Biochemical Journal

Research article

Identification and functional characterization of a novel Candida albicans gene CaMNN5 that suppresses the iron-dependent growth defect of Saccharomyces cerevisiae aft1Δ mutant

Chen BAI, Fong Yee CHAN, Yue WANG


In Saccharomyces cerevisiae, the transcription factor Aft1p plays a central role in regulating many genes involved in iron acquisition and utilization. An aft1Δ mutant exhibits severely retarded growth under iron starvation. To identify the functional counterpart of AFT1 in Candida albicans, we transformed a C. albicans genomic DNA library into aft1Δ to isolate genes that could allow the mutant to grow under iron-limiting conditions. In the present paper, we describe the unexpected discovery in this screen of CaMNN5. CaMnn5p is an α-1,2-mannosyltransferease, but its growth-promoting function in iron-limiting conditions does not require this enzymatic activity. Its function is also independent of the high-affinity iron transport systems that are mediated by Ftr1p and Fth1p. We obtained evidence suggesting that CaMnn5p may function along the endocytic pathway, because it cannot promote the growth of end4Δ and vps4Δ mutants, where the endocytic pathway is blocked at an early and late step respectively. Neither can it promote the growth of a fth1Δ smf3Δ mutant, where the vacuole–cytosol iron transport is blocked. Expression of CaMNN5 in S. cerevisiae specifically enhances an endocytosis-dependent mechanism of iron uptake without increasing the uptake of Lucifer Yellow, a marker for fluid-phase endocytosis. CaMnn5p contains three putative Lys-Glu-Xaa-Xaa-Glu iron-binding sites and co-immunoprecipitates with 55Fe. We propose that CaMnn5p promotes iron uptake and usage along the endocytosis pathway under iron-limiting conditions, a novel function that might have evolved in C. albicans.

  • Candida albicans
  • endocytosis
  • iron-binding protein
  • iron uptake
  • mannosyltransferase


Iron is an essential nutrient for almost all organisms. The facile nature of iron to gain and lose electrons enables it to participate as a co-factor in diverse enzymatic reactions [1,2]. Keeping free iron away from invading microbial pathogens is an important element of host innate immunity. To overcome this host defence, pathogens must develop efficient ways to compete for iron. Thus specialized iron acquisition tactics often constitute virulence factors of many pathogens such as Candida albicans, one of the top human fungal pathogens [3,4]. In human body fluids, largely because of the presence of iron-chelating proteins such as transferrin and lactoferrin, free iron concentration is ∼10−18 M, far below the minimal nutritional requirement of 10−6–10−7 M for growth [5]. On the other hand, iron reacts readily with oxygen-containing compounds to generate cytotoxic free radicals. Thus iron uptake requires tight regulation.

Iron transport has been studied extensively in Saccharomyces cerevisiae. The iron-uptake systems found so far can be divided into two categories on the basis of affinities for iron. Among the low-affinity transporters is a plasma membrane protein Fet4p that has a Km of 30 μM and functions under iron-replete conditions [6]. Fet4p has a broad specificity for bivalent metal ions, such as copper, cobalt, manganese and zinc, as well as iron [7]. Another low-affinity metal ion transport system is encoded by SMF1, 2 and 3 [8,9], orthologues of the Nramp family that play key roles in iron transport in mammals [10]. Under metal starvation, Smf1p accumulates at the cell surface, while Smf2p localizes to intracellular vesicles. Smf3p constitutively resides at the vacuolar membrane, transporting vacuolar iron into the cytosol [11].

Fet3p–Ftr1p is a plasma membrane high-affinity iron transporter with a Km of 0.15 μM [12]. Fet3p is a ferroxidase that converts ferrous into ferric iron and Ftr1p is a permease that translocates the ferric iron into cytoplasm [13,14]. Ftr1p contains a putative iron-binding motif Arg-Glu-Gly-Leu-Glu at amino acid positions 157–161 that is essential for iron uptake. Substituting alanine for either one of the glutamate residues completely abolishes the iron transport activity [14,15]. Another ferroxidase–permease complex resides at the vacuolar membrane, encoded by FET5 and FTH1, which are homologues of FET3 and FTR1 respectively. This complex performs vacuole–cytosol iron transport [16]. Many microbes produce ferric chelators called siderophores with high affinities for iron [17]. Microbes can take up siderophore–iron complexes via specific transporters. S. cerevisiae contains four siderophore–iron transporters, Arn1, 2, 3 and 4p, with distinct specificities [18,19]. Recently, Fischbach et al. [20] reported that the IroB protein of pathogenic Escherichia coli and Salmonella enterica strains catalyses the C-glucosylation of a siderophore, enterobactin, increasing the availability of enterobactin to the pathogen in serum.

Most genes involved in high-affinity iron transport in S. cerevisiae are activated when the iron supply is limiting and repressed when replete. This regulation is primarily controlled by a transcription factor Aft1p [21], which activates approx. 20 genes [2123]. Thus aft1-null mutants cannot grow in iron-limiting media [21].

Endocytosis also has a role in iron uptake in S. cerevisiae. The end4 mutant, which is defective in endocytosis, exhibits delayed growth under iron starvation [24]. A fraction of molecules brought into cells via endocytosis end up in the vacuoles for storage, usage, recycling or degradation. The vacuole was designated as an iron-storage organelle [25], from which iron is exported into cytosol by Fet5p–Fth1p and Smf3p [11,16].

In comparison with S. cerevisiae, much less is known about iron transport in C. albicans. A high-affinity system was found consisting of CaFet3p and CaFtr1p [4,26], homologues of Fet3p and Ftr1p of the budding yeast. Cafet3Δ mutants exhibit growth defects in iron-limiting media, but remains virulent [26], whereas CaFTR1 deletion causes severe iron-dependent growth defects and a complete loss of virulence in mice [4]. A single siderophore–iron transporter CaArn1p has also been found and characterized [27]. Caarn1Δ cannot use siderophore–iron, but is still virulent. Although CaFTR1 and CaARN1 expression is regulated by iron concentration [4,27], the transcription factors that are involved are not clear. A transcription repressor CaTup1p was shown to have a role, but the functional counterpart of S. cerevisiae Aft1p has not been found.

In the present study, we conducted a genetic screen to identify C. albicans genes that may suppress the iron-dependent growth defects of S. cerevisiae aft1Δ mutants. We present the discovery and functional characterization of CaMNN5 that promotes cell growth under iron-limiting conditions by a previously unknown mechanism.


Strains and growth conditions

S. cerevisiae and C. albicans strains are listed in Table 1. S. cerevisiae strains were routinely grown in YPD medium (1.0% yeast extract, 2.0% peptone and 2.0% glucose), minimal medium containing yeast nitrogen base without amino acids and 2.0% glucose (GMM or glucose minimal medium) or 2.0% galactose (GaMM or galactose minimal medium). All strains were grown at 30 °C, except end4 mutants which were grown at 22 °C. Iron-limiting media were prepared in most cases by adding 1 mM ferrozine and 50 μM ferrous ammonium sulphate to the minimal media. To create stringent iron-limiting conditions, 200 μM BPS (bathophenanthroline sulphate) was added to GMM or GaMM.

View this table:
Table 1 S. cerevisiae and C. albicans strains used in the present study

Gene disruption

AFT1-, FTH1- and SMF3-deletion mutants were generated from CRY2α by following the strategy described in [4]. A gene-deletion cassette was constructed by flanking a marker gene, HIS3 or LEU2, with two DNA fragments corresponding to the 5′- and 3′-untranslated regions of the target gene. The primers used for PCR-amplification of the DNA fragments are given in the Supplementary Data (available at Strain genotype was verified by Southern blotting.

Site-directed mutagenesis

Mutations of the Lys-Glu-Xaa-Xaa-Glu and Asp-Xaa-Asp motifs of CaMnn5p were created by using the QuikChange™ Site-Directed Mutagenesis kit (Stratagene). The oligonucleotides used for mutagenesis are listed in the Supplementary Data (available at Mutations were verified by DNA sequencing.

Indirect immunofluorescence staining and Western blotting

Mouse monoclonal antibody 12CA5 (Roche Applied Science) was used as a primary antibody in both indirect immunofluorescence staining of HA (haemagglutinin)-tagged proteins and Western blot analysis. For Western blotting, horseradish-peroxidase-conjugated sheep anti-mouse IgG antisera (Amersham Biosciences) was used, and, for indirect immunofluorescence staining, Cy3-conjugated goat anti-(mouse IgG) antibody was used. For Western blotting, yeast cells were grown to 1×107 cells/ml in 10 ml of GaMM. The culture was centrifuged at 3000 g for 5 min. The cell pellet was resuspended in 300 μl of ice-cold lysis buffer containing 1% (v/v) Triton X-100, 0.1% (w/v) SDS, 50 mM Tris/HCl (pH 7.2), 1% (w/v) sodium deoxycholic acid and one protease inhibitor mixture tablet (Roche Molecular Biochemicals)/25 ml. The cell suspension was then transferred to a 2-ml screw-cap tube. An equal volume of glass beads was added, and the cells were broken by using the Mini-Beadbeater (Biospec Products) at the maximum speed for two cycles of 2-min beating on ice. The cell lysate was spun at maximum speed in an Eppendorf microfuge for 5 min and the supernatant was transferred to a new 1.5-ml tube. Protein concentration was determined by the Bradford assay (Bio-Rad). Proteins were resolved by SDS/10% PAGE, blotted on to Hybond-C membrane (Amersham Biosciences) and detected by using the ECL® (enhanced chemiluminescence) Western blot detection kit (Amersham Biosciences). Indirect immunofluorescence staining was performed as described in [30].

Cell fractionation

Subcellular fractionation was performed as described in [31] with some modifications. Cells were grown in 500 ml of GaMM to a D600 of ∼0.9 before centrifugation at 3000 g for 10 min. The cell pellet was resuspended in 50 ml of 200 mM Tris/HCl (pH 8.0), 1 mM EDTA and 1% (w/v) 2-mercaptoethanol. The cell suspension was centrifuged again at 3000 g for 10 min, and the pellet was resuspended in 50 ml of 1.2 M sorbitol solution, containing 20 mg of lyticase, and digested at 30 °C for 60 min. The spheroplasts were washed once with 1.2 M sorbitol and then lysed with a Dounce homogenizer in 4 ml of an ice-cold buffer containing 50 mM potassium phosphate buffer (pH 7.5), 200 mM sorbitol, 1 mM EDTA and protease inhibitor mixture. The cell lysate was centrifuged at 500 g for 5 min, and then the supernatant was cleared by centrifugation at 13000 g for 15 min, loaded on to the top of a 22–60% sucrose gradient and centrifuged at 170000 g at 4 °C for 18 h. Serial fractions of 0.75 ml were collected, and 40 μl of each was resolved by SDS/10% PAGE.

59Fe uptake and 55Fe protein-binding assays

59Fe uptake was measured as described in [4] with some modifications. S. cerevisiae strains were first grown to saturation in GMM or GaMM according to strain background and then washed twice with 10 mM EDTA and once with GMM or GaMM containing 200 μM BPS. The cells were then resuspended in the GMM/BPS or GaMM/BPS to a density of 1×107 cells/ml and grown for 6 h to deplete intracellular iron. Cells were then counted, spun down at 3000 g for 5 min, and divided into 50 μl aliquots in GaMM or GaMM/BPS medium, each containing ∼2×107 cells. 59Fe was added to a final concentration of 2 μM, and the cells were incubated at 30 °C. One such aliquot of each strain was kept on ice to estimate background 59Fe binding. After incubation, cells were collected and washed twice with ice-cold 10 mM EDTA to remove free 59Fe. The cell pellet was then resuspended in 3 ml of scintillation fluid for radioactivity counting in an Amersham γ-counter. After subtracting the background, the iron-uptake rate is expressed as femtomoles of 59Fe/106 cells/min. For the assays in the presence of 200 μM BPS, iron uptake is expressed as total femtomoles of 59Fe/106 cells at the end of incubation.

For 55Fe-binding assay, the strains expressing HA-tagged proteins were first grown in the presence of 200 μM PBS for 6 h to deplete intracellular iron and then in GaMM containing 5 μM 55Fe for 12 h. Total protein was extracted as described above, and the cell lysate was mixed with 6 μl of anti-HA antibody-coated (ProFound™ HA Tag Co-IP kit; Pierce) or naked agarose beads in a Handee™ Mini-spin column. The mixtures were incubated at 4 °C for 3 h with gentle mixing. Then the beads were spun down and washed twice with TBS-T buffer (25 mM Tris/HCl, pH 7.2, 0.15 M NaCl and 0.05% Tween-20). The HA-tagged protein was eluted with 50 μl of elution buffer (from the kit) and mixed with 3 ml of scintillation fluid. An aliquot of the whole-cell lysate was used to determine 55Fe uptake into the cell and an aliquot of the eluted protein was examined by anti-HA Western blotting.

LY (Lucifer Yellow) endocytosis assay

To assay fluid-phase endocytosis, LY (Sigma) was added to exponential-phase yeast cells to a final concentration of 1 mg/ml. After 2 h, cells were washed three times with PBS containing 10 mM sodium azide and three times with PBS. Cells were then digested in 1.2 M sorbitol containing 0.2 mg/ml zymolase at 30 °C for 30 min. The spheroplasts were washed once with 1.2 M sorbitol before lysis by sonication for 30 s in ice-cold PBS containing 0.05% (v/v) Triton X-100. Fluorescence was determined by using a PerkinElmer LS-55 spectrofluorimeter with λexcitation at 423 nm and λemission at 530 nm, normalized to a LY standard curve.

Expression of CaMnn5p in Pichia pastoris

The expression of CaMnn5p was carried out using the Pichia Expression kit (Invitrogen) following the manufacturer's instructions.

Assay of CaMnn5p enzyme activity

Mannosyltransferase activity was assayed as described in [32]. The P. pastoris culture supernatant was used directly as the enzyme source in a buffer containing 50 mM Tris/HCl (pH 7.2), 10 mM MnCl2, 0.23 μM GDP–[14C]mannose (62 nCi; specific radioactivity 289 mCi/mmol), 20 mM 6-O-α-D-mannopyranosyl-D-mannopyranose acceptor (Sigma) and 5 mg/ml BSA. Standard reactions were performed for 60 min at 30 °C in a final volume of 50 μl. To terminate the reaction, the mixtures were passed through 0.8 ml of QAE-Sephadex anion-exchange resin (Amersham Biosciences) to remove unlabelled GDP–mannose. The 14C-labelled products were eluted with 0.8 ml of water, and radioactivity was counted in 3 ml of scintillation fluid.


Identification of C. albicans genes that suppress the growth defects of S. cerevisiae aft1δ under iron-limiting conditions

We transformed a C. albicans 2μ genomic DNA library into the aft1Δ mutant and screened for clones that grew on iron-limiting GMM plates. After 1 week of incubation at 30 °C, colonies were picked and grown for the recovery of the transforming plasmids and DNA sequencing of the inserts. The plasmids from 80% of the plasmids contained an identical ORF (open reading frame) of 1791 base pairs (bp). In the C. albicans sequence database (, this ORF (19.2347) is annotated as CaMNN5, a homologue of S. cerevisiae MNN5, encoding a Golgi α-1,2-mannosyltransferase. CaMnn5p is approx. 30% identical with Mnn5p and another S. cerevisiae homologue Mnn2p. To confirm that CaMNN5 is indeed responsible for suppressing the defects of aft1Δ, the ORF, together with ∼600 bp of its native promoter, was cloned in the 2μ vector. Figure 1(A) shows that this construct was able to allow aft1Δ to grow on the iron-limiting plate. CaMNN5 was also expressed from the Gal1-10 promoter in a CEN plasmid. This plasmid exhibited the same growth-promoting effect on aft1Δ on GaMM only, but not GMM iron-limiting plate (Figure 1B). The results confirmed that CaMNN5 is responsible for rescuing the growth defect of aft1Δ. When CaMNN5 with its own promoter was cloned in a CEN plasmid, it did not rescue the defects of aft1Δ (results not shown). However, when driven by the FTR1 (see Figure 6) or ADH1 (results not shown) promoter from the 2μ vector, it could effectively rescue the defects of aft1Δ. The results indicate that CaMNN5 expression level is important for the observed growth-promoting function.

Figure 1 CaMNN5 enhances aft1Δ growth on iron-limiting media

(A) CaMNN5 on a 2μ vector with its native promoter (aft1Δ:CaMNN5) suppressed the growth defect of aft1Δ on the iron-limiting GMM plate containing 1 mM ferrozine and 50 μM ferrous ammonium sulphate (FAS). The wild-type strain and aft1Δ transformed with the empty vector (CRY2α:vector and aft1Δ:vector respectively) were included for comparison. The GMM plate without ferrozine is iron-sufficient. (B) aft1Δ transformed with a CEN plasmid harbouring CaMNN5 driven by the Gal1-10 promoter (aft1Δ:CaMNN5) grew on the iron-limiting GaMM, but not on GMM plates. CRY2α and aft1Δ transformed with the empty plasmid were used as controls. Approx. 104, 103 and 102 cells of each strain were spotted on to the plates and incubated at 30 °C for 5 days.

Like other known Golgi mannosyltransferases, CaMnn5p contains a predicted transmembrane domain near the N-terminus. Interestingly, it also contains three Lys-Glu-Xaa-Xaa-Glu motifs, Lys131-Glu-Tyr-Leu-Glu135, Lys229-Glu-Phe-Cys-Glu233 and Lys587-Glu-Thr-Ala-Glu591 (referred to as motif I, II and III). Arg/Lys/His-Glu-Xaa-Xaa-Glu motifs are present in diverse iron-binding proteins, such as mammalian ferritin light chain, fungal iron permeases and the bacterial iron sensor PmrB (Figure 2A). Evidence for direct interaction between these motifs and iron has been reported for amphibian ferritin light chain and Salmonella PmrB [28,29]. Mutation of any one of the glutamate residues to alanine in a conserved Arg-Glu-Gly-Leu-Glu motif in Ftr1p and CaFtr1p completely abolishes the iron-uptake function of the protein [14,15]. S. cerevisiae Mnn2p and Mnn5p do not contain any such motifs.

Figure 2 CaMnn5p contains three potential iron-binding Lys-Glu-Xaa-Xaa-Glu motifs

(A) The Lys-Glu-Xaa-Xaa-Glu motifs of CaMnn5p were aligned with some conserved motifs of several known iron-binding proteins: S. cerevisiae (Sc) Ftr1p (amino acids 157–161) and Fth1p (amino acids 242–246), C. albicans (Ca) Ftr1p (amino acids 157–161), Salmonella enterica (Se) PmrB (amino acids 38–42, 63–67), Escherichia coli (Ec) PmrB (amino acids 38–42, 60–64), Klebsiella pneumoniae (Kp) PmrB (amino acids 36–40, 62–66) and Yersinia pestis (Yp) PmrB (amino acids 29–33, 54–58). (B) Each of the Lys-Glu-Xaa-Xaa-Glu motifs of CaMnn5p could functionally replace the Arg157-Glu-Gly-Leu-Glu161 motif of CaFtr1p. All strains were grown on iron-limiting GMM plates at 30 °C for 5 days. (C) The Lys-Glu-Xaa-Xaa-Glu motifs of CaMnn5p were mutated by site-directed mutagenesis. Cells of each strain were spotted on to iron-limiting GMM and GaMM plates for incubation at 30 °C for 5 days. The table summarizes the ability of each mutated gene in promoting aft1Δ growth on the GaMM plates. None of the strains grew on the GMM plates.

The Lys-Glu-Xaa-Xaa-Glu motifs of CaMnn5p can functionally replace the Arg157-Glu-Gly-Leu-Glu161 motif of CaFtr1p

Statistically, Lys-Glu-Xaa-Xaa-Glu motifs occur only once every 8000 amino acids. The presence of three such motifs in CaMnn5p suggests that they might have a role in suppressing the iron-dependent growth defect of aft1Δ. To evaluate the importance of these motifs in CaMnn5p's activity, we first determined whether each of the motifs can functionally replace the essential putative iron-binding motif Arg157-Glu-Gly-Leu-Glu161 of CaFtr1p [4]. CaFTR1 was cloned in plasmid CIp10 and then the Arg157-Glu-Gly-Leu-Glu161 motif was replaced by each of the three Lys-Glu-Xaa-Xaa-Glu motifs of CaMnn5p, yielding three mutated genes Caftr1-I, II and III. The three constructs were transformed into Caftr1Δ, and transformants were grown on the iron-limiting plates. We found that all three constructs were able to rescue the growth defect of Caftr1Δ as effectively as the wild-type gene (Figure 2B).

Next we asked whether these motifs are required for CaMnn5p to rescue the growth defects of aft1Δ. We systematically introduced mutations into the Lys-Glu-Xaa-Xaa-Glu motifs, individually or in combination, and tested each for its ability to promote aft1Δ growth under iron-limitation conditions. We mutated the first glutamate residue of each motif to alanine. Five mutated genes were generated. Camnn5-Imu, -IImu and -IIImu each had a single motif mutated, Camnn5-(I+II)mu had I and II mutated, and Camnn5-(I+II+III)mu had all three mutated. Each mutated gene was cloned in a CEN plasmid driven by the Gal1-10 promoter and transformed into aft1Δ. Figure 2(C) shows that the aft1Δ mutant expressing the CaMnn5p with a single motif mutated grew equally well as the one expressing the wild-type CaMnn5p on the iron-limiting plate. However, the aft1Δ mutants expressing the CaMnn5p with both I and II or all three motifs mutated exhibited retarded and no growth respectively. The results suggest that these three potential iron-binding sites in CaMnn5p may have a direct role in suppressing the growth defects of aft1Δ, although they appear to be functionally redundant. However, we cannot rule out the possibility that the mutations may have caused a structural change of the protein, which led to the loss of its activity.

CaMnn5p functions independent of the known high-affinity iron transporters of S. cerevisiae

We next set out to investigate the mechanism by which CaMNN5 promotes aft1Δ growth. The most likely mechanism responsible for iron uptake under the iron-limiting conditions we used is the high-affinity iron transporter Fet3p–Ftr1p. To determine whether CaMNN5 expression may increase the cellular level or activity of Fet3p–Ftr1p, we first used Northern blotting to examine FET3 and FTR1 expression, but did not detect any increase of the mRNA levels as a result of CaMNN5 expression (results not shown). We then tested whether CaMNN5 can enhance the growth of mutants defective in the high-affinity iron transport systems, including ftr1Δ, ftr1Δ fet3Δ and ftr1Δ fth1Δ. None of these mutants grew on the iron-limiting plates, whereas introducing CaMNN5 in these mutants enabled all of them to grow (Figure 3A). Since there is no high-affinity iron permease in ftr1Δ fth1Δ, we conclude that the function of CaMnn5p in promoting aft1Δ growth in iron-limiting conditions is independent of the high-affinity iron transport systems.

Figure 3 CaMNN5 promotes cell growth under iron-limiting conditions by a mechanism independent of the high-affinity iron transporters

(A) CaMNN5 expression promoted the growth on iron-limiting plates of S. cerevisiae mutants deleted of high-affinity iron transporter genes, such as ftr1Δ, ftr1Δ fet3Δ and ftr1Δ fth1Δ. CaMNN5 was transformed into each mutant on the CEN plasmid driven by the Gal1-10 promoter. The cells were spotted on to iron-limiting GaMM plates. (B) CaMNN5 promoted the growth of the wild-type strain under more stringent iron-limiting conditions. The same plasmid expressing CaMNN5 described in (A) was transformed into CRY2α, and the transformants were grown on GaMM containing 200 μM BPS. The plate was incubated at 30 °C for 5 days.

CaMNN5 can also enhance the growth of wild-type S. cerevisiae cells under iron-limiting conditions

We consistently noted that the aft1Δ mutant expressing CaMNN5 grew better than the isogenic wild-type strain on the iron-limiting media used (results not shown). Next we tested whether CaMNN5 would allow the wild-type cells to grow under the iron-limiting conditions where the high-affinity uptake system cannot function. We established previously that there is little iron uptake into the cells in GMM containing 200 μM BPS, a strong ferrous chelator [4]. Figure 3(B) shows that, under these conditions, the wild-type strain grew much more slowly than the strain expressing CaMNN5.

CaMnn5p has α-1,2-mannosyltransferase activity

We next examined whether CaMnn5p has α-1,2-mannosyltransferase activity. We used the P. pastoris expression system to produce soluble secreted protein. The region of CaMNN51 encoding amino acids 28–597 was cloned in an expression vector in-frame with the α-factor secretion signal peptide sequence, transformed into P. pastoris, and CaMNN5 integrated behind the AOX1 promoter. A copy of the empty vector was also transformed into P. pastoris for a negative control. CaMNN5 expression was induced for 24 h, and a single protein band of CaMnn5p was detected from 20 μl of culture supernatant by SDS/PAGE, whereas no protein band was detected from the negative control supernatant (Figure 4A). Aliquots of 10 μl (approx. 0.1 μg of CaMnn5p) of the supernatants were assayed directly for α-1,2-mannosyltransferase activity. This assay detects the transfer of 14C-labelled GDP–mannose to an acceptor 6-O-α-D-mannopyranosyl-D-mannopyranose, which is specifically catalysed by α-1,2-mannosyltransferases [33,34]. For control, 0.1 μg of BSA was added to the negative control supernatant. Figure 4(B) shows that the control supernatant produced a background 14C count of approx. 5×106 c.p.m./mg of protein per h, while the supernatant of the CaMnn5p-expressing strain resulted in a count of approx. 1.5×108 c.p.m./mg of protein per h, 30 times that of the control. The results confirm that CaMnn5p has α-1,2-mannosyltransferase activity. Two transferase-dead forms of the enzyme exhibited only a background level of 14C counts (see below, and Figures 4A and 4B).

Figure 4 Mannosyltransferase activity of CaMnn5p is not required for promoting cell growth

(A) CaMnn5p and two mutant forms were expressed in P. pastoris. Aliquots of 20 μl of culture supernatant from each of the strains expressing CaMnn5p, CaMnn5p (D282A) or CaMnn5p (D284A) [see below for the description of the mutants in (C)] were examined by SDS/10% PAGE. The gel was stained with Coomassie Blue. (B) The supernatants from the same cultures described in (A) were assayed directly for mannosyltransferase activity. The assay was conducted three times for each sample and results are means±S.D. The results are expressed as activity/mg of protein per h. Protein concentration was determined by the Bradford assay. (C) Amino acid sequence alignment of a segment of the catalytic region of three α-1,2-mannosyltransferases, S. cerevisiae Mnn2p and Mnn5p and C. albicans CaMnn5p. The arrows indicate two essential aspartate residues. (D) Expression of MNN2 and MNN5 did not promote the growth of aft1Δ on the iron-limiting plates. CaMNN5, MNN2 and MNN5 were cloned on the same CEN plasmid driven by the Gal1-10 promoter. Each was transformed into aft1Δ, and the transformants were spotted on to iron-limiting GaMM plates. FAS, ferrous ammonium sulphate. (E) HA-tagged CaMNN5, MNN5 and MNN2 were transformed into aft1Δ. The transformants were grown in liquid GaMM before total protein was extracted for Western blot analysis.

Mannosyltransferase activity of CaMnn5p is not required for suppressing the growth defect of aft1δ

We examined whether the mannosyltransferase activity is required for CaMnn5p to suppress the growth defects of aft1Δ. It is well-established that mannosyltransferases contain a conserved Asp-Xaa-Asp motif in the catalytic domain, and the mutation of either one of the aspartate residues completely inactivates the enzyme [34]. In CaMnn5p, this motif (amino acids 282–284) is present in a highly conserved region (Figure 4C). We substituted an alanine residue for the first or the second aspartate (D282A or D284A) and transformed the mutated genes into aft1Δ under the control of Gal1-10 promoter. We found that both mutated forms could still suppress the defects of aft1Δ (results not shown), and, when expressed from P. pastoris, they did not exhibit any detectable enzymatic activity (Figures 4A and 4B). The results indicate that the mannosyltransferase activity of CaMnn5p is not required for the promotion of cell growth under iron-limiting conditions.

We next asked whether S. cerevisiae MNN5 and MNN2 are able to rescue the growth defect of aft1Δ when similarly expressed. We found that neither one could enhance aft1Δ growth on the iron-limiting plates (Figure 4D). To ensure that the proteins were expressed to similar levels in the transformants, we constructed a new set of constructs by adding a HA tag to the C-terminus of CaMnn5p, Mnn2p and Mnn5p. Again, only the construct expressing CaMnn5p–HA rescued the defects of aft1Δ (results not shown). Western blot analysis showed comparable levels of expression for all three proteins (Figure 4E). These results suggest further that it is not the mannosyltransferase activity, but some novel activity unique to CaMnn5p, that is responsible for promoting cell growth under iron-starvation conditions.

CaMnn5p enhances a slow process of iron uptake

Next we tested whether CaMnn5p can enhance iron uptake. We compared iron uptake in the wild-type, aft1Δ and aft1Δ mutant expressing CaMNN5 driven by the Gal1-10 promoter. Iron uptake was first assayed in GaMM containing 2 μM 59Fe for 30 min. Figure 5(A) shows that aft1Δ and the aft1Δ mutant expressing CaMNN5 exhibited a similar low 59Fe-uptake rate of less than 3 fmol/106 cells per min. In comparison, the wild-type exhibited an uptake rate of approx. 150 fmol/106 cells per min, which is largely due to the activity of Ftr1p–Fet3p. This experiment shows that, under these conditions, CaMnn5p did not result in any significant increase in iron uptake. We next tested whether CaMnn5p might contribute to a less efficient iron-uptake process. Considering that the basal level of Ftr1p–Fet3p activity in aft1Δ might mask the low activity of CaMnn5p in iron uptake, we added 200 μM BPS to the assay medium and repeated the assay over an extended period of time. Under these conditions, the high-affinity iron uptake was blocked as shown by the drastically reduced iron uptake into the wild-type (Figure 5B). The amounts of 59Fe accumulated in the CaMNN5-expressing aft1Δ cells at 4 and 16 h were approx. 9 and 24 fmol/106 cells respectively, which are ∼3.2 and 3.1 times of those detected in aft1Δ cells at the same time points. A similar increase of iron uptake was also detected in the ftr1Δ and CRY2α strains expressing CaMNN5 (Figure 5C). The results demonstrate that CaMnn5p may enhance iron uptake by a mechanism much less efficient than and independent of the high-affinity iron-uptake system.

Figure 5 CaMNN5 enhances a slow process of iron uptake

(A) Strains CRY2α:vector, aft1Δ:vector and aft1Δ:CaMNN5 driven by the Gal1-10 promoter were first grown in GaMM to exponential phase. Then the same number of cells of each strain was assayed for iron uptake in the same medium supplemented with 2 μM 59Fe for 30 min. 59Fe uptake was determined by scintillation counting and expressed as fmol/106 cells per min. (B) The same strains used in (A) were incubated in GaMM containing 200 μM BPS and 2 μM 59Fe for 1, 4 and 16 h. The iron uptake is expressed as the total amount of 59Fe accumulated in 106 cells at the end of incubation. (C) Strains CRY2α and ftr1Δ transformed with CaMNN5 driven by the Gal1-10 promoter or the empty vector were incubated in GaMM containing 200 μM BPS and 2 μM 59Fe for 0.5, 2 and 8 h. Each assay was repeated three times.

The enhancement of cell growth by CaMNN5 depends on endocytosis

A significant fraction of vacuolar iron was brought in via endocytosis in S. cerevisiae [24]. The end4Δ mutant, which is defective in both receptor-mediated and fluid-phase endocytosis, accumulated much less vacuolar iron than the wild-type [24]. We next asked whether CaMnn5p function might be dependent on endocytosis. We expressed CaMNN5 in end4Δ under the control of the FTR1 promoter, which is activated by iron shortage [14], to see whether CaMNN5 can promote end4Δ growth in iron-limiting media. As shown in Figure 6(A), while CaMNN5 markedly enhanced the growth of the wild-type strain on the medium containing 200 μM BPS, it had no such effect on end4Δ. The results indicate that CaMnn5p's function is End4p- or possibly endocytosis-dependent. VPS4 encodes an AAA-type ATPase, which is required for efficient transport out of the pre-vacuolar endosome, a late step of endocytosis [35]. vps4 mutants are impaired in the transport of endocytosed fluorescent dyes, plasma membrane receptors and ligands from early to late endosomes [36]. We found that CaMNN5 expression in vps4Δ also failed to enhance cell growth (Figure 6A). Together, the results show that the intact endocytosis pathway is essential for CaMnn5p function.

Figure 6 CaMnn5p-mediated iron uptake in mutants defective of the endocytic pathway

(A) CRY2α, end4Δ and vps4Δ mutants were transformed with a 2μ plasmid carrying CaMNN5 under the control of FTR1 promoter or with the empty plasmid. The cells were spotted on to GMM plates containing no chelator or 200 μM BPS and incubated at 25 °C for 7 days. (B) CRY2α, fth1Δ, smf3Δ and fth1Δ smf3Δ strains were transformed with the same plasmids described in (A) and grown on GMM plates containing 200 μM BPS at 30 °C for 5 days. (C) 59Fe uptake assay was performed in GMM containing 200 μM BPS and 2 μM 59Fe for 30 and 120 min. The total amount of 59Fe accumulated at the end of incubation is shown. The amount of 59Fe associated with the cells kept on ice was subtracted. (D) LY uptake into CRY2α or end4Δ cells transformed with CaMNN5 or vector alone. There was no detectable LY uptake when the cells were kept on ice. The levels of cellular LY are expressed as μg of LY/mg of protein. (E) Co-immunoprecipitation of 55Fe with CaMnn5p–HA and Mnn5p–HA. The left-hand panel shows the amount of cellular 55Fe accumulation as c.p.m./g of cell pellet, and the inset shows the expression levels of the two HA-tagged proteins determined by Western blot analysis. The right-hand panel shows the amount of 55Fe (c.p.m./g of cell pellet) co-immunoprecipitated with CaMnn5p–HA and Mnn5p–HA. The HA-tagged protein was precipitated by mixing cell lysate with anti-HA antibody-coated agarose beads (open bars) or with naked agarose beads (closed bars). All assays were repeated three times.

Iron acquired through endocytosis is likely to reach the vacuole before being exported to the cytosol for cell use. We thought that if CaMnn5p's function is endocytosis-dependent, it may not be able to enhance the growth of mutants blocked in the vacuole–cytosol iron transport. The Fet5p–Fth1p complex and Smf3p are known to be responsible for iron export from the vacuoles. We created smf3Δ, fth1Δ and smf3Δ fth1Δ mutants to determine whether CaMNN5 expression can enhance their growth on iron-limiting plates. As shown in Figure 6(B), CaMNN5 expression in the wild-type, smf3Δ and fth1Δ strains enhanced cell growth to a similar extent, but had no such effect on smf3Δ fth1Δ. All of the strains grew equally well on iron-sufficient plates. Together, the results support our view that CaMnn5p may function along the endocytic pathway. We also examined 59Fe uptake into the CaMNN5-expressing end4Δ and smf3Δ fth1Δ strains and found that it enhanced 59Fe uptake into smf3Δ fth1Δ to the same extent as it did to the wild-type, aft1Δ and ftr1Δ strains (Figure 6C). However, it did not increase iron uptake into end4Δ. The results indicate that END4-deletion blocked the endocytosis-mediated iron uptake, while deletion of SMF3 and FTH1 impaired iron export from the vacuole without any detectable effect on iron uptake into the cell.

Next we addressed the possibility that the CaMnn5p-dependent iron uptake may be a result of increased liquid-phase endocytosis. We used LY as a marker for fluid-phase endocytosis. We found that, while end4Δ showed much less cellular LY accumulation than the wild-type, CaMNN5 expression did not cause any detectable increase in LY uptake in the wild-type or end4Δ cells (Figure 6D), indicating that CaMNN5 does not enhance the overall fluid-phase endocytosis.

To assess the possibility that CaMnn5p may be involved in iron binding, we expressed HA-tagged CaMnn5p in aft1Δ and examined whether iron can be co-immunoprecipitated with CaMnn5p-HA. A strain of aft1Δ expressing S. cerevisiae Mnn5p–HA was included for comparison. The strains were grown in the presence of 5 μM 55Fe for 12 h before immunoprecipitation of the HA-tagged proteins for radioactivity counting. Figure 6(E) shows that, while the cellular iron uptake and expression levels of the HA-tagged proteins are similar in the two strains, the amount of 55Fe co-immunoprecipitated with CaMnn5p–HA was approx. 8 times that co-immunoprecipitated with Mnn5p–HA, suggesting that CaMnn5p may bind iron directly or through an interacting protein.

Subcellular localization of CaMnn5p in S. cerevisiae

Next we examined the subcellular localization of HA-tagged CaMnn5p in aft1Δ. The HA-tagged protein is fully functional in promoting cell growth (Figure 7A). Indirect immunofluorescence microscopy revealed that CaMnn5p–HA localized to some intracellular vesicles with a punctate staining pattern (Figure 7B). On sucrose gradients, the CaMnn5p-containing organelles demonstrated a distribution pattern similar to that of Pep12p (Figure 7C), which is involved in Golgi–vacuole transport and localizes to the Golgi apparatus and endosome [37]. Although this experiment does not have a resolution to pinpoint the exact organelles for CaMnn5p, the results are consistent with our hypothesis of CaMnn5p functioning along the endocytic pathway.

Figure 7 Subcellular localization of CaMnn5p

(A) CaMnn5p was tagged with HA at the C-terminal end and expressed from a CEN plasmid under the control of the Gal1-10 promoter. (B) Indirect immunofluorescence staining of the CaMnn5p–HA-expressing cells. a, anti-HA antibody was used as the primary antibody; b, the primary antibody was omitted. (C) The subcellular localization of CaMnn5p–HA was determined by using sucrose gradient fractionation and Western blot analysis. The fractionation patterns of Mnn2p and Pep12p were also determined for comparison.


How does CaMnn5p provide a solution to the lack of sufficient iron uptake in aft1Δ under iron-limiting conditions? We have addressed several possibilities. First, CaMnn5p activity may indirectly increase the expression or activities of high-affinity iron transporters through glycosylation; secondly, it may function as an iron transporter; and thirdly, we thought that, under the iron-limiting conditions we used, there might still be a low level of iron uptake through processes such as liquid-phase endocytosis, albeit not sufficient to support cell growth. CaMnn5p may help the cells use this source of iron more efficiently.

We ruled out the first possibility, first because we did not detect increased FTR1 or FET3 expression in CaMNN5-expressing cells. Secondly, CaMNN5 expression was able to promote the growth of several mutants deleted of iron permease or ferroxidase genes or both, including ftr1Δ, ftr1Δ fet3Δ and ftr1Δ fth1Δ. Thus we conclude that CaMnn5p functions completely independently of the high-affinity iron transporters.

To determine whether CaMnn5p functions through glycosylation of other proteins, we first confirmed that CaMnn5p expressed in P. pastoris possesses α-1,2-mannosyltransferase activity and then created transferase-dead CaMnn5p mutants. We found that the mutated forms of CaMnn5p fully retained the ability to enhance aft1Δ growth under iron-limiting conditions. Furthermore, overexpression of either MNN2 or MNN5 in aft1Δ did not promote cell growth. Thus we conclude that CaMnn5p's growth-enhancing activity in iron-limiting environments does not require the mannosyltransferase activity and is a novel function that may have evolved in C. albicans.

To determine whether CaMnn5p enhances iron uptake, we measured 59Fe uptake into aft1Δ, ftr1Δ and wild-type cells transformed with CaMNN5. We used a medium containing 200 μM BPS where the Ftr1p–Fet3p complex does not function. Under these conditions, a slow gradual increase in 59Fe cellular accumulation was detected in all the strains tested, and all the strains expressing CaMNN5 consistently exhibited approx. 3-fold more 59Fe accumulation than the corresponding control strain lacking CaMNN5, indicating that CaMnn5p plays a role in enhancing cellular iron uptake.

The observation of a slow increase of iron accumulation over time in aft1Δ and ftr1Δ under iron-limiting conditions suggests the presence of other mechanisms for iron entry into the cells. We thought that constitutive endocytosis may mediate this slow iron uptake. Supporting this view, we found that CaMNN5 expression in end4Δ and vps4Δ, which are defective in different steps of endocytosis, did not promote cell growth, indicating that an intact endocytosis pathway is required for CaMnn5p's function. The vacuole is thought to be the end point of endocytosis and the iron storage organelle in S. cerevisiae [24,25]. Two mechanisms are responsible for the vacuole–cytosol iron transport: one involves the iron permease Fth1p and the other the Nramp family bivalent metal transporter Smf3p. If our hypothesis of CaMnn5p working along the endocytosis pathway is correct, it may be predicted that CaMNN5 expression would not be able to enhance the growth of mutants where the vacuole–cytosol iron transport is completely blocked. Indeed, CaMNN5 expression did not enhance the growth of the smf3Δ fth1Δ double mutant, whereas it promoted the growth of smf3Δ and fth1Δ single mutants. Moreover, CaMnn5p does not increase cellular accumulation of LY, a commonly used marker for fluid-phase endocytosis. Together with the observed co-immunoprecipitation of 55Fe with CaMnn5p, these results suggest that CaMnn5p may have a specific role in iron uptake. Thus we propose the following model. Constitutive endocytosis continuously brings into the cell from the environment a small amount of iron leading to vacuolar iron accumulation. The vacuolar iron can be exported to cytosol by the iron transporters Fet5p–Fth1p and Smf3p on the vacuolar membrane. Although this pathway serves as a source of iron supply, it alone may not be sufficient to support cell growth, especially when the external iron concentration is far below the Km of the Fet3p–Ftr1p high-affinity iron transporter. CaMnn5p may work along this endocytic iron-supply route, possibly by increasing the efficiency of iron uptake, transport or mobilization, resulting in cell growth under iron-limiting environments. However, the exact molecular mechanism needs further investigation.

The present study takes advantage of the power of S. cerevisiae genetics to discover and characterize a C. albicans protein that performs a novel function in iron metabolism. We have also characterized the functions of CaMNN5 in C. albicans. Our results show that CaMnn5p's activity is regulated by iron, and the deletion of the gene led to the loss of virulence in mice (C. Bai, F. Y. Chan and Y. Wang, unpublished work).


We thank Jerry Kaplan for providing us with the end4Δ mutant, Alexander D. Johnson for the C. albicans genome DNA library, and Greg Odorizzi for the vps4Δ mutant. We also thank members of the Y. W. laboratory for stimulating discussions and suggestions.

Abbreviations: BPS, bathophenanthroline sulphonate; GMM, glucose minimal medium; GaMM, galactose minimal medium; HA, haemagglutinin; LY, Lucifer Yellow; ORF, open reading frame


View Abstract