Sustained cell polarity and virulence in the phytopathogenic fungus Ustilago maydis depends on an essential cyclin-dependent kinase from the Cdk5/Pho85 family

Cell polarity is of fundamental importance for proliferation, differentiation and morphogenesis of both unicellular and multicellular organisms. The ability of cells to dynamically reorganize their cytoskeleton is crucial for rapid responses to environmental changes, as well as for the differentiation of distinct cell types during development. Therefore, cell polarity is controlled by both intra- and extracellular signals, which must be integrated to give the final shape of every particular cell. In mammals, Cdk5 represents an important example of a master element controlling different aspects of polarity, which allows efficient cytoskeleton remodeling in migrating neurons in response to external signals (Xie et al., 2006). Cdk5 belongs to a family of cyclin-dependent kinases that seem to be involved in the control of cell differentiation and morphology rather than cell division (Dhavan and Tsai, 2001). Cdk5 and its activating subunit, p35, have been implicated in neuronal differentiation (Nikolic et al., 1996; Xie et al., 2006). Loss of Cdk5 in neurons resulted in abnormal organization with no recognizable pattern of neurite arrangement (Chae et al., 1997). Among the different targets of Cdk5 there are Rho-like proteins such as Rac1 (Nikolic et al., 1998) and RhoA (Kawauchi et al., 2006), as well as cytoskeleton regulators such as Fak1 (Xie et al., 2003) and NUDEL (Niethammer et al., 2000).

Like neurons, fungal cells also provide an extraordinary example of polarized cell growth that efficiently responds to environmental cues. Interestingly, functional homologs of Cdk5 have been described in fungi (Huang et al., 1999; Nishizawa et al., 1999; Dou et al., 2003). The founding member of this family, the Pho85 cyclin-dependent kinase, was isolated in Saccharomyces cerevisiae because of its involvement in the regulation of phosphate-scavenging enzymes, and it has been implicated in many other cellular processes, including stress adaptation, glycogen storage, cell cycle progression and morphogenesis (for a review, see Carroll and O'Shea, 2002). The Pho85 kinase shows genetic interactions with genes encoding proteins involved in cell polarity such as Cdc42, its regulators (i.e. Bem1 and Cdc24) and downstream effectors (i.e. Cla4) (Lenburg and O'Shea, 2001; Huang et al., 2002; Moffat and Andrews, 2004). In the filamentous fungus Aspergillus nidulans two highly related Cdk5-like proteins, PhoA and PhoB, also play important roles integrating environmental conditions and developmental responses (Bussink and Osmani, 1998; Dou et al., 2003). Interestingly, although S. cerevisiae Pho85 is not essential for growth, deletion of both A. nidulans phoA and phoB caused lethality (Dou et al., 2003). The Cdk5 orthologs of A. nidulans seem to have roles in polarized growth. Conidia of a phoA^ts/ΔphoB mutant strain under restrictive conditions were unable to initiate polarized growth once germinated but instead continued isotropic growth (Dou et al., 2003).

The ability to induce strong polar growth is particularly dramatic in dimorphic fungi, which are able to produce a yeast-to-hypha transition in response to environmental conditions. Dimorphic transitions are common among pathogenic fungi and are often associated with a change in virulence (Gow et al., 2002). Therefore, proteins that control dimorphic transitions probably play major roles in microbial adaptation to new environments and in pathogenicity. Surprisingly, in spite of the apparent importance of the role of Cdk5-like kinases in morphogenesis, little is known about the role of these kinases in dimorphic fungi. To date, the presence of members of the Pho85/Cdk5 family has been described in dimorphic fungi such as Candida albicans (Miyakawa, 2000) and Sporothrix schenckii (de Jesús-Berrios and Rodríguez del Valle, 2002), but no further characterization of these regulators has been reported. In this study we took advantage of the plant pathogen Ustilago maydis, a known model system to study dimorphism and virulence (Bölker, 2001), to characterize the roles that Cdk5/Pho85-like kinases may have in pathogenic dimorphic fungi. During maize infection, U. maydis undergoes a dimorphic transition from budding, yeast-like cells to a filamentous dikaryon that proliferates in the host. This transition is regulated by mating and environmental signals. Here, we describe the identification and characterization of Cdk5, a Cdk5/Pho85 homolog in U. maydis. We provide evidence that this kinase is essential for growth and has a major role in the maintenance of cell polarity in this phytopathogen. Using temperature-sensitive mutants, we show that Cdk5 is required at all morphological stages of U. maydis – the yeast-like sporidia, the conjugation tubes and the dikaryotic hyphae – thereby affecting its ability to infect plants.

To identify a Cdk5/Pho85 homolog in U. maydis we searched its genome (Kämper et al., 2006) for open reading frames showing high sequence similarity with human Cdk5 as well as S. cerevisiae Pho85 sequences. We found two hits with E values lower than e^–80. The first one corresponded to an uncharacterized protein that we named Cdk5 (GenBank accession number EF190891). The second hit corresponded to Cdk1, the mitotic cyclin-dependent kinase (CDK) from U. maydis (García-Muse et al., 2004). Both Cdk1 and the predicted Cdk5 protein retain the PSTAIRE domain characteristic of CDK molecules involved in cyclin binding (Fig. 1A). Western blot assays using anti-PSTAIRE antibodies, showed that Cdk5 protein levels do not change throughout the cell cycle (Fig. 1C). We believe that Cdk5 represents a true Cdk5/Pho85 homolog because a dendrogram analysis placed Cdk5 and Cdk1 in different subfamilies, with Cdk5 in the Cdk5/Pho85 subfamily of CDK proteins (Fig. 1B). To further analyze the function of Cdk5, we replaced one cdk5 allele in the diploid FBD11 strain with a null allele carrying a nourseotricine resistance (Nat^R) marker. As a control we constructed a diploid strain carrying a Nat^R-marked cdk5 allele producing a functional C-terminal Myc-tagged protein. After sporulation, we analyzed the meiotic progeny of these strains and we did not find any Nat-resistant cells among the haploid offspring of the diploid strain carrying the Δcdk5 null allele (n=50 spores analyzed). By contrast, roughly half of the cells in the progeny of the diploid strain carrying the Myc-tagged cdk5 allele were Nat resistant (52.4% resistant in n=30 spores analyzed). These results indicated that cdk5 is essential in U. maydis.

To investigate the subcellular localization of the Cdk5 protein, we introduced a green fluorescent protein (GFP) fusion at the end of the chromosomal cdk5 gene in wild-type cells. The C-terminal tagging did not interfere with the function of the Cdk5 protein and the tagged strain grew as well as wild-type cells. GFP-tagged Cdk5 was distributed throughout the cytoplasm and concentrates within the nucleus (Fig. 1D). The GFP signal seems to be associated with the nucleus through the different cell cycle phases (Fig. 1D, upper panels), although the nuclear accumulation was lost in large-budded cells (i.e. those undergoing mitosis), which could be explained by the nuclear envelope breakage that occurs during mitosis in U. maydis (Straube et al., 2005).

To gain further insight into the role of Cdk5 in U. maydis we generated a temperature-sensitive mutant allele (cdk5^ts) that contained a point mutation (Y286H) in a conserved region at the C-terminus that confers temperature sensitivity to other CDKs, such as A. nidulans PhoA and NimX and S. pombe Cdc2 (Osmani et al., 1994; Dou et al., 2003). Cells carrying the cdk5^ts allele were unable to grow in solid complete medium (CMD) at the restrictive temperature (34°C) whereas at the permissive temperature (22°C) they grew at a similar rate as the wild-type strain (Fig. 2A).

We analyzed the morphology of cdk5^ts cells in liquid complete medium (CMD). At the permissive temperature cells had a wild-type appearance, with only minor defects such as a less-elongated cell body (Fig. 2B, left panel). After they were shifted to 34°C for 8 hours, mutant cells frequently arrested their growth as ellipsoidal cell doublets (Fig. 2B, right panel), with each cell carrying a single nucleus. Fluorescence-activated cell sorting (FACS) analysis indicated a 2C DNA content in each doublet, suggesting that each nucleus was arrested at G1 phase (see supplementary material Fig. S1). In wild-type cells, FITC-labelled wheat germ agglutinin (WGA), a lectin that binds to oligomeric chitin (Nagata and Burger, 1974), accumulated at the tip of the cells as well as in newly formed septa during cell separation. By contrast, in cdk5^ts cells, WGA reacted strongly with the lateral cell walls (Fig. 2B), suggesting an anomalous cell wall in cdk5^ts cells. To address whether the associated defects were reversed by temperature shift-down, the mutant cells grown at the permissive temperature of 22°C were shifted up to the restrictive temperature of 34°C and kept there for 6 hours, and then shifted down to 22°C and incubated for several hours. After the shift-down of the culture, the cell poles were re-established, and the morphology returned to a wild-type appearance (see supplementary material Fig. S2). However, we observed that cdk5^ts cells incubated at the restrictive temperature for long periods (more than 12 hours) had a strong lysis phenotype as evidenced by the frequent presence of cell `ghosts' and cell debris (Fig. 2C arrows). As these results were coherent with a cell wall defect, we tested the sensitivity of cdk5^ts cells to compounds described to affect cell wall integrity (i.e. caffeine, Congo Red, Calcofluor White, chlorpromazine and sodium dodecyl sulfate) (Levin, 2005). Interestingly, at the permissive temperature, cdk5^ts cells showed a greater sensitivity than wild-type cells to all these compounds (Fig. 2D). However, we found that the temperature-sensitive growth and cell lysis defect of the mutant were not rescued by 1M sorbitol, an osmotic stabilizer known to rescue cell wall defects (see supplementary material Fig. S3B,C), suggesting that the primary defect might not be a defective cell wall structure.

To further characterize the morphogenetic defects caused by the cdk5^ts mutation, samples incubated at restrictive conditions for different times were stained with calcofluor to visualize zones of active growth (Wedlich-Söldner et al., 2000). Early in the time course (Fig. 3A), between 2 and 4 hours after the temperature shift, cells were able to bud, although the mother cell tended to swell, giving thicker cells. This morphology change was accompanied by a misplaced accumulation of calcofluor staining (around 60% of the cells showed delocalization of calcofluor deposition at 6 hours). Later in the time course, cells stopped budding, became more spherical and formed septa, as well as accumulated calcofluor in the lateral cell walls. These results suggested that the observed cell wall defects and ellipsoidal morphology could be a consequence of impaired cell polarization in the absence of Cdk5.

In U. maydis the formation of the bud takes place at the G2 phase, the cell cycle stage at which polar growth reaches it maximum level (Straube et al., 2003). Since we observed that cdk5^ts cells incubated at restrictive conditions arrested mostly at G1, we reasoned that the apparent defect in polarization could be a consequence of the cell cycle arrest in a cell stage not suitable for polar growth. To distinguish whether the morphogenetic defect of cdk5^ts was a direct cause of a lack of polarity or just a consequence of a G1 cell cycle arrest, we arrested cells in G2 phase by overproduction of the mitotic inhibitor Wee1 (Sgarlata and Pérez-Martín, 2005) and then shifted them to 34°C. In a strain carrying a wild-type cdk5 allele, Wee1-induced G2 arrest resulted in a continuous polar growth in the bud apex leading to cells with elongated buds (Fig. 3B). However in the strain carrying the cdk5^ts allele, no elongated buds were observed, although a strong swelling at both cell tips was detected. These cells were G2 arrested as they showed a single nucleus (Fig. 3B, inset) with a 2C DNA content (not shown). This result indicated that depletion of Cdk5 resulted in a lack of polar growth even in the G2 phase of the cell cycle.

The polarity defective phenotypes caused by the cdk5^ts mutation could reflect an underlying defect in cytoskeletal organization or function. Therefore, we monitored the effect of cytoskeleton inhibitors on wild-type and cdk5^ts cells at the permissive temperature (Fig. 3C). In the presence of 1 μM benomyl, a microtubule destabilizer, no difference in growth on solid medium was observed between wild-type and cdk5^ts cells. By contrast, cdk5^ts cells were more sensitive than the wild type to the actin inhibitor cytochalasin D, suggesting that the actin cytoskeleton was affected.

It was previously demonstrated that polar growth of U. maydis requires an intact actin cytoskeleton (Fuchs et al., 2005). Therefore, we labeled F-actin by tagging the endogenous copy of the fimbrin-like gene fim1 (accession number XM755822.1). U. maydis Fim1 is 61% identical to Schizosaccharomyces pombe Fim1 and 62% identical to S. cerevisiae Sac6, and shares a similar domain structure (Fig. 4A; P-values for the predicted calponin-homology domains are indicated). Both ScSac6p and SpFim1 are components of actin patches (Adams et al., 1989; Huckaba et al., 2004; Pelham and Chang, 2001) in which they are thought to organize F-actin (Adams et al., 1991). Consistent with its assumed role in actin patches, a fusion protein of UmFim1 and GFP localized to patches in the growing bud in wild-type background cells (Fig. 4B, `con') and only a minor fraction was found in the mother cell (not shown). After treating the cells with 50 μM Latrunculin A for 30 minutes, this polar localization of Fim1-GFP was lost (Fig. 4B, LatA), suggesting that Fim1-GFP concentrated in actin patches. Consistent with this conclusion, Fim1-GFP patches showed a high turn-over (not shown) and occasionally moved within the cell (Fig. 4C, arrow), a behavior that is typical for actin patches in S. cerevisiae (Huckaba et al., 2004). In cdk5^ts mutant cells at 22°C, actin patches containing Fim1-GFP also concentrated in the bud (Fig. 4D), but this distribution was lost after 4-5 hours at the restrictive temperature and patches were randomly distributed within the mother cell and the bud (Fig. 4E,F; bud indicated by arrow).

The loss of polarization of actin patches observed in cdk5^ts cells could be explained either by a defect in the initiation of polar growth (i.e. establishment of polarity) or by a defect in the maintenance of the polar growth. To distinguish between these possibilities, we introduced a fusion protein of GFP and Myo5, a class-V myosin from U. maydis (Weber et al., 2003), into the cdk5^ts mutant background. In wild-type cells (Fig. 5A), the GFP-Myo5 protein concentrated at the bud tip and transiently appeared at the distal cell pole, where it is thought to mediate actin-dependent secretion (Weber et al., 2003). A similar distribution was found in cdk5^ts cells at 22°C (Fig. 5B, arrows) in which the polar GFP-Myo5 cluster remained stationary over time (Fig. 5B, compare T=0s and T=16s). By contrast, at the restrictive temperature (34°C, 4-5 hours) GFP-Myo5 polar accumulation was absent in many cells (Fig. 5C). Occasionally, GFP-Myo5 concentrated at cell poles but this localization was only transient (arrow in Fig. 5C, compare T=0s and T=16s). A more detailed analysis revealed that at the restrictive temperature, GFP-Myo5 signal appeared at the cell poles, but disappeared rapidly (Fig. 5C, inset; Fig. 5D, 34°C), a behavior that was never found in control cells (not shown) or cdk5^ts mutants at the permissive temperature (Fig. 5D, 22°C; see also supplementary material Movie 1). These results indicate that Cdk5 does not seem to be involved in the initial recruitment of Myo5 to polarization axes, but it is required for its subsequent retention. In other words, these data strongly suggest that Cdk5 is required to stabilize the polarization of the cell.

The amplification of polarization signals and subsequent stabilization of polarity axes have been recently shown in S. cerevisiae to be dependent on the Rho-like protein Cdc42 (Nern and Arkowitz, 2000; Irazoqui et al., 2003; Wedlich-Söldner et al., 2003). In dimorphic fungi, the Rho-like proteins Rac1 and Cdc42 play essential roles in cytoskeleton organization and polarization (Hurtado et al., 2000; Ushinsky et al., 2002; Boyce et al., 2003; Vallim et al., 2005; Bassilana and Arkowitz, 2006; Mahlert et al., 2006). In U. maydis, Rac1 is required to sustain polar growth, whereas Cdc42 has a role in cell separation (Mahlert et al., 2006). Since Cdk5 seems to be required for the stabilization of polarity axes in U. maydis, we wondered whether it acts in the same genetic cascade as Rac1. rac1 overexpression is sufficient to induce filament formation in U. maydis (Mahlert et al., 2006) (Fig. 6A). However, the overexpression of rac1 in a strain carrying the cdk5^ts allele at the restrictive temperature did not result in filament formation (Fig. 6A). Therefore, Cdk5 was required for Rac1-induced polar growth. To address whether Cdk5 is a downstream effector or an upstream activator of Rac1 we took advantage of the constitutively active rac1^Q61L allele, whose expression in wild-type cells produces a delocalized isotropic cell wall extension (Mahlert et al., 2006) (Fig. 6A). We overexpressed the rac1^Q61L allele in cdk5^ts cells and we found that it produced the same effect in cdk5^ts mutant cells as in wild-type cells (Fig. 6A). As rac1^Q61L-mediated activity is independent of upstream activation, this result suggests that Cdk5 might be acting as an upstream activator of Rac1. Rho-like proteins cycle between an inactive GDP-bound and an active GTP-bound state. GDP/GTP cycling is regulated positively by a guanine-nucleotide exchange factor (GEF). The U. maydis Cdc24 homolog is highly likely to be a Rac1 GEF: it localizes at the bud tip in growing cells; physically interacts with Rac1; and induces filamentation upon overexpression in a rac1-dependent manner (I.A.-T. and J.P.-M., unpublished results) (see also Mahlert et al., 2005). We wondered whether Cdk5 was acting at the level of Cdc24, and therefore we overexpressed cdc24 in cells carrying the cdk5^ts allele. Strikingly, we found that cdc24 overexpression was able to induce polar growth in cdk5^ts cells (Fig. 6A), supporting the notion that Cdk5 promotes Rac1 activation via Cdc24. To address whether cdc24 overexpression might counterbalance the failure of cdk5^ts cells to maintain the polarity, we analyzed the localization of Myo5-GFP in the cdc24-overexpressing cells. We found that high levels of Cdc24 suppressed the inability of the cdk5^ts cells to retain the Myo5-GFP signal at the growing tip (Fig. 6B and supplementary material Movies 2 and 3).

Since cdc24 overexpression was able to bypass the requirement of Cdk5 for the maintenance of the polar growth in U. maydis cells, we analyzed whether the cell growth defect of cdk5^ts was rescued by high levels of Cdc24. However, cdc24 overexpression did not suppress the growth defect of cdk5^ts cells at restrictive temperature (Fig. 6C). It is noteworthy that although cdk5^ts cells growing at 34°C and expressing high levels of cdc24 were able to polarize, they failed to form long filaments. Instead they lysed shortly after polarization has occurred (data not shown), suggesting that Cdk5 has additional roles, most likely related with the ability to stabilize the cell wall.

We analyzed the localization of Cdc24 to the incipient bud in cdk5^ts cells, as it could be a key determinant in the establishment and maintenance of an axis of polarity through activation of Rac1. Strikingly, in cdk5^ts cells incubated under permissive conditions it was possible to find Cdc24 at the growing pole of the cell, as in wild-type cells, whereas under restrictive conditions fluorescence was dispersed throughout the cell with no discrete accumulation at the cell tip (Fig. 7A,B). The levels of Cdc24-GFP fusion protein were not affected by the temperature shift, as assessed by western blot (not shown). These results supported the notion that Cdk5 could be required to stabilize Cdc24 at the cell tip. The ability to keep Cdc24 at the tip could be involved in the induction of a positive feedback loop that stabilizes polarity axes in this fungus, as it has been proposed for Cdc24 in budding yeast (Butty et al., 2002).

To determine the importance of Cdk5 in virulence, we infected maize plants with a mixture of compatible control strains FB1 and FB2 as well as the temperature-sensitive mutant strains FB1cdk5^ts and FB2cdk5^ts. To circumvent the associated growth defects of cdk5^ts cells at the restrictive temperature (34°C), infected plants were incubated at both permissive (22°C) and semipermissive (30°C) temperatures. We found that at 30°C, cdk5^ts cells were able to grow, the cell lysis defect was avoided but they still displayed clear morphological defects (see Fig. S3A in supplementary material). At 22°C, infection symptoms (measured as a percentage of tumor formation after 14 days, Fig. 8A) were found at roughly similar rates in plants infected with both wild-type and mutant cells. However, at the semipermissive temperature, cdk5^ts mutant strains failed to induce symptoms, whereas normal infection was observed in plants infected with control strains (Fig. 8A).

Plant infection by U. maydis requires the formation of conjugative hyphae, which, after cell fusion, establish an infective filament that invades the epidermis and continues the pathogenic program inside the plant. The formation of dikaryotic filaments can be monitored on charcoal-containing agar plates (Holliday, 1974). To further analyze the impaired virulence in cdk5^ts strains, we performed mating assays using compatible mutant strains as well as compatible control strains. At the permissive temperature, both compatible control and cdk5^ts strains fused to form a fuzzy white colony consisting of dikaryotic filaments. By contrast, at 30°C only the control cells formed fuzzy colonies, whereas cdk5^ts mutants were unable to form dikaryotic hyphae (Fig. 8B).

As polarized growth of conjugative and infecting hyphae is proposed to play a key role during infection, we sought to analyze the role of Cdk5 in these processes. For this, we introduced the cdk5^ts allele into FB1Fuz7DD and AB31 haploid strains. In FB1Fuz7DD cells, the overexpression of an activated form of the MAPK kinase Fuz7 induces the formation of conjugative hyphae mimicking an active pheromone signaling (Fig. 8C, left panel) (Müller et al., 2003). In AB31 cells, overexpression of a compatible bW/bE heterodimer induces the formation of infective hyphae (Fig. 8D, left panel) (Brachmann et al., 2001). Strikingly, upon Fuz7DD or bW2/bE1 induction, a clear defective polar growth was observed in cells carrying the cdk5^ts allele growing under semipermissive conditions (30°C, Fig. 8C,D, right panels). These results support the notion that Cdk5 is required for sustained polar growth in U. maydis during pathogenic conditions, and they help to explain the absence of virulence of cdk5^ts strains.

The functional characterization of the Cdk5/Pho85 homolog in U. maydis reported in this work indicated that it has an essential role most likely related to the cellular morphogenesis. Previous work has identified in U. maydis another essential cyclin-dependent kinase, Cdk1, a functional homolog of the cell-cycle-specific Cdc2/Cdc28 kinase (García-Muse et al., 2004). Although Cdk1 has a clear role in cell cycle progression in U. maydis, we believe that Cdk5 has a major role in the regulation of polarity in this fungus. In budding yeast, Pho85 appears to be required for cell cycle progression, collaborating with G1 cyclin-associated Cdc28 kinase in the degradation of the CDK inhibitor Sic1 (Nishizawa et al., 1998). We observed that U. maydis cdk5^ts cells incubated at the restrictive temperature arrested mostly at G1, suggesting a cell cycle defect. However, the accumulation of arrested cells occurs after several hours upon temperature shift, and we believe that the observed defects in cell integrity in this mutant could be responsible for the apparent G1 cell cycle arrest. Therefore, although we cannot discount a role of Cdk5 in cell cycle progression in U. maydis, we favor the hypothesis that defects in cell integrity could activate a sort of checkpoint mechanism arresting the cell cycle at G1 in cdk5^ts cells incubated at the restrictive temperature.

The reason for the essentiality of Cdk5-like kinases in A. nidulans and U. maydis versus the non-essential role of Pho85 in S. cerevisiae is unclear. An interesting possibility to explain this difference is the different amount of polar growth that occurs in these organisms. A. nidulans grows in a polar manner over its entire cell cycle. By contrast, budding yeast undergoes a brief period of polar growth during bud emergence, but then switches to isotropic growth, over the entire surface of the bud. In U. maydis, during the yeast-like phase, the bud expansion relies almost entirely on sustained polar growth (Steinberg et al., 2001). Assuming a major role of fungal Cdk5-like kinases in sustained polar growth, it is plausible that the absence of Cdk5-like activity will be more deleterious in A. nidulans and U. maydis, which depend on continued polar growth, than in budding yeast, which have only a brief period of polar growth. An alternative possibility may rely on functional redundancy with other cellular components. In S. cerevisiae it is well established that Pho85 becomes essential in the absence of the G1 Cdc28-associated cyclins Cln1 and Cln2 (Measday et al., 1994). In U. maydis there is a single G1 cyclin, which interacts with Cdk1 and appears to have cell cycle and morphogenetic roles (Castillo-Lluva and Pérez-Martín, 2005). However, we found that overexpression of this cyclin does not suppress the growth defects of cdk5^ts cells (J.P.-M., unpublished observations), suggesting that no such redundant function is present in U. maydis.

Our results indicated that Cdk5 is required for the accumulation of Cdc24 at the cell pole. We also observed that high levels of Cdc24 bypassed the requirement of Cdk5 to induce polarization. We believe that some of the polarization defects associated with cdk5^ts mutants at the restrictive temperature can be explained by defects in the localization of Cdc24. Cdc24 is essential for growth in U. maydis and cells carrying a conditional cdc24 allele, and growing at restrictive conditions recapitulates some of the phenotypes of cdk5^ts cells grown at 34°C, including swollen cells, often in form of doublets, with a highly delocalized deposition of chitinous material in the cell wall (see supplementary material Fig. S4). In S. cerevisiae, polarity mutants defective in cdc24 function exhibit similar chitin deposition as a consequence of a defect in the targeted localization of vesicles containing the enzymatic apparatus and precursors for chitin biosynthesis (Adams et al., 1990). The mechanism behind the Cdk5-dependent stabilization of Cdc24 at the cell pole is not understood. One possibility is that Cdc24 is a target of Cdk5, as it has been proposed for S. cerevisiae Pho85 (Moffat and Andrews, 2004), and its phosphorylation allows or enhances its interaction with some anchor structure at the membrane. Alternatively, Cdk5 may activate an anchor protein that mediates Cdc24 recruitment at the cell pole. We tried to demonstrate physical interaction between Cdc24 and Cdk5 with no reproducible results, suggesting some indirect interaction (J.P.-M., unpublished observations). Furthermore, we have no evidence of phosphorylation of Cdc24 by Cdk5 (J.P.-M., unpublished observations). In S. cerevisiae, the scaffolding protein Bem1 maintains Cdc24 at sites of polarized growth (Butty et al., 2002), and it is subjected to phosphorylation by different kinases (Loog and Morgan, 2005; Han et al., 2005). There is a putative U. maydis Bem1 homolog (accession number EAK85966), and future effort will be directed to address whether it plays a role in Cdc24 localization and whether it could be a target of Cdk5.

Strikingly, in spite of the inability to localize Cdc24 to the cell pole, in cdk5^ts cells growing at the restrictive temperature, Myo5 could be found at the cell poles, although this localization was only transient and usually disappeared rapidly. This result indicates that actin-mediated transport is still present in cdk5^ts cells at the restrictive temperature, most likely because other polarity determinants – for instance formins – can assemble actin cables. However, it seems that the elements transported to the cell end cannot be stabilized at the pole and they therefore diffuse freely. This could explain why actin patches concentrate at the tip of the bud in wild-type cells whereas in cdk5^ts mutant cells at the restrictive temperature they are distributed throughout the entire cell. It also may explain the progressive delocalization of the calcofluor staining in cdk5^ts cells incubated at restrictive conditions. In fungi, septins are diffusion barriers that specify the boundaries of the apical growth zone (Barral et al., 2000). The U. maydis genome contains four septin genes, from which only one (sep3) has been characterized (Boyce et al., 2005). Interestingly, deletion of sep3 results in cells with aberrant morphology which resembles that observed in cdk5^ts mutants (Boyce et al., 2005), and therefore it is tempting to propose that septins in U. maydis could be regulated, albeit indirectly, by Cdk5. Proper stabilization of septin assembly in S. cerevisiae requires the activity of the Cla4 kinase (Dobbelaere et al., 2003; Schmidt et al., 2003; Versele and Thorner, 2004), which is activated through binding to Cdc42-GTP. In U. maydis it is thought that Cla4 – which is required for sustained polarity and is activated by Rac1 (Leveleki et al., 2004) – could be involved in septin apparatus stabilization (Leveleki et al., 2004). Therefore, we believe that the inability to properly activate Rac1 because of the delocalization of Cdc24 in cdk5^ts cells, could result in the loss of diffusion barriers. We found that cdk5^ts shows a synthetic growth defect at the permissive temperature in cells carrying deletion of rac1 or cla4 (J.P.-M., unpublished observations).

It is noteworthy that although cdk5^ts cells growing at 34°C and expressing high levels of cdc24 were able to polarize, the cdc24 overexpression did not suppress the growth defect of cdk5^ts cells at the restrictive temperature. This result indicates that the Cdk5 kinase is important for some step additional to polarization, such as cargo delivery or fusion to the growth site, or local function of the cell wall remodeling machinery. During periods of polarized growth, cell wall remodeling must be carried out in a highly regulated manner: the growth site is loosened enough to allow expansion but not so much as to risk rupture. Additional roles of Cdk5 related with cell wall integrity are consistent with the finding that the cdk5^ts strain was more sensitive to cell wall stressors than the wild-type strain, even at the permissive temperature. We also found a synthetic lethality at the permissive temperature in strains carrying the cdk5^ts allele and a deletion of mpk1, encoding a MAPK kinase involved in the cell wall integrity cascade in U. maydis (N. Carbó, S.C.-L. and J.P.-M., unpublished results). This kind of interaction has been previously reported between PHO85 and genes encoding components of the cell wall integrity cascade in S. cerevisiae (Lenburg and O'Shea, 2001; Huang et al., 2002).

A speculative model is that Cdk5 stabilizes polar growth in U. maydis acting at least at two levels. On one hand, regulating the polar localization of Cdc24 it could allow the activation of Rac1 that, via Cla4, could help septin assembly. Septins may control the establishment of a specialized cortical domain facilitating the generation of positional information. However, our results also suggest that Cdk5 may be involved in the correct synthesis of the cell wall, although its role has not yet been established.

From our analysis, we found that cdk5^ts mutants showed drastically reduced virulence, probably because of its involvement in the induction of the dramatic polar growth required for the infection process. Pathogenic development of many fungi requires a yeast-hypha transition, which enables the pathogen to be passively spread or to actively invade substrates by polar hyphal growth (Gow et al., 2002). These morphological changes link polarized growth to pathogenesis, emphasizing the importance of growth dynamics regulation in dimorphic fungi. Based on this report we propose that Cdk5-like kinases may play a primordial role during the infective process in pathogenic fungi, and therefore may represent an `Achilles' heel' that can be exploited to limit fungal growth, as it has been proposed with other polarity regulators (Harris, 2006).

All strains had the genetic background FB1 (a1b1) or FB2 (a2b2) (Banuett and Herskowitz, 1989) and are summarized in supplementary material Table S1. U. maydis cells were grown in YEP-based or complete medium (CM) (Holliday, 1974). The expression of genes under the control of the crg1 and nar1 promoters, FACS analysis, western blot and cell cycle arrest were all performed as described previously (Brachmann et al., 2001; García-Muse et al., 2003; García-Muse et al., 2004; Garrido et al., 2004).

To generate the different strains, the constructs indicated were used to transform protoplasts as previously described (Tsukuda et al., 1988). The integration of the plasmids into the corresponding loci was verified by diagnostic PCR and subsequently by Southern blotting.

Deletion of cdk5 in the diploid strain FBD11 was performed by the construction of a deletion cassette consisting of a pair of DNA fragments obtained by PCR flanking the cdk5 open reading frame (ORF) and ligated to a nourseothricin (Nat) resistance cassette. The 5′ fragment spans nucleotide –218 to nucleotide +1 (considering the adenine in the ATG as nucleotide +1) and the 3′ fragment spans nucleotide +632 to nucleotide +1000. This cassette was introduced into the native cdk5 locus by homologous recombination.

To construct the cdk5-myc allele the cdk5 ORF without the stop codon was ligated to three copies of the myc epitope, marked with a Nat-resistance cassette, flanked at the 3′ end by 1 kb of the 3′ region downstream of the cdk5 open reading frame and introduced by homologous recombination into the cdk5 locus. The cdk5-GFP allele was constructed in a similar way, but instead of three copies of the myc epitope, the sequence encoding GFP was cloned at the 3′ end of the Cdk5 ORF.

The temperature-sensitive allele cdk5^ts was generated by PCR using a two-step mutagenesis protocol. The TAC codon (Tyr) at nucleotide position 856-858 was replaced by CAC (His). In addition, a silent point mutation (G to C) was introduced at position 870 to create a diagnostic SacII site. The mutated gene was marked with a Nat resistance at the 3′ end and flanked at the 3′ end of the Nat-resistance cassette by 1 kb of the 3′ region downstream of the cdk5 ORF and introduced by homologous recombination into the cdk5 locus of corresponding strains.

To overexpress rac1 under the control of crg1 promoter, the rac1 ORF plus 100 bp of the region downstream, were amplified by PCR from genomic DNA. The ATG codon start was exchanged with a BamHI site inserted in the same translational phase as a T7 epitope cloned under the control of crg1 in the pRU11T7 plasmid (S.C.-L., unpublished results). The T7-Rac1 protein was as functional as a wild-type untagged Rac1 protein. To construct the rac1^QL version we followed the two-step mutagenesis protocol as described (Mahlert et al., 2006) using T7-rac1 as a template.

To overexpress cdc24, a fragment carrying the cdc24 ORF as well as 160 bp of the 3′ downstream region was amplified by PCR, and cloned into the pRU11 plasmid (Brachmann et al., 2001) under the control of crg1 promoter. These plasmids were inserted by homologous recombination into the cbx locus as described in Brachmann et al. (Brachmann et al., 2001).

A bipartite cassette composed of a gfp ORF and an antibiotic resistance gene was used to construct the fim1-GFP and cdc24-GFP alleles. To this end, around 1000 bp fragment of the 3′ end of the corresponding ORFs without the stop codon were amplified by PCR and inserted in the same translational phase as gfp in the bipartite cassette. A second flanking region, corresponding to the 3′ region downstream of fim1 and cdc24 open reading frames was cloned in the 3′ end of the cassette and introduced by homologous recombination into the native loci of corresponding strains.

For microscopic analysis, cells were grown overnight at 22°C, 200 r.p.m., shifted to 34°C for 4-5 hours, and mounted on a 2% agar cushion for observation in a Zeiss Axioplan II imaging microscope (Oberkochen, Germany). Standard FITC fluorescent filters and a 100× Apochromat objective prewarmed to 34°C were used. For the analysis of GFP-Myo5 dynamics images series of 60 frames at 1-second intervals were taken. A CoolSNAP-HQ CCD camera (Photometrics, Tucson, AZ) controlled by the imaging software Metamorph (Universal Imaging, Downingtown, PA) was used.

To test for mating, compatible strains were co-spotted on charcoal-containing PD plates that were sealed with Parafilm and incubated at 22°C or 30°C for 48 hours (Holliday, 1974). Plant infections were performed with the maize cultivar Early Golden Bantam as described previously (Old Seeds, Madison, WI) (Gillissen et al., 1992). For the isolation of progeny and meiotic studies, spores from tumors were collected 10-14 days after inoculation and plated on plates containing 2% water agar. After 2 days, microcolonies (with about 200 cells) were removed and resuspended in YPD and dilutions plated onto YPD plates with or without nourseothricine as a selective agent.

The authors thank P. Sudbery for critical reading of the manuscript. The helpful suggestions of two anonymous reviewers are also thanked. This work was supported by Grants to J.P.M. from the Spanish government (BIO2005-02998) and EU (MRTN-CT-2005-019277). The authors declare no competing financial interests.