Background The small GTPase Rho1 has been shown to regulate the organization of the actin cytoskeleton and formation of the cell wall in the fission yeast Schizosaccharomyces pombe. Activity of Rho1 must be precisely regulated in vivo, since both increases and decreases in its activity affect cell growth and shape. Thus, it is important to clarify the mechanism by which the activity of Rho1 is regulated in vivo.
Results Seven genes encoding putative GAPs, GTPase-activating proteins, for the function of the Rho-family proteins were isolated from S. pombe. After disruption of these genes, rga1+ was found to play important roles in cell growth and morphogenesis. In rga1 null cells, delocalized F-actin patches and extraordinary thickening of the cell wall and the septum were observed. On the other hand, over-expression of Rga1 produced shrunken or dumpy cells. The phenotype of the rga1 null cells or the Rga1-over-expressing cells was similar to that of cells containing abnormally high or low Rho1 activity, respectively. Moreover, direct association of Rga1 with Rho1 was shown. Rga1 was localized to the cell ends and septum where Rho1 is known to function.
Conclusions In S. pombe, Rga1 is involved in the F-actin patch localization, cell morphogenesis, regulation of septation, and cell wall synthesis, probably functioning as a GAP for the function of Rho1.
In eukaryotic cells, small GTPases are involved in various signalling pathways (reviewed by Takai et al. 1992; Zerial & Huber 1995). These proteins alternatively bind GTP or GDP, and it is believed that the GTP-bound form is active, while the GDP-bound form is inactive. Activation of the small GTPases is thought to require the function of guanine nucleotide exchange factor (GEF). On the other hand, the inactivation depends on their GTPase activity, which is enhanced by GTPase-activating protein (GAP). Among the known small GTPases, the Rho-family proteins have been shown to be involved in various cellular activities, mainly through their ability to control the organization of the actin cytoskeleton (reviewed by Narumiya 1996; Hall 1998; Bishop & Hall 2000).
The fission yeast Schizosaccharomyces pombe has a cylindrical shaped cell which is enveloped with the cell wall. In the interphase wild-type cells, actin is organized as cortical F-actin patches that are localized at the growing ends of the cell where the cell wall is newly synthesized, and as F-actin cables running along the long axis of the cell (Marks & Hyams 1985; Kanbe et al. 1989; Arai et al. 1998). During mitosis, actin is recruited to the mid region of the cell and the F-actin ring is formed, which then divides the cytoplasm by its contraction (reviewed by Le Goff et al. 1999). This is followed by the progression of septation. The interphase F-actin structures reappear at the division site before and during the septation. Thus, controlled organization of the actin cytoskeleton is required for the cell to undergo polarized growth, cytokinesis and septation (Snell & Nurse 1993; Verde et al. 1995; Ishiguro 1998; Le Goff et al. 1999).
Six Rho-family proteins have been identified in the fission yeast, including Cdc42 (Fawell et al. 1992) and Rho1∼5 (Nakano & Mabuchi 1995; our unpublished data). Cdc42 and Rho1 are the counterparts of Cdc42Hs and RhoA in human, and Cdc42p and Rho1p in the budding yeast Saccharomyces cerevisiae, respectively. These proteins have been shown to be essential for the cell viability (Miller & Johnson 1994; Nakano et al. 1997). It has been reported that Cdc42 plays an important role in the determination of cell polarity (Miller & Johnson 1994), while Rho1 is involved in cell morphogenesis and septation (Arellano et al. 1996, 1997; Nakano et al. 1997). On the other hand, rho2+, rho3+, rho4+ and rho5+ are not essential for cell growth, but they also play some role in cell morphogenesis and septation (Hirata et al. 1998; our unpublished data).
Rho1 has been observed to localize at the growing ends of the interphase cells and the septating region in mitotic cells (Arellano et al. 1997; Nakano et al. 1997). Expression of a dominant-active Rho1 mutant (Rho1G15V or Rho1Q64L) produces swollen cells, branched cells and multiseptated cells, while that of a dominant-negative Rho1 mutant (Rho1T20N) produces shrunken or dumpy cells (Arellano et al. 1997; Nakano et al. 1997). These cells had defects in the organization of the actin cytoskeleton, formation of the cell wall, and septation. It has been demonstrated that Rho1 activates the cell wall synthesizing enzyme, β-glucan synthase (Arellano et al. 1996) as has been shown for the budding yeast Rho1p (Drgonova et al. 1996; Mazur & Baginsky 1996; Qadota et al. 1996). Moreover, it has recently been reported that Rho1 directly binds to PKC family protein kinases, Pck1 and Pck2, and functions as a positive regulator for these kinases (Arellano et al. 1999; Sayers et al. 2000). These kinases are considered to function redundantly in cell morphogenesis (Toda et al. 1993). Furthermore, the Rho1-Pck1/Pck2 pathway is involved in the control of septation which is antagonized by the calcineurin-like protein phosphatase Ppb1 (Nakano et al. 1997). Thus, Rho1 seems to play multifunctional roles by interacting with its targets, including the above-mentioned proteins in the fission yeast cell. However, it has not yet been shown how the activity of Rho1 is controlled in the cell.
In this study, we isolated seven novel genes probably encoding GAPs for the Rho-family proteins in the fission yeast and analysed in detail the function of Rga1 which seems to serve as a GAP for the function of Rho1.
Isolation and disruption of genes encoding RhoGAP in S. pombe
After searching the S. pombe genome sequence database at the Sanger centre, seven genes containing a putative GAP domain for the Rho-family proteins were revealed (Fig. 1A). Some of the predicted gene products contained distinct domains besides the GAP domain, such as the pleckstrin homology (PH) domain, the LIM domain, and the SH3-binding domain. In the budding yeast S. cerevisiae, seven genes encoding RhoGAP have been reported (Fig. 1A). Bem2p, Bem3p and Rga1p/Dbm1p act to establish the cell polarity in a redundant manner (Kim et al. 1994; Peterson et al. 1994; Zheng et al. 1994; Stevenson et al. 1995; Chen et al. 1996). These proteins stimulate the GTPase activity of Rho1p. Although Sac7p also interacts with Rho1p, it may play a role which is different from those of the above proteins (Schmidt et al. 1997). Rgd1p functions as a GAP for Rho3p and Rho4p (Doignon et al. 1999). The functions of Lrg1p and Bag7p have not yet been found (Muller et al. 1994; Schmidt et al. 1997). Based on the primary structures of the proteins, In particular those of the GAP domains, it was speculated that there exist orthologous relationships between the following pairs of the fission yeast and the budding yeast proteins: Rga1 and Lrg1p; Rga2 and Bem3p; Rga3 and Rga4, and Rga1p/Dbm1p; Rga5 and Rga6, and Sac7p and Bag7p; Rga7 and Rgd1p.
In order to study their cellular functions, we disrupted each gene in the cell (see Experimental procedures). After random spore analysis and tetrad analysis, it was found that none of these genes was essential for the cell viability. However, rga1 null cells showed a severe defect in cell growth (see below) and a swollen, multiseptated or branched shape (Fig. 1C). This phenotype was similar to that of the cell expressing a dominant-active Rho1 mutant (Rho1G15V or Rho1Q64L) (Nakano et al. 1997). Therefore, it is possible that Rga1 functions as a GAP for Rho1 in the fission yeast. Therefore, we mainly studied the function of Rga1 in this paper. In addition, cells lacking Rga4 displayed a dumpy shape. Shrunken cells were found in rga7 null cells (Fig. 1C).
Rga1 is involved in the actin cytoskeletal organization and the formation of the cell wall and septum
In order to examine the phenotype of the rga1 null cells in detail, we stained them with Calcofuor or Bodipy-phallacidin and DAPI. In the wild-type cells, both the newly synthesized cell wall region and the septum were clearly visualized with Calcofluor. In contrast, the whole cell surface of the rga1 null cells was strongly stained (Fig. 2). Judging from the Nomarski image in Fig. 1C, the cell wall of the rga1 null cells seemed to be thicker than that of the wild-type cells. These observations suggest that formation of the cell wall is probably enhanced by the Rga1 depletion.
It is well known that F-actin patches accumulate at the ends of interphase cells and around the F-actin ring in mitotic cells. Surprisingly, F-actin patches were abundantly formed all over the cell surface in the rga1 null cells, regardless of the cell cycle stage (Fig. 2). The fluorescence from the F-actin patches was so strong that F-actin cables were hardly observed at all. In addition, a small number of the rga1 null cells were binucleated without the F-actin ring, although most of the cells underwent cytokinesis. Thus, control of the actin organization requires the function of Rga1.
Next, we examined whether the growth and morphological defects of the rga1 null cells were suppressed by increased expression of Rga proteins (Fig. 3A). Introduction of pREP1rga1 restored the defects of the rga1 null cells in EMM containing 5 µm thiamine, while that of pREP1rga2, pREP1rga3, pREP1rga4, pREP1rga5, pREP1rga6 or pREP1rga7 did not in EMM containing thiamine at varying concentrations (0, 0.02, 0.1, 0.5 and 5 µm). Thus, none of the other Rga proteins could substitute for Rga1 in the cell.
We then made several plasmids containing rga1-truncated genes and studied which domains of Rga1 are important for its function (Figs 3B,C). After transformation of the rga1 null cells with these plasmids, it was found that the expression of the GAP domain of Rga1 is sufficient to rescue the growth defect of these cells. Therefore, Rga1 probably plays an important role in the cell through its GAP domain.
Over-expression of Rga1 mimics rho1 deficient cells
We next examined the effect of over-expression of Rga1 or its truncated proteins in the fission yeast cells. Over-expression of Rga1 inhibited cell growth and produced shrunken or dumpy cells (Fig. 4, also summarized in Fig. 3B). Almost the same effect was found with over-expression of Rga1C or Rga1GAP. The F-actin patches were delocalized on the surface of the dumpy cells, while no obvious F-actin structure remained in the shrunken cells (Fig. 4C). It has previously been reported that both rho1 null cells and cells expressing dominant-negative mutants of Rho1 display a phenotype which is similar to that described above (Arellano et al. 1997; Nakano et al. 1997). Therefore, Rga1 may function in the negative regulation of Rho1 in vivo.
Interestingly, over-expression of Rga1N produced elongated and curved cells, although it did not affect cell growth (Fig. 4). In these cells, F-actin was abnormally condensed at the cell ends. This result suggests that the N-terminus of Rga1 may play a role in cell morphogenesis which is independent of the GAP domain. On the other hand, cells over-expressing either Rga1N1 or Rga1N2 displayed a normal shape (Fig. 4B). It is possible that the three repeats of the LIM domains in the N-terminal half of Rga1 may function together for that activity.
Rga1 negatively regulates Rho1-activity
Rga1 may function as a negative regulator of Rho1 in the fission yeast, as described above. We next tried to estimate the GAP activity of Rga1 for Rho1 in vitro. However, this was not possible since we could not prepare a sufficient amount of Rga1 for the biochemical assay; the expression level of Rga1 in E. coli was very low, and almost all the Rga1 expressed was insoluble (data not shown). Moreover, only a very small amount of Rga1 was solubilized from the S. pombe membrane fraction, where almost all the Rga1 is present (see below). Therefore, we performed four genetic experiments in order to study the relationship between Rga1 and Rho1.
First, we examined the effect of co-over-expression of Rho1 and Rga1. Over-expression of Rho1 inhibited the cell growth as has previously been described (Arellano et al. 1996; Nakano et al. 1997) (Fig. 5A). In contrast, cells co-over-expressing Rho1 and Rga1GAP grew well. The GAP domain of Rga1 probably represses the activity of Rho1 in the cell.
Second, we examined whether the defect in cell growth of the rga1 null cells could be rescued by a decrease in the activity of Rho1. It has been shown that expression of Rho1T20N or the Rho1-binding region of Pck1 (Pck1N) lacking the kinase domain inhibits the cellular activity of Rho1 (Arellano et al. 1997; Nakano et al. 1997; Sayers et al. 2000). The growth rate of the rga1 null cells containing pREP1rho1T20N was between that of the cells containing pREP1 and of the cells containing pREP1rga1GAP in the presence of thiamine (Fig. 5B). Moreover, although wild-type cells containing pREP1rho1T20N could not grow in the absence of thiamine (Nakano et al. 1997), the rga1 null cells containing it were able to grow. In addition, the rga1 null cells containing pREP81pck1N grew as well as the wild-type cells in the absence of thiamine. Therefore, the decrease in Rho1 activity rescued the defect of the rga1 null cells.
Third, if the activity of Rho1 is very high in the absence of Rga1, the rga1 null cells would be more sensitive to increases in the expression level of Rho1 than the wild-type cells. We then examined the effect of over-expression of Rho1 from pREP41rho1, which does not affect cell growth and cell shape in the wild-type cells (Nakano et al. 1997). As expected, the rga1 null cells containing pREP41rho1 could not grow in the absence of thiamine (Fig. 5B).
Finally, we studied the association between Rho1 and Rga1 by using a two-hybrid system (Fig. 5C). Since Rho-family proteins specifically associate with their GAPs when they bind GTP (Aspenstrom & Olson 1995), we used a dominant-active Rho1 mutant, Rho1Q64L, instead of the wild-type Rho1. As shown in Fig. 5C, all transformants grew well in the presence of histidine. In contrast, only cells containing both pBTMrho1Q64L and pGADrga1GAP grew in the absence of histidine. This result indicates that Rho1 is able to bind directly to the GAP domain of Rga1.
Rga1 localizes to the cell ends and the division site
The cellular localization of Rga1 was studied by expressing the YFP fused protein in rga1 null cells. The rga1 null cells carrying pREP1YFP-rga1 grew as well as the wild-type cells and displayed a normal cell shape at 25 °C in YEA or EMM containing 5 µm thiamine (data not shown). After centrifugation of the cell lysate, almost all the YFP-Rga1 was found in the pellet, which may contain cell membranes (Nakano et al. 1997) (Fig. 6A). Some bands detected at positions lower than that of YFP-Rga1 may be due to the degradation products of YFP-Rga1, since these were not seen in the extracts derived from cells containing pREP1 (data not shown). In addition, treatment of the pellet with 1 m KCl solubilized a small amount of YFP-Rga1, while 1% Triton X-100 did not solubilize YFP-Rga1 at all (Fig. 6B). A similar result was obtained for cells expressing HA-Rga1 (data not shown). Therefore, Rga1 may be tightly associated with the cell membrane.
Next, we tried to observe the localization of YFP-Rga1 in live cells. At 16 h after removing thiamine from the medium, YFP-Rga1 was found to localize at the ends of the interphase cells and in the middle of the mitotic cells (Fig. 6C). It has previously been demonstrated that Rho1 also accumulates around the cell ends and the septum (Arellano et al. 1997; Nakano et al. 1997). Therefore, Rga1 may co-localize with Rho1 throughout the cell cycle. In addition, we also tried to determine a domain of Rga1 which is necessary for its cellular localization. However, neither YFP-Rga1N nor YFP-Rga1C localized at the cell ends or the septum region (data not shown). Thus, the full length of Rga1 may be required for cellular localization. However, we cannot exclude the possibility that either YFP-Rga1N or YFP-Rga1C was unstable in the cell.
Functions of the other Rga proteins in fission yeast
As described above, disruption of neither of the six genes, rga2+∼7+, caused severe defect in cell growth, in contrast to that of rga1+. In view of their similarities in the GAP region, we made double mutant rga2 rga3, rga2 rga4, rga3 rga4 and rga5 rga6 strains. However, none of the double mutant cells showed a marked phenotype (data not shown).
Meanwhile, we found that over-expression of Rga2, Rga3, Rga5 or Rga6 inhibited cell growth (Fig. 7A). This result suggested the possibility that these proteins may be involved in the negative regulation of Cdc42 or Rho1, since the other Rho proteins are not required for cell viability (see Introduction). Moreover, shrunken cells similar to those seen in the rho1 deficient cells were also seen in cells over-expressing Rga2, Rga3, Rga5 or Rga6 (Fig. 7B). Thus, these Rga proteins may also function as GAPs for Rho1 in vivo. Indeed, we found a direct interaction between Rho1Q64L and both Rga2 and Rga3 in the two-hybrid analysis (data not shown). In contrast, the growth of the Rga4- or Rga7-over-expressing cells was comparable to that of the control cells. However, the microscope appearance of these cells was abnormal: the Rga4-over-expressing cells were dumpy, while some of the Rga7-over-expressing cells contained multisepta (Fig. 7B).
We identified seven RhoGAPs from the fission yeast and studied the function of Rga1. Cells lacking Rga1 showed defects in growth and shape similar to cells expressing a dominant-active Rho1 mutant. These cells had an abnormally thick cell wall and increased the number of F-actin patches throughout the cortex. These defects were rescued by repressing the Rho1 activity. Rga1 was localized to the ends of interphase cells and the septum of dividing cells where Rho1 also localizes. Moreover, Rho1 binds the GAP domain of Rga1. Thus, it is likely that Rga1 negatively regulates Rho1 activity throughout the cell cycle. In addition, it was demonstrated that the N-terminus of Rga1 is involved in cell morphogenesis. It will be important to reveal the function of the N-terminal region of Rga1.
Rga1 may be an orthologue of the budding yeast Lrg1p, since the molecular structures of these proteins are somewhat similar to each other. It has been determined that the expression level of Lrg1p is elevated during mating and sporulation (Muller et al. 1994). However, its function has not been clarified. Lrg1p may also play a role in controlling Rho1p activity in the budding yeast.
We found that Rga1 is strongly associated with the insoluble fraction of the cell, which may contain cell membranes. Since Rga1 showed a tendency to be solubilized by treatment with 1 m KCl, but not with 1% Triton X-100, it may not be an intrinsic membrane protein. Although it has previously been considered that Rho1 is directly associated with the membrane fraction (Nakano et al. 1997), it is unlikely that Rga1 is localized on the cell membrane solely through interaction with Rho1, because Rho1 is efficiently solubilized with 1% Triton X-100. Thus, there may be another protein which anchors Rga1 to the cell membrane. Rga1 contains an SH3-binding domain and three repeats of LIM domains in its N-terminal half. Both of these domains are thought to function in protein–protein interactions (reviewed by Pawson 1998), and may thus play a role in the localization of Rga1.
In the budding yeast, it has been demonstrated that Bem2p, Bem3p, Rga1p/Dbm1p and Sac7p function as GAPs for Rho1p, respectively (Zheng et al. 1994; Chen et al. 1996; Schmidt et al. 1997). We isolated Rga2∼7, in addition to Rga1 and demonstrated the possibility that Rga2, Rga3, Rga5 and Rga6 may be negative regulators of Rho1, as judged from the phenotypes of their over-expressing cells. Interestingly, Rga2, Rga3, Rga5 and Rga6 are similar to Bem3p, Rga1p/Dbm1p and Sac7p, respectively. Although the fission yeast cells lacking each of these RhoGAPs grew well under normal conditions, the regulation of Rho1-activity by these proteins may become important under special conditions.
There are five Rho-family proteins in the fission yeast in addition to Rho1 (see Introduction). It is possible that Rga1∼7 may control their cellular activities. Indeed, cells lacking either rga4+ or rga7+, or cells over-expressing Rga4 or Rga7, showed an aberrant shape. Our preliminary experiment suggests that Rga7 is involved in controlling the activity of Rho3 in the cell (T.M., K.N & I.M., unpublished results). Moreover, it is possible that Rga4 functions as a GAP for Rho1 and Cdc42, since their primary structures are similar to that of the budding yeast Rga1p/Dbm1p. However, at this moment, we do not have more evidence on the interactions between the Rho-family proteins and the RhoGAPs. Further analysis of the relationships among the Rho-family proteins and the RhoGAPs would promote a better understanding of the mechanisms of cell morphogenesis and septation in the fission yeast.
Genetic techniques and DNA manipulations
S. pombe strains used in this study are listed in Table 1. The media used have been described previously (Moreno et al. 1991). Complete medium YEA and minimum medium EMM were used for growing S. pombe strains. MEA was used for the induction of conjugation and sporulation. All plates contained 2% agar. Standard procedures for S. pombe genetics were carried out as described in Alfa et al. (1993) and Moreno et al. (1991). Standard methods were used for DNA manipulations (Sambrook et al. 1989).
For the gene disruption, a part of the chromosomal locus of each gene was replaced with the marker gene ura4+ or LEU2 using homologous recombination. In detail, diploid cells constructed by mating the JY741 and JY746 strains were transformed with DNA fragments (rga1::ura4+ nucleotides spanning from 924 to 2289 in rga1+ were replaced by ura4+, rga2::ura4+ nucleotides 1–3825 in rga2+ were replaced by ura4+, rga3::ura4+ nucleotides 1–3450 in rga3+ were replaced by ura4+, rga4::ura4+ nucleotides 1146–1546 in rga4+ were replaced by ura4+, rga5::LEU2; nucleotides 307–729 in rga5+ were replaced by LEU2, rga6::ura4+ nucleotides 410–411 in rga6+ were replaced by ura4+, and rga7::ura4+ nucleotides 1–2031 in rga7+ were replaced by ura4+), and correct integrations were verified by Southern blotting. After the random spore analysis and the tetrad analysis were performed, each disruptant was acquired.
The fission yeast expression vectors, pREP1, pREP41, pREP81 and pREP2 (Maundrell 1993) were used in this study. The first three vectors contain the LEU2 marker gene, while pREP2 contains ura4+. pREP1 and pREP2 have the strongest nmt1 promoter, while pREP81 has the weakest (Forsburg 1993). Expression of exogenous genes from these plasmids is repressed by the introduction of 5 µm thiamine into the medium.
In order to over-express each protein, pREP1rga1 (for expression of amino acid residues 1–1150 of Rga1), pREP1rga1C (amino acid residues 570–1150), pREP1rga1GAP or pREP41rga1GAP (amino acid residues 696–911), pREP1rga1N (amino acid residues 1–572), pREP1rga1N1 (amino acid residues 1–315), pREP1rga1N2 (amino acid residues 266–572), pREP1rga2 (amino acid residues 1–1275 of Rga2), pREP1rga3 (amino acid residues 1–949 of Rga3), pREP1rga4 (amino acid residues 1–892 of Rga4), pREP1rga5 (amino acid residues 1–361 of Rga5), pREP1rga6 (amino acid residues 1–733 of Rga6), and pREP1rga7 (amino acid residues 1–695 of Rga7) were constructed. pREP1YFP-rga1 was constructed to express the yellow-green variant of the Aequorea victoria green fluorescent protein (YFP) fused at the N-terminus of Rga1.
pREP1rho1, pREP41rho1 and pREP1rho1T20N have been previously described (Nakano et al. 1997). pREP2rho1 corresponded to pREP1rho1, except for a difference in their marker genes. pREP81pck1N was constructed to express amino acid residues 1–400 of Pck1.
The interaction between two proteins was examined by a two-hybrid system utilizing the expression of the HIS3 reporter gene, as previously described (Vojtek et al. 1993). pBTMrho1Q64L has previously been described (Sayers et al. 2000). pGADrga1GAP contained the amino acid residues 570–911 of Rga1.
The preparation of cell lysates and subsequent fractionation were performed as previously described (Nakano et al. 1997). Proteins were electrophoresed on an SDS−10% polyacrylamide gel (Laemmli 1970), and transferred on to a PVDF membrane. Immunoblotting was performed as previously described (Nakano et al. 1997) using monoclonal anti-GFP antibodies (Molecular Probes Inc., Eugene, OR, USA).
Staining of the cells with Calcofluor (Sigma Chem. Co., St Louis, MO, USA) or Bodipy-phallacidin (Molecular Probes) and DAPI (Molecular Probes) was performed as previously described (Alfa et al. 1993; Arai et al. 1998). Images were obtained using a Zeiss Axioskop fluorescence microscope (Carl Zeiss Inc., Oberkochen, Jena, Germany) equipped with a Plan Apochromat ×63 objective lens and photographed on Kodak T-MAX ASA 400 film.
This work was supported by research grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (nos 50302815, 10213202 and 1249008).