The GIN4 family kinase, Cdr2p, acts independently of septins in fission yeast

Septins are a family of GTP-binding proteins originally identified in S. cerevisiae (Hartwell, 1971) as proteins required for cytokinesis. They are now known to exist in many if not all eukaryotes with the exception of plants (reviewed by Faty et al., 2002; Macara et al., 2002). Typically, septins exist in hetero-oligomeric complexes (Field et al., 1996; Frazier et al., 1998; Hsu et al., 1998; Sheffield et al., 2003) and when purified, they have the capacity to form filaments in vitro either alone (Mendoza et al., 2002) or in combination (Field et al., 1996; Frazier et al., 1998; Kinoshita et al., 2002). Although many septins have been implicated in the process of cell division because of their localization to the cell division site and/or because the disruption of their function leads to cytokinesis defects, septins are likely to be involved in additional biological processes if for no other reason than they are present in non-dividing cells such as neurons (reviewed by Faty et al., 2002; Field and Kellogg, 1999).

In S. cerevisiae, septins are thought to perform multiple functions at the mother-daughter bud neck. These include providing a boundary that restricts certain determinants to particular cortical domains (reviewed by Faty et al., 2002) and acting as a scaffold necessary for the proper localization of many factors involved in polarity and cell division (reviewed by Gladfelter et al., 2001). Whether septins perform all of these roles in other organisms has yet to be established.

In S. cerevisiae, septin ring organization is influenced by the functions of three related protein kinases, Gin4p, Kcc4p and Hsl1p (Barral et al., 1999; Longtine et al., 1998; Longtine et al., 2000). Although septin ring disorganization is slight in the absence of any one of them, combinations of deletion mutations lead to worsening phenotypes suggesting that they act together in a septin ring organization step (Barral et al., 1999; Longtine et al., 1998; Longtine et al., 2000). Gin4p, Kcc4p and Hsl1p localize to the bud neck in a septin-dependent manner (Barral et al., 1999; Longtine et al., 1998; Tanaka and Nojima, 1996), and Hsl1p is activated by septin binding (Hanrahan and Snyder, 2003). These protein kinases are involved in a checkpoint response that restrains mitotic progression in the event of septin defects. They prevent Swe1p degradation in the event of septin defects (Barral et al., 1999; Longtine et al., 2000; Ma et al., 1996; Shulewitz et al., 1999). One of two members of this protein kinase family in C. albicans also regulates formation of septin rings (Wightman et al., 2004).

Two relatives of the GIN4 protein kinase family, Cdr1p and Cdr2p, exist in the yeast Schizosaccharomyces pombe but they have not been linked to septin function. S. pombe cdr1 and cdr2 mutations were isolated based on their inability to respond appropriately to nitrogen starvation (Young and Fantes, 1987). cdr mutants are unable to re-set their size control and enter mitosis at a reduced length when faced with limiting nitrogen (Young and Fantes, 1987). It was anticipated that such a screen would identify genes involved in the control of G2/M progression and also those involved in nutritional sensing and monitoring. Indeed, cdr1 was isolated independently as a multicopy suppressor of cdc25-22 and called nim1 (new inducer of mitosis) (Russell and Nurse, 1987). Neither cdr1 nor cdr2 are essential genes but cycling cdr1Δ and cdr2Δ cells are longer than wild type, a phenotype indicative of altered mitotic control (Breeding et al., 1998; Feilotter et al., 1991; Kanoh and Russell, 1998; Russell and Nurse, 1987). Molecularly, Cdr1p phosphorylates and inhibits the Cdk1p inhibiting kinase, Wee1p (Coleman et al., 1993; Parker et al., 1993; Wu and Russell, 1993) and overproduction of Cdr1p causes premature activation of Cdk1p and a `wee' cell phenotype (Feilotter et al., 1991; Russell and Nurse, 1987). Cdr1p has been localized to the cytoplasm of S. pombe cells in no discrete pattern (Wu et al., 1996).

Although genetic analysis has firmly established that Cdr2p functions in G₂ as a negative regulator of Wee1p like Cdr1p (Breeding et al., 1998; Kanoh and Russell, 1998), it is not clear whether Cdr2p influences Wee1p activity directly. Further, in contrast to Cdr1p, overproduction of Cdr2p is lethal generating elongated, highly branched cells that contain two or more septa (Breeding et al., 1998). The lethality of Cdr2p overproduction is independent of an effect on Wee1p function because Cdr2p overproduction in wee1-deleted cells induces multiseptation and branching, although not cell elongation (Breeding et al., 1998). This suggested to us that Cdr2p might influence septin ring dynamics and cytokinesis in a similar manner to that of its S. cerevisiae homologs.

S. pombe septins are organized into a ring late in mitosis and participate in the completion of cytokinesis but not the formation of the actomyosin contractile ring (Berlin et al., 2003; Tasto et al., 2003). S. pombe septins known to be produced during vegetative growth (Spn1p, Spn2p, Spn3p and Spn4p) are not essential genes either alone or in combination (Berlin et al., 2003; Tasto et al., 2003) (H. An and K.L.G., unpublished). In their absence, however, cells are delayed in separation and grow in chains of typically 2-4 cell compartments. In this study, we have probed the potential connection between Cdr2p and the septins to further our understanding of Cdr2p and the factors influencing septin ring dynamics in S. pombe. Our data suggest that Cdr2p and septins are functionally independent. Cdr2p localizes to a wide cortical band overlaying the interphase nucleus that is destroyed by altering the structure of sterol-rich membrane domains. During mitosis, Cdr2p becomes diffusely localized throughout the cell and at septation it forms a thin ring that co-localizes with septins. Cdr2p localization to the cortex and ring is septin-independent and, conversely, septin ring structures are unaffected by the absence of Cdr2p. Structure-function analysis indicates that the non-catalytic C-terminus of Cdr2p is essential for its cortical localization and for its function during a nutritional shift. Overall, our data illustrate the functional divergence of the GIN4 protein kinase family between different species of yeasts with respect to interdependence of function with septins and raise the intriguing question of what defines the medial cortical region of S. pombe such that it binds Cdr2p.

S. pombe strains used in this study are listed in Table 1. Strains were grown in yeast extract medium, minimal medium with appropriate supplements, or minimal medium lacking ammonium chloride as the nitrogen source (Moreno et al., 1991). Crosses were performed on glutamate medium (minimal medium lacking ammonium chloride and containing 0.01 M glutamate, pH 5.6). Tetrad analysis was performed as described (Moreno et al., 1991). The cdr2⁺ and mid1⁺ genes were tagged at their chromosomal loci at the 3′ ends of their open reading frames with sequences encoding green fluorescent protein (GFP), cyan fluorescent protein (CFP), and/or yellow fluorescent protein (YFP) by a PCR-mediated strategy as described previously (Bahler et al., 1998). Proper integration of these epitope cassettes was confirmed by PCR.

All plasmid manipulations and bacterial transformations were by standard techniques. All sequencing was performed using the ThermoSequenase kit as directed (Amersham Pharmacia Biotech, Piscataway, NJ). All PCR reactions were performed using PfuTurbo polymerase (Stratagene, La Jolla, CA) in a PTC-100 programmable thermal controller (PTC-100; MJ Research, Watertown, PA). S. pombe DNA was prepared using the Nucleon MiY yeast DNA extraction kit as directed (Amersham Pharmacia Biotech).

cdr2 mutations and truncations under control of the S. pombe nmt1 (no message in thiamine) thiamine repressible promoter (Maundrell, 1993) (Fig. 6A) were created as follows. pSKcdr2⁺ (pKG1089) was digested with restriction enzymes NdeI and PvuII, releasing a 2184 bp fragment that was subcloned into pREP1, generating pREP1cdr2^Δ730-775 (pKG1050) (Fig. 6A). pKG1089 was digested with restriction enzymes NdeI and EcoRV, releasing a 2033 bp fragment that was subcloned into pREP1, generating pREP1cdr2^Δ679-775 (pKG790). pKG1089 was digested with restriction enzymes NdeI and Eco47III, releasing a 990 bp fragment, which was subcloned into pREP1, generating pREP1cdr2^Δ332-775 (pKG1593). Primer life del 5′-CCATTTGGCTACAGAATTCCAGCAAGCTTCGGCATCCAGACCTG-3′ (Operon Technologies; Alameda, CA) was used to generate an overexpression construct, pREP1cdr2^Δ730-746 (pKG1256). To generate a cdr2^Δ1-252 construct, the cdr2 NcoI site at position 755 in pKG1089 was replaced with NdeI by site-directed mutagenesis (Chameleon Double Stranded Mutagenesis Kit; Stratagene) using primer ncoswap 5′-ACTCGAATTCATATGGAACAA-3′ (Integrated DNA Technologies, Coralville, IA). The resulting construct, pKG1806, was digested with NdeI and EcoICR, releasing a 2761-bp fragment that was subcloned into pKG113, generating pREP1cdr2^Δ1-250 (pKG1593) (Fig. 1G). pSKcdr2^E177A (pKG1580) was generated from pKG1089 by site-directed mutagenesis (Chameleon Double Stranded Mutagenesis Kit; Stratagene) using primer cdr2E177A 5′-CATTACGCATCGCCGGCAATTATTATG-3′ (Integrated DNA Technologies). For localization and some overexpression studies, cdr2⁺, mutations of cdr2⁺ and fragments of cdr2⁺ were amplified from appropriate pSK plasmids using a 5′-end primer containing an NdeI site and a 3′end primer containing a BamHI site and subcloned into the pREP81GFP and the pREP1 plasmid. DNA sequencing confirmed the presence of all fragments and mutations.

To determine if the cdr2 constructs could rescue the nitrogen deprivation defect of the cdr2 null strain, the cdr2::ura4⁺ura4-D18 leu1-32 (KGY1519) strain was transformed with the various mutations expressed under control of the nmt1 promoter and the transformed strains were grown to mid-log phase at 32°C in minimal medium containing thiamine, collected on filters, released into minimal medium containing thiamine but lacking nitrogen, and incubated at 32°C for 48 hours. Rescue was determined by size. Samples were collected for immunoblotting and for documentation by phase microscopy.

Some light and fluorescence microscopy was performed on a Zeiss Axioskop20 microscope with appropriate filters, and images were captured and documented using a Optronics ZVS-47DEC image-capture system. To visualize DNA and/or cell and septal material, cells were fixed in 70% ethanol and stained with DAPI (Sigma) or Calcofluor (Sigma) as described (Balasubramanian et al., 1997). GFP, YFP or CFP-tagged proteins were visualized in live cells or in cells fixed with ethanol. Filipin (Polysciences, Warrington, PA) was dissolved in DMSO and added to the culture at a final concentration of 5 or 10 μg/ml. Cells were either imaged immediately after filipin addition or incubated for 1 hour prior to imaging to cause disruption of membrane domains (Wachtler et al., 2003). For detecting GFP-labeled proteins, microscopy was performed using a spinning disk confocal microscope (Ultraview LCI, PerkinElmer) and Ultraview LCI software (v5.2; PerkinElmer) for image acquisition. Images were processed using Velocity software (v1.4.2; Improvision). Z-series optical sections were taken at 0.5 μm spacing for 3D reconstructions.

To examine the localization of Cdr2p, its endogenous locus was modified to encode Cdr2p-GFP and live cdr2-GFP cells were imaged. Cdr2p-GFP was detected in three distinct patterns (Fig. 1A): a broad centrally placed band of speckles (cell a); no discrete localization (cell b); and a thin medial ring (cell c). Both the band and the ring of Cdr2p-GFP were cortical as determined by rotating the images to provide an en face view (Fig. 1B). To determine how the localization patterns of Cdr2p relate to the cell cycle, a cdr2-GFP sid4-GFP strain was constructed and imaged (Fig. 1C). Sid4p is a spindle pole body protein (Chang and Gould, 2000) and a useful marker of progression into and through mitosis. Cdr2p-GFP was present as a central broad band until cells separated their SPBs during mitosis (Fig. 1Ca,b). During metaphase and anaphase, no discrete Cdr2p-GFP localization could be detected (Fig. 1Cc,d). This is not due to the absence of Cdr2p because Cdr2p levels are constant through the cell cycle (Kanoh and Russell, 1998). Thus, we reason that at this time, Cdr2p might be in transit to a ring that forms at the end of anaphase (Fig. 1Ce). This ring appears not to be the equatorial microtubule organizing center (eMTOC) since it was observed in mbo1::ura4⁺ cells (data not shown), which lack an eMTOC (Venkatram et al., 2004; Sawin et al., 2004). Therefore it is likely to be the actomyosin contractile ring. Following septation, Cdr2p once again localized to a broad band of speckles centered over the two daughter nuclei (Fig. 1Cf).

The only described protein localization pattern in S. pombe similar to what we observe for Cdr2p is that of Mid1p. Mid1p resides in the nucleus during much of interphase but prior to mitosis it exits the nucleus and forms a broad band overlaying the nucleus that eventually coalesces into a tight ring. In the absence of Mid1p, actin rings and septa are misplaced (Chang et al., 1996; Sohrmann et al., 1996). Thus, we examined whether Cdr2p and Mid1p co-localized. During most of interphase, Mid1p is present within the nucleus while Cdr2p is present in a broad medial band as described above (Fig. 2Aa). Although distinct, these patterns appear to overlap when flattened images of 3D reconstructions are shown since Cdr2p overlies Mid1p at this cell cycle point (Fig. 2Aa). When Mid1p exits the nucleus and forms a cortical band, there is very little co-localization with Cdr2p (Fig. 2Bb). At the time Mid1p forms a tight ring, Cdr2p localization is not detected in any discrete pattern (Fig. 2Bc). Only when Cdr2p also forms a ring are Mid1p and Cdr2p seen to significantly co-localize (Fig. 2Ad). Mid1p was able to localize to both the broad band and the tight ring structures in the absence of Cdr2p (Fig. 2B). Similarly, Cdr2p did not depend upon Mid1p for its localization to speckles or to a ring structure (Fig. 2C). We conclude then, that Cdr2p and Mid1p localizations to the medial region are independent of one another.

The restriction of Cdr2p to a broad medial band during interphase prompted us to examine whether altering membrane domain organization affected Cdr2p localization. In S. pombe, the distribution of sterols in the membrane is cell-cycle regulated, with enrichment at growing cell tips and at the site of septation (Wachtler et al., 2003). Thus, during interphase, sterol-rich domains are found at the ends of the cells whereas Cdr2p is restricted to the medial region. The integrity of these sterol-rich membrane domains can be disrupted by extended incubation with filipin, an antibiotic that complexes with 3-β-hydroxysterols (Wachtler et al., 2003). In cdr2-GFP cells treated with 5 μg/ml filipin for 1 hour, Cdr2p localization became more punctate and was less confined to the medial region of the cell, compared with a solvent control (Fig. 3A). Higher concentrations of filipin (10 μg/ml) produced even more dramatic disruption of Cdr2p-GFP localization such that medial localization was nearly absent (Fig. 3A). Thus, we conclude that membrane domain organization is an important factor in Cdr2p localization during interphase. We also examined whether the length of the cortical Cdr2p band was proportional to the length of the cell. To determine this, we examined Cdr2p-GFP band length in a cdc25-22 strain shifted to 36°C for 0 and 3.5 hours. As these cells were shifted to the non-permissive temperature, they elongated from their length at birth of ∼10 μm up to an average of 32 μm. The length of the Cdr2p-GFP band increased proportionately remaining at approximately 30% the length of the cell (n=20 cells for 0 and 3.5 hours) (Fig. 3B). This observation suggests that the zone of Cdr2p localization is determined by cell length.

The Cdr2p localization pattern described above appeared to differ significantly from that of septins and this was confirmed by imaging a cdr2-YFP spn1-CFP strain where no overlap was detected (Fig. 4A). Nevertheless, we analyzed whether there was any interdependence of Cdr2p and septins for localization or function. To examine whether Cdr2p depended upon an intact septin ring for localization, Cdr2p-GFP was examined in a spn4Δ mutant. In these cells, none of the septins are localized to the medial region (H. An and K.L.G., unpublished) and cells exhibit a delay in cell separation (Berlin et al., 2003; Tasto et al., 2003). Cdr2p-GFP localization was normal in these cells including its ability to localize to a ring structure (Fig. 4B). Conversely, septin rings were normal in the absence of cdr2⁺ (Fig. 4C).

Overproduction of cdr2⁺ is lethal; it prevents cell separation giving rise to long chains of cells, each containing a single nucleus (Breeding et al., 1998). Thus, we examined whether Cdr2p overproduction disrupted septin localization. Even after 22 hours of Cdr2p overproduction, septin rings formed normally and appeared to remain at the cortex of cells although, in some cases, Spn3p spread away from the site of septum formation and was also observed at cell tips (Fig. 4D). Cdr2p overproduction had the same phenotypic consequence in cells lacking septins. It was lethal (Fig. 5A) and gave rise to highly branched and septated cells (Fig. 5B).

Cdr2p and Cdr1p functions are required for S. pombe cells to respond normally to nitrogen deprivation (Young and Fantes, 1987). Cells lacking either of these protein kinases are able to divide only once during nitrogen deprivation, and arrest in G₂ as elongated cells in contrast to wild type cells deprived of nitrogen that arrest in G₁ as short cells (Breeding et al., 1998; Kanoh and Russell, 1998). If septins were required for the function of either of these kinases, cells lacking septins would be expected to respond inappropriately to nitrogen deprivation. This was not the case. spn4Δ cells lack septin rings and fail to separate efficiently (Berlin et al., 2003; Tasto et al., 2003). However, they underwent the normal starvation response, arresting as small rounded cells (Fig. 5C). Thus, we conclude that Cdr2p and Cdr1p are functionally independent of septins.

Cdr2p shares a similar overall architecture with Gin4p and related kinases. It has a N-terminal catalytic domain and a C-terminal extension that by analogy with its family members is likely to regulate its kinase activity. To determine what portion(s) of Cdr2p is required for localization and function, a series of cdr2 mutations were made (Fig. 6A). For localization studies, the mutants were produced as GFP fusion proteins. When GFP-Cdr2p was expressed from the weak nmt81 promoter, an identical staining pattern to endogenously C-terminally tagged Cdr2p was obtained (Fig. 6B). Mutating the critical lysine residue in the Cdr2p catalytic domain to alanine, a change that severely impairs its protein kinase activity (Kanoh and Russell, 1998), had no effect on the ability of GFP-Cdr2p to localize normally (data not shown). Because K39A retains some protein kinase activity (Kanoh and Russell, 1998) and biological activity (C.B.N. and K.L.G., unpublished), a second mutation was made within the catalytic domain, E177A, which is predicted to abolish Cdr2p protein kinase activity (Hanks and Hunter, 1995). Cdr2p(E177A) also localized correctly (Fig. 6C) suggesting that kinase activity is not required for its proper targeting. By contrast, truncation of C-terminal residues abrogated the ability of Cdr2p to localize to the cortex. While GFP-Cdr2pΔ747-775 was able to localize correctly (Fig. 6D), neither GFP-Cdr2p^Δ679-775 nor GFP-Cdr2p^Δ730-775 (lacking 46 and 96 amino acids, respectively, from the C-terminus; Fig. 6A) was able to localize although both proteins were made (data not shown). Because loss of just 46 amino acids abrogated localization, the localization of the GFP-Cdr2p^Δ232-775 that lacks most of the C-terminal domain was not tested.

Given that the C-terminus was essential for Cdr2p localization, we asked whether C-terminal residues were sufficient to target GFP to the medial cortex of cells. GFP-Cdr2p(Δ1-262) was observed at all the positions normally occupied by the full-length protein but it was also observed prominently at cell ends, a localization not observed for the full-length protein (Fig. 6E). Taken together, these data indicate that (1) the C-terminus contains the Cdr2p cortical targeting domain and (2) the catalytic domain contributes to restricting the Cdr2p cortical localization to the medial region.

We next examined whether Cdr2p protein kinase activity and/or the C-terminus were important for eliciting the cdr2⁺ overexpression phenotype, which is an extreme branching and cell separation defect. Overproduction of Cdr2p(K39A) or Cdr2p(E177A) induced the same lethal phenotype as Cdr2p suggesting that this phenotype results from the titration of a factor(s) necessary for cell separation rather than hyperphosphorylation of such a factor (Fig. 6F and data not shown). While protein kinase activity is not necessary for Cdr2p to produce this branching phenotype, the catalytic domain is required (Fig. 6G). This suggests that the catalytic domain of Cdr2p has the capacity to interact stably with a protein(s). This interpretation is supported by the observation that overproduction of the catalytic domain in isolation also generates a phenotype consistent with its ability to titrate a factor(s) required for cell cycle progression (Fig. 6H).

We next examined whether the ability to localize correctly correlated with Cdr2p function. cdr2⁺ expressed from the repressed nmt1 promoter (low level expression) is able to rescue the nitrogen starvation defect of cdr2Δ cells (Breeding et al., 1998) (Fig. 7B) and we used this response as a measure of cdr2⁺ function. Truncation of the last 28 amino acids did not affect Cdr2p function (Fig. 7C). Indeed, this fragment corresponds to the original rescuing clone (Breeding et al., 1998). The constructs that were compromised in their ability to localize correctly were unable to rescue the nitrogen deprivation defect of the cdr2 null strain (Fig. 7D-G). Further, although not previously tested (Breeding et al., 1998; Kanoh and Russell, 1998), protein kinase activity is required for Cdr2p function since expression of cdr2-E177A did not rescue this defect (Fig. 7H). Thus, the functions of Cdr2p are probably performed at the cortex of the cell through phosphorylation of a target protein(s).

The Cdr2 protein kinase acts as a mitotic inducer in S. pombe and has an additional role in cytokinesis (Breeding et al., 1998; Kanoh and Russell, 1998). Here we have extended our characterization of Cdr2p and find that it localizes dynamically between two sites during the cell cycle. Both its protein kinase activity and residues within its C-terminal non-catalytic domain are essential for its function. Further, unlike members of the Gin4p protein kinase family in S. cerevisiae, Cdr2p localization and function are independent of septins and, conversely, S. pombe septins function independently of Cdr2p.

In S. cerevisiae, all members of the Gin4p protein kinase family localize to the bud neck in a septin-dependent manner and septin rings are disorganized in their absence (Barral et al., 1999; Bouquin et al., 2000). By contrast, the primary localization pattern of Cdr2p is a broad cortical band overlying the nucleus that is independent of septins. Since Cdr2p functions during G2 to induce mitosis, it is tempting to speculate that this is its primary site of action. In support of this, Cdr2p remained localized at the cortex for at least 2 hours after cells had been deprived of nitrogen (our unpublished observations). Cdr2p is also localized briefly to the medial cytokinetic ring at the end of mitosis, also in a septin-independent manner. Distinct targeting domains within the Cdr2p C-terminus for the cortical band and medial ring were not revealed during the course of our analyses. This raises the possibility that a single targeting motif directs the protein to both localizations. Further, since cells lacking Cdr2p function are defective for cytokinesis at elevated temperatures in a Wee1p-independent fashion (Breeding et al., 1998), Cdr2p might contribute to the process of cytokinesis when localized at the medial ring. This might also indicate that Cdr2p has more than one target.

The C-terminus of Cdr2p contains a cortical targeting domain. However, we have been unable to define it precisely. Further truncations abolish its ability to localize correctly and there is the additional caveat that the catalytic domain is important for its more restricted localization pattern. Gin4p has been found to dimerize (Mortensen et al., 2002) and therefore it is possible that some form of Cdr2p oligomerization must occur prior to its appropriate localization. Interestingly, we noticed that the C-terminal region of Cdr2p contains a region (amino acids 669-740) with significant sequence similarity to amino acids 965-1036 of S. cerevisiae Kcc1p. This region may be particularly important for an aspect of Cdr2p function because further truncation of the Cdr2p C-terminus from residue 747 to 730 correlated with the loss of its ability to function in the rescue of the cdr2 null strain. It will be interesting to test in future experiments whether this region is involved in dimerization or interaction with another factor.

Perturbing membrane organization with filipin indicated that membrane composition is critical for defining and maintaining the medial zone of Cdr2p localization. However, it remains to be determined whether Cdr2p binds directly to specific membrane components concentrated in this region or rather to peripheral membrane proteins. By examining the zone of Cdr2p staining in cells of different lengths, we have found that Cdr2p localizes to the middle 30% of the cell cortex in interphase cells. Further, our data show an inverse relationship between regions that stain with filipin, and are therefore rich in sterols, and those to which Cdr2p localizes. Interestingly, filipin staining domains expand inwards from the tips in cdc25-22 cells as they arrest and elongate (Wachtler et al., 2003) and as the band of Cdr2p expands outward. Thus, the sterol-rich domain may restrict Cdr2p to the medial portion of the cell.

In addition to establishing the independence of Cdr2p on septins, we have found that septins are independent of Cdr2p since septin localization and function were unaffected by the absence or the overproduction of Cdr2p. Although in S. cerevisiae significant abnormalities in septin rings are only evident when multiple copies of the GIN4 family are missing (Barral et al., 1999; Longtine et al., 1998; Longtine et al., 2000), Cdr1p is unlikely to contribute to septin ring formation in the absence of cdr2⁺ because the cdr1Δ cdr2Δ double mutant is longer than either single mutant but does not display cell separation defects (Kanoh and Russell, 1998) that would be expected if septin function were perturbed.

Like Cdr2p, Gin4p and related kinases have been implicated in the regulation of mitotic progression and cytokinesis in S. cerevisiae. Gin4p interacts with the Cdc28p-Clb2p complex (Altman and Kellogg, 1997; Okuzaki et al., 2003) and septins (Carroll et al., 1998; Longtine et al., 1998) and phosphorylates the Shs1p septin (Mortensen et al., 2002). Hsl1p, also called Nik1p, was first shown to regulate Swe1p by phosphorylation (Ma et al., 1996; Tanaka and Nojima, 1996) and has also been implicated in Swe1p degradation (McMillan et al., 1999; Shulewitz et al., 1999) although the role of phosphorylation in Swe1p degradation during G2/M has been contested (Harvey and Kellogg, 2003). Hsl7p functions as an adapter protein in Swe1p regulation, binding to both Hsl1p and Swe1p (Shulewitz et al., 1999). Biochemical data collected to date suggest that Cdr2p regulates Wee1p and it might do so directly since it binds and phosphorylates an N-terminal Wee1p fragment (amino acids 11-252) (Kanoh and Russell, 1998). Since Wee1p levels oscillate through the cell cycle (Aligue et al., 1997; Booher et al., 1993; Sia et al., 1998) it is likely that Wee1p levels are regulated in a manner similar to Swe1p. It will be interesting to see whether Cdr2p regulates Wee1p phosphorylation and/or levels in the medial region of the cell. If so, the role of this protein kinase family in mitotic control might be conserved despite the differences in their integration with septin functions. As of yet, however, there is no indication that Wee1p is ever targeted to the medial cortical region (Wu et al., 1996). So, while much remains to be learned about Cdr2p's role in regulating G2/M progression in S. pombe, understanding its intracellular localization is a significant step forward.

We are grateful to Sapna Mehta, Liping Ren and Srinivas Venkatram for performing some of the cell imaging presented in this study. This work was supported by NIH grant GM47728 to K.L.G., who is an investigator of the HHMI.