Pkl1, Cut7-dependent sliding forces mediate the long spindle microtubules protrusion at the minus-ends and, so, push segregated chromosomes to the site of cell division, resulting in chromosome breaks and chromosome loss (Syrovatkina and Tran, 2015) (Fig. 3D). Another S. pombe kinesin-14, Klp2, is also involved in nuclear positioning during interphase, organization of bipolar microtubules, chromosome alignment and anaphase spindle elongation (Akera et al., 2015; Braun et al., 2009; Mana-Capelli et al., 2012) (Fig. 3D).

In many solid tumors and in hematological malignancies, an increase in centrosome numbers, a common feature referred to as centrosome amplification, often strongly correlates with aneuploidy, chromosomal instability and malignant behaviors (Acilan and Saunders, 2008; Chandhok and Pellman, 2009; Nigg, 2002; Kwon et al., 2008). A crucial cellular mechanism for minimizing the deleterious consequences of multiple centrosomes is the clustering of the extra centrosomes into bipolar spindles to avoid the adverse levels of aneuploidy during mitosis (Acilan and Saunders, 2008; Basto et al., 2008; Holland and Cleveland, 2009).

In Drosophila S2 cells, depletion of kinesin-14 Ncd results in defects of centrosome clustering and in the numbers of multipolar spindles, especially in cells with supernumerary centrosomes (Basto et al., 2008; Kwon et al., 2008). There is a similar mechanism in mammalian cancer cells because human HSET is indispensable for focusing of acentrosomal spindle poles, bipolar spindle organization, bipolar cell division and the survival of cancer cells with extra centrosomes, although it has no obvious effect on wild-type diploid cells (Chavali et al., 2016; Kleylein-Sohn et al., 2012; Kwon et al., 2008; Mittal et al., 2016) (Fig. 3A-C).

In this Commentary, we mainly focused on the molecular kinetics and mechanisms of the minus-end-directed kinesin-14 motors in cell division. We have also discussed the counterbalance between kinesin-14 and kinesin-5 motors that is prevalent on both parallel and antiparallel microtubules, as well as at the MTOC during spindle assembly. Finally, we have provided a detailed view of the roles of kinesin-14 motors in different model systems that reflect the complexities of the mitotic or meiotic spindles and the cellular environment.

Furthermore, the studies of kinesins-14 across several model organisms have led to an understanding of the evolution of mitotic mechanisms and co-evolution of kinesins-14 (Braun et al., 2013; Olmsted et al., 2014), indicating that these motor proteins should not be considered as being constraint to exert a few standard capabilities. An important focus for future work will be to study the antagonistic interactions between kinesin-14 and kinesin-5 motors under more complex conditions, for example within multiple microtubule bundles or in the context of their associated partner proteins, as well as in vivo in a complex cellular environment. Further detailed in vitro and in vivo analyses are also required to elucidate the molecular kinetics and precise mechanisms of kinesin-14 motors at the both plus- and minus-ends of microtubules. In addition, the roles of kinesin-14 motors during development in model organisms remain largely unexplored. This raises the key scientific questions of whether loss-of-function of kinesin-14 motors results in defects of cell metabolism or morphogenesis, developmental abnormalities, or gives rise to any associated diseases.

The authors thank all members in the Sperm Laboratory at Zhejiang University for their helpful discussions. We sincerely thank our colleague Christopher Raymond Wood for the careful editing of the manuscript.