Loss of Cell Wall Mannosylphosphate in Candida albicans Does Not Influence Macrophage Recognition*

The outer layer of the cell wall of the human patho-genic fungus Candida albicans is enriched with heavily mannosylated glycoproteins that are the immediate point of contact between the fungus and cells of the host, including phagocytes. Previous work had identified components of the acid-labile fraction of N -linked mannan, comprising (cid:1) -1,2-linked mannose residues attached via a phosphodiester bond, as potential ligands for macrophage receptors and modulators of macrophage function. We therefore isolated and disrupted the Ca MNN4 gene, which is required for mannosyl phosphate transfer and hence the attachment of (cid:1) -1,2 mannose oligosaccharides to the acid-labile N -mannan side chains. With the mannosylphosphate eliminated, the mnn4 (cid:2) null mutant was unable to bind the charged cat-ionic dye Alcian Blue and was devoid of acid-labile (cid:1) -1,2-linked oligomannosaccharides. The mnn4 (cid:2) mutant was unaffected in cell growth and morphogenesis in

Candida albicans is now one of the most common agents of bloodstream infections in severely immunocompromised neutropenic patients where mortality may be as high 50% (1,2). In healthy immunocompetent individuals phagocytic leukocytes of the innate immune system are effective in clearing C. albicans. Neutrophils and activated macrophages play key roles in elimination of fungal pathogens (3). However, in vitro experiments with macrophages have often shown that yeast cells that are taken up into the phagolysosome develop germ tubes that can ultimately expand and lyse macrophages (4). In the macrophage C. albicans induces the expression of enzymes in the glyoxylate cycle (5,6) and other genes that promote its survival. Considerable interest therefore exists in the nature of the ligands and phagocyte receptors that are involved in recognition and stimulation of macrophages and other leukocytes by C. albicans (7)(8)(9) and in the molecular interactions between phagocytes and fungal cells in general (10 -12).
Interactions between C. albicans and macrophages involve the binding of target ligands on the C. albicans cell surface to receptors on the macrophage surface. Recognition of C. albicans by macrophages is followed by the induction of a variety of immunological mechanisms, whose function is to limit the damage caused by the organism (3,13). However, C. albicans also suppresses aspects of macrophage response to evade host defenses (3,4,14), and glycolipid-phospholipomannan of the fungus can induce macrophage apoptosis (15,16). Recognition of C. albicans by macrophages and subsequent modulation of macrophage function may therefore be an important early event in determining the subsequent course of disease and/or clearance of C. albicans in the infected host. Although mannan and ␤-1,3and ␤-1,6-glucan (7)(8)(9)17) have been shown to act as macrophage ligands, the relative contribution of each cell wall component to pathogenicity and/or the generation of a successful immune response has not been established.
The outer cell wall of C. albicans is enriched with glycoproteins that are modified by highly branched N-linked glycosylation including a mannosylphosphate-containing fraction that has been implicated in macrophage interactions (18,19). C. albicans N-linked mannan ( Fig. 1) consists of an ␣-1,6linked polymannose backbone with side chains comprising ␣-1,2and ␣-1,3-linked oligomannosides (18) and, in serotype A strains, ␤-1,2 mannose residues. Mannosylphosphate consists of chains of one to fourteen ␤-1,2-linked mannose residues attached to the side chains via phosphodiester linkages (19,20). N-linked mannan can be resolved by mild acid hydrolysis into acid-stable and acid-labile fractions, the former consisting of the phosphodiester bond and all proximal mannose residues and the latter of ␤-1,2-linked mannose chains only. In C. albicans serotype A strains the acid-stable fraction contains a number of ␣-1,2and ␤-1,2-linked mannose side chains, which constitute antigenic factor 6 and are absent in serotype B strains (21). C. albicans N-linked mannan is illustrated in Fig.  1. The other known source of ␤-1,2-linked mannose residues in the C. albicans cell wall is phospholipomannan, which consists of a lipid (ceramide) attached by a unique mannosylphosphate/ manno-inositol phosphate spacer to an unbranched ␤-1,2-oligomannoside chain of up to 19 residues in length (15).
In S. cerevisiae there are at least four potential mannosylphosphorylation sites. In N-linked mannan, mannosylphosphate may be attached to the terminal ␣-1,6-mannose residue in the ␣-1,6-linked polymannose backbone, ␣-1,2-linked mannose residues in side chains emerging from the ␣-1,6-linked backbone, and the Man 8 GlcNAc 2 oligosaccharide core. In Olinked mannan, mannosylphosphate may be attached to the second mannose residue of O-linked mannan (18,19,31). Much information concerning the control of mannosylphosphorylation in S. cerevisiae stems from the observation that mannosylphosphate-deficient cells have a reduced negative charge and consequently fail to bind the positively charged phthalocyanine dye Alcian Blue (32). This allowed the development of qualitative (33) and quantitative (34) assays of yeast cell mannosylphosphorylation. The role of the S. cerevisiae, ScMNN6, and ScMNN4 genes that are involved in mannosylphosphorylation of both O-and N-linked mannan (35,36) have been reviewed comprehensively (19). ScMnn6p shares significant homology with the Golgi ␣-1,2-mannosyltransferase Kre2p/ Mnt1p. The Scmnn6 mutant has diminished mannosylphosphate transferase activity and is proposed to encode the mannosylphosphate transferase (37). ScMnn4p does not display homology to known mannosyltransferases and has been proposed to act as a positive regulator of ScMNN6 (34). Mannosylphosphate transferase activity is reduced by ϳ80% in both the Scmnn4 and Scmnn4/MNN4 diploid strains, neither of which bind Alcian Blue, indicating that the Scmnn4 mutation is dominant (38). ScMnn4p has type II membrane protein topology and is characterized by a 130-amino acid region at the C terminus of the protein, consisting mainly of lysine-glutamic acid tandem repeats (KKKKEEEE), which is essential for Mnn4p activity (33).
Here we generate a mnn4 null mutant strain of C. albicans that is devoid of mannosylphosphate on its surface. The mutant was viable and unaffected in virulence, indicating that neither mannosylphosphate or the ␤-1,2-oligomannosides that are linked to it are involved in virulence or the interactions of C. albicans cells with macrophages.

EXPERIMENTAL PROCEDURES
Strains, Media, and Transformations-C. albicans was grown in shake culture in YPD (1% yeast extract, 2% mycological peptone, 2% glucose), GYEP (2% (w/v) glucose, 0.3% yeast extract, 1% peptone), SD (0.67% yeast nitrogen base, 2% (w/v) glucose), or NGY (0.1% yeast extract, 0.1% neopeptone, 0.4% glucose) with or without 1.5% agar. The media for growth of Ura Ϫ auxotrophs were supplemented with 25 g/ml uridine. For sensitivity testing, cell wall damaging agents were added to YPD agar at various concentrations. Germ tube formation was induced in 20% fetal bovine serum (Invitrogen) at 37°C, at a maximum cell density of 10 7 /ml. The C. albicans strains used or constructed in this study are shown in Table I. The plasmids were maintained in DH5-␣ Escherichia coli cells. C. albicans cells were transformed using the lithium acetate protocol (39), and transformants were selected on SD agar plates.
Creation of mnn4⌬ Null Mutant-The putative CaMNN4 (ORF) 1 was amplified by PCR with a single primer pair (Ca-MNN4 -3, 5Ј-CCACCA-CAACTATCAAAAAG-3Ј; Ca-MNN4 -4, 5Ј-CTAAGAAATAACGTCCA-AATC-3Ј) and cloned into pGEM-T Easy (Promega Ltd., Southampton, UK) to make plasmid pDH1. To make MNN4 disruption cassettes, the HindIII-KpnI fragment of CaMNN4 in pDH1 was replaced with the hisG-URA3-hisG cassette of pMB7 and p5921 to create plasmids pDH2 and pDH3, respectively (Table I). Plasmids pDH2 and pDH3 were digested with KspI (SacII) and SacI to yield a 4.8-kb disruption cassette containing the URA-blaster cassette, flanked by 320-bp upstream and 307-bp downstream sequences of MNN4. Plasmid pDH3 contained BamHI restriction sites at the downstream ends of the HisG repeats that were derived from p5921 and absent in pDH2. Sequential rounds of transformation of CAI-4, with cassettes derived from plasmids pDH2 and pDH3, respectively, were carried out to disrupt CaMNN4 (40). Ura Ϫ segregants were recovered on SD plates supplemented with 1 mg/ml 5-fluoro orotic acid 5-FOA and 25 g/ml uridine between transformations. Disruption was initially confirmed by PCR from independently derived transformants. Disruption was further confirmed by Southern analysis.
Reintegration of MNN4 into the Null Strain-Control strains were generated in which MNN4 was integrated into the RPS10 gene of the mnn4⌬ null mutant. The putative CaMNN4 ORF with 744-bp upstream and 27-bp downstream nucleotides was amplified from CAF2-1 genomic DNA by PCR with a single primer pair (Ca-MNN4 -7, 5Ј-GAACAAGAGCTCTCTTCTTTTTCTTTTATAAC-3Ј; Ca-MNN4 -10, 5Ј-FIG. 1. C. albicans N-linked mannan. Antigenic factor 6 consists of ␣-1,2and ␤-1,2-linked mannose residues and is specific to C. albicans serotype A. Antigenic factor 5 consists of the ␤-1,2-linked mannose residues within the acid-labile fraction and the ␤-1,2-mannobiose and -mannotriose (where present) within antigenic factor 6. Serotype B strains do not have ␤-1,2-linked mannose residues in the acid-stable fraction, therefore factor 5 in serotype B consists only of acid-labile ␤-1,2-linked mannose residues. Mannosylphosphate consists of the phosphate group and its associated ␤-1,2-linked oligomannosides. ␣-1,6-Linked side chain branches, which have been identified in serotype B C. albicans, are not shown. Chain lengths of up to 14 mannose residues have been reported in the acid-labile fraction (20). This figure is based on the work of Shibata et al. (63).
GAAGAAGCGGCCGCGTAAGTAAATATGTTTTATGC-3Ј) containing SacI and NotI restriction sites (underlined). The MNN4 PCR product was cloned into pGEM-T Easy and three independent recombinants, pDH4A/B and C, were selected. The MNN4 gene was excised from pDH4A by digestion with SacI and NotI and ligated into SacI-and NotI-digested CIp10 (41) to make the reintegration vector pDH5 (Table  I). The mnn4⌬ mutant was transformed with StuI cut pDH5, and transformants were screened by Southern analysis to detect single and multiple reintegrants of MNN4 at the RPS10 locus. Ura Ϫ mnn4⌬ null mutants were also transformed with StuI-digested CIp10 to create a control strain in which URA3 was expressed at the RPS10 locus.
Northern Analysis-Levels of MNN4 in parental strains, heterozygous and homozygous disruptants, and reintegrant strains with one or more copies of CaMNN4 placed at the RPS10 locus were measured by Northern analyses. RNA was isolated from yeast cells harvested in mid-log phase (optical density, 0.8) after growth in YPD at 30°C. RNA samples were heated at 50°C for 60 min in 55% Me 2 SO, 1.1 M glyoxal, separated on a 1.4% (w/v) nondenaturing agarose gel, and then transferred onto Hybond-N nylon membranes and hybridized with 32 P-labeled probes. The CaMNN4 transcript was detected using a 1.5-kb probe derived from the deleted region of CaMNN4. ACT1 mRNA was detected as a loading control.
DNA Sequencing-The putative MNN4 genes in three independent plasmids, pDH4A, pDH4B, and pDH4C, were amplified in Big Dye 2.0 Terminator Sequencing Cycle reactions (Applied Biosystems). The sequenced ORF was submitted to GenBank TM (accession number AF481861).
Alcian Blue Binding and Analysis of Whole Cell Mannan-Alcian Blue binding assays were adapted from Odani et al. (34). A suspension of ϳ1.4 ϫ 10 7 washed, stationary phase C. albicans cells were suspended in 1 ml of 30 g/ml Alcian Blue (George T. Gurr, London, UK) in 0.02 M HCl (pH 3.0), incubated at room temperature for 10 min, and pelleted by centrifugation. 〈 620 values of 100-l supernatant samples were determined in a spectrophotometer. Alcian Blue concentration was determined by reference to a standard curve. Alcian Blue binding (pg/cell) was calculated according to the formula: x ϭ [(u Ϫ v) Ϭn] ϫ 10 6 where x ϭ Alcian Blue bound (pg/cell), u ϭ original Alcian Blue concentration (g/ml), v ϭ final Alcian Blue concentration (g/ml), and n ϭ number of cells stained.
For analysis of whole cell mannan, C. albicans cells were labeled with [2-3 H]mannose. Acid-labile mannan was released by boiling cells in 10 mM HCl for 1 h. After neutralization with 10 mM NaOH products were analyzed by TLC according to the method of Hä usler et al. (42) and by fluorophore-assisted carbohydrate electrophoresis (FACE) as reported previously (43,44).
Macrophage Methods-Rat and mouse bone marrow-derived macrophages (RBMDM and MBMDM) were obtained by flushing the femoral marrow of male Sprague-Dawley rats or female BALB/c mice. Murine and rat macrophages, RAW264.7 cells (a murine macrophage cell line; ATCC TIB-71), and L929 mouse fibroblast cells (7,45) were cultured either in Dulbecco's modified Eagle's medium 10 (DMEM10; rat-derived cells) (Invitrogen) or RPMI 1640 medium (mouse-derived and RAW264.7 cells) (Invitrogen) supplemented with 10% heat-inactivated fetal calf serum, 2 mM L-glutamine, 50,000,000 units/liter penicillin, and 50 g/liter streptomycin (Sigma) at 37°C in air with 5% CO 2 . Macrophage growth media were supplemented with 10 -15% DMEM10/ RPMI 1640 in which L929 cells had been grown for 5-7 days, to provide a source of macrophage colony-stimulating factor (45). Macrophages were distributed into tissue culture plates (Corning) or Chamber slides (Fisher) for phagocytosis assays. For RBMDM phagocytosis assays C. albicans cells were incubated with RBMDM at a 1:1 C. albicans: macrophage ratio for 30 min. Macrophage monolayers were stained with 0.1 mg/ml acridine orange for 90 s and 1 mg/ml crystal violet for 20 s (crystal violet quenches the acridine orange staining of noninternalized Candida cells) (46). Phagocytosis was expressed as the percentage of macrophages containing one or more C. albicans yeast cells and as a phagocytic index (average number of C. albicans cells within each macrophage). For MBMDM and RAW264.7 adhesion/phagocytosis assays with live C. albicans, yeast cells were incubated with 20 M FUN1 (Molecular Probes) for 1 h at 30°C and added to MBMDM at a 20:1 C. albicans:macrophage ratio. After 1 h of incubation at 37°C in air with 5% CO 2 , adhesion/phagocytosis was quantified by washing off nonadhered cells, lysing macrophages in Triton X-100, and then measuring fluorescence at excitation/emission of 485/538 nm in a Fluoroskan II fluorometer (Titertek). For adhesion/phagocytosis assays with heatkilled C albicans, cells were killed by heating for 1 h at 100°C, washed in PBS, stained with Rhodamine green ϫ (Molecular Probes) as de-scribed by the manufacturer, washed three times in PBS, and stored at 4°C in PBS with 10 mM NaN 3 until use.
Nitrite Production by Macrophages-Nitrite production by RBMDM was detected with Greiss reagent (47) in 24-well tissue culture plates with DMEM10 growth medium with 5% CO 2 at 37°C. Macrophages were stimulated with rat ␥-interferon (R & D Systems, Abingdon, UK) for 24 h. After incubation with C. albicans, culture supernatants were removed and centrifuged to remove debris and cells, and the A 540 of 50-l samples was determined in an iEMS MF plate reader and compared with standard solutions of nitrite diluted in DMEM10 (46).
Ex Vivo Adherence Assay-The relative adherence of C. albicans strains for mouse splenic marginal zone macrophages was assayed as described previously (48). Fungal strains, including a wild type control (strain A9) used in other adherence studies, were grown in GYEP rotating (180 rpm) overnight at 37°C. Each strain was transferred to fresh GYEP, incubated for 24 h under the same conditions, and then transferred once more to fresh medium and incubated for 24 h before harvesting and testing for adherence. Fungal cells were harvested by centrifugation, washed three times in sterile Dulbecco's PBS, and suspended in Dulbecco's PBS containing 5% fetal bovine serum. The final suspension of each was adjusted to 1.5 ϫ 10 8 yeast cells/ml, and 0.1 ml of the suspension was overlaid at 4°C onto each cryosection of mouse (female, BALB/c) splenic tissue and incubated at 4°C for 15 min. The sections were fixed in cold 1.5% glutaraldehyde, washed in cold water, air-dried, and stained with crystal violet, and yeast cells' adherence to the tissue was assessed by bright field microscopy. Each fungal strain was tested on four cryosections, and the yeast adherence to the tissue sections was evaluated for all marginal zone areas of each tissue section. The marginal zones that were associated with yeast cells were scored in the range of 1ϩ (1-4 yeast cells/marginal zone) to 4ϩ (greater than 20 yeast cells/marginal zone).
Virulence Tests-For mouse inoculation, C. albicans strains were shaken in NGY medium for 18 -24 h at 30°C. The harvested cells were washed twice in water and resuspended in saline to give an inoculum of 2.5-6.0 ϫ 10 2 colony-forming units/g mouse body weight in a final volume of 100 l. The inoculum was adjusted for different batches of DBA/2 mice to give consistent mean survival times for control strains. Inoculum concentrations were verified by viable counting. Groups of six or seven DBA/2 mice (Harlan Sera-Lab Ltd., Loughborough, UK) were inoculated intravenously with each strain. Mice weighed ϳ20 g and were supplied with food and water ad libitum. Survival was monitored twice daily over 28 days. Animals that became seriously ill, showing hunched posture, ruffled fur, and reduced mobility, were humanely terminated, and their deaths were recorded as occurring on the following day. The kidneys were removed aseptically post mortem, and homogenized in water, and the C. albicans tissue burdens were determined by viable counting.

RESULTS
Identification of a Putative C. albicans Homolog of Sc-MNN4 -A putative homolog of ScMNN4 ORF 6.4390 (49) was found in the Stanford C. albicans sequence data base (Stanford Genome Technology Center website, www-sequence.stanford.edu/group/candida) by BLAST search. The ORF was 2991 bp in length and predicted to encode a 997-amino acid protein.
The protein contained a single predicted transmembrane domain (TMpred and TMAP) and a region of lysine/glutamic acid tandem repeats. The repeat region in the CaMNN4 gene was shorter (71 residues) and closer to the N terminus of the translated protein than the similar region in ScMnn4p. The MNN4 promoter and ORF sequences were PCR-amplified from genomic DNA and cloned into pGEM-T Easy. Three independent clones of the CaMNN4 ORF were sequenced and were identical but contained 11 nucleotide substitutions compared with ORF 6.4390 and an insertion of an extra glutamine codon in a Q 6 repeat. The size of the putative MNN4 gene is therefore 2994 bp, encoding a 998-amino acid protein. The sequence was deposited in GenBank TM (accession number AF481861).
Construction of Mannosylphosphate-deficient C. albicans Strains by Sequential Disruption-To generate a mutant specifically defective in mannosylphosphate, we disrupted the putative CaMNN4 gene in C. albicans strain CAI4. To disrupt this gene a 1467-bp fragment (49% of the ORF) was replaced by a URA-blaster cassette, consisting of a functional C. albicans URA3 gene flanked by Salmonella typhimurium hisG repeats (40). Wild type, disrupted, and reconstituted heterozygous MNN4 alleles are shown in Fig. 2. PCR was used to screen potential heterozygous and homozygous transformants. Screening of second round transformants made use of the fact that null strains deficient in mannosylphosphate fail to bind Alcian Blue. The genotypes of strains CDH1-CDH8 (Table I) were confirmed finally by Southern analysis (Fig. 2).
The plasmid, pDH5, was used to reintegrate MNN4 in mnn4⌬ null strains. Plasmid pDH5 contained MNN4 with a 744-bp promoter region, ligated into the Candida integrating plasmid CIp10, which integrates at the RPS10 locus (41). Integration of CIp10 and/or pDH5 at the RPS10 locus was confirmed by Southern analysis (Fig. 2). Some strains (CDH11 and CDH12) contained single copies of pDH5 (Table I). Two further pDH5 transformants (CDH13 and CDH14) produced hybridization bands of higher molecular weight corresponding to integrants with two or more copies of pDH5 (Table I).
Transformant colonies were screened for Alcian Blue-positive staining. Empty vector transformants remained deficient in Alcian Blue binding. The single and multiple reintegrants bound Alcian Blue to a variable extent.
CaMNN4 expression was not detected in the null strains (CDH5, CDH6, or CDH15) by Northern analysis. Expression was reduced in the MNN4/mnn4⌬ strains (CDH1, CDH2, and CDH17) and was equivalent to that in the MNN4/mnn4⌬ strains in the multi-copy reintegrant CDH13. Expression of CaMNN4 in the single-copy reintegrant CDH11 was reduced compared with the MNN4/mnn4⌬ strains and the multi-copy reintegrant (data not shown).
MNN4 Mutant Is Deficient in Alcian Blue Binding but Has Normal Growth in Vitro-Wild type, mnn4⌬ null, and reconstituted strains were indistinguishable with respect to colony morphology, growth kinetics in YPD, temperature sensitivity, hypha production in response to serum or Spider medium (50) and sensitivity to Calcofluor White, Congo Red, and SDS (data not shown). Growth of the mnn4⌬/mnn4⌬ strains CDH5, CDH6, and CDH15, but not of MNN4/MNN4 or MNN4/mnn4⌬ strains, was inhibited by hygromycin B at a concentration of 800 g/ml.
The results of quantitative Alcian Blue binding assays (Fig.  3) demonstrated that the mnn4⌬ strains were deficient in mannosylphosphorylation. Alcian Blue binding was almost completely abolished by disruption of two copies of MNN4 (CDH5, CDH6, and CDH15), whereas disruption of a single copy of MNN4 resulted in an intermediate level of Alcian Blue binding (CDH1, CDH2, and CDH17).
Reintegration of MNN4 partially restored the capacity to bind Alcian Blue. Integration of multiple copies on MNN4 at the RPS10 locus restored Alcian Blue binding to the level of heterozygous mnn4⌬ strains (CDH13). The location of URA3 had no effect on Alcian Blue binding. The acid-labile ␤-1,2oligomannosaccharide fraction was resolved by TLC analysis (Fig. 4). This demonstrated the complete absence of mannosylphosphate from the products of whole cell acid hydrolysis from strain CDH5 (mnn4⌬/mnn4⌬) and its restoration in strain CDH13 (mnn4⌬/mnn4⌬, RPS10::[CIp10-URA3-MNN4] n ) (Fig.  4). This result was confirmed by FACE (not shown). Therefore  Table II. For CDH5, with URA3 expressed at the MNN4 locus, mean survival was significantly longer than for CAF2 (p Ͻ 0.01). The kidney burden in mice infected with CDH5 was also three logs below wild type (CAF2-1), which represented a significant reduction in tissue burden (p Ͻ 0.001). However, the mean survival time for CDH15, where the URA3 gene was expressed at the RPS10, locus was not significantly different to CAF2-1, and no other differences in mean survival times were statistically different to the parental control strain. The longer survival times and lower tissue burdens for mice infected with CDH5 were therefore due to the location of URA3 rather than to disruption of MNN4. The mnn4⌬ null mutant was therefore not affected markedly in the ability to colonize and kill mice in a systemic disease model.
The Absence of MNN4 Does Not Affect Interactions with Macrophages-Mannosylphosphate has been implicated as a possible macrophage ligand; therefore the interaction of the mnn4⌬ mannosylphosphate-deficient mutants with macrophages was examined. Strains CAF2-1, CDH5 (mnn4⌬/ mnn4⌬), and CDH13 mnn4⌬/mnn4⌬, RPS10::[CIp10-MNN4-URA3] n ) were tested in the RBMDM phagocytosis assay and the mouse splenic adhesion assay. No significant differences between strains were found in the RBMDM phagocytosis assay (mean percentage of phagocytosis, 27.7 Ϯ 0.93 with a range of 27.1-28.8; mean phagocytic index, 0.39 Ϯ 0.01 with a range of 0.38 -0.40; analysis of variance, p Ͼ 0.05; n ϭ 4). In addition, no differences were observed in the pattern of binding in an ex vivo adherence assay of any of the test parental, null and reconstituted strains (data not shown). Adherence of each of  4. TLC of whole cell acid hydrolysis products from mnn4 mutant strains. Acid hydrolysis products from [2-3 H]mannose-labeled cells were run on a Whatman TLC plate, which was exposed to Kodak X-OMAT LS film for 10 days. these strains to the marginal zones gave a mean 3ϩ adherence score for marginal zones associated with yeast cells. In addition, binding of each strain showed a high degree of specificity for the marginal zones, with very few yeast cells associated with white and red pulp areas of the spleen (not shown). These results are similar to those obtained for wild type strains in previous ex vivo adherence studies (48,51,52).
Live C. albicans cells from strains CAF2-1, CDH5, CDH6, CDH13, CDH15, and CDH17 were tested in the MBMDM adhesion/phagocytosis assay. The results confirmed that there were no significant differences between strains (mean sample fluorescence, 1.22 Ϯ 0.10 with a range of 1.12-1.33; analysis of variance, p Ͼ 0.05; n ϭ 3). In separate experiments, both live and heat-killed samples of strains CAI-4/CIp10, CDH17, CDH15, CDH11, and CDH13 were tested in adhesion/phagocytosis assays with MBMDM and RAW 264.7 cells (Table III). Again, there were no significant differences in adhesion/phagocytosis between any strains tested with either live or heatkilled cells (p Ͼ 0.05).
The ␥-interferon-induced nitrite production by macrophages was inhibited by the presence of C. albicans yeast cells. This repression was similar for CAF2-1 (parent), CDH5, and CDH6 (data not shown). Therefore, deletion of MNN4 did not affect the interaction with macrophages or their metabolism. Therefore mannosylphosphate is not required for adhesion and phagocytosis and did not affect macrophage function. DISCUSSION The C. albicans MNN4 gene was isolated as the closest homolog to the ScMNN4 gene, which is involved in the regulation of mannosylphosphate production. Disruption and reintegration of CaMNN4 resulted in the absence and subsequent restoration of C. albicans cell wall mannosylphosphorylation, as detected by whole cell Alcian Blue binding, and the analysis of isolated acid-labile ␤-1,2 oligosaccharides by TLC and FACE. Alcian Blue binding and Northern analysis results indicated that MNN4 is expressed at an intermediate level in heterozygous MNN4/mnn4⌬ strains. This contrasts with the dominance of the Scmnn4 mutation, described by several authors (33,36) and suggests some difference in the control of mannosylphosphorylation by ScMNN4 and CaMNN4. There are eight MNN4-like sequences in the C. albicans genome data base (www:http// genolist.Pasteur.fr/candidaDB/) and only two MNN4-like genes in S. cerevisiae. However, only the ScMNN4 and CaMNN4 genes have the KE repeat region that has been shown to be essential for activity. In addition, the complete absence of mannosylphosphate in the Camnn4⌬ mutant suggests that this gene product is responsible for most mannosylphosphate synthesis under the conditions tested to date. We have demonstrated that the absence of mannosylphosphate does not lead to any increased mobility of extracted cell wall proteins, indicating that the Camnn4⌬ mutation does not have a gross effect on N-mannan assembly.
Many genes have been implicated in the virulence of C. albicans on the basis of reverse genetics (53). It has become apparent in recent years that gene reintegration can cause phenotypic effects that result from factors unrelated to the reintegrated gene. The most important example of this is the location of URA3, which has effects on both virulence (54 -58) and adhesion (59). Virulence was attenuated in the homozygous disruptant strain but was restored by integration of the empty CIp10 vector alone, indicating that disruption of MNN4 was not the cause of attenuation of virulence. Differences in orotidine 5Ј-monophosphate decarboxylase specific activity in the parental and mnn4⌬ mutant strains used in this study showed reduced activity in strains with URA3 at the MNN4 locus (data not shown). MNN4 was also poorly expressed at the RPS10 locus, as evidenced by Alcian Blue binding results and Northern analysis. These data underline the importance of chromosomal location on gene expression and difficulties in virulence testing with strains generated by the Ura-blaster protocol.
N-Linked mannan and its subfractions have been shown to have a role in the attachment of C. albicans to macrophages and macrophage-like cells. In a competitive binding assay, mannan inhibited the binding of C. albicans to splenic marginal cells (a macrophage population) (51). This model was refined in similar assays with subfractions of mannan. Splenic marginal zone adhesion was inhibited, and therefore mediated, by ␤-1,2-mannotetraose (60) and, to a lesser extent, ␤-1,2mannotriose (27). The property of adhesion was retained in (but not necessarily restricted to) the ␤-1,2-mannotetraose fraction of mannan released from a serotype A C. albicans strain by mild acid hydrolysis and therefore originating from mannosylphosphate. It was shown subsequently that the acid stable fraction of N-linked mannan exhibited adhesin activity. This activity is stronger than that of ␤-1,2-mannotetraose (52) and is mediated by both the ␣-1,6-linked oligomannoside backbone and the ␣-1,2and ␣-1,3-linked oligomannoside side chains (61). The potential roles of mannan adhesins in pathogenesis were investigated by raising monoclonal antibodies to a cell wall extract that contained C. albicans phosphopeptidomannan, designated mAb B.6 and mAb B.6.1. mAb B.6.1 protected mice against invasive candidiasis, whereas mAb B.6 did not (62). Subsequent epitope analysis revealed that the B6.1 epitope was a ␤-1,2-mannotriose, whereas the B6 epitope was in the acid-stable fraction of N-linked mannan. The B6.1 epitope was specific to ␤-1,2-mannotriose within mannosylphosphate, because its activity was not blocked by the acidstable fraction from a serotype A strain, which would contain the ␤-1,2-oligomannoside component of antigenic factor 6 (30). This provided evidence that mannosylphosphate per se was important in host interactions rather than the subfraction represented by ␤-1,2-oligomannosides. However, the strain used in this experiment (CA-1) was retested subsequently and identified as serotype B. 2 Serotype B strains do not express antigenic factor 6, so the latter conclusion is no longer supported. Further evidence of a role for ␤-1,2-oligomannosides as mediators of adhesion to other macrophage type cells was provided by the finding that incubation of J774 cells (a macrophage-like cell line) or mouse peritoneal macrophages with native and synthetic ␤-1,2-mannotetraose inhibited subsequent phagocytosis of C. albicans (26). The results of attachment/phagocytosis assays with RBMDM, MBMDM, and RAW264.7 cells and splenic cell adhesion assays with the mnn4⌬ strains generated here failed to demonstrate a role for mannosylphosphate in the adhesion of C. albicans to any of these cell types. These results support the finding that the property of macrophage adhesion resides mainly in the acid stable fraction of N-linked mannan (60), which should be unaffected by the mnn4⌬ mutation. The apparent role for the acid-labile fraction in macrophage adhesion described above may have resulted from the use of competitive inhibition assays in its demonstration. The results obtained from the use of isolated mannan subfractions in such assays may demonstrate the effects of saturating potential receptor sites with excessive concentrations of substrate rather than the physiological conditions encountered by C. albicans and macrophages in vivo. Recent work has demonstrated the importance of ␤-glucans in the activation of leukocytes and as inflammatory mediators (7). The nonopsonic macrophage receptor Dectin-1 has been shown to mediate binding to both ␤-1,3and ␤-1,6-glucans and to trigger the inflammatory responses in a Myd88 and Toll-like receptor-2-dependent manner (7). In Pneumocystis carinii the Dectin-1 receptor is essential for killing by alveolar macrophages (12). These observations suggest that fungal ␤-glucan may be a major ligand for macrophage recognition and activation, although they do not exclude the possibility of a role for mannan. Indeed, mannan binding activity has been shown on selected macrophage populations (9). Because serotype A strains, including the mutant lineage created here, still retain some ␤-1,2-mannan in the acid-stable fraction, the formal possibility remains that this nonmannosylphosphate-associated ␤-1,2-mannan is important in macrophage interactions. This hypothesis can be tested by creating mnn4⌬ mutants in a serotype B background. In addition it is important to recognize possible differences in the recognition mechanisms within different macrophage populations. However, the present study eliminates the possibility that mannosylphosphate per se is essential for the recognition and phagocytosis of C. albicans by macrophages.