The PAK family kinase Cla4 is required for budding and morphogenesis in Ustilago maydis



The phytopathogenic basidiomycete Ustilago maydis displays a dimorphic switch between budding growth of haploid cells and filamentous growth of the dikaryon. In a screen for mutants affected in morphogenesis and cytokinesis, we identified the serine/threonine protein kinase Cla4, a member of the family of p21-activated kinases (PAKs). Cells, in which cla4 has been deleted, are viable but they are unable to bud properly. Instead, cla4 mutant cells grow as branched septate hyphae and divide by contraction and fission at septal cross walls. Delocalized deposition of chitinous cell wall material along the cell surface is observed in cla4 mutant cells. Deletion of the Cdc42/Rac1 interaction domain (CRIB) results in a constitutive active Cla4 kinase, whose expression is lethal for the cell. cla4 mutant cells are unable to induce pathogenic development in plants and to display filamentous growth in a mating reaction, although they are still able to secrete pheromone and to undergo cell fusion with wild-type cells. We propose that Cla4 is involved in the regulation of cell polarity during budding and filamentation.


The phytopathogenic basidiomycete Ustilago maydis exhibits a dimorphic switch during its life cycle. Haploid cells are unicellular and multiply vegetatively by budding (Banuett, 1995). For initiation of sexual development, compatible cells have to fuse and form the filamentous dikaryon. Only in this stage is the fungus able to infect corn plants and to induce tumours in all green parts of the plant. In U. maydis, the ability to switch from yeast-like budding to hyphal tip growth is intimately linked with virulence (for review, see Martinez-Espinoza et al., 2002). Because similar dimorphic transitions occur in a variety of pathogenic organisms, the analysis of the underlying molecular and genetic mechanisms will improve understanding for one of the fundamental characteristics of fungal virulence.

Morphogenetic transitions are programmed changes of cell shape that require the temporal and spatial organization of cellular growth. In most organisms reorganization of cytoskeletal elements and directed transport of vesicles is responsible for the observed remodelling of the cell. Small GTPases of the Rho/Rac family play a prominent role in such processes (Hall and Nobes, 2000). They act as molecular switches and exist in two forms: the inactive GDP-bound form and the active GTP-bound form. The transition between these states is regulated by guanine nucleotide exchange factors (GEFs) and GTPase activating proteins (GAPs). An intriguing feature of these molecular switches is that they can interact with a variety of different effector molecules (Van Aelst and D’Souza-Schorey, 1997). Among those are many members of the growing family of p21-activated kinases (PAKs) and actin-organizing complexes of the WASP family. How the specificity of the different effectors is reached and whether these signalling complexes act simultaneously or in distinct regulatory networks have yet to be elucidated.

We have recently identified a Rho/Rac-containing signalling module triggering cytokinesis during budding. The guanine nucleotide exchange factor Don1 and the serine/threonine kinase Don3 are required for the formation of a secondary septum to promote cell separation. Results from the yeast two-hybrid system indicated that the U. maydis homologue of the highly conserved small GTPase Cdc42 is likely to be involved in this process (Weinzierl et al., 2002). The protein kinase Don3 belongs to the germinal centre (GC) kinase subfamily of PAK family. Interestingly, the members of these kinases do not contain a CRIB domain, which is normally characteristic for interaction with small GTPases of the Rho/Rac family (Kyriakis, 1999).

Here we describe the identification of another protein kinase of the PAK family in U. maydis, Cla4, which regulates cell polarity in haploid cells. cla4 mutant cells are unable to bud properly. Instead, they form septate cell clusters that divide by fission septal cross walls. Delocalized deposition of chitin in cla4 mutant cells indicate that Cla4 is required for maintaining cell polarity during the budding process. Expression of a constitutively active Cla4 derivative is lethal. The distinct phenotypes of mutants deficient for the Ste20-like kinases Don3 and Cla4 imply that in U. maydis at least two independent Rho/Rac-signalling modules exist, which regulate budding and cytokinesis.


Isolation of mutants affected in cytokinesis and morphology

A Rho/Rac GTPase containing signalling cascade regulates cell separation during budding in the dimorphic fungus U. maydis (Weinzierl et al., 2002). To identify additional components involved in the regulation of cell separation, we used UV-mutagenesis to screen for mutants affected in cytokinesis. Mutants unable to separate after nuclear division are expected to form large cell aggregates in liquid culture, which could be enriched by filtration through nylon mesh (Hartwell et al., 1974). Mutants of the wild-type strain BUB8 that showed a cytokinesis defect in liquid culture were tested for complementation in crosses with compatible don1 and don3 mutant strains respectively. Five novel complementation groups could be identified and were termed don4, don5, don6, dol1 and cla4(Fig. 1A). The phenotype of don4, don5 and don6 mutants resembled very much that of the original don1 and don3 mutants, which are only affected in cell separation but not in cell shape (Weinzierl et al., 2002). This suggests that these mutants might be affected in components involved in the same regulatory network. The dol1 mutant cells generated also star-like cell clusters that have been already observed in ubc1 mutants defective in the regulatory subunit of the cAMP-dependent kinase (Gold et al., 1994).

Figure 1.

Phenotype of cytokinesis mutants.
A. Cells were grown in liquid medium and observed by differential interference phase contrast. The bar corresponds to 15 µm.
B. Transmission electron micrograph of a septal cross wall of cla4 mutant cells.

One of the novel cytokinesis mutants, BUB8cla4, showed an additional morphological phenotype. Cells were unable to form buds and grew as unusually shaped and branched hyphae consisting of uninuclear septate compartments (Fig. 1A). Transmission electron microscopy revealed that hyphal compartments are separated by cross walls consisting of a single septum (Fig. 1B). The mutant strain was crossed with the compatible haploid wild-type strain FB1 and the progeny were tested for the mutant phenotype. The cytokinesis defect was detected in 50% of the progeny (data not shown) indicating that a single gene was responsible for the observed phenotype.

The cla4 gene codes for a p21-activated kinase

Using a cosmid library derived from wild-type strain FBD11 and by subcloning, we could identify an open reading frame that complemented both the cytokinesis and the morphology defect. The derived sequence of 827 amino acids contained a pleckstrin homology (PH) domain, a CRIB domain characteristic for proteins interacting with small GTPases of the Rho/Rac family and a highly conserved serine/threonine protein kinase domain (GenBank Accession No. AY616187). The Cla4 protein is highly similar to that of Ste20α from Cryptococcus neoformans and Cla4p from Saccharomyces cerevisiae(Fig. 2A). Both proteins belong to the family of PAKs, which are involved in many different signalling pathways. We termed the U. maydis gene cla4, because its derived gene product contains an N-terminal PH domain, which is characteristic for the Cla4 subfamily of Ste20-like kinases (Fig. 2A). The yeast Cla4 protein is involved in the regulation of cell polarity and shares an essential function with the related protein kinase Ste20 (Cvrckova et al., 1995).

Figure 2.

The Cla4 protein belongs to the family of p21-activated protein kinases.
A. Schematic representation of Cla4 domain structure and comparison with related proteins. Numbers in the kinase domain indicate the percentage of identical amino acids within the kinase domain.
B. Sequence alignment of the highly conserved nucleotide-binding region. The asterisk indicates the leucine residue, which is exchanged to proline in the cla4 mutant.

To identify the mutation, which is responsible for the observed phenotype, the DNA corresponding to the kinase domain of Cla4 was amplified by polymerase chain reaction (PCR) from wild-type and cla4 mutant cells. Comparison of the derived amino acid sequences revealed a single amino acid exchange in the highly conserved ATP-binding region within the kinase domain of Cla4. A transition from T to C was detected which leads to the replacement of leucine by proline at position 637. Because the leucine residue is highly conserved within this alpha-helical region and proline is known to act as a helix breaker, it is very likely that this exchange interferes with kinase activity (Fig. 2B).

cla4 deletion mutants are viable

To find out whether cla4 is essential in U. maydis, a deletion mutant was constructed, in which the complete open reading frame of cla4 was replaced by a hygromycin cassette. The disrupted allele of cla4 was introduced into different wild-type strains by homologous recombination. Gene replacement was confirmed by Southern analysis (data not shown).

The phenotype of FB1Δcla4 mutant cells resembled that of the originally identified cla4 mutant BUB8cla4. Cells were distorted and severely affected in cytokinesis. Clusters of septate hyphal cells formed irregular branched structures (Fig. 3A). Cells of the deletion mutant FB2Δcla4 displayed a slightly different phenotype from that of the BUB8cla4 or FB1Δcla4 mutants. In liquid culture, cells were elongated and often contained a central cross wall (Fig. 3B, arrowheads). Cell separation occurred by contraction at the centrally located septum (Fig. 3B, arrow). Restart of cell growth adjacent to the dividing septum often resulted in bending of daughter cells. Closer inspection of the BUB8cla4 and FB1Δcla4 strains revealed that cell separation by contraction could also be found in these mutants. But in general, cytokinesis and morphology were more severely affected in these strains than in FB2Δcla4. At least four independent Δcla4 mutants were constructed for each strain. In all cases the same difference in phenotype was observed. We take this as indication that the cla4 mutant phenotype is affected by the genetic background of strains FB1, BUB8 and FB2.

Figure 3.

Phenotype of cla4 deletion mutants.
A. FB1Δcla4 cells form branched hyphae which are septate.
B. FB2Δcla4 mutant cells divide by fission of a centrally located cross wall. Arrowheads indicate septa and the arrow points to a division site.

Cla4 is required for mating and pathogenicity but not for cell fusion

To initiate sexual development, U. maydis cells have to fuse to form the dikaryon, which is pathogenic for the host plant corn. Cell fusion events can be induced by co-cultivation of compatible haploid strains on charcoal-containing agar plates (Day and Anagnostakis, 1971). Dikaryotic cells grow as filaments and can be recognized as a white fuzzy mycelium on top of colonies. The Δcla4 mutant strains were crossed with compatible wild-type and mutant strains. Formation of dikaryotic mycelium could be clearly observed in combination of the FB2Δcla4 (a2 b2Δcla4) mutant with the wild-type strain FB1 (a1 b1) (Fig. 4A). However, when FB1Δcla4 (a1 b1Δcla4) mutant cells were mixed with the compatible wild-type strain FB2 (a2 b2), no filament formation occurred. The same mating deficiency was observed when compatible combinations of Δcla4 strains were mixed (Fig. 4A). To test whether cla4 mutants still secrete pheromone, which is required to induce cell fusion, cells were tested in combination with diploid strains that are homozygous for a but heterozygous at the b locus. These cells react on the presence of pheromone by filament formation without undergoing a cell fusion event (Laity et al., 1995). Thus, these strains have been used as pheromone tester strains to detect the presence of specific pheromone (Spellig et al., 1994). Both Δcla4 mutant strains are clearly able to fully induce filament formation in such tester strains (Fig. 4A). To test whether both strains are principally able to undergo cell fusion, we co-cultivated both Δcla4 mutants with strains heterozygous for a but homozygous for the respective compatible b allele. Here, we could detect a normal mating reaction for FB2Δcla4 (a2 b2Δcla4), but a significantly delayed reaction for FB1Δcla4 (a1 b1Δcla4) (Fig. 4A, arrowhead). This indicates that the mating reaction in the genetic background of FB1 strains is somehow impaired but not completely abolished. Thus, cla4 cannot be essential for the fusion event itself, at least with wild-type cells. This is supported by the observation that both mutant strains, FB1Δcla4 and FB2Δcla4, are able to infect corn plants when mixed with compatible wild-type strains (Fig. 4B). So far we have no conclusive explanation why cla4 deletion mutants behaved differently depending on their genetic background if crossed to wild-type cells.

Figure 4.

Cla4 mutants are affected in mating and pathogenicity.
A. Filament formation is observed in the plate mating assay only for the compatible combinations of FB2Δcla4 (a2 b2Δcla4) with FB1 (a1 b1) but not for the combination of FB1Δcla4 (a1 b1Δcla4) with FB2 (a2 b2). Both cla4 mutant strains are able to secrete pheromone because filament formation is induced in the pheromone tester strains FBD12-17 (a2 a2 b1 b2) and FBD11-7 (a1 a1 b1 b2) respectively. Fusion of FB1Δcla4 (a1 b1Δcla4) with strain FBD11-21 (a1 a2 b2 b2) that is heterozygous for a occurs only after incubation for 48 h (see arrowhead), whereas FB2Δcla4 (a2 b2Δcla4) reacts already after 24 h with strain FBD12-3 (a1 a2 b1 b1).
B. Tumours induced by the cross of FB2Δcla4 with FB1 are significantly larger than those induced by FB1Δcla4 in combination with FB2. The arrowhead indicates a small tumour at the stem of an infected plant.

cla4 mutants are non-pathogenic

To test whether the inability of cla4 mutants to show a mating reaction on charcoal plates affects the virulence, compatible combinations of Δcla4 mutant cells were used to infect young corn seedlings. Whereas the combination of cla4 mutants with compatible wild-type strains resulted in the induction of anthocyanine biosynthesis and tumour formation, a combination of compatible Δcla4 mutant strains was completely avirulent (Table 1). The tumours induced by the combination FB2Δcla4 with FB1 were significantly more prominent than those induced by the complementary combination FB1Δcla4 with FB2 (Fig. 4B). This parallels the inability of FB1Δcla4 mutants to form a filamentous dikaryon with wild-type strains on charcoal-containing agar plates. However, the presence of plant symptoms clearly indicate that FB1Δcla4 cells are able to fuse with compatible wild-type cells, although with significantly lower efficiency.

Table 1. Pathogenicity assay.
 Infected plantsTumours
FB1 × FB2 85
FB1 × FB2Δcla4128
FB1Δcla4 × FB2163
FB1Δcla4 × FB2Δcla4720

Cla4 regulates cell wall deposition in haploid cells

cla4 mutants exhibit abnormal morphology and do not form proper buds. In contrast to normal haploid cells, cla4 mutants grow as branched cell clusters that divide by fission at septal cross walls. To test whether this is caused by aberrant deposition of cell wall material, cells were stained with the dyes calcofluor white and fluorescently labelled wheat germ agglutinin (WGA). Calcofluor white is specific for 1,3-β-glucan that constitutes a major portion of fungal cell walls. WGA mainly detects free ends of chitin polymers. In wild-type cells both fluorescent dyes stain the two septa that were formed during cytokinesis (Fig. 5). In cla4 mutant cells, however, a completely different picture is observed: the septal cross walls between hyphal compartments were brightly stained with calcofluor white (Fig. 5A) but cannot be stained by WGA (Fig. 5B). Instead, WGA detects massive chitin deposition along the cell wall (Fig. 5B). This indicates that Cla4 is directly or indirectly involved in localizing chitin deposition to the expanding buds and to the growing septa or that Cla4 controls other components involved in cell polarization. Interestingly, the correct localization of β-glucan polysaccharides to septa appears not to be affected in cla4 mutant cells.

Figure 5.

Cell wall synthesis is altered in cla4 mutants.
A. Staining with calcofluor white reveals the presence of multiple cross walls in the branched hyphae of cla4 mutants.
B. Staining with WGA indicates a massive delocalized deposition of free chitin in cla4 mutants. In contrast to wild-type cells, septa appear to be free of chitin in cla4 mutant cells.

Expression of a constitutive active Cla4 derivative is lethal

Cla4 contains a N-terminal CRIB domain characteristic for members of PAKs. Unstimulated PAK kinases form homodimers in which the region that contains the CRIB domain binds and inhibits the catalytic activity of the kinase domain in trans (Parrini et al., 2002). Binding of Cdc42 or Rac1 in its GTP-bound active form stimulate kinase activity by alleviating the inhibitory function of the N-terminal region. To determine which of the known small GTPases might be involved in the activation of the PAK kinase Cla4, the CRIB domain was tested in the yeast two-hybrid assay for interaction with U. maydis Cdc42, Rac1, Rho1 and Rho3 respectively. As a control, the CRIB domain of the U. maydis Ste20 homologue Smu1 (Smith et al., 2004) was also tested in this assay. A clear interaction could be detected for both Cdc42 and Rac1, but not for Rho1 and Rho3 (Table 2). Interestingly, the CRIB domains of Cla4 and Smu1 interacted with comparable affinities with both Cdc42 and Rac1. Introduction of a single amino acid exchange, which renders the GTPases constitutively active (Q61L), resulted in stronger interactions with the CRIB domain (Table 2).

Table 2. Yeast two-hybrid assay.
  • a

    . LacZ activity (in Miller units) was determined in three independent replicate experiments.

Cdc4229 ± 7a67 ± 2
Cdc42Q61L56 ± 2.594 ± 17
Rac114 ± 223 ± 1
Rac1Q61L52 ± 1.380 ± 3.5
Rho1Q61L 3 ± 0.6 2 ± 0.3
Rho3Q61L 6 ± 1.3 2 ± 0.5

Overexpression of such constitutive active GTPases in U. maydis wild-type cells results in significantly different phenotypes. Whereas expression of constitutive active Rac1 is lethal, overexpression of constitutive active Cdc42 results in cell elongation and rescue of the don1 cell separation defect (M. Mahlert, in preparation). To get a clue which GTPase activates Cla4 in vivo, we constructed mutant Cla4 proteins that either lack the CRIB domain alone (Cla4ΔCRIB) or carry an N-terminal deletion including both the PH and CRIB domain (Cla4ΔNter) (see Fig. 6A). Deletion of the CRIB domain should result in a constitutive active kinase domain because this region is required both for autoinhibition and regulation (Hoffman and Cerione, 2000). Expression of Cla4ΔNter did not result in any significant phenotype. This could indicate that the presence of the PH domain might be necessary for correct localization and/or function of the Cla4 kinase. Overexpression of Cla4ΔCRIB, however, was lethal for the cells (Fig. 6B). Closer inspection revealed that cells stopped budding and died with a phenotype very similar to those expressing constitutive active Rac1 (data not shown). We take this as an indication that in vivo Cla4 is most probably an effector of Rac1 but not of Cdc42. We cannot exclude, however, the possibility that Cla4 might also serve as an effector for Cdc42 in a specific signalling complex.

Figure 6.

Expression of constitutive active Cla4 protein is lethal.
A. Schematic overview of N-terminal deletions of Cla4.
B. Expression of the Cla4ΔCRIB protein, which lacks the CRIB domain, is lethal in U. maydis wild-type cells. The overproduction of either full-length Cla4 or Cla4ΔNter did not affect viability.

Actin distribution during cell elongation in cla4 mutant cells

To find out whether deletion of Cla4 affects the polar growth of haploid cells, we followed the elongation of Δcla4 cells by time-lapse microscopy. In Fig. 7A, a single cell is visible that elongates only on one end. After 80 min, septum formation occurs in the central portion. At this time point, elongation starts also in opposite direction in the old mother cell (Fig. 7A, compare 80 min and 100 min). The direction of growth is often coordinated by the organization of the actin cytoskeleton. Therefore, we visualized the actin distribution in wild-type and cla4 mutant cells by immunofluorescence. In budding cells, actin is concentrated in the growth cone of the newly formed bud (Fig. 7B). At later stages actin is localized at both sides of the septum separating mother and daughter cells (Banuett and Herskowitz, 2002). This is reminiscent of cytokinesis in fission yeast where a contractile actomyosin ring is observed in dividing cells (Le Goff et al., 1999; Rajagopalan et al., 2003). In cla4 mutants, a similar polar localization of actin is observed at the elongating end of the growing daughter cell (Fig. 7B). However, no concentration of actin is observed in cla4 mutants at the cross wall separating mother and daughter cell (Fig. 7B). These observations point to a role of Cla4 in regulating the characteristic distribution and relocation of actin during the cell cycle. Whereas the polar accumulation at the growing tips is not affected in cla4 mutants, they appear to be unable to relocate actin to the central septum.

Figure 7.

Actin distribution in elongating Δcla4 cells.
A. Time-lapse microscopy of a Δcla4 mutant cell. Cells were placed on agar and microphotographs were taken every 20 min. Cell elongation occurs mostly on one side. Elongation at the opposite pole starts after initiation of septum formation (arrowheads).
B. Actin localization was detected by immunofluorescence microscopy. Actin accumulates at the newly formed bud in wild-type cells (left). In Δcla4 cells actin is localized at the growing end of the elongating cell (right). Some accumulation of actin is visible at the distal end of the mother cell. The arrowhead indicates the septum between mother and daughter cell.


The study of fungal dimorphism allows the molecular and genetic dissection of a morphological switch in a simple model system. Here we could show that in U. maydis Cla4, a member of the large family of p21-activated protein kinases, is required for budding and morphogenesis. These kinases are regulated by interaction with small GTPases of the Rho/Rac family (Sells and Chernoff, 1997), which is mediated by the conserved CRIB interaction domain located at the N-terminal portion of these kinases. In their unbound state, PAK family kinases dimerize and the CRIB domain acts as autoinhibitory repressor of kinase activity (Hoffman and Cerione, 2000; Lei et al., 2000; Parrini et al., 2002). Binding of activated GTPases of the Rho/Rac family relieves the intermolecular repression of kinase activity and promotes autophosphorylation (Parrini et al., 2002).

Which small GTPase activates the U. maydis Cla4 kinase?

The Cla4 kinase was first identified in the yeast S. cerevisiae where it shares an essential function with the related kinase Ste20 in the regulation of cell polarity (Cvrckova et al., 1995). Interestingly, in S. cerevisiae only Cdc42 is found but no Rac protein, whereas U. maydis contains homologues of both GTPases (Weinzierl et al., 2002). We could show that both Rac1 and Cdc42 are able to interact with the CRIB domains of Cla4 and Ste20 in the yeast two-hybrid system. Deletion of rac1 in U. maydis cells results in a morphological phenotype highly similar to that of cla4 mutants and expression of a dominant active Rac1 mutant is lethal (M. Mahlert, L. Leveleki, A. Hlubek, and M. Bölker, in preparation). Together with the observation that expression of the constitutive active Cla4ΔCRIB also results in a lethal phenotype, this strongly suggests that Cla4 acts in U. maydis as an effector of Rac1. At present, however, it is unclear how the small GTPases Cdc42 and Rac1 discriminate between the PAK kinases Cla4 and Ste20. In the human PAK1 kinase, which is normally activated by both Cdc42 and Rac, mutants have been identified that selectively couple this enzyme to Cdc42 (Reeder et al., 2001). Some of these mutations map outside of the p21-binding domain, indicating that binding specificity is not solely determined by the CRIB domain. Because in our experiments only isolated CRIB domains of U. maydis Cla4 and Ste20 have been used, it has to be tested whether full-length kinases may display a different binding specificity towards Rac1 and Cdc42.

It remains to be elucidated which targets are phosphorylated by Cla4 kinase in U. maydis. In yeast, both the Ste20 and Cla4 kinases are able to phosphorylate actin-dependent motor proteins of the myosin-I family (Wu et al., 1996). The phosphorylation site of myosin-I for PAK-like protein kinases was shown to be essential for its function (Wu et al., 1997). Whether myosin is a potential target for Cla4 in U. maydis is currently under investigation.

Cla4 is required for proper budding

Mutants deleted for cla4 display a more or less pronounced distorted cell shape and are deficient in the normal budding process. In particular, we have never observed cell separation by a duplicated septum characteristic for budding wild-type cells. Bud formation requires the coordinated switching from polarized growth at the apical end to isotropic expansion of the bud. Apical growth occurs by targeted fusion of vesicles containing cell wall material. This process is driven by motor proteins acting on the tubulin and actin cytoskeleton (Bretscher, 2003). In U. maydis, the unipolar growth of budding cells is accompanied by an asymmetric concentration of actin in the growth zone (Banuett and Herskowitz, 2002). The localization of actin in cla4 mutants indicates a function of Cla4 in the relocalization of the actin cytoskeleton to the septum during cytokinesis. cla4 mutant cells do not form regular buds and lack a visible constriction at the neck between mother and daughter cells. This might point to a function of Cla4 in regulating the assembly of a proper septin ring at the mother-bud neck. A similar dependence of neck integrity on Cla4 activity has been described for S. cerevisiae where Cla4 is required for the correct localization of the septin ring (Schmidt et al., 2003). In yeast, targeting of Cla4 to the septin ring is also required for phosphorylation and downregulation of Swe1, a negative regulator of mitotic cyclin-dependent kinase Cdc28 (Sakchaisri et al. 2004).

The septa observed in U. maydis cla4 mutants resemble more the cross walls of dikaryotic hyphae than double septa of budding cells. Similar septa have been observed also in budding cells that were blocked in cell cycle by feeding-starving cycles. In those cultures, binucleate cells were enriched that show a characteristic cross wall in the middle of the cell (Holliday, 1965; Snetselaar and McCann, 1997). Such cells obviously have decided to undergo a mitotic division in the absence of bud formation. It is feasible that cla4 mutant cells behave similarly and uses this type of cross wall to separate the dividing nuclei after mitosis.

Cla4 regulates cell morphogenesis

A characteristic feature of cla4 mutant cells is the highly delocalized deposition of chitinous material in the cell wall. This phenotype is reminiscent of that of temperature-sensitive cdc42 and cdc24 mutants in yeast, which exhibit similar overall chitin deposition at the non-permissive temperature (Adams et al., 1990). The delocalization of chitin is probably the result of a defect in the targeted localization of vesicles containing the enzymatic apparatus and precursors for chitin biosynthesis (Sietsma et al., 1996; Ziman et al., 1996). Interestingly, the localization of 1,3-β-glucans that can be stained with the dye calcofluor white appears to be undisturbed in cla4 mutants indicating a separate regulation for these different cell wall components. It is known that glucan synthesis requires the activity of another small GTPase, the Rho1 protein (Arellano et al., 1996; Drgonova et al., 1996; Qadota et al., 1996). In yeast, it was demonstrated that Rho1 is part of the enzyme 1,3-beta- d-glucan synthase, which is inactive in the absence of Rho1 protein (Mazur and Baginsky, 1996). Thus, it appears that during budding two independent mechanisms cooperate in organizing the growth of the fungal cell wall.

cla4 mutants are still able to fuse with wild-type cells of opposite mating type, although with low efficiency, indicating that Cla4 is not essential for induction of the pheromone response and cell fusion. However, cla4 mutants do not form filaments and are non-pathogenic if crossed with each other. This could indicate either that cla4 mutants are unable to fuse or that filament formation is blocked in these mutants. Yarrowia lipolytica cla4 mutants are also unable to grow as hyphae and display a similar aberrant distribution of chitin in the cell wall (Szabo, 2001). Whereas in U. maydis cell wall staining with calcofluor was nearly unchanged in cla4 mutants, in Y. lipolytica calcofluor staining reveals an abnormal distribution of cell wall material (Szabo, 2001).

We have observed characteristic differences in the morphological phenotype of cla4 mutants that depend on the genetic background of the strains used for the generation of mutants. In particular, filament formation was observed only after fusion of FB2Δcla4 cells with FB1 wild-type cells but not vice versa. In addition, infection symptoms differed significantly depending on the combination of strains. So far, we have no conclusive explanation for this phenomenon. Inspection of the genomic sequence revealed that cla4 resides within 80 kb distance of the a mating type locus. Quite intriguingly, a similar close linkage of cla4 to the mating type locus has also been observed in C. neoformans, where the Cla4 homologue exists in two different mating type-specific versions (Wang et al., 2002). Interestingly, mutants deleted for the U. maydis Ste20 homologue Smu1 show a similar but opposite dependence on the a mating type (Smith et al., 2004). Whereas cla4 mutants are able to fuse efficiently only in the a2 background, smu1 mutants do this only as a1 strains (Smith et al., 2004). This does not result from a lack of pheromone secretion because both mutant strains are still able to stimulate pheromone tester strains (Fig. 4A). A possible explanation could be that Cla4 and Smu1 might exert redundant functions during cell fusion but only in a specific a mating type background. Thus, if Cla4 is required for cell fusion only in a1 strains and Smu1 only in a2 strains, then deletion of cla4 would interfere with mating in a1 strains and deletion of smu1 would affect mating only in a2 strains. With this respect, it would be interesting to examine the phenotype of a double mutant deleted for both cla4 and smu1.

We have previously identified the Rho/Rac-specific guanine nucleotide exchange factor Don1, which regulates the induction of a secondary septum during cytokinesis. Don1 interacts in the yeast two-hybrid system specifically with Cdc42 but not with Rac1 (Weinzierl et al., 2002). In addition, expression of a constitutive active Cdc42 variant rescues the phenotype of don1 mutant cells (L. Leveleki, unpubl.). These data suggest that in this signalling module Cdc42 acts as central GTPase. Thus, in U. maydis at least two independent small GTPase signalling modules are involved in the regulation of cell polarity and cytokinesis.


Quite recently, the genome sequence of U. maydis has been made publicly available. Sequence inspection revealed that U. maydis contains seven small GTPases of the Rho/Rac family. Because U. maydis is an easily tractable genetic system with an intricate life cycle, it can serve as an interesting model system to dissect the molecular functions of small GTPases, in particular the different contributions of the highly related proteins Cdc42 and Rac1 to the organization of intracellular growth.

Experimental procedures

Strains, plasmids and culture conditions

Ustilago maydis haploid strains FB1 (a1 b1), FB2 (a2 b2), BUB8 (a2 b4) have been described (Schulz et al., 1990). The following diploid U. maydis strains were used: FBD12-17 (a2 a2 b1 b2), FBD11-7 (a1 a1 b1 b2), FBD11-21 (a1 a2 b2 b2) and FBD12-3 (a1 a2 b1 b1) (Banuett and Herskowitz, 1989). Cells were grown at 28°C in liquid YEPS (1% yeast extract, 2% peptone, 2% sucrose), in liquid potato dextrose broth (PDB) (2.4% PDB) or on solid potato dextrose agar (PDA). Solid media contained 1.5% (w/v) bacto-agar. For selection, PD plates were used containing 2 µg ml−1 carboxin or 200 µg ml−1 hygromycin. S. cerevisiae strain Y153 (Durfee et al., 1993) was used for yeast two-hybrid experiments. Escherichia coli strains DH5α and XL1-Blue MR were used for all DNA manipulations.

Isolation of the cytokinesis mutants

Cells of U. maydis strain BUB8 were grown overnight in liquid YEPS medium, centrifuged and resuspended in sterile water. A suspension (2 × 104 cells ml−1) was exposed to UV light (260 nm), at a rate of 200 µJ s−1 with a UV cross-linker (Stratagene). 2 × 103 cells were plated on each PD plate. Fifteen per cent of the cells survived. Mutants were grown in liquid culture and cell aggregates were enriched by filtration through a nylon mesh. After plating on low agarose medium, donut-shaped colonies were selected that indicate a cytokinesis defect (Weinzierl et al., 2002).

The cla4 gene was isolated by complementing the cla4 mutant strain with an autonomously replicating cosmid library. The complementing cosmid clone was recovered by transformation into E. coli and the cla4 gene was identified by subcloning on a 7.8 kb HindIII–NotI fragment.

To identify the UV-induced mutation responsible for the cla4-1 allele, genomic DNA was isolated from BUB8 and BUB8cla4-1 cells and used as template for PCR amplification. Primers MB299: CCACGCAGCTTTCCAT CGCC, MB300: GAAGGCTCCACTGATGGGAT, MB301: CTTTGTCCACCAGGTTCATG, MB302: CACGTTGTCCATG ATCTGCG, MB303: CGGAGCGCCGTATCAGCACC and MB304: GTGGCAGCATCGCTGGCTTG were used in pairs to generate overlapping fragments of ≈950 bp. The PCR fragments were sequenced and sequences were analysed using baseimager v.4.0 and sequencher 4.0.5 programs.

Construction of the cla4 mutant alleles

A PCR-based method similar to that published by Davidson et al. (2002) was used to generate a cla4 deletion construct in which the open reading frame of cla4 was replaced by the hygromycin resistance cassette, flanked by 1 kb of homologous sequences on both sides. For amplification of the 5′ flanking sequences, primers MB328: GAGCTCAATTCAGT TGTCGTTGC and MB329: GCGGCCGCAGACGCTGTCTG ATGG were used and for the 3′ flank MB330: GCGGCCG CAGGTGCACTCCCTTCCC and MB331: GGATCCGGAAT CAACCTACAAGC. The PCR products were digested with SacI and NotI (5′ region) or BamHI and NotI (3′ region), respectively, and ligated together with a hygromycin cassette, which carries NotI sites at both sides, into pUC18.

The plasmid pCRG-Cla4 expressing the wild-type Cla4 under control of the arabinose inducible crg-promoter (Bottin et al., 1996) was constructed by amplifying the Cla4 open reading frame with primers MB830 (GGATCCATGTCTG GCTTTCGTGG) and MB833 (GCGGCCGCTCAACGATCGG GCTTGTTGC). The PCR product was cloned into the BamHI and NotI sites of pCRG-KN-Cbx, which contains a small polylinker region downstream of the crg-promoter. For pCRG-Cla4ΔNter the corresponding DNA was amplified by PCR using the primers MB832 (CATATGGACATTCAGAAGCGC GAG) and MB833 (see above) and cloned into the NdeI and NotI sites of pCRG-KN-Cbx. pCRG-Cla4ΔCRIB was constructed by amplification of the first 600 nt using the primers MB830 (see above) and MB831 (CATATGGAGCGGCGAGC GAGAGTAG) and cloning into the BamHI and NdeI sites of pCRG-Cla4ΔNter. Transformed FB2 cells were grown overnight in rich medium in the presence of 2% glucose. Five microlitres of confluent cells were diluted in 1 ml H2O. This solution was diluted three times by 1:10 steps. Five microlitres of each dilution were spotted onto either glucose (OFF) or arabinose (ON) containing plates and incubated for 2 days.

Yeast two-hybrid system

Yeast two-hybrid assays were performed as described (Fields and Song, 1989). Construction of plasmids pGAD424-Cdc42, pGAD424-Cdc42Q61L, pGAD424-Rac1, pGAD424-Rac1Q61L, pGAD424-Rho1Q61L and pGAD424-Rho3Q61L were described by Weinzierl et al. (2002).

The CRIB domain of UmCla4 (aa 194–278) was amplified by PCR using primers MB496 (GATGGATCCTCGAGATC TACTCTCGCTCG) and MB497 (GCACTGCAGCTAATTCAT GGTAGGCGTGCC). The CRIB domain of Smu1 (aa 308–391) was amplified by PCR using primers MB498 (GAT GGATCCTCGAGGTCTTCTCGGCTCAG) and MB499 (GCACTGCAGCTAAGCGCCCATCTTTTTCC). Fragments were digested with BamHI and PstI and cloned into the BamHI and PstI sites of pGBT9 (Clontech) resulting in plasmids pGBT9-Cla4-CRIB and pGBT9-Smu1-CRIB respectively. Transformation of yeast strains was performed as described (Woods and Gietz, 2001).

Plant inoculations

Maize seedlings (variety Early Golden Bantam purchased from Olds Seed Company) were grown in a growth chamber (14 h light at 28°C, and 10 h dark at 20°C at 60% humidity) and after 9 days injected with c. 300 µl of cell suspensions. For the isolation of progeny, spores were collected 10–14 days after inoculation and plated on plates containing 2% water agar. After 2 days microcolonies (with about 200 cells) were picked and streaked on PD plates containing 1.5% agar. Mutants could be distinguished from wild type by their colony morphology.

Plate and drop mating assays

Plate mating assays were performed by placing mixtures of cultures on PD plates containing 1% activated charcoal and incubation at room temperature (Day and Anagnostakis, 1971).

Microscopy and staining

Cells from logarithmically growing cultures were analysed. WGA Oregon Green 488 was observed using a specific filter system I 3, blue (BP 450–490, LP 515); for visualization of Alexa Fluor 594 filter system N 2.1, green (BP 515–560, LP 590) of a Leica TCS SP2 Confocal Laser Scanning Microscope was used. Image processing was performed with Leica Confocal Software LCS (Leica) and Photoshop (Adobe).

For immunofluorescence of actin, cells were fixed for 30 min by addition of formaldehyde to a final concentration of 1% and, washed with phosphate-buffered saline (PBS; pH 7.2) and applied to slides precoated with 10 mg ml−1 poly- l-lysine. Cells were washed with PBS and treated with 3 mg ml−1 Novozyme for 50 min. After several washes with PBS, cells were incubated with 1% Triton X-100 for 1 min, followed by additional washes and incubation in blocking reagent (1% BSA in PBS, pH 7.2). Antibodies against actin (mouse IgG1 monoclonal C4; Lessard, 1988) and secondary antibodies [Alexa Fluor® 594 goat anti-mouse IgG (H+l), F(ab′)2 fragment conjugate, A-11020, Molecular Probes] were diluted in blocking reagent and applied for 15 min.

Electron microscopy

Cells were fixed with 3.7% formaldehyde and 0.2% glutaraldehyde grade I. After 1 h incubation cells were pelleted (4200 Upm, 5 min, 4°C) and resuspended in 3.7% formaldehyde in 0.1 M phosphate buffer. Fixed cells were transferred to 2.3 M sucrose and incubated overnight. Sections were made with a cryomicrotrome. Electron microscopy was done after contrasting the cells with uranyl acetate.


We are indebted to technical assistance by M. Piscator and M. Johannsen. We thank U. Kämper for critical reading of the manuscript. These studies were supported by grants from the Deutsche Forschungsgemeinschaft (BO 1237/1-1 and BO1237/1-2).