In the present work, we have investigated the function of DNA repair genes from C. albicans and the roles these genes play in genome stability and acquisition of drug resistance. We have shown that the HR mutants (mre11Δ/mre11Δ, rad50Δ/rad50Δ, and rad52Δ/rad52Δ) are very susceptible to several DNA-damaging agents, whereas the msh2Δ/msh2Δ, pms1Δ/pms1Δ, and yku80Δ/yku80Δ mutants are not. We also observed that the mre11Δ/mre11Δ, rad50Δ/rad50Δ, and rad52Δ/rad52Δ mutants are slow growing and produce wrinkled colonies with pseudohypha-like cells in YEPD+Uri at 30°C. We observed increased genome instability in the mre11Δ/mre11Δ and rad50Δ/rad50Δ mutants with an assay for chromosome 1 integrity. Surprisingly, deletion of some of the DSBR genes leads to an increased susceptibility to some antifungal drugs. We also observed an elevated frequency of appearance of drug-resistant colonies of the msh2Δ/msh2Δ, pms1Δ/pms1Δ, and rad50Δ/rad50Δ mutants inside the inhibition ellipse.
MSH2, PMS1, MRE11, RAD50, YKU80, and RAD52 are not essential genes in C. albicans, as we were able to construct null mutants for all of them. Nevertheless, RAD50 nulls were more difficult to obtain. This result is not likely to be due to the chromosomal location of RAD50; the ORF is not located near the centromere, telomeres, or one of the highly repetitive MRS elements, locations where gene disruptions can be more difficult to achieve. The difficulty in targeting the RAD50 sequence may be due to a local chromatin configuration that makes the sequence less accessible, to a specific function of the RAD50 protein required for HR, or to the oligonucleotide sequences.
As shown in Fig. and , mre11
Δ, and rad52
Δ are affected by oxidizing agents, UV, camptothecin, EMS, and MMS, whereas the msh2
Δ, and yku80
Δ mutants are not. Previous studies have shown that S. cerevisiae
utilizes HR over NHEJ, while in mammals the reverse is found (16
). We observed that mutations in genes involved in the HR pathway (MRE11, RAD50
, and RAD52
) affect the sensitivity of the mutants to DNA break-inducing compounds, while mutations in YKU80
(required for NHEJ) do not. From these data, we conclude that C. albicans
preferentially uses HR to repair DNA breaks, in agreement with the work done by Larriba et al. (2
) on RAD52
in C. albicans
We find that the mre11
Δ, and rad52
Δ mutants exhibit a slow-growth phenotype. In other organisms, various types of DNA damage activate specific cell cycle checkpoints that result in arrested cell cycle progression, providing more time for repair. Because the main DNA repair proteins are absent in these mutants, cells arrest for an even longer time to allow efficient DNA repair. This hypothesis has been confirmed in the rad52
Δ mutants by the Larriba group (3
), who showed that depletion of RAD52
in C. albicans
activates the DNA damage checkpoint and that cell cycle arrest generates a polarized-growth phenotype. As their rad52
Δ phenotype is comparable to the phenotypes we observed in our mre11
Δ, and rad52
Δ mutants, it is likely that the cell cycle checkpoint arrest hypothesis is true for RAD50
as well. Persistent DNA lesions in the absence of Mre11p, Rad50p, or Rad52p may trigger DNA checkpoints that result in changes in cell morphology. Work by other research groups has demonstrated that DNA checkpoint proteins are involved in morphological changes in response to a variety of DNA-damaging agents in C. albicans
and other fungi (7
A previous study (2
) showed that mutations in LIG4
, a gene shown to be involved in NHEJ in S. cerevisiae
, impair myceliation in C. albicans
. In our work, yku80
Δ NHEJ mutants did not show any defect in filament formation in response to serum and in Spider medium at 37°C. This observation suggests that the myceliation defect observed in lig4
Δ by Andaluz et al. is not the result of a defective NHEJ apparatus but rather may be due to a secondary function of the Lig4 protein in one of the signaling pathways controlling myceliation.
By using a GAL1/URA3 system on chromosome 1, we showed that deletion of MRE11 and RAD50 gives rise to an increased frequency of 2-DGr or 5-FOAr colonies compared to the parental strain when cells are grown on min-2-DG and min-5-FOA medium. This result indicates that the mre11Δ/mre11Δ and rad50Δ/rad50Δ mutants are more likely to lose GAL1 or URA3 function. The loss of GAL1 or URA3 function could be due to a point mutation in the ORF that would produce a nonfunctional protein or to the loss of the GAL1 or URA3 gene through gene conversion, BIR, reciprocal crossover, or segmental or total chromosome loss. To determine the relative frequencies of these events, we screened 20 2-DGr and 20 5-FOAr colonies of the parental, mre11Δ/mre11Δ, rad50Δ/rad50Δ, and rad52Δ/rad52Δ strains by PCR to detect the presence of the GAL1 and URA3 sequences. We showed that the vast majority of the cells lost the GAL1 or URA3 gene. SNPs located on both sides of GAL1/URA3 were then used to distinguish among gene conversion, BIR, reciprocal crossovers, and segmental or total chromosome loss. If the flanking SNPs remained heterozygous in the 2-DGr or 5-FOAr strains, this would indicate that the cells lost the GAL1 or URA3 function by localized gene conversion. If the SNPs became homozygous, this would suggest that the cells underwent full-length chromosome loss. If one of the SNPs is still heterozygous while the other becomes homozygous, this would suggest that a segmental chromosome loss or BIR event took place. We observed that the majority of the 2-DGr and 5-FOAr colonies of the parental, mre11Δ/mre11Δ, rad50Δ/rad50Δ, and rad52Δ/rad52Δ strains lost GAL1 or URA3 function as a result of LOH events. Because the GAL1/URA3 locus is located 450 kb away from the telomere, it is more likely that LOH results from BIR, in which one chromosomal arm is duplicated by using the homolog as a template, rather than segmental aneuploidy. The determination of the breakage point in all of these strains is ongoing work; the data will tell us whether there is a weak spot on chromosome 1 where chromosome breaks are favored in response to stress.
When we investigated the parental strain and the other mutants, we observed the same LOH mechanism for the appearance of 2-DGr and 5-FOAr colonies. On the basis of these results, we conclude that the spectrum of alterations on chromosome 1 is unchanged between the mutants and the parental strain, but the frequency of events is greatly increased in the mre11Δ/mre11Δ and rad50Δ/rad50Δ mutant strains. Thus, the genome alterations observed on chromosome 1 with the GAL1/URA3 assay system in wild-type cells are likely to arise from a complete failure of the DSBR pathway.
Interestingly, we saw an increase in the proportion of colonies that accumulated point mutations in the GAL1 or URA3 gene for the msh2Δ/msh2Δ, pms1Δ/pms1Δ, and rad52Δ/rad52Δ mutants. After gene sequencing, we showed that the majority of the mutations were located within repetitive DNA tracts. This type of repetitive-tract instability is a known phenotype of the msh2Δ/msh2Δ and pms1Δ/pms1Δ mutants, as the MMR pathway has been shown to be involved in maintaining the stability of DNA repetitive tracts. However, DNA repetitive-tract instability is a novel phenotype of rad52Δ/rad52Δ mutants.
Loss of MMR (msh2
Δ and pms1
Δ mutants) or loss of DSBR (rad50
Δ) leads to an increase in fluconazole-resistant colonies, as shown in Fig. . Colonies of these strains appear within the inhibition ellipse of the E-test strip, whereas the wild-type parental strain and strains bearing homozygous deletions of other DNA repair genes do not exhibit this phenotype. Upon retesting, the resistant isolates exhibit various degrees of antifungal drug resistance. In other organisms, loss of MMR leads to a mutator phenotype; an increase in resistant isolates of the msh2
Δ and pms1
Δ strains would be expected in these C. albicans
mutants. The appearance of colonies within E-test strip inhibition ellipses has been described in C. albicans
previously; a heterogeneous population gave rise to resistant isolates, although the isolates were only resistant transiently, possibly because of an epigenetic change in the expression of drug efflux pumps (19
The increase in resistant isolates of the mre11Δ/mre11Δ and rad50Δ/rad50Δ strains, but not of the rad52Δ/rad52Δ or yku80Δ/yku80Δ disruptant, indicates that a complete loss of DSBR is required for an increase in the frequency of antifungal drug resistance—loss of either HR or NHEJ alone is not sufficient. With regard to RAD52, this result is surprising, given the colony morphology, DNA damage sensitivity, and antifungal drug sensitivity phenotypes associated with the loss of RAD52. Examination of antifungal drug resistance development in other HR-specific gene disruptions might address this issue. If this is the case, compounds that specifically inhibit RAD52, or possibly HR, may be effective as companion drugs during treatment for Candida infections, as they would increase susceptibility to the antifungal drug without increasing the frequency of appearance of antifungal drug-resistant cells.
Our drug studies link genome instability to acquisition of drug resistance, as we have shown that cells that are defective in DSBR are generally more sensitive to the antifungal agent fluconazole but are more likely to give rise to a subpopulation of cells that have acquired resistance to fluconazole. Another link between antifungal drug resistance and genome instability has been identified by the Berman group (25
). They demonstrated that Candida
cells that become resistant to antifungal drugs can harbor an isochromosomal derivative of chromosome 5. Isochromosomes are chromosomal variants in which both arms of the chromosome are identical. Such derivatives may arise by aberrant HR events between sister chromatids during DSBR. Further investigation of isochromosome formation in DNA repair mutant strains might provide insight into the mechanisms of isochromosome formation. Finally, a large proportion of patients infected with C. albicans
are cancer treatment patients. Cancer patients are often treated with topoisomerase inhibitors, which act against rapidly dividing cells. Fluconazole treatment of a C. albicans
infection in a cancer patient undergoing therapy with topoisomerase inhibitors could favor the appearance of drug-resistant C. albicans
cells, as inhibition of topoisomerases can lead to an increase in recombination.
Genome plasticity is a hallmark of C. albicans and is believed to generate diversity in an organism that propagates by clonal mitotic division, as Candida has not been demonstrated to undergo meiosis. Our data demonstrate that DNA repair pathways, the acquisition of drug resistance, and genome plasticity are linked. However, it is still unclear how C. albicans tolerates such drastic genome changes (isochromosome formation, aneuploidy, a high level of heterozygosity) and if these changes illustrate a global adaptive response of C. albicans to the various stresses the fungus encounters during the course of infection.