Anillin is a conserved protein required for cell division (Field, C.M., and B.M. Alberts. 1995. J. Cell Biol. 131:165–178; Oegema, K., M.S. Savoian, T.J. Mitchison, and C.M. Field. 2000. J. Cell Biol. 150:539–552). One fission yeast homologue of anillin, Mid1p, is necessary for the proper placement of the division site within the cell (Chang, F., A. Woollard, and P. Nurse. 1996. J. Cell Sci. 109(Pt 1):131–142; Sohrmann, M., C. Fankhauser, C. Brodbeck, and V. Simanis. 1996. Genes Dev. 10:2707–2719). Here, we identify and characterize a second fission yeast anillin homologue, Mid2p, which is not orthologous with Mid1p. Mid2p localizes as a single ring in the middle of the cell after anaphase in a septin- and actin-dependent manner and splits into two rings during septation. Mid2p colocalizes with septins, and mid2Δ cells display disorganized, diffuse septin rings and a cell separation defect similar to septin deletion strains. mid2 gene expression and protein levels fluctuate during the cell cycle in a sep1- and Skp1/Cdc53/F-box (SCF)–dependent manner, respectively, implying that Mid2p activity must be carefully regulated. Overproduction of Mid2p depolarizes cell growth and affects the organization of both the septin and actin cytoskeletons. In the presence of a nondegradable Mid2p fragment, the septin ring is stabilized and cell cycle progression is delayed. These results suggest that Mid2p influences septin ring organization at the site of cell division and its turnover might normally be required to permit septin ring disassembly.
Cytokinesis is the final stage of mitosis, during which a cell is irreversibly split into two daughters. In many eukaryotic organisms, an essential element of this process is an actomyosin-based contractile ring whose constriction is required to generate the force necessary to complete cell division. The genetically tractable yeast Schizosaccharomyces pombe is an excellent organism for studying cytokinesis, as it divides by medial fission after assembly and contraction of an actomyosin ring (for review see Feierbach and Chang, 2001). Based on the isolation and characterization of numerous mutants, the process of cell division in S. pombe can be divided into several steps.
The initial phase of cytokinesis involves establishing the division site at the center of a symmetrical cell. This process begins at the onset of mitosis when components of the contractile ring are recruited to the cell cortex immediately adjacent to the nucleus (for review see Chang, 2001). The proper localization of these medial ring components depends on Mid1p (Chang et al., 1996; Sohrmann et al., 1996). The next steps of cytokinesis, medial ring constriction and septation, require a conserved signaling cascade, referred to as the septation initiation network, that is organized at the spindle pole body, a microtubule organizing center analogous to the mammalian centrosome (for review see McCollum and Gould, 2001; Pereira and Schiebel, 2001). Septation occurs concomitantly with medial ring constriction as the primary septum is laid down in a centripetal manner behind the constricting actomyosin-based ring, which is then flanked on each side by secondary septa (for review see Le Goff et al., 1999). Physical separation of the two daughter cells is achieved, in part, through the degradation of the primary septum, although the mechanistic details of this process remain obscure.
In S. pombe, septins are clearly required for normal cell separation because deletion of septin genes expressed during vegetative growth generates a nonlethal chained cell phenotype (J. Pringle, personal communication; Longtine et al., 1996). Septins are a group of conserved GTPases originally identified in budding yeast (Hartwell, 1971) and now described in many organisms (for review see Faty et al., 2002; Macara et al., 2002). In Saccharomyces cerevisiae, septins are thought to perform multiple functions at the mother–daughter bud neck, such as providing a boundary that restricts certain determinants to particular cortical domains (for review see Faty et al., 2002) and acting as a scaffold necessary for the proper localization of many factors involved in polarity and cell division (for review see Gladfelter et al., 2001). Whether septins perform all of these roles in other organisms has yet to be established. However, most septins studied to date localize to the site of cell division, and disruption of septin function leads to cytokinesis defects in many cell types (Hartwell, 1971; Neufeld and Rubin, 1994; Longtine et al., 1996; Kinoshita et al., 1997).
One factor important for septin organization is the conserved protein anillin, which is concentrated in the cleavage furrow of dividing cells where it links the actin and septin cytoskeletons (Field and Alberts, 1995; Oegema et al., 2000; Kinoshita et al., 2002). Further, the disruption of anillin function leads to defects in cytokinesis (Oegema et al., 2000). In S. pombe, there is no evidence that the one described anillin homologue, Mid1p, interacts with the septin cytoskeleton. Rather, the loss of Mid1p function results in misplaced actomyosin rings and septa (Chang et al., 1996; Sohrmann et al., 1996; Bahler et al., 1998a; Paoletti and Chang, 2000), whereas a septin deletion results in a cell separation defect (J. Pringle, personal communication; Longtine et al., 1996). Thus, these proteins appear to be involved in different stages of cell division.
Examination of the recently completed S. pombe genome sequence indicated the presence of a second ORF with significant homology to Hs anillin and mid1 that we have termed mid2 (Q9P7Y8 or SPAPYUG7.03C). In this study, we have examined the functional relationship of Mid2p with both Mid1p and the septins. Mid2p is regulated in abundance during the cell cycle, and we have examined the reasons for its periodicity. Our results suggest that Mid2p plays a role during late mitosis in organizing the septin ring and that Mid2p degradation may facilitate septin ring disassembly.
S. pombe mid2 encodes a second anillin homologue
Examination of the S. pombe protein database identified an uncharacterized protein with homology to Mid1p that has been termed Mid2p (Q9P7Y8) (Fig. 1
A). Mid2p has a predicted molecular mass of 78.7 kD. The overall similarity between Mid2p and either Mid1p or Hs anillin is 24.3% and 19.5%, respectively. Mid2p also shares moderate sequence similarity with two other fungal proteins, S. cerevisiae Bud4p (14.9%; Sanders and Herskowitz, 1996) and Candida albicans Int1p (12.8%; Gale et al., 1996). All of these proteins possess a carboxy-terminal pleckstrin homology (PH)*
Abbreviations used in this paper: APC, anaphase-promoting complex; LatA, latrunculin A; PEST, Pro/Glu/Ser/Thr; PH, pleckstrin homology; SCF, Skp1/Cdc53/F-box.
Sporulation and tetrad dissection of diploids that were confirmed to have the entire mid2 ORF successfully replaced by ura4+ produced four viable colonies, indicating that mid2 is not essential for viability. In contrast to the misplaced septum phenotype observed in mid1Δ cells, the loss of mid2 produced “chained” cells in which repeated rounds of nuclear division and septum synthesis occurred with delayed cell separation (Fig. 1 B). To determine if Mid1p and Mid2p share an overlapping essential function, a mid1Δ mid2Δ strain was constructed. This strain was viable at 25°C, exhibited both septum placement and cell separation defects (Fig. 1 B), and was temperature sensitive (unpublished data), as is the mid1Δ strain alone. Combined with the fact that both Mid1p–GFP and Mid2p–GFP were able to localize properly in the absence of the other (unpublished data), we conclude that mid1 and mid2 are not functionally redundant.
Mid2p localization and abundance are cell cycle regulated
To better understand the role of Mid2p in cell separation, its subcellular localization was examined. A carboxy-terminal Mid2p–GFP fusion protein was constructed by homologous recombination at the endogenous locus of mid2+ in order to maintain its physiological expression level. Mid2p–GFP localized to the medial region in 17% of the cells in an asynchronous population (Fig. 2
A). Using time-lapse microscopy, Mid2p–GFP was observed to initially form a single ring, split into two rings as the septum formed, and then disappear as cells separated (see Video 1). Mid2p–GFP was stable to a variety of fixation procedures, thereby permitting it to be visualized in fixed cells also stained with DAPI and antibodies to tubulin. In this case, Mid2p was only detected at the ring in late anaphase cells that lacked spindles (Fig. 2 B; unpublished data).
Because Mid2p–GFP localization was transient, we asked whether Mid2p levels were cell cycle regulated. A Mid2p–HA strain was constructed and used to examine protein levels in a culture synchronized by centrifugal elutriation. Mid2p–HA levels fluctuated considerably as cells traversed the cell cycle, peaking concomitantly with septation (Fig. 2 C). Also, Mid2p–HA appeared to migrate as several distinct species (Fig. 2 C). These multiple isoforms are due to phosphorylation because treatment of immunoprecipitated Mid2p–HA with λ-phosphatase collapsed them into a single, faster-migrating species (Fig. 2 D).
To determine if Mid2p periodicity was due, at least in part, to fluctuations in its mRNA, mid2 transcript levels were monitored from a synchronous culture of wild-type cells harvested by centrifugal elutriation. mid2 mRNA levels did vary significantly during vegetative growth, peaking as cells entered mitosis (Fig. 2 E).
Mid2p is destroyed by Skp1/Cdc53/F-box (SCF)–dependent proteolysis
Because of the dynamic changes in Mid2p–HA levels during the cell cycle (Fig. 2 C), we tested if Mid2p was targeted for destruction by ubiquitin-mediated proteolysis. First, we examined whether Mid2p was ubiquitinated in vivo. In an mts3-1 mutant strain, multi-ubiquitinated conjugates accumulate at the restrictive temperature due to the abrogation of proteosome function (Gordon et al., 1996). Therefore, polyubiquitinated Mid2p–Myc would be expected to accrue in an mts3-1 strain, but only if the relevant E3 ubiquitin ligase was active. S. pombe securin, Cut2p, was used as a positive control for a ubiquitin-conjugated protein (Berry et al., 1999). Both Cut2p–Myc and Mid2p–Myc were readily detected as multi-ubiquitinated proteins in isolates of ubiquitin-conjugated proteins from an mts3-1 strain overproducing tagged ubiquitin (Fig. 3
A, lanes 1 and 7).
Two ubiquitin ligases that participate in cell cycle–regulated proteolysis are the anaphase-promoting complex (APC) and the Skp1/Cdc53/F-box (SCF) (Willems et al., 1999; Zachariae and Nasmyth, 1999). To determine if either of these E3 ligases were responsible for Mid2p–Myc ubiquitination, we took advantage of available conditional lethal mutations that disrupt the function of each. Lid1p is a component of the APC that acts during mitosis and G1 (Peters, 2002), and a lid1-6 mutation prevents APC function at the restrictive temperature (Berry et al., 1999). Skp1p is a core member of the SCF complex, an E3 required for G1/S progression (Kominami et al., 1998; Krek, 1998; Willems et al., 1999), and the skp1-A4 allele abrogates SCF activity at 36°C (Yamano et al., 2000). Whereas the ubiquitination of Cut2p–Myc relied upon an active APC (Fig. 3 A, lanes 1 and 2), Mid2p–Myc did not (Fig. 3 A, lanes 7 and 8). Rather, Mid2p–Myc ubiquitination was strictly dependent upon functional Skp1p (Fig. 3 A, lanes 7 and 9), whereas Cut2p–Myc proteolysis was SCF independent (Fig. 3 A, lanes 1 and 3). An anti-ubiquitin immunoblot confirmed that ubiquitin conjugates were purified in all strains except the one lacking the His–ubiquitin vector (Fig. 3 B).
Consistent with the in vivo ubiquitination assay data, Mid2p–HA levels accumulated in an mts3-1 strain (Fig. 3 C) but not in an APC mutant at the restrictive temperature (unpublished data). Mid2p–HA abundance was also elevated in cells lacking two F-box–encoding genes, pop1 and pop2, components of SCF ubiquitin ligases (Fig. 3 C) (Kominami et al., 1998). Taken together, these results are consistent with the hypothesis that Mid2p is destroyed by SCF-dependent ubiquitination in vivo.
mid2 expression is dependent on the forkhead-like transcription factor sep1
The fact that mid2 mRNA levels oscillate during vegetative growth (Fig. 2 E), coupled with the observation that the mid2Δ cell separation defect resembles the loss of a putative transcription factor, sep1 (Fig. 4
A; Ribar et al., 1999), prompted us to examine whether mid2 transcript abundance depended on Sep1p function. A cdc25-22 block and release synchronization protocol (Moreno et al., 1990) was used to address this question because the sep1Δ strain cannot be synchronized by centrifugal elutriation due to its chained cell phenotype (Ribar et al., 1999). In these cells, progression through mitosis was monitored by examining fluctuations in Cdc13p–cyclin B levels that occur during this period (Alfa et al., 1989). Although readily detectable in cdc25-22 cells proceeding through mitosis (Fig. 4 B), Mid2p–HA levels were clearly diminished in sep1Δ cells during the same time course (Fig. 4 C). The drop in Mid2p levels correlated with a significant decrease in mid2 mRNA abundance in the absence of sep1, although delayed and reduced amounts of the mid2 transcript were detected (Fig. 4, D and E). These results indicate that Sep1p plays an important role in setting Mid2p levels.
To address whether mid2+ might be the critical target of Sep1p, we tested whether expression of mid2+ from a heterologous promoter was able to suppress the cell separation defect of a sep1Δ strain. At best, we observed a partial rescue (unpublished data), indicating that other targets of Sep1p must exist that, together with Mid2p, are required for efficient cell separation in S. pombe.
Medial ring localization of Mid2p–GFP requires F-actin
To determine if Mid2p required F-actin for its association with the medial ring, Mid2p–GFP localization was monitored in cells released from a cdc25-22 block into either DMSO or latrunculin A (LatA), which promotes F-actin depolymerization (Ayscough et al., 1997). In control cells, the first Mid2p–GFP rings appeared at 60 min and persisted (Fig. 5
A; unpublished data). In contrast, the LatA-treated cells failed to form Mid2p–GFP rings even after 150 min (Fig. 5 A). In an asynchronous culture of cells, however, Mid2p–GFP persisted at the medial ring 15 min after LatA addition (Fig. 5 B), suggesting that once recruited, Mid2p's association with the medial ring is actin independent.
Mid2p is required for proper septin ring organization
Because the loss of mid2 produced a strikingly similar phenotype to that of the spn4Δ mutant (Fig. 5 C) (J. Pringle, personal communication; Longtine et al., 1996), and the phenotype of a mid2Δ spn4Δ double mutant strain was similar to either single mutant (Fig. 5 C), it seemed likely that Mid2p and septins were involved in the same step of cell division. We therefore examined if Mid2p depended upon septins for its medial ring association. In spn4Δ cells, Mid2p–GFP was expressed (unpublished data) but failed to localize to the medial ring and was instead distributed throughout the cytoplasm (Fig. 5 D). We next determined whether or not septins and Mid2p colocalized. Images taken from a strain expressing endogenously tagged Mid2p–GFP and Spn1p–CFP show that their localization patterns are indistinguishable (Fig. 6
A). We also tested whether septin localization was affected by the lack of Mid2p function. To perform this experiment, the spn3 locus was modified to encode an Spn3p–GFP fusion protein. This strain was morphologically wild type, implying that the epitope does not disrupt septin function. By time-lapse microscopy, Spn3–GFP was observed to form a single ring, then form two rings, and finally disperse upon the completion of cell separation (Fig. 6 B; see Video 2). In a mid2Δ strain, Spn3p–GFP was recruited to the site of cell division in a loosely organized ring and then appeared to spread bilaterally across the septum as a disc over time (Fig. 6 C; see Video 3).
Because septa persist in a mid2Δ strain, we wondered whether the abnormal Spn3p–GFP localization was a direct consequence of the loss of mid2 or a byproduct of the cell separation defect in these cells. Therefore, we examined Spn3p–GFP localization in a different mutant that also displays multiple uncleaved septa. In the calcineurin deletion, ppb1Δ, which displays this phenotype (Yoshida et al., 1994), Spn3–GFP assembled into tight rings at the cell cortex (Fig. 6 D), as did Mid2p–GFP (Fig. 6 E). This observation indicates that septin rings are not necessarily disorganized in cell separation mutants but are specifically affected by the absence of Mid2p.
Delineation of Mid2p functional domains
To determine which regions of the protein dictated Mid2p localization and were important for its function, a series of mid2 constructs were generated and tested for their ability to rescue the mid2Δ cell separation defect, their localization pattern, and any overexpression phenotype (Fig. 7
B). Immunoblot analysis of anti-HA immunoprecipitations from denatured lysates confirmed that these fragments were all produced (Fig. 7 C).
The PH domain of Mid2p encompasses amino acids 582–685 (Fig. 7 A). A construct lacking most of this motif (Mid2p 1–595) does not rescue the mid2Δ phenotype or localize to the medial ring. However, two constructs containing the PH domain (Mid2p 569–704 and Mid2p 596–704) failed to direct GFP to the medial ring or rescue the null phenotype (Fig. 7 B; unpublished data). Therefore, the PH domain appears to be necessary, but not sufficient, for Mid2p localization and function. Indeed, the smallest fragment directing medial ring localization was Mid2p (336–704) (Fig. 7 D), although it also seemed to concentrate in or around the nucleus. Mid2p (336–704) was also able to partially rescue the mid2Δ phenotype.
In contrast to the denatured lysates analyzed directly in Fig. 2 C and Fig. 4 B, we noticed that HA–Mid2p was degraded during an immunoprecipitation procedure to produce a variety of smaller fragments (Fig. 7 C). However, deletion of the Mid2p amino terminus, which contains three putative PEST (Pro/Glu/Ser/Thr) domains that are predicted to contribute to protein instability by facilitating ubiquitin-mediated proteolysis (Rogers et al., 1986), largely prevented this degradation (Fig. 7 C). Coupled with the fact that Mid2p is multi-ubiquitinated in vivo (Fig. 3 A), these observations suggest that the putative PEST domains contribute to Mid2p instability.
To test whether deletion of the PEST motifs would yield a stable version of Mid2p, either full-length Mid2p or Mid2p (336–704) was produced in mid2Δ cdc25-22 cells under control of the moderate nmt41 promoter as amino-terminal HA fusion proteins. After 22 h at 25°C in the absence of thiamine, these cells were shifted to 36°C to block cells at G2/M before maximal expression was reached. The cells were then released to 25°C in the presence of thiamine to prevent further transcription, and samples were taken periodically for immunoblot analysis. As with endogenous Mid2p, the full-length fusion protein was degraded as cells septated and began a new cell cycle (Fig. 7 E). In contrast, Mid2p (336–704) remained unmodified and significantly more stable (Fig. 7 F).
Overproduction of Mid2p leads to ectopic septin structures
Given that mid2Δ cells had disorganized septin rings, we tested if excess amounts of wild type or the stable form of Mid2p would affect the septin cytoskeleton. Cells overproducing full-length Mid2p from the strong nmt1 promoter grew very slowly, became rounded, and had depolarized actin patches (Fig. 8
A), but did form colonies (unpublished data). In these cells, Spn3p–GFP was detected in aberrant filamentous structures that lacked F-actin, and very few cells were observed undergoing cytokinesis (Fig. 8 A). Overproduction of stable Mid2p (amino acids 336–704; Fig. 7 F) produced slightly elongated, rather than depolarized, cells that also grew very slowly (Fig. 8 A; unpublished data). In this case, septin rings did not disassemble on schedule. Instead, septin rings persisted at the new ends of cells well into the next cell cycle, and prominent remnants of septin structures could even be detected into a third cell division because they were observed at both tips in addition to the middle of the cell (Fig. 8 A).
To quantify this effect, five patterns of Spn3p–GFP localization were examined (Fig. 8 B). Septins are normally observed diffusely at the tips of interphase cells, but for the purpose of this quantification, tip localization was only considered if it was of equal intensity to the ring fluorescence. In the vector control, 83% of cells lacked Spn3p–GFP staining at the ring or tips while 17% of the cells contained a septin ring(s) at the center of the cell (n = 2,372 cells; Fig. 8 B, 1 and 2). By comparison, in the presence of stable Mid2p (336–704), the number of cells lacking Spn3p–GFP staining dropped to 33% while cells with only medial ring staining dropped to 9% (n = 1,982 cells; Fig. 8 B, 1 and 2). When Mid2p (336–704) was overproduced, the most prevalent pattern of Spn3p–GFP fluorescence was at one or both ends (32%; Fig. 8 B, 3). A small percentage of cells, 7%, exhibited Spn3p–GFP localization at the center of the cell in addition to the tips, implying that septin structures had persisted during one to two extra cell divisions (Fig. 8 B, 4). Finally, 19% of the cells overproducing Mid2p (336–704) had depolarized septin patches distributed throughout the cell (Fig. 8 B, 5). These three phenotypes were rarely observed in the vector control cells (<0.5%; Fig. 8 B, 3–5). Taken together, these data suggest that the carboxy terminus of Mid2p stabilizes septin ring structures and that the timely destruction of Mid2p facilitates the disassembly of septin rings normally observed during each cell cycle.
In this study, we have characterized a novel S. pombe protein, Mid2p, related to the cell division site placement factor, Mid1p. Our results suggest that Mid2p is involved in the establishment of the septin ring that permits efficient cell separation. Moreover, Mid2p degradation may facilitate septin ring disassembly.
Although our data indicate that both Mid1p and Mid2p act during cell division, it appears that their functions are nonoverlapping. There was no synthetic lethality observed in a double deletion strain, only an additive phenotype. Further, Mid1p localized appropriately in the absence of Mid2p and vice versa (unpublished data). Primary amino acid sequence and structure/function analyses also revealed some differences between these proteins. For instance, Mid1p possesses a nuclear localization signal and a large amino-terminal extension not found in Mid2p. On the other hand, there are a number of common elements. The amino-terminal portions of Mid1p and Mid2p contain a domain with significant similarity to the region of Drosophila anillin shown to directly bind and bundle actin filaments in vitro (Field and Alberts, 1995), although it remains to be determined whether or not either Mid1p or Mid2p interacts directly with actin filaments. Mid1p and Mid2p both contain PEST motifs conferring protein instability and similar carboxy-terminal regions that include a PH domain (∼36%). Although PH domains can mediate the protein–phospholipid interactions necessary to localize proteins to the plasma membrane (for review see Lemmon and Ferguson, 2001), this region of Mid2p is not sufficient to direct GFP to the cell cortex nor is it required for Mid1p localization to the medial ring (this study; Paoletti and Chang, 2000). Moreover, the Mid2p PH domain lacks the consensus sequence that predicts direct and avid binding to phosphoinositides (Isakoff et al., 1998). Thus, the PH domains of these proteins may be involved in protein–protein interactions. In the case of Mid2p, this region is critical for its localization and function, and it is tempting to speculate that it mediates interaction with septins, a hypothesis currently being explored. In the case of Mid1p, the carboxy-terminal region including the PH domain is dispensable for its function (Paoletti and Chang, 2000). Thus, the amino terminus of Mid1p and the carboxy terminus of Mid2p are likely to be the key determinants in mediating their unique roles during cytokinesis.
As mentioned above, Mid1p and Mid2p share sequence similarity not only with each other but with human anillin. Interestingly, there are multiple uncharacterized anillin homologues in Drosophila and C. elegans (Oegema et al., 2000). The regions of particular sequence similarity include the possible actin-interacting region noted above and an ∼200–amino acid stretch that includes the PH domain. Sequence alignments indicate that Mid1p (921 amino acids) is more similar to the larger anillin family members, such as Hs anillin (1,126 residues), Drosophila anillin (1201 residues), and the C. elegans gene product Y49E10.19 (1,205 residues). Conversely, Mid2p more closely resembles the smaller gene products, C. elegans Y43F8C.14 and Drosophila CG4530. Indeed, the characterized anillin family members in Drosophila and human seem to be more related to Mid1p than Mid2p in their localization pattern and loss of function phenotypes, with the exception of a connection to septins (Field and Alberts, 1995; Oegema et al., 2000). As these anillin homologues in higher eukaryotes are characterized, it will be interesting to learn whether they all function in cytokinesis. Further, it will be interesting to ascertain if they are functionally redundant or have distinct roles during cytokinesis as Mid1p and Mid2p do in S. pombe.
Several lines of evidence suggest that Mid2p affects septin ring organization and stability. In the absence of Mid2p, Spn3p–GFP is organized into a loose, rather than a tight, ring that disperses bilaterally across the septum as it forms. Conversely, in cells overproducing a stabilized Mid2p fragment, septin rings, or remnants thereof, persisted one to two cell divisions after they were formed. This observation implies that the turnover of full-length Mid2p might normally permit the timely disassembly of the septin ring. Furthermore, that these cells grew very slowly and were elongated upon the overproduction of nondegradable Mid2p indicates that septin ring disassembly may be important for normal cell cycle progression. In support of this possibility, cells producing a mutant form of the S. cerevisiae septin, Cdc3p, in which two Cdk sites have been altered to alanine, display two septin rings in G1 (Tang and Reed, 2002). Significantly, these cells are delayed in cell cycle progression until the old septin ring is disassembled (Tang and Reed, 2002).
In an independent study by Berlin et al. (2003), further evidence that Mid2p affects septin ring organization was obtained by FRAP analysis. The septin ring was found to be quite stable in wild-type cells, as assessed by the turnover of Spn4p–GFP (F. Chang, personal communication). In contrast, Spn4p–GFP was considerably more dynamic in mid2Δ cells, indicating that Mid2p function is required for the normal rigidity of the septin ring (F. Chang, personal communication).
Other anillin homologues have also been shown to influence septin ring organization. When C. albicans Int1p is overproduced in S. cerevisiae, highly ordered ectopic septin structures are observed in a manner strikingly similar to when Mid2p is overproduced (Gale et al., 2001). Further, overproduction of a carboxy-terminal fragment of Hs anillin induced the formation of ectopic septin containing structures with which Hs anillin was colocalized (Oegema et al., 2000). These abnormal cortical foci did not contain either actin or myosin-II (Oegema et al., 2000), nor did those formed upon Mid2p overproduction (Fig. 8 A; unpublished data). Thus, Mid2p shares with these proteins the ability to interfere with septin organization and function, although it is currently unknown whether the abnormal septin structures produced by Mid2p overproduction contain Mid2p.
An outstanding issue is how the events downstream of septation initiation network activation and septation are restricted temporally. Because all SCF targets identified to date are phosphorylated before being recognized by their cognate F-box protein (Willems et al., 1999), an appealing hypothesis is that the kinase that presumably modifies Mid2p in order to initiate its destruction might be conditionally activated upon the completion of septation. The identification of the protein kinase that phosphorylates Mid2p would significantly enhance our current understanding of the signaling pathways regulating cell separation.
Because Mid2p is only observed at the site of cell division, it follows then that the destruction machinery must also be recruited there. A candidate F-box protein is Pof6p, which was recently shown to localize to each side of the septum in dividing cells (Hermand et al., 2003). The loss of function of either Pof6p or the core SCF component Skp1p resulted in aberrant or multiple septa, implying that elevated levels of certain factors may also inhibit efficient cell division (Hermand et al., 2003). Both the loss of the transcription factor Sep1p (Ribar et al., 1999) or a slight increase in its abundance generates a cell separation defect (unpublished data). Thus, the proper coordination of cytokinesis appears to require a fine balance between the activation and inhibition of various factors involved in the process.
The loss of function of a wide array of proteins in S. pombe produces similar cytokinesis defects. For example, deletion of the septins, calcineurin (Ppb1p), a transcription factor (Sep1p), a MAP kinase (Pmk1p), a MAP kinase phosphatase (Pmp1p), members of the exocyst complex (Sec6p), and now Mid2p all cause a long delay in the physical separation of cells after septum synthesis, although the nuclear cycle remains unaffected (J. Pringle, personal communication; Yoshida et al., 1994; Longtine et al., 1996; Toda et al., 1996; Sugiura et al., 1998; Ribar et al., 1999; Wang et al., 2002). Whereas septin ring disorganization is likely to be the underlying cause of the cell separation defect of mid2Δ cells, this is not the case for ppb1Δ. In ppb1Δ cells, both Spn3p–GFP and Mid2p–GFP localize normally. This result suggests that even though a number of mutants inhibit cell separation in S. pombe, defects in multiple pathways may lead to this phenotype. Determining how septin ring organization is affected in other cytokinesis mutants may help elucidate the pathways regulating cell separation.
Materials And Methods
Strains, media, and methods
The S. pombe strains used in this study (Table I)
were grown in YE or minimal medium with the appropriate supplements as previously described (Moreno et al., 1990). DNA transformations were done by electroporation (Prentice, 1992) or lithium acetate transformation (Keeney and Boeke, 1994). Induction of the nmt promoter (Basi et al., 1993; Maundrell, 1993) was achieved by growing cells in thiamine (promoter repressed) and then washing cells three times in medium lacking thiamine (promoter induced).
The mid2, spn1, and spn3 ORFs were each tagged at their 3′ ends with either the HA3-KanR, myc13-KanR, or EGFP-KanR cassette, and the spn3 ORF was also tagged with a CFP-KanR cassette as previously described (Bahler et al., 1998b). KanR transformants were screened by whole cell PCR and then by immunoblotting to confirm the accurate integration and expression of the fusion protein.
The mid2, sep1, and spn4 ORFs were replaced with ura4+ by homologous recombination in a diploid strain (Bahler et al., 1998b). Ura+ transformants were screened for the proper gene disruptant by whole cell PCR. Heterozygous diploid strains were then sporulated at 25°C followed by tetrad dissection to determine if these genes were essential for viability.
Molecular biology techniques
PCR amplifications were performed with TaqPlus Precision polymerase (Stratagene) according to manufacturer's instructions. Oligonucleotides were synthesized by Integrated DNA Technologies, Inc., and all sequences are available upon request. Sequencing was done with Thermosequenase (USB) and Redivue 33P Terminator Kit (Amersham Biosciences).
The entire mid2 ORF was amplified from genomic DNA with oligos containing NdeI and BamHI sites on the 5′ and 3′ ends, respectively. The PCR product was cut with NdeI/BamHI and cloned into the thiamine-repressible pREP1, -41, -41HA, and -41GFP vectors to create pKG2207, pKG2208, pKG2491, and pKG2236. Primers containing 5′ NdeI/ATG and 3′ TAA/BamHI sites were also designed to amplify regions of Mid2p using pKG2208 as a template. These PCR products were also subcloned into the pREP series of vectors.
Cytology and microscopy
Strains producing GFP-tagged proteins were grown in YE medium, visualized live or fixed with ethanol or formaldehyde, and processed as previously described (Balasubramanian et al., 1997; Tomlin et al., 2002). To visualize cell walls, fixed cells were stained with aniline blue (methyl blue) (Sigma-Aldrich). Cells from a 1-ml culture were fixed in ethanol, resuspended in 1 ml of a 1:100 dilution of an aniline blue stock solution (100 mg/ml), incubated for 10 min, collected by centrifugation, washed three times with water, and resuspended in 20 μl water. LatA (Molecular Probes) was used at a concentration of 100 or 200 μM. Actin was visualized after formaldehyde fixation using AlexaFluor®594-phalloidin (Molecular Probes). Images were acquired digitally as previously described (Tomlin et al., 2002). For time-lapse experiments, cells were placed on a hanging drop glass slide (Fisher Scientific) containing solidified YE agar and covered with a coverslip. Time-lapse images of Spn3p–GFP and Mid2p–GFP were obtained using an Ultraview LCI confocal microscope equipped with a 488-nm Ar Ion laser (PerkinElmer). Images were captured using Ultraview LCI software (version 5.2; PerkinElmer) and processed using Velocity software (version 1.4.2; Improvision). Z-series optical sections were taken at 0.5-μm spacing at intervals of 2 or 3 min.
Immunoprecipitations and immunoblots
Whole cell lysates were prepared in NP-40 buffer followed by anti-HA or anti-Myc immunoprecipitations as previously described (Gould et al., 1991; McDonald et al., 1999). Denatured lysates were prepared as previously described (Burns et al., 2002). Protein samples were resolved on 4–12% NuPAGE gels in MOPS buffer and subsequently transferred to 0.2 μm nitrocellulose (Bio-Rad Laboratories) according to the manufacturer's instructions (Invitrogen). Immunoblotting was done with anti-HA (12CA5; 2 μg/ml), anti-Myc (9E10; 2 μg/ml), anti-Cdc13p (GJG56; 1:2,500 of serum), anti-Cdc2p (PSTAIRE; 1:5,000; Sigma-Aldrich), or anti-ubiquitin (1:100; Sigma-Aldrich). The primary antibodies were detected with HRP-conjugated goat anti–mouse or goat anti–rabbit secondary antibodies (0.4 mg/ml; Jackson ImmunoResearch Laboratories) at a dilution of 1:50,000 followed by ECL visualization using SuperSignal (Pierce Chemical Co.).
After an anti-HA immunoprecipitation from denatured lysate, beads were washed two times with 1 ml NP-40 buffer, four times with 1 ml of phosphatase buffer (25 mM Hepes-NaOH, pH 7.4, 150 mM NaCl, 0.1 mg/ml BSA), divided in half, pulsed down, and the supernatant aspirated off. 10-μl reactions composed of 1X phosphatase buffer, 2 mM MnCl2, plus 1 μl of λ-phosphatase (New England Biolabs, Inc.) or 1 μl H2O were then incubated at 30°C for 45 min with gentle mixing every 5 min. The beads were washed three times with NP-40 buffer and resuspended in 25 μl 2X lithium dodecyl sulfate sample buffer.
Total RNA was prepared by extraction with hot acidic phenol and SDS as detailed previously (Burns et al., 1999). Total RNA (20 μg per sample) was resolved on formaldehyde-agarose gels and capillary blotted to a Duralon-UV membrane. mid2+ and his3+ mRNA levels were detected by hybridization of 32P-labeled oligonucleotide probes (Rediprime II; Amersham Biosciences) from regions within their ORFs. The blots were exposed to Phosphor-Imager screens and visualized with ImageQuant 5.2 on an Amersham Biosciences Typhoon 9200 scanner.
In vivo ubiquitination assays
Strains carrying pREP1-His6–ubiquitin were grown at 25°C for 22 h in the absence of thiamine to induce the expression of His6–ubiquitin (Ub), shifted to 36°C for 4 h, and then 6 × 108 cells were harvested. Cell lysates were prepared in the presence of 8 M urea with 100 mM sodium phosphate, pH 8.0, and 5 mM imidazole followed by binding of His6–Ub-conjugated proteins to a nickel affinity resin (Ni-NTA Superflow; QIAGEN). The beads were washed as previously described (Benito et al., 1998) and ubiquitinated proteins eluted with 200 mM imidazole.
Online supplemental material
The supplemental material for this article is available. Three Quicktime movie files of the following live cells are available online: Mid2p–GFP (Video 1), Spn3p-GFP (Video 2), and Spn3p-GFP in a mid2Δ strain (Video 3).
The authors wish to thank F. Chang (Columbia University, New York, NY) and J. Pringle (University of North Carolina at Chapel Hill, Chapel Hill, NC) for communicating results prior to publication, T. Toda (Imperial Cancer Research Fund, London, UK) for providing strains, G. Burns, A. Feokistova, and L. Pundor for excellent technical advice and assistance, and J. Pringle for providing spn null strains that prompted our study of the relationship between Mid2p and septins as well as for helpful comments on the manuscript.
J.J. Tasto was supported by National Institutes of Health grant T32GM08554. This work was supported by the Howard Hughes Medical Institute of which K.L. Gould is an associate investigator.
The online version of this article includes supplemental material.