MNN5 Encodes an Iron-Regulated {alpha}-1,2-Mannosyltransferase Important for Protein Glycosylation, Cell Wall Integrity, Morphogenesis, and Virulence in Candida albicans

The cell walls of microbial pathogens mediate physical interactionswith host cells and hence play a key role in infection. Mannosyltransferaseshave been shown to determine the cell wall properties and virulenceof the pathogenic fungus Candida albicans. We previously identifieda C. albicans ${alpha}$ -1,2-mannosyltransferase, Mnn5, for its novelability to enhance iron usage in Saccharomyces cerevisiae. Herewe have studied the enzymatic properties of purified Mnn5 andcharacterized its function in its natural host. Mnn5 catalyzesthe transfer of mannose to both ${alpha}$ -1,2- and ${alpha}$ -1,6-mannobiose, andthis activity requires Mn²⁺ as a cofactor and is regulated bythe Fe²⁺ concentration. An mnn5 ${Delta}$ mutant showed a lowered abilityto extend O-linked, and possibly also N-linked, mannans, hypersensitivityto cell wall-damaging agents, and a reduction of cell wall mannosylphosphatecontent, phenotypes typical of many fungal mannosyltransferasemutants. The mnn5 ${Delta}$ mutant also exhibited some unique defects,such as impaired hyphal growth on solid media and attenuatedvirulence in mice. An unanticipated phenotype was the mnn5 ${Delta}$ mutant'sresistance to killing by the iron-chelating protein lactoferrin,rendering it the first protein found that mediates lactoferrinkilling of C. albicans. In summary, MNN5 deletion impairs awide range of cellular events, most likely due to its broadsubstrate specificity. Of particular interest was the observedrole of iron in regulating the enzymatic activity, suggestingan underlying relationship between Mnn5 activity and cellulariron homeostasis.

INTRODUCTION

Top

Introduction

Candida albicans, an opportunistic human fungal pathogen, frequentlycauses life-threatening systemic infections in immunocompromisedpatients (3, 35). The worldwide appearance of drug-resistantstrains and the limited options of anti-C. albicans drugs underscorean urgent need to identify new antifungal targets for drug development.Several virulence traits have been described for this pathogen,such as the transition between yeast and hyphal forms of growth(3, 41, 57), the production of lytic enzymes (23), phenotypeswitching (42), adherence to host cells (9), and iron acquisition(37). Since the cell wall provides functions at the interfacebetween microbial pathogens and host cells, it often criticallydetermines the outcome of host-pathogen interactions. The generalstructure of the fungal cell wall is conserved, containing aninner layer of structural polysaccharides, glucans, and chitinand an outer layer enriched for mannoproteins (25). The highlyglycosylated mannoproteins are involved in host cell adhesion,antigenicity, and modulation of host immune responses (10, 46,54). Previous studies showed that mannoproteins of the outerlayer mediate direct interactions of C. albicans with host cells(11, 47) and play important roles in pathogenesis (8, 48).

Studies with Saccharomyces cerevisiae have provided much ofour current understanding of protein glycosylation in fungi.Protein glycosylation starts in the endoplasmic reticulum, wherethe first mannose is transferred to the OH group of a serineor threonine residue in O-linked glycosylation (45) or an oligosaccharidecore structure is attached to the NH₂ group of an asparagineresidue in N-linked glycosylation (26). Then the glycoproteinsmove to the Golgi apparatus, where the elongation of O-linkedmannans and synthesis of complex N-linked glycans take place(13, 30). Many of the glycosylation steps in S. cerevisiae havebeen extensively characterized at the biochemical and geneticlevels (7, 16), providing valuable knowledge for the understandingof protein glycosylation in C. albicans. However, to date, onlya few C. albicans genes responsible for protein glycosylationhave been studied in detail, including MNT1 (8) and PMT1, -2,-4, -5, and -6 (36, 39, 48, 49). These genes are members ofthe MNT and PMT families specifically involved in the O-glycosylationpathway. Gene deletion experiments demonstrated that the O-linkedglycans are important for virulence (8, 39, 48, 49). MNN9 wasfound to have a role in extending N-linked glycan outer chainsand maintaining normal cell wall composition (44). MNN4 is requiredfor mannosylphosphate transfer to the acid-labile N-mannan sidechains. An mnn4 ${Delta}$ mutant exhibited drastically reduced cell wallmannosylphosphate content, but its virulence and interactionwith macrophages were not affected (20). CaVRG4 and CaSRB1 encodeproteins required for supplying the Golgi with the mannose donorGDP-mannose and are essential in C. albicans, indicating theimportance of overall protein glycosylation to cell viability(33, 55). The Golgi GDPase CaGDA1 has also been shown to beimportant in transporting GDP-mannose into the Golgi. A gda1null mutant is viable but has defects in cell wall biogenesis,hyphal formation, and O-mannosylation (19). Some metal ionsare known to regulate mannosyltransferase activity. Tkacz etal. (50) showed that manganese ions are an essential cofactorof Golgi-bound mannosyltransferases. CaPmr1 was recently foundto pump Ca²⁺/Mn²⁺ ions into the Golgi. A Capmr1 null mutantshowed defects in both O- and N-glycosylation, growth dependenceon supplemented calcium after entering stationary phase, andattenuated virulence (2).

We previously cloned a C. albicans gene encoding a Golgi ${alpha}$ -1,2-mannosyltransferasewith highest sequence homology to S. cerevisiae Mnn5 and Mnn2and thus named it CaMNN5 (1). In budding yeast, Mnn2 and Mnn5are responsible for the branching and extension of the sidechains of N-linked mannose oligomers (38). Deletion of MNN2or MNN5 in S. cerevisiae results in shortened ${alpha}$ -1,2-linked mannosebranches, blocking the subsequent addition of mannosylphosphate(12). CaMNN5 was initially identified for its ability to supportthe growth of S. cerevisiae mutants severely defective in ironacquisition under iron-limiting conditions (1). Heterologousexpression in S. cerevisiae revealed that CaMnn5 could enhanceiron uptake and coprecipitate with ⁵⁵Fe, an activity not foundfor any other fungal mannosyltransferases. These intriguingfeatures of Mnn5 prompted us to further characterize its functionand role in its natural host, C. albicans.

MATERIALS AND METHODS

Top

Materials and Methods

Strains and growth conditions. The C. albicans and S. cerevisiae strains used for this studyare listed in Table 1. The strains were routinely grown in YPDmedium (1% yeast extract, 2% peptone, and 2% glucose), GMM medium(yeast nitrogen base without amino acids and with 2% glucose),or GMM medium supplemented with required amino acids. For yeastgrowth in liquid media, all strains were grown at 30°C.Iron-depleted media were prepared by adding 100 µM bathophenanthrolinesulfate (BPS). Solid, hypha-inducing media were prepared asdescribed previously (57).

View this table:
[in this window]
[in a new window]
TABLE 1. Yeast strains used for this study

CaMNN5 complementation of S. cerevisiae mnn2 ${Delta}$ and mnn5 ${Delta}$ mutants. The coding sequence of CaMNN5 was cloned into an S. cerevisiae2µm plasmid (pRS426) (37) under the control of the ADH1promoter and transformed into S. cerevisiae mnn2 ${Delta}$ and mnn5 ${Delta}$ mutants.

Alcian blue binding assay. The Alcian blue binding assay was carried out using the methodof Herrero et al. (19). A series of solutions containing differentamounts of Alcian blue (Sigma) was prepared in 0.02 N HCl, andthe optical density at 600 nm (OD₆₀₀) of each solution was determined.A standard curve was plotted of the OD values versus amountsof Alcian blue. To quantify Alcian blue binding to the cellsurface, 500 µl exponential-phase yeast cells (OD₆₀₀,1) was centrifuged, and the cells were washed once with 2 mlof 0.02 N HCl and resuspended in 1 ml 0.02 N HCl containing50 µg Alcian blue. The cell suspension was allowed tostand for 10 min at room temperature and then centrifuged for2 min to pellet the cells. The OD₆₀₀ of the supernatant wasmeasured. The amount of Alcian blue was determined by usingthe standard curve. The amount of dye bound to the cells wascalculated by subtracting the amount of dye in the supernatantfrom 50 µg.

Expression and purification of GST-Mnn5 fusion protein. The region of C. albicans MNN5 encoding amino acid residues211 to 597 was cloned into pGEX-4T-1 (Amersham) in frame withthe glutathione S-transferase (GST) gene and transformed intoEscherichia coli strain BL21 (Amersham). Purification of theGST-Mnn5 fusion protein was performed by following standardprotocols (15).

Preparation of polyclonal antibodies. Two rabbits (New Zealand White) were purchased from the SembawangLaboratory Animals Centre (Singapore). Each rabbit was injectedwith 500 µg purified Mnn5 emulsified in complete Freund'sadjuvant (Sigma). Booster injections using the same amount ofantigen emulsified in incomplete Freund's adjuvant (Sigma) wereadministered every 2 weeks. The rabbits were bled 10 days afterthe fifth injection and after each subsequent boost. Affinitypurification of Mnn5-specific antibodies was performed usingthe purified GST-Mnn5 fusion protein coupled to glutathione-Sepharose4B beads. The serum was incubated with 1 ml of the beads at4°C for 3 h. The beads were then washed extensively withcold phosphate-buffered saline (PBS) (4°C), and the boundantibodies were eluted with ImmunoPure immunoglobulin G elutionbuffer (Pierce).

Western blot analysis. Protein concentrations were determined by the Bradford assay(Bio-Rad). Samples for Western blot analysis were resolved by10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis(SDS-PAGE) and transferred to a Hybond-C nylon membrane (Amersham).The purified anti-Mnn5 antibodies described above were usedas the primary antibody. An ECL Western blot detection kit (Pierce)was used to visualize the target protein.

Mnn5 expression in Pichia pastoris and FPLC purification of Mnn5. The expression of C. albicans Mnn5 using P. pastoris was doneas previously described (1). To purify the secreted Mnn5 protein,the culture was centrifuged to pellet the cells, and the supernatantwas concentrated by passage through a Centriplus spin filter(molecular size cutoff, 10 kDa) and washed three times withand then resuspended in a small volume ( $~$ 1/10 the original volume)of 20 mM Tris-HCl (pH 8) before being subjected to fast-performanceliquid chromatography (FPLC) separation using a Mono-Q column(10 x 100 mm; Amersham). A KCl gradient, which was increasedfrom 0 to 0.8 M from 0 to 50 min, was used to elute the protein.Fractions (4 ml/min) were collected for SDS-PAGE and Westernblot analysis to locate Mnn5. The elution buffer of the fractioncontaining Mnn5 was replaced with 20 mM phosphate buffer (pH6.0) by using a Centriplus filter as described above.

Assay of Mnn5 enzyme activity. The enzyme assay was carried out by using 10 µl (8 µg)of purified Mnn5 as the enzyme source in a buffer containing50 mM Tris-Cl (pH 7.2), 10 mM MnCl₂, 0.23 µM GDP-[¹⁴C]mannose(specific activity, 289 mCi/mmol), 20 mM 2- or 6-O- ${alpha}$ -D-mannopyranosyl-D-mannopyranoseacceptor (Sigma), and 5 mg/ml bovine serum albumin. Standardreactions were performed in 50 µl at 30°C for 60 min.To terminate the reaction, the mixtures were passed through0.8 ml QAE-Sephadex anion-exchange resin to remove the unlabeledGDP-mannose. The labeled product was eluted with 0.8 ml water,and radioactivity was counted in a scintillation beta counter(Amersham).

Gene deletion. The MNN5 gene was deleted from CAI4 by using the URA blastermethod (14), which is schematically illustrated in Fig. 3A.A pair of oligonucleotides (5'-CAATTGCCCAGTTCTACA-3' and 5'-AACATGCGTCTTCTTGAA-3')was used to PCR amplify the region upstream of the coding regionof MNN5, and other primers (5'-CGGCGAATGATTATGAGA-3' and 5'-CGCCCAAAGAGGTTTATG-3')were used for the region downstream of MNN5. S. cerevisiae mnn2 ${Delta}$ and mnn5 ${Delta}$ mutants were derived from strain CRY2 ${alpha}$ . A gene deletioncassette was constructed by flanking LEU2 or HIS3 with two DNAfragments corresponding to the 5'- and 3'-untranslated regionsof the target gene. The following pairs of primers were usedto PCR amplify the 5'- and 3'-flanking fragments of MNN2 andMNN5. For the MNN2 5' fragment, 5'-GAAGCATTCGTTATCAGG-3' and5'-CCTACCTACACTGATTTG-3' were used; for the MNN2 3' fragment,5'-CAACGGGTCGGACTATAG-3' and 5'-GCATTCGAATTCCTTAAC-3' were used;for the MNN5 5' fragment, 5'-CTATCTCCCACATGTTGGC-3' and 5'-CCTGAACTATTAGTATATC-3'were used; and for the MNN5 3' fragment, 5'-GAGAAAAACTAATGAAGTG-3'and 5'-GAATTGCTGTGCAGAAAC-3' were used. Appropriate restrictionsites were added to the PCR primers when needed. Transformantswere selected on histidine- or leucine-dropout medium accordingto the selection marker used. Correct deletion of each genewas verified by Southern blotting (data not shown).

View larger version (47K):
[in this window]
[in a new window]
FIG. 3. Chromosomal deletion of MNN5. (A) Schematic description of the strategy used to knock out (KO) the first and second copies of MNN5. The first base of the start codon ATG is designated nucleotide 1. The restriction sites for HindIII and SpeI are shown for the Southern blotting analysis below. (B) Strategy for reintegrating a copy of wild-type MNN5 into the KO mutants. A plasmid was constructed containing the MNN5 coding region with $~$ 1,500 bp of the 5'- and 600 bp of the 3'-untranslated region and the hisG sequence. A StuI restriction site was introduced into hisG by site-directed mutagenesis. This plasmid was linearized by StuI and transformed into the mnn5 ${Delta}$ mutant for site-specific integration at the genomic locus, where hisG replaced MNN5. (C) Southern blotting confirmation of MNN5 gene deletion and reintegration. Genomic DNA samples were digested with HindIII and SpeI. The region corresponding to nucleotides –236 to –644 of the MNN5 promoter region was PCR amplified and labeled with ³²P for use as a probe. Lane 1, CAI4; lane 2, CaWYBC1; lane 3, CaWYBC1.1; lane 4, CaWYBC2; lane 5, CaWYBC2.1; lane 6, CaWYBC1.2; and lane 7, CaWYBC2.2.

O-glycan analysis. Total cellular O-glycans were analyzed as described previously(18), with a minor modification. To radioactively label mannans,the yeast cells growing in 2 ml of YP plus 0.5% sucrose wereincubated with 1.85 MBq of [2-³H]D-mannose (777 GBq mmol^–1;Perkin-Elmer) at 30°C for 2 h. Cells were then pelleted,resuspended in 300 µl PBS, and lysed by bead beating aspreviously described (1). The cell wall was then pelleted bycentrifugation and lyophilized. Equal amounts of the lyophilizedcell walls from different strains were treated with 0.1 M NaOHfor 24 h at room temperature for ß-elimination ofO-glycans. The reaction was terminated by adding HCl to a finalconcentration of 0.1 M, and the mixture was then centrifugedat 80,000 rpm using a TLA-100.2 rotor (Beckman) for 2 h at 4°C.³H-labeled glycans in the supernatant were subjected to thin-layerchromatography (TLC) on Silica Gel 60G plates (Merck) in ethylacetate-butanol-acetic acid-water (3:4:2.5:4) with two ascents.The chromatograms were then treated with EN³HANCE reagent (Perkin-Elmer)and exposed to Kodak X-OMAT X-ray film at –70°C. Toquantify each labeled species, a 1-cm by 0.5-cm area of thesilica plate containing each species was cut out and subjectedto scintillation counting of ³H using a scintillation beta counter(Amersham).

RESULTS

Top

Results

Overexpression of MNN5 in C. albicans. Since MNN5 expression can rescue the iron acquisition defectsof an S. cerevisiae ftr1 ${Delta}$ mutant (1), we first wanted to determinewhether MNN5 overexpression will do the same for a C. albicansftr1 ${Delta}$ mutant. We used several relatively strong promoters, suchas the GAL1-10, ADH1, and MET3 promoters, to drive MNN5 expressionfrom either a single copy integrated at the RP10 locus or amultiple-copy plasmid (pABSK1), but none of these constructswas able to alleviate the growth deficiency of the C. albicansftr1 ${Delta}$ mutant (data not shown) under iron-limiting conditions.We then examined whether MNN5 expression in C. albicans cellsis regulated by iron and found that neither the mRNA nor theprotein level of MNN5 was affected by iron concentrations (datanot shown).

CaMNN5 rescues the defect of an S. cerevisiae mnn5 ${Delta}$ mutant but not an mnn2 ${Delta}$ mutant. Since CaMnn5 shares high sequence homology with two S. cerevisiaemannosyltransferases, Mnn5 and Mnn2, we wanted to know whetherCaMNN5 can functionally complement an S. cerevisiae mnn2 ${Delta}$ ormnn5 ${Delta}$ mutant. Negatively charged mannophosphate units are knownto be incorporated into N-linked oligosaccharides of mannoproteins,and thus defects in N-glycosylation are often associated witha decrease in the cell's affinity for the cationic dye Alcianblue (31, 34). Both mnn2 ${Delta}$ and mnn5 ${Delta}$ mutant strains showed a significantloss of Alcian blue binding (Fig. 1) (39). When a copy of CaMNN5driven by the ADH1 promoter in a 2µm plasmid was transformedinto the two mutants, Alcian blue binding was significantlyimproved in mnn5 ${Delta}$ but not in mnn2 ${Delta}$ mutant cells, suggesting thatCaMNN5 may be the counterpart of S. cerevisiae MNN5 and mayprovide similar functions in C. albicans.

View larger version (18K):
[in this window]
[in a new window]
FIG. 1. CaMNN5 restores Alcian blue binding in an S. cerevisiae mnn5 ${Delta}$ mutant. The assay was done as described in Materials and Methods. The amount of Alcian blue bound to the cells of a strain is shown as a percentage of that bound to the wild-type cells (CRY2 ${alpha}$ ). The strains used were CRY2 ${alpha}$ , ScWYBC8 (mnn2 ${Delta}$ ), ScWYBC8.1 (mnn2 ${Delta}$ CaMNN5), ScWYBC9 (mnn5 ${Delta}$ ), and ScWYBC9.1 (mnn2 ${Delta}$ CaMNN5). All assays described throughout the paper were done three times, and standard deviations are shown.

Mnn5 has ${alpha}$ -1,2-mannosyltransferase activity. To analyze the enzymatic activity of C. albicans Mnn5, we useda P. pastoris expression system to produce soluble, secretedprotein. Figure 2A (lane 2) shows that the culture supernatantof the P. pastoris strain expressing MNN5 contained a major64-kDa protein that matches the predicted size of Mnn5 but wasnot present in the culture supernatant of the strain transformedwith the empty vector (Fig. 2A, lane 3). To purify the protein,the supernatant was subjected to FPLC using a Mono-Q anion-exchangecolumn and eluted with a KCl gradient. SDS-PAGE and Westernblot analysis were performed to locate the fractions that containedMnn5. The purified protein (Fig. 2A, lane 4) was subjected toa mannosyltransferase activity assay. Figure 2B (top) showsthat the purified Mnn5 protein catalyzed the transfer of [¹⁴C]GDP-mannoseto both ${alpha}$ -D-1,6- and ${alpha}$ -D-1,2-mannobiose, with similar activities.We also found that Mnn5 exhibited high activities at pH 6.0and 9.0 under the assay conditions used (Fig. 2B, bottom).

View larger version (53K):
[in this window]
[in a new window]
FIG. 2. Expression, purification, and enzyme activity of CaMnn5. (A) SDS-PAGE (a) and Western blot analysis (b) of Mnn5 expressed in P. pastoris. Mnn5 expression in P. pastoris and FPLC purification of Mnn5 were carried out as described in Materials and Methods. Lane 1, molecular weight markers; lane 2, 20 µl of culture supernatant of the P. pastoris strain expressing Mnn5; lane 3, 20 µl of supernatant of the P. pastoris strain (KMvec) transformed with an empty vector; lane 4, FPLC-purified Mnn5. The Western blot was probed by using an anti-Mnn5 antibody as the primary antibody. (B) Mannosyltransferase assay (a) of Mnn5. The assay was performed at pH 6.0 as described in Materials and Methods. [¹⁴C]GDP-mannose was used as the mannose donor, and ${alpha}$ -1,2- and ${alpha}$ -1,6-mannobiose were used as acceptors. Purified Mnn5 (8 µg) was used in each reaction. The same amount of bovine serum albumin was used as a negative control. The radioactive products were isolated for quantification in a scintillation counter, expressed as counts per minute (cpm)/mg of protein/h. The pH dependence of Mnn5 activity (b) was evaluated using the same protocol, except with reaction buffers at different pHs. The same pattern was obtained using either ${alpha}$ -1,2- or ${alpha}$ -1,6-mannobiose as the acceptor. (C) Test of metal ion requirement for Mnn5's enzymatic activity. The same assay protocol was used as that described for panel B, except that the reaction buffer (pH 6.0) contained a 0.1, 1, or 10 mM concentration of one of the divalent metal ions, as indicated. (D) Ferrous iron regulates Mnn5 activity in the presence of Mn²⁺. The assay buffer (pH 6.0) contained 10 mM Mn²⁺ and was supplemented with different amounts of BPS and ferrous ammonium sulfate, as indicated at the bottom of the graph.

Mnn5 requires Mn²⁺ and Fe²⁺ for activity. Several Golgi mannosyltransferases are known to require Mn²⁺for activity (24, 38, 56). To test the metal ion requirementsof Mnn5, the transferase activity was assayed in buffers supplementedwith various divalent metal ions and chelators. We used purifiedMnn5 as the enzyme source and ${alpha}$ -1,2-mannobiose as the acceptorfor [¹⁴C]GDP-mannose. First, a metal ion such as Mn²⁺, Fe²⁺,Mg²⁺, Zn²⁺, or Cu²⁺ was included in the reaction buffer at arange of concentrations from 0.1 to 10 mM. Figure 2C shows thatwhen added alone, the only divalent metal ion supporting highenzyme activity was Mn²⁺, while all the other ions tested hadlittle effect, consistent with a role of Mn²⁺ as an essentialcofactor. Since iron was previously found to bind to Mnn5 (1),we tested whether it was required for the enzymatic activityof Mnn5 in the presence of Mn²⁺. Adding different amounts ofthe iron chelator BPS to the reaction buffer was found to inhibitthe enzyme activity in a concentration-dependent manner (Fig.2D). In the presence of 0.1, 0.2, and 0.4 mM BPS, the enzymaticactivity was lowered to 50, 30, and 18% that of the nonchelatedcontrol, respectively, suggesting that iron is required forthe mannosyltransferase activity of Mnn5. This observation wasfurther substantiated by the restoration of the enzymatic activitywhen Fe²⁺ was added to the buffer containing 0.4 mM BPS. AddingFe²⁺ to the reaction buffer without BPS slightly enhanced theenzyme activity when the concentration of the added iron waswithin 2 to 10 µM, suggesting that the reaction mix initiallycontained sufficient iron. The ferric iron chelator desferri-ferrichromehad no effect on Mnn5 activity (data not shown), suggestingthat ferrous but not ferric iron is required. Adding other metalions, such as Cu²⁺, Mn²⁺, and Mg²⁺, did not relieve the inhibitionby BPS. Depleting Cu²⁺ by adding the chelator 2,9-dimethyl-4,7-diphenyl-1,10-phenanthrolinedisulphonicacid (BCS) also had no effect on the enzyme activity (data notshown). Taken together, the results indicate that Mn²⁺ is anessential cofactor for the mannosyltransferase activity of Mnn5and that the activity is regulated by ferrous iron, consistentwith its iron-binding ability detected in our previous study(1).

Construction of MNN5 deletion mutant. To investigate MNN5's cellular functions in C. albicans, weused two different gene deletion cassettes, hisG-URA3-hisG andcat-URA3-cat, to sequentially delete the first and second copiesof MNN5 from strain CAI4, as described previously (14, 22) (Fig.3A). Figure 3B shows the strategy for reintroducing a wild-typeMNN5 copy into the mutants by integration at the hisG locus.The genotypes of the heterozygous and homozygous mutants andreintegrants were verified by Southern blot analysis (Fig. 3C).

We determined the growth rate of the mnn5 ${Delta}$ mutant over a widerange of iron concentrations but did not observe a significantdifference in comparison with the wild-type strain (data notshown). Also, the mnn5 ${Delta}$ mutant did not exhibit reduced ⁵⁹Fe uptakeunder all conditions used (data not shown). The results indicatethat unlike its function in the heterologous host S. cerevisiae,Mnn5 may not be involved in iron uptake in its natural host(see Discussion).

Study of Mnn5's role in N-linked and O-linked glycosylation. The S. cerevisiae mnn5 ${Delta}$ mutant is impaired in N-linked glycosylation.To test whether Mnn5 is involved in N-glycosylation, we performedan Alcian blue binding assay on mnn5 ${Delta}$ mutant cells. The assayshowed drastically decreased binding of the dye to mnn5 ${Delta}$ cellscompared to that for the wild-type cells (Fig. 4A). Approximately90% of the dye coprecipitated with the wild-type cells, in contrastto only 20% with mnn5 ${Delta}$ mutant cells. Reintroducing a wild-typecopy of MNN5 into mnn5 ${Delta}$ cells restored the dye binding activity.We previously showed that mutation of either one of the asparticacid (D) residues to alanine (A) in the conserved DXD^282-284motif of Mnn5 totally abolished the mannosyltransferase activity(1). Reintroducing a mutated mnn5 gene (D282A or D284A) intothe mnn5 ${Delta}$ mutant did not restore cell binding of the dye (Fig.4A), indicating that Alcian blue binding ability is dependenton the enzyme activity of Mnn5. Taken together, the resultsshow that the amount of cell surface mannosylphosphate is significantlyreduced in the mnn5 ${Delta}$ mutant, which is most likely the consequenceof blocked synthesis of N-linked mannan branches.

View larger version (39K):
[in this window]
[in a new window]
FIG. 4. Investigation of Mnn5's role in N-linked and O-linked glycosylation. (A) The mnn5 ${Delta}$ mutant exhibited greatly reduced Alcian blue binding. The following strains were included in the assay, as described in Materials and Methods: SC5314, CaWYBC2 (mnn5 ${Delta}$ ), CaWYBC2.2 (mnn5 ${Delta}$ MNN5), CaWYBC2.3 (mnn5 ${Delta}$ mnn5D282A), and CaWYBC2.4 (mnn5 ${Delta}$ mnn5D284A). (B) ß-Elimination of O-glycans. The mannans of SC5314 (WT) and CaWYBC2 (mnn5 ${Delta}$ ) were labeled with ³H by adding [2-³H]mannose to the growth medium. The base-sensitive O-glycans were released by ß-elimination, separated by thin-layer chromatography (ethyl acetate-butanol-acetic acid-water [3:4:2.5:4]), and visualized by subjecting the TLC plate to autoradiography. The different species of mannose oligomers, mannose (M1) to pentamannose (M5), are shown. The N-glycans were base resistant and thus were retained at the origin of sample loading. (C) An area (1 cm x 0.5 cm) of the TLC plate containing each mannose species was cut out for scintillation counting. The relative amount of each mannose species (³H count) is shown as a percentage of the total count for M1 to M5.

To determine whether the mnn5 ${Delta}$ mutant also has defects in O-glycosylation,we grew cells in the presence of [2-³H]mannose to label themannose units of glycosylated proteins. The O-linked mannanswere released by ß-elimination and resolved by TLC.The ß-elimination condition used was previously shownto completely release O-glycans from mannoproteins (18). Anautoradiograph of the TLC plate (Fig. 4B) displays five major³H-labeled molecules containing from one to five mannose residues(M1 to M5), as previously shown by Herrero et al. (19). Whenthe amounts of each ³H-labeled species were compared betweenthe wild type and the mnn5 ${Delta}$ mutant, we found that the mnn5 ${Delta}$ mutantshowed a markedly higher level of M1 and less M2, M3, and M5than the wild type, suggesting an impaired ability of the mutantto extend O-linked mannan chains. The area on the TLC platecorresponding to each radioactive spot was cut out and subjectedto quantification by scintillation counting of ³H (Fig. 4C).M1 was found to represent 16.6% of the total ³H count for allfive species (M1 to M5) in the wild-type strain, whereas thepercentage increased to 46.6% in the mutant, showing that alarge fraction of the first O-linked mannose residue cannotbe further extended in the absence of Mnn5. The radioactivesignal at the origin of sample loading should largely representN-glycans, which are resistant to ß-elimination andremain attached to the proteins. Quantification of the ³H-labeledglycans at the origin showed a 50% reduction in the mnn5 ${Delta}$ mutantin comparison with the wild-type strain. These results indicatethat Mnn5 is involved in extending O-linked, and likely alsoN-linked, mannans.

The mnn5 ${Delta}$ mutant exhibits markedly reduced sensitivity to LF. Lactoferrin (LF) is a mammalian iron-binding protein abundantlypresent in saliva, milk, and other exocrine secretions. It constitutesan important part of innate immunity, partly because its strongiron-chelating activity effectively removes free iron from theenvironment that is needed for the growth of nearly all microorganisms(40). In particular, LF and some LF-derived peptides are knownto inhibit the growth of C. albicans under certain culturalconditions (27, 53). We routinely test C. albicans mutants defectivein iron metabolism for their sensitivity to LF. An establishedprotocol (43) involves the treatment of C. albicans cells with100 µg LF in 1 ml of 0.05 mM KCl for 75 min at 37°Cbefore scoring the CFU on iron-sufficient plates. A mock treatmentof each strain without adding LF was done in parallel as a control.Figure 5 shows that if the cells were pregrown in GMM medium, $~$ 5% of the wild-type cells survived the LF treatment, whereasif they were pregrown in iron-depleting medium, $~$ 20% of the wild-typecells survived. This observation suggests that the level ofcellular iron is a determinant of a cell's sensitivity to LF.Strikingly, the mnn5 ${Delta}$ mutant exhibited markedly reduced LF sensitivity.Approximately 33% of the mutant cells pregrown in GMM and 50%of those pregrown in the iron-limiting medium survived the LFtreatment. Reintroducing a mutated mnn5 gene (D282A or D284A)into the mnn5 ${Delta}$ mutant did not restore sensitivity to LF (Fig.5), indicating that the mannosyltransferase activity is requiredfor Mnn5-mediated LF killing of C. albicans. For comparison,reintroducing a wild-type copy of MNN5 into the mnn5 ${Delta}$ mutantrestored LF sensitivity. The results reveal that Mnn5 has arole in determining a cell's sensitivity to LF which dependson cellular iron homeostasis. Is the increased resistance toLF killing a unique phenotype of the mnn5 ${Delta}$ mutant or one sharedby other mannosyltransferase mutants? To answer this question,we tested the LF sensitivity of a series of mannosyltransferasemutants, including pmt1 ${Delta}$ , pmt2 ${Delta}$ /PMT2, pmt4 ${Delta}$ , pmt5 ${Delta}$ , pmt6 ${Delta}$ , mnt1 ${Delta}$ ,and mnt2 ${Delta}$ mutants (8, 32, 36). We found that only the mnn5 ${Delta}$ mutantexhibited increased resistance to LF killing (Table 2), indicatinga unique role of Mnn5 in mediating LF killing of C. albicans.

View larger version (43K):
[in this window]
[in a new window]
FIG. 5. LF sensitivity assay. The following strains were used: SC5314, CaWYBC2 (mnn5 ${Delta}$ ), CaWYBC2.2 (mnn5 ${Delta}$ MNN5), CaWYBC2.3 (mnn5 ${Delta}$ mnn5D282A), and CaWYBC2.4 (mnn5 ${Delta}$ mnn5D284A). Each strain tested was grown in iron-sufficient (GMM) or iron-limiting medium (GMM plus 100 µM BPS) to mid-log phase before being treated with 100 µg LF in 1 ml of 0.05 mM KCl (pH 7.0). The cells were incubated in a shaking water bath at 37°C for 75 min. Cells in KCl buffer without LF were included as a control. The number of viable cells was enumerated by plating the LF-treated cells on YPD plates and counting the CFU after a 2-day incubation at 30°C. LF sensitivity is presented as CFU of LF-treated cells, expressed as a percentage of the CFU of the untreated control.

View this table:
[in this window]
[in a new window]
TABLE 2. Sensitivity of mannosyltransferase mutants to LF, Congo red, and hygromycin B

The mnn5 ${Delta}$ mutant shows a weakened cell wall. For both S. cerevisiae and C. albicans, the deletion of genesencoding mannosyltransferases has been reported to cause defectsin cell wall integrity. To determine whether the mnn5 ${Delta}$ mutanthas a similar defect, we tested the sensitivity of the mnn5 ${Delta}$ mutant to lyticase, an enzyme normally used to remove the fungalcell wall. Exponential-phase cells grown in YPD were dilutedto an OD₆₀₀ of 0.6 in a hypo-osmotic solution containing 10mM Tris and 0.2 mg/ml lyticase. The extent of cell lysis wasmonitored by measuring the OD₆₀₀ of the cell suspension at timedintervals. Figure 6A shows that the mnn5 ${Delta}$ mutant is much moresensitive to damage by lyticase than the wild-type strain. By1 h, the OD₆₀₀ of the mnn5 ${Delta}$ cell suspension had dropped from0.58 to 0.2. Microscopic examination of the cell suspensionrevealed that >90% of the mutant cells had been lysed (datanot shown). In contrast, the OD₆₀₀ of the wild-type cells was0.52 by 1 h and dropped to 0.2 by 5 h. The heterozygous MNN5/mnn5 ${Delta}$ mutant strain displayed an intermediate sensitivity to lyticase,indicating a gene dosage effect.

View larger version (36K):
[in this window]
[in a new window]
FIG. 6. The mnn5 ${Delta}$ mutant is hypersensitive to lyticase, Congo red, and hygromycin B. (A) Lyticase sensitivity assay. The following strains were included in the assay: SC5314, CaWYBC1 (MNN5/mnn5 ${Delta}$ ), CaWYBC1.2 (MNN5/mnn5 ${Delta}$ MNN5), CaWYBC2 (mnn5 ${Delta}$ ), and CaWYBC2.2 (mnn5 ${Delta}$ MNN5). Exponential-phase yeast cells of the strains tested were diluted to an OD₆₀₀ of 0.6 in 10 mM Tris-HCl (pH 7.5) containing 0.2 mg/ml lyticase. The cell suspension was incubated at 30°C, and the OD₆₀₀ was taken at timed intervals. Congo red (B) and hygromycin B (C) sensitivity assays were also performed. Exponential-phase yeast cells of the strains shown were serially diluted and dropped onto agar plates containing either 200 µg/ml Congo red or 200 µg/ml hygromycin B. The plates were incubated at 30°C for 3 days.

We also tested the sensitivity of the mutant to the cell wallconstruction inhibitor Congo red and found that the mnn5 ${Delta}$ mutantexhibited very high sensitivity to Congo red. The heterozygousMNN5/mnn5 ${Delta}$ mutant already showed significantly increased Congored sensitivity compared with the wild-type strain (Fig. 6B).This sensitivity was further increased in the homozygous mutant.Reintroducing a wild-type copy of MNN5 into the mutants partiallyrestored the resistance to Congo red. Again, a gene dosage effectwas observed, and it appears that the severity of Congo reddamage is more sensitive to the cellular level of Mnn5. Previousreports described that some pmt mutants are supersensitive tothe antifungal drug hygromycin B (36, 49). We found that themnn5 ${Delta}$ mutant also showed a significantly higher sensitivity tohygromycin B than the wild type, in a gene dosage-dependentmanner (Fig. 6C).

We also compared the sensitivities of the mnn5 ${Delta}$ mutant to Congored and hygromycin B with those of other mannosyltransferasemutants and found that several mutants, such as mnn5 ${Delta}$ , pmt1 ${Delta}$ ,and pmt2 ${Delta}$ /PMT2 mutants, exhibited increased sensitivity to bothcompounds (Table 2), suggesting that these phenotypes may bea general consequence of compromised cell wall integrity. Theresults of Congo red and hygromycin B sensitivity tests withthe PMT and MNT mutants are consistent with previous reports(8, 32, 36).

The mnn5 ${Delta}$ mutant is defective in hyphal morphogenesis. C. albicans mutants defective in O-glycosylation, such as pmt1 ${Delta}$ and pmt6 ${Delta}$ mutants, were shown to be partially impaired in hyphalmorphogenesis (48, 49). We examined the hyphal development ofthe mnn5 ${Delta}$ mutant in several common hypha-inducing media. Themnn5 ${Delta}$ mutant strain showed normal hyphal development in all theliquid inducing media used but no or poor hyphal growth on somesolid inducing media, such as RPMI, serum, Spider, and embeddingmedia (Fig. 7).

View larger version (66K):
[in this window]
[in a new window]
FIG. 7. The mnn5 ${Delta}$ mutant is defective in hyphal growth on some solid media. Yeast cells of strains SC5314 and CaWYBC2 (mnn5 ${Delta}$ ) were inoculated onto several hypha-inducing solid media, as indicated, and incubated at 37°C for 3 days.

MNN5 is required for C. albicans virulence. Next, we used the mouse systemic infection model to evaluatethe importance of MNN5 in C. albicans virulence. Each mousewas injected through the tail vein with 1 x 10⁶ yeast cellsof either a wild-type strain (SC5314), the mnn5 ${Delta}$ mutant, or thereintegrant strain CaWYBC2.2. Previous tests had shown thatall three strains grew at nearly identical rates in YPD andGMM media (data not shown). Injection with SC5314 killed allanimals within 1 week, whereas 80% of the mice injected withthe mnn5 ${Delta}$ mutant survived the full 35-day course of observation.The mice injected with the reintegrant strain showed a meansurvival time of 10 days (Fig. 8), again indicating a gene dosageeffect. The results show that MNN5 is required for the virulenceof C. albicans in this infection model.

View larger version (21K):
[in this window]
[in a new window]
FIG. 8. The mnn5 ${Delta}$ mutant showed markedly reduced virulence. Strains SC5314, CaWYBC2 (mnn5 ${Delta}$ ), and CaWYBC2.2 (mnn5 ${Delta}$ MNN5) were grown in GMM at 30°C to exponential phase. A 200-µl volume of 1 x 10⁶ cells of each strain in PBS was injected into each animal through the tail vein. Eight mice were used for each strain, and their survival was monitored daily for 35 days. The experiment was done twice, with similar results.

DISCUSSION

Top

Discussion

CaMnn5 shares nearly equal sequence similarities with S. cerevisiaeMnn5 and Mnn2, but our complementation test showed that CaMNN5restored Alcian blue binding to the cell surface of mnn5 ${Delta}$ butnot mnn2 ${Delta}$ mutant cells, suggesting that CaMNN5 may be the functionalcounterpart of MNN5. Camnn5 ${Delta}$ and mnn5 ${Delta}$ mutants share a range ofsimilar defects, such as a reduction of cell surface bindingof Alcian blue and hypersensitivity to cell wall-damaging agents.The C. albicans mnn5 ${Delta}$ mutant exhibited defects in O-linked glycosylation,as ß-elimination of O-linked glycans yielded a dramaticincrease in the mannose monomer (M1) and a decrease in mannoseoligomers (M2, M3, and M5) in comparison with the wild-typestrain. The mnn5 ${Delta}$ mutant may also be defective in N-linked glycosylation,as evidenced by the decreased Alcian blue binding and the amountof ß-elimination-resistant glycans (Fig. 4). In additionto the defects commonly associated with mutants of mannosyltransferasegenes in yeast, the mnn5 ${Delta}$ mutant also exhibited some phenotypesunique to C. albicans, such as a markedly reduced sensitivityto LF killing, a lack of hyphal growth on solid media, and attenuatedvirulence in mice. Of particular importance, Mnn5 activity wasfound to be regulated by iron, and manifestation of some ofthe phenotypes of the mnn5 ${Delta}$ mutant appears to be affected bythe cellular iron status.

Since MNN5 was first cloned for its unanticipated ability tosupport S. cerevisiae growth in the presence of high concentrationsof iron chelators, we were interested in knowing whether itmay perform similar functions in C. albicans. Surprisingly,overexpression of MNN5 in C. albicans did not enhance cell growth,and the mnn5 ${Delta}$ mutant did not exhibit any growth defects underiron-limiting conditions. This discrepancy might be due to thedifferent functions manifested by Mnn5 in the two differentbiological systems. The growth-enhancing activity of C. albicansMnn5 found in S. cerevisiae under iron-limiting conditions appearedto be largely, if not entirely, attributable to the novel iron-bindingability of the protein, because the growth enhancement did notrequire mannosyltransferase activity but was abolished whenthe putative iron-binding sites were mutated (1). On the otherhand, the effect of CaMnn5's mannosyltransferase activity onbudding yeast may be quite limited, possibly because of theredundant functions provided by host mannosyltransferases suchas Mnn5. Consistently, overexpression of either MNN2 or MNN5in S. cerevisiae also had no discernible effect (1). Nevertheless,the iron-dependent growth-enhancing activity of CaMnn5 foundin S. cerevisiae revealed a novel property of the protein, namely,iron binding. In this study, we established that the mannosyltransferaseactivity of C. albicans Mnn5 is regulated by iron, being activatedat low and inhibited at high iron concentrations (Fig. 2D).Also, the observation that high concentrations of BPS (100 to400 µM) were needed to significantly reduce the enzymaticactivity of purified Mnn5 suggests a high affinity of the proteinfor iron, consistent with the previously observed coprecipitationof ⁵⁵Fe with Mnn5 (1). It is an attractive possibility thatMnn5 may regulate certain cell functions by altering proteinglycosylation in response to the cellular or environmental ironstatus. In further support of this hypothesis, we observed aparallel between the mnn5 ${Delta}$ mutant's resistance to LF killingand the significantly enhanced LF resistance of wild-type C.albicans cells grown under iron-limiting conditions. Iron isan essential nutrient for almost all organisms. Keeping freeiron away from microbial pathogens is an important element ofhost innate immunity. Thus, tactics for the efficient acquisitionand use of iron constitute virulence factors for many microbialpathogens, including C. albicans (37). At the moment, we donot know whether the iron regulation of Mnn5 is of specificimportance for C. albicans survival in the host. Although themnn5 ${Delta}$ mutant is much less virulent than the wild-type strain,this could be the result of many compromised cellular functions.The isolation of MNN5 mutants that have lost the response toiron may help to answer this question in future. Nevertheless,for the purpose of drug development, Mnn5 may serve as a goodtarget.

The ability of C. albicans to switch from yeast to filamentousgrowth is important for pathogenicity (28, 57). The outer layerof the fungal cell wall is rich in mannoproteins, and mutationsin genes responsible for constructing mannan structures havebeen reported to cause various degrees of filamentous growthdefects in C. albicans (44, 48, 49). For example, all pmt mutationsexcept pmt5 were reported to suppress hyphal morphogenesis understandard inducing conditions (36). In this study, we found thatthe loss of MNN5 function blocked hyphal formation on severalsolid hypha-inducing media, but not in the corresponding liquidmedia. Although the reasons remain unclear, it might be thathyphal growth on and penetration into solid media are much moredemanding on cell wall strength and integrity than growth inliquid media or that hyphal growth on solid medium requirescertain cell surface properties which are lost in the mutant.Such conditional hyphal growth defects have been described formany C. albicans mutants (4, 48, 49), underscoring the highdegree of complexity of the signals that feed into the signaltransduction pathways for filamentous growth. The impaired hyphalgrowth observed as a result of various mannosyltransferase genemutations suggests that the defect might be a consequence ofthe loss of proper glycosylation of many cell surface proteinsimportant for the construction of hyphal morphology. However,it remains possible that some members of the mannosyltransferasefamily may have more specific roles in regulating cell morphogenesis.Indeed, Timple et al. (48, 49) found that overexpression ofgenes encoding components involved in hyphal formation suppressedthe hyphal growth defect of a pmt6 mutant, suggesting that Pmt6may act upstream of the morphogenetic signaling pathways.

LF is a 77-kDa iron-binding protein present in milk and mucosalsecretions. It is also released by specific granules of polymorphonuclearleukocytes during inflammation. LF has been ascribed many diversebiological functions, most of which are immunomodulatory orantibacterial (5, 51, 52). It inhibits microbial growth by effectivelysequestering free iron from the environment. The protein andits N-terminal fragment (amino acids 1 to 11) are also knownto have direct bactericidal and candidacidal activities (6,21). However, the mechanisms of cell killing are not understood.Some studies have suggested that the peptide targets energizedmitochondria, causing the production of ATP and reactive oxygenspecies. The reactive oxygen species may lead to a reductionin the cellular thiol level, whereas ATP interacts with theLF-interacting proteins in the cell envelope. These events arethought to collectively contribute to cell death (29). We demonstratedthat the mnn5 ${Delta}$ mutant lost sensitivity to LF killing and thatthis effect depends on the cellular iron status, rendering MNN5the first gene identified that mediates the candidacidal activityof LF. Mnn5 is an intracellular protein and thus unlikely tobe the direct target of LF. It is more plausible that Mnn5 modifiessome cell surface proteins which are responsible for the interactionwith LF or the peptide. The failure of enzyme-dead Mnn5 to restoreLF sensitivity to the mnn5 ${Delta}$ mutant supports this view. Importantly,among a total of eight C. albicans mannosyltransferase mutantstested, only the mnn5 ${Delta}$ mutant exhibited increased resistanceto LF. In contrast, several mutants exhibited increased sensitivityto either Congo red, hygromycin B, or both. These observationsstrongly suggest that Mnn5 has a specific role in determiningC. albicans sensitivity to LF, while the functions of multiplemannosyltransferases are required for resistance to cell wall-damagingagents or certain antifungal compounds. Our finding providesimportant information for further elucidation of the mechanismsunderlying the LF killing of C. albicans. Given the potent candidacidalactivity of LF, finding the LF target and designing compoundsthat may mimic LF activity could be a worthy strategy for developingnew anti-C. albicans therapies.

ACKNOWLEDGMENTS

We thank Neil A. R. Gow and Joachim F. Ernst for generouslyproviding us the MNT and PMT mutants, respectively. We alsothank members of the Wang laboratory for stimulating discussionsand suggestions.

Y.W. is an adjunct associate professor in the Department ofMicrobiology, National University of Singapore.

FOOTNOTES

* Corresponding author. Mailing address: Institute of Molecular and Cell Biology, 61 Biopolis Drive, Singapore 138673, Singapore. Phone: 65-6586-9521. Fax: 65-6779-1117. E-mail: mcbwangy{at}imcb.a-star.edu.sg.

REFERENCES

Top