We have analysed the effect of RAD52 deletion in several aspects of the cell biology of Candida albicans. Cultures of rad52Δ strains exhibited slow growth and contained abundant cells with a filamentous morphology. Filamentation with polarization of actin patches was accompanied by the induction of the hypha-specific genes (HSG) ECE1, HWP1 and HGC1. However, filament formation occurred in the absence of the transcription factors Efg1 and Cph1, even though disruption of EFG1 prevented expression of HSG. Therefore, expression of HSG genes accompanies but is dispensable for rad52Δ filamentation. However, deletion of adenylate cyclase severely impaired filamentation, this effect being largely reverted by the addition of exogenous cAMP. Filaments resembled elongated pseudohyphae, but some of them looked like true hyphae. Following depletion of Rad52, many cells arrested at the G2/M phase of the cell cycle with a single nucleus suggesting the early induction of the DNA-damage checkpoint. Filaments formed later, preferentially from G2/M cells. The filamentation process was accompanied by the uncoupling of several landmark events of the cell cycle and was partially dependent on the action of the cell cycle modulator Swe1. Hyphae were still induced by serum, but a large number of rad52 cells myceliated in G2/M.
A key question in biology is to know whether polarization of growth has a central core where several pathways converge or is the result of a number of independent processes. Fungi appear to be excellent models for this type of study, as changes in morphology are linked to changes in the polarization of growth. So far, these studies have been mostly restricted to Saccharomyces cerevisiae. However, recent advances in the molecular and cellular biology of other fungal species make it possible not only to test the universality of the currently prevailing models, but also to extend these studies to polymorphic organisms, where polarization of growth during hyphal induction requires new pathways and can be regulated by a number of specific environmental conditions. One such organism is Candida albicans, an opportunistic human fungal pathogen that causes superficial infections or even more severe systemic infections of blood and tissues in immunocompromised patients (Calderone, 2002).
Both internal stimuli as well as environmental stress can trigger polarization of growth. In S. cerevisiae and C. albicans, cell polarization during the budding cycle is restricted to a short period during the initial steps of budding. When the bud reaches a critical size, polarized growth is shut down and replaced by isotropic growth (Staebell and Soll, 1985; Lew and Reed, 1993). The sequential combination of polarized/isotropic growth results in the typical ovoid morphology of the yeast cell. These observations indicate that polarization of growth is intrinsically linked to the cell cycle. As with most of the cell cycle events, polarization of growth in S. cerevisiae has been shown to be controlled by cyclic modification of cyclin-dependent kinases. For instance, G1 cyclins activate the Cdk1 cyclin-dependent kinase (Cdc28 in S. cerevisiae) to promote apical growth through polarization of actin and polarized secretion to the tip; in later stages of the cell cycle, B-type cyclins bind to Cdk1 to promote isotropic growth through redistribution of actin and random secretion. A permanent activation of the G1 cyclins or abrogation of B-type cyclins resulted in elongated cells. Activation of the Cdc28 kinase by B-type cyclins also requires dephosphorylation of Cdc28 by the tyrosine phosphatase Mih1, as tyrosine inhibitory phosphorylation of Cdk1 by the kinase Swe1 (the S. cerevisiae member of the Wee1 kinase family) prevents both entry into mitosis and the polar/isotropic shift (Lew and Reed, 1995a,b). The Swe1-mediated inhibition of Cdc28 acts in the morphogenesis checkpoint apparently until the bud has reached a critical size. Deletion of Swe1 results in premature entry into mitosis and the formation of abnormally small cells (Harvey and Kellogg, 2003). Swe1 accumulates during the S phase and is hyperphosphorylated in order to be target for ubiquitin-mediated degradation (Sia et al., 1998; Shulewitz et al., 1999; Sreenivasan and Kellog, 1999; McMillan et al., 2002; Sakchaisri et al., 2004). Interestingly, a multikinase system, including Clb2-Cdc28 and the polo-like kinase Cdc5, contributes to this hyperphosphorylation (Asano et al., 2005; Harvey et al., 2005; Lee et al., 2005). In contrast, hypophosphorylation of Swe1 results in its stabilization and therefore in a Swe1-dependent inhibition of Clb-Cdc28 (Lew, 2003).
In S. cerevisiae, hyperpolarization of growth can be triggered by some environmental conditions, such as nitrogen deprivation or exposure to branched-chain alcohols, which result in the so-called filamentous or invasive growth. For instance, under nitrogen limitation, cells switch from the typical yeast to the pseudohyphal morphology, characterized by the presence of elongated cells arranged in chains (Gimeno et al., 1992). This condition activates two partially interconnected morphogenetic pathways, MAPK and cAMP, and results in an alteration of the gene expression pattern (i.e. FLO11). However, other parameters related to the cell cycle, including a G2/M cell cycle delay and actin polarization, are also affected. This was explained by the finding that the morphogenetic MAPK and cAMP pathways connect with well-known cell cycle regulators, including Clb2/Cdc28 and its inhibitor Swe1, suggesting that cell cycle components could serve as a link between environmental factors and filamentation (Rua et al., 2001). Although swe1Δ mutants form pseudohyphae in response to nitrogen-limited growth (Ahn et al., 1999), Swe1 seems to contribute to pseudohyphal formation under certain conditions (La Valle and Wittenberg, 2001), suggesting the existence of at least two pathways. When induced by branched-chain alcohols, filamentation has been shown unambiguously to be dependent on Swe1 (La Valle and Wittenberg, 2001; Martinez-Anaya et al., 2003).
Candida albicans possesses an even more complicated differentiation programme as, in addition to the yeast and pseudohyphal phases, the organism exhibits a hyphal morphology similar in a number of ways to filamentous fungi (Calderone, 2002). During the hyphal growth of C. albicans, deposition of new material leading to the enlargement of the cell wall is restricted to the apical tip. During the C. albicans yeast–hyphal transition or morphogenesis, polarization of growth is, at least in part, controlled by the environment. Several independent circuits, including PKA, MAPK, pH and microaerophilic signalling pathways, promote hyphal development by activating positive effectors or transcription elements, whereas others factors, such as a combination of repressors, Nrg1-Tup1 and Rfg1-Tup1, repress filamentation (Liu, 2001; Brown, 2002; Kadosh and Johnson, 2005). However, there is growing evidence that cell cycle components are also involved in hyphal formation. For instance, the G1 cyclin Cln1 is necessary to maintain hyphal growth under some conditions (Loeb et al., 1999), and a regulator of B-cyclins (Clb2 and Clb4) gene expression is required for hyperpolarized bud growth (Bensen et al., 2002). More recently, it has been shown that depletion of a homologue of the polo kinase Cdc5, which is involved in the adaptation to the DNA-damage checkpoint in S. cerevisiae (Toczyski et al., 1997; Sanchez et al., 1999), as well as loss of Nim kinases Hsl1 and Gin4 induces hyphal-like growth in C. albicans (Bachewich et al., 2003; Wightman et al., 2004). Compelling evidence for the involvement of cyclins in hyphal formation has been recently provided by Zheng et al. (2004), who discovered a new G1 cyclin-related protein, Hgc1, whose deletion abolishes hyphal growth. In addition, the G1 cyclin Cln3 regulates morphogenesis by negatively regulating the yeast hyphal transition (Bachewich and Whiteway, 2005; Chapa y Lazo et al., 2005), and depletion of either mitotic cyclin Clb2 or Clb4 results in hyperpolarization of growth (Bensen et al., 2005). Finally, cell cycle arrest during S or M phases also generates polarized growth (Bachewich et al., 2005).
Hyperpolarization of growth in S. cerevisiae can also be triggered by internal signals other than regular cyclic events of the cell cycle as a number of conditions that slow DNA synthesis and activate checkpoints (including treatment with hydroxyurea, MMS, nucleotide-analogues, or defects in DNA polymerase ∈ or DNA ligase Cdc9p) also induce filamentation (Jiang and Kang, 2003). The DNA damage signal occurs through the checkpoint kinase Rad53 which phosphorylates Swe1p, thus connecting with the morphogenetic checkpoint transmission pathway. In a recent study, we characterized the Rad52 from C. albicans, showing that it plays a role in homologous recombination (HR), DNA repair and genomic stability in C. albicans (Ciudad et al., 2004). In addition, RAD52 is required for virulence in a mouse model of disseminated candidiasis. Of interest, the RAD52 gene-reconstituted strain repairs a partially deleted chromosome during infection and, in doing so, virulence is restored to the level of single-gene strains (Chauhan et al., 2005). In the current study, we further show that null rad52 mutants are filamentous and constitutively express ‘hyphal’-specific genes (HSG). Interestingly, this filamentation was independent of the presence of transcription factors Efg1 and Cph1, even though disruption of EFG1 prevented expression of HSG, including the hypha-specific cyclin Hgc1. A large number of Rad52-depleted cells arrest in G2/M, suggesting that some intrinsic DNA damage caused by the absence of Rad52 triggers the DNA replication and/or DNA-damage checkpoints. Filamentation seems to occur through the subsequent uncoupling of several events of the cell cycle.
Effect of Rad52 disruption on growth rate, colony and cell morphology and agar invasiveness
Parental strain CAF2 and several rad52 Ura+ mutant derivatives, including heterozygous TCR1, null strains TCR2.1 and TCR2.2, and two reintegrants, TCR3.1.1 and TCR3.2.1, were subjected to a basic phenotypic analysis. As shown in Table 1 for strains grown in YEPD, null strains (TCR2.1 and TCR2.2) exhibited a generation time significantly longer (140 min) than the heterozygous TCR1 (80 min), which grew at the same rate as parental CAF2 [or CAI4 supplemented with uridine (Uri)]. Reintegrants TCR3.2.1 (Table 1) and TCR3.1.1 (not shown) grew at the same rate as the heterozygous TCR1 (g = 80 min). Similar conclusions were reached when these strains were grown in minimal synthetic medium (SC).
Table 1. Generation time and relationship of cfu to OD600 in wild type and rad52 mutants grown in YEPD.
cfu ml–1 at OD600 = 1 (exponential phase)
Generation time (min)
6 × 107
6 × 107
6 × 107
RAD52/rad52 Ura+ (URA3)
6 × 107
6 × 107
1 × 107
rad52/rad52 Ura– (+Uri)
1 × 107
Colonies from null strains TCR2.1 and TCR2.2 (not shown) (Ura+) were wrinkled with a thorny or spiny appearance and produced radiating filaments (Fig. 1A, c and Fig. 1B, b), whereas those of CAF2 and the heterozygote were smooth and lacked peripheral filaments (Fig. 1A, a and b). As shown for reintegrant TCR3.1.1, re-introduction of the gene in either null strain restored the colonial morphology of the wild type (Fig. 1A, d). When grown in liquid YEPD, rad52Δ cultures contained, in addition to yeast cells, elongated forms that included filamentous cells and pseudohyphal-like elements as well as malformed cells (Fig. 1A, g and Fig. 1B, a), whereas CAF2, heterozygous and reintegrant grew as a yeast (Fig. 1A, e, f and h). Interestingly, filamentation was significantly more abundant on agar (Fig. 1B, b) than in liquid media (Fig. 1B, a), suggesting that a solid surface favours the elongation of rad52Δ cells. The absence of URA3 did not change the phenotypes of these strains (not shown). These differences in morphology decisively influenced the value of colony-forming units (cfu) per volume of cell suspensions with identical OD600 (Table 1). For example, the cfu values of the rad52 null strains were sixfold lower than the Rad52+ counterparts in exponentially growing YEPD cultures. It is likely that, in the mutant, aggregations of several nucleated cells frequently serve as progenitors of a single colony. A second reason that may account for this behaviour is the presence of dead cells in rad52Δ cultures, as colonies of this mutant were slightly pink in phloxin B plates (Fig. 1A, i–l). Finally, the enlargement of rad52 cells could also contribute to an increase in the OD value.
The above-mentioned strains were also grown in two different media, M199 and Spider, known to promote hyphal development. As expected, parental and heterozygote strains gave rise to colonies with abundant peripheral projections indicative of hyphal growth, but, unexpectedly, these projections were significantly reduced in the null mutant (Fig. 1C). This phenotype could be attributed to the rad52 mutation, as the revertant TCR3.1.1 partially regained the wild-type pattern. These observations suggest that most filaments produced by rad52Δ cells are not true hyphae but result from an incomplete or even different developmental programme that, like hyphae, includes hyperpolarization of growth.
The filamentous nature of the rad52Δ strains was paralleled by a significant increase in their invasiveness in vitro. When grown in SC plates (minus Uri), rad52Δ Ura+ strains penetrated the agar whereas wild type or the heterozygote did not (Fig. 1D), and the same occurred in YEPD plates (not shown). Reintegrants TCR3.1.1 and TCR3.2.1 (Fig. 1D) behaved like the heterozygote, i.e. non-invasive.
Characterization of a conditional RAD52 strain
The fact that two rad52 strains are filamentous and exhibit a significantly lower growth rate and that re-introduction of RAD52 consistently restores both phenotypes to those of the wild type strongly suggests that those alterations are indeed due to the absence of Rad52 and not to unrelated mutations by cells accumulated during growth as a consequence of intrinsic defects in DNA repair. To further analyse the correlation of genotype and phenotype, we constructed and characterized a conditional rad52 strain, EAT4.1, in which a wild-type allele is regulated by the MET3 promoter (see Experimental procedures). Northern analysis indicated that the RAD52 message was present under non-repressive conditions but almost disappeared in less than 15 min following the addition of methionine (Met)/cysteine (Cys) and remained at almost undetectable levels during the next 5 h (Fig. 2A). This indicates that the half life of the RAD52 mRNA is very short and that the MET3 promoter was effectively repressed by Met/Cys. However, reverse transcription polymerase chain reaction (RT-PCR) experiments indicated the presence of small amounts of specific RAD52 mRNA following shift of the conditional EAT4.1 to repressive conditions for 15 min and 5 h (Fig. 2B) indicating that although repression was effective (compared with the unrepressed control) it was not absolute (as seen in the rad52Δrad52Δ control). In spite of this basal expression of RAD52, when incubated under repressive conditions, strain EAT4.1 showed phenotypes of the rad52Δ null strain suggesting that the levels of Rad52p are low enough to compromise cell function. Thus, repressed EAT4 was at least as sensitive to MMS (Fig. 2C) and UV light (not shown) as a null strain (TCR2.2). In addition, it was as invasive as the null strain, as shown by its capacity to penetrate agar (Fig. 2E). As expected, under non-repressive conditions, EAT4.1 behaved as the heterozygote in the same assays. When derepressed EAT4.1 yeast cells were plated on SC or YEPD plates containing Met/Cys, many of them developed into filamentous microcolonies (Fig. 2D, a and b), although the extent of filamentation was lower than with the null strain, probably due to the presence of low amounts of Rad52 or to the absence of secondary effects in the short-term culture of the depleted strain; the same was true in liquid medium (see Fig. 5D, e). In addition, as described for the null strain, filamentation of repressed EAT4.1 was much less prominent in liquid medium than in solid agar. In any case, the early appearance of these colonies suggests that the filamentous morphology yielded by rad52Δ cells is indeed due to Rad52 depletion and not to secondary mutations. On the other hand, both unrepressed EAT4.1 (Fig. 2D, c and d) and the heterozygote supplemented with Met/Cys (not shown) gave rise to smooth colonies formed almost exclusively by yeast cells. A decrease in the number of filaments was observed after long incubation periods (i.e. 23 h) in the presence of Met/Cys, when both Northern and RT-PCR analyses indicated the presence of significant amounts of RAD52 mRNA (Fig. 2A and B). Both effects are likely due to the consumption of Met/Cys or to the lysis of filaments. In the absence of Met/Cys, EAT4.1 grew as yeast in both liquid SC and YEPD (not shown).
Expression of hypha-specific genes in rad52Δstrains
To know whether filamentation of rad52Δ null strains was accompanied by the expression of HSG, we performed Northern hybridization using specific probes of ACT1, ECE1 and HWP1. ACT1 is expressed at similar levels in both yeast and hyphae, whereas ECE1 and HWP1 mRNAs are significantly increased in hyphae (Birse et al., 1993; Staab et al., 1999; Kadosh and Johnson, 2005). Parental CAF2 and the several C. albicans rad52 mutant strains were first grown under conditions that favour the yeast form (YEPD at 30°C). As shown in Fig. 3A, the mRNA levels of both HSG were almost undetectable in those strains that grew as yeast (CAF2, heterozygote TCR1, and reintegrant TCR3.2.1). In contrast, expression of these genes was very high in rad52Δ cells (TCR2.2), which, under the same conditions, produced extensive filamentation. As expected, the addition of 10% serum (at 37°C) to Rad52+ cells resulted in both a massive formation of filaments (not shown) and a significant induction of ECE1 and HWP1. However, no significant change in the expression of these genes was observed when rad52Δ cells were supplemented with 10% serum. Figure 3A also shows that the level of the ACT1 message remained constant in all the strains and was not modified by the addition of serum. These results demonstrate that the constitutive filamentation of rad52Δ cells is accompanied by the induction of genes that are usually derepressed during myceliation.
Recently, a novel C. albicans G1 cyclin, Hgc1, has been shown to be specifically expressed during hyphal induction, and a hgc1Δ mutant was defective in hyphal growth (Zheng et al., 2004). In order to know whether filamentation of Rad52-depleted cells was accompanied of the induction of this cyclin, we performed Northern analysis. As shown in Fig. 3B, parental SC5314 grown in YEPD at 30°C did not express HGC1 but did when supplemented with 10% serum and incubated at 37°C; in contrast, rad52Δ cells from mutant TCR2.1, incubated in YEPD at 30°C, showed significant levels of specific HGC1 mRNA, and the amount of this transcript remained at about the same level upon induction of hyphal growth at 37°C in the presence of serum. A second rad52Δ strain (TCR2.2) also constitutively expressed HGC1 at 30°C in the absence of serum (Fig. 3B). As expected HGC1-specific mRNA was absent in a hgc1Δ deletant, even in the presence of serum (Fig. 3B). Accordingly, filamentation of rad52Δ cells is accompanied by expression of several HSG, including ECE1, HWP1 and HGC1.
Filamentation of rad52Δ null cells occurs in the absence of Efg1, Cph1 and Hgc1, and is independent of an operative NHEJ pathway
Various signalling pathways, defined by specific transcription factors, have been reported to control myceliation in C. albicans. In particular, the transcription factor Efg1, a downstream effector of the Ras1/cAMP pathway, plays an important role in morphogenetic regulation in response to starvation, pH and serum stimulation, whereas the transcription factor Cph1, an effector of the MAP kinase pathway, only channels responses to a limited set of starvation conditions (Stoldt et al., 1997; Liu, 2001; Brown, 2002). In order to determine the role of these factors in the constitutive filamentation of rad52Δ cells, we deleted both copies of RAD52 in both efg1Δ and efg1Δcph1Δ genetic backgrounds (Ciudad et al., 2004). As shown in Fig. 3C (lower row), young colonies of both efg1Δ and efg1Δcph1Δ contained, like parental CAF2, exclusively yeast cells whereas colonies of either efg1Δrad52Δ or efg1Δcph1Δrad52Δ developed filaments, like rad52Δ. Furthermore, old colonies of wild type, efg1Δ and efg1Δcph1Δ were completely smooth whereas those of rad52Δ, efg1Δrad52Δ or efg1Δcph1Δrad52Δ had abundant peripheral filaments (not shown). In agreement with this, liquid cultures of efg1Δ or efg1Δcph1Δ contained blastospores almost exclusively whereas the introduction of the rad52 mutation in any of these backgrounds resulted in a significant amount of filamentous cells (Fig. 3C, upper row). We conclude that the filamentation caused by the absence of RAD52 may occur in the absence of the morphogenetic pathways defined by Efg1p and Cph1p.
It has been reported that transcription of ECE1 depends on the Efg1 pathway (Sharkey et al., 1999; Braun and Johnson, 2000). In agreement with this result, ECE1 was expressed in rad52Δ but not in efg1Δrad52Δ or efg1Δcph1Δrad52Δ strains (Fig. 3D), and the same was true for HWP1 (not shown). Expression of HGC1 has been reported to be also dependent on the cAMP/PKA signalling pathway and, accordingly, its mRNA was not produced in an efg1Δ null strain even in the presence of serum (Zheng et al., 2004; our own results). Furthermore, as shown in Fig. 3E, it was not expressed in the double mutant efg1Δrad52Δ. Therefore, the absence of Efg1p prevents expression of ECE1, HWP1 and HGC1 but not filamentation of rad52Δ strains, an indication that this morphogenetic event can occur in the absence of these HSG. In agreement with this, the double mutant hgc1Δrad52Δ filamented constitutively to the same extent as rad52Δ (Fig. 3C).
We have recently reported that a lig4Δ null mutant, which is defective in the non-homologous end-joining (NHEJ) recombination pathway, was slightly less mycelial than wild type (Andaluz et al., 2001). Accordingly, we tested the possibility that filamentation of rad52Δ was a consequence of an imbalance between both recombination pathways caused by deficiencies in HR. However, we could not find differences in morphology between rad52Δ and a double mutant rad52Δlig4Δ suggesting that filamentation in the former is not due to an exacerbation of the NHEJ pathway in the absence of HR (data not shown).
Filamentation of rad52Δ null cells is seriously compromised in cells lacking adenylate cyclase
Upstream components of the cAMP pathway, Ras1 and Cdc35, are also essential for yeast to hypha morphogenesis, but the loss of Cdc35 is more restrictive than the loss of Ras1 (Feng et al., 1999; Rocha et al., 2001). Furthermore, Bachewich et al. (2003) have shown that Cdc5-depleted and hydroxyurea-induced filaments could also form in the absence of Efg1 and Cph1, but required adenylyl cyclase signalling, as they were not formed in the absence of CDC35, the structural gene for the adenylate cyclase. In order to determine whether filamentation of rad52Δ strains showed a similar requirement, we investigated the phenotype of the double mutant cdc35Δrad52Δ. The mutant was constructed by deleting both copies of RAD52 in a cdc35 null strain, previously shown to lack both adenylyl cyclase and detectable levels of intracellular cAMP (Rocha et al., 2001). As shown in Fig. 4A (upper row), filamentation of rad52Δ cells occurred in the absence of Cdc35, but the number and length of the filaments were significantly reduced, whereas the number of yeast-like cells increased in parallel. Addition of dibutyryl-cAMP (10 mM) to cdc35Δrad52Δ cells improved filamentation to nearly the level of the rad52Δ strain. This was true on YEPD plates, especially in 48–72 h old colonies (Fig. 4A, lower row) as well as in liquid SC (Fig. 4B, lower row). As expected, cAMP did not influence the colony morphology of rad52Δ or cdc35Δ cells (Fig. 4A). Three additional cdc35Δrad52Δ independent double mutants behaved similarly (not shown).
rad52Δ cells exhibit a nuclear division defect and constitutive actin polarization
A large number of rad52Δ yeast cells (n = 200) in the stationary phase have small (14%) or large (43%) buds, suggesting that they are in the late S or G2/M phases of the cell cycle, whereas under the same conditions, a large percentage of wild-type cells (n = 213) are unbudded (86%). More important, DAPI staining of exponentially growing cultures indicated that whereas most wild-type (CAF2) yeast cells carrying medium size or big buds contain two nuclei (88%), one in the mother and the other in the daughter (Fig. 5A, c and d), a significant number of rad52 cells (from either null rad52Δ strain, TCR2.1 or TCR2.2) with large buds have only one nucleus (97%), either in the mother or in the daughter (54%), or, quite frequently, as a thread between mother and daughter cells (43%) (Fig. 5A, a and b;Table 2).
Table 2. Number and position of nuclei in G2/M cells from exponentially growing rad52Δ cultures.
One nucleus between both cells
SC5314 (wild type)
In addition to budded yeast and long filaments, a significant number of rad52Δ cells had elongated buds of variable sizes and shapes (Fig. 5B) that mimic incipient germ tubes. Interestingly, at a high frequency (≥ 70%) the undivided nucleus is located in the ‘germ tube’, in such a way that the rounded basal cell appears empty (Fig. 5B, a). Furthermore, in these cases, the marked constriction between both cells as well as the calcofluor staining pattern suggests that the septum is synthesized before nuclear division, preventing migration of one of the copies to the empty cell (Fig. 5B, a). Most (≥ 80%) of the elongated evaginations arising in rad52 cells had a tapered appearance suggesting that they were produced by G2/M cells. In some cases, it is likely that both mother and daughter cells carry one nucleus, but nuclear division is impaired in one of them (or later in the progeny) and that cell polarizes growth (Fig. 5B, b). Pseudohyphal elements, some of them significantly elongated, usually coexist with ‘germ tubes’, but again some cells in the chain lack nuclei (Fig. 5B, c).
The presence of more cell bodies than nuclei was also a constant feature of longer filaments (Fig. 5D, b and c), although filamentous cells with two nuclei were occasionally seen (Fig. 5D, a). Furthermore, not all the nuclei in the filaments showed the same fluorescence, suggesting that some of them do not contain a full set of genetic information (Fig. 5D, b, arrow). Again, about 70% of the basal cells were anucleated (Fig. 5D, a–c), and the marked constriction between the basal cell and the filament suggested that, in these cases, a regular septum had been formed. This was further confirmed by calcofluor white staining of this kind of cell (Fig. 5D, c). Long filaments formed axial buds, sometimes very elongated, at both poles, or even in the middle of the filament (Fig. 5C, a), and it was not infrequent to observe nuclei migrating into the buds (Fig. 5D, d). Although most long filaments exhibited features of pseudohyphae, some of them were as narrow as hyphae and, most importantly, did not show constrictions at the septa, suggesting that some true hyphae may be also present in rad52 cultures (Fig. 5C, b and c). In fact, after 24 h, liquid cultures showed some filaments as long as 100 µm and contained elongated cells 20 µm long. Also, some elongated cells in the filaments lyse, suggesting a weak cell wall (Fig. 5D, d). They probably correspond to anucleated filaments. As expected, the repressed conditional EAT4.1 showed the characteristics described for the null strain with regard to the distribution of nuclei and cell bodies (Fig. 5D, e).
Actin staining of wild-type yeast cells indicated the typical pattern (Anderson and Soll, 1986; Hazan et al., 2002), i.e. actin patches clustered at the tip in small or medium-sized buds, but dispersed over the surface in large buds, and forming a double ring at both sites of the septa (not shown). In elongated rad52Δ cells, a dense area of actin patches locates at the tip (Fig. 5E, a and b), and this was maintained in longer filaments (Fig. 5E, c and d). Still, as actin patches along the entire filament were frequently observed, it is feasible that some of them could be eventually recruited at the site of branching or budding.
These observations suggest that rad52Δ cells often fail to complete nuclear division, possibly as a consequence of the inefficient repair of spontaneous double-strand breaks (DSB), which triggers the DNA replication and/or DNA-damage checkpoint. This causes an uncoupling of other landmark events of the cell cycle, such as the periodic rearrangement of the actin cytoskeleton and alternative isotropic/polarized secretion of cell wall material that ultimately leads to filamentation.
Yeast cells present in rad52Δ cultures are still responsive to serum
Addition of 10% serum to rad52Δ cells yielded mycelial projections which were narrower than the spontaneous filaments seen in the absence of serum and had the appearance of true hyphae. However, filaments from mutant cells induced by serum exhibited a number of peculiarities when compared with those produced by wild-type cells under the same conditions. First, as expected from the cell cycle distribution of rad52Δ cells in liquid cultures, most cells (about 60% of the cells in the yeast form) formed a germ tube when they were in G2/M exhibiting the characteristic morphology shown in Fig. 6A, c–e. In contrast, in wild type, more than half of the germ tubes were produced by cells in G1 (Fig. 6A, a and b). In fact, as described before (Hazan et al., 2002), wild-type cells germinated regardless of the stage of their cell cycle, but G1 cells were more abundant in wild type than in the null mutant. Second, even after 2 h in serum, when almost all hyphae developed from wild-type cells had at least one septum (Fig. 6A, b), there were almost no septa in the serum-induced rad52 filaments, in spite of the fact that they had a similar length (Fig. 6A, d). This could reflect the longer generation time of rad52Δ cells. However, when a septum was formed it appeared regular (Fig. 6A, e). Third, before mitosis, the nucleus had migrated into the germ tube in both wild-type and rad52 cells (Fig. 6B). However, whereas in wild-type cells, one of the resulting nuclei from the nuclear division migrated back into the mother cell (Fig. 6B, a and b), in rad52 most basal cells were anucleated (Fig. 6B, c). Again, this is probably due to the fact that the chitin ring formed before nuclear division and/or its migration back into the mother cell. In fact, some nuclei seemed to be trapped in the septum (Fig. 6B, d). Other characteristics of the serum-induced rad52Δ filaments are similar to those described for wild-type cells (Hazan et al., 2002). For instance, no septum between the basal cell and the rest of the filament was observed when the germinating cell was in G1 (Fig. 6A, b). Also, when the basal cell was in G2/M, a chitin ring formed between mother and daughter cells (Fig. 6A, c–e), a situation similar to that described above for the spontaneous filamentation of rad52Δ. Finally, as expected (Anderson and Soll, 1986; Hazan et al., 2002), actin localized almost exclusively at the tip of the germ tube in both wild type (not shown) and rad52Δ cells (Fig. 6C, a and b). This situation is different from that observed during the spontaneous filamentation of rad52Δ cells where additional actin patches were also distributed along the filaments, suggesting the induction of a more specific and finely regulated hyperpolarization programme.
Incubation of rad52Δ cells in Lee's medium also induced germination. Again, many germ tubes arose from the buds of cells apparently arrested in G2/M and were shorter than their counterparts from parental CAF2 and revertant TCR3.2.1 which usually were formed from G1 cells (not shown).
Cell cycle in Rad52-depleted cells
The large number of filaments present in rad52Δ mutants makes elutriation difficult. Accordingly, we elutriated cells of the non-repressed conditional strain EAT4.1. Small unbudded EAT4.1 cells (G1) were collected and used to inoculate fresh SC medium supplemented with Met/Cys. As the absence of both amino acids increases the generation time of EAT4.1 (not shown), we used as a control the heterozygote RAD52/rad52 incubated in the presence of Met/Cys. The resulting synchronized cultures were monitored for 195 min (about two cell cycles). As shown inFig. 7A, downregulation of RAD52 significantly increases the length of the cell cycle, initially indicated by a delay in the completion of the first mitosis (transition of the G2/M cells to G1). This delay was exacerbated in the second cycle, suggesting the progressive induction of the DNA-damage checkpoint. As a consequence of this behaviour, the culture partially lost its synchrony and the proportion of G2/M cells never decreased below 25% of the total population, whereas in the control it reached values lower than 10%. In addition, during the second cycle, the transition of small buds to large buds was also significantly retarded, suggesting the additional presence of a replication block. Fluorescent-activated cell sorting (FACS) analysis of the same samples (DNA content; Fig. 7B) was consistent with these observations. The first DNA duplication started at about the same time in both cultures, but in the repressed strain, a delay in the transition from 4N (G2/M) to 2N (G1) occurred. For instance, at 105 min about 40% of the tetraploid cells in the control had returned to 2N, whereas in the repressed culture 90% still remained as 4N (G2/M). This delay not only subsisted but was accentuated in the second cycle. Thus, at 150 min, nearly 100% of the cells in the control were again 4N (G2/M), whereas in the repressed culture only 30% were 4N; it is likely that part of this 30% includes remnants of the G2/M cells of the first cycle that were arrested at that phase. Furthermore, as suggested by the budding pattern, a large percentage of the cells expressing the mutation could not finish the second cell cycle, as after 150 min transition from 2N to 4N occurred very slowly if at all in the mutant, whereas Rad52+ cells progressed at the same rate as in the first cycle (Fig. 7B). Determination of the position of the nuclei by DAPI staining in budded cells was also consistent with the above observations. Figure 7D illustrates the situation at 120 min. Most control cells had completed nuclear division and some had small buds, whereas a large percentage of repressed cells remained in G2/M with a single nucleus confined to the mother cell. As shown in Table 3, after 90 min the proportion of budded cells with a single nucleus remained very high in mutant cells whereas it fluctuated in Rad52+ cells as expected from a regular cell cycle. Concomitantly, the proportion of budded cells with two nuclei remained fairly constant in the mutant (likely because of a partial loss of synchrony) and yielded the expected pattern in the control, and the same was true for cells with one nucleus located in the neck. A similar picture was provided by the analysis of the cell size (FSC-H) (Fig. 7C). During the second cycle, a low percentage of mutant G1 cells shifted to the size expected for G2/M cells, whereas control cells behaved normally. In addition, at 195 min, a significant percentage of cells exhibited a significantly smaller size than the unbudded cells, probably representing anucleated cells, as they did not have a DNA signal. These results, together with our morphological observations, confirm that rad52 cells have aberrant DNA replication and likely induce the DNA-damage checkpoint. However, two cell cycles in the presence of Met/Cys were not enough to induce elongation of rad52 cells.
Table 3. Nuclei position in G2/M cells along the cell cycle progression in Rad52+ and Rad52– cells.a
One nucleus between both cells
For each sample, 200 budded cells (G2/M) were inspected and classified in the indicated category.
nd, not determined (G2/M cells were mostly absent).
Filamentation of rad52 mutants is partially dependent on Swe1
The observation that the G2/M arrest was an early phenotype derived from Rad52 depletion suggests that elongation could be a secondary event derived from the action of the cell cycle regulators involved in that arrest. In addition, as mentioned above, when induced by branched-chain alcohols, filamentation of S. cerevisiae has been shown unambiguously to be dependent on the cell cycle regulator Swe1, the inhibitory kinase of Clb2-Cdc28 (La Valle and Wittenberg, 2001; Martinez-Anaya et al., 2003). In order to determine whether a similar dependency occurred in rad52 mutants of C. albicans, we deleted both copies of RAD52 in a swe1 swe1 strain. In C. albicans, swe1Δ cells exhibit a 10% reduction in cell size, suggesting a role in the morphogenesis checkpoint, but otherwise form hyphae and pseudohyphe normally (Wightman et al., 2004). However, the pseudohyphal phenotype of cells deleted of the GIN4 kinase (which forms part of the intricate signalling network required for proper co-ordination of cell growth and cell division) was partially dependent on a Swe1-dependent checkpoint (Wightman et al., 2004). As shown inFig. 8, the double mutant swe1Δrad52Δ was still filamentous in both agar (Fig. 8A) and liquid media (Fig. 8B), but the number of filamentous cells was significantly reduced as compared with rad52, suggesting that elongation of rad52Δ cells is partially dependent on the action of Swe1. Similarly, the conditional strain TRS4.1 (MET3-RAD52/rad52Δswe1Δ/swe1Δ), in which the functional copy of RAD52 is under the MET3 promoter, was significantly less filamentous than EAT4.1 (MET3-RAD52/rad52Δ) under repressive conditions in either solid or liquid SC (Fig. 8C). As expected, this strain grew as yeast under non-repressive conditions, and the same was true for swe1Δ in liquid or solid YEPD (not shown). The reduction in filamentation caused by the absence of Swe1p in rad52 cells was also paralleled by a decrease in the invasiveness of agar medium, whereas the swe1Δ strain was non-invasive (Fig. 8D). Finally, the double mutant swe1 rad52 responded to serum forming true hypha as described for each of the single mutants (Wightman et al., 2004; this work) (not shown).
Rad52 depletion triggers the DNA-damage checkpoint in C. albicans
In this study, we have shown that rad52Δ mutants of C. albicans exhibit defects in growth and a marked filamentous morphology. Slow growth seems to be a common phenotype of rad52 mutants in other yeasts, such as S. pombe (Suto et al., 1999; van den Bosch et al., 2001). In S. cerevisiae, 13% of the rad52 cells were unable to form colonies (Toczyski et al., 1997; Melo et al., 2001). In C. albicans, it was not possible to determine this parameter as the filamentous morphology of rad52Δ complicates the determination of cfu. It is likely that the slower growth of rad52 mutants in yeast is due to their inability to repair spontaneous DSB. This results in an arrest in the cell cycle and, eventually, following adaptation, in spontaneous chromosome loss (Galgoczy and Toczyski, 2001). Under these conditions some cells may lose viability. We have shown that a significant percentage of the Rad52-depleted large budded cells of C. albicans do not progress normally through the next cell cycle but arrest at G2/M with a single presumably tetraploid nucleus, suggesting the presence of DNA-damage, and perhaps DNA-replication checkpoints. Given the role of Rad52 in DNA repair, these events are likely triggered by the accumulation of DSBs.
The absence of Rad52 induces polarization of growth accompanied by but independent of the expression of several HSG
The second relevant phenotype of rad52 strains of C. albicans is the formation of filaments in media that usually do not induce hyphal growth, such as YEPD and SC, a behaviour that has not been reported in other yeast. The rad52Δ filaments appears intermediate between pseudohyphae and hyphae, as although they are in general shorter and wider than true hyphae; some of them become thinner upon elongation and are in fact indistinguishable from hyphal elements. The latter phenotype is one of the several criteria proposed for distinguishing hyphal cells by Sudbery et al. (2004). Interestingly, filamentation of rad52Δ cells was also accompanied by derepression of HSG, including ECE1, HWP1 and HGC1. Two of these genes, HWP1 and ECE1, were also significantly derepressed in cells depleted of the polo-like kinase Cdc5 (which also develops highly polarized filaments), at least during the later stages of cell elongation (Bachewich et al., 2003; 2005). However, filamentation occurred also in both efg1Δrad52Δ and cph1Δefg1Δrad52Δ mutants, indicating that the process was independent of the MAP kinase and cAMP/PKA signalling pathways that include these transcription factors. As deletion of EFG1 prevents expression of ECE1, HWP1 and HGC1, we conclude that expression of the HSG accompanies but is dispensable for filamentation of rad52Δ cells. In agreement with this, the double mutant hgc1Δrad52Δ was also filamentous. It follows that the actin polarization signal activated by the Rad52 defect is either independent or downstream of the execution points of Ece1, Hwp1 and Hgc1 and, accordingly, bypasses the possible contribution of these factors to cell elongation. However, the rad52Δ-induced signal somehow activates the cAMP/PKA pathway and, consequently, the Efg1-dependent HSGs, accentuating the similarities between the rad52Δ filamentation and the hyphal programme (Fig. 9). A detailed comparison of the filamentation patterns yielded by rad52Δ and efg1Δrad52Δ strains will reveal the contribution of the Efg1-controlled HSG to morphogenesis. On the other hand, filamentation was severely impaired on M199 and Spider agar in the rad52Δ strain, which was still responsive to serum or Lee's medium at 37°C. These observations emphasize the difficulties inherent to the establishment of cultural, morphological and transcriptional parameters that define the different forms that C. albicans may adopt. In fact, the majority of the HSG seem to be also induced in pseudohyphal cells (Kadosh and Johnson, 2005).
Efg1 only controls a limited subset of all the genes controlled by the cAMP/PKA pathway, in particular those activated in the yeast-to-hyphal transition. Harcus et al. (2004) have shown that loss of upstream components of the pathway, in particular the adenylate cyclase, influences expression of, in addition to those controlled by Efg1, a much larger number of genes, including some known as well as putative transcription factors. The reduction in the number and length of filamentous cells in the cdc35Δrad52Δ strain indicates that an adequate expression level of some of those genes may facilitate the initiation and maintenance of the polarized growth in rad52Δ cells.
Finally, the observation that filamentation was significantly more prominent in solid media than in liquid cultures suggests that the agar-induced filamentation pathway may be activated in rad52 strains. It has been recently reported that contact activation of a kinase (Mkc1p) signals invasive hyphal growth in C. albicans (Kumamoto, 2005). However, the fact that disruption of both copies of either CDC35 or SWE1 reduced significantly the number and length of rad52Δ filaments, even on solid agar, suggests that only part of the elongation would be dependent on the contact activation mechanism, unless both Cdc35 and Swe1 also play some role in this process. Again, a detailed analysis of the nature of the filaments formed by the double mutants cdc35Δrad52Δ and swe1Δrad52Δ appears necessary.
Filamentation of rad52Δcells is likely derived from the DNA-damage checkpoint rather than from secondary mutations
How is the absence of Rad52 transduced to the growth polarization machinery in C. albicans? One possibility is that the latter derives from the genetic instability of rad52Ä cells. Many of the phenotypes exhibited by rad52 mutants of S. cerevisiae, including chromosome loss or an increased frequency of point mutations (Mortimer et al., 1981; Yoshida et al., 2003), probably are derived from the inability of these mutants to repair some form of intrinsic DNA damage (Paques and Haber, 1999; Galgoczy and Toczyski, 2001). In fact, nearly 60% of the S. cerevisiae rad52 cells, versus 8% of the wild-type cells, showed fluorescent Ddc1-GFP foci which are indicative of DNA damage (Melo et al., 2001). We have evidence that rad52 mutants of C. albicans lose chromosomes at an increased rate (our unpublished results), and this may result in new phenotypes, given the high degree of heterozygosity of this organism (Janbon et al., 1998; Forche et al., 2004; Jones et al., 2004). Another cause of genetic alteration is the introduction of permanent genetic changes as a consequence of defective repair using alternative recombination pathways. However, the early filamentation of many cells from repressed cultures of the conditional strain and the fact that many cells of a rad52 culture are able to form filaments suggest that the reason for filamentation is not a consequence of secondary mutations, unless a particular open reading frame (ORF) related to that process behaves as a hot spot.
A second, more attractive possibility is that growth polarization of rad52Δ cells is derived from the response of the cells to DNA damage, which, in turn, triggers the checkpoints that precede filamentation. In S. cerevisiae, cells with an unrepaired DSB arrest in anaphase with a single nucleus, but, unlike C. albicans, no polarization of actin has been reported. However, after 8–12 h S. cerevisiae-arrested cells resume progression through the cell cycle, a process termed adaptation (Sandell and Zakian, 1993; Toczyski et al., 1997; Lee et al., 1998). An early landmark event of C. albicans rad52Δ filamentation is the uncoupling between actin polarization and nuclear division; this uncoupling is maintained for some time, as filaments with a single nucleus continue to enlarge and even give rise to new cell compartments in the absence of nuclear division. We have not investigated the existence of a phenomenon similar to adaptation in C. albicans, but it is clear that, after some time, at least some nuclei of the elongated, uninucleated cells undergo karyokinesis. However, nuclear division remains uncoupled from other events of the cell cycle, such as the rearrangement of actin cytoskeleton and formation of a new cell, as it is possible to observe single-celled filaments with two nuclei, empty axial buds and nuclei entering big buds. A likely possibility is that these uncoordinated events are symptoms of apoptosis, a phenomenon that can be triggered by the arrest in cell division that follows a DNA-damage response (Kastan and Bartek, 2004). In fact, the frequency of dead of filaments supports this possibility.
Is there a pathway connecting DNA damage with filamentation in C. albicans?
Hyperpolarization of C. albicans growth in response to DNA damage has also been observed in the presence of hydroxyurea (HU), an inhibitor of ribonucleotide reductase that inhibits DNA replication and arrests (400 mM) or delays (200 mM) S. cerevisiae cells in S phase (Clarke et al., 1999; Jiang and Kang, 2003). In the presence of 200 mM HU (a concentration that does not arrest completely but slows down DNA replication), each C. albicans cell forms a single uninucleated filament lacking septa or buds and does not proceed further (Bai et al., 2002; Bachewich et al., 2003; our own observations). rad52 filaments also mimic those formed by C. albicans cells depleted of the polo-like kinase CDC5, which in C. albicans is required for spindle elongation. These observations suggest that a common mechanism may be responsible for the polarized growth under several conditions. Interestingly, Cdc5 is involved in the hyperphosphorylation and subsequent degradation of Swe1 in S. cerevisiae (Asano et al., 2005; Harvey et al., 2005; Lee et al., 2005) (see Introduction). If this role was maintained in C. albicans, its depletion would result therefore in the stabilization of Swe1 and the maintenance of the subsequent Swe1-dependent inhibition of Clb-Cdc28 (see also below). However, once again and in contrast to rad52 mutants, filaments induced by the depletion of Cdc5 did not show buds nor septa (Bachewich et al., 2003), suggesting that additional specific mechanisms operate under each condition.
In different organisms, DNA damage is detected by sensor kinases (Rad17, Rad24, Mec1/Rad3/ATR and Tel1/ATM) and then transmitted to effector kinases (Chk1 and Rad53/Cds1) (Melo and Toczyski, 2002) that activate transcription of the DNA repair regulon and connect with components of the cell cycle, halting transiently or delaying its progression. On the other hand, recent results have also shown that filamentous differentiation in S. cerevisiae occurs in response to slow DNA synthesis. This process involves Mec1 and Rad53 kinases, as well as the cell cycle regulators Cdc28 and its inhibitor Swe1 (Jiang and Kang, 2003). Interestingly, Cdc28 and Swe1 are also involved in the filamentous differentiation of S. cerevisiae triggered by nitrogen starvation or short-chain alcohols (Liu et al., 1993; Lorenz and Heitman, 1997; Lorenz et al., 2000; Martinez-Anaya et al., 2003). Our finding that deletion of Swe1 decreases significantly the ability of rad52 cells of C. albicans to form filaments suggests that the Swe1-mediated inhibition of Cdk1 (Cdc28) that mediates the morphogenesis checkpoint is also partially responsible for the filamentation process (Fig. 9). Although inactivation of Cdc28 in C. albicans does not seem to play any role in hyphal growth in response to serum (Hazan et al., 2002), Cdc28 could be involved in other kinds of processes requiring cell polarization, such as rad52 filamentation. For instance, it has been reported that Cdc28 functionally interacts with Hgc1 (Zheng et al., 2004). A signal could occur triggered by the Rad52 defect could relieve the Cdc28 function on growth polarization from the Hgc1 interaction. Residual filamentation of swe1 rad52 mutants may result from alternative mechanisms derived from the G2/M arrest. For instance, in S. cerevisiae the DNA damage and replication checkpoints may also block mitosis independently of the inhibitory phosphorylation of Cdc28-Y19 by Swe1 (Amon et al., 1992; Sorger and Murray, 1992). In the absence of Swe1, the induction of HSG by a component of the putative pathway (marked with an asterisk in Fig. 9) could promote and maintain some polarization. As mentioned above, a number of recent reports have indicated that deletion of components that regulate cell cycle progression or induction of cell cycle arrest during S or M phases results in pseudohyphal or even hyphal growth (see Introduction) as a consequence of the persistent maintenance of cell polarity. Our results support this concept and further indicate that a disturbance in the DNA-damage repair machinery may cause a similar effect through the G2/M arrest. Furthermore, they also suggest that cAMP signalling facilitates both the induction and the maintenance of growth polarization. The elucidation of the possible connection between the Rad52 depletion and the activation of a putative C. albicans counterpart of the central core, shown to be involved in the filamentous differentiation of S. cerevisiae caused by slow DNA synthesis (Rad53, Swe1, Cdc28), is one of our present objectives.
Strains and media
The C. albicans strains used in this study are listed in Table 4. C. albicans SC5314 is a prototrophic strain (Gillum et al., 1984). Therefore, its derivative CAF2-1 (URA3/ura3Δ) is used as a control for strains carrying one copy of URA3 (Fonzi and Irwin, 1993). C. albicans cells were grown routinely at 30°C in YEPD (2% glucose, 1% yeast extract, 2% Bacto Peptone), or SC medium [0.67% nitrogen base (Difco), 2% glucose and a mixture of amino acids] supplemented (Uri–) or not (Uri+) with Uri (25 µg ml–1). Spider and M-199 (Gibco-BRL, adjusted to pH 7.5) media were prepared according to Gimeno et al. (1992) and Ramon et al. (1999) respectively. Lee's medium was prepared as described (Lee et al., 1975). To induce germination, cells of C. albicans were transferred to YEPD pre-warmed at 37°C and supplemented with 10% serum. To study the cell morphology of the several strains, frozen cells (–70°C) were streaked on YEPD agar plates and grown for 48 h at 28°C. The cell mass was then suspended in water and used to inoculate a liquid YEPD culture that was incubated for 24 h at 28°C. These cultures were then used to inoculate agar plates or liquid media.
Sequential disruption of both alleles of RAD52 in the indicated genetic background (efg1Δ/efg1Δ, efg1Δ/efg1Δcph1Δ/cph1Δ, hgc1Δ/hgc1Δ, swe1Δ/swe1Δ and cdc35Δ/cdc35Δ) was performed as described before (Ciudad et al., 2004) using the URA blaster method (Fonzi and Irwin, 1993). All the strains were verified using Southern blot hybridization. To generate the CaMET3-RAD52 cassette, a 320 bp fragment, comprising positions –39 to +281 in relation to the first nucleotide of the CaRAD52 ORF (Ciudad et al., 2004), was amplified by PCR using the appropriate oligonucleotides (R1, 5′-CGCGGATCCCCACCCACACTTAAAATACACAG-3′, which includes a restriction site for BamH1, and R2, 5′-AACTG CAGCGAGTTCCACCTGGACCATCACG-3′, which includes a restriction site for PstI. The PCR product was cloned in plasmid MET3-URA3 (pCaDis) (Care et al., 1999) following digestion both enzymes (BamH1 and Pst1) which are also present in the polylinker of the vector. The construct was digested with SacI and used to transform the heterozygous RAD52/rad52 Ura– (TCR1.1). Transformants carrying a conditional allele were analysed by PCR using oligonucleotides MET-R3, 5′-CGACATAGATCCATTAATGCGCC-3′, located in the MET3 promoter, and RV4, 5′-CCGCTACCACCATGAG TATCT-3′, internal to RAD52 (Ciudad et al., 2004), whose PCR product should correspond to a 0.9 kb fragment. Positive transformants were then analysed for the null allele by PCR, using primers flanking the deleted region of CaRAD52 (RV1, 5′-CAACCACGACCACCACAACAA-3′, complementary to positions +25 to +45, and RV2, 5′-TGCGGTATAC CCGAAGAAGGA-3′, complementary to positions 1587 to 1607 on the complementary strand). As shown before, wild-type and disrupted alleles should yield 1.6 kb and 1.4 kb fragments respectively (Ciudad et al., 2004). Following this approach, a conditional strain rad52Δ/MET3-RAD52 (EAT4.1) yielding the expected fragments was selected and characterized. Using the same approach, we constructed an additional conditional strain in the swe1Δ null background (TSR4.1, swe1Δ/swe1Δrad52Δ/MET3-RAD52) (Table 4). Both conditional strains were routinely grown in SC medium lacking Met, Cys and Uri. To repress RAD52 expression, agar plates were supplemented with 2.5 mM Met and 2.5 mM Cys. Similarly, liquid cultures were supplemented with 2.5 mM Met and 2.5 mM Cys and inoculated with exponentially growing cells (OD600 lower than 2) to reach an initial OD600 of about 0.01.
Liquid cultures were started by inoculation of a 250 ml flask containing 50 ml of YEPD or SC with a suspension of cells previously grown in the same medium to reach a final OD600 = 0.05. Samples were taken at the indicated times, and the OD600 was measured in a spectrophotometer. Appropriate dilutions of each sample were plated in duplicate to YEPD or SC plates, respectively, to determine the number of cfu. UV and MMS treatment was carried out as described (Ciudad et al., 2004).
Nucleic acids extraction and analysis
Standard techniques were routinely used for DNA preparation, cell transformation and Southern blot hybridization (Andaluz et al., 2001; Ciudad et al., 2004). To analyse the expression of filamentation genes, overnight cultures were used to inoculate YEPD and YEPD supplemented with 10% bovine serum at a final OD600 = 0.5, and these cultures were incubated for 1.5 h at 30°C and 37°C, respectively, in a rotatory shaker. To analyse the expression of RAD52 in the conditional strain EAT4.1, cells were inoculated into 50 ml of fresh SC with/without Cys/Met at an OD = 0.5. At the indicated times a volume equivalent to 8 × 108 cells was used for RNA preparations. RNA extraction and Northern analysis have also been described (Ciudad et al., 2004). Probes were generated by PCR from genomic DNA using the following oligonucleotides. For ECE1, F: 5′-TGGCAACATTCCACAAG TAATC-3′ and R: 5′-AGC CGGCATCTCTTTTAACTGG-3′. For HWP, F: 5′-TGCTC CAGGTACTGAATCCGC-3′ and R: 5′-GGCAGATGGTTG CATGAGTGG-3′. For HGC1, F: 5′-CCAACAACAACCCCC AAGCTTCTGGC-3′ and R: 5′-GCCAGTAGAACTAGTTGGT GTAGTAC-3′. For ACT1, F: 5′-GCTGCTTTAGTTATCGAT AACGG-3′ and R: 5′-TGAACAAAGCTTCTGGAGCTCTG-3′. These primers amplify fragments of 482 bp, 223 bp, 1198 bp and 772 bp, respectively, which were cloned in the pGEM-7Zf(+) vector. For determination of RAD52 expression in the conditional strain EAT4.1, we used a 1453 bp SacI–SpeI fragment internal to the RAD52 ORF as a probe (Ciudad et al., 2004). Hybridization bands were visualized using a Molecular Imager (Bio-Rad Laboratories). RT-PCR was performed according to the manufacturer (Promega). RNA extracts (as above) were incubated with DNase and then subjected to amplification in the presence of transcriptase and polymerase or only polymerase to demonstrate the absence of any contaminating DNA. For amplification of RAD52 mRNA in the conditional strain EAT4.1, we used the following oligonucleotides: F: 5′-CGT GATGGTCCAGGAACT-3′ and R: 5′-TGAGCAACAATGGTC TTGTCG-3′ that amplify a 365 bp fragment internal to the RAD52 ORF.
Elutriation and cell cycle analysis
A 3 l flask containing 1.5 l of SC medium lacking both Cys and Met was inoculated with the conditional mutant EAT4.1. Following an overnight incubation in a rotatory shaker, cells were harvested at an OD600 = 1 and subjected to elutriation using a Beckman centrifuge. About 600 ml of unbudded cells at a OD600 = 0.2 was collected and split into two cultures of 300 ml, one of which was made 2.5 mM Met and Cys. These cultures were maintained for 30 min in a rotatory shaker to relieve cells from the elutriation stress and adapt them to the new conditions. Samples were then taken at the indicated times to determine cell cycle progression (relative proportion of unbudded cells, cells with small buds and cells with big buds). Cell size and DNA content were determined by FACS using a FacsScan (Becton-Dickinson, Franklin Lakes, NJ) flow cytometer.
Cell staining and microscopy
The cell morphology of fresh or fixed cells was inspected microscopically using a Nikon Eclipse 600 fluorescence microscope with a 60x DIC objective. A CC-12 digital camera interfaced with Soft Imaging System software was used for imaging. When cells were fixed (70% ethanol), they were re-hydrated and sonicated to disperse clumps before microscopy. For fluorescence microscopy 108 cells were fixed in 70% of ethanol and supplemented with 5 µl of a DAPI solution (1 mg ml–1). After 5 min, cells were washed several times in water and resuspended to 106 cells ml–1 for microscopy. Calcofluor staining of the chitin ring and rhodamine-phalloidin were performed as described (Amberg, 1998; Loeb et al., 1999). Colonial morphology on agar plates was followed in a Nikon microscope (× 10).
This study was supported by a grant from the Public Heath Service, NIH-NIAID 1 R01 AI51949, to R.C. and G.L. We thank the reviewers for their comments. We also thank W. Fonzi for critical reading of the manuscript, J. Correa for his advice in cell elutriation, M.C. López-Cuesta for her help in the FACS analysis, J. Ernst for strains efg1Δ, cph1Δ and efg1Δcph1Δ, Y. Wang, P. Sudbery, and D. Harcus and M. Whiteway for the hgc1Δ, swe1Δ and cdc35Δ strains, respectively, and Belén Hermosa for her careful technical support. J.G.-R. is the recipient of a fellowship from the NIH-NIAID grant and Belén Hermosa is supported by contracts from a Junta de Extremadura grant, 2PR03A044, to E.A. and the NIH-NIAID grant.