Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Although increases in chromosome copy number typically have devastating developmental consequences in mammals, fungal cells such as Saccharomyces cerevisiae seem to tolerate trisomies without obvious impairment of growth. Here, we demonstrate that two commonly used laboratory strains of the yeast Candida albicans, CAI-4 and SGY-243, can carry three copies of chromosome 1. Although the trisomic strains grow well in the laboratory, Ura+ derivatives of CAI-4, carrying three copies of chromosome 1, are avirulent in the intravenously inoculated mouse model, unlike closely related strains carrying two copies of chromosome 1. Furthermore, changes in chromosome copy number occur during growth in an animal host and during growth in the presence of growth-inhibiting drugs. These results suggest that chromosome copy number variation provides a mechanism for genetic variation in this asexual organism.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Genomic stability is of great importance for the normal growth of organisms. In abnormal states such as cancer, losses or gains of whole chromosomes, chromosome translocations or gene amplifications are frequently observed (Lengauer et al., 1998). During development, the consequences of gaining or losing a chromosome are often severe. For example, in humans, only certain trisomics are viable, e.g. XXY (Klinefelter syndrome); XYY (XYY syndrome); trisomy 21 (Down syndrome); trisomy 13 (Patau syndrome); and trisomy 18 (Edwards syndrome). Although these trisomics are viable, profound physical abnormalities and mental retardation are commonly found in these individuals (Nicolaidis and Petersen, 1998; Griffiths et al., 2002). It is believed that altered gene dosage, caused by the extra copy of a chromosome, leads to altered gene expression and suboptimal cellular physiology.
In the model eukaryote Saccharomyces cerevisiae, changes in chromosome copy number have also been observed. In a study of 290 S. cerevisiae deletion mutants, Hughes et al. (2000) found that ≈ 8% of the deletion mutants (22/290) were aneuploid for whole chromosomes or chromosomal segments. Interestingly, five of the aneuploid mutants appeared to have gained a copy of a chromosome encoding a close paralogue of the gene that had been deleted, suggesting that fitness of the mutant strain was improved by increasing the copy number of a closely related gene.
Previous reports have described changes in the copy number of C. albicans chromosomes as a response to selective conditions. For example, when an ADE2/ade2 heterozygous strain was treated with the antimitotic agent methyl benzimidazole carbamate, adenine-requiring strains that appeared to have lost the homologue of chromosome 3 carrying ADE2 were observed (Barton and Gull, 1992). These strains grew very poorly and frequently gave rise to healthy derivatives that appeared to have duplicated the remaining copy of chromosome 3, regenerating the normal copy number.
Janbon et al. (1998) found that cells plated on a normally non-utilizable sugar, l-sorbose, gave rise to colonies that lacked one homologue of chromosome 5. These Sou+ colonies grew poorly and gave rise to Sou– revertant colonies that grew well and were disomic for chromosome 5. Although the SOU1 gene responsible for the utilization of l-sorbose is not located on chromosome 5, its expression is controlled by the copy number of chromosome 5.
Strains of C. albicans (Perepnikhatka et al., 1999) or Candida glabrata (Marichal et al., 1997), another pathogenic yeast, that are resistant to the antifungal drug fluconazole were found to have altered chromosome copy number. In the case of C. glabrata, the chromosome carrying CYP51, the target of fluconazole, was duplicated in a FlcR strain, and the duplication was lost in FlcS revertant isolates. It has been suggested that changes in the copy number of chromosomes are an important source of genetic variability, allowing the evolution of organisms with new properties such as the ability to use sorbose or grow in the presence of fluconazole (Suzuki et al., 1989; Rustchenko-Bulgac et al., 1990; Rustchenko et al., 1997).
In this paper, we demonstrate that, unexpectedly, certain strains of C. albicans, including the most commonly used laboratory strain, CAI-4, carry three copies of chromosome 1. Although the trisomic strain CAI-4 grew well under laboratory conditions, the trisomic strains were avirulent in an intravenously inoculated mouse model, suggesting that, within a host, their growth was compromised. We also observed that, under non-stress conditions, the copy number of chromosome 1 was stable but, during growth in the presence of selective agents, strains with altered numbers of chromosome 1 were obtained. This is the first report of an effect of a specific chromosomal change on the fitness of a C. albicans strain during growth within a host.
Some stocks of strain CAI-4 contain three copies of genes found on chromosome 1
During the course of genetic studies on a gene termed YAL36, which encodes a GTPase, we demonstrated that strain CAI-4 contains three copies of YAL36. Using an NsiI polymorphism in the YAL36 region, the three copies of YAL36 were detected in two ways. First, when chromosomal DNA from strain CAI-4 was analysed by Southern blotting, one allele of YAL36 yielded a 4.5 kb band, whereas the other allele of YAL36 yielded a 7 kb band as a result of the lack of an NsiI restriction site (Fig. 1B, lane a). Quantification of the intensities of these two bands, as described in Experimental procedures, showed that the ratio of the 7 kb band to the 4.5 kb band was 1.8:1, not 1:1 as would be expected if there were one copy of each allele. The observed ratio, which is close to 2:1, suggested that this stock of strain CAI-4 carried three copies of the YAL36 gene.
To demonstrate the presence of three copies of YAL36 in a second way, strain CAI-4 was transformed with a construct that would alter the NsiI restriction map of YAL36 by replacing the YAL36 coding sequence with other DNA sequences. As illustrated in Fig. 1A, in some transformants, the transforming DNA recombined with the 4.5 kb allele of YAL36, shifting the 4.5 kb band to 5.9 kb (Fig. 1B, lane b). The ratio of the 7 kb to 5.9 kb bands in this strain was 1.9:1. In other transformants, the transforming DNA recombined with one of the 7 kb alleles, and this 7 kb band shifted to 5.9 kb. In such strains, three distinct bands corresponding to the 7 kb, 5.9 kb and 4.5 kb alleles were observed, and the ratios of the three bands were 1:1.0:1.1 respectively (Fig. 1B, lane c). The observation of three bands in some transformants and the ratios of band intensities were consistent with the presence of three copies of YAL36 in strain CAI-4. Deviations in the ratios from theoretical expectations were probably due to differences in transfer efficiency for fragments of different sizes. Therefore, we conclude that this stock of CAI-4 contains three copies of YAL36 and that the number of copies of a gene can be determined without transformation by quantifying the relative band intensities of polymorphic fragments following Southern hybridization.
The location of the YAL36 gene on the genetic map of Candida albicans was determined by identifying the YAL36 open reading frame (ORF) within the partially assembled contigs that comprise the C. albicans genome. Interestingly, a second gene that we had studied previously, CRD1, which encodes a pump that confers resistance to copper toxicity, was located on the same SfiI fragment and appeared to have three copies in our stocks of strains CAI-4 and SGY-243 (Riggle and Kumamoto, 2000). These results suggested that this stock of strain CAI-4 carried either three copies of chromosome 1 or a duplication of part of chromosome 1 somewhere in the genome.
To examine these possibilities, the locations of other genes previously reported to have three copies were determined. As shown in Table 1, the majority of these genes are also located on chromosome 1. Most of these studies were performed using strain CAI-4 as the starting strain and, therefore, these results suggested that stocks of strain CAI-4 used in several laboratories carried three copies of genes located on chromosome 1. In addition, as the FKS1 and YAL36 genes mapped to different SfiI fragments, the results suggested that CAI-4 may carry a third copy of the entire chromosome. In addition, another laboratory-generated strain, SGY-269, was found to carry three copies of GLY1, suggesting that this strain may similarly carry three copies of chromosome 1.
Table 1. . Genes reported to have three copies in C. albicans.
Consistent copy numbers of genes on chromosome 1 in several strains of C. albicans indicate that the entire chromosome is present in three copies in some strains
Previous studies of other genes that map to chromosome 1, e.g. CPP1 (Csank et al., 1997) and CPH1 (Liu et al., 1994), reported only two copies of the genes. These results may indicate that different stocks of strain CAI-4 differ in the number of copies of chromosome 1 or that the entire chromosome is not duplicated. To distinguish between these possibilities, the copy number of several genes was determined in several strains using polymorphic restriction sites and determining the ratios of the hybridized bands.
Five genes GLY1, FKS1, CPH1, CPP1 and YAL36 were chosen for this study. A refined map of chromosome 1 demonstrated that these five genes map to several SfiI fragments (Fig. 2A). The addition of more genes in the C. albicans database allowed finer mapping of the chromosome in comparison with the map of Chu et al. (1993). In addition, fosmid contig information was used to join SfiI fragment B-A-C2 to L. In the refined map, FKS1 is located on fragment S, GLY1 is located on fragment E, YAL36 is located on fragment J, CPP1 is located on fragment J, and CPH1 is located on fragment B (Fig. 2A).
As CPP1 and CPH1 have no reported polymorphisms, we first identified useful polymorphisms. Analysis of genomic sequence allowed the prediction of possible polymorphisms, which were then tested by Southern hybridization, as described in Experimental procedures. Several polymorphisms were found in or near CPP1 and CPH1 using this approach (data not shown), and one was chosen for each gene. The restriction fragment length polymorphisms (RFLPs) used to study these five genes are described in Table 2.
Table 2. . RFLPs used for FKS1, GLY1, YAL36, CPH1, CPP1 and CZF1.
To determine the copy number of the five genes in wild-type C. albicans strains, we first performed Southern hybridization and band intensity quantification on several clinical strains recovered from patients. In all strains in which there was a polymorphism in the locus of interest, the ratio of band intensities was close to 1:1, indicating that these strains contained two copies of all five genes (data not shown). In addition, the prototrophic parent of strain CAI-4, SC5314, also contained two copies of all five genes (Fig. 2B, lane a). For example, for the GLY1 locus in strain SC5314, the ratio of band intensities ranged from 0.90 to 1.21 in five determinations. Therefore, we concluded that there are two copies of these genes in wild-type C. albicans, as expected for a diploid organism.
In the next experiment, various laboratory strains were studied to determine how many copies of the genes of interest were present in each strain. As shown in Fig. 2B, the wild-type parental strain SC5314 was found to contain two copies of all five genes. Strain CAF2, produced by deletion of one copy of the URA3 gene from SC5314 (Fonzi and Irwin, 1993), similarly contained two copies of all five genes (Fig. 2B, lane b). In contrast, two different laboratory stocks of strain CAI-4, which was generated by deleting the second copy of URA3 from CAF2 (Fonzi and Irwin, 1993), contained three copies of all five genes (Fig. 2B, lanes c and d). For example, in strain CAI-4 (K), the ratio of GLY1 band intensities ranged from 1.97 to 2.34 in three determinations. However, strain CAF4-2, constructed in parallel with strain CAI-4, contained two copies of all five genes (Fig. 2B, lane e), and strain BWP17, a multiply marked strain derived from CAI-4 (Wilson et al., 1999), also contained two copies of all five genes (Fig. 2B, lane g).
In addition to studying the stocks of CAI-4 used in the Kumamoto and Magee laboratories, we also studied the original stock of CAI-4, which was kindly provided by Dr William Fonzi. Analysis of the copy number of GLY1 in several clones purified from the original CAI-4 sample showed that some clones (e.g. F2; Fig. 2B, lane i) yielded a ratio of 1.1:1, whereas other clones (e.g. F3; Fig. 2B, lane h) yielded a ratio of 1.9:1, and CAI-4 (F3) contained three copies of all five genes. Therefore, the original culture of CAI-4 contained strains that carried three copies of all five genes.
As these five genes all map to chromosome 1, are located throughout the chromosome and have the same copy number in each strain, we concluded that there is an extra copy of chromosome 1 in some stocks of strain CAI-4. In addition, we analysed the electrophoretic karyotypes of SC5314 and CAI-4 (K). In the contour-clamped homogeneous electric field (CHEF) separation, no differences were seen in the size of chromosome 1 between the strains (data not shown). Furthermore, in the SfiI separation, no differences were observed in the sizes of the SfiI fragments for the genes tested (e.g. CPH1) between the strains (data not shown), suggesting that the genes are not tandemly duplicated. Together, these results further support the conclusion that there is an extra copy of the entire chromosome 1 in some stocks of CAI-4.
Another independently constructed Δura3 strain, SGY-243, was found to contain three copies of all five genes (Fig. 2B, lane f), but three Ura– strains, CAF4-2, CNC43 (data not shown) and BWP17, which are related to CAI-4, were found to carry two copies of these five genes. Other Ura– strains, including strains 1006 and WO-1 (Ura–), were also found to carry only two copies of the genes on chromosome 1 (data not shown). These results demonstrate that not all Ura– strains carry an extra copy of chromosome 1. Therefore, it is unlikely that the presence of the third copy of chromosome 1 is directly related to the Ura– phenotype of strains CAI-4 and SGY-243.
Strains CAI-4 and SGY-243 are trisomic, not triploid
To study whether strains CAI-4 and SGY-243 contained three copies of all chromosomes, i.e. are triploid, the copy number of a gene not located on chromosome 1 was determined. CZF1, a gene that maps to chromosome 4, has an XcmI polymorphism allowing analysis of copy number by Southern hybridization (Fig. 2B). There were only two copies of CZF1 in all the strains tested, including strains CAI-4 and SGY-243.
Candida albicans strains with three copies of chromosome 1 are avirulent
As strain CAI-4 is commonly used for the construction of mutant strains and the analysis of mutant virulence, we examined whether chromosome 1 copy number influenced virulence. To this end, the virulence of strains containing either two or three copies of chromosome 1 was analysed in the haematogenously disseminated murine model. In the first study, we analysed two strains purified from the original culture of CAI-4, CAI-4 (F3), which carries three copies of chromosome 1, and CAI-4 (F2), which carries two copies. These strains were transformed with a vector encoding the wild-type URA3 gene, and the chromosome 1 copy number of the transformants was shown to be unchanged by the transformation. Cells from each strain were injected into five or six mice per strain via the lateral tail vein, and survival of the mice was monitored as a function of time. As shown in Fig. 3A, strain F2U (two copies) killed mice whereas strain F3U (three copies) did not.
In the second experiment, strain CAF4-2U (two copies of chromosome 1) was compared with two derivatives of CAI-4, strains XC101 and XC102. XC101 and XC102 are transformants generated in the same transformation when a Ura+ vector was introduced into trisomic CAI-4. During this process, XC101 appeared to lose one copy of chromosome 1 and thus carried only two copies. XC102, and most transformants, maintained the three copies of chromosome 1 found in the starting strain. Therefore, XC101 and XC102 are two closely related strains derived from the same parent that differ in copy number of chromosome 1. These strains were inoculated intravenously into mice, and mouse survival was monitored as a function of time. Consistent with the results shown in Fig. 3A, both strains with two copies of chromosome 1 (CAF4-2U and XC101) were capable of killing mice, whereas trisomic strain XC102 failed to kill the mice (Fig. 3B). In addition, the virulent strains typically yielded ≈ 5 × 105 cells per kidney at day 3, whereas the trisomic strain yielded less than 100 cells or no cells in the kidney. Therefore, these studies demonstrated that two closely related pairs of strains differing in chromosome 1 copy number exhibited a striking difference in virulence.
To determine whether changes in chromosome 1 copy number occurred during growth within the host, C. albicans cells were cultured from homogenates of kidneys dissected from mice inoculated with CAF4-2U, XC101 or XC102 at 3 and 11 days after infection. The copy numbers of GLY1 and FKS1 were determined by Southern hybridization and quantification of band intensity. As a control, cells from the same culture used to inoculate the mice were plated, and 20 of these colonies that had not been passaged in the mouse were analysed for chromosome 1 copy number as above.
As shown in Table 3, in the overnight culture before passage in the mouse, the copy number of chromosome 1 was unchanged in 20 out of 20 colonies from each strain. At both day 3 and day 11 of growth within mice, the copy number of chromosome 1 in strains with two copies of chromosome 1 was also unchanged in 20 out of 20 colonies from strains CAF4-2U and XC101. However, the copy number of chromosome 1 in trisomic strain XC102 changed. At day 3, six out of 49 colonies (12%) recovered from the kidney had undergone a change in chromosome 1 copy number, and the ratio of the intensities of the two bands on a Southern blot changed from 2:1 to 1:1. At day 11, six out of 20 XC102 colonies (30%) recovered from the kidney had changes in chromosome 1 copy number.
Table 3. . Changes in chromosome 1 copy number after growth in mice.
. Number of clones exhibiting a change in chromosome 1 copy number divided by total number of clones studied.
CAF4-2U (two copies)
XC101 (two copies)
XC102 (three copies)
Surprisingly, chromosome 1 was not the only chromosome to undergo changes during growth in mice. As shown in Fig. 4, we found that, for trisomic strain XC102, only two copies of CZF1 (on chromosome 4) were observed in 20 out of 20 colonies that had not been passaged in the mouse (Fig. 4, lane a). There was also no change in the copy number of CZF1 in the 49 colonies recovered from the kidney at 3 days after infection (data not shown). However, among the 20 colonies recovered from the kidney at 11 days after infection, 13 out of 20 colonies (65%) gained an extra copy of CZF1 (Fig. 4). These changes could be caused by duplication of the CZF1 gene or by the presence of an extra copy of chromosome 4. Four of the 20 colonies had changes in both chromosome 1 and CZF1 copy number (e.g. Fig. 4, lanes b and c). These results demonstrated that strains with multiple changes in chromosome or gene copy number arise during growth within the host.
Thus, trisomic strains failed to kill mice and appeared to have a growth disadvantage that allowed strains with altered chromosome copy number to accumulate. These two findings indicate that the presence of a third copy of chromosome 1 compromises the virulence of C. albicans.
Growth in the presence of 5-fluoroorotic acid results in alterations in the copy number of chromosome 1
When CAI-4 strains, containing three copies of chromosome 1, were grown in liquid rich medium and passaged for ≈ 120 generations, 15 individual colonies purified from five independent populations after growth were found to contain three copies of chromosome 1 with no changes in chromosome 1 copy number (data not shown). Therefore, during growth in the absence of a stress condition, the copy number of chromosome 1 appeared to be stably maintained.
However, during genetic manipulations, C. albicans strains are often exposed to the compound 5-fluoroorotic acid (5-FOA) in order to select for Ura– strains and, in particular, this selection method was used during the generation of strain CAI-4. Therefore, growth in the presence of 5-FOA may result in changes in chromosome copy number. When a strain in which one copy of a gene of interest was replaced by the URA3 gene flanked by direct duplication of Salmonella typhimurium hisG (hisG-URA3-hisG) is plated on 5-FOA plates, the Ura– colonies that arise may have undergone intrachromosomal recombination between the hisG repeats, loss of the hisG-URA3-hisG-bearing mutation by gene conversion or mitotic recombination and segregation or loss of the entire chromosome carrying the hisG-URA3-hisG mutation. In most cases, the intrachromosomal recombination event is the most frequently observed event. However, when a trisomic yal36 mutant strain (Δyal36::hisG-URA3-hisG/YAL36/YAL36) was plated on 5-FOA plates, ≈ 90% of the Ura– colonies that arose appeared to have lost one copy of YAL36, along with URA3 (data not shown), possibly as a result of the loss of one copy of chromosome 1, consistent with the hypothesis that growth in the presence of 5-FOA results in changes in chromosome copy number.
To determine whether changes in chromosome copy number would be observed in the absence of direct selection against the chromosome, we asked whether changes in the copy number of chromosome 1 would occur during selection for an intrachromosomal recombination event on chromosome 4. The starting strain for this experiment, XC1, was derived from strain CAF4-2 (two copies of chromosome 1; Fig. 5, lane a) by deleting one allele of CZF1 (on chromosome 4) and replacing its sequences with a copy of the URA3 gene flanked by direct duplication of S. typhimurium hisG. This heterozygous strain XC1 contained two copies of chromosome 1 (Fig. 5, lane b).
Cells of strain XC1 were grown in YPD + uridine liquid medium and plated on either YPD plates or 5-FOA plates to select for Ura– strains. At day 3, 30 colonies were picked from both types of plates. At day 8, new colonies arose on the 5-FOA plates that were not present at day 3, and 30 of these colonies were picked. At day 8 on the YPD plates, no new colonies were observed, and 30 different colonies were picked in addition to the 30 previously picked colonies. Most of the colonies picked from 5-FOA plates at day 8 grew slowly in the presence or absence of 5-FOA. The copy number of chromosome 1 in these colonies was analysed by studying the copy number of YAL36, FKS1, GLY1, CPP1 and CPH1 (Fig. 5).
The results of two experiments are summarized in Table 4. At day 3, only two out of 60 colonies grown on 5-FOA plates had undergone a change in chromosome 1 copy number (e.g. Fig. 5, lanes c and d). However, at day 8, 18 of 60 newly appearing colonies had changes in chromosome 1 copy number. Twelve colonies gained an extra copy of chromosome 1 (e.g. Fig. 5, lanes e–h), while six colonies lost heterozygosity on chromosome 1 possibly as a result of a chromosome loss event (e.g. Fig. 5, lanes i and j). The 90 control colonies from YPD plates all contained two copies of chromosome 1 at day 3 and day 8.
Table 4. . Chromosome 1 copy number in strains grown on 5-FOA.
Therefore, these results demonstrated that the copy number of chromosome 1 changed during exposure to 5-FOA, and that the colonies could either gain an extra copy of chromosome 1 or possibly lose one copy of chromosome 1 (at least transiently). Furthermore, longer exposure to 5-FOA resulted in a very high percentage of colonies having changes in copy number (≈ 30% of the colonies).
To determine whether the copy number of chromosome 1 in trisomic CAI-4 was stable during exposure to 5-FOA, we tested three independently constructed CZF1 heterozygous strains and six Ura– strains derived from them by growth in the presence of 5-FOA. All three CZF1 heterozygotes contained three copies of chromosome 1 (data not shown). However, after exposure to 5-FOA, one out of six Ura– colonies exhibited a 1:1 ratio of fragment intensities for both GLY1 and FKS1, suggesting that a copy of chromosome 1 had been lost. These results further support the hypothesis that changes in chromosome copy number occur during growth in the presence of 5-FOA.
Linkage map of chromosome 1
By studying colonies that underwent changes in chromosome 1 copy number during exposure to 5-FOA, we constructed a linkage map of the five genes on chromosome 1 (Fig. 6). One homologue of chromosome 1 (designated 1a) contains the L, L, L, S, L alleles of genes FKS1, GLY1, YAL36, CPP1 and CPH1 respectively (L denotes the allele that yields the larger fragment after digestion of chromosomal DNA with the appropriate restriction enzyme, and S denotes the allele that yields the smaller fragment). The other homologue of chromosome 1 (designated 1b) carries the S, S, S, L, S alleles of genes FKS1, GLY1, YAL36, CPP1 and CPH1 respectively. The stocks of trisomic CAI-4 in the Kumamoto laboratory, Magee laboratory and Fonzi laboratory all gained an extra copy of 1a.
Of the colonies derived from CAF4-2 that gained an extra copy of chromosome 1 during growth in the presence of 5-FOA, half gained an extra copy of 1a, and the other half gained an extra copy of 1b. In contrast, all six colonies that lost heterozygosity of chromosome 1 lost the alleles corresponding to 1b. These results suggest that there was a bias for losing a particular homologue of chromosome 1 in these strains, and that 1b in CAI-4 may carry a recessive lethal mutation.
Exposure to fluconazole also resulted in changes in the copy number of chromosome 1
To determine whether other growth-inhibiting drugs would affect chromosome copy number, we studied the effect of exposure to fluconazole, a commonly used antifungal drug. Trisomic strain CAI-4 (three copies of chromosome 1) and strain PR913 (two copies of chromosome 1) were grown for 120 generations in the presence of gradually increasing concentrations of fluconazole (P. Riggle and C. A. Kumamoto, unpublished). Highly drug-resistant strains arose, and one to three individual colonies from each of six independent populations were analysed for chromosome 1 copy number.
Study of the copy number of GLY1 and FKS1 showed that the majority of the strains derived from trisomic strain CAI-4 appeared to lose one copy of chromosome 1 and exhibited a 1:1 ratio of band intensities (12 out of 14). In contrast, the fluconazole-resistant strains derived from strain PR913 all maintained two copies of chromosome 1 (8 out of 8). As noted above, control strains grown for the same length of time without fluconazole maintained the same copy number as the starting strain (data not shown). We conclude that changes in copy number of chromosome 1 can occur during selection with growth-inhibiting drugs, especially in the trisomic strain.
Here, we show that some stocks of two commonly used laboratory strains of C. albicans carry a third copy of chromosome 1. In addition, we demonstrate that the presence of an extra copy of chromosome 1 leads to avirulence of the strain in an animal model. These two results provide the first report of a connection between a specific chromosomal change and strain fitness during growth within a host.
During laboratory growth in rich medium, the copy number of chromosome 1 was stably maintained. However, changes in copy number of chromosome 1 were detected after growth within an animal host. These results suggest that the fitness of strains with two or three copies of chromosome 1 may be similar during laboratory growth in the absence of stress conditions but different during growth within an animal. This difference in fitness may arise because higher expression of one or more genes located on chromosome 1 due to trisomy decreases the virulence of the organism. In addition, as strains carrying an additional copy of the CZF1 locus arose after growth of the trisomic strain in the host, a gene(s) linked to CZF1 on chromosome 4 may enhance virulence so that increased expression favours virulence. These differences in fitness can account for the observation that strains with altered chromosome and gene copy numbers were isolated after growth within a host.
Another factor that may favour the production of strains with altered copy number during growth within a host is an increase in the rate at which variants with altered chromosome copy number arise. Perhaps, during growth within an animal, when an organism encounters conditions that restrict its growth, the organism enters a ‘hypermutable state’ (Hall, 1992) in which changes in chromosome copy number occur at a higher rate than during normal growth. This model accounts for the finding that strains with multiple changes in chromosome copy number were isolated in our study and previously (e.g. Perepnikhatka et al., 1999). As Suzuki and Rustchenko pointed out, the ability to undergo chromosomal changes may be a strategy for C. albicans to generate genetic diversity (Suzuki et al., 1989; Rustchenko-Bulgac et al., 1990; Rustchenko et al., 1997).
The most straightforward explanation for the chromosomal changes described in this report is that they were caused by mitotic non-disjunction because, in all cases, strains either gained or lost copies of all the genes studied co-ordinately. Although the precise mechanism underlying mitotic non-disjunction remains largely unknown, recent studies indicate that defects in sister chromatid cohesion lead to genetic instability (Nasmyth, 2002), and defects in the mitotic spindle may also cause abnormal chromosome segregation (Mountain and Compton, 2000).
The effect of gaining a copy of one chromosome on protein levels was studied by Marichal et al. (1997) using the haploid yeast Candida glabrata. When the expression of proteins was compared between an azole-resistant strain that had gained an extra copy of one chromosome and its derivative, azole-sensitive revertant strain that had lost the gained copy of the chromosome, 25 proteins were found to be upregulated and 76 proteins were downregulated by more than a factor of three out of 1337 proteins identified. As C. albicans chromosome 1 is one of the largest chromosomes (≈ 3.4 Mb), it could encode 1000 genes or more. We expect that expression of these genes and their protein levels would be affected by amplification of the entire chromosome. Furthermore, some of the genes on chromosome 1 encode transcription factors, e.g. CPH1, and such transcription factors may affect the expression of genes located on the other chromosomes. Despite the large numbers of genes potentially affected by the chromosome 1 trisomy, the trisomic strains grow well under laboratory conditions. However, within an animal host, the trisomic strains exhibited a striking difference in growth in comparison with non-trisomic strains.
Our findings may provide an explanation for puzzling results obtained in previous virulence studies. For example, during the study of the AAF1 gene, located on chromosome 3 and involved in flocculation and adhesion, researchers found that different mutant isolates constructed from strain CAI-4 varied in their growth rates and their virulence in the mouse model (Rieg et al., 1999). Some heterozygous aaf1/AAF1 mutants exhibited relatively slow doubling times and reduced virulence, whereas other heterozygous mutants exhibited faster doubling times and were virulent. Surprisingly, homozygous aaf1/aaf1 mutants were virulent. The puzzling results of this experiment may be explained by our findings because we have observed that trisomic strains have a doubling time ≈ 10–15% longer than closely related diploid strains (X. Chen and C. A. Kumamoto, unpublished observations). As both transformation and exposure to 5-FOA, standard steps in C. albicans gene disruption, are associated with changes in chromosome copy number, some of the heterozygous aaf1/AAF1 mutant isolates may have carried three copies of chromosome 1 while others carried two, leading to differences in growth rates and virulence. The homozygous aaf1/aaf1 mutants may have carried two copies of chromosome 1, resulting in faster growth and virulence.
These effects complicate virulence studies using strain CAI-4 as a null mutant lacking a gene of interest and carrying three copies of chromosome 1 would be avirulent but not because of the deletion mutation. When the wild-type gene of interest was reintroduced into the null mutant by transformation, the healthiest, fastest growing transformants may have lost one copy of chromosome 1 and would be virulent. In this case, the recovery of virulence would not reflect the reintroduction of the gene of interest but rather the change in the copy number of chromosome 1.
In addition, we demonstrated that exposure of Ura+ strains to 5-FOA for long periods of time resulted in a very high frequency of strains carrying altered numbers of chromosome 1. Chromosomal changes after 5-FOA exposure were also observed by Wellington and Rustchenko (2002), who showed that 1/100 5-FOAR mutants derived from C. albicans strain 3153A carried either a duplication of chromosome 4b or an increase in the length of one homologue of chromosome 5. Demonstration that a null mutant and its derivative carrying the reintroduced wild-type gene both have two copies of chromosome 1, for example by analysing the copy number of GLY1 and CPH1 as described in this paper, will be essential before the results of virulence studies can be interpreted.
Our results and the results of others indicate that multiple changes in C. albicans chromosome copy number can occur during growth under stress conditions. As demonstrated in this study, these stress conditions include growth within a host, providing a mechanism for adaptation of the pathogen to the host. Thus, changes in chromosome copy number provide a means for genetic variation in this organism.
Candida albicans strains and plasmids are listed in Tables 5 and 6. Strains XC111 and XC112 were constructed by transforming strain CAI-4 (Δura3/Δura3; three copies of chromosome 1) with a DNA fragment from plasmid pXC110, encoding the Δyal36::hisG-URA3-hisG deletion/replacement allele, using lithium acetate transformation. Transformants containing a replacement of wild-type YAL36 with the yal36 deletion mutation were identified by Southern blotting. Plasmid pXC110 was constructed from pBB510 (http://www.sacs.ucsf.edu/home/JohnsonLab/burk/Candisrupt.html) by the introduction of fragments of DNA derived from the YAL36 locus (X. Chen and C. A. Kumamoto, unpublished).
To construct strain XC1, plasmid pDB114 (Brown et al., 1999) was digested with HindIII and SacI and used to transform strain CAF4-2 (Δura3/Δura3; two copies of chromosome 1) using lithium acetate transformation. Transformants containing the desired czf1 mutation were identified by Southern blotting.
To generate the strains used in the virulence study, plasmid pRC3915 (Cannon et al., 1990), encoding URA3, was digested with BstEII to direct integration to the LEU2 gene. The Δura3/Δura3 strains CAI-4, CAF4-2, CAI-4 (F2) and CAI-4 (F3) were individually transformed with the digested pRC3915. The copy number of chromosome 1 in the URA3+ transformants was confirmed by Southern blotting.
Media and growth conditions
For C. albicans, cells were routinely grown in either YPD medium (1% yeast extract, 2% bacto peptone, 2% glucose) or CM-Ura (CM lacking uridine and uracil) (Ausubel et al., 1989). YPD medium was supplemented with 60 µg ml−1 uridine for growth of Ura– strains. For selection of Ura– clones, CM-Ura medium was supplemented with 60 µg ml−1 uridine and 1 mg ml−1 5-fluoroorotic acid (5-FOA) (Kelly et al., 1987). Escherichia coli strains were cultured in L broth or on L plates (Miller, 1972) with ampicillin added to a concentration of 100 µg ml−1.
For long-term growth in the presence or absence of fluconazole, six individual colonies of strains CAI-4 or PR913 were grown separately in Sabouraud + uridine liquid medium (250 µg ml−1 uridine). After overnight growth, the cultures were diluted and grown in the same medium in the presence of 2 µg ml−1 fluconazole. Over time, the concentration of fluconazole in the liquid medium was gradually increased from 2 µg ml−1 to 64 µg ml−1. One to three individual colonies from each culture were purified and used for this study. As a control, cultures of the same starting strains were diluted and passaged in Sabouraud + uridine liquid medium every day, without the addition of fluconazole. From each culture, three individual colonies were purified and used for this study.
Chromosome 1 map
Southern blots of CHEF gels of strains SC5314, CAI-4 and 1006 were probed with isolated genes or polymerase chain reaction (PCR) fragments of genes and assigned to chromosomes, SfiI fragments and the fosmid library using the method of Chibana et al. (1998).
Growth on 5-FOA
Strain XC1 was grown overnight in liquid YPD + 18 µg ml−1 uridine at 30°C and then plated. Approximately 1 × 101 cells were plated on YPD plates and 1 × 104 cells were plated on 5-FOA plates to select for Ura– strains. At day 3, 30 colonies were picked from both types of plates. At day 8, new colonies arose on the 5-FOA plates that were not present at day 3, and 30 of these colonies were picked. All colonies picked from 5-FOA plates were tested to confirm their Ura– phenotype. At day 8 on the YPD plates, no new colonies were observed, and 30 different colonies were picked in addition to the 30 previously picked colonies.
RFLPs in the CPP1 and CPH1 loci
Two contigs encoding the CPP1 gene (contigs 6–2491 and 6–1671) and one contig encoding the CPH1 gene (contig 6–1667) were observed in assembly 6 of the unfinished C. albicans genomic sequence (http://www-sequence.stanford.edu/group/Candida/index.html). The two CPP1 contigs may correspond to the two alleles of each gene present in the diploid strain. When the restriction maps of the two contigs of CPP1 were compared, possible polymorphisms were observed. When the restriction map of contig 6–1667 was compared with the C. albicans CPH1 DNA sequence deposited in GenBank (U15152) (Malathi et al., 1994), possible polymorphisms were also observed. To identify RFLPs that existed in the genomic DNA of wild-type C. albicans, probes were designed for Southern analysis, and the sizes of DNA restriction fragments were predicted based on information from the contigs. This method allowed the efficient identification of RFLPs. For each locus, one RFLP was chosen for the analysis of gene copy number from several RFLPs identified.
PCR, restriction digestion and gel electrophoresis were performed by standard methods as described previously (Sambrook et al., 1989). Automated DNA sequencing was performed by Michael Berne and coworkers at the Tufts University Core Facility.
Methods for chromosomal DNA isolation and Southern blot hybridization were performed by standard methods as described previously (Sambrook et al., 1989; Burke et al., 2000). Restriction enzymes used for the analysis of each gene are shown in Table 2. DNA probes were labelled with the Bethesda Research Laboratories random primer labelling kit and [α-32P]-dATP (New England Biolabs). C. albicans FKS1, YAL36, CPP1, CPH1 and GLY1 PCR products were used for probing Southern blots. The primers used were 5′-TGAATGAGGACGAGGAACCA and 5′-TCATCATGGCAT TCATACCAG for FKS1; 5′-TTGACAGGAGCCTTAGATGGA and 5′-AGAAGAAGAACAGCGTAGGTT for YAL36; 5′-AAGAACTGACATCAATAAATGCA and 5′-AGGACATATTC TATCACAGGCA for CPP1; 5′-GTTGTTGGTGAGGTG GAATC and 5′-TCACTACAGAGTGCTAATGGT for CPH1; and 5′-CTAGCCAGCTAATAGAGGCT and 5′-CAACTTC CGTCCATCGCCT for GLY1. A SacI–XcmI fragment from pDB112 was used as probe for CZF1.
To study the changes in the copy number of CZF1 on chromosome 4 in colonies recovered from mouse kidneys, chromosomal DNAs were digested with SspI, and a C. albicans CZF1 PCR product was used for probing a Southern blot. The primers used were 5′-CACAATCTGTAGGTTAC CTAG and 5′-TGCTGCTCTGATGGAGACAA for the CZF1 probe.
The sizes of DNAs on Southern blots correlated with the expected lengths based on information from the C. albicans genome database.
Quantification of band intensity
Band intensity was quantified using the Storm PhosphorImager (Amersham Biosciences) and imagequant software. Band intensities were determined by calculating the areas of the peaks.
Candida albicans strains were grown overnight in CM-Ura and harvested by centrifugation at 2500 g. The cells were washed once with 1× PBS, suspended in 1× PBS, counted and adjusted to a density of 5 × 106 cells ml−1. Cell suspension (0.2 ml), containing 1 × 106 cells, was injected into the lateral tail vein of female CF1 mice (18–20 g; Charles River Laboratories). Each strain was tested in five or six mice. Surviving mice were observed daily after infection with C. albicans. At day 3 and day 11 after infection, the kidneys were removed aseptically from one mouse per strain. At day 11 after infection, mice infected with strains CAF4-2U or XC101 were moribund. The organs were weighed, homogenized, diluted in PBS, and the homogenate was plated on a YPD + amp plate. Colonies were counted after incubation of the plates at 30°C for 48 h.
We are grateful to William Fonzi, Aaron Mitchell, Elena Rustchenko and Perry Riggle for providing laboratory strains. We also thank Susan Hadley for providing clinical isolates. We thank Ralph Isberg, Malcolm Whiteway, Claire Moore, Andrew Wright, Gavin Schnitzler, Perry Riggle, Marcelo Vinces and Igor Bruzual for helpful discussion and critical reading of the manuscript. We thank Neil Gow for sharing information on CCH1 before publication. We also thank the Tufts University Division of Laboratory Animal Medicine for technical assistance. We thank the Candida albicans Genome Sequencing Project, supported by the NIDR and Burroughs Wellcome Fund, and the Stanford DNA Sequencing and Technology Center and the University of Minnesota for establishing and maintaining the C. albicans genome sequence database. This work was supported by NIH grant AI38591 (to C.A.K.) and NIH grants AI46351 and AI05406 (to P.T.M.).