A Screen for Genes of Heme Uptake Identifies the FLC Family Required for Import of FAD into the Endoplasmic Reticulum*

Although Candida albicans and Saccharomyces cerevisiae express very similar systems of iron uptake, these species differ in their capacity to use heme as a nutritional iron source. Whereas C. albicans efficiently takes up heme, S. cerevisiae grows poorly on media containing heme as the sole source of iron. We identified a gene from C. albicans that would enhance heme uptake when expressed in S. cerevisiae. Overexpression of CaFLC1 (for flavin carrier 1) stimulated the growth of S. cerevisiae on media containing heme iron. In C. albicans, deletion of both alleles of CaFLC1 resulted in a decrease in heme uptake activity, whereas overexpression of CaFLC1 resulted in an increase in heme uptake. The S. cerevisiae genome contains three genes with homology to CaFLC1, and two of these, termed FLC1 and FLC2, also stimulated growth on heme when overexpressed in S. cerevisiae. The S. cerevisiae Flc proteins were detected in the endoplasmic reticulum and the FLC genes encoded an essential function, as strains deleted for either FLC1 or FLC2 were viable, but deletion of both FLC1 and FLC2 was synthetically lethal. FLC gene deletion resulted in pleiotropic phenotypes related to defects in cell wall integrity. High copy suppressors of this synthetic lethality included three mannosyltransferases, VAN1, KTR4, and HOC1. FLC deletion strains exhibited loss of cell wall mannose phosphates, defects in cell wall assembly, and delayed maturation of carboxypeptidase Y. Permeabilized cells lacking FLC proteins exhibited dramatic loss of FAD import activity. We propose that the FLC genes are required for import of FAD into the lumen of the endoplasmic reticulum, where it is required for disulfide bond formation.

Iron is an essential nutrient for virtually all microorganisms because iron is required for the activity of a variety of enzymes that bind this metal in the form of free ions, heme, and ironsulfur clusters. Microbes living in different environments encounter iron in different forms, and pathogenic bacteria and fungi have evolved specific strategies for obtaining iron from their mammalian hosts. Heme is the most abundant source of iron in the mammalian host and heme uptake systems have been described in both Gram-negative and Gram-positive bac-teria. Gram-negative pathogens, such as Shigella dysenteriae, express an outer membrane receptor for heme, ShuA, which facilitates the movement of heme into the periplasmic space, where it is bound by the periplasmic heme-binding protein, ShuT, which delivers the heme to an ABC transporter consisting of ShuU and ShuV (1,2). The Gram-positive bacterium Staphylococcus aureus appears to prefer heme over transferrin as an iron source when infecting an animal host (3). S. aureus does not contain an outer membrane, but instead expresses two cell wall-associated proteins, IsdA and IsdC, that bind and transport heme, respectively, across the cell wall before it is taken up at the cytoplasmic membrane by a transporter of the ABC family (4). Although heme transport systems have been described for many bacterial species, much less is known about heme transport in eukaryotes.
Two mammalian proteins with heme transport activity have recently been described. The receptor for feline leukemia virus type C is a transporter required for the development of erythroid cells, and the human ortholog of this transporter can facilitate the export of intracellular heme (5). Shayeghi and colleagues (6) reported the identification of a murine heme transporter, HCP1, expressed on the apical surface of duodenal epithelial cells. This transporter has similarity to bacterial metal-tetracycline transporters, but no clear ortholog has been identified in non-mammalian species. Free-living worms such as Caenorhabditis elegans and parasitic helminths have an absolute requirement for dietary heme transport, as these organisms completely lack the capacity to synthesize heme de novo (7). Heme transporters in these organisms have yet to be identified.
Candida albicans is a dimorphic fungus that is part of the commensal flora of the human gastrointestinal tract and is frequently associated with mucocutaneous infections. In patients with compromised immune systems, Candida species can disseminate hematogenously and cause life-threatening infections. This fungus efficiently uses heme as a nutritional source of iron and exhibits strategies to acquire heme from host erythrocytes. Candida spp. secrete hemolytic factors that lead to the release of erythrocyte hemoglobin (8) and hemoglobin binds to the cell wall of both yeast and hyphal forms of the fungus (9). Recently, a cell surface heme-binding protein, Rbt5, was shown to be involved in heme-iron utilization (10). Although a plasma membrane transport system for heme has not been identified, intracellular degradation of heme through the heme oxygenase encoded by CaHMX1 is required for hemeiron utilization (11,12).
Although C. albicans and Saccharomyces cerevisiae express very similar systems of iron uptake (13), they differ in their capacity to use heme as a nutritional source of iron. Both species utilize reductive systems of iron uptake and express plasma membrane metalloreductases of the FRE family coupled to a ferrous iron transporter complex that consists of a permease (Ftr1) and a multicopper oxidase (Fet3). Both species also utilize non-reductive systems consisting of siderophore-iron transporters of the ARN/SIT family. Heme uptake in C. albicans occurs independently of the reductive system and submicromolar concentrations of heme or hemoglobin will support vigorous growth of a strain lacking high affinity ferrous iron transport (12). Similar concentrations of heme and hemoglobin do not stimulate growth of a S. cerevisiae strain lacking ferrous uptake (15). We exploited this difference in heme utilization to identify genes from C. albicans that, when expressed in S. cerevisiae, would enhance heme uptake activity. We identified a family of fungal-specific proteins that are required for the transport of flavin adenine dinucleotide (FAD) into the endoplasmic reticulum (ER), 2 where it is required for oxidative protein folding.
Geneticin-resistant clones were selected on YPD plates containing 80 mg/liter G418 (Invitrogen). To delete FLC2 the HISG-URA3-HISG cassette was PCR amplified from plasmid pMPY-ZAP (19) using the primers del-053-F and del-053-R. YPH499 was transformed with that PCR product to generate the flc2⌬ strain. The same procedure was used to delete FLC2 in the flc1⌬ strain transformed with plasmid pMET3-FLC1. Clones containing the correctly targeted deletion cassette were plated on 5-fluoroorotic acid-containing plates and 5-fluoroorotic acid-resistant clones were selected. The TET-FLC1flc2⌬flc3⌬ strain was constructed in R1158 (MATa his3-1 leu2-0 met15-0 URA3::CMV-tTA) (20). The promoter replacement of the FLC1 gene was made as described (20) using primers YPL221W-upRP188 and YPL221W-downRP188. The complete coding sequence for FLC2 was replaced by the nurseothricin resistance gene cassette natMX (21). The complete coding sequence for FLC3 was replaced by the URA3 gene from plasmid Yep352, using primers YGL139W-up (URA3) and YGL139W-down (URA3). All constructs and strains generated were verified by extensive PCR analysis and selection on the appropriate media. The fet3⌬ and fet3⌬ hmx1⌬ strains were constructed as described (22,23). Rich medium (YPD) and synthetic defined medium (S.D.) were prepared as described (24). Defined iron media were prepared as described (25) using yeast nitrogen base without iron and 1 mM ferrozine, an iron chelator, in addition to the indicated hemin supplement. Lee-Buckley-Campbell media used to induce hyphal growth of C. albicans strains was made as described (26) with addition of 50 M bathophenanthroline disulfonate and 1 M hemin. Milk-Tween agar media was prepared as described (27). To create hypoxic conditions, cultures were incubated in an anaerobic chamber using a BBL-GasPak (BD Biosciences) for 5 h.
Plasmids and Libraries-A C. albicans genomic DNA library was constructed in a S. cerevisiae 2-m vector and was a kind gift from J. Berman. The S. cerevisiae genomic DNA library in 2-m vector YEp13 was obtained from ATCC. The CaFLC1 open reading frame (ORF) was amplified using primers CA_BMST_F and ca_NCSM_R, digested with SacI and NcoI, and ligated into the SacI and NcoI sites of the vector YIpDCE51 (23) to construct plasmid YIpDCE1-CaFLC1. This plasmid was linearized with StuI for integration at ADE2. Plasmids pCaFLC1, pFLC1, pFLC2, and pFLC3 were constructed using vector pYX242 and the PCR-amplified ORFs of the corresponding genes. For pCaFLC1, primers CA_BMST_F and ca_NCSM_R were used and the library clone containing CaFLC1 was the template. The PCR product was digested with BamHI and SmaI and inserted into the same sites of the vector. Primer pair YPL221-F and YPL221-R was used to amplify FLC1, YAL053-F and YAL053-R for FLC2, and YGL139-F and YGL139-R for FLC3. Genomic DNA from S. cerevisiae YPH499 was used as the template to amplify FLC1, FLC2, and FLC3. FLC1 was inserted into EcoRI/SmaI sites, FLC2 into HindII/ SmaI, and FLC3 into NcoI/ApaI of pYX242.
To construct epitope-tagged genes, a triple copy of the HA tag was amplified by PCR using primers 221-cHA-F and 221-cHA-R for FLC1 and in vivo recombination in yeast was used to insert this PCR product into the C terminus of FLC1 in plasmid pYX-FLC1 cut with SmaI. The resulting plasmid, pYX-FLC1-HA, was subsequently rescued from yeast and amplified in Escherichia coli. The same method was used to make plasmids expressing pYX-FLC2-HA and pYX-FLC3-HA. Primers 221-cHA-R and YAL053-cHA-F were used for pYX-FLC2-HA. Primers 221-cHA-R and YGL139-cHA-F were used for pYX-FLC3-HA. Functionality of carboxyl terminus HA-tagged versions of FLC1, FLC2, and FLC3 was confirmed by complementation of the flc1⌬flc2⌬ pMET3-FLC1 strain on media supplemented with methionine.
To make plasmid pRS416-FLC1-HA, the promoter region of FLC1 was PCR amplified from genomic DNA using primers prom-YPL221-F and prom-YPL221-R and digested with KpnI and EcoRI. This KpnI/EcoRI fragment and EcoRI/SmaI fragment from pYX-FLC1-HA were subsequently ligated into Kpn/ EcoRI and EcoRI/SmaI sites of pRS426. KpnI/NotI fragment was cut from this plasmid and ligated into Kpn/NotI sites of pRS416. To make plasmid pRS415-FLC2-HA, a PstI/BamHI fragment from the FLC2 library clone containing the FLC2 upstream sequences and a BamHI/SacII fragment of the FLC2 ORF from plasmid pYX-HUF2-HA were ligated into the PstI/ SacII sites of pRS415. To make plasmid pRS413-FLC3-HA, a HindIII/EcoRI fragment containing the promoter region of FLC3 from the FLC3 library clone and the EcoRI/SmaI fragment from pYX-FLC3-HA were ligated into HindIII/SmaI sites of pRS415. Then the ApaI/SmaI fragment of this plasmid was ligated into the same sites of pRS413. To make plasmid pMET3-FLC1, the EcoRI/SmaI fragment of the FLC1 gene from pYX-FLC1-HA was inserted into vector pRS313-MET3 cut with the same restriction enzymes.
To make plasmid pVAN1, the ORF of VAN1 was amplified by PCR from the VAN1 library clone using primers Van1-pr-F and Van1-pr-R and digested with XbaI and SmaI. This fragment was inserted into XbaI/SmaI sites of YEp351. The plasmid pKTR4 was made by inserting the KTR4 gene into the PstI/ HindIII sites of YEp351. KTR4 was amplified by PCR using primers KTR4-pr-f and KTR4-pr-r from the KTR4 library clone. Plasmid pPTC1 was made by inserting the PTC1 gene into the XbaI/SmaI sites of YEp351 using primers PTC1ϩpr-F and PTC1ϩpr-R from the PTC1 library clone. Plasmid pHOC1 was made by inserting HOC1 into XhoI/SacI sites of pRS426 using primers Hoc1-t-F and Hoc1-t-R and genomic DNA of S. cerevisiae YPH499 as the template. To make plasmid pMNN9, the MNN9 gene was inserted into the XbaI/ClaI sites of pRS426 using primers MnnI-t-F and MnnI-t-R and genomic DNA as template. Plasmid pMET3-CaFLC1 was constructed by subcloning the SmaI/BamHI fragment of pCaFLC1 with CaFLC1 ORF into vector pCaEXP (28). Vector was digested with PstI, treated with Klenow fragment, and then digested with BamHI. Plasmid was linearized with StuI and integrated into the RP10 locus of the C. albicans genome. Plasmids pGAL-FLC1 (pBG1800-YPL221W) and pGAL1-FAD1 (pBG1805-YDL044C) were obtained from the YEAST ORF collection (Open Biosystems). Functionality of the cloned genes was confirmed by sequencing and/or phenotypic analysis. Plasmid pJS401 with UPRE-LacZ was a gift from D. Griffith (29).
Immunofluorescence, Fractionation, and Western Blotting-The strain CRY (MATa his3 ura3 leu2 trp1 ade2) (gift from The FLC Genes of S. cerevisiae E. Cabib) was transformed with centromeric plasmids containing HA-tagged FLC genes and transformants were grown to mid-log phase and the cells were prepared for immunofluorescence microscopy as described (30). Subcellular fractionation was performed as described (31) with the following modifications. Cells were disrupted with glass beads, and unbroken cells were removed by centrifugation at 500 ϫ g for 2 min. Cell lysate was applied to the top of 20 -60% continuous sucrose gradient. Samples were centrifuged at 28,500 ϫ g for 17 h and 0.9-ml fractions were collected. Lysates and gradients were prepared with EDTA or 2 mM MgCl 2 . Western blotting was performed using a 1:4000 dilution of HA.11 (Covance) as the primary antibody followed by 1:3000 dilution of horseradish peroxidase-conjugated sheep anti-mouse antibody (Amersham Biosciences). Antibody to Dpm1, porin, and Vps10 were purchased from Molecular Probes and used according to the manufacturer's manual. Anti-Gas1p antibody was used at 1:5000 dilution (32). Antibody was detected using enhanced chemiluminescence (Amersham Biosciences).
Radiolabeling and Immunoprecipitations-Metabolic labeling and immunoprecipitation was performed as described (33). Briefly, cells were grown overnight in SD (Ϫura) with 2% raffinose as the carbon source and 0.2% galactose. Glucose was added to 2% and cells were incubated 24 h more. Cells were pulsed-labeled with L-[ 35 S]methionine for 12 min and chased with unlabeled cysteine and methionine for the indicated number of minutes. Immunoprecipitation was performed using a monoclonal anti-CPY antibody (Molecular Probes). The immunoprecipitate was analyzed by SDS-PAGE and phosphorimaging.
Electron Microscopy-S. cerevisiae R1158 and the triple mutant TET-FLC1flc2⌬flc3⌬ strains were grown to an A 600 of 0.2 in liquid YPD at 30°C with shaking. Cultures were divided, and doxycycline was added to a final concentration of 10 g/ml; the other half did not receive any additions and was used as control. Subcultures were incubated an additional 9 h. Cells were pelleted and washed twice with water and then fixed in 1.5% (w/v) potassium permanganate for 20 min at room temperature with occasional mixing. The cells were then washed with water, incubated in 0.5% sodium periodate for 20 min at room temperature, washed with water, incubated for 15 min in 1% ammonium chloride, washed with water again, and left overnight in 1% uranyl acetate at 4°C. The next day cells were washed several times with water, dehydrated in an ethanol/ water series, and finally embedded in Epon 812 resin. Ultrathin sections were viewed with a JEOL 2000 electron microscope.
Assays-Northern blot analyses were performed as described (25). Probes for CaFLC1 and ACT1 were prepared from PCR products corresponding to the open reading frame. Heme uptake was measured as described (34) with the following modifications. Yeast were grown in SD medium overnight and then diluted to an absorbance of 0.1 and incubated 5 h. Washed cells were suspended in phosphate-buffered saline containing 5% glucose, 0.05% Tween 80, and 0.5% bovine serum albumin and preincubated for 10 min at 37°C. 55 Fe-hemin was added at a final concentration of 2.3 M to 0.1 ml of cell suspension with a final absorbance of 1.0 and incubated 60 min at 37 or 4°C. To stop the reaction, 20 l of 1 mM cold hemin was added and cell suspensions were transferred to microfilter plates on ice. The cells were washed 6 -7 times with buffer without glucose. Accumulation of 55 Fe was measured by scintillation counting. Heme uptake was reported as the difference between 55 Fe accumulation at 37 and 0°C. 55 Fe-hemin was synthesized as described (35) from protoporphyrin IX (Porphyrin Products, Logan, UT) and 55 FeCl 3 (PerkinElmer Life Sciences).
Alcian blue binding was measured as described previously (36 -38). Briefly, the stain (0.1%) was prepared in 0.02 N HCl and the suspension was centrifuged to eliminate insoluble precipitates. The cells were grown for 48 h in S.D. (ϪMet), then diluted to an absorbance of 0.2 and incubated 2 days in SD with or without methionine. An absorbance of 1.0 of cells was applied to microfilter plates. Cells were washed two times with 0.2 ml of 0.9% NaCl and then supplemented with 0.2 ml of Alcian blue solution. The mixture was maintained statically at room temperature for 10 -15 min and then washed twice with 2 ml of 0.02 N HCl. ␤-Galactosidase assays were performed as described (39).
FAD uptake assays were performed using transport-competent membranes in permeabilized cells, and were prepared essentially as described (40,41) with the following modifications. Cells from 50 -100 ml of culture were harvested and suspended at a density of 50 A 600 units/ml in 100 mM Tris-HCl, pH 9.4, and 10 mM dithiothreitol and incubated 5 min at room temperature. Cells were centrifuged and re-suspended in 1 ml of 0.75 ϫ SC with or without 1 mM methionine, 0.5% glucose, 0.7 M sorbitol, and 10 mM Tris-HCl, pH 7.5, with 0.5 mg/ml zymolyase T-100 . After 30 min incubation at 30°C, spheroplasts were collected and resuspended in 0.75 ϫ SCM containing 0.7 M sorbitol and 1% glucose with or without 1 mM methionine. Cells were allowed to recover at 30°C for 20 min, then washed with lysis buffer (0.4 M sorbitol, 20 mM HEPES, pH 6.8, 150 mM potassium acetate, 2 mM magnesium acetate). Cells were resuspended in lysis buffer at 200 A 600 units/ml. Aliquots of suspensions were slowly frozen over liquid nitrogen for 60 min and stored at Ϫ80°C. Uptake of FAD was measured as described (42). Permeabilized yeast cells were thawed quickly and washed three times with ice-cold reaction buffer (20 mM HEPES, pH 6.8, 150 mM potassium acetate, 250 mM sorbitol, 5 mM magnesium acetate) and then resuspended in the same buffer. Protein concentrations in the samples were normalized using the bicinchoninic acid method (Pierce). Membranes were incubated in 1 mM FAD at 30°C for the indicated period of time. Uptake was terminated by the addition of 1 ml of ice-cold reaction buffer and samples were quickly centrifuged 16,000 ϫ g for 1 min. Membranes were washed twice in reaction buffer and solubilized in 2% Triton X-100. FAD content was measured fluorometrically at an excitation wavelength of 450 nm and an emission wavelength of 530 nm on an ISS PC fluorescent spectrophotometer. Fluorescence of vesicles incubated in the absence of FAD served as a background control.

RESULTS
Identification of a C. albicans Gene Conferring Heme Uptake-S. cerevisiae strains bearing deletions of FET3 are deficient in ferrous iron transport and do not grow on media containing low concentrations of iron salts and ferrous iron chelators (Ref. 43 and Fig. 1A), and the addition of low concentrations of hemin to these media do not appreciably stimulate growth of a fet3⌬ strain (Fig.  1B). Using a C. albicans genomic library cloned into a S. cerevisiae high copy expression vector, we identified a plasmid that stimulated the growth of the fet3⌬ strain only in the presence of hemin. An analysis of the nucleotide sequence of this clone indicated the presence of two C.albicans genes, orf19.2501 and orf19.2503. Subsequent subcloning of these genes revealed that only orf19.2501 conferred the utilization of hemin as an iron source in the S. cerevisiae fet3⌬ strain (Fig. 1B). This gene was termed CaFLC1 for flavin carrier 1 after subsequent characterization revealed a role in FAD transport. To determine whether CaFLC1 stimulated the uptake of intact hemin or whether it stimulated the extracellular release of iron from hemin, we expressed CaFLC1 in a strain from which both the intracellular heme degrading enzyme, HMX1 (22), and FET3 had been deleted. Expression of CaFLC1 in a strain lacking Hmx1p resulted in slower growth on hemin when compared with the congenic strain expressing Hmx1p, suggesting that CaFLC1 expression led to the uptake of intact hemin and that intracellular degradation of this hemin was needed for growth.
CaFLC1 Iron Regulation, Localization, and Involvement in Heme Uptake-In most organisms, iron uptake systems are homeostatically regulated by iron. We tested whether iron supplementation or deprivation influenced the expression of CaFLC1. Using Northern blot analysis, we measured CaFLC1 mRNA levels under different growth conditions. CaFLC1 transcripts were induced when cells were grown under iron deprivation ( Fig. 2A, lanes 1 and 4) and repressed when iron was replete (lanes 2 and 5). A temperature shift from 30 to 37°C did not affect transcript levels. To address whether the product of CaFLC1 was involved in heme uptake in C. albicans, we measured heme transport activity in strains lacking this gene or A, iron regulation of CaFLC1 transcripts. C. albicans strain SC5314 was grown in defined-iron medium supplemented with 10 or 200 M iron and 1 mM ferrozine (lanes 1, 2, 4, and 5) or SC without ferrozine with 1.2 M iron (lanes 3 and 6) at 30 or 37°C. Northern blot analysis was performed on total RNA with sequential hybridization of the indicated probes. B, altered hemin uptake in C. albicans FLC1-deleted and -overexpressing strains. C. albicans congenic strains of the indicated genotype with an integrated copy of MET3-CaFLC1 or the empty parent vector pCaEXP were grown in SC medium without methionine to induce CaFlc1p expression and assayed for hemin uptake. The experiments were replicated three times, and data from a representative experiment are shown. The error bars indicate the standard deviations. C-E, reduced formation of filaments in CaFLC1 deletion strains. C. albicans congenic strains of the indicated genotype were plated in serial dilutions on Lee-Buckley-Campbell medium supplemented with 1 M hemin and 50 M bathophenanthroline disulfonate. Colonies from a single dilution were photographed after 4 days of growth. expressing it at a high level. We sequentially deleted both alleles of CaFLC1 and confirmed their deletion by PCR, generating a Caflc1/Caflc1 deletion strain. For overexpression of CaFLC1, the coding region was cloned under the control of the MET3 promoter and integrated at the RP10 locus of the Caflc1/Caflc1 deletion strain. Cells were grown in methionine-free medium to stimulate CaFlc1p expression and heme uptake activity was measured by the accumulation of 55 Fe-labeled heme. The Caflc1/Caflc1 deletion strain accumulated 48% less heme than the wild-type strain, whereas the CaFlc1p overexpressing strain exhibited an 89% increase in heme uptake as compared with the wild type strain (Fig. 2B). These results indicated that CaFlc1p played a role in heme uptake in C. albicans, but the significant amount of heme uptake activity that remained in the Caflc1/ Caflc1 deletion strain indicated that CaFlc1p did not comprise the sole heme uptake system.
Because heme is potentially an important iron source for C. albicans when it invades the circulatory system of the mammalian host (9) and because C. albicans undergoes morphological changes during infection (44), we tested the effects of CaFLC1 deletion on filamentation when the fungus was maintained in hyphal growth conditions. Strains were plated onto iron-depleted Lee-Buckley-Campbell medium containing heme, and filamentation was assessed. The Caflc1/Caflc1 deletion strain exhibited reduced filamentation when compared with the wild type strain (Fig. 2, E versus C) and the CaFLC1/ Caflc1 heterozygous strain exhibited an intermediate degree of filamentation (Fig. 2D). This defect in filamentation was also observed in strains grown on milk-Tween agar, a medium that did not contain heme (data not shown). These results suggested that the impaired filamentation was not simply due to diminished heme uptake, but that CaFLC1 was likely to be involved in other cellular processes that influence filamentous growth.
We examined the cellular localization of CaFlc1p by constructing a strain in which a triple HA tag was introduced to the carboxyl terminus of FLC1 and the fusion gene under the control of the MET3 promoter was integrated at the RP10 locus. By indirect immunofluorescence, Flc1-HA was localized exclusively to punctate, intracellular vesicles, with no detectable signal on the plasma membrane or vacuolar membranes (Fig. 3, A and B).
Yeast and Fungal Homologues of CaFLC1-We examined protein sequence databases to identify homologues of CaFLC1 and found that clear homologues of CaFLC1 were present in essentially every fungal genome examined, and that multicellular eukaryotes contained only more distantly related genes. The C. albicans genome contains two ORFs that display significant homology over their entire length to CaFLC1, Orf19.1813 (35% identity) and Orf19. 19.2198 (36% identity), and these were designated CaFLC2 and CaFLC3, respectively. The S. cerevisiae genome contained three homologues of CaFLC1, YPL221W, YAL053W, and YGL139W, which exhibited 48, 36, and 47% amino acid identity to CaFLC1 and were designated FLC1, FLC2, and FLC3, respectively. A fourth ORF in S. cerevisiae (YOR365C) exhibited slightly less homology to CaFLC1 (28% identity). Each of the predicted amino acid sequences contained a signal peptide and 9 -10 transmembrane domains, suggesting that the FLC genes represent a conserved fungal gene family of integral membrane proteins that could function as transporters or carriers. Hsaing and Baillie (45) recently identified YOR365C and the FLC gene family as one of 17 "core fungal" genes that are represented in all fungal species but not in prokaryotes or nonfungal eukaryotes.
We questioned whether overexpression of S. cerevisiae FLC1, FLC2, or FLC3 in S. cerevisiae could also confer growth on hemin in a manner similar to that of CaFLC1. The ORFs corresponding to FLC1, -2, and -3 were cloned into a high copy vector under the control of the strong TPI1 promoter and the resulting plasmids were transformed into the fet3⌬ strain. Transformants were grown in iron-depleted medium, then  plated in serial dilutions on iron-depleted medium containing hemin. Overexpression of both FLC1 and FLC2, but not FLC3, could stimulate the growth of the fet3⌬ strain when hemin constituted the sole source of iron (Fig. 4), whereas overexpression had no effect on growth in the fet3⌬ strain in the absence of heme (data not shown), suggesting that the FLC genes may function similarly in S. cerevisiae and C. albicans. As S. cerevisiae is a well established and convenient model for studying eukaryotic gene function, we examined the function of the FLC genes endogenous to S. cerevisiae.
Localization of Flc Proteins to the Endoplasmic Reticulum-Because deletion and overexpression of CaFLC1 was correlated with changes in heme uptake activity in C. albicans and because the predicted amino acid sequences of the FLC genes suggested a polytopic integral membrane protein, we initially hypothesized that the FLC genes were directly involved in heme uptake on the cell surface. Although CaFlc1p was expressed in intracellular vesicles and not detected on the plasma membrane, other transporters involved in metal uptake in yeast are expressed on intracellular vesicles (30,46). We therefore examined the cellular localization of the Flc proteins in S. cerevisiae. Overexpression of Flc1p from a strong promoter resulted in the detection of Flc1p on membranes throughout the cell (data not shown). Therefore, we constructed centromeric plasmids containing FLC1, -2, or 3 with a carboxyl-terminal triple HA tag in which each gene was expressed from its endogenous promoter. Functionality of the tagged alleles was confirmed by complementation of the flc1⌬flc2⌬ strain (see Fig. 6). Strains transformed with these plasmids were examined by indirect immunofluorescence, and Flc1p-HA was detected at the periphery of the cell and in a ring-like intracellular structure (Fig. 5, A and B). This intracellular structure co-localized with the nucleus by 4Ј,6-diamidino-2-phenylindole staining, and this pattern of peripheral and perinuclear signal is typical of the yeast ER, and typically does not indicate plasma membrane localization (47)(48)(49). These data suggested that Flc1p was localized to the ER; however, localization of Flc2p and Flc3p could not be confirmed by this method, as the fluorescent signal was not detectable.
To confirm the localization of Flc1p and to determine the localization of Flc2p and Flc3p, strains bearing the centromeric plasmids were grown in SC medium to mid-log phase and the expression levels of the tagged proteins were determined by Western blotting. Although Flc1p-HA was detected under these conditions, Flc2p-HA and Flc3p-HA were barely detectable (data not shown). Published microarray studies of the yeast genome indicate that the transcription of FLC2 is increased in response to dithiothreitol (50) and that the transcription of FLC3 is increased in response to hypoxia (51). We confirmed the increased expression of Flc2p and Flc3p under these conditions by Western blot and used these conditions for membrane fractionation studies. Lysates prepared from Flc1p-, Flc2p-, and Flc3p-expressing cells were separated on sucrose density gradients and proteins were detected by Western blotting (Fig. 5C). Fractions containing Flc1p, Flc2p, and Flc3p co-migrated with fractions containing Dpm1p, a resident protein of the ER, and did not co-migrate with marker proteins for mitochondria (porin), late Golgi (Vps10p), or plasma membrane (Gas1p).
Fractionation of a Flc2p-HA lysate in the presence of Mg 2ϩ resulted in a shift of Flc2p to denser fractions that also contained plasma membrane. This shift is characteristic of ER membranes that are associated with ribosomes in the presence of Mg 2ϩ (31). These results indicate that despite different pat- Epitope-tagged copies of FLC1, FLC2, and FLC3 under the control of their endogenous promoters were cloned into the low-copy plasmids pRS416, pRS415, and pRS413, respectively, and the S. cerevisiae strain CRY was transformed with the resulting plasmids. A and B, indirect immunofluorescence of Flc1p-HA. The S. cerevisiae strain carrying pRS415-FLC1-HA was grown in SC medium and examined by indirect immunofluorescence. HA-11 was used as the primary antibody and Cy3-conjugated donkey anti-mouse was the secondary antibody. Anti-HA images are on the left, 4Ј,6-diamidino-2-phenylindole images are in the center, and the merged images are on the right. C, co-sedimentation of Flc1-HA, Flc2-HA, and Flc3-HA with Dpm1, a resident protein of the endoplasmic reticulum. Strains were grown in SC medium. Cell extracts were prepared and fractionated on 20 -60% (w/w) sucrose gradients as described and the fractions were subjected to SDS-PAGE and Western blotting. All cell extracts and gradients were prepared in the presence of EDTA except for Flc2p-HA ϩ Mg, which was prepared with 2 mM Mg 2ϩ . Fractions were numbered from the top of the gradient, and unfractionated extract served as a control (C ). JULY 28, 2006 • VOLUME 281 • NUMBER 30 terns of regulation, most of the Flc proteins are located in the ER and not on the plasma membrane, although a small amount could be located in other vesicles or be cycling between the ER and the Golgi. These results also suggested that the Flc proteins were normally not involved in heme uptake at the cell surface.

The FLC Genes of S. cerevisiae
Synthetic Lethality of FLC1 and FLC2 Deletion-To further characterize the FLC gene family, we constructed strains containing deletions of FLC1, FLC2, and FLC3. Although each of these genes was deleted individually, attempts to delete FLC2 in the flc1⌬ strain or FLC1 in the flc2⌬ strain failed, and we considered the possibility that FLC1 and FLC2 might be synthetically lethal. To test this, we cloned the FLC1 ORF into a low copy plasmid under the control of the methionine-regulatable MET3 promoter, and transformed the resulting plasmid, pMET-FLC1, into the flc1⌬ strain. FLC2 was successfully deleted in this strain when it was maintained in methioninefree medium and Flc1p was expressed from the plasmid. We tested whether deletion of both FLC1 and FLC2 was lethal by plating the congenic strains containing pMET-FLC1 on either medium without methionine (on which plasmid-born Flc1p is expressed) or on medium with methionine (on which plasmidborn Flc1p is not expressed at significant levels). Although all strains grew equally well on medium without methionine (Fig.  6A), the flc1⌬flc2⌬ strain failed to grow on medium with methionine (Fig. 6B), indicating that FLC1 and FLC2 were synthetically lethal and together encoded an essential function in yeast.
Notably, the addition of hemin in various concentrations to plates containing methionine did not rescue the lethality of the flc1⌬flc2⌬ strain. Similarly, the addition of long chain fatty acids and ergosterol, the products of essential heme-dependent biosynthetic pathways, did not rescue the lethality of the flc1⌬flc2⌬ strain. These results again suggested that the transport of heme was not the essential function of Flc1p and Flc2p and that the essential function of the FLC genes was unknown.
Impaired Cell Wall Integrity in Strains Deleted for FLC1 and FLC2-To gain clues as to the essential function of Flc1p and Flc2p, we tested various compounds for their effects on the growth of FLC-deleted strains. Addition of 1 M sorbitol to media containing methionine resulted in a partial restoration of growth to the flc1⌬flc2⌬ strain (Fig. 7B). Osmotic stabilizers such as sorbitol can rescue the growth of strains with defective cell walls. Glucosamine, which can be used by the cell for the synthesis of chitin, glycosylphosphatidylinositol anchors, and N-glycosylated proteins (52), also partially rescued the growth of the flc1⌬flc2⌬ strain (Fig. 7C). Chitin is an essential carbohy-drate component of the cell wall (53). Calcofluor White is a fluorescent dye that binds to chitin, and mutant strains with defects in cell wall synthesis frequently exhibit sensitivity to this agent as well as increased deposition of chitin in the cell wall (53). The flc2⌬ strain exhibited sensitivity to Calcofluor White (Fig. 7D) and another chitin-binding dye, Congo Red (data not shown). Furthermore, microscopic examination of the flc1⌬flc2⌬ strain stained with Calcofluor White revealed increased chitin deposition in the cell wall (Fig. 7E), especially at the bud neck, the site of maximal cell wall synthesis and chitin deposition, and bud scars. This increase in staining was not present in the flc1⌬flc2⌬ strain or the wild type strain when Flc1p was overexpressed (Fig. 7, F and G). These results all suggested that the flc1⌬flc2⌬ strain was impaired in some aspect of cell wall biosynthesis.
High Copy Suppressors of flc1⌬flc2⌬ Synthetic Lethality-To further investigate the essential function of Flc1p and Flc2p, we screened a yeast genomic library for genes that, when expressed from a high copy plasmid, could rescue the lethality and permit the growth of the flc1⌬flc2⌬ strain on methionine-containing medium. Plasmids identified in this screen were isolated and retransformed into the flc1⌬flc2⌬ strain to confirm the sup-  pressor phenotype. Plasmids were then sequenced and the individual ORFs within each plasmid were subcloned into the high copy vector and retested. The genes identified in this screen are presented in Table 2. In addition to FLC1 and FLC2, we determined that both FLC3 and CaFLC1 rescued the lethality of the flc1⌬flc2⌬ strain, which indicated that each of these genes encoded a protein of similar function.
We identified VAN1, a component of the mannan polymerase I enzyme complex (54,55), as a suppressor of the lethality of the flc1⌬flc2⌬ strain. Mannoproteins form the outer layer of the yeast cell wall and two ␣-1,6-mannan polymerase complexes (M-Pol I and M-Pol II) operate in the Golgi apparatus to attach the long mannan backbone found on certain glycoproteins (56). We tested and found that, in addition to VAN1, overexpression of HOC1, encoding a component of M-Pol II (55), could also suppress the lethality of the flc1⌬flc2⌬ strain. However, overexpression of MNN9, encoding a component of both M-Pol I and M-Pol II, did not. In a separate screen for S. cerevisiae genes that could facilitate the uptake of hemin when overexpressed, we identified KTR4, a gene that exhibits homology to the KTR family of ␣-1,2-mannosyltransferases involved in O-linked glycosylation and N-linked outer chain mannosylation (57). Overexpression of KTR4 also suppressed the lethality of the flc1⌬flc2⌬ strain. MSG5, a protein phosphatase involved in signaling through the cell integrity pathway (58), and PTC1, a protein phosphatase involved in signaling through the mitogen-activated protein kinase osmosensing cascade (59) were also identified as suppressors. Strains with defects in cell wall biogenesis have been reported to become dependent on stress-response signaling pathways to remain viable (60,61). Overexpression of these phosphatases could alter the activity of stress-response signaling pathways and permit the continued growth of the flc1⌬flc2⌬ strain. APM2 encodes an adaptor-like protein that may be involved in protein or vesicular trafficking, and may suggest that alterations in these processes facilitate heme uptake or contribute to the lethality of the flc1⌬flc2⌬ strain. Taken together, the genes identified in the suppressor screen suggested that the lethal-ity of the flc1⌬flc2⌬ strain could be rescued by an increase in the activity of outer chain mannosyltransferases.
Impaired Synthesis of Cell Wall Components in the flc1⌬flc2⌬ Strain-In yeast, subsequent to the attachment of the long ␣-1,6-mannose polymers by M-Pol I and II, branches are added by additional mannosyltransferases, and phosphomannose residues are attached at some of these branches (56). Negatively charged phosphomannose residues can be detected in the cell wall by staining with the cationic dye Alcian blue (36), and we tested the congenic wild type, flc1⌬, flc2⌬, and flc1⌬flc2⌬ strains for the presence of phosphomannose residues in the cell wall by Alcian blue staining (Fig. 8A). The flc1⌬ strain exhibited a modest decrease in staining when compared with the wild type, whereas the flc1⌬flc2⌬ strain exhibited a more dramatic decrease in staining, indicating that the loss of FLC gene expression led to a decrease in the phosphomannose content of the cell wall.
We examined the cell wall of a FLC-depleted strain by transmission electron microscopy. A strain in which FLC2 and FLC3 were deleted and FLC1 was controlled by the tetracycline-regulatable promoter (20) and the congenic parent strain were grown in the presence of doxycycline to shut off FLC1 expression. Electron microscopic analysis revealed that the FLC-depleted strain exhibited a thickened cell wall and an accumulation of amorphous material in the developing bud (Fig. 8B). In vitro synthesis of ␤-1,6-D-glucan, a major component of the yeast cell wall, was also impaired in this strain (data not shown). These data indicated that in the absence of FLC gene expression, yeast cells exhibit defects in both the mannoprotein component and the glucan component of the cell wall.
Impaired FAD Transport and Oxidative Protein Folding in the flc1⌬flc2⌬ Strain-Both N-linked core glycosylation of and disulfide bond formation in proteins within the secretory pathway occur in the ER and these processes are highly conserved among eukaryotes (53,62). Core N-glycosylation precedes and is required for outer chain mannosylation in the Golgi. Disulfide bond formation requires the oxidizing environment of the ER and proteins lacking properly formed disulfide bridges are a Capacity to stimulate growth of flc1⌬flc2⌬ strain on SC medium when expressed on high copy plasmid from endogenous promoter. b Capacity to stimulate growth of fet3⌬ strain on SC low iron medium supplemented with hemin when expressed on high copy plasmid from endogenous promoter or from the TPI1 promoter (in parentheses). c Identified as suppressor of flc1⌬flc2⌬ lethality. d NA, not applicable. e Identified as stimulator of growth on hemin. f ND, not done.

The FLC Genes of S. cerevisiae
retained in the ER. We examined the flc1⌬flc2⌬ strain for defects in early steps of N-glycosylation and oxidative protein folding by following the maturation of carboxypeptidase Y (CPY). Native CPY contains five disulfide bonds and undergoes N-glycosylation in the ER and Golgi before undergoing proteolytic processing to its mature form in the vacuole (63). Defects in N-glycosylation do not prevent the sorting of CPY to the vacuole, whereas failure to form disulfide bonds causes CPY to accumulate in the ER. Using strains bearing a copy of FLC1 under the control of the GAL1,10 promoter, CPY maturation was monitored using pulse-chase and immunoprecipitation (Fig. 8C). In wild type cells, most of the ER form of CPY was processed after 5 min and none of the ER form was detectable after 10 min of chase. In contrast, in the flc1⌬flc2⌬ strain most of the ER form of CPY was still present after 5 min and the ER form remained readily detectable after 10 min of chase. The electrophoretic mobilities of the ER and Golgi forms of CPY were similar to those of the wild type, however, suggesting that although exit from the ER was delayed, core N-glycosylation was intact.
The delay in processing of CPY raised the possibility that protein disulfide formation was impaired in the flc1⌬flc2⌬ strain. Oxidative protein folding in the ER is driven by a protein relay that transfers oxidizing equivalents to folding proteins, and the essential, ER luminal protein Ero1p initiates this process (64). Ero1p is an FAD-dependent enzyme, and Ero1p activity is highly sensitive to changes in free FAD levels. FAD is briskly transported into the ER, but the gene(s) encoding this transport activity have not been identified in any organism. Because ero1-1 mutants can be rescued by overexpression of FAD synthetase (65), we tested and observed that overexpression of this enzyme also partially rescued the lethality of the flc1⌬flc2⌬ strain (Fig. 9A), and hypothesized that the FLC genes might be involved in the transport of FAD into the ER.
We measured the transport of FAD into microsomes by growing congenic FLC deletion strains in medium with or without methionine supplementation for 5 h, then incubating permeabilized cells in FAD and measuring retained FAD by fluorescence spectrophotometry. After 5 h in methioninecontaining medium, the flc1⌬flc2⌬ culture was still increasing in density, albeit very slowly, and the cells retained a trace amount of Flc1p expression from the methionine promoter (data not shown). We found that, whereas wild type cells exhibited a robust uptake of FAD, the flc1⌬flc2⌬ strain exhibited very low uptake of FAD when grown in methionine-containing medium (Fig. 9B, left panel). Transport of FAD into wild type microsomes was time dependent and reached a maximum at 10 min (Fig. 9B, right panel), which was similar to the FAD transport kinetics identified in rat liver microsomes (42). We measured only a small amount of time-dependent FAD transport into microsomes from the flc1⌬flc2⌬ strain. Similar to rat liver microsomes, FAD uptake into wild type microsomes was inhibited by the anion transport inhibitor 4,4Ј-diisothiocyanostilbene-2,2Ј-disulfonic acid, and insensitive to EDTA (data not shown). This loss of FAD transport activity was apparent after only 5 h of methionine-induced Flc1p shut-off, whereas other phenotypes of the flc1⌬flc2⌬ strain were observed after 16 -24 h after shut-off, suggesting that the loss of FAD transport might be a direct consequence of Flc protein depletion.
Activation of the Unfolded Protein Response in the flc1⌬flc2⌬ Strain-Defects in FAD-dependent disulfide bond formation in the ER can lead to misfolding, aggregation, and retention of proteins within the ER. Yeast cells respond to the accumulation of misfolded proteins in the ER by activating the unfolded protein response (UPR), a signal transduction pathway between the ER and the nucleus (66). Activation of the UPR leads to the transcription of genes containing a UPR response element (UPRE) in their promoter regions. We addressed whether oxida- FIGURE 8. Cell wall defects in the flc1⌬flc2⌬ strain. A, decreased mannose phosphate content in the flc1⌬flc2⌬ strain. S. cerevisiae congenic strains of the indicated genotype bearing plasmid pMET3-FLC1 were grown for 2 days in medium supplemented with 1 mM methionine prior to staining with Alcian blue. B, bud cell wall defects in strains lacking FLC1, -2, and -3. R1158 and the triple mutant TET-FLC1flc2⌬flc3⌬ cells were treated for 9 h with doxycycline, fixed, and analyzed by electron microscopy. Left panels, R1158 (Wild Type) cells at ϫ6,000 (top) and ϫ11,000 (bottom) magnification. Right panels, TET-FLC1flc2⌬flc3⌬ cells at ϫ6,000 (top) and ϫ16,000 (bottom) magnification. Bars in the top panels represent 2 m; in bottom panels, 1 m. C, delayed processing of CPY in the flc1⌬flc2⌬ strain. S. cerevisiae congenic strains of the indicated genotype bearing pGAL1-FLC1 were grown in SC medium supplemented with 2% raffinose and 0.2% galactose. To shut off the GAL1,10 promoter, glucose was added to 2% and strains were additionally incubated 24 h. Cells were collected and labeled with [ 35 S]methionine for 12 min and chased with an excess of the methionine/cysteine mixture for the indicated number of minutes. CPY was immunoprecipitated and subjected to SDS-PAGE and autoradiography. The mobility of the ER-modified form (E ), the Golgi-modified form (G), and the mature vacuolar form (V ) is indicated.
tive protein folding defects in strains lacking FLC genes would activate the UPR by transforming congenic FLC deletion strains with a UPRE-LacZ reporter plasmid (29) and measuring ␤-galactosidase activity in media with and without methionine (Fig. 9C). The flc1⌬flc2⌬ strain exhibited approximately five times the activ-ity of the wild type strain when Flc1p was not expressed, indicating the accumulation of misfolded proteins in the ER.
The Role of Flc1p in Hemin Uptake-Our data suggested an essential role for the FLC genes in FAD transport into the ER, the mechanism by which FLC genes stimulated heme uptake remained unclear. CaFlc1p overexpression did not seem to alter the binding of hemin to the cell wall in S. cerevisiae (data not shown), and we postulated that Flc1p could be acting as a low specificity/low affinity transporter for hemin. Cells overexpressing Flc1p expressed the protein on membranes throughout the cell, including the plasma membrane (data not shown), yet uptake of hemin remained too low for accurate, direct measurement. However, if hemin were a low specificity substrate of a transporter encoded by FLC1, then FAD could be expected to competitively inhibit the uptake of hemin via Flc1p. We tested this hypothesis by expressing CaFlc1p in the S. cerevisiae fet3⌬ strain, and determining whether FAD could inhibit growth of this strain on medium containing hemin as the source of iron (Fig. 10). Whereas CaFlc1p could stimulate the growth of the fet3⌬ strain on hemin (Fig. 10A), the addition of excess FAD completely blocked the stimulatory effects of CaFlc1p (Fig.  10B). FAD had no effect on the slow growth of the fet3⌬ strain in the absence of CaFlc1p. These data are consistent with a mechanism in which heme is a low affinity substrate for an FAD transporter encoded by FLC family members.

DISCUSSION
A strategy designed to identify C. albicans genes involved in heme uptake unexpectedly led to the identification of a fungal FIGURE 9. Impaired FAD transport in strains lacking FLC1 and FLC2. A, partial rescue of flc1⌬flc2⌬ lethality by FAD1 overexpression. Congenic strains of the indicated genotype carrying pMET3-FLC1 were transformed with pGAL-FAD1 or the empty vector and transformants were plated in serial dilutions on SC medium supplemented with 2% raffinose, 0.2% galactose, and 0.5 mM methionine. B, impaired FAD transport in strains lacking FLC1 and FLC2. Congenic strains of the indicated genotype carrying pMET3-FLC1 were grown for 5 h in medium supplemented with 1 mM methionine. Cells were collected and permeabilized spheroplasts were assayed for FAD uptake activity. Left, FAD content of microsomes measured after 30 min of incubation in 1 mM FAD. Right, FAD content measured after the indicated times of incubation (30 s to 30 min) in 1 mM FAD. The collection of frequent time points required an abbreviated washing procedure, and resulted in higher nonspecific binding of FAD. C, activation of the UPRE in strains lacking FLC1 and FLC2. Congenic strains of the indicated genotype carrying pMET3-FLC1 were transformed with pUPRE-lacZ and transformants were incubated in SC medium with or without 1 mM methionine for 5 h. Cells were collected and assayed for ␤-galactosidase activity. The experiments were replicated two times, and data from a representative experiment are shown. The error bars indicate the S.D. The FLC Genes of S. cerevisiae JULY 28, 2006 • VOLUME 281 • NUMBER 30 gene family required for the transport of FAD into the ER. Whereas heme transport does not appear to be the primary function of the Flc proteins in S. cerevisiae, they do contribute significantly to heme uptake in C. albicans. Whether this uptake activity is direct, e.g. through low specificity uptake activity, or indirect, by affecting the folding and expression of secreted or plasma membrane proteins directly involved in heme uptake, remains to be determined. Because each of the C. albicans and S. cerevisiae Flc proteins identified in these studies can rescue the lethality of the flc1⌬flc2⌬ strain, and thereby supply the essential function lacking in this strain, they are all strongly predicted to have similar, if not identical, biochemical activities.
The Role of the FLC Genes in FAD Transport-Ample biochemical evidence supports the existence of a robust FAD transport system localized to the membranes of the ER. In yeast, FAD rapidly enters purified microsomes, where it can directly bind to and induce the oxidation of luminal proteins (see below) (65,67). Biochemical characterization of FAD transport in rat liver microsomes suggests a similar type of transport mechanism in which uptake is initially rapid, then plateaus. FAD transport is bidirectional, as an equally rapid FAD efflux system is also present. Neither process requires exogenous sources of energy and both are inhibited by anion transport inhibitors (42). The FAD transport activities measured in these yeast studies exhibits kinetic characteristics nearly identical to those reported for the mammalian microsomes. These kinetic observations also raise the possibility that microsomal FAD transport may represent facilitated diffusion or a carrier-mediated transport process in which the transporter permits free cytosolic FAD to equilibrate within the lumen of the ER. This is supported by our observation that overexpression of Flc1p did not result in higher levels of microsomal FAD accumulation in wild type cells (Fig. 9B). Although a mitochondrial carrier for FAD (Flx1p) has been identified in yeast (68), genes encoding a microsomal FAD carrier have not been identified previously in any eukaryote. Because the loss of FAD transport in our studies occurred within 5 h of methionine-induced Flc1p shut-off, and because this phenotype preceded the appearance of other phenotypes by several hours, loss of FAD transport may be a direct result of FLC gene deletion. The Flc proteins do not exhibit homology to any known family of carrier proteins, but members of this family exhibit significant sequence conservation and marked similarity in their hydrophobicity profiles. Depending on the prediction program used, each family member is predicted to have 9 -10 transmembrane domains, and thus could function directly as an FAD transporter. Alternatively, the Flc proteins could indirectly affect FAD transport through interactions with an unidentified downstream partner.
The Role of FAD in Oxidative Protein Folding-Protein disulfide formation is a crucial step in protein folding and is required by many proteins that traverse the secretory pathway. A protein relay delivers oxidizing equivalents to folding proteins in the ER. This process is initiated by the essential ER luminal protein Ero1p (69,70), which catalyzes a disulfide exchange to oxidize protein-disulfide isomerase (PDI) (71). Protein-disulfide isomerase can then catalyze the oxidation and rearrangement of disulfide bonds in folding proteins (72). Ero1p is an FAD-binding protein (65), and the activity of Ero1p is highly dependent on free FAD levels (65). Depletion of cellular riboflavin (and subsequently FAD) leads to a profound defect in oxidative protein folding, and mutations in Ero1p can be rescued by increased synthesis of FAD through overexpression of FAD synthetase. Two other proteins involved in oxidative protein folding, Erv1p and Fmo1p, also require FAD for activity (64).
Loss of FAD transport activity at the ER membrane, as occurred in the FLC-deleted strains, would be predicted to result in depletion of intraluminal FAD and loss of Ero1p activity. In the absence of active Ero1p, misfolded proteins would be expected to accumulate and trigger the UPR. Accordingly, deletion of the FLC genes resulted in activation of a UPR reporter construct (Fig. 9C). Loss of Ero1p function, which is essential in yeast, also explains the synthetic lethality of the flc1⌬flc2⌬ strain. Loss of Ero1p function would ultimately lead to depletion of active forms of many proteins within the secretory pathway, and likely accounts for the pleiotropic phenotypes observed in the flc1⌬flc2⌬ strain, including morphologically aberrant cell walls with loss of ␤-1,6-glucan synthase activity and depletion of phosphomannoproteins. Depletion of a Schizosaccharomyces pombe FLC orthologue, PKD2, also resulted in altered cell morphology and reduced glucan levels (73). Strains carrying mutations in PDI1 have also been reported to exhibit phenotypes associated with cell wall defects, such as sensitivity to caffeine and zymolyase (14).
We observed that the processing of CPY in the ER, which requires the formation of five disulfide bonds, was slightly delayed in the flc1⌬flc2⌬ strain in which expression of GAL-FLC1 was repressed by glucose (Fig. 8C). Strains lacking Ero1p or Pdi1p typically exhibit more pronounced delays in the processing of CPY, suggesting that the ER is not completely depleted of FAD and there is residual Ero1p activity after deletion of FLC1 and FLC2. This could be explained by trace levels of expression of pGAL-FLC1 or native FLC3, or by a process in which the cells stop dividing before the ER becomes fully depleted of FAD.
Members of the FLC family of genes are present in essentially every fungal genome examined to date (45). Yet no clear homologues to these genes are present in other eukaryotic genomes, despite the kinetically similar FAD transport process in mammals and clear conservation of the FAD-dependent oxidative protein folding machinery in higher eukaryotes. One possibility is that sequence homology has fallen below a level that is detected in conventional blast searches. Further investigation will be required to identify FAD transporters in these organisms.