In S. cerevisiae, spindle orientation is linked to the inheritance of the `old' spindle pole by the bud. A player in this asymmetric commitment, Bud6p, promotes cortical capture of astral microtubules. Additionally, Bud6p stimulates actin cable formation though the formin Bni1p. A relationship with the second formin, Bnr1p, is unclear. Another player is Kar9p, a protein that guides microtubules along actin cables organised by formins. Here, we ask whether formins mediate Bud6p-dependent microtubule capture beyond any links to Kar9p and actin. We found that both formins control Bud6p localisation. bni1 mutations advanced recruitment of Bud6p at the bud neck, ahead of spindle assembly, whereas bnr1Δ reduced Bud6p association with the bud neck. Accordingly, bni1 or bnr1 mutations redirected microtubule capture to or away from the bud neck, respectively. Furthermore, a Bni1p truncation that can form actin cables independently of Bud6p could not bypass a bud6Δ for microtubule capture. Conversely, Bud61-565p, a truncation insufficient for correct actin organisation via formins, supported microtubule capture. Finally, Bud6p or Bud61-565p associated with microtubules in vitro. Thus, surprisingly, Bud6p may promote microtubule capture independently of its links to actin organisation, whereas formins would contribute to the program of Bud6p-dependent microtubule-cortex interactions by controlling Bud6p localisation.
Spatial coordination between chromosomal segregation and the division plane in an asymmetrically dividing cell is achieved by coupling mitotic spindle orientation to the cell polarity axis. The yeast S. cerevisiae adheres to this general principle: the yeast preanaphase spindle aligns according to the mother-bud axis so that chromosomal segregation may occur across the bud neck (Pearson and Bloom, 2004). Spindle orientation is achieved through a complex program that controls the spindle pole body (SPB), the yeast counterpart of the centrosome. The SPB is imbedded in the nuclear envelope and exposes nuclear and cytoplasmic faces that organise the spindle and astral MTs, respectively (Byers, 1981).
SPB duplication generates a `new' SPB side-by-side to the `old' SPB from the preceding cell cycle (Jaspersen and Winey, 2004). The old SPB is dynamically tethered to the emerging bud by its existing astral MTs. Astral MT organisation at the new SPB awaits spindle assembly (Shaw et al., 1997), a delay dependent upon S-phase cyclin-dependent kinase activity (Segal et al., 2000b; Huisman et al., 2007). Thus, the new SPB can move away from the bud neck as the spindle assembles. A similar temporal asymmetry underlies the movement of the new centrosome in stem cell divisions of Drosophila male germline (Yamashita et al., 2007). In yeast, inherent SPB asymmetry translates into spindle polarity by the action of two mechanisms ensuring that the old SPB becomes the SPBbud, i.e. the bud-ward pole (reviewed by Huisman and Segal, 2005), accounting for an invariant pattern of SPB inheritance (Pereira et al., 2001).
The first mechanism involves the polarity determinant Bud6p that marks the incipient bud cortex for capture of astral MTs from the old SPB. Astral MTs contacting Bud6p sites prolong their interaction with the cell cortex and can couple MT growth and shrinkage with SPB movement (Segal et al., 2002). Later, Bud6p accumulates at the bud neck in time to limit the access to the bud of astral MTs newly formed by the new SPB (Amberg et al., 1997; Segal et al., 2000a). Thus, spindle polarity is established within this restricted temporal window, with the old SPB intended for the bud. The second mechanism centres on Kar9p, a protein that provides guidance of astral MT plus ends along actin cables to reach the growing bud. Kar9p is recruited by both SPBs at onset of spindle assembly and becomes progressively polarised to the SPBbud. From the SPB, Kar9p translocates to MT plus ends in association with the plus end-tracking protein Bim1p. Kar9p-bound MTs are then delivered to the bud as cargoes of the type V myosin Myo2p (Korinek et al., 2000; Lee et al., 2000; Miller et al., 2000; Beach et al., 2000; Yin et al., 2000; Hwang et al., 2003) (for a review, see Huisman and Segal, 2005).
Both mechanisms enforcing spindle polarity engage the activity of formins, actin filament nucleators that play important roles in cell polarity and morphogenesis with concomitant impact in a wide range of developmental processes (Faix and Grosse, 2006). Organisation of polarised actin cables in yeast involves two formins, Bni1p and Bnr1p, which are downstream effectors of Rho-like GTPases (Evangelista et al., 1997; Imamura et al., 1997; Pruyne et al., 2002; Sagot et al., 2002; Dong et al., 2003; Pruyne et al., 2004; Moseley and Goode, 2006). Part of a family related to Diaphanous, they may exist in an autoinhibited conformation involving an intramolecular interaction that is relieved upon binding of an active Rho GTPase close to an N-terminal inhibitory site (Otomo et al., 2005). Bni1p generates actin cables from the bud tip as part of the polarisome, a complex assembled at the site for bud emergence upon Cdc42p activation (Chang and Peter, 2003) that also includes Bud6p, Spa2p and Pea2p. Bnr1p organises actin cables from the bud neck (Pruyne et al., 2004; Buttery et al., 2007). At least one formin is necessary for viability (Ozaki-Kuroda et al., 2001). A recombinant C-terminal fragment of Bud6p (amino acids 489-788) can stimulate actin cable formation by Bni1p in vitro. Stimulation depends on Bud6p binding to a site between amino acids 1750 and 1824 of Bni1p, although further C-terminal sequences are required for stimulation (Moseley and Goode, 2005). Bud6p may bind Bnr1p (Kikyo et al., 1999), but no functional interaction was detected in vitro (Moseley and Goode, 2005).
Consistent with its importance for cell polarity, Bni1p has been implicated in establishment of spindle polarity (Fujiwara et al., 1999; Lee et al., 1999; Miller et al., 1999; Segal et al., 2000a; Yeh et al., 2000; Evangelista et al., 2002). Astral MTs of bni1Δ cells tether the duplicated SPBs to the incipient bud (Segal et al., 2000a); however, preanaphase spindle alignment along the mother–bud axis fails (Lee et al., 1999). This may be due to compromised Kar9p function (Miller et al., 1999) compounded by a tendency to symmetric interactions from both SPBs with the bud neck that perturbs specification of SPB identity (Segal et al., 2000a; Yeh et al., 2000). This latter defect may correlate with the advanced and exaggerated accumulation of Bud6p at the bud neck in bni1Δ cells (Segal et al., 2000a). By contrast, a bnr1 mutation did not affect spindle orientation (Evangelista et al., 2002).
Interestingly, a bni1CTΔ mutant that expresses a Bni1p truncation lacking amino acids 1749-1953, thus excluding the putative Bud6p-binding site and DAD domain, shared the spindle orientation and bud site selection phenotypes of bni1Δ cells, even though actin organisation was deemed to be unperturbed (Evangelista et al., 1997; Lee et al., 1999). Yet, the same mutant formin was dominant active for actin cable formation in vivo upon overexpression [then termed BNI1-FΔD by Ozaki-Kuroda et al. and Sagot et al. (Ozaki-Kuroda et al., 2001; Sagot et al., 2002)]. Under these conditions, cable formation by Bni1FΔDp (but not by full-length Bni1p) was independent of Bud6p (Sagot et al., 2002). Similarly, S. pombe Bud6p binds For3p in a C-terminal region overlapping with its DAD domain. Mutations disrupting DAD function, allowed for3(DAD-) to bypass NETO defects of a bud6Δ mutant. Thus, lack of autoinhibition rendered For3p independent of Bud6p for control of actin integrity and cell polarity in vivo (Martin et al., 2007).
Formins and Bud6p may orient the spindle, in part, via Kar9p as they organise the actin cables necessary for MT guidance. Yet, Bud6p contributes to spindle orientation by a separate mechanism. Indeed, the dynamic properties of MT–cortex interactions promoted by Bud6p are unchanged in a kar9Δ mutant (Huisman et al., 2004). However, this observation does not preclude that Bud6p-dependent capture of MTs may involve its known partners: the formins.
Here, we explored the key contributions of formins to Bud6p-mediated control of spindle polarity and orientation by determining whether formins control Bud6p cortical localisation and whether they mediate Bud6p-dependent capture of MTs. We also attempted to separate Bud6p role in MT capture from actin organisation via formins.
We found that both formins regulated Bud6p localisation. bni1 mutations advanced recruitment of Bud6p at the bud neck, ahead of spindle assembly, whereas bnr1Δ reduced Bud6p association with the bud neck. Accordingly, in formin mutants, MT capture was redirected to sites mirroring the disruption in cortical distribution of Bud6p. Furthermore, cells expressing a Bni1p truncation that formed actin cables independently of Bud6p as the sole formin could not support cortical capture of MTs in the absence of Bud6p. Conversely, Bud61-565p, a truncation lacking the region required for formin binding and stimulation, supported MT capture without restoring actin organisation to bud6Δ cells. Finally, Bud6p or Bud61-565p associated with microtubules in vitro.
Our data indicate that Bud6p-dependent capture may not involve formins beyond their contribution to cell cycle control of Bud6p cortical partition, with Bud6p acting as the link to MTs.
Bud6p correct partitioning between the bud and bud neck requires the yeast formins
To undertake the genetic analysis of formin involvement in Bud6p-mediated MT capture, we first determined whether mutating either formin gene perturbed Bud6p localisation. Our previous studies showed that Bni1p, the formin at the bud tip, was crucial for the temporal program of Bud6p localisation (Segal et al., 2000a) but a corresponding role for Bnr1p was not addressed.
To determine the contribution of both formins to Bud6p localisation, the partitioning between the bud tip and bud neck of Bud6p-GFP expressed at endogenous levels was compared in different formin mutants. Cells additionally expressed CFP-Tub1p to offer an independent landmark of cell cycle progression. After early association with the bud tip, Bud6p-GFP accumulated at the bud neck in wild-type cells coincident with initiation of spindle assembly (Fig. 1A,B). A bnr1Δ mutation resulted in very weak or absent localisation of Bud6p-GFP at the bud neck as cells completed spindle assembly. Faint label of the bud neck appeared in large budded cells with elongated spindles (Fig. 1A-B) and reached wild-type levels as cells underwent cytokinesis. This observation was validated by comparing fluorescence intensity at the bud tip and bud neck cortex in wild type or bnr1Δ budded cells expressing Bud6p-GFP as a function of bud growth (Fig. 1C) or by comparing overall labelling of the bud and bud neck regions in mid-size budded cells (1.5-3.0 μm bud length) (Fig. 1D). Labelling of the bud cortex was comparable yet labelling at the bud neck was significantly reduced in bnr1Δ cells prior to cytokinesis.
bni1Δ cells (Segal et al., 2000a), or bni1CTΔ cells that expressed a formin truncation lacking amino acids 1749-1953 eliminating the putative Bud6p-binding site (Evangelista et al., 1997; Lee et al., 1999; Sagot et al., 2002), exhibited advanced accumulation of Bud6p-GFP at the bud neck without otherwise affecting association with the bud. Label at the bud neck was already prominent in cells lacking a spindle (Fig. 1A,B). Finally, the bnr1Δ mutation also delayed Bud6p-GFP labelling at the bud neck in combination with bni1CTΔ (Fig. 1A,B). Thus, Bnr1p may tether Bud6p to the bud neck whereas Bni1p may act from the bud tip to restrict Bud6p arrival at the bud neck until a spindle has formed. At the same time, Bni1p was dispensable for Bud6p localisation at the bud tip.
Importantly, Bud6p association with the bud neck exhibited only partial temporal overlap with the presence of each formin (see Fig. S1 in the supplementary material). Indeed, Bnr1p was recruited at the bud neck following bud emergence before Bud6p (98%, n=200). Both proteins then colocalised until only Bnr1p disappeared prior to cytokinesis. Conversely, Bni1p and Bud6p colocalised at the bud cell cortex, yet, association of Bud6p with the bud neck in mid-size budded cells was not accompanied by Bni1p (100%, n=200 cells). Finally, Bni1p colocalised with Bud6p during cytokinesis (see Fig. S1 in the supplementary material) whereas Bnr1p was absent from the division site (Buttery et al., 2007).
Thus, formins may influence Bud6p program of cortical localisation; however, their distribution may not be the sole factor accounting for the temporality of Bud6p cortical partition.
Distinct perturbation of actin organisation by bni1Δ vs. bni1CTΔ in spite of a common effect on Bud6p localisation
Bni1p and Bnr1p may define two axis directing polarised actin cable organisation from the bud tip and bud neck, respectively (Pruyne et al., 2004), that might contribute to target Bud6p (Jin and Amberg, 2000). However, the similar perturbation in Bud6p cortical accumulation observed in bni1Δ and bni1CTΔ cells, prompted us to reassess the actin organisation phenotypes of the different formin mutants.
Bulk F-actin organisation was examined in fixed cells stained by rhodamine-phalloidin (Fig. 2). Wild-type cells exhibited the characteristic polarised actin patches and thick actin cables lining the mother cell. Consistent with Bnr1p function as a potent actin nucleator and bundling factor positioned at the bud neck (Pruyne et al., 2004; Moseley and Goode, 2005), bnr1Δ cells exhibited fewer and thinner cables (Fig. 2A,B). By contrast, bni1Δ or bni1CTΔ exhibited thick actin cables from the neck (Lee et al., 1999). Cells expressing solely the temperature sensitive Bni1-FH2#1p (Sagot et al., 2002) exhibited completely depolarised patches and no cables following a 30 minute shift to 34°C (see Fig. S2A,B in the supplementary material).
The perturbation provoked by bni1CTΔ, however, became apparent in the double mutant bnr1Δ bni1CTΔ. Indeed, contrary to bnr1Δ cells that lacked the bulk of actin cables in the mother cell, the bnr1Δ bni1CTΔ mutant maintained robust actin cables, although their polarity was affected (Fig. 2B). Thus, consistent with previous findings based on overexpression (Sagot et al., 2002) the formin encoded by endogenous bni1CTΔ promoted excessive actin cables. However, scoring of bipolar budding pattern in diploid cells, a highly sensitive indicator of perturbations in actin and cell polarity (Yang et al., 1997; Sheu et al., 2000), showed that heterozygous bni1CTΔ/bni1Δ diploids supported bipolar budding within 15% of the value for a BNI1/bni1Δ (Fig. 2C,D). Thus, in spite of the proposed hyperactivity (Sagot et al., 2002), Bni1CTΔp at endogenous levels significantly supported cell polarity. It is noteworthy that bni1Δ and bni1CTΔ mutants shared a common behaviour regarding Bud6p accumulation at the bud neck, but exhibited distinct actin organisation phenotypes. It follows that the influence of Bni1p on Bud6p localisation at the bud neck may not solely involve actin.
Actin organisation in bni1CTΔ bnr1Δ cells may be independent of Bud6p
To determine the relative requirement of Bud6p for actin organisation versus cytoplasmic microtubule capture in cells expressing Bni1CTΔp as the sole formin, we first proceeded to test for synergistic impairment of actin organisation upon combining the different formin mutations and a bud6Δ.
A bud6Δ mutation decreased thick actin cables as did a bnr1Δ (Amberg et al., 1997; Feierbach et al., 2004), suggesting a link between Bud6p and Bnr1p-dependent cable organisation in the mother cell. In support of such a link, synergistic effects between bni1Δ and bud6Δ were observed. Relative to single mutants, overall polarity and actin organisation were severely compromised in bni1Δ bud6Δ cells. This was the only instance in which we observed a marked increase in cells with depolarised patches (Fig. 2E). Yet, this synergism was not apparent by combining bud6Δ with bni1CTΔ bnr1Δ mutations (Fig. 2F), in contrast to the genetic interaction observed between bud6Δ and single bni1CTΔ or bnr1Δ mutations, respectively (see Fig. S2C,D in the supplementary material), indicating that formin organisation of actin cables in the bni1CTΔ bnr1Δ mutant may be indeed Bud6p independent.
Supporting this view, colocalisation between Bud6p-CFP and Bni1CTΔp-GFP at the bud cortex was less precise with ∼25% of all foci containing one of the two signals, in marked contrast to the precise colocalisation with Bni1p-GFP (see Fig. S3 in the supplementary material). Taken together, these data may point to the importance of the C-terminal region deleted in Bni1CTΔp in mediating an interaction with Bud6p in vivo.
In conclusion, Bud6p may cooperate with Bnr1p to organise actin, a functional interaction that may also account for Bnr1p influence on Bud6p localisation at the bud neck. In addition, Bud6p distribution in formin mutants might reflect the differential disruption of actin cables from the bud tip or bud neck sustaining Bud6p delivery (Jin and Amberg, 2000; Pruyne et al., 2004). However, the fact that Bni1CTΔp supported cell polarity but could not impede Bud6p advanced accumulation at the bud neck may also point to the importance of the physical interaction between Bni1p and Bud6p via their respective C termini in tethering Bud6p to the bud cortex. Finally, a bni1CTΔ bnr1Δ strain may organise actin cables independently of Bud6p.
Bud6p-dependent microtubule-cortex interactions persist in formin mutants
Preanaphase spindle orientation along the mother-bud axis requires correctly partitioned contacts by astral MTs emanating from each SPB. MTs from the SPBbud interact with three cortical domains: the bud, bud neck and mother cell cortex. The mother-bound SPB is restricted to interact with the mother cell cortex. Bud6p may be important for partitioning these interactions (Segal et al., 2000a; Yeh et al., 2000; Huisman et al., 2004).
To determine whether formins might participate in Bud6p-dependent capture of astral MTs and to assess the impact of Bud6p displacement in formin mutants, time lapse analysis of wild type or formin mutants expressing GFP-Tub1p was carried out.
MT contacts by cortical domain were monitored from SPB separation to assembly of a preanaphase spindle (Huisman et al., 2004). Compared with wild-type cells, bnr1Δ cells increased the number of MT interactions within the bud at the expense of interactions with the bud neck region. Conversely, bni1CTΔ cells favoured MT contacts with the bud neck (Fig. 3A,B), as shown for bni1Δ cells (Segal et al., 2000a), an effect suppressed by disruption of bnr1Δ (Fig. 3A) or bud6Δ [not shown but see Segal et al. (Segal et al., 2000a)].
MT growth and shrinkage at the cell cortex (causing SPB movement relative to the cell cortex contact point) constitute two modes of interaction linked to cortical Bud6p (Segal et al., 2002). To address whether formins could mediate this dynamic behaviour, the various modes of MT–cortex interaction were scored in time lapse series from mitotic exit to bud emergence (Fig. 3C). During this interval, localisation of Bud6p in different formin mutants would be comparable. MT shrinkage and growth at the cell cortex were well represented in wild-type and bnr1Δ bni1CTΔ cells, even though the double mutant expressed a formin that was presumably unable to bind Bud6p. More importantly, deletion of BUD6 in this mutant disrupted MT growth and shrinkage at the cell cortex (Fig. 3C). In order to verify that these findings did not result from any indirect effect of the deletion of BUD6 on Bni1CTΔp position at the cell cortex, a bni1CTΔ-GFP allele was introduced replacing endogenous BNI1 in otherwise wild-type, bud6Δ and bud6Δ bnr1Δ backgrounds. As shown in Fig. 3D, Bni1CTΔp-GFP was localised to the bud cortex and the division site as expected from its ability to support cell polarity (Fig. 2). This localisation was unaffected in combination with bud6Δ and bnr1Δ mutations. Yet, the same formin construct depended on Spa2p for efficient cortical localisation, consistent with Bni1CTΔp having the putative binding site of Spa2p but not that of Bud6p. By contrast, cortical localisation of full-length Bni1p-GFP expressed at endogenous levels was impaired by deleting both BUD6 and SPA2 in a synergistic manner (see Fig. S4 in the supplementary material).
Thus, Bni1CTΔp cannot bypass the requirement for Bud6p in cortical capture of MTs, even though it seemed to support Bud6p-independent organisation of actin cables (Sagot et al., 2002) (Fig. 2F). Moreover, bni1Δ cells that are devoid of formins at the cytokinesis site and presumptive bud site supported proficient Bud6p-dependent MT capture during that cell cycle interval (see Fig. S5 in the supplementary material).
Finally, preanaphase spindle orientation was determined in asynchronous populations of the different formin mutants to compare the overall disruption of the program for spindle orientation (Fig. 3E). Both bud6Δ and bnr1Δ exhibited subtle defects related to lack of spindle retention at the bud neck (Evangelista et al., 2002; Miller et al., 1999; Segal et al., 2000a; Yeh et al., 2000). Indeed, bnr1Δ cells exhibited frequent transits of preanaphase spindles across the bud neck during time lapse analysis (not shown), as previously shown for bud6Δ cells (Segal et al., 2000a). Both bni1Δ and bni1CTΔ exhibited misaligned spindles tethered to the bud neck (Miller et al., 1999; Lee et al., 1999; Segal et al., 2000a) and were additionally affected by deleting BUD6. Interestingly, bnr1Δ in combination with bni1CTΔ significantly increased misaligned spindles away from the bud neck, highlighting the combined contributions of disorganised actin cables that discourage MT targeting to the bud (Fig. 3A) and the absence of Bud6p from the bud neck (Fig. 1). Accordingly, bud6Δ had a modest effect in combination with a double formin mutation.
It follows that Bud6p may promote MT-cortex interactions coupled to SPB movement, independent of its association with formins. However, formins are still crucial to effect Bud6p localisation and to allow Kar9p to deliver MTs to Bud6p-decorated areas.
A Bud6p truncation unable to promote actin organisation via formins supports MT capture
To gain further support to the view that formins may not be mediators in Bud6p-dependent capture of MTs, we then attempted to genetically separate Bud6p functions in actin organisation via formins from any links to MTs.
In order to outline regions of Bud6p supporting cortical capture of MTs, a series of GFP-tagged truncations of Bud6p was assayed in bud6Δ cells. In particular, a truncation still encoding amino acids 1-565 of Bud6p (Bud61-565p) was localised to the bud and bud neck cortex (Fig. 4A), while shorter versions of the N-terminal region marked the incipient bud but became delocalised as the bud grew (not shown), consistent with Jin and Amberg (Jin and Amberg, 2000).
Bud61-565p may lack functional domains for binding and stimulation of formins (Amberg et al., 1997; Evangelista et al., 1997; Moseley et al., 2005). Indeed, this truncation did not interact with Bni1p by yeast two-hybrid analysis (not shown) and was unable to correct actin organisation in a bud6Δ mutant (Fig. 4B,C) or to restore fully bipolar budding to a bud6Δ diploid strain (not shown). Significantly, live imaging analysis (Fig. 4D) showed that MT shrinkage at the cell cortex, a mode of cortical interaction that is linked to Bud6p (Segal et al., 2002), was restored to wild-type levels in bud6Δ cells by Bud61-565p. Accordingly, bud61-565 suppressed characteristic synthetic spindle orientation phenotypes observed in bud6Δ kar9Δ and bud6Δ dyn1Δ cells (Huisman et al., 2004). Preanaphase spindle orientation in kar9Δ bud61-565 cells was comparable with that in a kar9Δ mutant (Fig. 4E). Similarly, bud61-565 dyn1Δ cells did not accumulate late anaphase cells with both poles of the spindle contacting the bud, contrary to bud6Δ dyn1Δ cells (Fig. 4F). Finally, bud61-565 suppressed premature mitotic exit of bud6Δ dyn1Δ cells in the presence of elongated spindles in the mother cell. Indeed, asynchronous cultures of a bud6Δ dyn1Δ mutant accumulated 12% of cells with multiple SPBs (n=500), a reflection of premature mitotic exit (Huisman et al., 2004; Nelson and Cooper, 2007). By contrast, bud61-565 dyn1Δ cells did not show excess SPBs (n=1500).
Thus, Bud6p may contain a functional domain that promotes cortical capture of MTs away from the region required for promoting actin organisation via formins. To further determine the possible involvement of this region in interaction with MTs, extracts from bud6Δ cells expressing either HA3-Bud6p or HA3-Bud61-565p were prepared to perform MT-binding assays. Both full-length Bud6p and truncated Bud61-565p sedimented with MTs (Taxol-MTs) through a sucrose cushion but not in the presence of free tubulin (Noco-tub) or in the absence of tubulin (Fig. 5A,B). Following sedimentation, treatment of the MT fraction with 8 mM ATP, 10 mM GTP or 0.5 M NaCl did not release Bud6p from MTs whereas 2 M urea caused partial release (Fig. 5C). Thus, Bud6p appeared to bind tightly to MTs in a nucleotide-insensitive manner. Yet, we failed to deplete Bud6p from the supernatant in this assay, indicating that a pool of Bud6p may not readily interact with MTs. Interestingly, full-length Bud6p sedimented with MTs was consistently enriched for high mobility bands as observed by western blot analysis, suggesting that phosphorylation may play a role in controlling these pools (Fig. 5A).
Finally, extracts from bni1CTΔ bnr1Δ cells (Fig. 5D) or from bni1Δ cells (not shown) expressing HA3Bud61-565p yielded a similar result, indicating that Bud6p association with MTs may occur independently of formins.
Both yeast formins influence Bud6p localisation
Here, we addressed whether formins are players in Bud6p-dependent MT capture, by analysing formin mutations for their impact on Bud6p localisation and any associated effects. Bud6p association with the incipient bud is actin independent (Ayscough et al., 1997; Jaquenoud and Peter, 2000), whereas the secretory pathway and actin integrity have been implicated in maintaining an axis of cell polarity and Bud6p delivery that may involve formins (Ayscough et al., 1997; Jin and Amberg, 2000; Pruyne et al., 2004; Irazoqui et al., 2005).
Our data showed that both formins are required for the correct cortical partition of Bud6p. Bnr1p promoted Bud6p association with the bud neck, whereas Bni1p from the bud restricted the arrival of Bud6p at the bud neck until spindle assembly began (Fig. 1).
The involvement of Bni1p was further addressed by comparing bni1Δ with a bni1CTΔ mutant (Lee et al., 1999). Bni1CTΔp is a truncated formin lacking the putative Bud6p-binding site (Evangelista et al., 1997). Previous studies based on overexpressed Bni1p and Bni1CTΔp demonstrated that the truncation promoted actin cable formation in a Bud6p-independent manner in vivo (Sagot et al., 2002). Indeed, the behaviour of Bni1CTΔp at endogenous levels presented here is best explain by the failure of this mutant formin to bind Bud6p. First, there was no synergism between bud6Δ and bni1CTΔ bnr1Δ mutations (Fig. 2), indicating that actin cable formation by Bni1CTΔp was independent of Bud6p. Second, precise colocalisation at the bud cortex was no longer observed between Bni1CTΔp and Bud6p (see Fig. S3 in the supplementary material). Third, Bni1p efficient localisation at the bud cortex depended on both Spa2p and Bud6p. However, Bni1CTΔp (which only contains the putative binding site for Spa2p) (Fujiwara et al., 1998) localised efficiently in bud6Δ cells but not in spa2Δ cells (see Fig. S4 in the supplementary material). As our analysis was performed in live cells with formins expressed at endogenous levels, we could detect a contribution of Bud6p to Bni1p localisation not observed before (Sagot et al., 2002).
Regarding Bud6p localisation, bni1CTΔ behaved like a bni1Δ by exhibiting premature accumulation at the bud neck (Fig. 1). However, both mutants differed in overall actin organisation (Fig. 2). Moreover, Bni1CTΔp was correctly polarised (Fig. 3) and supported cell polarity (Fig. 2). Thus, Bud6p binding to Bni1p may play a crucial role in Bud6p distribution irrespective of actin organisation.
It remains possible that Bud6p accumulation in bni1 mutants partly reflects the importance of Bni1p-driven actin cable organisation from the tip. Likewise, Bnr1p may promote Bud6p transport to the bud neck. Yet, Bud6p association with the bud neck was delayed in relation to Bnr1p (see Fig. S1 in the supplementary material) indicating that Bud6p delivery along actin cables cannot account for the timing of Bud6p recruitment at the bud neck (Segal et al., 2000a). Indeed, localisation of Bud6p may additionally depend on the axial determinant Bud3p that accumulates at the bud neck in late S phase (Segal et al., 2000a; Lord et al., 2000).
Integrating formin functions in spindle orientation via Kar9p and Bud6p
The dual impact of formins on actin organisation and Bud6p distribution highlights their importance in spindle orientation. Kar9p-mediated MT delivery along actin cables and Bud6p accumulation will be responsive to formin mutations (Fig. 6). Here, we showed that cortical domains for MT capture favoured in bni1 or bnr1 mutants paralleled the perturbation in Bud6p accumulation (Fig. 6A). For example, bni1Δ (Segal et al., 2000a) or bni1CTΔ cells exhibited prominent MT interactions with the bud neck matching the accumulation of Bud6p.
In addition, Kar9p function would rely on the remaining set of actin cables in the respective formin mutants (Fig. 6B; see Fig. S6 in the supplementary material). Surprisingly, a bnr1Δ mutant exhibits a slight defect in spindle alignment (Evangelista et al., 2002) and virtually no impairment in Kar9p-mediated delivery (see Fig. S6 in the supplementary material). However, a bnr1Δ mutant is significantly perturbed for vesicle transport within the mother cell (Pruyne et al., 2004), indicating that Kar9p function might be comparatively less sensitive to disruption of bulk actin organisation in the mother cell. Still, spindle retention at the bud neck was compromised in bnr1Δ and bud6Δ cells (Segal et al., 2000a). By contrast, Kar9p function would be severely disrupted once bnr1Δ is combined with bni1CTΔ, reducing overall MT delivery to the bud (Fig. 3E; see Fig. S6 in the supplementary material) owing to excessive and partly disorganised actin cables.
Finally, even though Bud6p and Bni1p act together at the bud tip, the spindle orientation phenotypes of single and double mutants are quite distinct, underscoring the dual contribution of Bni1p to actin organisation and Bud6p localisation. A bud6Δ strain still favours Kar9p-mediated transport of MTs to the bud while reducing MT contacts with the bud neck (Huisman et al., 2004) (Fig. 6C). By contrast, MT contacts with the bud neck in a bni1Δ mutant persist upon KAR9 deletion (Segal et al., 2000a) (Fig. 6D). Deleting instead BUD6 impairs MT delivery to the bud and eliminates the prominent contacts with the bud neck (Segal et al., 2000a).
Bud6p functions in actin organisation and microtubule capture are separable
The fact that Bni1CTΔp organised actin cables independently of Bud6p (Sagot et al., 2002) (and our results) allowed us to explore Bud6p requirement for MT capture beyond formins. A bnr1Δ bni1CTΔ strain still required Bud6p function to support MT growth and shrinkage at the cell cortex (Fig. 3C). This result was confirmed in a bni1Δ mutant assessed during a cell cycle interval in which no formin would be cortically localised (see Fig. S5 in the supplementary material). Therefore, MT capture at Bud6p sites may be independent of its links to formins.
These findings were validated by the characterization of Bud61-565p. This truncation may still interact with Spa2p (Sheu et al., 1998), but would not contain actin- and forming-binding sites (Amberg et al., 1997; Evangelista et al., 1997). Accordingly, Bud61-565p did not support actin organisation (Fig. 4B,C) demonstrating that this truncation was unable to stimulate formins. However, Bud61-565p mediated MT shrinkage at wild-type levels (Fig. 4D), and suppressed genetic interactions between kar9Δ or dyn1Δ and bud6Δ mutations (Fig. 4E,F). Finally, both Bud6p and Bud61-565p co-sedimented with MTs in vitro in support of a role for Bud6p in cortical capture of MTs. Taken together, our analysis showed the existence of separable functional domains of Bud6p, both required for establishment of spindle polarity and orientation. One participating in formin-dependent organisation of actin cables and another mediating astral MT-cortex interactions. The molecular partners for this latter function remain to be elucidated.
Materials and Methods
Yeast strains, plasmids and genetic procedures
Yeast strains used in this study are listed in Table 1 and were isogenic to 15DauA–a ade1 his2 leu2-3,112 trp1-1a ura3Dns arg4, except when indicated. The alleles bud6Δ, bni1Δ (Segal et al., 2000a), bnr1Δ and spa2Δ were generated using KANR cassettes amplified by PCR (Wach et al., 1994). The truncation bni1CTΔ was constructed by disrupting the BNI1 locus with a KANR cassette that eliminated the sequence encoding amino acids 1749-1953 of Bni1p. An analogous bnr1CTΔ allele did not support viability in combination with bni1Δ (not shown) and was not analysed further. A bud6::LEU2 disruption was used when indicated. Deletions were confirmed by PCR analysis. A bnr1Δ bni1-FH2#1ts strain (PY3744) and the control carrying the single bnr1Δ mutation (PY3505) were a gift from D. Pellman. Both strains (Sagot et al., 2002) were isogenic to the wild-type strain S288C (a his3Δ1 leu2 ura3 met15).
pAFS91 or its derivative pOB1 were used to express GFP-Tub1p (Straight et al., 1997). pbud6::LEU2, used for one-step disruption of BUD6, contained a 2 kb sequence of LEU2 between BUD6 flanking 5′ (400 bp) and 3′ (300 bp) sequences generated by PCR as a 2.7 kb BamHI-XhoI fragment in pBluescript(KS). pCFP-TUB1 or its derivative pRS404CFP-TUB1 were used to express a CFP-Tub1p fusion (Jensen et al., 2001). Tagging constructs were derived from pKGFP, pKHA3 or pKCFP (Jensen et al., 2002; Huisman et al., 2004). BUD6 was amplified from yeast genomic DNA from 15DauA. The sequence predicts two changes with respect to Bud6p protein sequence based on the Saccharomyces Genome Database: an insertion of TGA at nucleotide position 699 of the ORF resulting in an additional Asp residue at position 234 in the protein sequence and a Thr to Met substitution at position 482. YIplac211-BUD6tGFP and YIplac128-BUD6tCFP contained a 565 bp EcoRI-NotI fragment generated by PCR for 3′ in-frame fusion of BUD6 to GFP or CFP, respectively. Linearisation with SacI targeted integration at the endogenous BUD6. pRS404 (Sikorski and Hieter, 1989) carrying a 210 bp KpnI-XhoI fragment containing the HIS3 promoter followed by a 780 bp XhoI-BamHI fragment encoding GFP was used to express N-terminally tagged versions of full-length or truncated Bud6p. A 2370 bp sequence corresponding to full-length BUD6, and fragments encoding amino acids 1-565, 1-414, 1-364 or 1-233 flanked by BglII-NotI sites, were generated by PCR for in-frame fusion after GFP. To derive HA3 or CFP-tagging constructs, the GFP sequence was replaced by a XhoI-BamHI fragment encoding either HA3 or CFP. Constructs were linearised with Bsu36I before transformation. YIp211-BNR1tGFP contained a 640 bp sequence corresponding to the 3′ end of the BNR1 ORF as an EcoRI-NotI fragment generated by PCR for 3′ in-frame fusion to GFP. Linearization with BseRI targeted integration at the endogenous BNR1. pRS406-BNI1tGFP contained a 1000 bp SacI-NotI fragment generated by PCR for 3′ in frame fusion of BNI1 to GFP. Linearisation with HindIII targeted integration at the endogenous BNI1. pRS406-BNI1CTΔtGFP contained a 504 bp SacI-NotI fragment generated by PCR in which the NotI site truncated the BNI1 ORF sequence at amino acid position 1749 for 3′ in frame fusion to GFP. Linearisation with HindIII targeted integration at the endogenous BNI1, introducing instead the bni1CTΔ allele fused to GFP. pRS406-KAR9tGFP contained a 495 bp SacI-NotI fragment generated by PCR for 3′ in-frame fusion to GFP. Linearisation with BamHI targeted integration at the endogenous KAR9 (Huisman et al., 2004). Standard yeast genetic procedures and media were used (Guthrie and Fink, 1991). Yeast cultures were grown at 25°C unless otherwise stated. Cells were fixed in 3.7% formaldehyde for 30 minutes and processed for calcofluor or rhodamine-phalloidin staining as previously described (Adams and Pringle, 1991; Pringle, 1991).
Digital imaging microscopy
Time lapse recordings were carried out as previously described (Huisman et al., 2007) using a Nikon Eclipse E800 with a CFI Plan Apochromat 100×, N.A. 1.4 objective, Chroma Technology filter sets and a Coolsnap-HQ CCD camera (Roper Scientific). Five fluorescence images were acquired at a z-distance of 0.8 μm between planes using 2×2 binning along with a single differential interference contrast image in the middle focal plane. When indicated, still images were also obtained by compiling five-plane z-stacks. Images of cells co-expressing GFP and CFP-tagged constructs were obtained by a protocol that discriminates between CFP and GFP using a CFP/YFP filter set (Huisman et al., 2007). Images were processed using Metamorph software and digital overlays were used for scoring.
Modes of astral MT-cell cortex interaction were scored by following the history of individual MTs during the cell cycle interval from mitotic exit to bud emergence. Interactions were categorised as previously described (Carminati and Stearns, 1997; Segal et al., 2002). Interactions by cell compartment during the preanaphase interval (Huisman et al., 2004) were scored using the definition for cortical interaction by Carminati and Stearns (Carminati and Stearns, 1997). Bud neck was defined as the region within a 0.5 μm distance from the point of constriction between the mother and the bud. Measurements in digital images were performed using Metamorph software tools calibrated with a stage micrometer. Linescan analysis for maximal fluorescence (six-pixel width) was carried out along the cell polarity axis to determine the relative fluorescence label between bud cortex and bud neck in cells expressing solely Bud6p-GFP acquired as five-plane z-stacks using a GFP filter set. Fluorescence intensity at the bud neck or bud was measured as mean fluorescence intensity within a fixed box (10×5 pixel) spanning the bud neck or a region including the whole bud excluding the bud neck, respectively. Mean background intensity in the mother cell was subtracted from all measures.
Spindles were scored as oriented if an imaginary line drawn through the long axis of the spindle intersected the bud neck, a measure reflecting both the spindle angle relative to the mother-bud axis and its distance from the bud neck (Theesfeld et al., 1999). Spindle polarity at anaphase was scored as previously described (Huisman et al., 2004).
Extracts were prepared from pellets of 100 ml yeast cell cultures (1-2×107 cells /ml) suspended in PEMap lysis buffer (100 mM K-PIPES pH 6.8, 1 mM MgSO4, 2 mM EGTA, protease inhibitors) (Infante et al., 1999) +0.1% NP40 by disruption with glass beads and centrifugation to obtain a clear lysate. Bovine brain tubulin (Cytoskeleton) was polymerized at a concentration of 7 μg/μl in PEMap containing 1 mM GTP and 40 μM Taxol at 36°C for 5 minutes. The preparation was then passed three times through a 25-gauge needle attached to a 1 ml insulin syringe to generate more MT ends (Taxol-MT). An equal amount of tubulin was incubated instead in the presence of 40 μM nocodazole (Nocodazole-tubulin) to prevent polymerization. Extracts at a final concentration of 1.2 μg/μl in PEMap were incubated at 16°C for 15 minutes in a final volume of 210 μl with one of the following: (1) 1 mM GTP, 40 μM Taxol and 10 μl Taxol-MT; (2) 40 μM nocodazole and 10 μl nocodazole-tubulin; (3) buffer as a control. Samples were then centrifuged through a 10% sucrose cushion (20% for strains expressing HA3Bud61-565p) at 30,000 g for 30 minutes at 16°C. Pellet (5 μl out of a resuspended volume of 30 μl) and supernatant (7.5 μl out of 200 μl) fractions were subjected to western blot analysis by probing with anti-HA monoclonal antibody (1:1000 dilution, Roche) for detection of Bud6p constructs and monoclonal antibody B-5-1-2 (1:1000 dilution, Sigma) to detect α-tubulin.
We thank Sanne Jensen (Department of Cell Cycle and Cancer, Danish Cancer Society, Copenhagen, Denmark) and David Pellman (Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA) for providing strains and plasmids, Gavin Burns for generating part of the BUD6 deletion constructs, and members of the Segal laboratory for many fruitful discussions. N.D. was partly supported by a fellowship from the Fondation pour la Recherche Médicale (France) and C.S.J.L. by the Fundação para a Ciência e a Tecnologia (Portugal). In addition, M.S. acknowledges the support from Cancer Research UK, The Wellcome Trust and the BBSRC.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/121/22/3803/DC1
- Accepted September 1, 2008.
- © The Company of Biologists Limited 2008