Ectopic Expression of URA3 Can Influence the Virulence Phenotypes and Proteome of Candida albicans but Can Be Overcome by Targeted Reintegration of URA3 at the RPS10 Locus

Abstract

Uridine auxotrophy, based on disruption of both URA3 alleles in diploid Candida albicans strain SC5314, has been widely used to select gene deletion mutants created in this fungus by “Ura-blasting” and PCR-mediated disruption. We compared wild-type URA3 expression with levels in mutant strains where URA3 was positioned either within deleted genes or at the highly expressed RPS10 locus. URA3 expression levels differed significantly and correlated with the specific activity of Ura3p, orotidine 5′-monophosphate decarboxylase. Reduced URA3 expression following integration at the GCN4 locus was associated with an attenuation of virulence. Furthermore, a comparison of the SC5314 (URA3) and CAI-4 (ura3) proteomes revealed that inactivation of URA3 caused significant changes in the levels of 14 other proteins. The protein levels of all except one were partially or fully restored by the reintegration of a single copy of URA3 at the RPS10 locus. Transcript levels of genes expressed ectopically at this locus in reconstituted heterozygous mutants also matched the levels found when the genes were expressed at their native loci. Therefore, phenotypic changes in C. albicans can be associated with the selectable marker rather than the target gene. Reintegration of URA3 at an appropriate expression locus such as RPS10 can offset most problems related to the phenotypic changes associated with gene knockout methodologies.

Genes encoding enzymes in the uridine biosynthetic pathway are used as positive selectable markers in gene disruption strategies for a wide range of eukaryotic organisms. For the major human pathogenic fungus Candida albicans, this approach, based on the URA3 gene, was the only means available for gene disruption for more than 10 years. C. albicans is diploid and has no known natural sexual cycle that leads to a haploid form. Therefore, molecular genetic evidence associating phenotypic changes with individual gene functions requires the successive disruption of two copies of the gene of interest. “Ura-blasting,” used for gene disruption in C. albicans (28), is based on a similar approach in Saccharomyces cerevisiae (1) and makes use of a C. albicans ura3 auxotroph. Transformants are selected on the basis of uridine prototrophy. The marker is then recycled by use of 5-fluoroorotic acid to select for Ura⁻ segregants that have lost URA3 through a reciprocal crossover between identical flanking DNA repeats. The “Ura-blaster” can then be used to disrupt the second allele in these ura3 segregants. PCR-based gene disruption protocols also exploit the URA3 marker but often do not recycle it (90).

One area of investigation that has made extensive use of specific gene disruption based on the URA3 selection strategy is the study of virulence in C. albicans. At least 50 genes have been implicated as putative virulence factors in this fungus based on results of mouse intravenous challenge experiments in which the homozygous C. albicans null mutants of interest were shown to be less lethal than wild-type controls (61). In most of these studies, the strain tested in mice had been the Ura⁺ transformant obtained after the second round of Ura-blasting (e.g., yfgΔ::hisG/yfg1Δ::hisG-URA3-hisG).

“Molecular Koch's postulates” suggest that confirmation of a microbial virulence factor requires that (i) inactivation of the gene should attenuate virulence and (ii) reintegration of the gene into the null mutant should restore wild-type virulence. In C. albicans, this process depends predominantly on the use of URA3 as a selectable marker. However, the use of Ura selection creates a technical problem. These procedures result in the transfer of URA3 from its normal location on chromosome 3 to a nonnative chromosomal location. Hence, mutant and control strains used in virulence experiments often differ in two (not one) respects: disruption of the target locus and the genomic location of URA3.

A number of publications have documented problems arising from the use of URA3 as a selectable marker following gene disruption. Lay et al. (48) demonstrated that specific activities of orotidine 5′-monophosphate (OMP) decarboxylase (Ura3p) differed between strains of C. albicans harboring one or two copies of URA3 in a way that was not simply related to the gene dosage. They showed an inverse correlation between OMP decarboxylase activity and the survival times of mice infected with these mutants. Sundstrom et al. (82) showed that the location of URA3 at the disruption site affected the virulence phenotype of hwp1/hwp1 mutants. Bain et al. (7) demonstrated that the URA3 status of cell wall mutants affected their ability to adhere to buccal epithelial cells. Cheng et al. showed that changing the location of URA3 in the C. albicans genome affected OMP decarboxylase activity, hyphal morphogenesis, buccal cell adherence, and lethality for mice (19). In addition, the phenotypic effects that have been correlated with the disruption of four C. albicans genes were shown to be a consequence of the genomic relocation of URA3 in three of these cases (19). Staab and Sundstrom reviewed the importance of URA3 positional effects in influencing phenotypes of C. albicans mutants (81). They propose that a “wild-type” strain of C. albicans comprising CAI-4 with URA3 reintegrated at a highly expressed locus, ENO1, is a more appropriate control for Ura-blasted mutants than either of the CAI-4 parental strains, SC5314 and CAF2-1.

We have further assessed the significance of URA3 expression as an influence on virulence and other phenotypic characteristics in C. albicans. Our experiments make use of the CIp10 vector, which reintegrates URA3 at the high-expression RPS10 locus (59), producing mutant and control strains with isogenic backgrounds. We show that expression levels of URA3 correlate with OMP decarboxylase activities in three mutants created using the Ura-blaster protocol. We have tested the association between OMP decarboxylase activities and mouse virulence in several C. albicans mutants and clinical isolates and studied the global effects of URA3 deletion on the C. albicans proteome in a pairwise comparison of SC5314 and CAI-4. We show, for a wide range of mutant strains, that ectopic expression of URA3 is likely to have an effect on the virulence phenotype in as many as 30% of published cases but that the phenotype can be corrected by expression of URA3 at the RPS10 locus.

MATERIALS AND METHODS

C. albicans strains, culture media, and growth conditions.

The C. albicans strains used in this study are listed in Table 1. C. albicans strains were routinely grown and maintained on Sabouraud agar. For reverse transcription-PCR (RT-PCR) experiments, strains were grown with shaking at 37°C in SD (6.7 g of yeast nitrogen base per liter, 20 g of glucose per liter) to an optical density at 600 nm (OD₆₀₀) of 0.95. For analysis by two-dimensional (2D) gel electrophoresis, strains were grown to an OD₆₀₀ of 0.65 at 37°C in YEPD (10 g of yeast extract [Oxoid] per liter, 20 g of mycological peptone [Oxoid] per liter, 20 g of glucose per liter) supplemented with 25 μg of uridine per ml. For inoculum preparation in animal experiments, strains were grown for 24 h in 5-ml lots of NGY (1 g of neopeptone [Difco] per liter, 4 g of glucose per liter, 1 g of yeast extract [Difco] per liter) rotated at 30°C at 20 rpm in a test tube wheel set at an angle of 5° from the horizontal. For OMP decarboxylase assays, cells were either grown in SD, as for RT-PCR experiments, or shaken in flasks of NGY at 37°C until growth reached an OD₆₀₀ of 0.6. Pilot experiments showed that there was no significant difference in enzyme activity from cells grown under the two conditions.

TABLE 1.

C. albicans strains used in this study

Strain	Description or name	Genotype	Source or reference
ATCC 44858	Isolate from parrot		Originally from Janssen Pharmaceutica
NCPF 3153	Clinical isolate		Reference antigen strain, Public Health Mycology Reference Laboratory
RV4688	Clinical isolate		Originally from Janssen Pharmaceutica
AM2003-020	Isolate from healthy carrier
73/034	Isolate from healthy carrier
SC5314	Clinical isolate		28
CAF2-1		ura3Δ-iro1Δ::imm434/URA3	28
CAI-4		ura3Δ-iro1Δ::imm434/ura3Δ-iro1Δ::imm434	28
NGY152	CAI-4/CIp10	ura3Δ-iro1Δ::imm434/Δura3Δ-iro1Δ::imm434/Clp10	60
NGY23	mnt1Δ/mnt1	As CAI-4 but mnt1Δ::hisG/mnt1Δ::hisG-URA3-hisG	15
NGY158	mnt1Δ/Clp10	As CAI-4 but mnt1Δ::hisG/mnt1Δ::hisG/Clp10	Bates et al., unpublished
NGY148	mnt1Δ/Clp10-MNT1	As CAI-4 but mnt1Δ::hisG/mnt1Δ::hisG/Clp10-MNT1	Bates et al., unpublished
NGY337a	mnt1Δ/mnt2Δ/Clp10	As CAI-4 but mnt1-IPF6318-mnt2Δ::hisG/mnt1-IPF6318- mnt2Δ::hisG/Clp10	Bates et al., unpublished
NGY105	mnt2Δ	As CAI-4 but mnt2Δ::hisG/mnt2Δ::hisG-URA3-hisG	Bates et al., unpublished
NGY145	mnt2Δ/Clp10	As CAI-4 but mnt2Δ::hisG/mnt2Δ::hisG/Clp10	Bates et al., unpublished
NGY103	MNT2/mnt2Δ	MNT2/mnt2Δ::hisG-URA3-hisG	Bates et al., unpublished
NGY336	MNT2/mnt2Δ/Clp10	MNT2/mnt2Δ::hisG/Clp10	Bates et al., unpublished
NGY63	mnt3Δ	As CAI-4 but mnt3Δ::hisG/mnt3Δ::hisG-URA3-hisG	Thomson et al., unpublished
NGY146	mnt3Δ/Clp10	As CAI-4 but mnt3Δ::hisG/mnt3Δ::hisG/Clp10	Thomson et al., unpublished
NGY71	mnt4Δ	As CAI-4 but mnt4Δ::hisG/mnt4Δ::hisG-URA3-hisG	Bates et al., unpublished
NGY313	mnt4Δ/Clp10	As CAI-4 but mnt4Δ::hisG/mnt4Δ::hisG/Clp10	Bates et al., unpublished
NGY71	mnt5Δ	As CAI-4 but mnt5Δ::hisG/mnt5Δ::hisG-URA3-hisG	Thomson et al., unpublished
NGY147	mnt5Δ/Clp10	As CAI-4 but mnt5Δ::hisG/mnt5Δ::hisG/Clp10	Thomson et al., unpublished
NGY151	mnt5Δ/Clp10-MNT5	As CAI-4 but mnt5Δ::hisG/mnt5Δ::hisG/Clp10-MNT5	Thomson et al., unpublished
DH5	mnn4Δ	As CAI-4 but mnn4Δ::hisG/mnn4Δ::hisG-URA3-hisG	Hobson et al., unpublished
DH15	mnn4Δ/Clp10	As CAI-4 but mnn4Δ::hisG/mnn4Δ::hisG/Clp10	Hobson et al., unpublished
DH11	mnn4Δ/Clp10-MNN4	As CAI-4 but mnn4Δ::hisG/mnn4Δ::hisG/Clp10-MNN4	Hobson et al., unpublished
DH13	mnn4Δ/Clp10-MNN4nb	As CAI-4 but mnn4Δ::hisG/mnn4Δ::hisG/Clp10-MNN4n	Hobson et al., unpublished
DH17	MNN4/mnn4Δ	MNN4/mnn4Δ::hisG/Clp10	Hobson et al., unpublished
GTC43	gcn4Δ	As CAI-4 but gcn4Δ::hisG/gcn4Δ::hisG-URA3-hisG	84
GTC41	GCN4/gcn4Δ	As CAI-4 but GCN4/gcn4Δ::hisG-URA3-hisG	Tournu et al., unpublished
GTC45	gcn4Δ/Clp10	As CAI-4 but gcn4Δ::hisG/gcn4Δ::hisG/Clp10	Tournu et al., unpublished
GTC49a	gcn4Δ/Clp10-GCN4(a)	As CAI-4 but gcn4Δ::hisG/gcn4Δ::hisG/Clp10-ACT1p-GCN4	Tournu et al., unpublished
GTC49b	gcn4Δ/Clp10-GCN4(b)	As CAI-4 but gcn4Δ::hisG/gcn4Δ::hisG/Clp10-ACT1p-/GCN4	Tournu et al., unpublished
GTC47	gcn4Δ/pMET-GCN4	As CAI-4 but gcn4Δ::hisG/gcn4Δ::hisG/pMET-GCN4	Tournu et al., unpublished
MMC3	nrg1Δ	As CAI-4 but nrg1Δ::hisG/nrg1Δ::hisG-URA3-hisG	60
NGY205	och1Δ	och1Δ::hisG/och1Δ::hisG	Thomas et al., unpublished
BH1-1-1	sap3Δ	As CAI-4 but sap3Δ::hisG/sap3Δ::hisG-URA3-hisG	35
VC3.9	sap3Δ/Clp10	As CAI-4 but sap3Δ::hisG/sap3Δ::hisG/Clp10	Copping et al., unpublished
rad52 heterozygote	RAD52/rad52Δ	As CAI-4 but RAD52/rad52Δ::hisG-URA3-hisG	Ciudad et al., unpublished
rad52 reintegrant	rad52Δ/RAD52	As CAI-4 but rad52Δ::hisG/RAD52::URA3-hisG	Ciudad et al., unpublished

^aDeletion spans MNT1, MNT2, and intervening ORF IPF6318 (putative β-glucosidase).

^bMulticopy reintegrant.

RNA isolation and cDNA synthesis.

Total RNA was extracted by the method of Schaller et al. (72). mRNA was isolated from total RNA using a Dynabeads mRNA purification kit (Dynal A. S., Oslo, Norway). cDNA was synthesized using 400 U of Superscript II RNase H⁻ reverse transcriptase kit (Invitrogen) as specified by the manufacturer. cDNA was quantified by measurement of OD₂₆₀ and diluted to give a stock concentration of 500 μg/ml.

Multiplex RT-PCR and estimation of URA3 mRNA abundance.

Fragments of URA3, GCN4, MNN4, and MNT5 cDNA were each coamplified with the cDNA of elongation factor B1 (54, 72). Primers for PCR are listed in Table 2. PCR conditions were 94°C for 10 s followed by 27 cycles of 94°C for 45 s, annealing temperature for 60 s (Table 2), and 72°C for 60 s. Approximately 500 ng of cDNA was used as template. The PCR products from sequential PCR cycles were visualized on 2% agarose gels, and the difference in cycle number between the appearance of the first band for EFB1 and that of the gene of interest was recorded for three independent samples for each strain. Within each RT-PCR experiment, the relative abundance of URA3 mRNA was estimated from the mean cycle number difference.

TABLE 2.

Primers used for RT-PCR

Gene amplified	Primer name	Primer sequence (5′-3′)	Product size (bp)	Annealing temp (°C)	Reference
EFB1	EFB1F	ATTGAACGAATTCTTGGCTGAC	526 (cDNA)		72
	EFB1R	CATCTTCTTCAACAGCAGCTTG	891 (genomic DNA)
MNT5	MNT5F	AGCATTGGAATCAATGCGATC	460	53	This study
	MNT5R	TTTCATAAATGATTCTACTGTGG
MNN4	MNN4F	AGATTCAGACGCTGATAGAGC	338	56	This study
	MNN4R	ATTCACGGATATGAACCTTC
GCN4	GCN4F	TTGGAGATGACAATGATGATGG	276	56	This study
	GCN4R	CAACTTCACCACCAGCAGTG
URA3	URA3F	TAATGCTCATGGTGTCCATG	243	54	This study
	URA3R	CAAATCCTTCTTCTTGTCCA

OMP decarboxylase assays.

Protein extraction from cells that had been disrupted by vortexing with 0.45-mm-diameter glass beads (Sigma-Aldrich Ltd, Poole, United Kingdom) and OMP decarboxylase assays were based on the methods described by Lay et al. (48) and Yashimoto et al. (95). The assay mixture, in a total volume of 850 μl, contained approximately 100 μg of cell lysate protein, 0.1 M phosphate buffer (pH 6.0), 1.0 μmol of β-mercaptoethanol and 0.15 μmol of OMP substrate (Sigma-Aldrich). Conversion of OMP to UMP was measured spectrophotometrically at room temperature as the decrease in absorbance of the substrate at 285 nm over a specified period. The concentration of OMP in the assay was calculated from a molar extinction coefficient for orotidine 5′-monophosphate of 1.65 × 10³ liters cm⁻¹ M⁻¹. One unit of enzyme was defined as the amount of enzyme required to convert 1 μmol of OMP to UMP per min. Specific activity was defined as units of enzyme activity per milligram of protein. The protein concentration was determined with a Coomassie protein assay reagent kit (Pierce, Rockford, Ill.). Enzyme activity was assayed for three independent samples per strain, and C. albicans SC5314 was included as a reference control in each experiment. To compare data from different experiments, specific activities of OMP decarboxylase were normalized to the activity measured for SC5314, which was set at 100%. The OMP decarboxylase specific activities reported previously for CAF2-1 (48) differed by approximately 1,000-fold from those we have measured.

Protein extraction and 2D gel electrophoresis.

The COGEME Proteome Research Facility 1 (http://www.abdn.ac.uk/cogeme) undertook all protein extraction and 2D gel electrophoresis experiments. Samples were resolved in the first dimension on 24-cm Immobiline DryStrip gels (pH 4 to 7) (Amersham Biosciences, Chalfont St. Giles, United Kingdom) in an Ettan IPGphor isoelectric focusing unit (Amersham Biosciences) and in the second dimension on precast sodium dodecyl sulfate-12.5% polyacrylamide gels with an Ettan Dalt system (Amersham Biosciences).

Image analysis and protein identification.

Images from three independent samples per strain were analyzed with Phoretix 2D software (Nonlinear Dynamics Ltd., Newcastle-upon-Tyne, United Kingdom) to identify features that were reproducibly altered between strains. Spots that displayed more than a twofold change in relative volume with a statistical significance of P ≤ 0.05 (Student's t test) were analyzed further. Peptide mass fingerprints were generated from spots by MALDI-TOF mass spectrometry and used to interrogate the CandidaDB database (http://genolist.pasteur.fr/CandidaDB/). Data are available at the COGEME PRF website (http://www.cogeme.abdn.ac.uk).

Experimental infections in mice.

Experimentation was done under the terms of United Kingdom Home Office licenses for research on animals. Female DBA/2 and BALB/c mice (Harlan) with a weight range from 17 to 23 g were maintained in individually ventilated cages under conditions specified by the Health and Safety Executive for level 2 biohazard containment. The animals were supplied with food and water ad libitum. Challenge inocula were grown overnight at 30°C in NGY, centrifuged, washed twice with sterile distilled water, and resuspended in sterile saline. The inocula were standardized by spectrophotometry, and their concentrations were adjusted according to the experimental design. The true concentrations of yeast inocula were routinely determined by viable counts made from the suspensions used for (intravenous) challenge.

Mice were challenged by i.v. inoculation of C. albicans suspensions via a lateral tail vein. For BALB/c mice the challenge dose was in the range of 4.2 to 5.0 log₁₀ CFU/g of body weight (median, 10^4.5 CFU/g), and for DBA/2 mice the dose range was 2.4 to 3.3 log₁₀ CFU/g of body weight (median, 10^2.8 CFU/g). Differences in mouse survival and organ burdens between C. albicans strains were not attributable to the variations in challenge dose. Correlation coefficients between challenge dose and mean survival times or mean organ burdens ranged from −0.1 to 0.3 for both mouse strains.

For all animals, the postchallenge day of death or of euthanasia for animals exhibiting signs of serious illness was recorded and the C. albicans tissue burden in both kidneys and brain was determined at the time of demise by viable counting of 100-μl samples homogenized in 0.5-ml volumes of sterile saline. Detection limits for tissue burdens under these conditions were 150 CFU (kidneys) and 80 CFU (brains); culture-negative samples were scored as 0.5 log₁₀ unit lower than these limits for statistical purposes. All experiments ended 28 days after challenge; animals surviving at this time were humanely killed, and tissue burdens were determined by plating on Sabouraud agar.

RESULTS

OMP decarboxylase activity correlates with URA3 mRNA abundance.

Ura3p specific activities were measured for cell extracts from a range of strains created with the Ura-blaster methodology and containing one or two copies of URA3. OMP decarboxylase activity reflected the URA3 gene copy number in the control strains SC5314 and CAF2-1 and was not affected by growing the cells in increasing concentrations of exogenous uridine (Table 3). No activity was detected in strain CAI-4, consistent with the lack of a functional URA3 allele in this strain. In mutant strains derived from the SC5314-CAF2-1-CAI-4 lineage, OMP decarboxylase activity varied depending on the genomic location of the URA3 gene and was lower than that of CAF2-1 in all mutants. However, OMP decarboxylase activity was restored to the level of CAF2-1 when URA3 was reintegrated at the RPS10 locus by using CIp10. There was one exception (NGY145), which had significantly higher activity than CAF2-1 (Student's t test; P = 0.022), although this was only 73% of the activity of SC5314. A selection of five wild-type clinical isolates of C. albicans had OMP decarboxylase specific activities of the same order as that of SC5314 (Table 3). Therefore, the chromosomal location of URA3 significantly influenced Ura3p specific activity.

TABLE 3.

OMP decarboxylase activity in control and mutant strains that were grown in SDG or NGY, with and without uridine supplementation

C. albicans strain	OMP decarboxylase activity (nmol/mg/min)a	Activity relative to SC5314 (%)	Strain transformed with Clp10	OMP decarboxylase activity (nmol/mg/min)a	Activity relative to SC5314 (%)
SC5314	29.7 ± 3.7	100.0
CAF2-1	15.4 ± 1.4	51.7
CAF2-1 + uridine (25 μg/ml)	14.7 ± 0.8	49.5
CAF2-1 + uridine (250 μg/ml)	15.0 ± 2.0	50.5
CAI-4	<1.0b	<1.0b	NGY152	16.7 ± 1.3	56.2
CAI-4 + uridine (25 μg/ml)	<1.0b	<1.0b	NGY152	13.0 ± 1.2	43.8
CAI-4 + uridine (250 μg/ml)	<1.0b	<1.0b	NGY152	13.0 ± 2.8	43.8
NGY23 (mnt1Δ)	8.8 ± 0.5	29.6	NGY158	16.8 ± 0.7	56.6
NGY105 (mnt2Δ)	13.6 ± 1.5	45.8	NGY145	21.6 ± 2.9	72.7
NGY63 (mnt3Δ)	2.6 ± 0.8	8.7	NGY146	12.9 ± 1.9	43.4
NGY87 (mnt4Δ)	2.7 ± 1.1	9.1	NGY313	14.4 ± 1.4	48.5
NGY71 (mnt5Δ)	5.9 ± 0.3	19.9	NGY147	17.7 ± 1.2	59.6
DH5 (mnn4Δ)	7.0 ± 0.9	23.6	DH15	12.3 ± 1.1	41.4
GTC43 (gcn4Δ)	8.8 ± 0.9	29.6	GTC45	15.5 ± 0.8	52.1
BH1-1-1 (sap3Δ)	14.1 ± 0.9	47.5	VC3.9	15.4 ± 0.0	51.9
73/034	38.7 ± 0.2	130.3
AM2003-020	29.4 ± 1.0	99.0
3153	25.0 ± 3.2	84.2
B2360	24.9 ± 2.2	83.8

^aResults are expressed as mean ± standard error of the mean.

^bLimit of detection.

URA3 expression levels depend on chromosomal location but are restored to the CAF2-1 equivalent by reintegration at the RPS10 locus.

To establish whether the expression of ectopically placed URA3 was influenced by the normal level of transcriptional activity at target loci, MNT5, MNN4, and GCN4 mRNA abundance was measured relative to URA3 mRNA abundance for strain CAF2-1 (URA3/ura3) by semiquantitative RT-PCR. The results (Table 4) showed that GCN4 mRNA levels were considerably higher than those for URA3. The MNN4 mRNA levels were similar to those of URA3, and expression of MNT5 was 25-fold lower than that of URA3.

TABLE 4.

Expression of C. albicans genes and expression of URA3 in ectopic gene loci measured on the basis of relative RNA abundance by RT-PCR

Gene	Expression of target genes in CAF2-1a	URA3 expression at different loci		URA3 expression at the RPS10 locus
		Strain	URA3 mRNA abundance relative to CAF2-1	Strain	URA3 mRNA abundance relative to CAF2-1
URA3		CAF2-1		NGY152 (CA1-4/CIp10)	Increased 1.34-fold
MNT5	Decreased 26-fold	NGY71 (mnt5Δ)	Decreased 6.8-fold	NGY147 (mnt5Δ/CIP10)	Decreased 1.33-fold
MNN4	Decreased 0.6-fold	DH5 (mnn4Δ)	Decreased 2.0-fold	DH15 (mnn4Δ/Clp10)	Decreased 1.33-fold
GCN4	Increased 435-fold	GTC43 (gcn4Δ)	Decreased 8.0-fold	GTC45 (gcn4Δ/Clp10)	Decreased 2.0-fold

^aMeasured as mRNA abundance relative to URA3 mRNA.

Notwithstanding these differences in the expression of the native genes, URA3 mRNA abundance was relatively low at all of these loci compared with URA3 expression at its native locus (in CAF2-1 [Fig. 1; Table 4]). This indicated that expression levels of URA3 at ectopic loci do not reflect the expression level of the gene native to each locus. When CIp10 was used to transform the mnt5, mnn4, and gcn4 mutants, producing strains with URA3 at the RPS10 locus, URA3 mRNA abundance was restored essentially to the same levels as in CAF2-1 (Fig. 1). Hence, the reinsertion of URA3 at the RPS10 locus restored URA3 mRNA to normal (heterozygote) levels, regardless of which chromosomal locus had been disrupted by Ura-blasting. There was a strong positive correlation between URA3 mRNA level and OMP decarboxylase activity in the same strains (Tables 3 and 4; r = 0.75).

FIG. 1.

URA3 mRNA abundance when URA3 is positioned at its native locus (CAF2-1) (A), the GCN4 locus (GTC43) (B), or the RPS10 locus after reintegration using the CIp10 vector (GTC45) (C). cDNA was synthesized from mRNA isolated from late-exponential-phase cells. URA3 mRNA abundance was measured relative to an internal control gene, EFB1, using multiplex RT-PCR.

Influence of URA3 expression on virulence of C. albicans mutants.

To determine the importance of URA3 expression levels for the virulence of C. albicans, a series of strains derived from CAI-4 were tested by i.v. challenge in DBA/2 mice for differences in survival times and tissue burdens of C. albicans. The strains tested carried a single copy of URA3: CAF2-1 (URA3/ura3), NGY152 (CAI-4/CIp10), GTC41 (GCN4/gcn4Δ), GTC43 (gcn4Δ), GTC45 (gcn4Δ/CIp10), GTC49 (gcn4Δ/CIp10-GCN4). The results are shown in Fig. 2 and Table 5. A marked reduction in virulence for both the heterozygous GCN4/gcn4Δ mutant (GTC41) and the homozygous gcn4Δ mutant (GTC43) was found relative to the parent strain (CAF2-1). The effect was equally evident from both the long survival (Fig. 2) and the relatively low tissue burdens (Table 5) of mice challenged with these strains. However, integration of CIp10 containing only the URA3 gene into the gcn4Δ null mutant (GTC45) restored wild-type virulence, with survival curves and tissue burdens matching those of CAF2-1 and NGY152. Reintegration of one copy of GCN4 at the RPS10 locus using CIp10 also produced a mutant with wild-type virulence restored.

FIG. 2.

Survival curves for female DBA/2 mice challenged iv. with C. albicans CAF2-1 (solid circles) and the following mutants derived from this strain: NGY152 (solid squares), GTC41 (GCN4/gcn4Δ; open circles), GTC43 (gcn4Δ; open squares), GTC45 (gcn4Δ/CIp10; solid triangles) and GTC49 (gcn4Δ/CIp10-GCN4; open triangles). Mice were in groups of 6 animals, except for CAF2-1, with 10 animals challenged.

TABLE 5.

Tissue burdens of C. albicans in left kidneys and brains from mice challenged i.v. with mutants derived from CAF2-1^a

Strain	C. albicans burden (log10 CFU/g)b in:
	Left kidney	Brain
CAF2-1	6.6 ± 0.6	4.4 ± 0.5
NGY152 (CAI-4/CIp10)	6.4 ± 0.6	4.3 ± 0.4
GTC41 (GCN4/cgn4Δ)	3.2 ± 1.2	2.9 ± 1.1
GTC43 (gcn4Δ)	3.2 ± 1.2	1.7 ± 1.0
GTC45 (gcn4Δ/CIp10)	6.5 ± 0.2	4.5 ± 0.1
GTC49 (gcn4Δ/CIp10/GCN4)	6.7 ± 0.1	4.3 ± 0.2

^aThere were 6 mice per group except for CAF2-1, which had 10 mice per group.

^bResults are expressed as mean ± standard deviation.

Figure 3 shows scatter plots of OMP decarboxylase specific activities for a range of C. albicans clinical isolates and mutants versus their mean survival times in the mouse and versus mean fungal burdens in the kidneys and brains of infected animals. For mutants derived from the SC5314-CAF2-1-CAI-4 lineage by Ura-blasting, a weak negative association was apparent between relative OMP decarboxylase specific activity and mean survival time (r = −0.38) and a weak positive association was apparent between enzyme activity and kidney burdens (r = 0.40) and brain burdens (r = 0.38). Data for strains RV4688 and 73/034 show clearly that strains of relatively low virulence can have wild-type OMP decarboxylase levels. This suggests that factors other than OMP decarboxylase levels influence the virulence of these strains.

FIG. 3.

Scatter plots of OMP decarboxylase specific activity in wild-type and mutant C. albicans strains versus three parameters of virulence in a mouse i.v. C. albicans challenge model: mean survival time (a), mean kidney burden (b), and mean brain burden (c). Each point represents the result with a single strain tested for BALB/c mice (open circles) or DBA/2 mice (solid circles) in groups of 6 animals, except for CAF2-1, NGY152, SC5314, BH1-1-1, and VC3.9, where group sizes of 12 to 30 animals were pooled from two to five replicate experiments. The strains tested were all derived originally from SC5314, whose OMP decarboxylase activity was taken as 100%. Details of the strains are shown in Table 1. Additional results for five natural isolates of C. albicans that differed in inherent virulence are individually identified on the graphs as solid squares; their virulence was tested in BALB/c mice. Data for brain burdens were not available for ATCC 44858.

Proteomic analysis of SC5314 and CAI-4.

To investigate the global effects of inactivating the URA3 gene, we compared the proteomes of SC5314 and CAI-4 cells grown in YPD supplemented with uridine. Because the gene deletion in CAI-4 includes a portion of IRO1 as well as URA3 (31), we also examined the proteome of strain NGY152, in which one copy of URA3 is reintegrated at the RPS10 locus. As predicted, Ura3p was present on SC5314 gels at a position consistent with its pI and molecular mass (pI = 5.24, molecular mass = 24 kDa) but was not detected on CAI-4 gels. Exogenous uridine was supplied for all strains to facilitate growth of CAI-4. The presence of Ura3p in SC5314 gels under these growth conditions was consistent with the finding in this study that exogenous uridine supplied to growing cells did not affect OMP decarboxylase activity. Fourteen other proteins showed altered expression in CAI-4 (Table 6). Levels of all proteins except the transcription factor, Toa2p, were partially or fully restored by the reintegration of a single copy of URA3 in strain NGY152, indicating that the deletion of URA3, and not the truncation of IRO1, accounts for the majority of proteomic changes observed for CAI-4. Of the 13 proteins displaying URA3-dependent changes, 3 (Hpt1p, Ade2p, and Ura5p) are involved in purine and pyrimidine metabolism. These pathways may be perturbed by an altered flux of metabolic intermediates in the absence of OMP decarboxylase, the last of six enzymes in the uridine de novo biosynthetic pathway. Other changes in expression suggest that heme biosynthesis (Ald5p and Hem13p) and aromatic amino acid turnover (Aro8p and Aro10p) may also be perturbed by the loss of Ura3p. Another group of proteins with related function that showed altered expression in CAI-4 comprised Rps12p, Eft2p, and IPF6037p, which are involved in translation or transcription. Therefore, inactivation of URA3 appears to influence C. albicans metabolism in unexpected ways.

TABLE 6.

Effect of URA3 deletion on the cytosolic proteomer^a

Protein (Reference)	Change in expression in CAI-4	Expression in NGY152 (URA3 reintegrated at RPS10 locus)
Ura3p (33)	Not detected	Partially restored
Hpt1p (34)	Not detected	Fully restored
Ald5p (2 spots) (46, 68)	Only 1 spot present	Fully restored
Sgt2p (22, 45)	Only 1 spot present	Fully restored
Pmm1p (78)	4.5-fold increase	Fully restored
IPF6037	4.5-fold increase	Fully restored
Aro8p (40)	3.8-fold increase	Fully restored
Ade2p (73)	3.8-fold increase	Partially restored
Ura5p (25)	3.3-fold increase	Fully restored
Eft2p (55)	2.9-fold increase	Partially restored
Aro10p (87)	2.9-fold increase	Partially restored
Hem13p (96)	2.1-fold increase	Partially restored
Rps12p (63)	4.2-fold decrease	Fully restored
Toa2p (43)	3.2-fold decrease	Not restored (same as CAI-4)
IPF4328	2.6-fold decrease	Fully restored

^aResults are shown for proteins with a greater than twofold change in expression between SC5314 and CAI-4. Cells were grown in YPD with 25 μg of uridine per ml at 30°C to OD₆₀₀ = 0.6. Total solubilized protein was resolved by 2D gel electrophoresis and subjected to proteomic analysis. Proteins were identified by MALDI-TOF mass spectrometry and interrogation of the CandidaDB database.

DISCUSSION

We show that the chromosomal location of URA3 has a significant impact on its expression, the resulting specific activity of the gene product, the cellular proteome, and consequently the virulence of C. albicans. The Ura-blaster approach to disruption of diploid genes is highly effective as a molecular tool, but phenotypic alterations recorded in mutants created by Ura-blasting cannot be reliably attributed to the target gene unless care is taken to ensure that the mutant expresses adequate levels of Ura3p. Ectopic expression of URA3 from within a disrupted gene cannot be depended on to provide adequate Ura3p levels, presumably because of positional effects of the target locus on URA3 gene expression. Only by specific insertion of the marker gene at a common genomic locus in different strains can a scientifically comparable set of parents and mutants be created. Sundstrom and her colleagues have recommended (81) and implemented (4, 5) specific insertion of the URA3 gene at the ENO1 locus as a means of circumventing positional effects on URA3 expression. Our own preference is for placement of URA3 at the RPS10 locus, since the plasmid CIp10 makes the necessary transformation technically very simple and we have confirmed that this locus enables near-normal levels of gene expression to occur.

One area of study in which the difficulties associated with ectopic expression of URA3 may have greatly affected the outcome of experiments is in the delineation of molecular virulence factors in C. albicans. At least 70 C. albicans genes (Table 7) have been implicated in virulence on the basis of evidence of attenuation in the mouse i.v. challenge model for Ura-blasted homozygous null mutants. However, in the majority of these animal studies, URA3 was expressed from within the locus of the disrupted gene (Table 7). We show a dramatic example of this, where attenuated virulence in C. albicans was attributable to decreased URA3 expression rather than disruption of the GCN4 locus (Fig. 2, compare NGY152 [wild type], GTC43 [gcn4Δ] and GTC45 [gcn4Δ/CIp10]). Our evidence supporting a role for Gcn4p in virulence satisfies the molecular Koch's postulates. However, the virulence of the mutant containing only URA3 reintegrated at the RPS10 locus proves that the attenuated phenotype in the homozygous null mutant was the result of ectopic expression of URA3 and not of deletion of GCN4 (Fig. 2, compare NGY152 [wild type], GTC45 [gcn4Δ/CIp10], and GTC49 [gcn4Δ/CIp10-GCN4]). We have similar data for two other C. albicans loci (unpublished data). It seems likely that many of the attenuated virulence phenotypes attributed to particular genes in the studies listed in Table 7 may also have resulted from (inadequate) ectopic expression of URA3, and hence these virulence phenotypes require reexamination. In some of the instances tabulated, the evidence for a virulence role of the gene concerned has been confirmed in experiments not dependent on the ectopic expression of URA3, including reintegration of a disrupted gene without altering the ectopic locus of URA3 (11), but there remain many examples where this is not the case. Based on the analysis of the number of strains in which Ura3p specific activities were below 40% of that found when the URA3 was at the native locus and which were significantly attenuated as measured by mean survival time or kidney burdens (Fig. 3), we estimate that the virulence of at least 30% of all mutants generated to date may have to be formally reassessed after correction for ectopic URA3 expression. Of the 73 genes for which attenuated virulence has been associated with specific disruptions (Table 7), at least 60 were tested in experiments where URA3 was expressed ectopically. Our data suggest that more than 20 of these genes may have been incorrectly designated virulence factors in C. albicans.

TABLE 7.

Publications on C. albicans describing attenuated virulence related to specific gene disruptions

Gene	Function	Locus containing URA3a	Reference(s)
AAF1	Unknown function	AAF1	66
ADE2	Phosphoribosylaminoimidazole carboxylase	ADE2	27
ALO1	d-Arabinono-1,4-lactone oxidase	ALO1	36
ALS1	Surface adhesin	ALS1	29
ASH1	Transcription factor	ASH1	39
BGL2	(1-3) β-Glucosyltransferase	BGL2	71
CAP1	Adenyl cyclase regulator	eno::URA3	4, 5
CAT1	Catalase	CAT1	91
CDC10	Septin	ARG4	88
CDC11	Septin	ARG4	88
CDC24	GDP:GTP exchange factor	RPS10	8
CDC35	Adenyl cyclase regulator	part of plasmid pVEC	67
CDC42	G protein (GTPase)	RPS10	8
CHS1	Chitin synthase	CHS1 under mal promoter	58
CHS3	Chitin synthase	CHS3	14, 58
CLA4	Ser/Thr protein kinase	CLA4	50
CNA	Calcineurin A	CNA	70
CNB1	Calcineurin B	CNB1	11
COS1	Histidine kinase	COS1?	75
CPP1	Protein phosphatase	CPP1	20
CRK1	Cdc2 kinase	CRK1	18
CSP37	37-kDa surface protein	CSP37	76
CST20	MEKK kinase	CST20	49
EFG1	Transcription factor	EFG1	52
FAS2	Fatty acid synthase	FAS2	98
FTR1	High-affinity iron permease	FTR1	64
GNA1	Glucosamine-6-phosphate acetyltransferase	GNA1	56
GP17	Transcription factor	GP17	65
HEM3	Uroporphyrin I synthase	URA3	44
HK1	Histidine kinase	HK1	17, 93
HOG1	Mitogen-activated protein kinase	MKC1	2
HWP1	Transglutaminase	HWP1, ENO1	21, 80, 82, 85
ICL1	Isocitrate lyase	ICL1	53
INT1	Surface protein	INT1	10, 30
KEX2	Kexin (subtilase)	KEX2	62
LIG4	DNA ligase	LIG4	3
MAD2	Spindle assembly factor	MAD2	6
MDR1	Membrane efflux pump	MDR1	9
MKC1	Mitogen-activated protein kinase	MKC1	26
MNT1	α-1,2-Mannosyl transferase	MNT1	15
NAG1	N-Acetylglucosamine-6-phosphate deaminase	NAG1	77, 94
NAG2	N-Acetylglucosamine phosphate deacetylase	NAG2	94
NAG6	N-Acetylglucosamine kinase	NAG6	94
NIK1	Histidine kinase	NIK1	93
NMT1	N-Myristoyl transferase	NMT1	89
NRG1	Transcriptional repressor	NRG1	60
PHR1	pH-regulated expression	PHR1	24, 32
PLB1	Phospholipase B	PLB1	51, 57
PMT1	Protein O mannosylation	PMT1	83
RAS1	Regulates gene expression	Part of plasmid pVEC	49
RBT1	Hypha-specific wall protein	RBT1	12
RBT4	Unknown function	RBT4	12
RFG1	Transcription factor	RFG1	42
RIM101	Transcription factor	NotI-digested pRS-ARG-URA-BN	23
RIM8	Transcription factor	NotI-digested pRS-ARG-URA-BN	23
RSR1	GTPase	RSR1	92
SAP1	Aspartyl proteinase	SAP1	35
SAP2	Aspartyl proteinase	SAP2	35
SAP3	Aspartyl proteinase	SAP3	35
SAP4-6	Aspartyl proteinases	SAP5	69
SLN1	Histidine kinase	SLN1	93
SOD1	Cu/Zn superoxide dismutase	SOD1	38
SPT3	Transcriptional activator	SPT3	47
SSK1	Two-component response regulator	SSK1	16
SSN6	Transcriptional corepressor	SSN6	37
TEC1	Hyphal transcription factor	Part of plasmid pVEC	74
TOP1	Topoisomerase	Upstream of TOP1, part of inducible promoter construct	41
TPK2	Protein kinase A subunit	TPK2	79
TPS1	Trehalose phosphate synthase	TPS1	97
TPS2	Trehalose phosphate phosphatase	TPS2	86
TUP1	Transcription regulator	TUP1	60
URA3	Nucleotide synthesis		44
VPS34	Phosphoinositol-3-kinase	VPS34	13

^aThe URA3 position in the mutants is indicated according to the information published.

Ura3p enzyme activities correlated well with URA3 mRNA levels in C. albicans mutants. Our results relating relative OMP decarboxylase activity with mean survival time (Fig. 3) are consistent with those of Lay et al. (48). However, any relationship between the enzyme activity and virulence can be interpreted only for mutants in the C. albicans SC5314 lineage. The existence of naturally attenuated wild-type strains with normal levels of OMP decarboxylase shows that a virulence phenotype (as measured by mouse mean survival times) is polygenic. Our experiments with an isogenic set of mutants have allowed us to determine specifically the influence of URA3 expression on virulence. We interpret the data in Fig. 3 as indicating that a minimum necessary level of OMP decarboxylase activity (URA3 expression) of around 40% of wild-type (SC5314) levels is required for an SC5314-derived mutant used in a mouse virulence assay.

Iron uptake is thought to be important for the growth and survival of a pathogen such as C. albicans (64). Hence, the fact that the disruption of URA3 in CAI-4 also removed a portion of the IRO1 locus (31) could be argued as another cause of the phenotypic changes that we and others have attributed to URA3 underexpression. However, we could reestablish wild-type virulence in our homozygous gcn4 and mnn4 mutants by reintegrating URA3 alone at the RPS10 locus. Proteomic analysis showed that changes in the expression of only one protein appeared to be URA3 independent, suggesting that its lowered expression in CAI-4 and NGY152 was related to the truncation of IRO1. Taken together, these data strongly suggest that URA3 and not IRO1 is the more important regulator of phenotype, at least for the mouse lethality phenotype.

Our proteomic data (Table 6) show that the expression of 13 proteins is altered in CAI-4 in a URA3-dependent manner, because reintegration of a single copy of URA3 at the RPS10 locus either partially or fully restored expression levels to those of SC5314. While some of the proteins that are collaterally affected by URA3 deletion are obviously related to purine and pyrimidine metabolism (Hpt1p, Ade2p, and Ura5p), others are not. These findings further reinforce the main conclusion of this study and its predecessors, where phenotypic effects resulting from URA3 disruption were demonstrated (7, 19, 48, 81, 82). Namely, the use of auxotrophic selectable markers can result in misleading phenotypes that arise from unpredicted collateral effects on cell metabolism and physiology.