Critical Role of DNA Checkpoints in Mediating Genotoxic-Stress–induced Filamentous Growth in Candida albicans

Strains and Growth Conditions

The strains used in this study are listed in Table 1. Except where noted, C. albicans was grown at 30°C in YPD medium (2% yeast extract, 1% bactopeptone, and 2% glucose) or in GMM (2% glucose and 6.79 g/l yeast nitrogen base without amino acids) or in GMM supplemented with the required nutrients for auxotrophic mutants. For hyphal induction, 20% serum was added to liquid YPD medium, and the culture was incubated at 37°C.

To prepare synchronized cultures by centrifugal elutriation, cells were grown overnight at 30°C in a medium containing 2% galactose and 6.79 g/l yeast nitrogen base supplemented with amino acids. G₁ cells were obtained using a Beckman (Fullerton, CA) JE or J6-MC elutriator. G₁ cells prepared using different elutriators can differ in average size and uniformity, which may significantly influence the timing of entry into the cell cycle, leading to variation between experiments. However, cells to be compared directly for cell-cycle–related functions were always prepared using the same elutriator.

Sensitivity to DNA-damaging agents was tested on solid or liquid medium. For the former, cells were grown in YPD overnight at 30°C and serially diluted into fresh YPD at concentrations of 5 × 10⁶, 5 × 10⁵, and 5 × 10⁴ cells/ml; after 2 h of growth at 30°C, 5 μl of each culture was dropped onto YPD plates containing different concentrations of HU or methylmethane sulfonate (MMS) and incubated at 30°C. Plates were photographed at 12 and 24 h. For tests in liquid medium, cells were grown in YPD at 30°C overnight and diluted into fresh YPD at a concentration of 5 × 10⁶ cells/ml. After 2 h of growth at 30°C, HU or MMS was added to a final concentration of 50 mM or 0.02%, respectively, for further growth. Aliquots of cells were collected at 2-h intervals, and cell viability was evaluated by spreading diluted samples onto YPD plates and counting colony-forming units after 2 d of incubation at 30°C. Aphidicolin (Sigma-Aldrich, St. Louis, MO) was dissolved in DMSO at 10 mg/ml (∼30 mM), stored at −20°C, and diluted appropriately into liquid growth medium.

Gene Deletion and Mutagenesis

C. albicans deletion mutants were constructed by sequentially deleting the two copies of the target gene from BWP17 (Wilson et al., 1999). The deletion cassettes were constructed by flanking a selectable marker gene (ARG4, HIS1, or URA3) with the AB and CD fragments, which correspond to the 5′- and 3′-untranslated regions (UTRs) of the target gene, respectively. Strain genotypes were verified by Southern blot analysis or PCR. To reintegrate wild-type RAD53 into rad53Δ cells (also see Figure 8, A and B), the genomic region from nucleotide (nt) −1106-2890 (the first nt of the start codon is 1), which contains the promoter, coding sequence, and 3′-UTR, was PCR-amplified using the forward primer 5′-GGGCTCGAGTGACAGTTTTTGACAGTG-3′ (underlined letter is an added XhoI site) and the reverse primer 5′-GGGATCGATTAAGAGATCGAATGATAT G-3′ (underlined letter is a ClaI site). The PCR fragment was cloned between XhoI and ClaI sites in a CIp10-based plasmid (Murad et al., 2000) containing HIS1 as selectable marker. The resultant plasmid was linearized at a unique NdeI site (−857) in the promoter region and transformed into rad53Δ cells (strain WYS3). The same integration strategy was used to introduce various FHA-domain-mutation alleles of RAD53 into rad53Δ cells. Correct integration was verified by PCR using two primers targeting nt −1161 to −1142 (5′-AACATGTTTGTTAGTTCTG-3′) and 53-71 (5′-GTTTGCGTCTGAGTTGTC-3′). The sequence targeted by the first primer is outside of the promoter region included in the integration construct.

A similar strategy was used to reintroduce wild-type RAD9 into the rad9Δ mutant and wild-type MRC1 into the mrc1Δ mutant. For RAD9, the promoter and complete coding region were PCR-amplified from nt −995-3535 and cloned into a CIp10-based, HIS1-marked plasmid. The resulting plasmid was linearized at the unique AatII site at nt −785 and transformed into WYS1 cells. For MRC1, the genomic region from nt −872-3750 was PCR-amplified and cloned into a CIp10-based, URA3-marked plasmid. The resulting plasmid was linearized at the unique HpaI site at nt −227 and transformed into WYS2 cells.

To construct the RNR2-shut-off strain WYRL, one copy of RNR2 was first replaced with ARG4 using the strategy described above. To replace the promoter of the second copy with the MET3 promoter (Care et al., 1999), the selectable marker HIS1 was first inserted upstream of the 5′ end of the MET3 promoter in pGEM-Teasy plasmid (Promega, Singapore). The HIS1-P_MET3 fragment was then flanked with AB and CD fragments at the 5′- and 3′-ends, respectively; these fragments corresponded to nt −833 to −327 and nt 1-386, respectively. The AB-HIS1-P_MET3-CD cassette was cloned into a plasmid for amplification, cut out, and used to transform the rnr2Δ::ARG4/RNR2 heterozygous mutant.

Site-directed mutagenesis of amino acid residues in the FHA domains of Rad53p was performed using the QuickChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA).

Protein Methods

To extract proteins, cells were harvested by brief centrifugation, and ∼100 mg of cell pellet was resuspended in 300 μl of ice-cold lysis buffer containing 9 M urea, 4% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (Sigma-Aldrich, St. Louis, MO), 1.5 M Tris (pH 9.5), 1 mg/ml Pefabloc (Roche, Basel, Switzerland), 1 mg/ml leupeptin (Sigma-Aldrich), 1 mg/ml pepstatin (Sigma-Aldrich), and 0.15% DTT. After adding an equal volume of acid-washed glass beads (Sigma-Aldrich), the cells were broken by three rounds of 45-s beating at 5000 rpm in a MicroSmash MS-100 bead beater (TOMY Medico, Minato-ku, Japan) with 2 min of cooling on ice between rounds. After centrifugation in a microfuge at 13,000 rpm for 10 min at 4°C, the supernatant was collected. The protein concentration of the supernatant was determined using the Bradford protein assay (Bio-Rad, Hercules, CA) after neutralizing the pH using 0.1 M HCl. Approximately 30 μg of protein was separated by 10% SDS-PAGE and transferred to Hybond-C nylon membrane (Amersham, Buckinghamshire, United Kingdom). For Western blot analysis, the membrane was first immersed with gentle shaking in phosphate-buffered saline containing 0.1% Tween-20 (PBST) and 5% nonfat dry milk for 1 h at room temperature; then the membrane was incubated in PBST containing 1% milk and the primary antibody followed by incubation in PBST containing 1% milk and the horseradish peroxidase–coupled secondary antibody. The membrane was developed using the ECL system (Amersham). Anti-CaRad53p antibody was raised against the peptide LRPLDSERNRKSSKQ (amino acids 485-499) by Biogenes GmbH (Berlin, Germany).

Protein dephosphorylation was carried out using the method described by Vialard et al. (1998), and λ-phosphatase was purchased from New England Biolabs (Ipswich, MA).

Microscopy and Flow Cytometry

Staining of nuclei and actin structures was carried out as previously described (Zheng et al., 2003). A Leica DMR fluorescence microscope (Deerfield, IL) interfaced with METAMORPH software (Molecular Devices, Sunnyvale, CA) was used for imaging.

For flow cytometry, ∼10⁷ cells were fixed in 70% ethanol for 1 h at room temperature. Cells were then centrifuged, washed twice with a buffer containing 50 mM Tris-HCl (pH 8.0) and 5 mM EDTA, and then treated with 1 mg/ml RNase in the same buffer for 3 h at 37°C. The cells were then harvested by centrifugation, washed twice with PBS, and incubated overnight at 4°C in 50 μg/ml propidium iodide in PBS. The cells were then diluted 10-fold with PBS and sonicated briefly to break up cell aggregates. The DNA contents of 1 × 10⁴ cells were analyzed in a FACScan flow cytometer (Becton Dickinson, Franklin Lakes, NJ).

Both DNA-Replication Stress and DNA Damage Cause Constitutive Filamentous Growth of C. albicans

As a preliminary step to determine whether DNA checkpoints are involved in the filamentous growth induced by cell-cycle perturbations in C. albicans, we tested the effects of a variety of genotoxic insults. Although HU has been reported to cause filamentous growth (Bachewich et al., 2005), it remains unclear whether cell elongation is triggered by the inhibition of DNA synthesis or by disturbance of other cellular processes. To ask if alternative means of inhibiting DNA replication would also trigger filamentous growth, we examined the effects of the distinct DNA-synthesis toxins HU and aphidicolin (AC), which inhibit ribonucleotide reductase and DNA polymerase α, respectively, and of shutting off the expression of RNR2, which encodes a ribonucleotide-reductase subunit.

The high concentration of HU (200 mM) used by Bachewich et al. (2005) proved to be lethal. However, we found that at lower HU concentrations, the buds of yeast cells first elongated significantly during cell-cycle arrest; later, the cells were able to re-enter the cell cycle and continue to grow in the filamentous form with apparently normal cell and nuclear division, although with a longer cell-cycle time (Figure 1, A–F). Fluorescence-activated cell sorter (FACS) analysis showed that cells treated with 50 mM HU in liquid medium had arrested with a 1C DNA content by 1 h and that the arrest continued for >5 h (Figure 1A). By 7 h, however, the cell cycle had resumed, as shown by the presence of divided nuclei and new buds (Figure 1, B and D). Similar results were obtained on solid medium (Figure 1, E and F) and with 75 mM HU, although the arrest was longer in this case (Figure 1D). The tubular buds were much wider in diameter than hyphae (Figure 1, B and F) and contained clear constrictions at septal junctions (Figure 1C), a feature typical of pseudohyphae (Sudbery et al., 2004). New rounds of nuclear division and bud formation occurred in both the apical and subapical cells, forming branched filaments (Figure 1, B, C, and E). Despite its many normal-looking features, the cell cycle was much slower after its reinitiation. By counts of new bud formation, the first cell cycle after exit from the arrest in 50 mM HU was estimated to be ∼190 min, compared with ∼90 min in HU-free medium.

In parallel experiments, we tested the effects of AC and depletion of Rnr2p. At 100 μM, AC produced effects very similar to those of 50 mM HU: nuclear division was arrested transiently, and bud elongation began during arrest and continued after reinitiation of the cell cycle (Figure 1G). Because RNR2 is essential, we constructed a strain in which one copy of the gene was deleted and the other was placed under control of the MET3 promoter (Care et al., 1999). G₁ cells were inoculated into GMM medium or into GMM containing 1 mM each of methionine and cysteine to switch off RNR2 expression. These conditions effectively blocked DNA replication as evaluated by both flow cytometry and microscopic examination: cells maintained a 1C DNA content for at least 4 h (Figure 2A; compare left and right panels) and arrested with a single nucleus for extended periods (Figure 2B). During the arrest, buds elongated dramatically (Figure 2B). However, like wild-type cells treated with high concentrations of HU, the Rnr2p-depleted cells could not exit from the arrest, probably because they have no other means to raise cellular dNTP levels in the absence of ribonucleotide reductase. Taken together, the results above strongly suggest that DNA-replication stress triggers filamentous growth in C. albicans.

To test whether DNA damage also causes filamentous growth, we examined the effects of the DNA-alkylating agent MMS and of UV radiation. Cells were either inoculated into liquid or spread onto solid YPD medium supplemented with 0.01–0.03% MMS. In both cases, the cells were found to switch to filamentous growth (Figure 3A). After arrest with a single nuclear mass for ∼6 h, the cells re-entered the cell cycle, but they continued to produce elongated cells, although these were less elongated than those produced in response to HU or AC. To test the effect of UV radiation, cells were spread onto YPD plates and exposed to a 3-s pulse of UV radiation. Strikingly, this treatment caused extensive cell elongation (Figure 3B). Taken together, the results show that C. albicans can also switch to filamentous growth when DNA damage occurs.

Identification of the MRC1, RAD53, and RAD9 Genes of the C. albicans DNA-Replication and DNA-Damage Checkpoint Pathways

In eukaryotes, DNA-replication stress and DNA damage activate, through respective checkpoint pathways, a range of cellular responses including cell-cycle arrest, activation of repair-gene transcription, stabilization of replication forks, and delay of late-origin firing (Zhou and Elledge, 2000). To ask if the induction of filamentous growth by genotoxic stress is another of the cellular responses regulated by the checkpoint pathways, we first needed to identify the C. albicans checkpoint genes, which have been little studied to date.

In S. cerevisiae, the serine/threonine protein kinase Rad53p is the downstream effector of the DNA-replication checkpoint (Sanchez et al., 1999), Mrc1p is the signal transducer that transmits the DNA-replication-stress signal to Rad53p (Katou et al., 2003; Osborn and Elledge, 2003), and Rad9p (Weinert and Hartwell, 1989; Zhou and Elledge, 2000) is the signal transducer for the DNA-damage checkpoint. The C. albicans genome (http://www.candidagenome.org) contains candidate orthologues of RAD53 (Contig: Chr3_ Ctg10259, orf 19.6936), MRC1 (Contig: Chr1_ctg10064, orf19. 658), and RAD9 (Contig: Chr5_Ctg10202, IPF8741.3f and IPF8741.5f). CaRAD53 encodes a 699-amino acid (aa) protein that has 46% identity with ScRad53p. Each protein contains a cluster of serine/glutamine (SQ) and threonine/glutamine (TQ) pairs near the N-terminus and two forkhead-associated (FHA) domains at equivalent positions; these flank a highly conserved kinase domain (Figure 4A). CaMRC1 encodes a protein of 1234 aa, which is 138 aa longer than ScMrc1p. The two proteins have only 17% overall sequence identity, but higher identity (32%) is found in the C-terminal third of the proteins. A cluster of 12 [S/T]Q residues, which are known to be the phosphorylation sites of the upstream PI3-family protein kinases Mec1p and Tel1p (Sun et al., 1998; Osborn and Elledge, 2003), exists from aa 90-272 in ScMrc1p. Interestingly, CaMrc1p contains two [S/T]Q clusters, one between aa 1 and 185 with 10 and the other between aa 630 and 742 with 12 (Figure 4B). CaRAD9 encodes an 1165-aa protein that has significant sequence identity (∼30%) to ScRad9p only over the C-terminal 400 aa, which contains two BRTC domains (Figure 4C), which are found predominantly in proteins involved in cell-cycle checkpoints responsive to DNA damage (Glover, 2006). In the N-terminal two thirds of each protein, multiple [S/T]Q residues are present. Strikingly, CaRad9p contains a total of 24 such pairs, 20 of which are located between aa 180 and 373, whereas ScRad9 contains only 10 such pairs in the region from aa 1-690. Between aa 180 and 310 of CaRad9p are 13 TQ pairs, each embedded in extended, conserved motifs of PDTQV/I, which are regularly spaced by 5-7 aa.

Functional Characterization of RAD53, MRC1, and RAD9

To investigate the roles of the putative DNA-checkpoint genes in cellular responses to DNA damage and replication stress, we constructed deletion mutants of RAD53, MRC1, and RAD9 (see Materials and Methods and Table 1). None of the heterozygous mutants showed increased sensitivity to HU or MMS or any other discernible defects (data not shown). Interestingly, in contrast to the lethality of Scrad53Δ, homozygous Carad53Δ cells were viable, although they grew noticeably more slowly than wild-type cells and exhibited hypersensitivity to both HU and MMS (Figure 5A). Flow cytometry showed that in contrast to wild type, the mutant cells were unable to slow down DNA replication in response to MMS treatment (Figure 5B, 0.5- to 2.0-h samples), consistent with a defect in the intra-S checkpoint. Because mother and daughter cells of the mutant often failed to separate during MMS treatment (see Figure 5E), the flow cytometry data did not reveal whether the mutant might also be defective in G2/M arrest (see, however, below). Western-blot analysis of wild-type cells treated with HU or MMS showed that CaRad53p exhibited a retarded electrophoretic mobility that was abolished by λ-phosphatase treatment of the cell lysates (Figure 5C), indicating that CaRad53p, like ScRad53p, becomes hyperphosphorylated in response to DNA damage. Taken together, the results suggest that CaRad53p is indeed the orthologue of ScRad53p.

We next examined the rad53Δ cells for filamentous growth in reaction to various genotoxic stresses. In liquid YPD medium containing 50 mM HU, ∼90% of rad53Δ cells grew in yeast form, whereas the rest showed some elongation that was greatly reduced in comparison with wild-type cells (Figure 5D, left half); also, nuclear division was not arrested in the mutant cells, indicating a defect in G₂/M arrest. The lack of HU-induced filamentous growth was also observed on solid medium (Figure 5D, right half). In addition, rad53Δ cells grew exclusively in yeast form upon MMS or UV treatment (Figure 5E). The lack of cell elongation did not seem to be the result of cell death, because most cells could go through more than one budding cycle, producing microcolonies on plates (Figure 5, D and E), and methylene-blue staining showed no increase in the number of dead rad53Δ cells compared with wild-type cells after 12 h of HU or MMS treatment (data not shown). Thus, Rad53p is critically required for the filamentous growth induced by either DNA-replication stress or DNA damage. All the defects of the rad53Δ cells were fully corrected by reintegrating a copy of wild-type CaRAD53 at the RAD53 locus as described in Materials and Methods and Figure 8A (data not shown).

When rad53Δ cells were examined for serum-induced hyphal growth, we unexpectedly observed a significant defect. Yeast cells were inoculated into YPD containing 20% serum and incubated at 37°C. By 1 h, 70% of wild-type cells but only 1.4% of rad53Δ cells (n = 100) had grown germ tubes with average lengths of 2.8 μm and 1.7 μm, respectively. By 3 h, 86% of wild-type cells had developed true hyphae, with an average length of ∼23.5 μm, whereas only 9% of rad53Δ cells had grown true hyphae, with an average length of 9.3 μm. Could the reduced hyphal growth of rad53Δ cells be the result of their slow growth rate? In S. cerevisiae, Rad53p regulates the availability of dNTPs during S phase (Zhao et al., 1998), so that slow DNA synthesis is one of the reasons for the slow proliferation of rad53Δ cells. Furthermore, Scrad53 cells are delayed in the completion of mitosis in an unperturbed cell cycle (Pike et al., 2003). Therefore, we thought that a comparison of germ-tube formation in G₁ cells might better indicate whether Carad53Δ mutants have a specific defect in hyphal growth. We prepared G₁ cells and first released them into YPD at 30°C for yeast growth to determine budding indices. We found that both wild-type and rad53Δ cells started to bud at ∼2 h (data not shown), indicating similar cell growth rates in G₁ phase. We then subjected the same G₁ cells to hyphal-induction conditions. Strikingly, 25 and 82% of wild-type cells had grown germ tubes at 1 and 2 h, respectively, whereas only 2 and 10% of rad53Δ cells had done so. By 4 h, 18% of the mutant cells had formed hyphae, whereas the rest either remained as yeast or exhibited varied degrees of abnormal elongation (Figure 5F). Together, these data reveal that Rad53p contributes to true hyphal morphogenesis.

Camrc1Δ cells grew significantly more slowly than wild-type cells and were sensitive to HU and MMS (Figure 6A), consistent with defects in DNA checkpoints. Even under unperturbed conditions, 30% of the mutant cells were highly elongated (Figure 6B) and were similar in morphology to cells treated with HU or AC. Interestingly, a fraction of the elongated cells contained a single nucleus (Figure 6B), suggesting cell-cycle arrest caused by checkpoint pathways that do not depend on Mrc1p. It is known that DNA-damage-like structures are generated during DNA synthesis in Scmrc1Δ cells, which causes cell-cycle arrest via the DNA-damage checkpoint (Alcasabas et al., 2001; Katou et al., 2003). Consistent with the hypothesis that the same events occur in Camrc1Δ cells, Rad53p was hyperphosphorylated in these cells (Figure 6C), and flow cytometry analysis of mrc1Δ G₁ cells released into YPD at 30°C showed a much slower S phase than in wild-type cells (Figure 6D), reminiscent of the defect reported for Scmrc1Δ cells (Alcasabas et al., 2001; Katou et al., 2003; Calzada et al., 2005; Szyjka et al., 2005). Thus, the Camrc1Δ mutant exhibits phenotypes similar to those of Scmrc1Δ mutants, supporting their orthologous relationship. The observed constitutive activation of Rad53p and cell elongation in the CamrcΔ mutant is consistent with our hypothesis that activation of DNA-checkpoint pathways leads to filamentous growth.

Carad9Δ cells showed a normal growth rate and resistance to HU but were hypersensitive to MMS (Figure 7A). Flow cytometry showed that the rad9Δ cells were impaired in arresting cells at G₂/M upon MMS treatment (Figure 7B). These results are consistent with Rad9p's expected role in the DNA-damage checkpoint. We next examined whether rad9Δ cells displayed filamentous growth in response to various inducing agents. The mutant showed normal cell-cycle arrest and filamentous growth in both liquid and solid media containing HU (Figure 7C), as well as true hyphal growth when induced by serum (data not shown). However, the rad9Δ cells were significantly compromised for filamentous growth in response to MMS or UV radiation (Figure 7D), demonstrating that Rad9p has a specific role in the DNA-damage checkpoint and is critically required for DNA-damage–induced filamentous growth. Notably, ∼15% of the MMS- or UV-treated rad9Δ cells exhibited various degrees of elongation, which might be due to the presence of alternative signaling pathways that activate Rad53p in C. albicans. In S. cerevisiae, two partially overlapping Rad53p-activation pathways function in response to DNA damage, one containing Rad9p and the other Rad17p, Rad24p, Mec3p, and Ddc1p (Kondo et al., 1999). Consistent with this explanation, we found a lower level of Rad53p phosphorylation in rad9Δ cells than in wild-type cells upon MMS treatment (Figure 7E).

Role of FHA Domains in Rad53p-mediated Cellular Functions

The results above are consistent with a role for Rad53p as the effector kinase in both the DNA-damage and DNA-replication checkpoint pathways and in mediating the induction of filamentous growth by genotoxic stress. The two FHA domains of ScRad53p are known to interact with different upstream and downstream proteins, resulting in different responses to different genotoxic insults (Durocher et al., 1999; Durocher and Jackson, 2002; Emili et al., 2001; Duncker et al., 2002; Tsai, 2002; Leroy et al., 2003; Schwartz et al., 2003). Thus, we asked whether the FHA domains of CaRad53p are required for the filamentous growth induced by genotoxins. To address this, we selected six highly conserved amino acids in each FHA domain for alanine substitution. Each mutant rad53 allele was transformed into rad53Δ cells by integration at the RAD53 chromosomal promoter (Figure 8A). A wild-type copy was similarly integrated as a control (strain WYS3.1). The transformants were then examined for cell-cycle arrest, filamentous growth, and survival upon exposure to HU and MMS as described in Table 2. Among the 12 single-residue mutations, N109A was the only one that had no discernible deleterious effect on any of the Rad53p functions examined. The H84A substitution was similar to N109A except that it considerably reduced the number of elongated cells upon HU or MMS treatment. All other mutations, to widely varied extents, impaired multiple Rad53p-mediated functions. The G65A and N104A mutations are interesting in that they only moderately impaired cell-cycle arrest and survival upon exposure to HU or MMS but almost completely abrogated cell elongation, indicating that the mutations resulted in a genetic separation of cell-cycle arrest and cell elongation. Similarly, several mutant alleles, such as rad53-R586A, were able to support both cell-cycle arrest and cell elongation without providing resistance to HU or MMS, consistent with a separation of the cell-cycle arrest and DNA-damage-repair functions of Rad53p.

Because it is important to understand the causal relationship between cell-cycle arrest and cell elongation, we carried out more detailed analysis of the rad53-G65A and rad53-N104A alleles using elutriated G₁ cells. On release into YPD, the cells of wild-type, rad53Δ, rad53-G65A, and rad53-N104A strains started to bud at a similar time of ∼2 h (data not shown), but nuclear division was significantly delayed in rad53Δ relative to wild-type cells (Figure 8B), a phenotype also observed in Scrad53Δ cells (Pike et al., 2003). rad53-G65A and rad53-N104A cells underwent nuclear division at a time earlier than rad53Δ but later than wild-type cells, indicating that the mutant rad53 alleles are partially functional in regulating nuclear division kinetics. In YPD containing HU or MMS, rad53Δ cells started to show divided nuclei between 4 and 5 h, and nearly 50% of HU-treated and 80% of MMS-treated cells had completed nuclear division by 8–9 h (Figure 8B). In contrast, more than 80% of wild-type, rad53-G65A, and rad53-N104A cells still contained a single nuclear mass at 9 h. The percentages of cells with divided nuclei were slightly higher in the mutants than in wild-type cells, indicating that the G65A and N104A mutations compromise slightly the cell-cycle-arrest function of Rad53p. However, in marked contrast to the highly elongated wild-type cells, rad53-G65A, and rad53-N104A cells exhibited little cell elongation in response to HU or MMS (Figure 8C). The results confirm that the G65A and N104A mutations preferentially impair the function of Rad53p required for cell elongation without significantly compromising its function in mediating cell-cycle arrest. The rad53-G65A and rad53-N104A cells also showed defects in cell separation, which were especially prominent in fresh cultures started from elutriated G₁ cells (Figure 8C). However, the defect became less severe when the cultures reached higher density, when many separate cells were present (data not shown).

To evaluate MMS-induced cell-cycle arrest in rad53-G65A and rad53-N104A cells, elutriated G₁ cells were grown in the presence of 0.02% MMS, and their DNA contents were examined by flow cytometry. Figure 8D shows that MMS-treated rad53-G65A and rad53-N104A cells exhibited much slower S-phase progression than untreated cells and then arrested in G₂/M phase with a 2C DNA content, indicating a functional checkpoint response for cell-cycle arrest. The flow-cytometry profiles of the untreated cells started to show an additional peak after 4 h because of the formation of cell aggregates. Western blot analyses showed that cellular levels of the mutant versions of Rad53p were comparable to that in wild-type cells, and the mutant proteins underwent similar hyperphosphorylation in response to HU or MMS (Figure 8E), indicating that the defects of the mutants are not due to failure in Rad53p hyperphosphorylation.

In summary, mutation analyses of Rad53p functions identified two mutations, G65A and N104A in the FHA1 domain, that nearly completely abolish filamentous growth in response to either DNA-replication inhibition or DNA damage without significantly impairing cell-cycle arrest, demonstrating a critical role of the FHA1 domain in mediating the DNA-damage–induced filamentous growth. Because all the mutations in the FHA2 domain also compromise filamentous growth to varied extents (Table 2), both FHA domains contribute to the induction of filamentous growth by genotoxic stress. Our results also demonstrate that both FHA domains play essential roles in the cell's resistance to genotoxins.

Conservation of the DNA-Replication and DNA-Damage Checkpoint Pathways

Using the completed C. albicans genome database, we identified candidate genes for several key proteins of the DNA-replication and DNA-damage checkpoints, including the Chk2-family kinase Rad53p and the signal transducers Mrc1p and Rad9p. CaRad53p is highly similar to ScRad53p in both sequence and domain organization. ScRad53p is activated by phosphorylation and acts as a downstream effector of both the DNA-replication and DNA-damage checkpoints (Zhou and Elledge, 2000). Our data show that a Carad53Δ mutant, like Scrad53 mutants, is hypersensitive to genotoxins and fails to arrest cell-cycle progression, and CaRad53p is also hyperphosphorylated upon DNA damage. These results indicate that CaRad53p is an orthologous Chk2 kinase. However, unlike the lethality of Scrad53Δ mutations, CaRAD53 is not essential for cell survival. Although the candidates for CaMrc1p and CaRad9p only share localized sequence similarity with ScMrc1p and ScRad9p, their relatedness was supported by the presence of similar functional domains and motifs such as the [S/T]Q clusters in both proteins and the BRCT repeats in Rad9p. Moreover, Camrc1Δ and Carad9Δ mutants exhibited a range of defects, many of which are also found in the corresponding S. cerevisiae mutants, supporting their roles in the DNA-replication and DNA-damage checkpoints, respectively. Identification of these key components of the checkpoint pathways in C. albicans allowed us to address the role of each pathway in mediating the filamentous growth induced by genotoxic stress, as discussed below.

S. cerevisiae contains another conserved checkpoint kinase, Chk1p, which primarily mediates response to DNA damage (Sanchez et al., 1999). Using the ScChk1p sequence to BLAST-search the C. albicans genome database, we were not able to identify a candidate Chk1p. Similar scores of e-31 to e-34 were obtained for several protein kinases with similar domain organizations, suggesting either that one of these kinases is the true orthologue of ScChk1p or that functionally redundant kinases are present for Chk1p functions in C. albicans. We are currently investigating these possibilities.

Role of the DNA Checkpoints in the Filamentous Growth Induced by Genotoxic Stress

The observation that a multitude of conditions that exert genotoxic stress via different means and cellular targets all lead to filamentous growth in C. albicans suggests that a common mechanism may be involved. The DNA checkpoints seemed likely to be part of this mechanism, because they are activated under these various circumstances. We tested this hypothesis by blocking the checkpoint pathways by gene deletion. Deletion of RAD53 nearly completely blocked the filamentous growth caused by either inhibition of DNA replication or DNA damage. In contrast, deletion of RAD9 blocked the filamentous growth caused by DNA-damaging agents but not that caused by DNA-replication inhibitors, consistent with the evidence that CaRad9p functions specifically in the DNA-damage checkpoint. The lack of filamentous growth in the checkpoint mutants is unlikely to be the result of cell death, for two reasons. First, methylene-blue staining did not reveal increased cell death in the mutants relative to wild-type cells. Second, the mutant cells treated with HU or MMS could undergo multiple cell cycles to produce microcolonies on plates. Taken together, the data indicate that intact DNA checkpoints are indeed required for the filamentous growth induced by genotoxic stress in C. albicans.

Role of Rad53p in True Hyphal Growth

A rather unexpected finding was the impaired ability of rad53Δ cells in true hyphal growth. Initially, we thought that this effect might reflect the slow growth of the mutant cells. Previous studies of Scrad53 mutant cells indicated that the slow growth is attributable to several defects: limitation of dNTPs during S phase (Zhao et al., 1998), accumulation of excess histones (Gunjan and Verreault, 2003), and slower nuclear-division kinetics (Pike et al., 2003). An important feature of C. albicans is the inducibility of germ-tube formation in early-G₁ yeast cells. This allowed us to compare the responses of mutant and wild-type cells to hyphal induction before the G₁/S transition, thus avoiding the phases presumably responsible for the slow growth of rad53Δ mutants. We found that elutriated G₁ cells of both wild-type and rad53Δ strains indeed started to bud at about the same time. However, the rad53Δ cells exhibited a significant decrease in the number of germ tubes formed. Moreover, the majority of the rad53Δ cells remained as yeast or exhibited abnormal elongation after prolonged hyphal induction. These data suggest a role for Rad53p in true hyphal development. Although how loss of Rad53p function causes the hyphal defects remains unclear, several studies in S. cerevisiae have suggested possible functional connections of Rad53p with cell morphogenesis. Pike et al. (2004) reported the identification of a new checkpoint protein, Mdt1p, and showed that even under unperturbed conditions, mdt1Δ cells exhibited a delay in anaphase completion, which causes cell elongation. Interestingly, Mdt1p is constitutively threonine-phosphorylated and binds to the FHA1 domain of Rad53p. In addition, Wu and Jiang (2005) recently found that a S. cerevisiae mutant expressing cdc42-Y40F was specifically defective in HU-induced filamentous growth and thus proposed a direct interaction between Rad53p and Cdc42p, a Rho GTPase essential for polarized cell growth (Etienne-Manneville, 2004). Moreover, ScRad53p activation is known to lead to G₂/M arrest by inhibiting the Polo-like kinase Cdc5p (Sanchez et al., 1999), and depletion of CaCdc5p has been shown to cause cell elongation (Bachewich et al., 2003). Further studies of the molecular events linking CaRad53p to cell morphogenesis might well follow these leads from S. cerevisiae.

Genetic Separation of Rad53p-mediated Responses to Genotoxic Stress

Activation of ScRad53p results in diverse cellular responses, partly through FHA-domain–mediated interactions with different upstream and downstream proteins (Durocher et al., 1999; Pike et al., 2004). The two FHA domains of ScRad53p are known to mediate associations with different partners (Durocher and Jackson, 2002; Tsai, 2002; Schwartz et al., 2003), and FHA-domain mutations can alter their substrate specificity (Durocher et al., 2000; Yongkiettrakul et al., 2004). We reasoned that if CaRad53p promotes filamentous growth through FHA-domain–mediated interactions with other proteins, it might be possible to find FHA-domain mutations that block or compromise the genotoxic-stress–induced filamentous growth without affecting other Rad53p-controlled responses. Indeed, we found two single-amino-acid mutations in FHA1 that almost completely blocked the filamentous growth induced by HU, MMS, or UV but left the cell-cycle-arrest function largely intact. Similar functional separation was also observed between cell-cycle arrest, filamentous growth, and resistance to genotoxic insults. For example, rad53-G586A cells exhibited largely normal cell-cycle arrest and significant cell elongation but were as sensitive to HU and MMS as rad53Δ cells. One plausible explanation of such separation of function is that the different substitutions differentially affect Rad53p's interactions with various downstream effector proteins. In this case, the filamentous growth might constitute a specific Rad53p-mediated response to genotoxic stress. Although we have shown that cell-cycle arrest can occur without concurrent cell elongation in some RAD53 mutant cells, it remains unclear whether the genotoxic-stress–induced filamentous growth can happen in the complete absence of cell-cycle arrest.