The regulation of cyclin-dependent kinase (CDK) activity through inhibitory phosphorylation seems to play an important role in the eukaryotic cell cycle. We have investigated the influence that inhibitory phosphorylation of the catalytic subunit of mitotic CDK has on cell growth and pathogenicity of the corn smut fungus Ustilago maydis. This model pathogen is worthy of attention since it is well suited to analyze the relationships between the cell cycle, morphogenesis and pathogenicity. We set out to study these relationships by producing a cdk1 mutant allele that was refractory to inhibitory phosphorylation. The expression of this mutant in U. maydis cells dramatically altered their morphology. Since this kind of mutation makes the CDK catalytic subunit resistant to regulation by Wee1-related kinases in other organisms, we characterized the orthologous Wee1 kinase from U. maydis. We found that Wee1 is essential in U. maydis. Overexpression of wee1 produces cell cycle arrest in G2, the target of Wee1 apparently being the Cdk1/Clb2 complex, which is required specifically for the onset of mitosis. Given the connection between the cell cycle control and pathogenesis in U. maydis, we also analyzed whether cells with impaired inhibitory phosphorylation of Cdk1 were able to infect plants. We found that inhibitory phosphorylation was required for mating, a prerequisite to initiate pathogenic development. By examining plant-specific expression of the constitutively unphosphorylated cdk1AF allele, we also found that appropriate levels of inhibitory phosphorylation were required at stages of infection subsequent to penetration by the fungus. These data reinforces the connections between cell cycle, morphogenesis and virulence in this smut fungus.
The correct development of all eukaryotic organisms requires nuclear division and cytokinesis to be strictly coordinated during cell growth. In fungi, the coordination of these two processes raises unique problems. The mechanisms employed must be compatible with simple uninucleate cells in which a single nuclear division is followed by cytokinesis, and more complex multinuclear filamentous fungi in which several rounds of nuclear division must precede septum formation (Harris, 1997; Lew, 2000; Cerutti and Simanis, 2000). A common feature of these processes is the regulation of the cell cycle machinery, including the modulation of the cyclin-dependent kinases (CDKs) via inhibitory phosphorylation of their catalytic subunit (Kellogg, 2003). In fission yeast, the Wee1 and Cdc25 proteins play crucial roles in coordinating cell growth and cell division at the G2/M transition. Wee1 is a protein kinase that phosphorylates and inhibits Cdc2. Phosphorylation of this CDK catalytic subunit on conserved tyrosine and threonine residues prevents entry into mitosis until cells reach a critical size (Russell and Nurse, 1987; Gould and Nurse, 1989; Featherstone and Russell, 1991; Lundgren et al., 1991; Parker et al., 1992). Conversely, Cdc25 is a phosphatase that removes the inhibitory phosphate added by Wee1, thereby promoting entry into mitosis (Russell and Nurse, 1986; Gautier et al., 1991; Millar et al., 1991). Loss of Wee1 activity causes cells to enter into mitosis before sufficient growth has occurred and therefore, cytokinesis produces two abnormally small daughter cells. Increasing the gene dosage of wee1 causes delayed entry into mitosis and an increase in cell size. These responses indicate that the levels of Wee1 activity determine the timing of entry into mitosis (Nurse, 1975; Russell and Nurse, 1987).
In Saccharomyces cerevisiae, the budding yeast homolog of Wee1, Swe1, has been linked to a checkpoint that delays cell cycle progression in response to morphogenetic defects (Booher et al., 1993; Sia et al., 1996). Swe1 also modulates pseudohyphal growth in a broad spectrum of conditions (La Valle and Wittenberg, 2001; Martínez-Anaya et al., 2003). A role for Swe1 in the control of cell size was suggested by the finding of a swe1 mutant allele in a screening for whi phenotype (Jorgersen et al., 2002) and more recently it was shown that Swe1 delays entry into mitosis and is required for the control of cell size (Harvey et al., 2003). The common role of the Wee1 and Swe1 kinases may also apply to the filamentous fungus Aspergillus nidulans, where the Wee1-like kinase AnkA controls septum formation (De Souza et al., 1999; Kraus and Harris, 2001). In A. nidulans cells, predivisional hyphae cannot septate until a specific cell size is attained, indicating that septum formation is coordinated with cell growth (Wolkow et al., 1996).
Here, we set out to investigate the influence of inhibitory phosphorylation of the catalytic subunit of mitotic CDK on cell growth and pathogenicity of the corn smut fungus Ustilago maydis. This pathogen is perfectly suited to analyze the relationships between cell cycle, morphogenesis and pathogenicity (Basse and Steinberg, 2004). Haploid cells of this fungus are unicellular and divide by budding. Induction of the pathogenic phase requires the mating of two compatible haploid cells and after cell fusion, the generation of an infective dikaryotic filament that invades the plant (Kahmann et al., 2000). The different morphological changes that the fungal cells undergo during the pathogenic process provide evidence of the tight control of the cell cycle in these transitions. Indeed, we recently showed that manipulation of the transcription of mitotic cyclins affects hyphal proliferation within the plant, resulting in fungal cells unable to produce a successful infection (García-Muse et al., 2004). We also reported that mutations in the Fizzy-related APC activator Cru1 disrupted different stages of plant infection by U. maydis (Castillo-Lluva et al., 2004). As a further step toward elucidating the role of cell cycle regulation in the virulence of this fungus, we have analyzed whether impairing the inhibitory phosphorylation of CDK affects the ability of U. maydis to infect plants. To achieve this, we took advantage of cells expressing a constitutively unphosphorylated cdk1 allele to show that deficient inhibitory phosphorylation affects morphogenesis, the coordination of cell growth with the cell cycle, the S phase checkpoint control and virulence. We also characterized the U. maydis Wee1 kinase ortholog and found that Wee1 is required to control the G2/M transition in U. maydis. The data reported here reinforce the proposed connection between the cell cycle, morphogenesis and pathogenicity in U. maydis.
Materials and Methods
Strains and growth conditions
The yeast strains used in this study are listed in Table 1. The U. maydis cells were grown in YPD (Sherman et al., 1986), complete medium (CM; 0.25% casaminoacids, 0.1% yeast extract, 0.15% NO3NH4, 1% Holliday vitamins solution, 6.25% Holliday salts solution) or minimal medium (MM; 0.3% KNO3, 6.25% Holliday salts solution) (Holliday, 1974). Controlled expression of genes under the crg1 and nar1 promoters, FACS analysis and cell cycle arrest were all carried out as previously described (Brachmann et al., 2001; García-Muse et al., 2003; García-Muse et al., 2004).
DNA, RNA and protein analysis
U. maydis DNA and RNA isolation, the preparation of protein extracts, northern and western blotting, and immunoprecipitations were all performed as described previously (Tsukuda et al., 1988; Garrido and Pérez-Martín, 2003; García-Muse et al., 2004; Garrido et al., 2004). The anti-PSTAIRE (Santa Cruz Biotechnology), anti-myc 9E10, and anti-VSV (Roche Diagnostics Gmb) antibodies were all used at a dilution of 1:10,000 in phosphate-buffered saline + 0.1% Tween + 10% dry milk. The anti-phospho-Cdc2 (Tyr15) antibody (Cell Signaling) was used according to the manufacturer's instructions. The anti-mouse Ig horseradish peroxidase and anti-rabbit Ig horseradish peroxidase (Roche Diagnostics Gmb) secondary antibodies were used at a dilution of 1:10,000. All western blots were visualized by enhanced chemiluminescence (Renaissance®; Perkin Elmer).
Plasmids and constructs
Plasmid pGEM-T easy (Promega) was used for cloning, subcloning and sequencing of genomic and PCR fragments. The plasmids pRU11, pRU2, pCU2 and pCU3 were used to express genes under the control of Pcrg1, Pnar1, Pscp and Ptef1 promoters, respectively, and the plasmids pBS-MYC-HYG and pGNB-myc to produce C-terminal myc-tagged protein fusions as already described (Brachmann et al., 2001; Brachmann, 2001; García-Muse et al., 2004; Garrido et al., 2004). The oligonucleotides used for PCR amplification are shown in Table 2 and the PCR fragments generated were analyzed using an automated sequencer (ABI 373A) and standard bioinformatic tools. To construct the different U. maydis strains, protoplasts were transformed with the constructs indicated using the protocol described by Tsukuda et al. (Tsukuda et al., 1988). Integration of the plasmids into the corresponding loci was verified by diagnostic PCR and Southern blot analysis in all cases.
The mutant cdk1AF allele was constructed by assembling two PCR fragments that carried the T14A and Y15F mutations, generated with the primer pairs CDK1/CDK2-b and CDK3-b/CDK4. The wild-type cdk1 allele was amplified using the CDK1 and CDK4 primers. In both cases, the resulting 1.1 kb fragment was inserted into pGNB-myc to obtain the myc-epitope-tagged protein. The tagged alleles were then inserted into pCU3 (Ptef1-dependent expression) or pRU2 (Pnar1-dependent expression) to construct the pCU3-Cdk1, pCU3-Cdk1AF, pRU2-Cdk1 and pRU2-Cdk1AF plasmids. These plasmids were linearized and integrated into the cbx1 locus by homologous recombination, as described by Brachmann et al. (Brachmann et al., 2001). To express the cdk1 and cdk1AF alleles under the control of the plant-specific promoter Pmig1, a 2 kb fragment carrying the Pmig1 promoter was inserted into pCU3-Cdk1 and pCU3-Cdk1AF to replace a 0.25 kb fragment carrying the Ptef1 promoter. This fragment was obtained by PCR amplification with primers MIG1-1 and MIG1-2, using U. maydis genomic DNA as the template. The resulting plasmids, pMIG1-Cdk1 and pMIG1-Cdk1AF were integrated by homologous recombination into the mig1 locus.
To produce a C-terminal tagged version of Wee1, a 4.5 kb fragment encompassing the wee1 ORF without the stop codon was amplified by PCR from U. maydis genomic DNA with the primers WEE1-32 and WEE1-31. This fragment was cloned into pGEM-T easy to produce the pGEMT-Wee1 plasmid. A 2.6 kb fragment from pGEMT-Wee1 was inserted into pBS-MYC-HYG to produce the endogenous myc-tagged wee1, and the resulting pBS-Wee1-myc plasmid was integrated into the wee1 locus by homologous recombination. To overexpress wee1, a C-terminal myc-epitope tag was first introduced by inserting the 4.5 kb pGEMT-Wee1 fragment into pGNB-myc. From the resulting pGNB-wee1-myc plasmid, a 4.7 kb fragment carrying the tagged allele was inserted into the pRU11 vector (Pcrg1-dependent expression). The resulting pRU11-Wee1-myc plasmid was linearized and integrated into the cbx1 locus by homologous recombination as described by Brachmann et al. (Brachmann et al., 2001).
To delete the wee1 gene, we used the pKOWee1 plasmid generated by ligating two DNA fragments flanking the wee1 ORF into pNEB-HYG (+), a U. maydis integration vector containing a hygromycin resistance cassette (Brachmann et al., 2001). The 5′ fragment was produced by PCR with the primers WEE1-7 and WEE1-8 using U. maydis genomic DNA as the template, and the 3′ fragment using the primers WEE1-9 and WEE1-10. After linearization, the pKOWee1 plasmid was integrated into the wee1 locus by homologous recombination.
To produce a conditional wee1nar allele we inserted two PCR fragments into pRU2. The 5′ fragment was produced using the primers WEE1-35 and WEE1-36 and spanned from nucleotide –872 to nucleotide –120 (considering the adenine in the ATG as nucleotide +1). The 3′ fragment was obtained with the primers WEE1-32 and WEE1-27 and spans from nucleotide +1 to nucleotide +1524. The resulting plasmid pWEE1nar was linearized and integrated into the wee1 locus by homologous recombination.
To produce the wee1scp allele, a 2.28 kb fragment from pWEE1nar was inserted into pCU2 (Pscp-dependent expression). The resulting plasmid pWEE1scp was linearized and integrated into the wee1 locus by homologous recombination.
Microscopy was carried out using a Leica DMLB microscope with phase contrast optics. Standard FITC and DAPI filter sets were used for epifluorescence analysis of nuclear staining with DAPI (see García-Muse et al., 2003) and WGA staining, performed as described by Castillo-Lluva et al. (Castillo-Lluva et al., 2004). Photomicrographs were obtained with a Leika 100 camera and the images were processed with Photoshop (Adobe).
Mating and plant infection
To test for mating, compatible strains were co-spotted on charcoal-containing PD plates that were sealed with parafilm and incubated at 21°C for 48 hours (Holliday, 1974). Plant infections were performed with the maize cultivar Early Golden Bantam as described previously (Old Seeds, Madison, WI, USA) (Gillissen et al., 1992). Filaments inside the plant tissue were stained with Chlorazole Black E as described by Brachmann et al. (Brachmann et al., 2003).
Expression of a cdk1 allele insensitive to inhibitory phosphorylation dramatically alters the morphogenesis of U. maydis cells
The catalytic subunit of the mitotic CDK from U. maydis, encoded by the cdk1 gene, contains the conserved amino terminal Thr14 and Tyr15 residues. These residues are the targets of inhibitory phosphorylation by Wee1-related kinases in other systems (Fleig and Gould, 1991; García-Muse et al., 2004). A specific antibody raised against the phosphorylated human Cdc2-Y15P peptide recognized one major band in total protein extracts of U. maydis. Moreover, this phosphorylated protein co-migrated with a protein recognized by the anti-PSTAIRE antibody (Fig. 1A). To determine whether this band corresponded to the Cdk1 subunit, we introduced a cdk1-1 allele that produces Cdk1 carrying a C-terminal copy of the FLAG epitope into fungal cells by homologous recombination (García-Muse et al., 2004). In protein extracts from these strains the phospho-specific antibody recognized a single band with a slightly lower electrophoretic mobility to wild-type Cdk. This retardation was due to the increase in size provoked by adding the FLAG tag, indicating that the protein recognized by the phospho-specific antibody in both wild-type and recombinant strains was Cdk1.
To address what role this phosphorylation might play in U. maydis cells, we constructed a mutant allele, cdk1AF, in which these putative inhibitory phosphorylation sites in Cdk1 were replaced with residues that could not be phosphorylated (Thr14 to Ala and Tyr15 to Phe, respectively). However, we were unable to replace the endogenous cdk1 locus with this cdk1AF allele (not shown) and therefore, we overexpressed this mutant allele and the wild-type allele under the control of the tef1 promoter that produces high levels of transcription (Spellig et al., 1996). We used 3xMyc-tagged versions of both the wild-type and the constitutively unphosphorylated Cdk1 to discriminate between the levels of protein produced by the endogenous locus and the ectopically expressed alleles (Fig. 1B). Overexpression of a wild-type allele had no discernible effect on cell growth or the morphology of the U. maydis cells (Fig. 1C, left column). Cells grew by budding and newly formed buds emerged at the pole of the elongated mother cell. Furthermore, when the cell wall of these cells were stained with FITC-labeled wheat germ agglutinin (WGA), a lectin that binds to oligomeric chitin, the bud scar as well as newly formed septa at the stage of cell separation were labeled. In contrast, typical budding pattern was lost in the cells expressing the constitutively unphosphorylated allele and the morphology of these cells was dramatically altered (Fig. 1C, right column). Large aggregates of these cells formed, composed of long chains of rounded cells that divided by septation but which remained attached to each other. Each cell compartment in these chains of cells contained a single nucleus, and WGA reacted strongly with septa and the lateral cell wall. These data indicate that the inability to phosphorylate Cdk1 at its amino terminus interferes with normal growth in U. maydis.
S phase checkpoint control in U. maydis could be mediated through inhibitory phosphorylation of Cdk1
Inhibitory phosphorylation of the catalytic subunit of CDK is part of the S phase checkpoint control in response to damage or incompletely replicated DNA in several organisms. For instance, S. pombe, A. nidulans and human cells are dependent of this regulation. In contrast, S. cerevisiae cells do not require inhibitory phosphorylation to arrest cell cycle after activation of a S phase checkpoint (Lew and Kornbluth, 1996). To address whether in U. maydis the inhibitory phosphorylation is involved in the response to S phase checkpoint, we firstly examined the levels of Tyr15 phosphorylation in response to an arrest in S phase caused by addition of 1 mg/ml hydroxyurea (García-Muse et al., 2004). We found that the level of tyrosine-phosphorylated Cdk1 increased upon addition of hydroxyurea (HU) (Fig. 2A), suggesting an involvement of this negative regulation in the S phase checkpoint as in other organisms. To reinforce this conclusion, we tested cells expressing high levels of the unphosphorylated cdk1AF allele for sensitivity to HU at sub-lethal concentrations. As control we used wild-type cells and cells expressing high levels of wild-type cdk1 alelle. Consistently, we found that cells with impaired inhibitory phosphorylation showed a high degree of HU sensitivity (Fig. 2B).
U. maydis contains a single Wee1-related protein
The dramatic phenotype observed by overexpression of the constitutively unphosphorylated Cdk1 protein indicates that the inhibition induced by Wee1-related kinases may be critical for the correct growth of U. maydis. Therefore, we searched the available U. maydis genomic sequence for potential homologues of Wee1-related kinases (http://www.broad.mit.edu/annotation/fungi/ustilago_maydis/index.html). We found a single gene, wee1, encoding a putative protein of 1525 amino acids (accession no. AY995171). Sequence analysis indicated that Wee1 contained a kinase domain near the carboxyl terminus which shared 43-28% sequence similarity with that in other fungal Wee1-related proteins (Fig. 3A). In addition, Wee1 contained the Glu-Gly-Asp triplet motif characteristic of kinase subdomain VIII in the Wee1 protein kinase family (Fig. 3C) (Booher et al., 1993). A dendrogram analysis (Fig. 3B) indicated that the Wee1-related proteins from fungi fall in two different branches, U. maydis Wee1 lies in a group with S. pombe Wee1 and A. nidulans AnkA.
We created a fungal strain in which the genomic copy of wee1 was replaced with a C-terminal myc-tagged Wee1 protein. Cell growth and morphology was not noticeably affected by this replacement (not shown). Using this strain, we investigated whether the levels of Wee1 fluctuated in different phases of the cell cycle. Because no reproducible synchronization method is so far available for U. maydis, we performed western blotting on protein extracts isolated from cultures of wild-type cells enriched in G1 phase, or cells arrested in S phase by the presence of hydroxyurea, or M phase with benomyl. We detected Wee1 under all conditions, but it appeared to undergo a mild down-regulation that coincided with M phase arrest (Fig. 3D), as described for Swe1 in S. cerevisiae (Harvey and Kellogg, 2003) and Wee1 in S. pombe (Aligue et al., 1997). We also analyzed the mRNA levels and found that wee1 mRNA is preferentially expressed at G1 phase with the lowest levels again being found in M phase-arrested cells (Fig. 3E), as described for SWE1 in S. cerevisiae (Rey et al., 1996).
High levels of wee1 expression causes cell cycle arrest at G2 phase
Overexpression of Wee1-related kinases delays entry into mitosis in several systems, resulting in G2 arrest as a consequence of accumulation of phosphorylated inactive CDK complexes (Russell and Nurse, 1987; Booher et al., 1993; McGowan and Russell, 1995). To determine whether overexpression of wee1 has similar inhibitory activity in U. maydis, an extra copy of wee1 was introduced into wild-type U. maydis cells. This additional wee1 allele was under the control of the crg1 promoter which can be induced by arabinose and repressed by glucose (Bottin et al., 1996). These cells were unable to grow in solid medium containing arabinose (Fig. 4A) indicating that the overexpression of wee1 was deleterious to the cells.
We evaluated this response in more detail, characterizing the morphology of the cells as well as their DNA content in liquid cultures. After 6 hours in arabinose-containing medium, virtually all wee1-overexpressing cells ceased dividing and remained as single, uninuclear cells with a bud-like protrusion that continued elongating over time (Fig. 4B). These cells contained long microtubules that reached the tip of the growing pole, thereby supporting active polar growth (Fig. 4C) a hallmark of G2 phase in U. maydis cells (Steinberg et al., 2001; Banuett and Herskowitz, 2002). Flow cytometry showed these cells to have a 2C DNA content (Fig. 4D), all of which was compatible with G2 arrest. To determine whether the Wee1-induced cell cycle arrest correlated with the increased tyrosine phosphorylation of Cdk1, we examined the phosphotyrosine content of Cdk1 during the growth of the wild-type and wee1-overexpressing cells (Fig. 4E). We consistently found an increase in the levels of Tyr15P-Cdk1 when wee1 overexpression was induced with respect to the wild-type cells.
In summary, high levels of Wee1 induce G2 arrest, correlated with a high level of Tyr15 phosphorylation of Cdk1 in U. maydis. On these grounds, we consider U. maydis Wee1 to be a Wee1-related protein kinase.
Wee1 is essential for the survival of U. maydis
To further analyze the function of Wee1, we inactivated one wee1 allele in the diploid FBD11 strain, replacing it with a hygromycin-resistance cassette, generating the Δwee1 null allele. When the meiotic progeny of this strain were analyzed after sporulation, no hygromycin-resistant cells were found, indicating that wee1 is an essential gene. Therefore, we constructed another strain in which the Wee1 protein could be conditionally ablated. This was achieved using the U. maydis nar1 promoter that is induced by growing the cells in nitrate as the nitrogen source, and strongly repressed when the cells are grown in rich medium (YPD) (Brachmann et al., 2001). A chimeric allele (wee1nar) was constructed by fusing the nar1 promoter to the coding region of wee1 and the native allele was replaced by this conditional allele. In the conditional mutant, a three-fold increase was observed in wee1 mRNA with respect to the levels in wild-type cells in permissive (minimal medium containing nitrate, MMNO3) but not in restrictive conditions (YPD medium; Fig. 5A). Conditional mutant cells grew on solid medium containing nitrate at a rate similar to control wild-type cells. However, in accordance with the essential role of the Wee1, they were unable to form colonies when shifted to solid YPD medium (Fig. 5B).
We analyzed the growth of this conditional strain in liquid medium. In restrictive conditions the conditional cells generated chains of rounded cells that divided by septation (Fig. 6). The cell compartments were shorter than normal (6 μm on average versus 17 μm in wild-type U. maydis cells), and contained at least one nucleus. Nevertheless, when the DNA content was analyzed by FACS we found that in restrictive conditions, wee1nar cells accumulated DNA in genome sized multiples (Fig. 4D). Cell analysis using microdensitometry (Snetselaar and McCann, 1997) indicated that the relative intensity of nuclei fluctuate between two peaks, presumably corresponding to 1C and 2C DNA content (not shown), which is consistent with normal DNA replication. Accordingly, these cell aggregates were reminiscent of those seen in cells expressing the constitutively unphosphorylated cdk1AF allele. Our interpretation is that impairing the inhibitory phosphorylation of Cdk1 promoted premature entry into mitosis, resulting in the inability to produce a bud and followed by division through septation.
Strikingly, when grown in permissive conditions the conditional strain did not display a wild-type phenotype. Mother and bud cells were more elongated than the wild-type cells (Fig. 5E), and no clear G1 phase was observed (Fig. 5D). We believe that the unscheduled expression of the wee1nar allele augmented Wee1 activity to above normal levels (compare the amount of wee1 mRNA in wild-type and wee1nar cells growing in permissive conditions, Fig. 5A). This increase in Wee1 activity could have delayed entry into mitosis resulting in an increase in cell size at division and as a consequence, a shorter G1. An additional phenotype found in these conditional cells was that about 30% of the budding cells had lateral buds. Wild-type cells bud in a polar manner whereby the daughter cell emerges from one end of the mother cell. During exponential growth, less than 1% of wild-type cells had lateral buds. Thus, although preliminary, this result suggests that correct Wee1 activity could be required for bud site selection in U. maydis.
To correlate Wee1 function with inhibitory phosphorylation in Cdk1, wild-type and conditional cells were grown in permissive conditions until the mid-exponential phase and then the cultures were shifted to restrictive conditions. Samples were collected at different times and the degree of Tyr15 phosphorylation was analyzed by Western blotting with the anti-Cdc2-Y15P antibody. A dramatic reduction in the levels of Tyr15 phosphorylation was clearly evident in extracts obtained from conditional cells growing in restrictive conditions (Fig. 5C).
Taken together these results support the notion that Wee1-mediated inhibitory phosphorylation of Cdk1 controls entry into mitosis and hence the length of the G2 phase in U. maydis.
Wee1 controls the Cdk1-Clb2 complex
We previously demonstrated the specificity of the B-type cyclin Clb2 to the G2/M transition where it appears to be rate-limiting for entry into mitosis in U. maydis (García-Muse et al., 2004). Cell growth was arrested in G2 phase when Clb2 function was impaired, these cells developing an elongated bud that displayed active polarized growth, a phenotype resembling that generated by wee1 overexpression. In contrast, high levels of clb2 expression resulted in short cells that divide by septation, a phenotype reminiscent of cells that do not express wee1 or in which a constitutively unphosphorylated cdk1AF allele was overexpressed. Together, these observations suggest that the Cdk1-Clb2 complex could be a target of Wee1.
To evaluate whether Wee1 antagonizes the Cdk1-Clb2 complex, we generated a double conditional mutant wee1nar clb2nar. When we compared this mutant with the respective single wee1nar and clb2nar conditional mutants in restrictive conditions, the concurrent down-regulation of wee1 and clb2 transcription produced a phenotype identical to that caused by the down-regulation of clb2 expression alone (Fig. 7A). This epistastic relationship located clb2 genetically downstream of wee1. We then analyzed whether high levels of Wee1 could overcome the effects of high levels of Clb2 on cell morphology. Cells with high constitutive levels of clb2 expression (replacing the native clb2 promoter with the strong constitutive hsp70 promoter) and carrying an ectopic arabinose-inducible wee1 allele expressed the phenotype of cells overexpressing clb2 when wee1 expression was not induced (YPD, Fig. 7B). However, when these cells were transferred to arabinose-containing medium, the cells elongated, resembling wild-type cells that overexpress wee1 (Fig. 7B, YPA). This result provides further evidence of antagonism between clb2 and wee1.
Finally, we directly tested whether the Cdk1/Clb2 complex is a target of inhibitory phosphorylation in U. maydis. Using strains that carry epitope-tagged U. maydis cyclins [the B-type cyclins Clb1 and Clb2 (García-Muse et al., 2004) and the G1 cyclin, Cln1 (S. Castillo-Lluva and J.P.-M., unpublished)], these cyclins were immunoprecipitated and the levels of Tyr15P-Cdk1 with which they were associated were analyzed. We found that Cdk1 was not phosphorylated on the inhibitory residues when complexed with the G1 cyclin, Cln1. However, when Cdk1 was associated with either Clb1 or Clb2 it was recognized by the phospho-specific antibody (Fig. 7C). Thus, an important determinant for Wee1 substrate recognition appears to be the presence of the Clb1 and Clb2 subunits, but not the Cln1 subunit.
Transcriptional regulation of wee1 is important to adapt cell size to different nutritional conditions
Fungal cells adjust their cell cycle depending on environmental conditions. For instance, in response to poor nutritional conditions, this adjustment results in an enlargement of the generation time avoiding the generation of abnormally small daughter cells (Rupes, 2002). In U. maydis, this enlargement takes place by increasing the length of the G1 phase until the cell reaches a minimum size to enter a new S phase (Snetselaar and McCann, 1997). We have previously shown that the APC adaptor Cru1 is important in U. maydis to determine the length of G1 and cell size in response to nutritional conditions. Indeed, the mRNA levels of cru1 are regulated by the quality of the growth medium (Castillo-Lluva et al., 2004). In this sense, cell size control in U. maydis seems to be more similar to S. cerevisiae, in which it takes place mainly in G1, than to fission yeast, in which G2/M transition is the primary cell size control point (Rupes, 2002). However, the above results showed that the levels of Wee1 seem to be important in determining the length of the G2 phase and hence cell size. Therefore, to examine whether Wee1 has a role in coordinating cell growth, cell division and nutritional conditions in U. maydis, we first addressed whether the levels of wee1 mRNA were affected by the growth medium. Total RNA was extracted from wild-type cells cultured in YPD, CMD and MMD, and the levels of wee1 mRNA analyzed. It was readily apparent that wee1 mRNA levels increased as the quality of the medium decreased (Fig. 8A).
How important is wee1 transcriptional regulation for cells to adapt to different nutritional conditions? To address this question we exchanged the native wee1 promoter with the scp promoter, a weak U. maydis constitutive promoter that produces low levels of expression (Bölker et al., 1995a). The abundance of the mRNA produced by this promoter did not vary as a function of the nutritional conditions, and was similar to the wee1 levels produced in wild-type cells growing in rich medium (YPD, Fig. 8A). Moreover, the cells carrying the constitutive wee1scp allele were smaller than wild-type cells (Fig. 8B), and this difference was more obvious in less nutritional medium. This result suggests that up-regulation of Wee1 levels play some role in the adjustment of cell cycle in response to nutritional conditions, particularly in less favorable medium. Analysis of the FACS profile of wild-type and mutant wee1scp cells grown in the different media (Fig. 8C) showed an increase in the number of cells with 1C DNA content (i.e. cells in G1 phase). This bias towards cells in G1, suggests that the G1 period must be lengthened in these small cells until the minimum size required for overriding the Start-restraint is attained One possible interpretation of these results is that U. maydis uses two different cell size controls at the same time in response to nutritional conditions: one in G1 phase (related with the APC-Cru1 complex) and other in G2/M (related with the Wee1 levels), and that the absence of one of these controls must be compensated by the other. To reinforce this interpretation, we tried unsuccessfully to delete the cru1 gene in cells carrying the wee1scp allele. To determine if the deletion of the cru1 gene in wee1scp cells was lethal, we crossed haploid Δcru1 (UMP9, marked with a hygromycin-resistance gene) cells with compatible haploid wee1scp cells (UMC44, marked with a carboxine resistance gene) and afterwards we analyzed the meiotic progeny. From 100 analyzed resulting haploid cells, none was found to carry the two mutant alleles, indicating a synthetic lethality.
Inhibitory phosphorylation could be required for polar growth during the infection process
Several results suggested an involvement of the inhibitory phosphorylation of Cdk1 in the maintenance of polarized growth in U. maydis. The conditional depletion of Wee1 or the expression of the cdk1AF allele generated cell aggregates composed of cells that often lost polarity and became almost spherical. In contrast, the overexpression of wee1 generated a highly polarized growth. The initiation of pathogenic development in U. maydis involves the activation of a strong polarized growth, as yeast-like budding cells switch, after mating, to tip-growing hyphae, so-called infective tubes (Kahmann et al., 2000). Therefore, we were interested in determining whether cells deficient in inhibitory Cdk1 phosphorylation were able to produce the infective tube. Since wee1 is essential in U. maydis, we took advantage of the cells expressing the cdk1AF allele to mimic defective inhibitory phosphorylation. The induction of infective tube formation can be easily scored by co-spotting compatible strains (i.e. with different a and b loci) on solid medium containing charcoal. On these plates, cell fusion and the development of an infective dikaryotic filament results in the formation of a white layer of aerial hyphae on the surface of the growing colony [Fuz+ phenotype (Holliday, 1974)]. We generated compatible strains that expressed an ectopic copy of the cdk1AF allele under the control of the nar1 promoter. Mixtures of compatible wild-type and mutant strains were spotted on charcoal-containing minimal medium with ammonium (non-inducing conditions) or nitrate (inducing conditions) as a nitrogen source. Mixtures of wild-type cells produced a clear Fuz+ phenotype in both conditions. In contrast, when compatible strains carrying the cdk1AF allele were spotted or mixed with wild-type compatible strains, the Fuz+ phenotype did not develop (Fig. 9A). Furthermore, microscopic observation of the mutant mixtures growing on nitrate-containing charcoal plates confirmed the absence of dikaryotic hyphae (not shown). These results indicated that inhibitory phosphorylation is required for the formation of the dikaryotic infective filament. However, dikaryotic hyphal formation requires the fusion of compatible cells, and we therefore considered it possible that this phenotype is a secondary consequence of a defect in mating. To circumvent the need for cell fusion we generated the respective mutations in the solo-pathogenic strain SG200 (Bölker et al., 1995b). This strain, which carries the information of the two mating types, is able to form infective tubes without a mating partner. Consistently with the charcoal assay, SG200 cells expressing the cdk1AF allele were unable to produce the infective filament (Fig. 9B).
Once it has entered the plant, U. maydis proliferates in a hyphal form. Therefore, we wished to address whether the impaired inhibitory phosphorylation of Cdk1 also affected the ability to grow inside the plant. Since plants inoculated with a mixture of cdk1AF compatible strains were asymptomatic and indistinguishable from uninoculated plants (Table 3), most likely because of the inability of mutant fungal cells to invade the plant, we constructed an a1 b1 strain carrying an ectopic copy of the cdk1AF allele under the control of the mig1 promoter (as well as a control strain with the wild-type cdk1 allele). This promoter directs expression only when the fungus is growing inside the plant (Basse et al., 2000). These strains were crossed with wild-type a2 b2 compatible cells. Strikingly, the plants inoculated with the mixture carrying the ectopic constitutively unphosphorylated cdk1AF allele showed evidence of chlorosis emanating from the point of inoculation, but fungal infection was unable to induce either anthocyanin stains or tumor formation (Fig. 10A). In contrast, the plants inoculated with mixtures carrying an ectopic wild-type cdk1 allele under mig1 promoter control displayed all the usual symptoms of infection (Banuett and Herskowitz, 1996). To analyze the fungus growing inside the plant, symptomatic leaves were stained and examined microscopically. In contrast to control infections that showed the typical hyphal growth inside the plant, the fungal cells expressing the cdk1AF allele were present in clumps composed of long chains of rounded cells (Fig. 10B).
Taken together, these results strongly suggest that inhibitory phosphorylation of Cdk1 plays an important role at different points of the infection process and that this role could be related with the ability to support a sustained polarized growth.
The main objective of this study was to examine the inhibitory effect that phosphorylation of Tyr15 of Cdk1 has on U. maydis cells. To pursue this goal, we produced a cdk1 mutant allele refractory to inhibitory phosphorylation, which promoted aberrant morphogenesis when expressed in U. maydis cells. Since this kind of mutation makes the CDK catalytic subunit resistant to negative regulation by Wee1-related kinases in other organisms, we isolated the Wee1-related kinase gene for U. maydis. This gene encodes a polypeptide containing features common to all protein kinases, as well as certain motifs specific to the Wee1 kinase family. As expected, the U. maydis wee1 gene functions as a dose-dependent inhibitor of mitosis. When wee1 was overexpressed in U. maydis, uninuclear cells with a 2C DNA content underwent arrest apparently during G2 phase. The activity of Wee1 as a mitotic inhibitor seems to be related to its ability to phosphorylate Cdk1. There is a clear correlation between the degree of Cdk1 Tyr15 phosphorylation and Wee1 activity. In addition, the G2 cell cycle arrest induced by Wee1 overexpression is largely suppressed in strains expressing a copy of the cdk1AF allele (J.P.-M., unpublished).
The fact that wee1 overexpression causes arrest of U. maydis cells in G2 suggests that the Cdk1-cyclin complexes targeted by Wee1 are required for the G2/M transition. In U. maydis, the G2/M transition requires the participation of both the Cdk1-Clb1 and the Cdk1-Clb2 complexes (García-Muse et al., 2004). Accordingly, we have shown that the phosphorylated Cdk1 is associated to these B-type cyclins. The Tyr15 phosphorylated Cdk1 cannot be found in complexes with Cln1, indicating that as in other organisms, different cyclins target CDKs for distinct negative regulatory controls in U. maydis (Devault et al., 1992; Booher et al., 1993; Watanabe et al., 1995). We believe that the primary target for Wee1 is the Cdk1-Clb2 complex. While Cdk1-Clb1 is required for the G1/S and G2/M transitions, Cdk1-Clb2 appears to be specific for G2/M transition and it seems to be a rate-limiting for entry into mitosis (García-Muse et al., 2004). Indeed, clb2 overexpression produces a phenotype that resembles those produced by overexpression of the cdk1AF allele or by downregulation of wee1. In contrast, overexpression of clb1 produces lethal chromosome missegregation (García-Muse et al., 2004). Furthermore, we found that high levels of Wee1 overcome the effects on cellular morphology imposed by high levels of Clb2. In contrast, the toxic effect of high levels of Clb1 (García-Muse et al., 2004) cannot be suppressed by an abundance of Wee1 (J.P.-M., unpublished). However, since the Cdk1-Clb1 complex is also subjected to Tyr15 phosphorylation (albeit apparently less than the Cdk1-Clb2 complexes), we believe a fraction of the Cdk1-Clb1 complexes involved in the G2/M transition, are also targeted by the Wee1 kinase. As such, we assume that elements specific to the G1/S transition modify the capacity of the Wee1 kinase to phosphorylate the Cdk1-Clb1 complex.
Inhibitory phosphorylation seems to be essential for cell cycle regulation in U. maydis. This is similar to S. pombe in which two related kinases (Wee1 and Mik1) phosphorylate Cdc2, the deletion of either gene being lethal (Lundgren et al., 1991). In contrast, the gene encoding the Wee1-like kinase in S. cerevisiae (SWE1) is dispensable for growth (Booher et al., 1993). The requirement in U. maydis for wee1 reflects the importance of controlling G2 phase length in this organism. Once DNA replication has occurred, U. maydis cells must decide whether to bud or to enter into a mating program, a decision that is taken in response to external stimuli (García-Muse et al., 2003). Controlling the length of G2 seems to be primordial for U. maydis to make the correct decision. In addition, our data are in accordance with an interesting model proposed by Kellog (Kellog, 2003), which suggests that Wee1-related kinases monitor the total amount of polar growth that occurs. Fission yeast and U. maydis are rod-shaped cells and all growth occurs at the ends of the cell (polar growth). In contrast, S. cerevisiae cells undergo a brief period of polar growth during bud emergence, but then growth occurs over the entire surface of the bud (isotropic growth). In agreement with Kellog (Kellog, 2003), the loss of wee1 function causes a more severe phenotype in U. maydis than in S. cerevisiae, despite the fact that both organisms divide by budding. We believe that this may be because U. maydis relies almost entirely on polar growth that occurs during G2, whereas budding yeast only undergo a brief period of polar growth and then switch to isotropic growth.
Also, again in contrast with S. cerevisiae, two observations reported here also suggest that inhibitory phosphorylation of Cdk1 could be involved in the modulation of a S phase checkpoint response. We found that the amount of Tyr15 phosphorylation increases after treatment with HU, and that cells expressing the cdk1AF allele were hypersensitive to sub-lethal concentrations of HU (i.e. the S phase checkpoint defect is presumably the cause of the HU hypersensitivity).
Progression through the cell cycle is regulated principally before the onset of S phase and of mitosis. In both cases, a critical cell mass must be attained before progression occurs (Nurse, 1975; Fantes, 1977). Recently, we described that in U. maydis, the absence of the APC adaptor Cru1 resulted in the inability to adjust G1 phase and hence in smaller than normal cells (Castillo-Lluva et al., 2004). Here, our results show that there is a close relationship between wee1 expression and cell size. We generated two different wee1 alleles, wee1nar and wee1scp, that are transcribed more or less than the wild-type allele. Altering the expression of wee1 produced cells that were larger and smaller than normal. Wild-type cells can adjust their size to the exterior environment, such as the availability of nutrients (Fantes and Nurse, 1978). Thus, by regulating Wee1 function, growth signals could conceivably produce such an effect. We found that wee1 mRNA levels increased as the quality of the medium decreased. In poor medium cells spend longer in G1 phase than in rich medium (Castillo-Lluva et al., 2004) and wee1 seems to be expressed preferentially in G1 phase. Hence, the differences in mRNA levels imposed by the growth medium could be easily explained without having to invoke mechanisms by which wee1 transcription is directly controlled by the nutritional conditions. Nevertheless, this latter possibility cannot be discarded.
Although minimal medium has been shown to extend the G1 phase of the U. maydis cell cycle (Castillo-Lluva et al., 2004), we propose that poorer medium may also delay progression to G2, and that this delay is dependent on Tyr15 phosphorylation. Such coordination in the length of the G1 and G2 phases is not only useful in conditions of limited nutrition but also to compensate for defects in each phase. For instance, G1 phase is not appreciable in cells carrying the wee1nar allele, which leads to a delay in mitosis and oversized buds (compare the FACS profile of wee1nar cells with wild-type cells growing in MM-NO3, Fig. 4B). In contrast, in wee1scp cells, which have low levels of wee1 mRNA and are small, there is a clear increase in the time spent by the cell in G1 phase. Furthermore, Cru1 is indispensable for the growth of wee1scp cells, probably because these double mutants cannot lengthen the G1 period to restrain S phase until the critical size to override the Start control is attained.
In addition to the influence on cell size, the cdk1AF allele or the down-regulation of wee1 had a marked effect on septation and cell pattern formation during vegetative growth. This indicates that septation in axenic cultures of U. maydis is normally prevented through a mechanism involving tyrosine phosphorylation of Cdk1. Indeed, septation in A. nidulans is regulated via tyrosine phosphorylation of nimXcdc2 (Harris and Kraus, 1998; Kraus and Harris, 2001). Another possibility is that the failure to inhibit Cdk1 through phosphorylation impairs the coordination between cell cycle regulation and the transition to the budding mode of growth. In S. cerevisiae a checkpoint system monitors the emergence of the bud. This morphogenetic checkpoint coordinates bud emergence with the cell cycle through the activity of the Swe1 protein kinase and tyrosine phosphorylation of Cdc28 (Lew and Reed, 1995; Sia et al., 1996). This system is linked to the organization of the cytoskeleton and septin function through a protein kinase cascade that eventually inhibits Cdc28 by tyrosine phosphorylation (Carroll et al., 1998; Longtine et al., 1998; Barral et al., 1999). If such a system were operative in U. maydis, it may become critical in axenic cultures where U. maydis cells grow by budding. The failure to inhibit Cdk1 by phosphorylation may prevent the normal synchronization and eventually contribute to the developmental defects we have observed.
One of the aims of this work was to define the interaction between the inhibitory phosphorylation of the catalytic subunit of mitotic CDKs and the program of virulence in U. maydis. We found that the expression of a constitutively unphosphorylated Cdk1 abolishes the ability of U. maydis to produce infective filaments. This inability could be explained assuming a role of Wee1 protein in the regulation of polarized growth. Alternatively, since the infective filament is cell cycle arrested in G2 phase (C.S. and J.P.-M., unpublished), it is tempting to speculate that G2 cell cycle arrest is mediated by the inhibition of Cdk1 by phosphorylation, and that the failure to induce arrest of the cell cycle impairs the formation of the infective tube. This interesting possibility will be clarified in future studies. By restricting the expression of the constitutively unphosphorylated cdk1 allele to fungi growing inside the plant, we bypassed the requirement of inhibitory phosphorylation for plant penetration. Hence, we were able to show that Tyr15 phosphorylation is also required for infection to progress inside the plant, since the symptoms of mutant cell infection do not progress beyond chlorosis. Interestingly, fungal cells expressing high levels of clb2 (García-Muse et al., 2004) showed a similar infection defect. Taken together, these data indicate that the accurate control of G2/M transition seems to be important for successful infection by U. maydis. It is worth bearing in mind that U. maydis cells grow in yeast-like unicellular form in saprophytic conditions. The induction of the pathogenic phase requires two compatible haploid cells to fuse and the generation of an infective dikaryotic filament that invades and proliferates inside the plant (Kahmann et al., 2000). Studies of pseudohyphal development in S. cerevisiae have revealed the importance of controlling the G2/M transition for the production of the filamentous growth (for a review, see Rua et al., 2001).
In summary, the results presented here reinforce the connections between the cell cycle and the induction of the pathogenic program in U. maydis. Moreover, these data highlight the potential significance of cell cycle regulation in microbial pathogenesis.
We thank Prof. Correa-Bordes for advice regarding the phosphorylation western blots. This work was supported by Grant BIO2002-03503 from MCyT and Grant 07B/0040/2002 from CAM.
- Accepted May 13, 2005.
- © The Company of Biologists Limited 2005