Characterization and functional analysis of the siderophore-iron transporter CaArn1p in Candida albicans.

Siderophores are small organic compounds with high affinity for ferric iron. Microorganisms commonly acquire iron via siderophore secretion and uptake. Here we report the characterization of the siderophore transporter CaArn1p in the fungal pathogen Candida albicans. Deletion of CaARN1 reduced the ability of C. albicans to use iron bound to the hydroxamate-type siderophore ferrichrome and abolished it when two high-affinity iron permease genes (CaFTR1 and CaFTR2) were also deleted, indicating a role of CaArn1p as well as the permeases in ferrichrome-iron uptake. Caarn1Delta (but not Caftr1DeltaCaftr2Delta) assimilated iron from another hydroxamate-type siderophore, ferrioxamine B, suggesting that iron uptake from this compound depends on the permeases, but not on CaArn1p. Northern blot analysis revealed that the transcription repressor CaTup1p repressed CaARN1 expression under iron-replete conditions via the DNA-binding protein Rfg1p. Green fluorescent protein-tagged CaArn1p was observed predominantly in the plasma membrane, with some in the cytoplasm as distinct spots. The number of these spots increased with the increase in ferrichrome concentration, suggesting that CaArn1p internalization might be a mechanism for ferrichrome-iron uptake or for recycling the transporter. Caarn1Delta did not show reduced virulence when injected into the blood stream of mice, implying that CaArn1p is not required for iron uptake along this route of infection.

Iron is an essential nutrient for nearly all organisms because it serves as an obligate component of many indispensable enzymes and other proteins (1). As a general antimicrobial defense mechanism, mammals possess a sophisticated iron-withholding system consisting of high-affinity iron-binding proteins in body fluids, iron uptake proteins on the cell surface, and iron storage proteins inside the cells (1,2). This system effectively maintains the free iron concentration in body fluid at an extremely low level insufficient for microbial growth. However, microbial pathogens have collectively evolved a diverse repertoire of highly effective iron acquisition mechanisms, some of which appear to be specifically "designed" to defeat the ironwithholding system or to target iron-rich niches of the host (3)(4)(5)(6). For example, pathogenic Neisseria species have surface receptors specific for the host extracellular iron-binding proteins transferrin and lactoferrin, turning the two major components of the host iron-withholding system into their iron sources (7,8). Candida albicans possesses mechanisms for attracting and lysing erythrocytes and utilizing the iron in hemoglobin (9,10). These iron acquisition schemes constitute important virulence factors. A more commonly used strategy for microorganisms to acquire iron is the secretion of highaffinity iron chelators called siderophores (11)(12)(13)(14)(15). Siderophores are able to extract iron from host iron-binding proteins, and the siderophore-iron complexes are then assimilated by the pathogens via specific receptors and transporters. Some microorganisms such as Saccharomyces cerevisiae and pathogenic Neisseria species do not produce their own siderophores, but are able to use the ones secreted by other organisms (7,16).
C. albicans is currently the most prevalent human fungal pathogen, causing infections in immunocompromised hosts (17). In some patients, C. albicans cells enter the blood stream, establishing fatal systemic infection. How does C. albicans acquire iron for growth in the iron-restricted blood stream and internal organs? We have previously shown that the highaffinity iron permease CaFtr1p is indispensable for C. albicans to establish systemic infection in mice (18). However, there is no evidence that CaFtr1p or its interacting partner oxidase, CaFet3p, can withdraw iron directly from iron-binding proteins of the host. According to studies of iron uptake mechanisms in S. cerevisiae and other organisms, there are at least two possible ways by which C. albicans may dissociate iron from the iron-binding proteins. First, C. albicans produces extracellular ferric reductases, which may convert the bound ferric iron to unbound ferrous iron. The free ferrous iron is then transported into the cells by the high-affinity iron uptake system (1, 19 -21). Second, C. albicans may produce siderophores, which directly extract the bound ferric iron. Genes encoding ferric reductases have been cloned from C. albicans (22), and there have been several reports on possible siderophore production by C. albicans (23)(24)(25). Either the siderophore-bound iron may be released by the ferric reductases and then picked up by the high-affinity uptake system (18, 26 -28), or the siderophoreiron complex is transported into the cell via receptor-mediated mechanisms (28,29). Thus, it is likely that the siderophoremediated iron binding may function upstream of and provide free iron to the oxidase-permease iron uptake system. It has been reported that disruption of tonB in Bordetella bronchiseptica and Bordetella pertussis prevents utilization of ferric siderophores, hemin, and hemoglobin as iron sources (30). Thus, it is possible that the siderophore uptake mechanism is related to the use of hemoglobin iron by C. albicans.
Four siderophore transporters (Arn1p, Arn2p/Taf1p, Arn3p/ Sit1p, and Arn4p/Enb1p) have been reported in S. cerevisiae, collectively responsible for the uptake of a range of siderophores of different structures and chemical properties (16,(31)(32)(33)(34). These proteins belong to the major facilitator superfamily, whose members have been found in many organisms. Currently, one siderophore transporter gene (CaARN1) has been found in C. albicans, and its function has been investigated by heterologous expression in S. cerevisiae (28,29). Ardon et al. (29) reported that CaArn1p expressed in S. cerevisiae specifically mediates the uptake of FC 1 -iron, but does not transport iron bound to FOB. This result seems to contradict the report by Lesuisse et al. (28), who observed intracellular accumulation of fluorescent analogs of both FC and FOB in C. albicans.
In this study, we constructed C. albicans mutants with deletions of either CaARN1 alone or together with the highaffinity iron permease genes and compared them for iron uptake from different siderophores. We also investigated the cellular localization of CaArn1p, identified transcription regulators of CaARN1 expression, and examined the virulence of the CaARN1 deletion mutant using a mouse model for systemic candidiasis.

C. albicans Strains and Growth
Conditions-The C. albicans strains used in this study are listed in Table I. The C. albicans strains were routinely grown in either YPD medium (1.0% yeast extract, 2.0% peptone, and 2.0% glucose) or GMM medium (yeast nitrogen base without amino acids and with 2.0% glucose). For the growth of strains defective in uracil, arginine, or histidine synthesis, the required nutrient was added to GMM medium accordingly. Iron-depleted media were prepared by adding 100 -500 M BPS to YPD, GMM, or YNB/glucose medium (identical to GMM medium, but iron-limiting; Bio 101, Inc.). To prepare iron-replete media, 100 -200 M FeCl 3 or FeSO 4 was added to GMM or YNB/glucose medium.
Chromosomal Deletion of CaARN1-The two copies of the CaARN1 gene were sequentially deleted from strains CAI4 and CaWYRN2 (Caftr1⌬Caftr2⌬) using the URA blaster method (35). The gene deletion strategy is shown in Fig. 1A. PCR primers were designed based on the genome sequence released by the C. albicans Genome Sequence Project at Stanford University. 2 Primers 5Ј-TTCTAAGTGCTGCGTTTA-3Ј and 5Ј-GGATCCGATGCCGTTTGAGTTAT-3Ј amplified a 437-bp 5Ј-end fragment (nucleotides 344 -781; the first base of the HindIII site upstream of the CaARN1 coding sequence is designated nucleotide 1), and primers 5Ј-GGATCCTCGCGCTTTAATGTTTAT-3Ј and 5Ј-TGCCATAT-TCTGTTCCAG-3Ј amplified a 502-bp 3Ј-end fragment (nucleotides 2703-3205). During the PCR, a BamHI restriction site (underlined in the primer sequences) was added to the 3Ј-and 5Ј-ends of the two DNA fragments, respectively. The two DNA fragments were joined at the BamHI site in pGEM-Teasy. Then the hisG-URA3-hisG cassette with compatible ends was inserted at the BamHI site of the above plasmid. The Caarn1⌬::hisG-URA3-hisG cassette was released from the plasmid by NotI digestion, gel-purified, and used to transform CAI4 and CaWYNR2 cells by electroporation as previously described (18). Clones of the CaARN1/Caarn1⌬::hisG-URA3-hisG genotype were obtained, and cells were spread onto fluoroorotic acid plates to isolate CaARN1/Caarn1⌬::hisG clones. To disrupt the second copy of CaARN1, a new cassette was constructed in which the hisG tandem repeats were replaced by an 850-bp Escherichia coli chloramphenicol acetyltransferase gene fragment. Also, two new pairs of primers were used to generate the 5Ј-and 3Ј-end CaARN1 fragments, which are internal to the ones used in the hisG-URA3-hisG cassette. The primers used were as follows: 5Ј-TAAAGCCTATAACTCAAA-3Ј and 5Ј-GGATCCCGGTT-GTTAATAAAC-3Ј, which amplified a 413-bp fragment (nucleotides 757-1170), and 5Ј-GGATCCGCATATGGCTGTGATTTC-3Ј and 5Ј-AA-CATTAAAGCGCGAGTA-3Ј, which yielded a 444-bp fragment (nucleotides 2273-2714). Two independent CaARN1/Caarn1⌬::hisG clones were transformed with the Caarn1⌬::cat-URA3-cat cassette to obtain Caarn1⌬::hisG/Caarn1⌬::cat-URA3-cat clones. The URA3 gene was then looped out on fluoroorotic acid plates to produce Caarn1⌬::hisG/Caarn1⌬::cat clones. The genotypes of all the gene deletion mutants were verified by Southern blot analysis as exemplified in Fig. 1B.
Construction of Plasmids for Reintroduction of CaARN1 into Mutant Cells-A genomic DNA fragment containing the CaARN1 open reading frame and ϳ900-bp and ϳ500-bp 5Ј-and 3Ј-flanking regions was PCRamplified and cloned at the XhoI site of the integration plasmid pCIp10 (36) and at the multiple cloning site of the autonomous plasmid pABSK1 (18). The CaARN1 gene was cloned into pCIp10 containing the RP10 gene for integration and CaURA3 as a selection marker. The resultant plasmid was linearized by cleavage at the unique StuI site in RP10 and transformed into the Ura Ϫ homozygous Caarn1⌬ mutant (CaWY3-3). To tag CaArn1p with GFP, the GFP coding sequence was joined in-frame to the C-terminal end of CaARN1 by PCR to introduce compatible restriction sites to the ends of each DNA fragment. The reading frame was verified by sequencing. Chromosomal Deletion of CaTUP1, RFG1, and CaNRG1 Genes-CaTUP1, RFG1, and CaNRG1 deletion mutants were all derived from strain BWP17 (37) using the same gene deletion strategy. A gene deletion cassette was constructed by flanking a marker gene, CaHIS3 or CaURA3, with two DNA fragments corresponding to the 5Ј-and 3Ј-untranslated regions of the target gene, respectively. For sequentially deleting the two copies of a gene, two cassettes containing different selection markers were constructed for each gene. For example, Catup1⌬::HIS3 and Catup1⌬::URA3 cassettes were used sequentially to delete the CaTUP1 gene. The following pairs of primers were used to PCR-amplify the 5Ј-and 3Ј-flanking DNA fragments of the three genes: the 5Ј-fragment of CaTUP1, 5Ј-AAGTCAGACATTACTGAG-3Ј and 5Ј-ATGGAAGAAGTTGTGTTG-3Ј; the 3Ј-fragment of CaTUP1, 5Ј-AAAA-AATAAGTGTGTAG-3Ј; the 5Ј-fragment of RFG1, 5Ј-CATTTGGGATA-CTACTAC-3Ј and 5Ј-GCAGTAGACATAATAAATTG-3Ј; the 3Ј-fragment of RFG1, 5Ј-ACCTCCACAATAATTTCTAC-3Ј and 5Ј-GCAACTA-TATGGATTAACCC-3Ј; the 5Ј-fragment of CaNRG1, 5Ј-CAACTAGGG-ATTCATCAT-3Ј and 5Ј-gaaactagcagggaaaaa-3Ј; and the 3Ј-fragment of CaNRG1, 5Ј-GGTTAAATTTGGATGG-3Ј and 5Ј-AGAAGATCTATGGC-AATG-3Ј. To join the 5Ј-and 3Ј-fragments to the selection marker gene, appropriate restriction sites were added to the PCR primers. Transformants were selected on histidine-or uracil-dropout medium according to the selection marker used. Correct deletion of each gene was verified by Southern blotting and examination of phenotypes characteristic of each mutant.
Siderophore-Iron-dependent Growth Assay-Desferri-FC and desferri-FOB were purchased from Sigma. Ferric FC and ferric FOB were prepared by overnight incubation of the desferric siderophore with FeCl 3 at an equal molar concentration as described (16). Agar plates were prepared using the iron-limiting YNB/glucose medium supplemented with 500 M BPS, followed by the addition of different amounts of siderophore. C. albicans strains to be tested were first grown in YPD medium to saturation. The cells were washed twice with and diluted to a density of 1 ϫ 10 5 cells/ml in YNB/glucose medium containing 100 M BPS and grown for an additional 5 h. 10-fold serial dilutions of the cultures were prepared, and cells were spotted onto the agar plates. The plates were incubated at 30°C for 2 days before photographing.

59
Fe Uptake Assay-59 Fe uptake assay was performed as described previously (18) with some modifications. C. albicans strains were first grown in YPD medium to saturation and then washed twice with 10 mM EDTA and once with LIM0 medium (18) containing 100 M BPS. The cells were resuspended in LIM0 medium to a density of 1 ϫ 10 7 cells/ml and grown for 5 h. Cells were then counted; spun down; and divided into 50-l aliquots in LIM0 medium, each containing ϳ2 ϫ 10 7 cells. 59 Fe, FC ϩ 59 Fe, or FOB ϩ 59 Fe were added to each aliquot to a final concentration of 2 M, and cells were incubated at 30°C for 30 min. One such aliquot of each strain was kept on ice for 30 min to estimate the background binding of 59 Fe to the cell surface. After incubation, cells were washed twice with ice-cold 10 mM EDTA to remove unincorporated 59 Fe, spun down, and air-dried briefly before determining the amount of intracellular accumulation of 59 Fe using an Amersham Biosciences Compugamma counter (Model 1282). After subtracting the background binding, the iron uptake was expressed as picomoles of 59 Fe/1 ϫ 10 6 cells/min.
Northern Blot Analysis-C. albicans strains were grown in YPD medium at 30°C to saturation, and each strain was used to inoculate an iron-rich and an iron-limiting medium at a density of 1 ϫ 10 5 cells/ml. The iron-rich medium was GMM medium supplemented with 200 M FeCl 3 , and the iron-limiting one was GMM medium containing 100 M BPS. The cells were grown for 8 h before harvesting for the preparation of total RNA. RNA extraction and Northern blotting were performed as described (18). The coding regions of the CaARN1 and CaACT1 genes were PCR-amplified from SC5314 genomic DNA and used as probe.
Western Blot Analysis-Cells were grown to 1 ϫ 10 7 cells/ml in 5 ml of GMM medium containing either 200 M FeCl 3 or 100 M BPS at 30°C. Cells were spun down by centrifugation at 3000 rpm for 5 min. The cell pellet was resuspended in 200 l of ice-cold lysis buffer containing 1% Triton X-100, 0.1% SDS, 50 mM Tris (pH 7.2), 1% sodium deoxycholic acid, and one protease inhibitor mixture tablet (Roche Molecular Biochemicals)/25 ml. The cell suspension was transferred to a 2-ml screw-cap tube. An equal volume of glass beads was added, and the cells were broken using a Mini-Beadbeater (Biospec Products, Inc.) at maximum speed for two cycles of 2-min beating, leaving the sample on ice for 2 min in between. The lysate was spun at maximum speed in an Eppendorf microcentrifuge, and the supernatant was transferred to a new 1.5-ml tube. Protein concentration was determined by the Bradford assay. The proteins were resolved by denaturing polyacrylamide gel electrophoresis and transferred onto Hybond-C nitrocellulose mem-brane (Amersham Biosciences). For Western analysis, we used ECL Western blot detection reagents (Amersham Biosciences) following the manufacturer's procedure. The primary anti-GFP antibody (JL-8) was purchased from CLONTECH and used at 1:500 dilution. The secondary antibody was a horseradish peroxidase-conjugated sheep anti-mouse antibody (NA931, Amersham Biosciences) used at 1:2000 dilution.
Mouse Systemic Candidiasis Model-Female BALB/c mice (Jackson ImmunoResearch Laboratories, Inc.) 8 weeks old were used in the virulence test. Cells from each C. albicans strain were grown in YPD medium to log phase and washed three times and resuspended in PBS at a concentration of 1 ϫ 10 7 cells/ml. For each strain, two groups of eight mice were used; one group was inoculated with 1 ϫ 10 6 cells/ animal, and the second with 2 ϫ 10 6 cells through the tail vein. The animals were monitored for death over a period of 35 days. Eight animals were used for each strain tested at each inoculum. Two animals from each group were killed for histological examination of C. albicans growth in the kidney. Kidney infection was examined as described (18).

RESULTS
Identification of the C. albicans Siderophore Transporter Gene CaARN1-To identify C. albicans siderophore transporter genes, we used ARN1-4 sequences to BLAST-search the C. albicans genome sequence for homologous genes. This search identified one open reading frame conceptually encoding a 604-amino acid polypeptide. Amino acid sequence alignment of this sequence with the known Arnp sequences showed significant homology over the entire alignment, and the highest identity value of 46% was scored with Arn1p. During the course of this study, two other groups published the same gene discovered using similar strategies (28,29) and demonstrated that the encoded protein indeed functions as a siderophore transporter by heterologous expression in S. cerevisiae. Thus, in this study, we only report the characterization of this gene, CaARN1, in C. albicans.
Chromosomal Deletion of CaARN1-To study the effect of deleting CaARN1 alone and together with genes encoding the high-affinity iron uptake system, we used the URA blaster strategy (Fig. 1A) to sequentially delete the two copies of the gene from strain CAI4 and from a Caftr1⌬Caftr2⌬ mutant (CaWYNR2) that is completely defective in high-affinity iron uptake mediated by the oxidase-permease complex (18). The correct deletion of each copy of the gene was verified by Southern blot analysis (Fig. 1B).
Caarn1⌬ Mutant Exhibits a Growth Defect on Iron-limiting Medium Supplemented with Ferric FC-Ardon et al. (29) showed that when expressed in S. cerevisiae, CaArn1p is specific for FC-iron, but does not transport FOB-iron. Thus, we chose FC to test the function of CaArn1p in C. albicans and also included FOB for comparison. YNB/glucose medium was prepared by the manufacturer to contain a low amount of iron. However, all the mutant strains (Caarn1⌬, Caftr1⌬Caftr2⌬, and Caftr1⌬Caftr2⌬Caarn1⌬) were found to grow as well as the wild-type strain on the agar plates prepared using this medium. Thus, we added increasingly higher amounts of the nonpermeable ferrous iron chelator BPS to the medium to determine a concentration at which the wild-type strain exhibited little growth after 2-3 days of incubation at 30°C unless FC was added. We reasoned that at such a concentration, chelation of ferrous iron by BPS would lower the level of iron uptake through the reductive pathway to a minimum level so that it would not interfere much with the assays for iron uptake mediated by the siderophore transporter. At BPS concentrations between 10 and 50 M, the Caftr1⌬Caftr2⌬ mutant exhibited increasingly slower growth (data not shown), whereas the wild-type strain still grew normally. This observation is consistent with the lack of the high-affinity iron uptake system in the mutant strain (18). When the BPS concentration was increased to 500 M, the wild-type strain exhibited little growth after 3 days of incubation. 5-10 M ferric FC, ferric FOB, or FeCl 3 was added to this iron-depleted medium, and serially diluted cultures of each strain were spotted onto the agar plates for siderophore-dependent growth assay. Fig. 2 shows that the wild-type strain grew on the plate supplemented with 10 M FC-iron, but not on the ones supplemented with the same amount of FOB-iron or FeCl 3 , demonstrating that C. albicans utilizes FC-bound iron under this condition. Deletion of CaARN1 abolished this growth, and this growth defect was corrected when a copy of CaARN1 was reintroduced into the Caarn1⌬ mutant. These results indicate that CaARN1 is responsible for FC-iron utilization. Like the wild-type strain, the Caftr1⌬Caftr2⌬ mutant also exhibited FC-dependent growth, and this growth was abolished when the CaARN1 gene was deleted, suggesting that FC utilization does not require the high-affinity iron uptake system under the experimental condition used. We noticed that the Caarn1⌬ mutant consistently exhibited some growth on the plates supplemented with either FC or FOB when the plates were incubated for longer times (data not shown and see below). In contrast, this slow growth was not observed for the Caftr1⌬Caftr2⌬Caarn1⌬ mutant. This observation suggests that iron chelated by FC and FOB can be used via a route dependent on the oxidase-permease iron uptake system under this condition, albeit much less efficiently.
Caarn1⌬ Mutant Is Defective in FC-Iron Uptake-To evaluate iron uptake in a more quantitative manner, we determined the rate of 59 Fe accumulation in cells of various strains. 59 Fe was added to a final concentration of 2 M in the form of either 59 Fe alone or together with an equal molar concentration of FC or FOB. The results are summarized in Fig. 3. In the Caftr1⌬Caftr2⌬Caarn1⌬ mutant, which is defective in both high-affinity iron uptake and siderophore transport, there was little iron uptake. Strikingly, the wild-type cells exhibited more than twice the amount of cellular 59 Fe accumulation when treated with FC ϩ 59 Fe than with 59 Fe alone. When FOB was added with 59 Fe, the wild-type cells also accumulated a significant amount of 59 Fe to a level ϳ80% of that in cells to which only 59 Fe was added. When the CaARN1 gene was deleted, the iron uptake from 59 Fe and FOB ϩ 59 Fe was nearly unchanged, whereas the uptake from FC ϩ 59 Fe was markedly reduced to a level similar to that of the uptake from 59 Fe and FOB ϩ 59 Fe. These results indicate that, first, CaARN1 is responsible for a large fraction of the total 59 Fe uptake from FC ϩ 59 Fe; and that, second, the high-affinity iron uptake system also contributes to the 59 Fe uptake from FC ϩ 59 Fe and FOB ϩ 59 Fe. In the Caftr1⌬Caftr2⌬ mutant, the iron uptake from 59 Fe and FOB ϩ 59 Fe was abolished, indicating that the iron uptake from these sources is entirely mediated by the oxidase-permease highaffinity iron uptake system. In contrast, the iron uptake from FC ϩ 59 Fe was increased in Caftr1⌬Caftr2⌬, suggesting that the siderophore-iron uptake system may function independently of the oxidase-permease system and is enhanced in the absence of the latter. Reintroduction of CaARN1 into Caftr1⌬Caftr2⌬Caarn1⌬ either as a single copy or on pABSK1 restored iron uptake from FC ϩ 59 Fe, but not from 59 Fe and FOB ϩ 59 Fe, confirming that CaARN1 is responsible for the uptake of FC-bound iron. The difference in the levels of iron uptake between cells containing a single copy of CaARN1 and cells containing the gene on pABSK1 can be explained by gene dosage effect.
Deletion of CaARN1 Has Little Effect on Virulence in the Mouse Systemic Candidiasis Model-Because the oxidase-permease iron uptake system was shown to be essential for C. albicans growth and virulence in the host (18), we wanted to determine whether CaARN1 is similarly required. We employed the mouse systemic candidiasis model to examine the virulence of the Ura ϩ Caarn1⌬ mutant. Although the Ura ϩ Caftr1⌬Caftr2⌬ strain was again found to be avirulent, the Caarn1⌬ mutant exhibited similar levels of virulence and kidney infection as the wild-type strain, suggesting that the CaARN1 gene is not required along this route of infection (data not shown).
Iron-responsive Regulation of CaARN1 Expression-A number of genes responsible for iron uptake are known to be regulated by iron concentration, normally being increased when iron supply is limited and repressed when it is replete (16,18,20,28). Northern blot analysis of CaARN1 mRNA levels in cells grown in iron-depleted and iron-replete media revealed the same pattern of iron-responsive expression (see below). In S. cerevisiae, the transcription factor Aft1p has been shown to activate the expression of genes responsible for high-affinity iron uptake (16,38,39). However, so far, an AFT1 homolog has not been found in C. albicans. Recently, Murad et al. (40,41) used DNA array to profile genes that show higher levels of expression in a C. albicans strain with a deletion of the transcription repressor gene CaTUP1. Interestingly, they found that the expression of CaFTR1 and CaCFL2 was increased by 16-and 61-fold, respectively. They also found significantly elevated levels of expression of these two genes in a strain with a deletion of CaNRG1, which encodes a DNA-binding subunit for the Tup1p-Ssn6p repressor complex (42)(43)(44). This result prompted us to examine whether CaTup1p also regulates the expression of CaARN1. Because Rfg1p has also been implicated in mediating Tup1p-Ssn6p binding to DNA (45,46), we wanted to investigate which of the two proteins, CaNrg1p or Rfg1p (or both), is involved in CaTup1p regulation of CaARN1. For this experiment, we first constructed gene deletion mutants for each of the three genes. The strategy for gene deletion is described under "Experimental Procedures," and the correct gene deletion was verified by Southern analysis (data not shown). As well documented by others (42)(43)(44)(45)(46), a common phenotype of Catup1⌬, Canrg1⌬, and rfg1⌬ mutants is constitutive filamentous growth. Fig. 4A shows that all three mutant strains we constructed exhibited filamentous growth in YPD medium at 30°C, a condition under which the wild-type strain normally grows in the yeast form. The cell morphology and the extent of filamentous growth of each mutant were in good agreement with previous descriptions, which further confirmed the correct deletion of each gene. Fig. 4B shows the expression patterns of CaARN1 in the wild-type strain and the mutants grown in iron-replete and iron-depleted media. The CaARN1 probe detected a single band of ϳ2.1 kb (sufficiently large to encode the protein) in all RNA samples except the ones prepared from strain Caarn1⌬, indicating that the detected band represents CaARN1 mRNA. The wild-type cells exhibited a CaARN1 expression pattern typical of many iron-responsive genes, being expressed under iron-depleted conditions, but repressed under iron-replete conditions. In contrast, the Catup1⌬ mutant apparently had lost the transcription repression of CaARN1, which was expressed to a similar level under both conditions, indicating that CaTup1p is responsible for the repression of the CaARN1 gene under iron-sufficient conditions. A similar observation was recently reported by Lesuisse et al. (28). In the Canrg1⌬ mutant, the expression of CaARN1 exhibited the same response to iron concentration as in the wild-type strain, whereas in the rfg1⌬ mutant, CaARN1 displayed the same expression pattern as in Catup1⌬, the loss of repression at high iron concentration. This result indicates that CaTup1p repression of CaARN1 expression is mediated by Rfg1p, but not by CaNrg1p.
Cellular Localization of CaArn1p-In this study, we used GFP tagging to examine CaArn1p localization in living cells. GFP was fused to the C terminus of CaArn1p and expressed from its own promoter. This gene was expressed either as a single copy or on pABSK1. We used Western blot analysis to confirm the expression of a fusion protein of the expected size (ϳ93 kDa) and in an iron-dependent manner in the transformed cells using anti-GFP antibody (Fig. 5A). We then determined the functionality of the protein by testing whether it could rescue the siderophore-dependent growth defect of Caarn1⌬. Fig. 5B shows that the fusion protein, expressed either as a single copy or on pABSK1, fully rescued the FC utilization defect of Caarn1⌬ on the iron-depleted plates. Fluorescence microscope examination of the cells grown in ironlimiting medium revealed a bright fluorescent periphery in every cell, either in the yeast or the filamentous form (Fig. 5C), reminiscent of the localization of Ftr1p and Fet3p (16, 18, 27,   FIG. 4. Iron-responsive regulation of CaARN1 expression. A, C. albicans CaTUP1, RFG1, and CaNRG1 deletion mutants exhibited constitutive filamentous growth under conditions that normally support the yeast form of growth. All strains were grown in GMM medium at 30°C. The strains used were CAF2-1 (WT), CaWY5 (rfg1⌬; RZ-1 exhibited a similar phenotype (not shown)), CaWY6 (tup1⌬), and CaWY7 (nrg1⌬). B, Northern analysis revealed loss of iron-responsive regulation of CaARN1 expression in Catup1⌬ and rfg1⌬ mutants. Each strain was grown separately under high (H) and low (L) iron conditions. In addition to the four strains used in A, strain CaWY3-2 (arn1⌬) was also included as a negative control. PCR-amplified CaARN1 and CaACT1 DNA fragments were used as probes. 47). A few bright spots were also seen in the cytoplasm of some cells. The same cells grown in iron-replete medium showed undetectable fluorescence (data not shown). The results suggest that CaArn1p is primarily localized to the cell surface, presumably the plasma membrane. This localization implies that CaArn1p may serve as receptor and/or transporter of siderophore-iron at the cell surface. We wished to investigate whether CaArn1p transports siderophore-iron by internalization of the receptor upon ligand binding in a way similar to transferrin-iron uptake in mammalian cells (2). As a first step to answer this question, we determined whether exposure of cells to FC would result in an increase in the number of intracellular fluorescent spots. We treated aliquots of the cells ex-pressing CaArn1p-GFP with 10 M FC, FOB, or FeCl 3 for 2 h before visualization by fluorescence microscopy. Fig. 5D shows that there were significantly more intracellular fluorescent spots in the cells treated with FC than in the cells treated with FOB or FeCl 3 . To further confirm the FC-induced increase in intracellular CaArn1p, we determined whether the number of spots would increase with an increase in FC concentration. Fig.  6 demonstrates that, indeed, the number of intracellular fluorescent spots increased with an increase in FC concentration, which again was not seen in cells treated with FOB or FeCl 3 (data not shown).

DISCUSSION
In this study, we investigated the properties and functions of the siderophore transporter CaArn1p in C. albicans. CaARN1 has been characterized in the heterologous host S. cerevisiae (28,29), which provided very useful information for the design of our experiments to study the gene in its native host. Our study confirmed some previous results, such as the role of CaArn1p in siderophore uptake, its FC specificity, and ironregulated expression of the gene. In addition, we made some new observations that may shed light on the understanding of CaArn1p function and roles in C. albicans virulence.
Comparison of siderophore-iron uptake in mutants with deletions of CaARN1 alone and of both the CaARN1 and CaFTR genes showed that siderophore-iron can be used via both the siderophore transporter and the oxidase-permease iron uptake system. The oxidase-permease system can assimilate iron from both FC and FOB. This route of uptake would require the function of ferric reductases to release iron in ferrous form from the siderophores. The role of ferric reductases in the utilization of siderophore-iron by the oxidase-permease system has been well documented (16,28,29,47). Our results support the conclusions that siderophore-iron uptake occurs through both reductive and non-reductive pathways in yeast. In agreement with the FC specificity of CaArn1p reported by Ardon et al. (29), we also observed that CaArn1p transported FC-bound iron, but had undetectable activity for FOB-bound iron. However, these observations do not agree with the intracellular accumulation of a fluorescent FOB analog reported by Lesuisse et al. (28). One explanation could be that the intracellular fluorescence was due to the slow entry of the compound via liquid-phase endocytosis over the 24-h growth of cells before fluorescence microscopy. This slow entry of FOB would not be detected as significant by the 15-30-min 59 Fe uptake assay used by us and by Ardon et al. (29).
Using GFP tagging, we were able to determine CaArn1p localization in living cells. Because the tagged protein could fully correct the defect of Caarn1⌬ in using FC-iron, the observed GFP fluorescence pattern should reflect the physiological cellular localization of CaArn1p. In living C. albicans cells, the majority of CaArn1p-GFP was found in the plasma membrane, and a small amount was found in the cytoplasm with a punctate appearance. This localization is apparently different from the predominant localization of Arn1p and Arn3p in intracellular vesicles in S. cerevisiae previously observed by others (16,47). However, these authors did not exclude the possibility that a small amount of Arn1p and Arn3p could also be localized on the plasma membrane. Although we have not determined the nature of the intracellular vesicles containing CaArn1p in C. albicans, we assume that they are likely to be the same as those containing Arn1p and Arn3p in S. cerevisiae. The two different observations do not necessarily contradict each other. These proteins may be localized in both the plasma membrane and intracellular vesicles in a dynamic equilibrium, which is shifted toward different ends in the two organisms to suit each one's specific need for siderophore-iron in their respective natural environmental niches. These two organisms are well known to differ in the number of ARN genes, specificity for siderophores, and ability in siderophore production. A similar situation is the ability of C. albicans (but not S. cerevisiae) to shift a dynamic cell cycle-dependent distribution of a pool of actin toward a persistent localization at the hyphal tips when cells switch growth from the yeast form to the filamentous form (48). Then, what is the relationship between the CaArn1p molecules in the two locations? One possibility is that CaArn1p in the intracellular vesicles may come from the internalization of the ones in the plasma membrane. This internalization might be a result of CaArn1p-mediated transport of siderophore-iron complexes into the cells, a mechanism similar to the receptor-mediated cellular uptake of transferrin-iron complexes in mammals (1). Supporting this explanation, we observed an increase in the number of fluorescent spots in cells treated with FC in a concentration-dependent manner, but not in cells treated with FOB or Fe. However, it is equally possible that the receptors are internalized alone as a mechanism to down-regulate CaArn1p levels on the cell surface after translocating iron across the plasma membrane. At this stage, we are unable to distinguish between these two possibilities. Other researchers have also suggested that Arn3p/Sit1p is likely to act as a transporter at the cell surface and be rapidly recycled by internalization (20,34). These authors mentioned that a mutant completely defective for endocytosis showed normal siderophore uptake, a result strongly challenging the hypothesis of Arnp-mediated internalization as the mechanism for FC uptake. However, the role of endocytosis in FC utilization needs to be tested in C. albicans. Although the mechanism of Arnp-mediated siderophore uptake is still not clear, our observation of a prominent cell-surface localization of CaArn1p strongly suggests that the main site of function for this protein is at the plasma membrane in C. albicans, where it is likely to play a direct role in binding and transporting siderophore-iron. The observation of an FC-induced increase in the intracellular localization of CaArn1p suggests testable hypotheses for the mechanisms of siderophore-iron transport and recycling of the transporter.
The expression of many genes involved in iron uptake is regulated by iron concentration in both S. cerevisiae and C. albicans (16, 18, 20, 38 -40). They are normally repressed under iron-replete conditions and activated when the iron supply is limited. In S. cerevisiae, the transcription factor Aft1p activates the expression of genes such as FTR1, FET3, and ARN1-4 (16,38,39). Another transcription factor (Aft2p) was recently reported to have a role in the regulation of some iron transport genes (49). However, an AFT homolog does not seem to be present in C. albicans. Currently, it is not clear whether there is any transcription factor in C. albicans that activates transcription in an iron-dependent manner. Recently, evidence, including this study, is accumulating that the global transcription repressor Tup1p is responsible for the repression of at least some iron uptake-related genes in both S. cerevisiae and C. albicans (20,21,40). Tup1p associates with Ssn6p to form a general transcription repressor. However, this complex requires additional DNA-binding subunits for its specificity for different promoters (42)(43)(44). In C. albicans, two DNA-binding proteins, CaNrg1p and Rfg1p, have been found to mediate CaTup1p repression of target genes (43)(44)(45)(46). Therefore, we examined CaARN1 expression in both Canrg1⌬ and rfg1⌬ mutants to ascertain which one mediates CaTup1p function. We discovered that deletion of CaNRG1 had no discernible effect on the iron-responsive regulation of CaARN1, whereas deletion of RFG1 resulted in loss of repression under iron-replete conditions to the same extent as deletion of CaTUP1. These results indicate that CaTup1p repression of CaARN1 is primarily, if not entirely, mediated by Rfg1p. Because CaNrg1p mediates the repression of CaFTR1 and CaCFL2 in C. albicans (40,41), it would be interesting to know why the repression of genes for different iron uptake systems are mediated by different DNAbinding proteins.
To determine whether CaARN1 is required for C. albicans virulence, we tested the ability of the Ura ϩ Caarn1⌬ mutant in causing death in mice via intravenous injection. The mutant exhibited a similar level of virulence as its isogenic wild-type strain, suggesting that CaArn1p function is not critically required for iron acquisition during this route of infection. This result indirectly supports our previous report of the essentiality of the oxidase-permease iron uptake system (18), the only other alternative high-affinity iron uptake system known so far in C. albicans. Although the systemic candidiasis model is useful in testing factors important for the pathogen to grow or survive in the bloodstream and internal organs, it does not necessarily reflect the entire natural route of infection. C. albicans mainly colonizes the mucosal surfaces along the gastrointestinal and vaginal tracts. The pathogen is exposed to various environmental niches with limited iron supply and is in contact with different resident microbial communities. CaArn1p function may be required for iron uptake in these niches. The use of different infection models is required to unveil the role of siderophore-iron uptake in C. albicans infection.