Formins are conserved eukaryotic proteins playing key roles in regulating cell polarity. We have characterized the roles of a formin CaBni1p in the polymorphic fungus Candida albicans. CaBni1p localized persistently to hyphal tips during hyphal growth but to distinct growth sites at different cell cycle stages during yeast growth. Cabni1Δ yeast cells exhibited several morphological defects, such as round and enlarged cells, widened bud necks and a random budding pattern. Although Cabni1Δ cells could still undergo yeast-hypha growth switch, the hyphae were markedly swollen. Cabni1Δ also showed defects in spindle and cytoplasmic microtubule orientation and positioning. Coincidentally, the spindle orientation protein CaKar9p in Cabni1Δ yeast cells appeared as multiple random cortical spots, in contrast to the single spot at the bud tip of many wild-type cells. Interestingly, several defects manifested in Cabni1Δ yeast cells were partially corrected during hyphal growth. We found that the second formin CaBnr1p was recruited to hyphal tips, while it localized only to the bud neck during yeast growth. This behavior of CaBnr1p may play a key role in correcting Cabni1Δ defects during hyphal growth. Cabni1Δ exhibited reduced virulence in mice. These results indicate that the formins play an important role in Candida albicans polarized growth and CaBni1p's function is required for virulence.
In all unicellular and multicellular organisms, the establishment and maintenance of cell polarity are crucial for various cellular functions such as differentiation, morphogenesis, motility and signal transduction (Cid et al., 1995; Drubin and Nelson, 1996). Regulation of cell polarity is a complex process requiring precise spatial and temporal coordination of functions of many proteins (Drubin, 2000; Pruyne and Bretscher, 2000). Formins are a family of highly conserved eukaryotic proteins implicated in a wide range of actin-based processes, including cell polarization and cytokinesis (Wasserman, 1998; Zeller et al., 1999; Evangelista et al., 2003). Saccharomyces cerevisiae (Sc) has two formins Bni1p and Bnr1p. BNI1 was first identified as a gene required for bipolar bud site selection (Zahner et al., 1996). Substantial evidence indicates that Bni1p and other formins function as molecular scaffolds that link Rho-type GTPases with components of the actin cytoskeleton. Bni1p interacts through an N-terminal domain with the GTP-bound forms of Rho1p (Kohno et al., 1996) and Cdc42p (Evangelista et al., 1997), which regulates the organization of actin structures. The proline-rich formin homology domain 1 (FH1) of Bni1p binds to the G-actin-binding protein profilin (Evangelista et al., 1997; Imamura et al., 1997) and this interaction delivers ATP-bound actin monomers to the growing barbed ends of actin filaments (Theriot and Mitchison, 1993; Carlier and Pantaloni, 1997; Wear et al., 2000). The FH2 domain of Bni1p is a unique and defining feature of formin proteins. It controls actin nucleation and cable assembly both in vitro and in vivo (Copeland and Treisman, 2002; Evangelista et al., 2002; Pruyne et al., 2002; Sagot et al., 2002a; Sagot et al., 2002b). The C-terminal end of Bni1p binds Bud6p/Aip3p (Evangelista et al., 1997), which has been identified as an actin-binding protein (Amberg et al., 1997). Bni1p also binds Spa2p (Fujiwara et al., 1998), a component of polarisomes that plays an important role in morphogenesis (Snyder, 1989; Sheu et al., 2000). Bni1p is also involved in certain microtubule (MT) functions required for the positioning and orientation of mitotic spindles (Lee et al., 1999; Lee et al., 2000). This function was first thought to be attributable to its role in determining the cortical localization of the microtubule orientation protein Kar9p, because Kar9-GFP was observed to mislocalize in bni1Δ mutants (Miller and Rose, 1998; Miller et al., 1999). Recently, it was found that Kar9p cross-links the ends of cytoplasmic MTs to the motor protein Myo2p, which drives the migration of MT ends along actin cables towards the bud cortex (Hwang et al., 2003; Liakopoulos et al., 2003). These observations provide a good mechanistic explanation for the spindle positioning and orientation defects in bni1Δ mutants. Since Bud6p, an interacting partner of Bni1p, has an important role in capturing MT ends at cell cortex (Segal et al., 2000; Huisman et al., 2004), the altered properties of Bud6p in the absence of Bni1p may also contribute to the spindle defects of bni1Δ mutants.
Candida albicans (Ca) is the most prevalent human fungal pathogen. An important virulence trait of this pathogen is its ability to switch between several morphological forms: budding yeast, pseudohyphae and true hyphae (Odds, 1985; Liu, 2001; Zheng et al., 2004). It has been well documented that the yeast-hypha switch allows the pathogen to penetrate epithelial tissues and rupture phagocytes (Lo et al., 1997; Calderone and Fonzi, 2001; Bai et al., 2002). C. albicans is a good model for the study of polarized cell growth because of the presence of a hyphal program where cell polarity is maintained and growth is persistently restricted at hyphal tips. Studies using this model have revealed important mechanisms underlying polarized growth that are either unique to C. albicans or have not been found in the study of the budding yeast form (Zheng et al., 2003; Zheng et al., 2004). During yeast growth, cells proliferate by budding, and polarized growth occurs at distinct sites in different stages of the cell cycle (Anderson and Soll, 1986). By contrast, cell growth in true hyphal cells occurs in a cell cycle-independent manner, being persistently confined to hyphal tips (Staebell and Soll, 1985; Anderson and Soll, 1986; Hazan et al., 2002; Zheng et al., 2003; Zheng et al., 2004). It has been shown that actin cortical patches and cables polarize towards hyphal tips throughout hyphal growth (Hazan et al., 2002; Zheng et al., 2003).
In a previous study we showed that the polarisome protein CaSpa2p is important for polarity establishment and maintenance and virulence in C. albicans (Zheng et al., 2003). In this study, we have investigated the role of CaBni1p, a possible interacting partner of CaSpa2p, in the growth and virulence of C. albicans.
Materials and Methods
Strains, media and growth conditions
The Candida albicans strains used are listed in Table 1. The strains were grown in either YPD (2% yeast extract, 1% Bacto peptone and 2% glucose) or GMM medium (2% glucose, 1× yeast nitrogen base). To grow strains defective in uracil, arginine or histidine synthesis, the required nutrient was added to GMM. Liquid and solid media for hyphal growth were prepared as previously described (Zheng et al., 2003). G1 cells were prepared by centrifugal elutriation as previously described (Zheng et al., 2003).
Gene deletion mutants were constructed by sequentially deleting the two copies of a gene from BWP17 (Enloe et al., 2000). A gene deletion cassette was constructed by flanking a marker gene, CaHIS1 or CaARG4, with AB and CD DNA fragments (∼400 bp each) corresponding to the 5′ and 3′ untranslated regions (UTR) of the target gene, respectively. The oligonucleotide primers used to PCR-amplify the AB and CD fragments, together with all the primers used in this study, are listed in Table 2. Appropriate restriction sites were added to some primers (underlined in Table 2) to facilitate ligation. Transformants were selected using appropriate dropout GMM media and correct deletion was verified by Southern blotting or PCR (data not shown). To construct a C. albicans strain deleted of both CaBNI1 and CaBNR1 genes, a URA3 flipper cassette was constructed by flanking the 4.2 kb URA3 flipper (Morschhauser et al., 1999) with the AB and CD DNA fragments corresponding to the 5′ and 3′ UTR of the CaBNI1 gene. This URA3 flipper cassette was used to disrupt the first copy of the CaBNI1 gene in the Cabnr1Δ strain (WYL22) as described previously (Morschhauser et al., 1999), yielding strain WYL23. The coding sequence of the second copy of CaBNI1 in WYL23 was placed under the control of the MET3 promoter by a promoter replacement strategy as described previously (Care et al., 1999) to yield strain WYL25.
Construction of GFP-tagged genes and CaBNI1 domain deletion strains
To construct a GFP-Bni1p fusion protein, the C. albicans MET3 promoter sequence was amplified by primers Met3p-fw and Met3p-re. The PCR product was cleaved with KpnI and XhoI and cloned in frame at the KpnI/XhoI sites of the plasmid pCIpGFPutr (Zheng et al., 2003), yielding plasmid pMet3-GFPutr. For the CaBni1p N-terminal GFP tagging, the N-terminal region of 1828 bp was amplified by PCR with primers 804Cla-F and 2584Nsi-R, cleaved with ClaI and NsiI, and cloned in frame at the ClaI/PstI sites of the plasmid pMet3-GFPutr. The resulting plasmid was linearized by PstI and transformed into WYL1. For the N-terminal GFP-tagging of CaKar9p, the entire ORF of CaKAR9 together with 900 bp of its 3′ UTR was amplified by PCR with primers Kar9-F and Kar9-R, cleaved with ClaI and PstI, and cloned in frame at the ClaI/PstI sites of the plasmid pMet3-GFPutr. The resulting plasmid was linearized by StuI and integrated into the RP10 locus of CAI4 or WYL2. For the CaBnr1p N-terminal GFP tagging, the N-terminal 1771 bp of the gene was amplified with primers Bnr1-F and Bnr1-R, cleaved with NarI and PstI and cloned in frame at the ClaI/PstI sites of the plasmid pMet3-GFPutr. The resulting plasmid was linearized by ClaI and transformed into WYL2 or CAI4. For the C-terminal GFP tagging of CaBud6p, the C-terminal 781 bp of CaBUD6 were amplified with primers Bud6-F and Bud6-R, cleaved with KpnI and XhoI and cloned in frame at the KpnI/XhoI sites of the plasmid pGFPutr (Zheng et al., 2003). The resulting plasmid was linearized by SalI and transformed into WYL2 or CAI4.
For the deletion of the Rho1-binding domain (RBD) domain of CaBNI1, the two regions immediately adjacent to the RBD domain were first PCR-amplified by using primer pairs Bni1-5′P/RBD-R and RBD-F/Bni1-3′P, and then fused together with primers Bni1-5′P and Bni1-3′P. This final PCR product was cleaved with ClaI and NsiI and cloned in frame at the ClaI/PstI sites in the plasmid pMet3-GFPutr. The resulting plasmid was linearized by StuI and integrated at the RP10 locus of Cabni1Δ mutant strain WYL2. Constructs for other domain deletion mutants of CaBNI1 were obtained by using the same strategy.
Actin, nuclear and cell wall staining and fluorescence microscopy
Actin, nuclear and cell wall staining were performed as previously described (Zheng et al., 2003). A Leica DMR fluorescence microscope with 100× objective and a Hamamatsu digital camera interfaced with METAMORPH software (Universal Imaging) were used for imaging.
Mouse systemic candidiasis
Systemic infection of mice and histology were carried out as previously described (Bai et al., 2002). Each animal was injected with 1×106 yeast cells and two groups of 10 mice were used for each strain. Two animals from each strain were killed 2 days after injection for histology and the rest were kept for 30 days to monitor survival.
C. albicans homologue of ScBNI1
CaBNI1 was identified through a BLAST search of the C. albicans genome database (http://www-sequence.stanford.edu) using the ScBni1p sequence. CaBni1p contains 1732 amino acids (aa) and shares 35% aa identity with ScBni1p. By using the Simple Modular Architecture Research Tool (SMART, http://smart.embl-heidelberg.de) and aa sequence alignment with ScBni1p we identified several functional domains in CaBni1p (Fig. 1). CaBni1p FH2 domain exhibits 50% aa identity with ScBni1p FH2, consistent with the core function of this domain in actin nucleation. Other domains are 20-35% identical between CaBni1p and ScBni1p. The FH1 domain, CaSpa2p-binding domain (SBD) and CaBud6p-binding domain (BBD) are the corresponding regions of ScBni1p defined in a previous study (Ozaki-Kuroda, 2001). C. albicans genome also contains a second formin CaBnr1p of 1485 aa residues, which shares overall aa identities of 21% and 14% with ScBnr1p and ScBni1p, respectively. The SMART analysis identified the FH2 and coiled-coil domains at similar positions as the corresponding domains in ScBni1p, ScBnr1p and CaBni1p. A proline-rich FH1 domain was found by sequence alignment. However, outside these three domains sequence similarities are not significant between CaBnr1p and the other formins.
Subcellular localization of CaBni1p in C. albicans yeast and hyphal cells
Previous studies found that ScBni1p and ScSpa2p are interacting partners in regulating polarized cell growth in S. cerevisiae. We observed that CaSpa2p localized to the site of cell growth in yeast cells in a cell cycle-dependent manner but was persistently present at the hyphal tips during hyphal growth (Zheng et al., 2003). To determine whether CaBni1p has the same subcellular localization as CaSpa2p, we constructed a GFP-CaBni1p fusion protein. GFP was fused to the N-terminal end of CaBni1p, because C-terminal fusion was earlier found to severely compromise CaBni1p function (our unpublished observation). When the GFP-CaBni1p fusion protein was expressed from the native promoter, the GFP signal was too weak for detection. Therefore, we placed the fusion protein under the control of the MET3 promoter. Since C. albicans is diploid, we first deleted one copy of CaBNI1 and then tagged the second with GFP so that the expressed fusion protein was the only source of CaBni1p function in the cell. The cell morphology, growth rates and hyphal growth of this strain are indistinguishable from the wild-type strain when grown in medium without methione and cysteine (data not shown), suggesting that GFP-CaBni1p is fully functional. This was also confirmed by its ability to rescue all the defects of the Cabni1Δ mutant (see below).
To characterize CaBni1p subcellular localization as a function of cell cycle progression, we examined GFP-CaBni1p localization using synchronous cultures. G1 yeast cells were obtained by centrifugal elutriation, released into GMM at 30°C and aliquots collected at 30-minute intervals for microscopic inspection. By ∼60 minutes a small fluorescent spot was clearly observed at the site of the emerging buds in many cells (Fig. 2A). Later, in small-budded cells the fluorescence was concentrated at the bud tip. In cells with a medium-sized bud, the label remained associated with the tip but became more scattered. Cells collected near cytokinesis were mostly large budded and the fluorescence disappeared from the tip and redistributed to the mother-daughter neck. In the next cell cycle, GFP-CaBni1p repeated this pattern of subcellular localization. When the G1 yeast cells were inoculated into the hypha-inducing medium (GMM containing 10% serum at 37°C) a bright fluorescent spot started to appear at 30 minutes in the majority of cells at the site of germ tube emergence (Fig. 2B). Throughout the next 7 hours under this condition, the fluorescence was seen to be continuously present at all hyphal tips. In less than 5% of the hyphal cells a band of fluorescence was also observed at the septum. Taken together, CaBni1p exhibits a pattern of cell cycle-controlled subcellular localization during yeast growth, whereas at least a fraction of the protein exhibits persistent, cell cycle-independent tip localization throughout hyphal growth, suggesting that it has a role in cell polarity. The patterns of CaBni1p localization in yeast and hyphal cells are similar to those of CaSpa2p and actin structures reported previously (Hazan et al., 2002; Zheng et al., 2003).
Cabni1Δ mutant exhibits multiple morphological defects
To further investigate the role of CaBNI1 in polarized cell growth, we constructed a Cabni1Δ mutant (WYL2) by sequentially replacing the two copies of the gene by CaARG4 and CaHIS1. Then a copy of CaURA3 was introduced into WYL2 by integrating the plasmid CIP10 at the RP10 gene locus yielding an ura+ Cabni1Δ strain (WYL3). A wild-type copy of CaBNI1 together with a copy of CaURA3 was also introduced into WYL2 via integration at the promoter region of the CaBNI1 locus, yielding the rescued strain WYL4. In WYL4 the defects observed in the Cabni1Δ mutants were fully rescued (see below). Cabni1Δ mutant cells exhibited a range of morphological defects. First, the mutant yeast cells were round instead of the typical ellipsoidal shape of the wild-type cells (Fig. 3A). The mutant cells were also considerably bigger. Second, the mother-daughter bud necks were markedly widened. Third, the mutant cells exhibited a random budding pattern in contrast to the bipolar budding pattern of the wild type (Fig. 3C). Next, under liquid hypha-inducing conditions, Cabni1Δ cells could produce germ tubes, which later grew into long hyphae in a manner similar to the wild-type cells. However, the germ tubes and hyphal cells of the mutant were significantly swollen (Fig. 3B). The results indicate that CaBni1p has a critical role in confining cell growth to a small cell surface area at the hyphal tips, which determines the diameter of hyphal cells. Cabni1Δ cells exhibited relatively normal filamentous growth on some solid media, such as YPD plus serum and RPMI medium (data not shown), but showed reduced filamentous growth on the solid medium containing only 1.5% agar and 0.05 mM ammonium sulphate (Fig. 3D). All the morphological defects of Cabni1Δ are reminiscent of those observed in the Caspa2Δ mutant (Zheng et al., 2003), consistent with the expected cooperative roles of the two proteins in regulating cell morphogenesis and polarity. However, the defects of the Cabni1Δ mutant are much less severe than those of Caspa2Δ cells. For example, a large fraction of Caspa2Δ yeast cells have highly elongated bud necks with misplaced septa (Zheng et al., 2003), which is not seen in Cabni1Δ cells. Furthermore, Caspa2Δ exhibited little filamentous growth on all solid media tested, whereas Cabni1Δ showed this defect only on certain solid media. The differences may be explained by the distinct roles of CaSpa2p and CaBni1p in cell polarity. Another reason for the less severe phenotype of the Cabni1Δ mutant might be the presence of CaBNR1, whereas unlike S. cerevisiae that has a SPA2-related gene SPH1, there is no SPH1 orthologue in C. albicans. Deletion of CaBNR1 alone had little effect on cell polarity (data not shown). We could not obtain a Cabni1Δ Cabnr1Δ mutant, suggesting that the two genes cannot be simultaneously deleted. We then used the MET3 promoter to control the expression of the single copy of CaBNI1 in a strain where one CaBNI1 and both CaBNR1 copies had been deleted. This strain grew well on the permissive GMM plate without methionine and cysteine but did not grow on the repressive plate supplemented with 2.5 mM of each of the two amino acids (Fig. 3E). Also, the strain grown in the permissive condition exhibited normal yeast morphology and formed true hyphae in response to serum (Fig. 3F). However, when the yeast cells were inoculated into the repressive condition, they became round and significantly swollen and failed to respond to serum. The results show that at least one formin must be present for viability and polarized growth.
Cabni1Δ mutants are defective in actin localization and nuclear positioning
Bni1p plays a critical role in actin cable assembly in S. cerevisiae. Next we wanted to determine the actin localization in Cabni1Δ cells. We first examined the yeast cells of a random population. In the wild-type cells, actin cortical patches polarized to the sites of cell growth such as the presumptive budding site, bud tip in small and medium-sized buds and bud neck in large budded cells (Fig. 4A). However, actin patches were delocalized in nearly all Cabni1Δ cells, exhibiting a more or less even cortical distribution in all phases. A similar random distribution of cortical actin patches was also observed in the Cabnr1Δ pMET3-CaBNI1 (strain WYL25) cells when the expression of CaBNI1 was shut off (Fig. 4D), even in the presence of serum. These results indicate that CaBni1p is required for normal actin localization in C. albicans yeast cells. The lack of polarized actin localization to the presumptive budding site and bud tip is consistent with the formation of round yeast cells. We next examined the actin localization in the hyphal cells (Fig. 4B). In contrast to the yeast cells, there was no significant difference in the general actin localization in hyphae between the wild type and Cabni1Δ except that the actin patches localized to a broader area at the swollen hyphal tips in the mutant. Actin patches were observed to polarize to hyphal tips persistently from the onset of germ tube formation throughout the hyphal growth in both strains. Similarly, actin cables were clearly seen to run along the long axis of the hyphal cells. The results indicate that CaBni1p is not indispensable for actin patch polarization to hyphal tips and the formation and orientation of actin cables in hyphal cells. It is striking that the activation of the hyphal program restored, to a considerable extent, the actin polarization in Cabni1Δ mutant. We speculated that CaBnr1p might be recruited to the hyphal tips in Cabni1Δ. To test this hypothesis, we created a GFP-CaBNR1 fusion gene under the control of the MET3 promoter and introduced it into Cabni1Δ. Fluorescence microscopy revealed that GFP-CaBnr1p was only observed at the bud neck in medium and large-budded yeast cells but never seen at the bud tips or presumptive budding sites (Fig. 4C). This localization pattern correlates well with that of ScBnr1p, recently described by Pruyne et al. (Pruyne et al., 2004). Strikingly, during hyphal growth the fluorescence could be clearly seen at the hyphal tips. The results indicate that during hyphal growth CaBnr1p can be recruited to the hyphal tips, providing a plausible explanation for the partial restoration of polarized actin localization to hyphal tips. The polarisome components CaSpa2p-GFP and CaBud6p-GFP were also observed to persistently localize to the hyphal tips in Cabni1Δ hyphal cells (Fig. 4C).
Since ScBni1p is known to have a role in positioning of the nucleus in the budding yeast (Miller et al., 1999), and the Caspa2Δ mutant has a severe nuclear localization defect (Zheng et al., 2003), we next performed nuclear staining. The exponential yeast cultures of Cabni1Δ were found to have 4.2% multinucleate cells (Fig. 5), approximately three-quarters of which were budded cells with two nuclei in the mother side and the rest were large single cells containing three to four nuclei. Such cells were not seen in the wild-type strain. The nuclear division within the mother compartment is a clear indication of mislocalized spindle elongation. The large multinucleate single cells suggest failure of bud formation and more than one round of nuclear division.
Cabni1Δ mutant is defective in mitotic spindle and cytoplasmic MT alignment
The S. cerevisiae bni1Δ mutant exhibits defects in cytoplasmic MT alignment and the positioning of mitotic spindle (Lee et al., 1999; Fujiwara et al., 1999). To determine whether CaBni1p has a similar function in C. albicans, we tagged one copy of the C. albicans tubulin gene CaTUB2 with GFP in both the wild-type and Cabni1Δ strains to visualize microtubule structures. Previous studies showed that this CaTub2p-GFP fusion is functional (Hazan et al., 2002; Zheng et al., 2003). Synchronized G1 cells were released into YPD liquid medium and incubated at 30°C for up to 7 hours. Aliquots were collected for microscopic examination at 30-minute intervals. In the majority of the large-budded wild-type cells with elongated spindles, the spindles were observed running across the bud neck along the mother-daughter axis (Fig. 6A, left) and the cytoplasmic MTs radiating from one spindle pole body nearly always extended toward the pole region of the mother or daughter cells. The spindles were rarely seen to have elongated within one cell body (<2%). In stark contrast, ∼60% of the large-budded Cabni1Δ cells had misoriented spindles that had elongated within the mother cell and the cytoplasmic MTs often extended in a random fashion (Fig. 6A, right). Furthermore, in ∼10% of the large-budded Cabni1Δ cells the cytoplasmic MTs emanating from one spindle pole extended separately into mother and daughter cells. The results indicate that CaBni1p is required for the normal alignment and positioning of mitotic spindles and cytoplasmic MTs.
To examine the MT structures in hyphal cells, synchronized G1 yeast cells were incubated in YPD containing 10% serum at 37°C for up to 7 hours. In nearly all wild-type hyphal cells, cytoplasmic MTs appeared to form a single bundle that extended all the way from the nucleus to hyphal tip (Fig. 6B). In stark contrast, three to four cytoplasmic MT bundles were observed in a majority of Cabni1Δ hyphal cells. Interestingly, these MT bundles were largely oriented towards the hyphal tip, though they appeared to end at separate sites in a large area of the tip. Since the mutant hyphal cells were severely swollen, it was difficult to capture an entire microtubule cable in the same focal plane. Thus the microtubule cables appeared broken, but by moving the focal plane up and down, we could see that the cables were continuous. Again, similar to what was previously observed in Caspa2Δ mutants, the results indicate that the hyphal program seems to be able to partially correct the MT defects of Cabni1Δ seen during yeast growth.
CaKar9p is mislocalized in Cabni1Δ mutants
In S. cerevisiae Bud6p and Kar9p play different roles in directing interactions between cytoplasmic MTs and cell cortex (Segal et al., 2000; Hwang et al., 2003; Liakopoulos et al., 2003; Huisman et al., 2004). Since we had found that in the Cabni1Δ mutant Cabud6p subcellular localization appeared normal in both yeast and hyphal cells (Fig. 4C and our unpublished data), we then examined CaKar9p cellular localization to see whether there is any correlation with the MT alignment defects in Cabni1Δ. We constructed a GFP-CaKAR9 fusion gene under the control of the MET3 promoter because the native promoter proved to be too weak (data not shown). GFP-CaKar9p was found to show cell cycle-dependent localization similar to that described for ScKar9p (Miller and Rose, 1998). In about 60% of small to medium-budded wild-type yeast cells GFP-CaKar9p was seen as a single small dot at the bud tip (Fig. 7A). By contrast, in nearly all Cabni1Δ cells GFP-CaKar9p appeared as multiple random cortical dots. In approximately 80% of the large-budded wild-type cells GFP-CaKar9p was observed as a short band at the bud neck and a dot in the cortex of both poles. In comparison, in ∼80% of large-budded Cabni1Δ cells GFP-Kar9p appeared as multiple random cortical dots. Under hypha-inducing conditions, GFP-CaKar9p signal was seen as a single dot at the hyphal tip in the wild-type cells, whereas multiple dots were found in almost all Cabni1Δ hyphal cells (Fig. 7B). Interestingly, unlike the totally random distribution pattern observed in Cabni1Δ yeast cells, these dots tended to cluster, albeit rather loosely, towards the hyphal tips. However, we are not clear whether the CaKar9p dots coincide with microtubule ends. Taken together, deletion of CaBNI1 dramatically altered GFP-CaKar9p localization, which correlates well with the MT misalignment defect. The clustering of GFP-CaKar9p dots towards the hyphal ends of Cabni1Δ mutants is consistent with the partial restoration of actin and MT structures under hypha-inducing conditions.
We next tested the dependence of CaKar9p cellular localization on actin and microtubules. We treated both yeast and hyphal cells expressing GFP-CaKar9p with either the actin depolymerizing drug latrunculin-A (LAT-A) or the microtubule disassembly drug nocodazole. We found that treating cells with 200 μM LAT-A or 40 μM nocodazole for 30 minutes resulted in a complete loss of the site-specific localization of CaKar9p in both yeast and hyphal cells (Fig. 7C). The results indicate that the cellular localization of CaKar9p is dependent on both actin and microtubules, consistent with the current understanding of the role of Kar9p in connecting actin cables and microtubules (Hwang et al., 2003; Liakopoulos et al., 2003; Huisman et al., 2004).
The role of different domains of CaBni1p in C. albicans growth
Formins have several conserved functional domains. To dissect the functions of CaBni1p, a series of mutant genes were generated by deleting an entire or part of a domain described in Fig. 8A. All the mutant proteins were tagged with GFP at the N terminus, placed under the control of the MET3 promoter and introduced into Cabni1Δ. Most of these proteins were expressed to comparable levels as estimated by both FACS analysis of GFP fluorescence intensity (Table 3) and visualization under a fluorescence microscope using the same exposure setting. First, we sought to define the region required for CaBni1p localization. Surprisingly, all the deletion mutants tested could polarize to cortical sites in manners similar to the wild-type protein during both yeast and hyphal growth, indicating that none of the regions deleted is essential for proper CaBni1p localization. We observed that if a mutant, such as RBDΔ and FH3Δ, rescued the `wide neck and swollen hypha' phenotype of the Cabni1Δ mutant, the fluorescence was highly concentrated at the hyphal tip, as in the wild-type strain. If a mutant failed to rescue this phenotype, as was the case with FH1Δ, the fluorescent label was in a broad crescent (Fig. 8B, left). One exception is the FH2Δ mutant (Fig. 8B, right). This mutant exhibited the phenotype of wide neck and swollen hyphae but the GFP fluorescence was in a small dot at the hyphal tip, suggesting that the FH2 domain may have a role in determining the organization of CaBni1p molecules at the bud and hyphal tips.
Next, we evaluated the effect of each deletion on CaBni1p function by examining the ability of each mutant to correct three defects of Cabni1Δ: swollen hyphae, multinucleate cells and random budding pattern (Table 3). Three mutants, 1-183Δ, RBDΔ and FH3Δ fully corrected the swollen hyphae of Cabni1Δ and only the FH3Δ mutant rescued the multinucleate phenotype. Surprisingly, all the domain deletion strains rescued the random budding pattern of Cabni1Δ. In summary, these results indicate that, (1) the RBD and FH3 domains were not required to define the diameter of hyphal cells. In fact, the RBDΔ and FH3Δ mutants did not exhibit any discernible defects under our experimental conditions, consistent with the observation by Ozaki-Kuroda et al. (Ozaki-Kuroda et al., 2001) that the RBD domain is apparently not required for any aspect of ScBni1p function and localization, (2) the FH3 domain was not required for nuclear positioning and (3) no single domain was indispensable for the bud site selection.
CaBNI1 is required for virulence
We next used the mouse systemic candidiasis model to determine whether CaBNI1 is required for infection and virulence. The growth rates of Cabni1Δ (WYL3) and CaBNI1 (WYL4) strains in yeast form were first compared in GMM and YPD media at 30°C and no significant difference was found (data not shown). Each mouse was injected with 1×106 yeast cells of either strain through the tail vein and monitored for death for 30 days. We found that the Cabni1Δ mutant exhibited reduced virulence (Fig. 9A). There was a delay in the death of animals injected with the mutant compared with the ones inoculated with the wild-type strain. Also, 30% of the mice infected with the mutants survived to the end of the 30-day observation, whereas the animals infected with the wild-type cells all died within 5 days. Histological examination of kidney sections of infected mice at the end of day 2 showed significantly less filamentous growth in mice infected with the mutant than in those infected with the wild type (Fig. 9B).
The polymorphic fungus C. albicans undergoes a phase of persistent polarized cell growth at a defined site of the cell surface during hyphal growth, making it an excellent model for functional characterization of the roles of polarity proteins. In an earlier study we characterized such a protein, CaSpa2p, and identified new roles of this protein in cell polarity control that were previously unknown from the studies of its budding yeast orthologue (Zheng et al., 2003). Here, we have investigated the role of CaBni1p, a possible interacting partner of CaSpa2p, in C. albicans yeast and hyphal growth. Consistent with its role in cell polarity, CaBni1p was found to specifically localize to the sites of cell growth. CaBNI1 deletion caused multiple defects in cell polarity, hyphal development and virulence.
Localization of CaBni1p is distinct in yeast and hyphal cells
By monitoring GFP-tagged CaBni1p in living cells, we observed a dynamic, cell cycle-dependent localization of the protein during yeast growth. CaBni1p localizes to the presumptive budding site in G1 cells, the bud tip of small to medium-budded cells and the bud neck of large-budded cells. This behavior indicates that CaBni1p cellular localization corresponds to sites of active cell growth and is regulated in a cell cycle-dependent manner. By contrast, during hyphal growth at least a fraction of CaBni1p was seen to persistently localize at the tips of germ tubes and hyphae independently of cell cycle progression. This specific localization at sites of cell growth suggests that CaBni1p has a role in polarized cell growth. CaSpa2p was previously shown to have a similar pattern of subcellular localization (Zheng et al., 2003), consistent with the involvement of Spa2p and Bni1p in controlling cell polarity and morphology that has been reported in the budding yeast (Sheu et al., 2000). Furthermore, the cell cycle-independent localization of CaBni1p and CaSpa2p during hyphal growth indicates that the hyphal signaling pathways have a dominant role over the cell cycle signals in determining the localization of these proteins.
CaBni1p is required for cytoplasmic MT and spindle alignment and positioning
Mutants defective in polarity control often show abnormal interactions of cytoplasmic MTs with cortical positional cues, leading to misalignment and mispositioning of mitotic spindles (Segal and Bloom, 2001; Ahringer, 2003). We observed that in a significant fraction of the large-budded Cabni1Δ cells the elongated spindle was either mispositioned and/or misaligned. Coincidently, the cytoplasmic MTs in the mutant cells appeared to make random contact with cell cortex and sometimes the MT bundles emerging from one spindle pole entered different cell bodies. Furthermore, the microtubule end-interacting protein CaKar9p in Cabni1Δ was found in multiple, randomly distributed spots in the cortical regions of both mother and daughter cells throughout the cell cycle, which is in marked contrast to a single cortical spot seen at the bud tip of many wild-type yeast cells. The results show a good correlation between the defects in MT orientation and positioning and the loss of CaKar9p localization to defined cortical regions in Cabni1Δ yeast cells. Although the possibility that this may be an artifact resulting from GFP-CaKar9p overexpression, as was described for Kar9p (Liakopoulos et al., 2003), the observed distinct patterns of GFP-Kar9p localization in the wild-type and Cabni1Δ cells may reflect altered interactions, direct or indirect, between CaKar9p and some cortical elements whose localization depends on CaBni1p function. ScBud6p also has an important role in capturing MT ends and is an interacting partner of Bni1p (Segal et al., 2000; Huisman et al., 2004). Although we have observed apparently normal bud and hyphal tip localization of CaBud6p and CaSpa2p in Cabni1Δ (Fig. 4C and our unpublished data), it is entirely possible that the organization and functions of the proteins in polarisomes in the absence of CaBni1p may be altered in a way that compromises MT-cortex interaction.
Suppression of Cabni1Δ polarity defects during hyphal growth
We observed that under hypha-inducing conditions, some of the defects of Cabni1Δ were partially suppressed, most notably the random actin localization, cytoplasmic MT misalignment and random CaKar9p distributions. First, in Cabni1Δ yeast cells actin patches are distributed largely at random in the cortex of both mother and daughter cells, indicating a loss of polarized localization. However, upon hyphal induction the mutant cells apparently regained the ability to polarize actin structures. The actin patches were found to localize towards hyphal tips, although the patches in Cabni1Δ hyphae appeared less concentrated than those in the wild-type hyphae. Given the essential role of actin structures in polarized cell growth, we speculated that other functionally redundant proteins, most likely the second formin CaBnr1p, might be recruited to hyphal tips. Indeed, using GFP tagging we observed persistent hyphal tip localization of GFP-CaBnr1p (Fig. 8B). In comparison, GFP-CaBnr1p was only detected at the bud neck in the yeast cells of Cabni1Δ. In addition, the polarisome proteins CaSpa2p and CaBud6p also exhibited persistent tip localization in Cabni1Δ hyphal cells. Second, Cabni1Δ yeast cells often exhibited two to three cytoplasmic MT bundles emanating from a single spindle pole body and extending towards random cortical sites in one or both cell bodies of large-budded cells, whereas in the wild-type cells only one or occasionally two such bundles were observed which normally extended towards the pole area of the mother or daughter cell. However, during hyphal growth the mutant cells reestablished the polarized alignment of cytoplasmic MT bundles with all the cytoplasmic MT bundles extending along the hyphal cells towards the hyphal tips, although these bundles appeared to end at different sites within a broad region of the hyphal tip. Third, the partial suppression of MT defects during hyphal growth seems to correlate well with CaKar9p localization. Unlike the random cortical localization of multiple CaKar9p spots in Cabni1Δ yeast cells, the majority of these spots were seen at or near hyphal tips. Based on current knowledge of the roles of the proteins discussed above and the suppression of the defects of Cabni1Δ during hyphal growth, we propose that the recruitment of CaBnr1p to the hyphal tip might play a crucial role. Localization of CaBnr1p to the hyphal tip may restore actin assembly towards the bud tip. The actin cables would then provide the tracks for the CaKar9p-mediated MT end transport to the hyphal tip. However, the partial suppression of the defects of Cabni1Δ also indicate that CaBni1p has other functions that cannot be fulfilled by CaBnr1p, such as the role of CaBni1p in restricting polarized cell growth to a small surface area. Recently, Pruyne et al. (Pruyne et al., 2004) reported that ScBnr1p specifically localizes to the bud neck and is responsible for the assembly of actin cables radiating from the bud neck into the mother cells. The neck localization of ScBnr1p depends on the septins. It is interesting that during C. albicans hyphal growth some septins are persistently present at the hyphal tips (our unpublished observation), which might provide the mechanism for CaBnr1p tip localization.
Functional domains of CaBni1p
Formins contain several conserved functional domains. The most highly conserved domains are the two juxtaposed FH1 and FH2 domains (Evangelista et al., 2003). The FH1 domain binds to the G-actin-binding protein profilin and the FH2 domain controls actin cable assembly (Evangelista et al., 2003). The FH1 domain of ScBni1p has been found to be essential for all its functions in apical growth, bud neck formation, bipolar bud site selection and distal bias at first bud site selection (Ozaki-Kuroda et al., 2001). However, in this study, the FH1 domain of CaBni1p was found to be not required for bud site selection (Table 3). In fact, no single domain alone was indispensable for bud site selection during yeast growth. Given that mutations in many genes encoding cytoskeleton components have been shown to cause random budding (Casamayor and Snyder, 2002), the lack of effect of the FH1 and FH2 domain deletion on bud site selection seems puzzling. There are at least two possible explanations. First, the loss of a single domain of CaBni1p may still allow the assembly of polarisomes, whose integrity may be important for bud site selection. Consistently, deletion of any one of the polarisome genes results in random budding (Sheu et al., 2000). Second, deletion of a single domain, such as FH2Δ, is expected to cause the loss of actin assembly at the growth site. This effect may be relatively more localized than the effects of the loss-of-function mutation of a cytoskeleton component gene. For instance, nearly all of the cytoskeleton components documented to have a role in bud site selection are associated with actin patches or both the patches and cables (Casamayor and Snyder, 2002). In the domain-deletion experiment, we also observed that none of the regions deleted appears to be essential in determining CaBni1p's localization to the sites of cell growth during both yeast and hyphal growth. This seems contradictory to the reported role of the FH3 domain of S. pombe formin Fus1p, which was shown to be necessary and sufficient for its subcellular localization (Petersen et al., 1998). However, the FH3 domain is weakly conserved among formin homology proteins such that its boundaries may be difficult to define. Ozaki-Kuroda et al. (Ozaki-Kuroda et al., 2001) proposed that the coiled-coil domain of ScBni1p is required for its localization at bud tips. However, this conclusion was based on the study of a truncation that removed the entire region from the coiled-coil domain to the C terminus. We propose that at least for CaBni1p, two or more domains might be functionally redundant in bud site selection and subcellular localization of the protein. We have previously found a similar functional redundancy between the N-terminal SHD-I domain and C-terminal SDH-V domain of CaSpa2p in determining cell morphology and polarity in C. albicans (Zheng et al., 2003). Although we were unable to pinpoint the domains responsible for CaBni1p localization, we noted that FH2 deletion had effects on the appearance of the GFP-CaBni1p fluorescence distinct from all the other domain deletion mutants. The GFP signal of the FH2Δ mutant appeared as a small spot at the bud and hyphal tip, whereas all the other mutant strains showed GFP signals as a crescent at the hyphal tip. Since the FH2 domain is directly involved in actin assembly, this activity may play a key role in determining how CaBni1p molecules are organized at the hyphal tips. A prominent feature of Cabni1Δ mutant is its swollen hyphae. RBDΔ, FH3Δ and 1-183Δ mutants grew wild-type-like hyphae, indicating that none of these domains is required for hyphal morphology. However, deletion of any one of the coiled-coil, SBD, FH1, FH2 and BBD domains resulted in the swollen hypha phenotype. The function of the coiled-coil domain is currently unknown. The SBD and BBD domains of ScBni1p have been shown to interact with ScSpa2p and ScBud6p. We have found that both Caspa2Δ (Zheng et al., 2003) and Cabud6Δ (our unpublished data) mutants showed the same hyphal morphology as Cabni1Δ. Thus, the interaction of CaBni1p with the polarisome components CaSpa2p and CaBud6p may play an important role in determining C. albicans hyphal morphology. The actin cable assembly activity of FH1 and FH2 domains is also crucial for the hyphal morphology, consistent with previous observation that inhibition of actin functions by cytochalasin A and latrunculin-A in hyphal cells caused swelling of the hyphal tips (Akashi et al., 1994; Hazan and Liu, 2002).
This work is supported by a fund from the Agency for Science, Technology and Research of Singapore. Y.W. is an adjunct associate professor in the Department of Microbiology, National University of Singapore.
- Accepted March 21, 2005.
- © The Company of Biologists Limited 2005