Haemin uptake and use as an iron source by Candida albicans: role of CaHMX1-encoded haem oxygenase

Iron is required by most living systems. Iron availability plays a critical role in the hostpathogen relationship and in the virulence of bacteria and fungi [reviewed by Howard (1999) and Payne (1993)]. In the host, iron is predominantly intracellular, occurring as haem, ironsulphur proteins or ferritin, with smaller quantities in other iron proteins. The limited amounts of extracellular iron are tightly held by proteins such as transferrin in the serum and lactoferrin on body surfaces. The withholding of iron by the host has been shown to block virulent infections by many pathogens, and high-affinity iron acquisition by pathogens often promotes virulent infections (Ratledge & Dover, 2000). Candida albicans is a saprophytic organism for humans; it can be found inhabiting external surfaces of the body, including the skin and the mucosae of the gut and mouth in more than half of the normal population. In some settings C. albicans invades tissue and disseminates systemically. Prolonged neutropenia, following cancer chemotherapy or bone marrow transplantation, is a major risk factor for invasive fungal infections, and Candida species are the most frequently implicated organisms (Rex et al., 1995). Recurrent oropharyngeal and oesophageal candidiasis is encountered in most AIDS patients (Laine & Bonacini, 1994). There has been a rapid rise in the frequency of drug resistance in clinical C. albicans infections (Rex et al., 1995); hence, there is a pressing need to identify new drug targets.

The first studies about iron uptake by C. albicans emphasized the possible role of siderophore production by this organism to fulfil its iron requirement (Holzberg & Artis, 1983). Some authors described the excretion of hydroxamates, phenolates or both kinds of siderophores by C. albicans (Ismail et al., 1985; Sweet & Douglas, 1991). However, these studies were never reproduced, and siderophores secreted by C. albicans were never isolated and identified. Although siderophore production by Saccharomyces cerevisiae has never been shown and has not been confirmed for C. albicans, both S. cerevisiae and C. albicans have the ability to take up some siderophores non-reductively, via specific siderophore transporters (Ardon et al., 2001; Heymann et al., 1999, 2000; Lesuisse et al., 1998, 2002; Yun et al., 2000). The first eukaryotic siderophore transporter, Sit1, was identified in S. cerevisiae (Lesuisse et al., 1998). A gene homologous to SIT1 (CaSIT1/CaARN1) was then identified and characterized in C. albicans and was shown to encode a transporter for ferrichrome-type siderophores (Ardon et al., 2001; Heymann et al., 2002; Hu et al., 2002; Lesuisse et al., 2002). Siderophore-mediated iron uptake is not the only iron transport mechanism in fungi. Reductive iron uptake (i.e. iron removal from its ligands by reduction outside the cell prior to transport) was shown to occur in Ustilago maydis (Emery, 1987) and S. cerevisiae (Lesuisse et al., 1987). Reductive iron uptake by S. cerevisiae was extensively studied at the molecular level [reviewed by Eide (2000) and Van Ho et al. (2002)]. Two plasma membrane reductases (Fre1 and Fre2) are involved in releasing iron from its ligands by reduction; the free iron is then transported into the cell via a permeaseoxidase complex (Fet3Ftr1). Morrissey et al. (1996) showed that a very similar mechanism of reductive iron uptake was present in C. albicans. At the molecular level, reductive iron uptake by C. albicans involves proteins that are homologous to the components of the S. cerevisiae reductive uptake system. Cfl95/CaFre1 is a plasma membrane ferrireductase, while CaFtr1 and CaFet3 are the components of the permeaseoxidase complex (Hammacott et al., 2000; Knight et al., 2002; Ramanan & Wang, 2000). A knockout strain lacking CaFTR1 was shown to be avirulent in a mouse model for systemic infection (Ramanan & Wang, 2000), confirming the importance and non-redundancy of reductive iron uptake.

Interestingly, the reductive and siderophore iron uptake systems of C. albicans are regulated differently. Both systems are induced when the cells are grown under iron-deficient conditions, but transfer of the cells from a synthetic medium (YNB) to a serum-based medium results in repression of the reductive uptake system and in induction of the siderophore uptake system (Lesuisse et al., 2002). This observation was the first evidence that C. albicans can adapt its strategy for iron uptake to the physiological context. In the case of systemic infection, a possible iron source for C. albicans is haem, since this organism is known to secrete haemolytic factors (Manns et al., 1994; Moors et al., 1992). Indeed, recent preliminary studies have shown that haemin can be used by C. albicans as an iron source (Weissman et al., 2002). Up until now, haem uptake and use was studied mainly in bacteria (reviewed by Genco & Dixon, 2001). Here, we investigate this third strategy of iron acquisition by the pathogenic fungus C. albicans, and show that it is independent from the reductive and siderophore mechanisms of iron uptake.

Strains, media and iron compounds.
We used the following strains of C. albicans: SC5314 (wild-type); BWP17 (ura3Δ : : imm434/ura3Δ : : imm434 his1 : : hisG/his1 : : hisG arg4 : : hisG/arg4 : : hisG) (Wilson et al., 1999); BCa02-10 (tup1Δ : : hisG/tup1Δ : : hisG : : p405-URA3, ura3Δ : : imm434/ura3Δ : : imm434) (Braun & Johnson, 1997); SS4 (efg1 : : ADE2/PCK1p : : efg1 ura3Δ : : imm434/ura3Δ : : imm434) (Stoldt et al., 1997). For control experiments, we used the following wild-type strains of S. cerevisiae: S150-2B (MATa his3-Δ1 leu2-3,112 trp1-289 ura3-52); YPH499 [MATa ura3-52 lys2-801 ade2-101(ochre) trp1-Δ63 his3-Δ200 leu2-Δ1].

Unless otherwise stated, cells were grown at 30 °C in YPD medium (1 % yeast extract, 1 % peptone, 2 % glucose). For iron-deficient and iron-rich cultures, cells from an overnight pre-culture in YPD were diluted 10-fold in fresh YPD containing either 200 µM bathophenanthroline disulfonic acid (BPS) (iron-deficient culture) or 10 µM ferric citrate (iron-rich culture) and grown for 5 h at 30 °C. Cells were then harvested and washed with water before being resuspended in 50 mM citrate (tri-Na⁺) buffer (pH 6·5) containing 5 % glucose. Cells were then used for experiments. For experiments requiring RNA isolation, cells were grown in minimal YNB/glucose (without copper and iron) medium (Bio 101) plus the required amino acids. After overnight pre-culture, the cells were diluted 10-fold in the same fresh medium added with various supplements [10 µM ferric citrate, 10 µM ferrichrome, 5 µM iron-saturated transferrin, 50 µM haemin, 50 % (final) fetal bovine serum or 200 µM BPS]. The cells were grown for 5 h at 30 °C, before RNA isolation.

Radiolabelled iron compounds were prepared from ⁵⁵FeCl₃ (50 mCi mg^-1; 1·85 GBq mg^-1). ⁵⁵Fe-haemin was synthesized chemically from protoporphyrin IX and ⁵⁵FeCl₃ in pyridine/acetic acid (1 : 50) under a nitrogen atmosphere, as described by Galbraith et al. (1985). ⁵⁵Fe-haemin was taken into ether, washed extensively with water and 2·7 M HCl to remove any remaining Fe and protoporphyrin. Ferric citrate was obtained by mixing FeCl₃ in sodium citrate buffer (pH 6·5) to get a final Fe/citrate ratio of 1 : 20. Ferrichrome and transferrin were purchased from Sigma.

Haemin uptake assays.
Cells were suspended in 50 mM citrate (tri-Na⁺) buffer (pH 6·5) containing 5 % glucose and 0·05 % Tween 80, and pre-incubated for 15 min at 30 °C under agitation. ⁵⁵Fe-haemin was added at a final concentration of 1 µM to the cell suspension. Aliquots (100 µl) were withdrawn as a function of time and added to 20 µl of 1 mM cold haemin kept on ice in the wells of a microtitre plate. The cells were collected with a cell harvester (Brandel) and washed on the filter.

Strain construction.
Molecular cloning techniques and gel electrophoresis were performed as described (Sambrook et al., 1989). To disrupt the two alleles of the CaHMX1 gene from strain BWP17, we used the primer-directed integration of Cahmx1Δ : : URA3 and Cahmx1Δ : : ARG4 as described by Wilson et al. (1999). PCR primers 5'-CGTCAAGGTTTGCAAGCATTCTATCATGTATTTGCTAGTATTGAAAAGGCCTTGTACAGACAGCTTGAAAAGTGGAATTGTGAGCGGATA-3' and 5'-GTCATGTTCGAAAATGTATTTTGATTCTTCAATGATTTCCAACTTTTGTTCTTCCGTCAAACCATTTCTTGTTTTCCCAGTCACGACGTT-3' were used to amplify the URA3 and ARG4 cassettes, from plasmids pGEM-URA3 and pRS-ARG4ΔSpeI, respectively, flanked by 71 nt of CaHMX1. After homologous recombination, a 397 nt deletion (nucleotides 252649, starting from ATG) was created. Successful integration was verified by PCR using the external primers 5'-ATGCAATACAAACTGAGTGGAGCTACATCG-3' and 5'-TTAGGTTAATTTATTGACAACTCTTCTTAA-3'. For re-introduction of the CaHMX1 gene into the double disrupted strain, a 2 kb fragment containing the entire ORF was amplified by PCR using the primers 5'-GGCGGATCCGAGGGCAATGATACTGATTGGGCCATTATTTGG-3' and 5'-GGAATCAACGGCATGCGTTGATTGGCATTGTTGTGATATTTTCC-3', which carry a BamHI and a SphI restriction site (underlined), respectively. After BamHI/SphI digestion, the amplified fragment was inserted into the BamHISphI sites of the vector pGEM-HIS1 (Wilson et al., 1999). The resultant plasmid was linearized by using NruI, used for transformation of the double CaHMX1 knockout mutant and His⁺ clones were selected. Correct insertion at the HIS1 locus was verified by PCR using the primers 5'-CTCGTGCCGTGTTGAATGTTTGCTTC-3' and 5'-CGAGTACCAATATATCGGTTGCACCAGC-3'. The C. albicans CaHEM14, CaFTR1 and CaSIT1 knockouts were obtained using the same method. The primers used to disrupt these genes were: CaHEM14, 5'-GATTTGGATTCTCAAATAGAAGTAATTAATGAAAAATGTAATGCCAATAAGAAATATATTCTTGATTCTTCGTGGAATTGTGAGCGGATA-3' and 5'-TAATTTTGATGTCAATACATCTTTAACAATTTTCAAATTCACTGACGAAGGAATTGTCCAATTTGTATATTTCCCAGTCACGACGTT-3'; CaFTR1, 5'-CGTTCAAATTTTCTTCATCGTTTTCAGAGAATCTTTGGAAGCTATCATTGTTGTTTCAGTGCTTTTGGCGTGGAATTGTGAGCGGATA-3' and 5'-GTCTCTTGCCTTATTCTTTTAGTTGTTGAATAATAATTAACTAAGTTTATTTGTTTTCTTTGGATTCGTTTCCCAGTCACGACGTT-3'; CaSIT1, 5'-CCAGTCTTCCAATAATCATTCTTCAGAAGAAGATAAACACTTGTCCGGAGATGAAAAGACGTTTTCGTGGAATTGTGAGCGGATA-3' and 5'-GCTACTCTTTTCTTCTTGAAATTGCCGAAGAAATTGGCCAACGAGTCCTTCTCTTCTTGCTTTTCTTTTCCCAGTCACGACGTT-3'. Genotypes of all strains were confirmed by PCR and Southern blot hybridization (Sambrook et al., 1989).

RNA analysis.
RNA was extracted as described by Kohrer & Domdey (1991). Northern blotting and hybridization, at 42 °C in 50 % (v/v) formamide, were done essentially as described (Knight et al., 2002; Lesuisse et al., 2002). The DNA fragments used as probes for each gene were amplified by PCR using primers 5'-ATGCAATACAAACTGAGTGGAGCTACATCG-3' and 5'-TTAGGTTAATTTATTGACAACTCTTCTTAA-3' for CaHMX1, and primers 5'-GATTCTTATGTTGGTGATGA-3' and 5'-TCGTCGTATTCTTGTTTTGA-3' for CaACT1. Probes for CFL95/CaFRE1, CaFTR1 and CaSIT1 have been described previously (Knight et al., 2002; Lesuisse et al., 2002). After Northern blotting, CaHMX1, CaFRE1, CaFTR1, CaSIT1 mRNA and 25S rRNA levels were quantified using the IMAGEQUANT software (version 1.2; Molecular Dynamics). Volume values for mRNA signals were normalized with volume values for 25S rRNA and ACT1 mRNA.

Haemin is an iron source for C. albicans
Free haemin or haem bound to proteins as the prosthetic group (cytochrome c, haemoglobin) could be used by C. albicans as the sole iron source. When a high concentration (1 mM) of the iron chelator BPS was added to a complete growth medium (YPD), growth of C. albicans cells was completely inhibited. Addition of free haemin or haem proteins restored growth. Similar growth arrest in BPS and growth restoration by haemin were seen for strains disrupted for both copies of the high-affinity ferrous iron permease CaFTR1 gene (reductive uptake system) or of the siderophore receptor CaSIT1/CaARN1 (Fig. 1). We can rule out the possibility that ΔCasit1/ΔCasit1 cells in Fig. 1 were able to take up iron from haemin via the reductive uptake system, after extracellular removal of iron from haemin. Indeed, in haemin iron is tightly bound to the porphyrin ring and cannot be removed by simple reduction (Buchler, 1975). Moreover, the presence of BPS in the medium functionally inactivated the reductive uptake system, as evidenced by the fact that addition of 10 µM ferric citrate to the BPS-containing plates did not restore growth (not shown). Finally, we showed that haemin uptake was copper-independent in a wild-type as well as in a ΔCasit1/ΔCasit1 strain (see below). S. cerevisiae did not behave in the same manner; haemin was unable to restore growth of cells in iron-deficient medium (Fig. 1). We conclude that in C. albicans, but not in S. cerevisiae, there is a third iron uptake system in addition to the reductive and siderophore uptake systems described previously. This is consistent with the recent observation of Weissman et al. (2002) that haemin and haemoglobin are potential iron sources that can be used by C. albicans in a CaCcc2-independent manner.

(73K): Fig. 1. C. albicans, unlike S. cerevisiae, can use haemin as the sole iron source. Iron-deficient cells (pre-grown on YPD+200 µM BPS) were sequentially fivefold-diluted and then plated onto YPD (+uridine) agar with (A) no addition, (B) 1 mM BPS or (C) 1 mM BPS+50 µM haemin. Lanes: 1, S. cerevisiae (YPH499); 2, C. albicans (SC5314, wild-type); 3, C. albicans (double ΔCaftr1 mutant); 4, C. albicans (double ΔCasit1 mutant).

Characteristics of haemin uptake kinetics
We determined the kinetic parameters of haemin uptake by C. albicans using chemically synthesized ⁵⁵Fe-haemin. Two phases were observed (Fig. 2), a rapid phase of haemin binding followed by a slower uptake phase. The uptake phase was more temperature dependent than the binding phase, but both phases strongly depended on the iron status of the cells (Fig. 2). Binding and uptake of haemin were very low when the cells were previously grown in iron-rich medium (10 µM ferric citrate), and both phases were induced when the cells were previously grown in iron-deficient medium (Fig. 2). Similar kinetics of haemin uptake were observed with a mutant deficient for reductive uptake of ferric citrate (ΔCaftr1/ΔCaftr1), and with a mutant deficient for non-reductive uptake of ferrichrome (ΔCasit1/ΔCasit1). When the siderophore pathway was inactivated by knockout of CaSIT1/CaARN1 and the reductive pathway was inactivated by copper chelation (Knight et al., 2002) there was still no effect on haemin uptake (not shown). A Δtup1/Δtup1 mutant previously shown to misregulate both reductive and siderophore iron uptake also showed normal kinetics for haemin uptake (not shown). In contrast, no haemin uptake was observed with S. cerevisiae cells grown either in iron-rich or in iron-deficient conditions (Fig. 2). Thus, C. albicans but not S. cerevisiae has a specific and inducible, iron-regulated haemin uptake system independent from the reductive and non-reductive uptake systems. Protoplasts of C. albicans showed the same kinetics of haemin uptake as intact cells, but treatment of protoplasts by proteinase K decreased the amount of haemin bound to the cells, while transport of ferrichrome by proteinase-treated protoplasts was unaffected (not shown). The same effect of proteinase K treatment on cell-surface binding of haemin was recently observed in the fungus Histoplasma capsulatum (Foster, 2002). In that case, the authors concluded that haemin uptake by this organism involved a first stage of haemin binding at the cell surface. The same is probably true for C. albicans: the two phases that are observed in the kinetics (Fig. 2) probably correspond to two steps of transport. The first step is energy-independent and probably consists of the binding of haemin to a cell-surface receptor expressed under low-iron conditions. The second step is energy-dependent and probably consists of the transport of cell-surface bound haemin into the cells.

(20K): Fig. 2. Kinetics of haemin uptake by C. albicans and S. cerevisiae. C. albicans cells (strain SC5314) grown under iron-deficient (, ) or iron-rich () conditions were incubated in 50 mM citrate buffer (pH 6·5) containing 5 % glucose, 0·05 % Tween 80 and 1 µM 55Fe-haemin at 30 °C (, ) or 4 °C (). S. cerevisiae cells (strain S150-2B) grown in iron-deficient (x) or iron-rich (+) conditions were incubated at 30 °C in assay buffer as above. Haemin uptake was followed by withdrawing aliquots at the indicated intervals.

We further studied the binding phase of haemin uptake. Haemin binding at the cell surface at 4 °C was measured as a function of extracellular haemin concentration (Fig. 3). The contribution of non-specific binding of haemin was determined by measuring ⁵⁵Fe-haemin binding in the presence of a 1000-fold excess of unlabelled haemin. The values for non-specific binding were subtracted from those for total binding (binding in the absence of unlabelled haemin) to give specific binding (Fig. 3). Experimental data for specific binding were fitted to the equation y=c[x]^a/b^a+[x]^a, where b and c are the K_d and B_max values, respectively, and a is a constant. Resolution of this equation gives the following values, K_d=195·4±6·1 nM, B_max=1·21±0·017 pmol per 10⁶ cells, a=2·02±0·08. The B_max value allows one to calculate that a maximum of about 7x10⁵ molecules of haemin can be bound specifically to one cell. Also, considering that 10⁶ cells represent about 1 mg total protein and 0·01 mg plasma membrane protein, a putative receptor protein of 30 kDa (arbitrary value) would represent about 1/5000 of total membrane protein. The data on specific binding did not fit to standard Scatchard curves. The curve representing haemin dependence of specific haemin binding was sigmoidal, indicating that a process involving positive co-operativity occurs at low haemin concentrations (Fig. 3).

(12K): Fig. 3. Saturation of haemin binding at the cell surface. Iron-deficient C. albicans cells (strain SC5314) were incubated for 5 min at 4 °C with increasing concentrations of 55Fe-haemin, with (, non-specific binding) or without (•, total binding) a 1000-fold excess of cold haemin. Specific binding of haemin at the cell surface was calculated ().

CaHmx1 is required for iron assimilation from haemin
In haem, iron is tightly bound to the porphyrin ring. Haem-bound iron can be released by enzymic ligand destruction. An alternative possibility, enzymic iron removal (reverse ferrochelatase), has not been described. In mammals, iron re-utilization from cellular haemoproteins requires haem oxygenase activity (Poss & Tonegawa, 1997). Haem oxygenase catabolizes haem to biliverdin, carbon monoxide and free iron. A gene encoding a putative haem oxygenase is present in the C. albicans genome (orf6.7617) and has been named CaHMX1. This gene encodes a 291 residue protein with a theoretical M_r of 33 973 and a pI value of 7·7. The CaHmx1 protein showed 25 % identity with human haem oxygenase-1 (HO-1) and 35 % identity with S. cerevisiae Hmx1p for the conserved core of the enzyme. Crystal structures of the phylogenetically distant human HO-1 and Neisseria meningitidis haem oxygenases (Schuller et al., 1999, 2001) show that the haem binding pocket is formed by two α-helices. Even though the structure of the active site is similar, the sequence of the distal helices of these enzymes is different (Schuller et al., 1999, 2001). Accordingly, sequence comparison (not shown) revealed that only the proximal helices residues Thr35, His39, Asp40, Ala42 and Asp43 are conserved in the CaHmx1 sequence.

We tested the role of CaHmx1 in iron utilization from haemin by deleting two copies of the CaHMX1 gene from the C. albicans genome. A wild-type allele of CaHMX1 was then re-introduced into the Cahmx1Δ : : URA3/Cahmx1Δ : : ARG4 mutant (Fig. 4). Cells from the wild-type, from the single Cahmx1 and double Cahmx1 disruption strains and from the reconstituted Cahmx1/CaHMX1 strain were plated onto complete medium with haemin as the sole iron source. The double Cahmx1 deletion strain grew normally on YPD medium (not shown) but was unable to grow in conditions where haemin was the sole iron source (Fig. 5, column 3). In the same conditions, the reconstituted knockout grew normally (Fig. 5, column 4). This result shows that CaHmx1 is required for iron assimilation from haemin in C. albicans. Although CaHmx1 is involved in iron utilization from haemin, it is not involved in haemin uptake. Kinetics of haemin binding and transport were not significantly changed in the double Cahmx1 deletion strain compared to wild-type cells (not shown).

(43K): Fig. 4. Confirmation of chromosomal disruption and re-introduction of gene CaHMX1 by PCR. The primers used are described in Methods. Amplified fragments were run on a 0·8 % agarose gel and stained using ethidium bromide. Lanes: 1, phage λ digested with HindIII; 2, BWP17 (wild-type); 3, BWP17 Cahmx1Δ : : URA3/CaHMX1 (single knockout); 4, BWP17 Cahmx1Δ : : URA3/Cahmx1Δ : : ARG4 (double knockout); 5, BWP17 Cahmx1Δ : : URA3/Cahmx1Δ : : ARG4/CaHMX1 (reconstructed strain).

(81K): Fig. 5. CaHMX1 knockout prevents growth of C. albicans with haemin as the sole iron source. Iron-deficient cells (pre-grown on YPD+200 µM BPS) were sequentially fivefold-diluted and then plated onto YPD (+uridine) agar containing either 1 mM BPS (A, iron-deficient condition) or 1 mM BPS+50 µM haemin (B, haemin as the sole iron source). Strains: 1, BWP17 (wild-type); 2, BWP17ΔCahmx1/CaHMX1; 3, BWP17ΔCahmx1/ΔCahmx1; 4, BWP17ΔCahmx1/ΔCahmx1/CaHMX1.

Regulation of CaHMX1 transcription
We tested various conditions for effects on CaHMX1 expression by Northern blot (Fig. 6A). In standard conditions of growth (minimal medium at 30 °C), the CaHMX1 transcript was difficult to detect due to low expression, and this level was further repressed when iron was added as ferric citrate, ferrichrome or transferrin. Transcript levels were strongly induced by iron deprivation and by haemin exposure. A shift of the temperature from 30 to 37 °C was also a strong inducer. Incubation of the cells in fetal calf serum had no effect on the level of CaHMX1 transcripts (Fig. 6A). The observations that CaHMX1 expression was upregulated by both iron deprivation and haemin suggest that this gene has a complex pattern of regulation, since haemin is itself an iron source. To our knowledge, CaHMX1 is the first C. albicans gene found to be regulated by haem. While extracellular haemin induced transcription of CaHMX1, intracellular haem was not required for induction. The Cahem14Δ : : URA3/Cahem14Δ : : ARG4 haem-deficient mutant contained no trace of haem but responded to iron-deficient conditions as the wild-type by inducing CaHMX1 transcription (not shown).

(59K): Fig. 6. Regulation of CaHMX1 compared to genes of the reductive (CFL95, CaFTR1) and non-reductive (CaSIT1/CaARN1) uptake systems. Cells were grown in minimal medium (YNB without iron) with various additions. Northern blotting of the indicated genes was performed as described in Methods. (A) CaHMX1 expression in the wild-type (BWP17) grown under various conditions. Lanes: 1, no addition; 2, 200 µM BPS; 3, 10 µM ferric citrate; 4, 10 µM ferrichrome; 5, 5 µM iron-saturated transferrin; 6, 25 µM haemin; 7, 50 % (v/v) fetal bovine serum; 8, no addition, 5 h shift at 37 °C. (B) CaHMX1 expression in various strains grown under different conditions. Lanes: 1, Wild-type (BWP17), 200 µM BPS; 2, BWP17ΔCahmx1/CaHMX1 (single knockout), 200 µM BPS; 3, BWP17ΔCahmx1/ΔCahmx1 (double knockout), 200 µM BPS; 4, Δefg1/Δefg1, no addition; 5, Δefg1/Δefg1, 200 µM BPS; 6, Δefg1/Δefg1, 10 µM ferric citrate; 7, Δefg1/Δefg1, 25 µM haemin; 8, Δefg1/Δefg1, 50 % (v/v) fetal bovine serum; 9, Δtup1/Δtup1, no addition; 10, Δtup1/Δtup1, 200 µM BPS; 11, Δtup1/Δtup1, 10 µM ferric citrate; 12, Δtup1/Δtup1, 25 µM haemin; 13, Δtup1/Δtup1, 50 % (v/v) fetal bovine serum. (C) Expression of CFL95, CaFTR1 and CaSIT1/CaARN1 in wild-type (WT), Δtup1/Δtup1 and Δefg1/Δefg1 isogenic strains. Lanes: 1, no addition; 2, 200 µM BPS; 3, 10 µM ferric citrate. (D) Relative abundance of CaHMX1, CFL95, CaFTR1 and CaSIT1 transcripts in wild-type (WT), Δtup1 and Δefg1 strains. 1, 200 µM BPS; 2, no addition; 3, 10 µM ferric citrate.

We also investigated the pattern of CaHMX1 expression in mutants of the major transcription factors Tup1 and Efg1 (Brown & Gow, 1999; Liu, 2001; Whiteway, 2000). Both the reductive and the siderophore pathways of iron uptake were influenced by the Tup1 co-repressor (Knight et al., 2002; Lesuisse et al., 2002) and by the Efg1 activator (S. Knight & A. Dancis, unpublished data). These effects may be linked to the serum response, which is associated with a shift from reductive to siderophore uptake pathways (Knight et al., 2002). In contrast, CaHMX1 expression was almost unchanged in a double Δtup1 mutant compared to wild-type (Fig. 6B). This suggests that the haemin uptake pathway is regulated independently from the reductive and non-reductive pathways of iron uptake. The effect of Efg1 was also unique. CaHMX1 expression was deregulated in the efg1 mutant (Fig. 6B), which fits well with the observation that the promoter region of CaHMX1 contains four E-box sequences (Leng et al., 2001) (CATGTG, -684 to -679; CACTTG, -548 to -543; CATCTG, -409 to -404; CAATTG, -385 to -380). In unsupplemented media, the expression of the CaHMX1 gene was strongly induced in the efg1 mutant compared to the wild-type (Fig. 6B). Induction by BPS and haemin was almost identical in both strains. However, addition of fetal serum strongly induced the gene in the efg1 mutant compared to the wild-type strain (Fig. 6B). These results suggest that the transcriptional regulator Efg1 is a repressor of CaHMX1. Fig. 6(C) shows the expression patterns of CFL95/CaFRE1, CaFTR1 and CaSIT1/CaARN1 in wild-type, Δtup1 and efg1 mutants. Expression of CFL95 was downregulated by iron exposure in the wild-type and derepressed in the Δtup1 mutant, as observed previously (Knight et al., 2002). In the efg1 mutant CFL95 expression was unresponsive to iron restriction and detectable at low levels under all conditions. Likewise, CaFTR1 expression did not respond to iron depletion in the efg1 mutant and was present at low, but not fully repressed, levels similar to that in the Δtup1 mutant. The expression of CaSIT1 was repressed under all conditions in the Δtup1 mutant and was barely detectable in the efg1 mutant. The effect of EFG1 and TUP1 deletions on the relative transcript abundance of CFL95/CaFRE1, CaFTR1, CaSIT1 and CaHMX1 is summarized in Fig. 6(D). Thus, our results show that the three different iron uptake systems are regulated independently and in a complex manner, although they were all induced by iron deprivation.

Haemin effects on colony and cell morphology
C. albicans colonies forming on agar plates with haemin as the sole iron source showed a very unusual morphology. Colonies were made up of worm-like, tubular structures organized into a complex network (Fig. 7A). When observed microscopically, some cells within the colonies were seen to form a network of filaments enclosing other cells in the yeast form (Fig. 7B). When the colonies grew older, the proportion of filaments increased and the colonies took on the consistency of a dried sponge (not shown). These colonies with tubular structures appeared only when haemin was the sole iron source in the medium (haemin plus 1 mM BPS). Haemin added to complete medium without the iron chelator BPS induced filamentation (not shown), but the morphological change of the colonies was not as striking (Fig. 7C). Others have previously reported induction of filamentation by haemin (Casanova et al., 1997). This effect of haemin on filamentation was specific, since it was not observed with any other iron source (ferric citrate, ferrichrome) or with protoporphyrin IX (not shown). Induction of filamentation by haemin increased with increasing extracellular concentrations of haemin, and it was always more pronounced in the double ΔCahmx1/ΔCahmx1 mutant than in the wild-type (Fig. 7D). The double ΔCahmx1/ΔCahmx1 mutant showed unchanged haemin uptake and decreased haemin degradation; thus more haemin is expected to accumulate in these cells. This suggests that the inducer of filamentation may be intracellular haemin. The mechanisms by which intracellular haemin promotes filamentation and morphological change of the colonies, and the physiological significance of these processes, remain to be investigated.

(102K): Fig. 7. Haemin-induced morphological change of C. albicans colonies. (A) Colony of a wild-type strain (BWP17) grown with haemin as the sole iron source (YPD+1 mM BPS+50 µM haemin). (B) Light microscope view of cells from the colony shown in (A). (C) Colony of a wild-type strain (BWP17) grown in the presence of haemin (YPD+50 µM haemin). (D) Colonies of the wild-type (1) and of mutant strains BWP17ΔCahmx1/CaHMX1 (2) and BWP17ΔCahmx1/ΔCahmx1 (3) grown on YPD+25 µM haemin (upper view) or on YPD+75 µM haemin (lower view).

Several recent studies have greatly improved our knowledge of the mechanisms of iron uptake by C. albicans (Ardon et al., 2001; Eck et al., 1999; Hammacott et al., 2000; Heymann et al., 2002; Hu et al., 2002; Knight et al., 2002; Lesuisse et al., 2002). These studies, together with the present one, show that there are at least three independent iron uptake systems in this pathogenic fungus. The three systems are induced under conditions of iron deprivation; however, fine regulation is probably complex, involving several transcription factors, including Tup1 and Efg1. The effect of Efg1 on CaHMX1 transcription is especially interesting. Efg1 is well known as a transcriptional activator of many hyphal-specific genes. CaHMX1 seems to be the first gene shown to be directly under the negative control of Efg1, though others have previously suggested that it may function as both repressor and activator based on the effects of shut-down and overexpression of EFG1 on hyphal growth (Stoldt et al., 1997). The complex regulatory patterns of the genes involved in the three iron uptake pathways of C. albicans probably reflect the ability of C. albicans to readily adapt to various ecological niches where iron may be present in very different chemical forms. The reductive uptake system is able to release iron from various ferric chelates at the surface of the cells, and the resulting free ferrous iron is taken up via a permeaseoxidase complex of the plasma membrane homologous to that described in S. cerevisiae (Dancis et al., 1992; Stearman et al., 1996). The siderophore uptake system is able to mediate uptake of siderophores of the ferrichrome family, which enter the cell by a specific receptor, the CaSIT1/CaARN1 gene product. This uptake system is homologous to the siderophore uptake system described in S. cerevisiae (Lesuisse et al., 1998; Yun et al., 2000). The haemin uptake system described in this work mediates the use of extracellular haem as an iron source, a process involving specific binding at the cell surface, transport into the cell and degradation by haem oxygenase. This iron utilization system does not seem to be present in S. cerevisiae. In this yeast, there is a gene (HMX1) expected to encode a haem oxygenase, but its physiological role remains unexplored. Haemin can enter S. cerevisiae cells, as shown by the ability of haemin added to the extracellular medium to restore growth of haem-deficient mutants. This effect may represent passive diffusion of haemin into the cells. Haem can diffuse passively across model lipid bilayers (Genco & Dixon, 2001), and we could not detect any transport of haemin into S. cerevisiae cells by kinetic studies. Moreover, S. cerevisiae was unable to grow with haemin as the sole iron source, unlike C. albicans. Thus, the role of haem oxygenase in S. cerevisiae, if any, could be restricted to re-utilization of iron from intracellular haemoproteins, as in higher eukaryotes. The presence of a haemin uptake system able to provide C. albicans cells with iron could have important implications. It is well known that this pathogenic fungus can secrete haemolytic factors, and thus haem proteins may provide an iron source for cells during systemic infection (Manns et al., 1994; Moors et al., 1992). CaHmx1 is not directly involved in haemin uptake, but it allows C. albicans to use haem as an iron source. The study of CaHMX1 regulation could be important, in this respect. CaHMX1 is involved in a crucial step of iron assimilation from haemin and, as such, it is not really surprising that it is induced under iron-deficient conditions. More interesting is the observation that haemin itself induces expression of CaHMX1. This observation will form the basis of a strategy to find new components of the haemin uptake system, i.e. a search for haemin-induced genes. It would be surprising that CaHMX1 could be the only gene positively regulated by haemin that is involved in haemin iron utilization. The observation that incubation of cells with haemin increases haemin binding to the cell surface is consistent with this hypothesis. Haemin most likely induces expression of several genes involved in the haemin uptake/utilization pathway, and these genes may be identified by DNA micro-array experiments. Identification of such new genes by this strategy is one of our priorities now. Another important field to investigate based on the results of the present study is the involvement of haem iron utilization in virulence. We are currently testing the effect of CaHMX1 double disruption on systemic infection in mice. CaHmx1 itself is an interesting protein to study, as such. In a recent review, haem oxygenase-1 of mammals was characterized as the emerging molecule (Morse & Choi, 2002). This is because haem-oxygenase-related mechanisms play an important role in several aspects of different diseases, and increasing importance is given to haem oxygenases in the defence of organisms against oxidative stress. The mechanism by which haem oxygenase confers its protective effect is, as yet, poorly understood. In micro-organisms, the role of haem oxygenases in iron assimilation from haemin was often postulated, for evident mechanistic reasons, but rarely shown experimentally. To our knowledge, a direct role of haem oxygenase in iron assimilation has only been shown in Corynebacterium diphtheriae (Chu et al., 1999). Our study provides the first experimental evidence that haem oxygenase is required for iron assimilation from haem by a pathogenic fungus. We thank R. Bryce Wilson and Aaron Mitchell (Columbia University, New York, USA) for the BWP17 strain and for the plasmids used for gene disruption, Burk Braun and Alexander Johnson (University of California, San Francisco, USA) for the Bca02-10 strain, and Joachim Ernst (Institut für Mikrobiologie, Heinrich-Heine-Universität, Düsseldorf, Germany) for the SS4 strain. Some of the sequence data for C. albicans were obtained from the Stanford DNA Sequencing and Technology Center web-site (), which is supported by funds from the NIDR and The Burroughs Wellcome Fund. This work was supported by grants from CNRS and the Ministère de la Recherche et de l'Enseignement Supérieur (Programme de Recherches Fondamentales en Microbiologie et Maladies Infectieuses and Réseau Infections Fongiques), and by the American Cancer Society (Grant RPG-00-101-01-MBC).