Cortical capture of microtubules and spindle polarity in budding yeast - where's the catch?

Orientation of the mitotic spindle along the cell polarity axis ensures correct segregation of polarized cell fate determinants with respect to the division plane in asymmetrically dividing cells. This prevalent principle for cell patterning in metazoan development (Rhyu and Knoblich, 1995; Betschinger and Knoblich, 2004; Macara, 2004) is also crucial for division in unicellular eukaryotes undergoing polarized cell divisions, such as fission and budding yeast (Hayles and Nurse, 2001; Segal and Bloom, 2001).

Saccharomyces cerevisiae represents a unique genetic model to explore mechanisms of spindle orientation in a cell dividing asymmetrically. This yeast divides by budding off progeny according to a patterning program that distinguishes haploid and diploid cells (Chant, 1999; Chang and Peter, 2003): haploid cells bud adjacent to a previous budding site (an axial pattern); and diploid cells bud either adjacent to or opposite the previous budding site (a bipolar pattern). As is the case in animal cells, orientation of the spindle follows the position of the spindle poles, which interact with the cell cortex through astral (cytoplasmic) microtubules (MTs). Budding yeast is uniquely suited to address how astral MTs orientate the spindle along the cell polarity axis, since in this organism they do not participate in polarized cell growth or control of cell shape (Huffaker et al., 1988; Palmer et al., 1992). This contrasts with Schizosaccharomyces pombe, in which MTs and actin promote the characteristic rod shape. Indeed, perturbation of MTs in S. pombe results in abnormally bent and T-shaped cells (Hayles and Nurse, 2001).

Control of astral MT capture at the cell cortex is intimately linked to the progression of the spindle cycle. Spindle morphogenesis and orientation are governed by the yeast MT-organizing centre - the spindle pole body (SPB). The SPB is inserted in the nuclear envelope and organizes the intranuclear spindle and astral MTs throughout the cell cycle (Byers et al., 1978; Hoyt and Geiser, 1996; Winey and O'Toole, 2001). Progression of the yeast cell cycle is driven by a single cyclin-dependent kinase (Cdk), Cdc28p (also known as Cdk1p), bound to stage-specific cyclins: Cln1p, Cln2p and Cln3p in G1 phase; Clb5p and Clb6p in S phase; and Clb1p, Clb2p, Clb3p and Clb4p in M phase (Nasmyth, 1993).

As cells proceed through G1 phase, astral MTs probe the cell cortex in a search for sites specialized for cortical capture. This process exploits the dynamic instability of MTs (Carminati and Stearns, 1997; Desai and Mitchison, 1997; Shaw et al., 1997; Pearson and Bloom, 2004). MTs emanating from the SPB find their way to the incipient bud in response to asymmetrically positioned polarity determinants (Fig. 1). Captured MTs can then generate movement of the SPB towards the new budding site coupled to MT depolymerization (Shaw et al., 1997; Segal et al., 2002; Pearson and Bloom, 2004). These events prime spindle polarity because the SPB interacting with the incipient bud cortex will later occupy the bud (SPB_bud). The SPB is duplicated by a conservative mechanism at the G1/S transition (Winey and O'Toole, 2001). The new SPB is driven to interact with the opposite cell cortex compartment (in the mother cell) through newly formed astral MTs during spindle assembly (Fig. 1). This asymmetry is imparted by Cdk, which inhibits the organization of MTs at the new SPB in early S phase (Segal et al., 2000b). As the cortical landscape changes during S/G2 phase, this delay effectively confines MTs from the new SPB to the mother cell. As discussed below, controls enforcing this correctly established polarity continue to select MTs at the SPB_bud for delivery along polarized actin cables to the bud. The short spindle then orientates along the mother-bud axis, with the old SPB ready to enter the bud (Segal and Bloom, 2001). Translocation of the SPB into the bud during anaphase is monitored by the mitotic exit network (MEN) (Bardin and Amon, 2001; McCollum and Gould, 2001; Simanis, 2003), which couples successful chromosomal segregation with cytokinesis (Bardin et al., 2000; Pereira et al., 2000).

As part of the ongoing quest for mechanisms of cortical capture linking actin organization with spindle orientation, a large body of research has focused on Kar9p, a proposed functional counterpart of the mammalian tumour suppressor protein adenomatous polyposis coli (APC) (Bloom, 2000), and the actin interactor Bud6p/Aip3p (Amberg et al., 1997; Moseley et al., 2004). More-recent work has revealed novel aspects of spindle polarity control through Kar9p and, possibly, Cdk (Kusch et al., 2003; Liakopoulos et al., 2003; Maekawa et al., 2003; Huisman et al., 2004). We review these studies here, outlining the possible relationship between Bud6p- and Kar9p-dependent modes of spindle orientation and the basis for Cdk control of spindle polarity. In addition, we consider whether Kar9p mediates polarity and/or MT dynamic behaviour under Cdk control (Maekawa and Schiebel, 2004; Huisman et al., 2004).

Genetic analysis of spindle orientation in budding yeast has assigned motors, MT-binding proteins and cortical determinants to two partially overlapping `early' and `late' pathways that promote positioning of the spindle near the bud and its insertion into the bud neck, respectively (Heil-Chapdelaine et al., 1999; Schuyler and Pellman, 2001b). The early pathway is inactivated by actin-depolymerizing drugs or mutations disrupting polarized actin cables. By contrast, the late pathway is independent of actin and is driven by the MT motor dynein (Palmer et al., 1992; Theesfeld et al., 1999; Heil-Chapdelaine et al., 2000b).

A key element in the early pathway, Kar9p, was first identified through genetic screens for karyogamy-defective mutants. Mutant kar9 strains fail to orientate astral MTs towards the tip of the shmoo projection in mating cells or the bud in dividing cells (Miller and Rose, 1998). In addition, a kar9 mutation perturbs karyogamy (the fusion of haploid nuclei into a single diploid nucleus initiated at the SPBs brought together by astral MTs).

Cells overexpressing a fusion protein of Kar9p and green fluorescent protein (GFP) exhibit dot-like fluorescent structures at the bud tip cortex and along MTs. The cortical localization is sensitive to the actin-depolymerizing agent latrunculin A, as well as inactivation of Bud6p or the formin Bni1p (Miller et al., 1999). Kar9p was therefore proposed to be anchored at the bud tip (dependent upon Bud6p and formins) and capture MTs by virtue of its physical interaction with the MT-binding protein Bim1p (Lee et al., 1999; Korinek et al., 2000; Lee et al., 2000; Miller et al., 2000). Bim1p is related to human EB1, a plus-end-tracking protein (+TIP) that binds to APC (Tirnauer and Bierer, 2000; Schuyler and Pellman, 2001a). Kar9p shares limited sequence similarity with APC within the domain mediating binding to EB1.

More-recent studies showed that a Kar9-GFP fusion expressed at endogenous levels is instead recruited to the SPB (Fig. 2) and travels towards MT plus ends in a manner dependent upon Bim1p and the kinesin Kip2p (Carvalho et al., 2003; Liakopoulos et al., 2003; Maekawa et al., 2003). However, Kip2p might be more directly involved in the partially overlapping dynein pathway (see below). Kar9p-bound MTs then interact with the cargo domain of the type V myosin Myo2p and are transported along actin cables towards the bud (Beach et al., 2000; Yin et al., 2000). Accordingly, a Bim1p-Myo2p fusion can bypass the requirement of Kar9p for spindle orientation (Hwang et al., 2003). Kar9p is recruited to the SPB_bud following spindle assembly. In this way, only MTs organized by this SPB are guided to the bud tip (Kusch et al., 2003; Liakopoulos et al., 2003; Maekawa and Schiebel, 2004). This aspect of Kar9p function might be regulated by Cdk (Liakopoulos et al., 2003); however, as we discuss below, the relationship between Cdk and Kar9p is complex. In fact, asymmetrically positioned (by an as-yet-undetermined mechanism) Kar9p appears to be responsible for the translocation of Cdc28p-Clb4p (Fig. 2) to the plus ends of MTs emanating from the SPB_bud, which precludes a role for this mitotic Cdk complex in generating polarity through Kar9p (Maekawa and Schiebel, 2004).

The studies discussed above favour a role for Kar9p as a bridge between MTs and Myo2p-bound actin transport, rather than as a cortical receptor for MT capture (Liakopoulos et al., 2003; Gundersen et al., 2004), and explain the dependency of spindle orientation on the actin cytoskeleton (Palmer et al., 1992; Theesfeld et al., 1999; Liakopoulos et al., 2003; Gundersen et al., 2004). However, this view raises the question of the identity of the elusive cortical target for MT capture (Gundersen et al., 2004). Another point that remains unclear is whether Myo2p-dependent transport of unbound Kar9p (Fig. 2) coexists with delivery of Kar9p while bound to MT plus ends. Actin-mediated transport generating a gradient of Kar9p towards the bud might explain the reported latrunculin-A-sensitive localization of Kar9-GFP to the bud cortex under overexpression conditions (Miller et al., 1999). Live-cell imaging has also revealed the potential for MT plus ends to initiate interactions with overexpressed Kar9-GFP already located in the bud (Beach et al., 2000). Nevertheless, binding of Kar9p to cell cortex factors participating in the early pathway remains to be shown by biochemical methods.

Recent studies in Drosophila have shown the evolutionary conservation of mechanisms controlling spindle orientation. These studies implicate APC, which shares limited sequence similarity with Kar9p, in cortical capture and control of spindle orientation in symmetric versus asymmetric divisions in neural precursors and the germline (Lu et al., 2001; Yamashita et al., 2003). Analysis of centrosome positioning in divisions supporting germline stem cell self-renewal reveals that spindle orientation might be primed during interphase possibly prior to centrosome duplication (Wallenfang and Matunis, 2003). This strikingly correlates with the cell-cycle timing for priming yeast mitotic spindle polarity through control of MT capture at the incipient bud, which occurs well before spindle assembly (Segal and Bloom, 2001).

Bud6p/Aip3 was initially characterized as an actin interactor required for correct bud site selection of diploid cells (Zahner et al., 1996; Amberg et al., 1997). More recently, a truncation of Bud6p has been shown to stimulate formin-dependent polymerization of actin in vitro (Moseley et al., 2004). However, in most genetic backgrounds, bud6Δ mutants form actin cables and exhibit mild defects in cell morphogenesis (Amberg et al., 1997; Evangelista et al., 1997; Sheu et al., 2000; Goehring et al., 2003), indicating that Bud6p might not be crucial for actin cable formation in vivo. Continued localization of Bud6p also depends on Myo2p-bound actin transport (Jin and Amberg, 2000). This partly explains the functional relationship with the formin Bni1p, a Bud6p-interactor required for organization of actin cables and for retention of Bud6p at the bud tip cortex (Lee et al., 1999; Segal et al., 2000a; Wallar and Alberts, 2003).

Bud6p follows a striking program of cortical localization during spindle morphogenesis (Fig. 3). Coupled to the specification of a new cell polarity axis and localized activation of the small GTPase Cdc42p, Bud6p is recruited to the pre-bud site and remains at the bud tip after bud emergence (Amberg et al., 1997; Chang and Peter, 2003). As the spindle assembles, Bud6p also accumulates at the bud neck. Following mitotic exit, it remains at the recent division site until a new bud site is selected. Capture of MTs by Bud6p and its relationship with polarity and bud-site-selection determinants functionally couple spindle orientation with specification of the division plane (Segal and Bloom, 2001).

Live-cell imaging demonstrates the ability of astral MTs to follow the cortical program set by Bud6p (Segal et al., 2002). This is already noticeable during G1 phase, when astral MTs encounter Bud6p at the incipient bud cortex to commit the old SPB to being destined for the bud, priming spindle polarity in a Kar9p-independent manner (Fig. 3). Significantly, shrinkage of MTs coupled to SPB movement towards the bud cortex, a hallmark of MT capture, occurs almost exclusively at Bud6p sites. A bud6Δ mutation abolishes MT shrinkage at the bud and bud neck cortex. However, this mode of cortical capture is unperturbed in the mother cell in these mutants (Segal et al., 2002), which suggests the presence of an alternative anchor for similar pulling force on the SPB retained in the mother cell (Ross and Cohen-Fix, 2004).

MT capture at the bud neck begins at the onset of spindle assembly (Fig. 1) and ensures the new SPB a mother-bound fate. This observation led to the proposal that Bud6p accumulation (Fig. 3) initiates capture events at the bud neck (Segal and Bloom, 2001). Bud6p fails to localize to the bud neck following heat inactivation of the septin ring in cdc3^ts cells (Huisman et al., 2004), which explains the disruption of spindle orientation observed in septin mutants (Kusch et al., 2002). However, Kusch et al. attributed capture at the bud neck to Kar9p (Kusch et al., 2002), even though kar9Δ mutants still orientate the spindle through MT interactions with the bud neck (Segal et al., 2000a). Furthermore, live-cell imaging of Kar9p-bound MTs confirms that they interact with either the bud tip or bud neck; yet, interactions with the bud neck are selectively eliminated by a bud6Δ mutation (Huisman et al., 2004).

In bni1Δ cells, Bud6p prematurely accumulates at the bud neck, which leads to excessive MT capture in this region, an effect abolished by additionally deleting BUD6. These findings single out Bud6p as the prime mediator in cortical capture (Segal et al., 2000a). Bni1p is indirectly linked to Kar9p-dependent delivery of MTs by promoting actin cable formation (Evangelista et al., 2003). By contrast, mDia-family formins might have more-direct functions in the capture of MTs in mammalian cells through their proposed interaction with APC (Palazzo et al., 2001; Yarm et al., 2001; Wen et al., 2004).

Bud6p from S. pombe is also an actin interactor (Jin and Amberg, 2001) participating in cell morphogenesis (Glynn et al., 2001). Studies in this yeast have identified partners of Bud6p underlying the cooperation between actin and MTs in the control of cell shape. S. pombe divides by medial fission to generate daughter cells of equal size. Initially, cells grow by extending the old end (i.e. the end distal to the site of septation that exists prior to division). Later in the cell cycle, cells initiate apical extension of the recent division site or `new' cell end, a transition known as `new-end take-off' (NETO). Spatial control of polarized growth requires the MT-mediated delivery of polarity factors such as the kelch-repeat protein Tea1p to the cell tips (Behrens and Nurse, 2002; Chang and Peter, 2003). In turn, Tea1p recruits Bud6p and the formin For3p to regulate actin organization (Feierbach et al., 2004). This process is highlighted by the colocalization of Bud6p with 62% of the MTs that reach the cell ends (Glynn et al., 2001). It is unclear whether Tea1p mediates dynamic interactions between MTs and Bud6p directly or whether additional proteins are required (Martin and Chang, 2003; Snaith and Sawin, 2003). Another question is whether Bud6p has any role in orientating Tea1p cortical delivery. In this respect, it is significant that bud6 mutants do not support NETO (Glynn et al., 2001).

These observations illustrate the importance of Bud6p in linking MTs with cell polarity in these two highly divergent yeasts and might further suggest relevant partners for S. cerevisiae Bud6p.

Previous models assumed that Bud6p is required for actin organization to support Kar9p-dependent MT delivery as part of the `early pathway'. However, the studies reviewed here indicate Bud6p and Kar9p function independently in spindle orientation by engaging different astral MTs generated by the SPB_bud, as summarized in Fig. 4 (Huisman et al., 2004). This notion is also consistent with the dynamic behaviour of MTs associated with Kar9p-dependent delivery to the bud or Bud6p-mediated capture (Fig. 4B,C). The two systems cooperate throughout the actin-sensitive period to achieve pre-anaphase spindle orientation (Theesfeld et al., 1999); yet, Bud6p participation in MT capture extends beyond this point (Segal et al., 2002).

Kar9p- and Myo2p-dependent transport along actin cables (Fig. 4B) occurs without changes in MT length (Hwang et al., 2003; Liakopoulos et al., 2003). This mode of MT delivery to the bud is proficient even in bud6Δ cells (Huisman et al., 2004). Similarly, many aspects of Bud6p-driven capture occur in kar9Δ mutants (Segal et al., 2002), including MT depolymerization coupled to SPB movement towards the cortex (Huisman et al., 2004). More importantly, live-cell imaging shows that Kar9p-bound MTs do not couple shrinkage with SPB movement and that Kar9p is not always present at MT plus ends interacting with the cell cortex (Huisman et al., 2004). This indicates that Kar9p does not mediate capture at the cell cortex per se.

To maintain dynamic MT bud-cortex interactions primed by Bud6p during G1 phase at the incipient bud (Segal et al., 2002), Kar9p might also affect MT dynamic behaviour (as shown by the fixed length of Kar9p-bound MTs during processive actin-based transport; Fig. 4B). Indeed, a tub2^C354S allele, which decreases MT dynamics (Gupta et al., 2002), improves spindle orientation in kar9Δ cells in a Bud6p-dependent manner (Huisman et al., 2004). This indicates that Kar9p-independent cortical capture at Bud6p sites (Segal et al., 2002) is sufficient to orientate the spindle, provided MT turnover is decreased. It might also point to an additional role for Kar9p that can be emulated by reduced MT dynamics. Kar9p could modulate Bim1p-dependent MT dynamicity (Tirnauer et al., 1999; Segal and Bloom, 2001). Indeed, Cdk-dependent phosphorylation has been proposed to control Kar9p-Bim1p interaction (Liakopoulos et al., 2003). Moreover, MTs entering the bud undergo repeated cycles of recovery (a new phase of regrowth following depolymerization) in a Kar9p-dependent manner, preventing MTs from shrinking past the bud neck. This has the net effect of maintaining astral MTs targeted to the bud compartment (Huisman et al., 2004). Significantly, human APC (the counterpart of yeast Kar9p) is also subject to phosphorylation by Cdk, which controls its association with EB1 and thereby modulates MT polymerization (Nakamura et al., 2001). Thus, this mode of regulation of MT dynamics might be evolutionary conserved.

Insertion of the mitotic spindle into the bud neck during anaphase requires pulling force exerted by astral MT sliding along the bud cortex, which is independent of actin cables (Adames and Cooper, 2000; Heil-Chapdelaine et al., 2000b). This process is driven instead by cytoplasmic dynein (Fig. 5), a minus-end-directed MT motor (Yeh et al., 1995; Bloom, 2001), associated with the cortical anchor Num1p (Heil-Chapdelaine et al., 2000a; Farkasovsky and Kuntzel, 2001). Targeting of dynein to MT plus ends requires Bik1p and Pac1p (Lee et al., 2003; Sheeman et al., 2003), the yeast homologues of the human +TIP CLIP-170 and the dynein activator LIS1, respectively (Olson and Walsh, 2002; Carvalho et al., 2003). In turn, cell-cycle-dependent build-up of Bik1p at MT plus ends is regulated by Kip2p, a putative plus-end-directed kinesin (Carvalho et al., 2004) previously shown to promote MT stability and placed genetically in the `dynein pathway' (Cottingham and Hoyt, 1997).

These findings parallel results obtained in studies of the role of Tea2p, the fission yeast Kip2p counterpart, in the localization of the CLIP-170 family member Tip1p. The presence of Tip1p at MT plus ends inhibits catastrophe and ensures delivery of the polarity factor Tea1p to the cell tips. Yet some differences between the two model systems are apparent. In the case of fission yeast, Tip1p and Tea2p localization requires Mal3p, the fission yeast EB1 homologue, which suggests a mechanism for a combined action of kinesin and +TIPs in inhibition of MT catastrophe (Browning et al., 2003; Busch and Brunner, 2004; Busch et al., 2004). However, the budding yeast EB1 homologue, Bim1p, is not required for Kip2p localization. In fact, deletion of BIM1 enhances targeting of Kip2p and Bik1p to MT plus ends, highlighting the potential for competition between Kar9p- and dynein-dependent pathways at the level of MT plus end occupancy (Segal and Bloom, 2001; Carvalho et al., 2004). Thus, the functional relationship between the various +TIPs and/or the effects on MT dynamic behaviour might not be entirely conserved between the two yeasts. Note also that a functional counterpart of Kar9p has not been identified in S. pombe.

The Bud6p localization program has been proposed to play an important role in control of SPB inheritance. Capture at the incipient bud singles out the `old' SPB for segregation to the bud (Pereira et al., 2001; Bornens and Piel, 2002; Segal et al., 2002; Smeets and Segal, 2002). Subsequently, Cdc28p-Clb5p kinase inhibits de novo astral MT organization at the SPB outer plaque until a short spindle assembles (Segal et al., 2000b). This built-in delay coincides with the accumulation of Bud6p at the bud neck, which generates a barrier that effectively prevents new astral MTs formed by the new SPB from contacting the bud cortex (Segal et al., 2002). The correct execution of this program sets the identity of each SPB - the old SPB remains targeted to the bud whereas the new SPB interacts exclusively with the mother cell cortex.

The asymmetric recruitment of Kar9p might underscore spindle polarity. Either Cdc28p-Clb5p or Cdc28p-Clb4p kinase has been proposed to phosphorylate Kar9p and limit its function to MTs emanating from the SPB_bud following spindle assembly (Kusch et al., 2003; Liakopoulos et al., 2003; Maekawa et al., 2003). However, other studies discussed below indicate that asymmetric recruitment of Kar9p might lie downstream of spindle polarity establishment.

In cdc28-4 clb5Δ cells, which have impaired Cdk activity, Kar9p-bound MTs do not undergo capture at the bud neck cortex. Maekawa et al. interpreted this phenotype as indicating that Cdc28p-Clb5p kinase restricts Kar9p-mediated capture of MTs to the bud neck to ensure the new SPB remains in the mother cell (Maekawa et al., 2003). However, cdc28-4 clb5 mutants establish symmetrical interactions between MTs emanating from both SPBs and the bud because of the advanced timing of astral MT organization by the new SPB relative to the accumulation of Bud6p at the bud neck. MTs generated by the new SPB prior to the presence of a Bud6p barrier interact with the bud tip rather than with the bud neck. Failure to specify the correct fate for the new SPB is not linked to Kar9p function. Indeed, a dynein-GFP fusion that localizes to the SPBs in an astral-MT-dependent manner initially marks solely the old SPB but begins to accumulate at the new SPB after a lag (as de novo MTs form) (Shaw et al., 1997; Sheeman et al., 2003). This intrinsic asymmetry is abolished in cdc28-4 clb5Δ cells, in which both SPBs become labelled simultaneously (Segal et al., 2000b), but is unperturbed by a kar9 mutation (Yeh et al., 2000). Moreover, Bud6p rather than Kar9p promotes bud neck capture (Segal et al., 2000a). More importantly, Kar9p is present at both SPBs during spindle assembly in wild-type cells, becoming asymmetrically recruited once their identities (daughter-destined versus mother-destined) are defined (Huisman et al., 2004). This correlation points to a mechanistic link between orientation of cortical interactions and asymmetric recruitment of Kar9p to the SPB_bud, as suggested previously in the case of MEN regulators (Pereira et al., 2001). Significantly, a kar9 mutation does not prevent Bud6p-dependent MT capture at the incipient bud that commits the old SPB to later occupying the bud (Segal et al., 2002) and is not sufficient to randomize SPB inheritance fully (Pereira et al., 2001). Thus, Kar9p may enforce but not initiate spindle polarity.

Liakopoulos et al. have suggested that Cdc28p-Clb4p controls asymmetric recruitment of Kar9p after spindle assembly (Liakopoulos et al., 2003). Their finding that overexpressed Clb4p localizes to the mother-bound SPB prompted the proposal that Cdc28p-Clb4p-mediated phosphorylation prevents recruitment of Kar9p at the SPB retained by the mother cell. However, this notion has been challenged by another study (see below).

Kar9p asymmetry is partially lost in clb3 and clb4 mutants (Liakopoulos et al., 2003). Yet, these mutants exhibit a delay in SPB separation in early spindle assembly, during which Kar9p is present at both poles (Huisman et al., 2004). Thus, Cdk might exert an indirect effect on asymmetric recruitment of Kar9p.

The fact that Kar9p might not be the effector in control of polarity by Cdk, in particular Cdc28p-Clb5p, is further demonstrated by the behaviour of spindles in cells expressing the Bim1p-Myo2p fusion protein mentioned earlier. This fusion can bind to MTs generated by both SPBs (through its Bim1p component) and link them directly to actin cables (through the Myo2p domain), bypassing Kar9p (normally the bridge between Bim1p and Myo2p). This bypass, however, does not confer diploid-specific lethality in which spindles are fully positioned in the bud (Hwang et al., 2003), a phenotype associated with the loss of spindle polarity in cdc28-4 clb5Δ cells (Segal et al., 1998). Thus, control of Kar9p recruitment lies downstream of intrinsic SPB functional asymmetry (the delay in MT organization at the new SPB) imparted by Cdc28p-Clb5p kinase in S phase, which dictates the differential behaviour of the SPB retained by the mother cell.

By contrast, genetic analysis implicates Clb3p or Clb4p in the control of SPB separation not spindle polarity (Segal et al., 1998; Segal et al., 2000b; Maekawa and Schiebel, 2004). Indeed, an entirely different picture emerges from a new study demonstrating that Kar9p mediates the localization of Cdc28p-Clb4p to plus ends of MTs generated by the SPB_bud (Maekawa and Schiebel, 2004). This study showed that Clb4p at endogenous levels does not accumulate at the mother-orientated SPB or significantly affect asymmetric recruitment of Kar9p. Instead, Kar9p-mediated delivery of this Cdk complex appears to modulate differential MT-cortex interactions and might therefore be important to redirect dynamic attachments from the bud tip to the bud neck. In clb4Δ mutants, exaggerated MT interactions with the distal portion of the bud cortex appeared to exert excessive pulling force on the otherwise correctly aligned spindle. Failure to direct interactions with the cortex in the vicinity of the neck compromises both the retention of the spindle and the timely initiation of spindle elongation (Segal et al., 2000a; Yeh et al., 2000; Maekawa and Schiebel, 2004). In light of these findings, it is tempting to speculate that Cdk acts on a range of targets close to MT plus ends either to disengage MTs associated with Kar9p-bound transport or to modulate the transition from actin-dependent orientation to dynein-driven positioning of the pre-anaphase spindle. The cell-cycle-regulated abundance of the kinesin Kip2p, and the ensuing increase in dynein recruitment through Bik1p (Carvalho et al., 2004), might interface with this mode of control to mark this transition between capture mechanisms.

Establishment of spindle polarity by regulated MT capture exhibits many layers of control. That Cdk plays a significant role in this process through multiple effectors is beginning to emerge. Identification of the Cdk targets involved will provide exciting leads to help us understand the temporal and spatial control of cortical capture of MTs. Another outstanding issue is how polarized association of SPBs with distinct cortical compartments translates into correct SPB fate through asymmetric recruitment of Kar9p. This additional layer of extrinsic control also marks the SPB_bud through the recruitment of MEN regulators (Bardin et al., 2000; Hoyt, 2000; Pereira et al., 2000; Pereira et al., 2001; Smeets and Segal, 2002).

The pattern of SPB inheritance can be randomized; for example, through transient exposure to MT poisons (Pereira et al., 2001; Maekawa et al., 2003), a treatment that induces symmetric recruitment of Kar9p and MEN elements to SPBs (Liakopoulos et al., 2003; Maekawa and Schiebel, 2004). However, upon recovery from such treatment, the spindle shows the amazing ability to rebuild polarity to direct a single SPB to the bud. Irrespective of its original identity (new versus old), the SPB that happens to become orientated towards the bud recruits Kar9p, Cdc28p-Clb4p and MEN components. As a result, it is always the pole that enters the bud that carries the signals to trigger mitotic exit upon completion of chromosomal segregation. The mechanism by which this asymmetry is established remains to be elucidated, although it must involve MT-mediated transfer of information from the cortex to the SPB. This capacity for `symmetry breaking' is not restricted to re-establishment of spindle polarity. Indeed, symmetry-breaking mechanisms have been demonstrated in yeast cells devoid of cortical landmarks for selection of a new budding site. These cells build up a cell polarity axis to divide (Irazoqui et al., 2003).

Spontaneous cell polarization in the absence of pre-established asymmetric cues also occurs in a variety of eukaryotic cells (Wedlich-Soldner and Li, 2003). Spindle orientation in higher eukaryotes shares many common molecular themes with budding yeast. Further studies might determine whether similar underlying principles of crosstalk between centrosomes and the cell cortex contribute to control of spindle polarity and cell division in animal cells.

We thank R. Sinka, M. Bettencourt-Dias, N. Delgehyr and D. Glover for fruitful discussions and comments on the manuscript. We apologize to colleagues whose work could not be cited in this commentary due to space limitations. Work in our laboratory is supported by Cancer Research UK, The Wellcome Trust and The Isaac Newton Trust.