A member of the Fizzy-related family of APC activators is regulated by cAMP and is required at different stages of plant infection by Ustilago maydis

In the fungus Ustilago maydis, pathogenicity and sexual development are intimately intertwined. Haploid cells grow in yeast-like unicellular form, dividing by budding, and induction of the pathogenic phase requires the mating of two compatible haploid cells and the generation, after cell fusion, of an infective dikaryotic filament that invades the plant (Kahmann et al., 2000). Inside the vegetal tissue, the dikaryotic filament proliferates and induces formation of plant tumors, which consist of an increased number of enlarged host cells. Lately, the hyphae undergo fragmentation to release individual cells that will produce the diploid spore (teliospore). Eventual germination of the teliospores is associated to meiosis that will produce four haploid cells. All these different steps are though to be the result of a complex fungal-plant signaling (Kahmann et al., 1999). At present the nature of these signals is unclear, but data from several laboratories showed that a regulated cAMP cascade is crucial for the development of U. maydis inside the plant. In particular, mutations in adr1, encoding the catalytic subunit of protein kinase A (PKA), or in gpa3, encoding a Gα subunit that feeds the cAMP pathway, generated U. maydis cells unable to cause any pathogenicity symptoms, whereas mutations in ubc1, encoding the regulatory subunit of PKA, produce fungal cells that grow within the plant but are unable to induce tumors (Gold et al., 1994; Gold et al., 1997; Regenfelder et al., 1997; Dürrenberger et al., 1998). Furthermore, strains in which cAMP signaling is frozen at different levels are disturbed at distinct stages of development inside the plant (Krüger et al., 2000).

The different morphological changes that the fungal cells undergo during the pathogenic process advocates for an accurate control of the cell cycle in these transitions. Mating, which is the first step in the infection process, is linked to cell cycle, and cells of U. maydis, when exposed to pheromone, undergo a cell cycle arrest in G2 phase before the cells are fused and to form the infective filament (García-Muse et al., 2003). Once the filament enters the plant, the cell cycle arrest is released, probably in response to some unknown plant signal, and cells proliferate to form filaments in which septa partition cell compartments, containing a pair of nuclei each. This morphological transition requires a clear connection with the cell cycle control, and, for example, we had found that manipulation of the transcriptional levels of mitotic cyclins affects the hyphal proliferation inside the plant as well as the cell shape in axenic conditions (García-Muse et al., 2004). Finally, inside the tumor, the formation of the teliospores implies first karyogamy and then hypha fragmentation. How this process is controlled is currently unknown, but it is predicted that cell cycle adjustment must be required (Kahmann et al., 2000).

To date, little is known about the mechanisms responsible for the integration of cell cycle machinery into the distinct differentiation stages of the pathogenic program in U. maydis. In a similar way, potential links between the cAMP/PKA signal transduction pathway and the cell cycle machinery remain poorly defined. In eukaryotes, the control of cyclin levels involves, among others, stage-specific protein degradation mediated by the anaphase promoting complex (APC) (reviewed by Harper et al., 2002). Strikingly, there are several reports in the literature suggesting that, in other organisms, the components of the Fizzy-related family of APC activators are required for cell differentiation. For example, Schizosaccharomyces pombe cells lacking Ste9 are unable to undergo cell differentiation (Yamaguchi et al., 1997; Kitamura et al., 1998). In Saccharomyces cerevisiae cells, Cdh1/Hct1 is important to mating response and hct1 mutants are resistant to mating pheromone (Schwab et al., 1997). In the plant Medicago sativa, the Fizzy-related protein Ccs52 is involved in the differentiation of the nodule primordium (Cebolla et al., 1999; Vinardell et al., 2003). Drosophila fizzy-related is required at specific stages of embryogenesis when cells stop proliferating (Sigrist and Lehner, 1997). In summary, all these findings suggest that the APC is required to allow cell differentiation and that defects in this complex impair the ability of the cells to undergo differentiation.

Because of the previously described connections between cell cycle and the induction of the pathogenesis program in U. maydis, we sought to analyze the relationships between the cyclin destruction machinery and the ability to infect plants by U. maydis. For this, we have isolated and characterized a new member of Fizzy-related family of APC activators, which we called Cru1. We found that cells lacking the Cru1 regulator are severely impaired in the ability to infect corn plants. We also studied the connections between cAMP/PKA pathway and cru1 expression, and our results show that the cell cycle engine is a critical target of the cAMP/PKA pathway. To our knowledge this is the first report showing that an APC activator is regulated by cAMP and plays a role in the virulence of a fungus.

For cloning purposes, the Escherichia coli K-12 derivative DH5α (Bethesda Research Laboratories) was used. The U. maydis strains used in this study are listed in Table 1. Strains were grown at 28°C in YPD (Kaiser et al., 1994), YEPS [modified after Tsukuda et al. (Tsukuda et al., 1988)], potato dextrose broth (PD, Difco), complete medium (CM) or minimal medium (MM) (Holliday, 1974). For induction of crg1 promoter, strains to be tested were grown in CM medium with glucose as a carbon source (CMD) until OD₆₀₀ of 0.2, pelleted by centrifugation, washed twice with water and resuspended in CM with 2% arabinose as a carbon source (CMA). Hygromycin B was purchased from Roche, nourseothricin from the Hans-Knöll Institute (Jena, Germany) and carboxin from Riedel de Haen. All chemicals used were of analytical grade and were obtained from Sigma or Merck.

Two sets of degenerate oligonucleotides were synthesized according to the nucleotide sequences that encode two conserved regions in members of Fizzy-related family from different fungi: QEVCGLE/KW (MOTIF-1: 5′CARGARGTNTGYGGNYTNRARTGG3′) and DETLRFWK (MOTIF-2: 5′YTTCCARAANCKNTRNGTYTCRTC3′). The MOTIF-1 and MOTIF-2 oligonucleotides were used for amplification with 50 ng FBD11 DNA as a template in a volume of 50 μl. PCR products were generated in the following reaction mixture: 10 mM Tris/HCl pH 8.0, 50 mM KCl, 1.2 mM MgCl₂, 100 μM dNTP, 50 μM of each primer and 2.5 units Taq polymerase. Conditions for PCR cycling included denaturation at 94°C for 1 minute, annealing at 45°C for 1 minute and extension at 72°C for 2 minutes. Selected fragments (around 0.5 kb in size) were isolated and cloned into pGEM-T Easy (Promega). Positive clones containing inserts were chosen and the nucleotide sequence of each plasmid insert was determined in both directions by using the ABI model 373A Auto Sequence system (Perkin Elmer/Applied Byosystems). Sequences flanking these fragments were obtained with a PCR-walking strategy (Siebert et al., 1995) using the Genome Walker system (Clontech) as directed by the manufacturer.

Plasmids pGEM-T easy (Promega), and pBS-SK(-) (Stratagene) were used for cloning, subcloning and sequencing of genomic fragments and fragments generated by PCR. Sequence analysis of fragments generated by PCR was performed with an automated sequencer (ABI 373A) and standard bioinformatic tools.

Plasmid pNEBHyg(+) was used as source of the hygromycin resistance cassette (Brachmann et al., 2001). Plasmid pRU11 contains the crg1 promoter as a 3.5-kb NotI-NdeI fragment (Brachmann et al., 2001). Plasmid pTUBGFPHyg carries a GFP-tubulin a fusion and was constructed after ligation of a 4.6-kb fragment from potefGFPTub1 (Steinberg et al., 2000) carrying the GFP fusion into the pSMUT plasmid, a U. maydis integration vector containing a hygromycin B resistance cassette (Bölker et al., 1995). pCLB1VSVCbx, carrying a carboxin-resistant epitope-tagged version of Clb1 was produced after removal of a 4.4-kb EagI fragment from pRU11-CLB1 (García-Muse et al., 2004).

Plasmid pRU11cru1 is a pRU11 derivative that carries the cru1 open reading frame under the control of crg1 promoter. A 1.8-kb fragment generated by PCR with primers HCT1 (5′GACTAGTCATATGACGAGCCCCCCCATACCAATA3′) and HCT2 (5′GGGAATTCTCGCAGCTTGGCGAATGGGTTAAA3′) was inserted after NdeI-EcoRI digestion into the corresponding sites of pRU11. Digestion of pRU11cru1 with XcmI before transformation of U. maydis directs the integration into the ip locus.

Plasmid pCru1KO, carrying the disruption allele cru1Δ, was constructed by ligation of a pair of DNA fragments flanking the cru1 ORF into pSMUT. The 5′ fragment spans from nucleotide -494 to nucleotide -9 (considering the adenine in the ATG as nucleotide +1) and it was produced by PCR amplification using the primers FR2 (5′CCACAACACAAGGGTACCATATCTCCA3′) and FR3 (5′CCGCTCGAGAGCCTGTTAGGAGTCTGTCAGTT3′). The 3′ fragment spans from nucleotide +1811 to nucleotide +2204 and it was produced by PCR amplification using the primers FR4 (5′CGGGATCCATCTGTATCATGTATAATCTG3′) and FR5 (5′GGGGTACCGGAATGATGTTGCTTGCGGCAGT3′). The pCru1KO was linearized with KpnI before transformation of U. maydis, and transformants were screened for loss of the wild-type copy of cru1 by PCR analysis and confirmed by Southern analysis.

Standard molecular techniques were used. U. maydis DNA isolation and transformation was performed as previously described (Tsukuda et al., 1988). RNA isolation and northern analysis were performed as described previously (Garrido and Pérez-Martín, 2003). For mfa1 probe a 0.67-kb EcoRV fragment was used as described previously (Bölker et al., 1992). For cru1 probe, a 0.66-kb SalI fragment was used.

Extraction of U. maydis protein, and western analysis were performed as described previously (García-Muse et al., 2004). Anti-PSTAIRE (Santa Cruz Biotechnology), anti-myc 9E10 (Roche Diagnostics Gmb) and anti-VSV-G (Roche Diagnostics GmbH) antibodies were used at 1:10000 dilution in phosphate-buffered saline +0.1% Tween + 10% dry milk. Anti-mouse-Ig-horseradish peroxidase and anti-rabbit-Ig-horseradish peroxidase (Roche Diagnostics Gmb) were used as a secondary antibody at 1:10000 dilution. All western analyses were visualized using enhanced chemiluminescence (Renaissance®, Perkin Elmer).

Cell cycle arrests with benomyl and FACS analysis were carried out as described previously (García-Muse et al., 2003).

For microscopic observation, we used a Leika DMLB microscope with phase contrast. Frames were taken with a Leika 100 camera. Epifluorescence was observed using standard FITC and DAPI filter sets. Image processing was performed with Photoshop (Adobe). Nuclear staining was done using DAPI staining as described previously (Garrido and Pérez-Martín, 2003). WGA staining was performed as described (Wedlich-Söldner et al., 2000).

To test for mating, compatible strains were co-spotted on charcoal-containing PD plates (Holliday, 1974), which were sealed with parafilm and incubated at 21°C for 48 hours. Pheromone induction of autocrine strain was performed as described by García-Muse et al. (García-Muse et al., 2003).

Plant infections were performed as described previously (Gillissen et al., 1992) with the maize cultivar Early Golden Bantam (Old Seeds, Madison, WI). Filaments inside the plant tissue were stained with Chlorazole Black E as described previously (Brachmann et al., 2003).

Protein sequences of fungal Fizzy-related components were downloaded from PubMed (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi) with the exception of Candida albicans Cdh1, which was generously provided by J. Correa-Bordes (University of Badajoz, Spain). Alignments and phylogenetic dendrograms were constructed using ClustalW and NJPlot programs (Thompson et al., 1997).

We took advantage of the high sequence conservation at amino acid level between components of the Fizzy-related family of APC activators from diverse organisms, and performed a PCR-based approach to clone U. maydis genes that may encode for APC activators in this fungus (see Materials and Methods for experimental details). From this approach, we obtained the sequence of a single gene that we named cru1 (cell cycle regulator in Ustilago 1). Comparison of genomic and cDNA sequences indicated that cru1 was intronless (not shown). The conceptual translation of the cru1 sequence produces a putative protein of 592 amino acids (accession number AY118173) that contained typical motives shared among all members of the APC activators (Fig. 1A), such as the seven WD-repeats and the C-box, a motif required for APC association (Schwab et al., 2001). Cru1 showed high sequence similarity to Schizosaccharomyces pombe Ste9/Srw1 (40.9% amino acid identity) and to Saccharomyces cerevisiae Cdh1/Hct1 (38.8% amino acid identity) (Fig. 1B). A dendrogram analysis including fungal components from the Fizzy and Fizzy-related families (Fig. 1C) indicates that the predicted Cru1 protein fall in the same branch as the Fizzy-related family, strongly suggesting that Cru1 could be a new member of this protein group.

Fungal Fizzy-related proteins promote the degradation of mitotic cyclins and therefore ectopic expression of these APC activators resulted in the arrest of the cells in a G2-like state, as a consequence of the unscheduled depletion of mitotic cyclins (Schwab et al., 1997; Visitin et al., 1997; Yamaguchi et al., 1997; Kitamura et al., 1998). We introduced an ectopic copy of the cru1 gene under the control of the arabinose-induced crg1 promoter (Bottin et al., 1996) in the U. maydis UMP28 strain, which carries a GFP-Tub1 fusion that allows the visualization of the microtubule organization. In the resulting strain, TAU54, the level of cru1 mRNA increased to more than 25-fold in the presence of arabinose as carbon source (not shown). TAU54 cells were unable to grow on solid medium containing arabinose (Fig. 2A), indicating that the overexpression of cru1 was deleterious to the cells. To evaluate the response in detail, we characterized in liquid cultures the morphology of the cells, nuclei and MT cytoskeleton after cru1 overexpression. We have observed that after 9 hours of incubation in arabinose-containing medium, essentially all TAU54 cells arrested with an elongated shape, carrying a single nucleus and containing long microtubules that reached the tip of the growing pole (Fig. 2B). This morphology was neither present in TAU54 cells growing in glucose-containing medium nor in UMP28 cells growing in any condition (not shown). Analysis of the DNA content by flow cytometry (Fig. 2C) indicated that TAU54 cells arrested with a 2C DNA content. All these results were compatible with a G2-like arrest in response to a high level of cru1 expression.

U. maydis cells have two B-type cyclins, which are essential for growth. Clb1 is required for both S- and M-phase transitions, whereas Clb2 is only required for M-phase transition (García-Muse et al., 2004). To check whether high levels of cru1 expression correlated with mitotic cyclin depletion, we introduced the cru1 ectopic copy into two strains carrying epitope-tagged active versions of the two U. maydis mitotic cyclins Clb1 (UMP19 cells) and Clb2 (UMP27 cells) (García-Muse et al., 2004). The resulting mutant strains, TAU27 and TAU51, displayed the above-described G2-like arrest in arabinose-containing medium (not shown). Western analysis of TAU27 and TAU51 cells, as well as their respective control strains, UMP19 and UMP27, incubated in glucose- or arabinose-containing medium, showed that the protein levels of Clb1 and Clb2 dramatically decreased in conditions of cru1 overexpression (Fig. 2D).

Overall, the simplest interpretation of these results is that high levels of Cru1 interfere specifically with the accumulation of mitotic cyclins and thereby cause a cell cycle arrest in a G2-like state, as reported with other fungal Fizzy-related components (Schwab et al., 1997; Visitin et al., 1997; Yamaguchi et al., 1997; Kitamura et al., 1998).

To obtain insight into the function of Cru1 we had generated loss of function mutants by replacing the entire cru1 ORF with a hygromycin resistance cassette in the haploid strains FB1 (a1 b1) and FB2 (a2 b2), as well as in the solopathogenic strain SG200 (a1mfa2 bW2bE1) (not shown). The respective mutant strains UMP7 (a1 b1 cru1Δ), UMP9 (a2 b2 cru1Δ) and TAU10 (a1mfa2 bW2bE1 cru1Δ) were first characterized in axenic cultures. The data presented below correspond to the FB1 background, although no significant differences were found in the other genetic backgrounds. Mutant strains were viable, indicating that cru1 was not essential. However, cru1Δ cells growing in solid medium produced smaller colonies than wild-type cells, although the absence of Cru1 factor did not change substantially the generation time of the cells in liquid medium (not shown). In liquid cultures, cru1Δ cells appear to be shorter than wild-type cells, and frequently (around 8% of the population) the cells defective in Cru1 function exhibited at least two buds attached to the same mother cell pole (Fig. 3A). These cells were connected by a septum that could be stained with wheat germ agglutinin (WGA) that recognizes chitin (not shown). Because we never found the typical tree-like structures that have been described in cells unable to perform cell separation (Weinzierl et al., 2002; Wedlich-Söldner et al., 2002), and cells could be separated by mild sonication (not shown), we believe that multibudded cru1Δ cells are the result of a delay in late steps of cell separation, and that eventually the daughter cell separates from the mother. We measured the length of mother and bud cells in asynchronous cultures of wild-type and cru1Δ strains, and then generated a histogram showing numbers of cells as a function of cell length (Fig. 3B). We found that cru1Δ mother cells show a smaller length distribution than wild-type mother cells. By contrast, the distribution of the length of the bud appeared to be similar in both kinds of cells, although buds in the cru1Δ strain were always shorter than 8-9 μm. Where we found cru1Δ buds larger than this size they already had a second bud emanating from their pole, like the cells showed in Fig. 3A, and they were not included in these measurements.

We also analyzed the DNA content of cru1Δ cells growing in liquid medium. This analysis revealed that the apparent 1C peak corresponding to the G1 population (around 10-15% of total cells) was seen in the wild-type strain, but no such a population was found in the cru1Δ strain. This defect was more apparent in cultures growing in less abundant media such as complete medium (CMD) or minimal medium (MMD) (Fig. 3C). Enlargement of the G1 phase is part of the adaptive response to decreasing nutritional condition in U. maydis wild-type cells (Garrido and Pérez-Martín, 2003). One of the proposed roles of Fizzy-related APC activators in fungi is to decrease the protein levels of B-type cyclins during G1, thus ensuring a proper G1 length (Blanco et al., 2000; Yamaguchi et al., 2000). In U. maydis, Clb1, which is a target of Cru1, is involved in S-phase promotion (Garcia-Muse et al., 2004). It could be that in cru1Δ cells, the absence of downregulation of Clb1 during G1 promoted a premature entry in S-phase. To evaluate this possibility, we deleted the cru1 gene in the TAU61 strain, which carries an epitope-tagged version of Clb1. The resulting strain, TAU56, and the TAU61 control strain were arrested at G2/M transition with benomyl, and after being released in benomyl-free medium, the DNA content and the Clb1 levels were followed as these cells proceeded through mitosis and entered G1 (Fig. 3D). We observed that on release, wild-type cells accumulated with a 1C DNA content over time, and Clb1 protein levels abruptly declined to increase again between 100 and 120 minutes later, at the time DNA replication started as evidenced by the steadily increasing 2C peak. By contrast, in the cru1Δ strain, the accumulation of cells in the 1C peak was less apparent on release, suggesting that cells entered in S-phase and duplicated its DNA as soon as they exited from mitosis. Strikingly, the Clb1 protein levels did not decrease so abruptly as in wild-type cells, and Clb1 started to accumulate as early as 80 minutes after release. We also checked for Clb2 levels, but they were undetectable in both wild-type and mutant cells after mitosis exit (not shown). These results indicated a correlation between Clb1 levels and entry in S-phase, and they are consistent with a role of Cru1 avoiding premature entry in S phase by decreasing the levels of the cyclin Clb1 during G1.

In response to nutrient depletion, fungal cells exit from the proliferating phase and they enter into a quiescent status (G0 state) that allows the maintenance of cell viability. We examined the behavior of the cru1Δ cells in response to nitrogen starvation. DNA content of wild-type and cru1Δ cells cultured in nitrogen-free medium (MM-N) was analyzed by flow cytometry. On nitrogen starvation, cells with a 1C DNA content gradually accumulated in the wild-type strain, indicating that most cells arrest in the G1 phase as they cease to proliferate. By contrast, cru1Δ cells accumulated with 2C DNA content (Fig. 4A). The lack of cells with 1C DNA content in the mutant strain was not caused by a failure of cell separation. cru1Δ cells cultured for longer than 8 hours were stained with DAPI, and the nuclear morphology was microscopically inspected. Most of the cells were mononucleated (Fig. 4B). We also examined whether cru1Δ cells retained viability under nitrogen depletion. Wild-type and mutant cells were grown to exponential phase in nutrient-rich medium (YPD) and then transferred to nitrogen-free medium (MM-N). At different times, cells were removed and plated in YPD plates. We found that wild-type cells maintained a high viability for at least 3 days in nitrogen-free medium, whereas the viability of cru1Δ cells decreased dramatically to less than 2% of the cells during the same period of time (Fig. 4C).

The correlation of medium quality and length of G1 (measured as number of cells with 1C DNA content in asynchronous cultures) and the loss of this correlation when Cru1 is absent suggests some link between Cru1 levels and quality of growth medium. Therefore, we evaluated the levels of expression of cru1 gene in the different growth conditions. To this end, we cultured wild-type cells in YPD, CMD and MMD, and total RNA was extracted and checked for the cru1 mRNA levels. Northern analysis revealed that cru1 mRNA levels increased as the quality of the medium decreased (Fig. 5A). Because the cAMP/PKA pathway has been proposed to transmit nutritional signals in U. maydis (Kronstad et al., 1998), we investigated the influence of the cAMP pathway in the levels of cru1 mRNA. We found that the exogenous addition of cAMP resulted in a lineal increase in the mRNA levels of cru1 (Fig. 5B). Consistently, we found that deletion of the ubc1 gene, which encodes the negative regulatory subunit of PKA (Gold et al., 1994), resulted in high levels of both cru1 mRNA (Fig. 5C). In an opposite way, the deletion of the adr1, which encodes one of the catalytic subunits of PKA (Dürrenberger et al., 1998), resulted in low levels of cru1 mRNA (Fig. 5C). These data were consistent with a positive role of PKA in the expression of the cru1 gene.

To assess the role of cru1 in pathogenesis, haploid strains carrying deletions in the cru1 gene were inoculated into maize seedlings. The plant inoculations involved combinations of cru1Δ cells of opposite mating type, combinations of cru1Δ mutants with compatible wild-type strains and combinations of compatible wild-type strains. Symptoms were scored at 14 days after inoculation (Table 2). When compatible combinations of cru1Δ strains (UMP7 and UMP9) were co-injected into maize plants, only 5% of the plants produced tumors compared with 90% in comparable wild-type infections. A more dramatic reduction was seen after infection with the cru1Δ solopathogenic-derivate strain, TAU10, where no tumors were observed at all (Table 2). In the few cases where tumors were produced after co-inoculations of mutant strains, the tumor size was considerably smaller than those observed after infection with compatible wild-type strains (Fig. 6A,B). The mutant tumors contained teliospores, although less in number and irregular in shape (Fig. 6C,D). Moreover, they neither germinated nor produced haploid sporidia (not shown). Co-inoculations of cru1Δ UMP7 strain and the wild-type strain FB2 produced infections that were indistinguishable from wild-type crosses, albeit with less efficiency (Table 2).

Although a low frequency of tumor formation was obtained after infection with mutant strains, we found that 38% of the plants co-injected with the mutant combinations showed chlorosis around the site of inoculation, an early symptom of infection (Christensen, 1963) (Fig. 7A). The observation of these symptoms suggested that cru1Δ cells were capable of infecting the plant issue. Therefore, symptomatic leaves obtained from both mutant and wild-type crosses were sampled after 1 week of inoculation, stained and examined microscopically for the presence of the fungus. Septated hyphae that proliferate massively were observed in plants inoculated with compatible wild-type strains (Fig. 7B). When material obtained from mutant crosses was analyzed, the fungal network observed in wild-type cells was not observed. Careful analysis of the area allowed the detection of short hyphae, composed of no more than four to five cell compartments, suggesting that cru1Δ cells were not able to proliferate massively inside the plant (Fig. 7B).

Taken together, our results indicated that cru1 is required during the infection process at several stages in U. maydis.

The fact that more than half of the plants inoculated with mutant crosses did not develop symptoms led us to investigate the ability of cru1Δ cells to mate, a prerequisite to initiate the pathogenic development. The mating reaction in U. maydis can be easily scored by co-spotting compatible strains (i.e. with different a and b loci) on solid media containing charcoal. In these plates, cell fusion and development of the infective dikaryotic filament resulted in the formation of a white layer of aerial hyphae on the surface of the growing colony (Fuz⁺ phenotype) (Holliday, 1974). Control mating reactions between compatible wild-type strains produced a clear Fuz⁺ phenotype (Fig. 8A). By contrast, when compatible strains carrying the cru1Δ allele were co-spotted or mixed with wild-type compatible strain, formation of the Fuz⁺ phenotype was attenuated (Fig. 8A).

The reduced formation of filaments could be the result of defects in the growth or formation of conjugation tube, which precede the formation of dikaryotic hyphae. To analyze this possibility, we took advantage of the TAU3 strain, an a1 mating type haploid strain that constitutively expresses the pra2 pheromone receptor gene (which recognizes the a1 pheromone). The addition of tiny amounts of synthetic a1 pheromone (0.1 ng/ml) to this strain induces an autocrine loop that reproduces the pheromone response in U. maydis and induces the formation of conjugative tubes (García-Muse et al., 2003). We deleted the cru1 gene in the pheromone-responsive haploid strain TAU3 and treated the resulting strain, TAU7, as well as the control TAU3 cells, with the addition of 0.1 ng/ml of synthetic pheromone. After 8 hours, the autocrine cru1Δ cells were unable to induce the formation of conjugation tubes, whereas in the TAU3 cells, conjugative tubes were produced (Fig. 8B). Strikingly, we observed that the addition of 10⁴-fold more exogenous pheromone (1 μg/ml) to TAU7 cells induced the formation of structures that resembled conjugative tubes (Fig. 8B). Because the ability of TAU3 cells to respond to tiny amounts of pheromone relies on the autocrine production of a1 pheromone - as a result of the transcriptional activation of the gene encoding the a1 pheromone, mfa1, in response to pheromone (Garcia-Muse et al., 2003) - we reasoned that the inability to respond to tiny amounts of pheromone in the TAU7 cells could be related to defects in the pheromone production, more than defects in conjugative tube formation or in the inability to transmit the pheromone signal. To test this assertion we investigated the expression levels of the pheromone precursor gene, mfa1, in TAU3 and TAU7 cells after the addition of 0.1 ng/ml of synthetic pheromone, and we found that, as expected, the cells defective in Cru1 function were unable to express the mfa1 gene (Fig. 8C). Further support of a requirement of Cru1 for pheromone gene expression comes from two additional experiments. First, we observed that the deletion of cru1 gene in the solopathogenic strain SG200 - which carries the genetic information from the two different mating types and as a consequence this genetic background does not require cell fusion to produce the infective hypha (Bölker et al., 1995) - resulted in the inability to produce the Fuz⁺ phenotype, ruling out the absence of cell fusion events as being solely responsible for the mating defect in the cru1Δ cells (Fig. 8A). Second, when the cru1Δ strain is co-spotted with the strain FBD12-17 - a tester strain used to check for pheromone production (Spellig et al., 1994) - the Fuz⁺ phenotype was clearly attenuated (Fig. 8A).

In summary, all these data strongly support a requirement of Cru1 for the expression of the gene encoding the pheromone. Because the pheromone production is required for both cell fusion and the expression of the bW and bE genes - which are necessary for the induction of the filamentous growth - these results help to explain the drop in mating and subsequently plant invasion efficiency observed in cru1Δ cells.

Because of the role of the APC in targeting proteins to proteolysis, it was tempting to speculate the existence of some negative factor acting in the mfa1 expression, which might be removed by the action of the Cru1-APC complex. To support this working hypothesis, we investigated whether high levels of Cru1 factor induced the expression of mfa1 in nutrient-rich medium, a condition in which the mfa1 gene is silenced (Hartmann et al., 1999). UMP17 cells, carrying an ectopic copy of cru1 ORF under the control of Pcrg1, and control FB1 cells were incubated in either nutrient-rich glucose medium (YPD, noninducing conditions) or nutrient-rich arabinose medium (YPA, inducing conditions). We detected elevated levels of mfa1 mRNA in conditions of high expression of cru1, whereas in the control strain FB1 the levels of mfa1 were undetectable in any condition (Fig. 9A).

We also tried to bypass the requirement of cru1 for the expression of mfa1 gene by downregulating the expression of the genes encoding the mitotic cyclins, clb1 and clb2, in cru1Δ cells. We introduced in cells defective for Cru1 function the conditional alleles, clb1^nar and clb2^nar (García-Muse et al., 2004), and we grew the conditional cru1Δ cells as well as the respective controls in restrictive conditions. No bypass of the Cru1 function was obtained when we depleted the cells of Clb2 cyclin (not shown), but the depletion of Clb1 resulted in a clear expression of mfa1, even in cells defective in Cru1 (Fig. 9B).

In this study we provide evidence for a role of Cru1 as a virulence factor in U. maydis. Cru1 belongs to the Fizzy-related family of APC activators and is involved in the promotion of the degradation of mitotic cyclins. In axenic cultures, Cru1 is involved in the coordination between nutritional conditions and the cell cycle. In addition to this role, Cru1 is required in different stages of plant infection by U. maydis. Why is a cell cycle regulator required during the pathogenic process? Our results, discussed below, support a model in which Cru1 is required for cell cycle adaptation to changing environmental conditions. During the infection process, fungal cells are challenged by the surrounding conditions. We propose that the virulence defects associated with a mutation in the cru1 gene are related to the inability of the cru1Δ cells to `interpret' and to adapt the cell cycle to these challenging conditions.

Functional analysis indicated that Cru1 is a new member of the Fizzy-related subfamily of APC activators. First, we found that the levels of the U. maydis B-type cyclins, Clb1 and Clb2, decreased when Cru1 is overproduced. This decrease in B-type cyclin contents correlated with a G2-like cell cycle arrest during which polar growth of the cell continued, resulting in elongated cells. Our results agree with those of other authors showing that overexpression of fungal Fizzy-related components in their respective hosts also resulted in a G2-like cell cycle arrest (Schwab et al., 1997; Visitin et al., 1997; Yamaguchi et al., 1997; Kitamura et al., 1998). Second, using experiments of release of benomyl-arrested cells we have shown that Cru1 is required to keep a low level of Clb1 cyclin during the G1 phase. S. pombe Ste9 is required for the degradation of mitotic cyclins during G1 phase (Blanco et al., 2000) and cells of S. cerevisiae hct1 strains had substantial amounts of mitotic cyclins in G1 phase (Schwab et al., 1997). Finally, we also showed that Cru1 is not essential for the proteolysis of mitotic cyclins at the end of mitosis. We showed that cru1Δ cells exit from mitosis at a rate similar to wild-type cells after release from benomyl arrest. Moreover, in these cells a decrease in mitotic cyclin levels still occurs after release from the G2/M arrest. Because expression of nondegradable versions of Clb1 and Clb2 resulted in a cell cycle arrest at the end of mitosis in U. maydis (García-Muse et al., 2004), a minor role of Cru1 in mitotic exit is in accordance with the fact that the cru1 gene is not essential for growth. This is reminiscent of the situation in S. pombe where proteolysis of mitotic cyclins at the end of mitosis is independent of Ste9 (Blanco et al., 2000). These results imply that another APC complex might be responsible for the degradation of mitotic cyclins at the end of mitosis. We recently identified in the publicly available sequence of U. maydis (http://www.broad.mit.edu/annotation/fungi/ustilago_maydis/index.html) a homologue of Cdc20 in U. maydis, which is essential for growth and is required for mitotic exit (J. Torreblanca and J. P.-M., unpublished).

Cell cycle control by trophic factors has a key role in the regulation of cell proliferation in all organisms. Nutrients are among the most important trophic factors for fungal cells and, hence, mechanisms must exist that couple nutrient availability to crucial cell cycle transitions. Progression through the cell cycle is regulated principally before the onset of S phase and the onset of mitosis. In both cases a critical cell mass must be attained before progression occurs (Nurse, 1975). In U. maydis cells growing in nutrient-rich medium like YPD, only the mitotic size control is operational because cell division produces daughter cells with a mass near to the minimum required to initiate S phase, and consequently cells have a very short G1. However, in less abundant media (like CMD or MMD), mitosis is initiated at reduced cell size, producing smaller daughter cells that must delay the initiation of S phase until a critical mass is achieved, resulting in a longer G1 phase (C. Sgarlata and J.P.-M., unpublished). Our results suggested a role for Cru1 when cells need to delay the cell cycle in G1 phase. This delay probably takes place by decreasing the levels of mitotic cyclins, particularly Clb1, which is required at S phase. In experiments of cell cycle arrest/release with benomyl, we found that in cru1Δ cells the Clb1 cyclin accumulated faster than in wild-type cells, and that this accumulation correlated with a premature entry in S phase. The ability to delay G1/S transition could be important for small cells that had to lengthen the G1 phase until they reach the minimum cell size required to initiate DNA replication or to prevent entry into mitosis from G1. Such a role explains the smaller size of cru1Δ cells, as well as the absence of adjustment of the G1 phase length in nutrient-poor medium. Cells lacking cru1 also showed an apparent delay in cell separation. In U. maydis, cell separation is a complex process that requires the formation of two septa delimiting a fragmentation zone at which disarticulation of cells occurs (Weinzierl et al., 2002). In S. cerevisiae several genes known to be involved in cell separation, such as CTS1, encoding a chitinase, SCW11, encoding a protein with similarities with glucanases, and ENG1, encoding an endoglucanase, are expressed at the M/G1 transition (Spellman et al., 1998; Balandrón et al., 2002). We believe that, in a similar way, the cell separation process in U. maydis must be coordinated to the exit and the entry into a new cell cycle and that a premature entry could uncouple these events.

The inability to adapt the cell cycle to changing environmental conditions would have fatal consequences for the cell. Cells lacking the cru1 gene lost significant viability when they were incubated in medium lacking nitrogen. The G1 phase of the cell cycle is an important decision point in eukaryotic cells. In this phase, a cell determines whether it enters the next cell cycle or ceases proliferation entering in a differentiation process. To make this decision, cells in the G1 phase monitor environmental information such as the presence or absence of nutrients, developmental signals, stress conditions, etc. We believe that because Cru1-deficient cells abrogate G1 arrest induced by nitrogen depletion, the cells are unable to make the decision to enter into an alternative developmental program dedicated to adaptation to starvation conditions.

Our results showing an increase in the cru1 mRNA levels parallel to the decrease in the quality of the medium strongly suggested a way to couple nutrient availability with cell cycle via Cru1 levels: the poorer the nutrients, the higher the Cru1 levels and, subsequently, the longer the G1 phase. Although this simple model accounts for the results reported in this work, additional mechanisms are likely to be involved in the coordination of nutritional conditions and cell cycle progression in U. maydis. Recently, we described a new protein kinase, Crk1, which is also involved in the adaptation to changing nutritional conditions (Garrido and Perez-Martin, 2003). Additional work will be necessary to determine the relationships between Crk1 and Cru1 proteins. Strikingly, the cAMP/PKA pathway seems to have a key role in the control of mRNA levels of both crk1 and cru1 genes (Garrido and Pérez-Martín, 2003). In this work, we provided genetic evidence that the levels of cru1 mRNA are positively regulated by the cAMP/PKA pathway. In other fungi, cAMP has been implicated in the regulation of cell size under different growth conditions. In S. cerevisiae, cell cycle progression in response to growth conditions occurs in late G1 phase (Johnston et al., 1979), and G1 cyclins Cln1 and Cln2 have been identified as targets of cAMP regulation (Baroni et al., 1994; Tokiwa et al., 1994). However, it is unclear how during G1 phase the cAMP signal is transmitted to the cell cycle machinery. For instance, a cAMP- and PKA-mediated inhibition of APC activity has been described recently in S. cerevisiae (Irniger et al., 2000), suggesting a different regulatory circuit. By contrast, in S. pombe, ste9 mRNA and protein levels increased in cells submitted to nitrogen starvation (Blanco et al., 2000), although so far no studies have addressed whether this increase requires a functional cAMP/PKA pathway.

In this study we have shown that an APC adaptor is an important virulence determinant in a plant pathogenic fungus. Absence of the Cru1 factor resulted in a lower rate of plant infection, based on the number of observed disease lesions, and affected both the mating as well as the formation and germination of teliospores. The Cru1 APC activator, therefore, appears to play several distinct cellular roles during infection.

The reduced virulence of cru1Δ mutants is probably attributable, at least in part, to the reduction in mating ability as a consequence of low pheromone expression. The induction of the sexual development in U. maydis could be considered to be part of an alternative fate in response to environmental cues. The expression of the a mating-type genes is regulated by various signals, including nutritional conditions (Hartmann et al., 1999). The requirement of Cru1 protein for pheromone expression could be explained by nonexclusive interpretations. For instance, the Cdk1-Clb1 complex might promote the activity of unknown proteins that repress pheromone expression. Because of the low levels of cru1 mRNA in rich-medium, the activity of Cdk1-Clb1 is not downregulated and these putative factors are active. Transfer of U. maydis cells to less abundant media resulted in higher levels of Cru1 and therefore in lower levels of Cdk1-Clb1 complex. Alternatively, the expression of pheromone genes might be restricted to the G1 phase, and the inability of cru1Δ cells to keep an accurate G1 length could result in a low expression of those genes. The ability to bypass the requirement of Cru1 by depletion of Clb1 levels supports this explanation, because Clb1 is required for the G1 to S and G2 to M transitions, and depletion of Clb1 resulted in cell cycle arrest in either G1 or G2 phases. Future experiments will be necessary to address these possibilities.

In spite of the defects in pheromone expression, a good proportion of mutant cells were able to enter the plant (around 40%), and with the exception of three cases, no progression of symptoms further to chlorotic spots was detectable. In fact, fungal proliferation was arrested at early steps, with filaments composed of no more of four to five cell compartments. Microscopic studies of early stages of infection (Banuett and Herskowitz, 1996) indicated that after plant penetration, fungal cells proliferate extensively, supposedly after recognition of some signals produced by the plant. Because such a morphological switch implies cell cycle adjustments, we could hypothesize that Cru1 is essential at the interphase between plant signals and the induction of the developmental program that allow the proliferation of cells inside the plant. In support of this interpretation it is worth noting that the expression of the cru1 gene appears to be controlled by the cAMP/PKA pathway, and that this same pathway plays an important role in the proliferation of the fungus inside the plant (Gold et al., 1994; Gold et al., 1997; Regenfelder et al., 1997; Dürrenberger et al., 1998; Krüger et al., 2000).

In three cases, we detected progression of the disease until tumor formation. Those tumors were considerably smaller and they contained very few teliospores, most of them with abnormal shape. No viability was detected in these teliospores. Furthermore, we found that teliospores obtained from cru1Δ cells were able to accumulate propidium iodide, whereas wild-type teliospores were impermeable to this molecule (García-Muse, unpublished observations). We believe that these results could be interpreted as a defect in the teliospore maturation process in cru1Δ cells. In S. pombe spore formation requires the participation of a specific APC activator, Fzr1/Mfr1, which is required for the coordination between meiotic cycle and sporulation (Asakawa et al., 2001; Blanco et al., 2001). S. pombe cells lacking fzr1 were defective in the maturation of the spore envelope resulting in low spore viability. Because Cru1 is the only Fizzy-related member we detected in the available sequence of the U. maydis genome, we could conclude that it recapitulates the roles of Ste9/Srw1 and Fzr1/Mfr1 from S. pombe.

Studies from both yeast and mammalian systems indicated that G1 is the phase of cell cycle in which cells respond to extracellular signals to commit to another round of cell division, to withdraw temporarily from cell cycle and become quiescent, or to terminally differentiate (Pardee, 1989). Hence, it is probable that the control of the length of G1 will play an important role in pattern formation and differentiation. A short G1 phase could mean a short window to load a new developmental program and therefore to adapt to changing environmental conditions. On the basis of our work, we propose that the role of Cru1 in U. maydis is to keep an accurate G1 length to elaborate the appropriate response. The progression of the fungal cells through the infection process implies several steps. First, haploid cells have to mate, then the dikaryotic hypha must penetrate the plant; afterwards, hyphal cells must proliferate inside the plant and finally they must produce teliospores. We believe that in each of these steps, fungal cells have to take distinct developmental decisions in response to the surrounding conditions. In other words, cells need time to `think' about the next step and this time could be provided in G1 phase by the activity of Cru1-APC. In the absence of Cru1, because of a narrow G1 phase, only a limited number of cells would be able to guess correctly, so the further they progress in the infection, the lower the probability to take the right decision, resulting in a dramatic drop in the infectivity of the mutant population. In this sense Cru1 might act like a car gear, keeping the G1 phase running while environmental signals are received and the responses are carried out.

In summary, the results reported in this work reinforces the connections between cell cycle and the induction of the pathogenesis program in U. maydis, and highlight the potential significance of cell cycle regulation in microbial pathogenesis.

We thank Prof. J. Correa-Bordes for providing the sequence of C. albicans Cdh1 and critical reading of the manuscript. Dr Gero Steinberg is also appreciated for critical comments on the manuscript. This work was supported by Grant BIO2002-03503 from MCyT and Grant 07B/0040/2002 from CAM. Sonia Castillo-Lluva is a recipient of a FPI fellowship from MCyT.