Candida albicans Pmr1p, a Secretory Pathway P-type Ca2+/Mn2+-ATPase, Is Required for Glycosylation and Virulence*

The cell surface of Candida albicans is the immediate point of contact with the host. The outer layer of the cell wall is enriched in highly glycosylated mannoproteins that are implicated in many aspects of the host-fungus interaction. Glycosylation of cell wall proteins is initiated in the endoplasmic reticulum and then elaborated in the Golgi as the protein passes through the secretory pathway. Golgi-bound mannosyltransferases require Mn2+ as an essential cofactor. In Saccharomyces cerevisiae, the P-type ATPase Pmr1p transports Ca2+ and Mn2+ ions into the Golgi. To determine the effect of a gross defect in glycosylation on host-fungus interactions of C. albicans, we disrupted the PMR1 homolog, CaPMR1. This mutation would simultaneously inhibit many Golgi-located, Mn2+-dependent mannosyltransferases. The Capmr1Δ null mutant was viable in vitro and had no growth defect even on media containing low Ca2+/Mn2+ ion concentrations. However, cells grown in these media progressively lost viability upon entering stationary phase. Phosphomannan was almost completely absent, and O-mannan was severely truncated in the null mutant. A defect in N-linked outer chain glycosylation was also apparent, demonstrated by the underglycosylation of surface acid phosphatase. Consistent with the glycosylation defect, the null mutant had a weakened cell wall, exemplified by hypersensitivity to Calcofluor white, Congo red, and hygromycin B and constitutive activation of the cell integrity pathway. In a murine model of systemic infection, the null mutant was severely attenuated in virulence. These results demonstrate the importance of glycosylation for cell wall structure and virulence of C. albicans.

Candida albicans is the most common fungal agent of invasive disease in humans (1,2). It is responsible for superficial epithelial infections and, in the immunocompromised host, lifethreatening systemic infections (3,4). The cell wall of C. albicans is the immediate point of contact between the fungus and host and hence is important in host-fungus interactions. The cell wall is composed of an inner layer of structural polysaccharides, ␤1,3and ␤1,6-glucans and chitin, and an outer layer that is enriched for mannoproteins (5,6). The highly glycosylated mannoproteins play important roles in adhesion, antige-nicity, and modulation of the host immune responses (7)(8)(9)(10)(11). Both the carbohydrate epitopes and the protein components have been implicated in these roles (7,8,12), although the exact epitopes involved are still unclear. The study of glycosylation in C. albicans will therefore increase our understanding of the host-fungus interaction. To determine the role of glycosylation in the virulence of this fungus, we deleted the Golgi P-type ATPase, which transports divalent cations into the Golgi, where they act as essential cofactors for mannosyltransferases.
In Saccharomyces cerevisiae, glycosylation is initiated in the endoplasmic reticulum by the transfer of the first mannose residue to serine or threonine in O-linked glycosylation (13) and by the transfer of the N-linked core structure to asparagine residues (14). The construction and transfer of the N-linked core (14,15) and the initiation of O-linked glycosylation (16) are essential processes. Glycosylation is continued in the Golgi with the extension of the linear O-linked glycans (17) and the extensive elaboration of the branched outer chains of the Nlinked glycans (18).
The process of glycosylation has been extensively studied in S. cerevisiae (13)(14)(15)17), and this has provided a resource for understanding glycosylation in C. albicans. However, key differences exist between the O-and N-glycan structures present in S. cerevisiae and C. albicans. For example, in C. albicans, the terminal O-linked glycans that are attached by ␣1,2-linkages, as opposed to ␣1,3-linkages in S. cerevisiae (19), and ␤1,2linked mannose residues are present in both the acid-labile and acid-stable N-linked glycans (20). Following the completion of the C. albicans genome project, it has also become clear that there has been expansion and contraction of the number of genes in different mannosyltransferase gene families of C. albicans. 1 Hence, the process of glycosylation has adapted in C. albicans, perhaps as a result of selective pressure from its interaction with a mammalian host.
To date, few genes involved in glycosylation in C. albicans have been analyzed in detail. Those studied include members of the PMT and MNT gene families, all of which act in O-glycosylation, and the studies have demonstrated the importance of O-glycans in virulence (19,(21)(22)(23)(24). Also, Mnn9p is involved in extension of the N-linked glycan outer chain and hence is required for normal cell wall composition (25). Phosphomannan has been implicated in the interaction of C. albicans with macrophages. However, deletion of the C. albicans MNN4 gene, which is required for mannosyl phosphate transfer, demonstrated that phosphomannan is not required for macrophage interactions or virulence of C. albicans (26). Genes such as CaVRG4 2 and CaSRB1, which encode proteins required for supplying the Golgi with GDP-mannose, the mannose donor, are essential in C. albicans, indicating the overall importance of glycosylation to cell viability (27)(28)(29)(30). However, to date, no single glycosylation event in the Golgi has been shown to be essential.
As well as requiring GDP-mannose as the mannose donor, Golgi-bound mannosyltransferases require manganese ions as essential cofactors. In S. cerevisiae, the P-type ATPase Pmr1p is a Ca 2ϩ /Mn 2ϩ ion pump that supplies the Golgi with these ions (31)(32)(33). Homologs of Pmr1p are classed as secretory pathway Ca 2ϩ -ATPases and are distinct from the sarco/endoplasmic reticulum Ca 2ϩ -ATPases and plasma membrane Ca 2ϩ -ATPases (34). The secretory pathway Ca 2ϩ -ATPase class of P-type ATPases has been identified in a wide range of fungal and animal cells, including S. cerevisiae. Disruption of PMR1 in S. cerevisiae results in a range of phenotypes, including some protein-sorting defects such as elevated secretion of heterologously expressed proteins (35,36) and incomplete processing of the ␣-factor (32). The null mutant also has severe glycosylation defects affecting both N-and O-linked mannoproteins (31)(32)(33), which are associated with manganese deficiency (33).
To determine the effect of a gross defect in glycosylation on host-fungus interactions and virulence, we constructed a null mutant of the C. albicans PMR1 homolog. We reasoned that this mutant would down-regulate glycosylation in general, allowing the importance of this protein modification to fungal virulence to be assessed. The null mutant had severe defects in both O-and N-glycosylation and was hypersensitive to cell wall stress. In a murine model of systemic infection, the strain was severely attenuated in virulence. These results demonstrate that normal glycosylation is not required for growth in vitro, but is essential for the virulence of C. albicans.

EXPERIMENTAL PROCEDURES
Strains, Media, and Culture Conditions-All strains used and constructed during this work are listed in Table I. Strains were grown at 30°C in YEPD medium (1% (w/v) yeast extract, 2% (w/v) mycological peptone, 2% (w/v) glucose) or in SD medium (0.67% (w/v) yeast nitrogen base with ammonium sulfate without amino acids, 2% glucose) with uridine (50 g/ml) as required. Calcium-deficient medium was prepared as SD medium but with the omission of CaCl 2 from the yeast nitrogen base. The only source of calcium ions in this medium was from the calcium salt of pantothenoate (present at 0.8 M). Solid media were prepared with 2% technical agar no. 3. Hyphal cells were grown in YEPD medium ϩ 20% newborn calf serum, RPMI 1640 medium (Invitrogen, Paisley, UK), or Lee's medium (pH 6.5) (37) at 37°C or in solid Spider medium (38) at 30°C. To induce acid phosphatase expression, cells were grown in Sabouraud glucose medium depleted of phosphate. Phosphate ions from 1% neopeptone were allowed to precipitate for 30 min after the addition of 0.01 M MgSO 4 and 0.28% NH 4 OH, followed by filtration and adjustment to pH 5.4 with concentrated HCl before the addition of glucose to a final concentration of 4% (w/v). For virulence testing, strains were grown in NGY medium (0.1% neopeptone, 0.4% glucose, and 0.1% yeast extract) at 30°C.

Construction of Capmr1⌬ Null Mutant and Re-integrant Strains-
The CaPMR1 gene was disrupted by the standard "ura-blaster" protocol (39). To make the disruption cassette, the 5Ј-and 3Ј-flanking regions of the gene were amplified by PCR (5Ј-primer pair 5Ј-GAAGAGCTCACG-GCAATGTAAGTAAGTGCG-3Ј and 5Ј-GAAAGATCTGTGATGGCTA-ATGTGATCG-3Ј, with the SacI and BglII restriction sites underlined, respectively; and 3Ј-primer pair 5Ј-GAAGTCGACCAGTATTGCTGCA-TTGAC-3Ј and 5Ј-GAAGCATGCTGCAAGCTATACACCATAAC-3Ј, with the SalI and SphI restriction sites underlined, respectively) and cloned into the complementary restriction sites of pMB-7 (39). The disruption cassette was released by digestion with SacI and SphI and contained the ura-blaster cassette flanked by complementary sequences 733 bp upstream and 620 bp downstream of CaPMR1. CaPMR1 was disrupted by sequential rounds of transformation of strain CAI-4 and the recycling of the URA3 marker by selection on SD medium plus 5-fluoroorotic acid (1 mg/ml) and uridine (50 g/ml). To avoid potential problems associated with the ectopic expression of URA3 (40), the Ura Ϫ Capmr1⌬ null strain was transformed with StuI-digested CIp10 plasmid (41) so that URA3 was expressed at the RPS1 locus (orf19.3002, formally referred to as RPS10). To create a re-integrant strain to act as a control, the CaPMR1 open reading frame plus 1055 bp of its promoter and 765 bp of its terminator were amplified by PCR (primer Pair 5Ј-CAGACCTAGTCCGACATTGGC-3Ј and 5Ј-ATGAAGCAAGTAT-CATTGGAGC-3Ј), and the 3.6-kb product was cloned into pGEM-T-Easy (Promega Ltd., Southampton, UK). The insert was released by NotI digestion and subcloned into the NotI site of CIp10. The resulting plasmid was digested with StuI and transformed into the Ura Ϫ Capmr1⌬ null strain. As a positive control for experiments, strain CAI-4 was also transformed with StuI-digested CIp10; hence, all strains analyzed in this work have URA3 expressed at the RPS1 locus. The two RPS1 alleles in strain CAI-4 can be distinguished by the presence of a polymorphism resulting in the presence or absence of an XbaI restriction site downstream of RPS1. As an added control, all strains used were confirmed to have CIp10 inserted into the non-XbaIcontaining RPS1 allele.
Sensitivity Testing-To test strains for sensitivity to specific wallstressing agents, strains were initially grown for 24 h in YEPD medium and then washed with water and resuspended at A 600 ϭ 1. These cells were inoculated into YEPD medium at A 600 ϭ 0.01, and 95-l volumes were pipetted into microdilution plate wells. Test agents in 5-l volumes were added at a range of doubling dilutions. Plates were incubated for 16 h at 30°C, and absorbance was read at 600 nm. All strains were tested in duplicate. The agents tested were Calcofluor white (100 g/ml), Congo red (100 g/ml), SDS (0.1%), hygromycin B (500 g/ml), a high salt concentration (NaCl 2 , KCl, CaCl 2 , MnCl 2 , or MgCl 2 ; each at 1 M), caffeine (50 mM), vanadate (80 mM), and tunicamycin (100 g/ml). The concentrations listed are the maximal concentration tested for each agent. Antifungal susceptibility testing was carried out by standard methods (42), except that agents were diluted beyond the stated ranges because derivatives of the C. albicans SC5314 genetic background are known to be highly susceptible to antifungal agents.
Protein Extracts and Western Blotting-To test for activation of the cell integrity pathway (43), cells were grown in YEPD medium at 30°C and collected in mid-exponential growth. For positive controls, the strains were stressed by the addition of 100 g/ml Calcofluor white 2 h before collection. Cells were washed and resuspended in extraction buffer (100 mM Tris-HCl (pH 7.5), 0.01% (w/v) SDS, 1 mM dithiothreitol, 10% (w/v) glycerol, protease inhibitor mixture (Roche Applied Science, Lewes, UK)) and then disrupted with glass beads in a FastPrep machine (Qbiogene, Inc., Cambridge, UK). The lysate was clarified by centrifugation at 21,500 ϫ g for 10 min. Protein extracts were quantified using the Coomassie protein assay reagent (Pierce, Cramlington,  . Prior to Western blotting, 50 g of protein was separated on a 4 -12% NuPAGE bis-Tris gel (Invitrogen) before blotting onto a polyvinylidene difluoride membrane. The membrane was blocked in phosphate-buffered saline plus 0.1% Tween 20 and 5 mg/ml bovine serum albumin for 2 h at room temperature. Detection was then carried out with the PhosphoPlus p44/42 MAPK antibody kit (New England Biolabs Inc., Hertfordshire, UK) according to the manufacturer's instructions. This antibody cross-reacts with C. albicans Mkc1p (Slt2p) in its phosphorylated form.
Cell Wall Analysis-Alcian blue binding assays were carried out as described previously (26). Briefly, ϳ1 ϫ 10 7 stationary phase cells were washed with water, resuspended in 1 ml of Alcian blue (30 g/ml) in 0.02 M HCl, and incubated at room temperature for 10 min. Cells were then spun down, and the absorbance of the supernatant at 620 nm was measured to determine the level of non-bound Alcian blue from a calibration curve. From these values, the specific amount of Alcian blue bound to cells was calculated (1 unit ϭ 1 g bound per A 600 unit of cell suspension). Total cell wall carbohydrate composition was analyzed by hydrolyzing cell walls with sulfuric acid and determining the monosaccharide composition by high performance anion-exchange chromatography with pulsed amperometric detection as described previously (44,45).

Metabolic Labeling of Glycans and TLC-For the analysis of Olinked and acid-labile glycans, yeast cells were initially labeled with D-[2-
3 H]mannose. Cells growing in 2 ml of YP medium (1% (w/v) yeast extract, 2% (w/v) mycological peptone) ϩ 0.5% sucrose were incubated with 1.85 MBq of D-[2-3 H]mannose (555 GBq/mmol; PerkinElmer Life Sciences, Beaconsfield, UK) at 30°C for 90 min. Cells were then harvested, resuspended in 100 l of phosphate-buffered saline, and disrupted with glass beads as described above. The cell walls were collected and washed twice with 1 M NaCl and twice with water. O-Linked glycans were released by ␤-elimination with 100 mM NaOH for 16 h at room temperature. Cell walls were pelleted at 21,500 ϫ g for 10 min, and the supernatant containing O-linked glycans was retained for TLC analysis. The cell walls were washed twice with water and then boiled for 1 h in 10 mM HCl to release the acid-labile glycans. The remaining cell walls were pelleted as described above, and the supernatant was retained for TLC analysis.
Acid Phosphatase Zymogram-In situ acid phosphatase activity assays were performed following a modified method of Schweingruber et al. (46). Briefly, cells were grown overnight in phosphate-depleted Sabouraud glucose medium to induce acid phosphatase expression. Cells were collected and washed before resuspension in 200 l of lysis buffer (62.5 mM Tris-HCl (pH 6.8), 1 mM EDTA, 0.1 mM dithiothreitol, protease inhibitor mixture). Cells were then disrupted with glass beads as described above. The lysate was clarified by two rounds of centrifugation at 21,500 ϫ g for 10 min. For endoglycosidase H treatment, the native sample was treated with 25 milliunits of endoglycosidase H (Roche Applied Science) in 50 mM sodium acetate (pH 5.2) for 16 h at 37°C. Samples were then mixed with nondenaturing loading dye (62.5 mM Tris-HCl (pH 6.8), 0.01% bromphenol blue, 15% glycerol) and run on a 6% Tris/glycine-polyacrylamide gel (Invitrogen) under nondenaturing conditions for 6 h at 125 V. The gel was rinsed in 100 mM sodium acetate (pH 5.2) for 10 min at room temperature and then incubated with substrate solution (0.05% ␣-naphthyl phosphate in 100 mM sodium acetate (pH 5.2)) for 30 min at 37°C. Finally, the gel was stained with 0.05% ␣-naphthyl phosphate and 0.03% fast blue in 100 mM sodium acetate (pH 5.2) at 60°C for 10 -30 min until the color developed.
Adhesion to Buccal Epithelial Cells-Adhesion to buccal epithelial cells (BEC) was assessed by a modification of a method previously described for measuring adhesion to vaginal epithelial cells (47). Briefly, BEC were collected from a healthy volunteer, washed twice with physiological saline, and resuspended at 5 ϫ 10 5 BEC/ml. C. albicans strains were grown in NGY medium, washed twice with physiological saline, and suspended at 5 ϫ 10 6 cells/ml. The C. albicans strain and BEC were then mixed in equal proportions (200 l) to achieve a 10:1 ratio of yeast cells to BEC and incubated at 30°C for 1 h. The cells were fixed with formalin, and the BEC were scored for the number of yeast cells adhered. At least 150 BEC were assessed for C. albicans adherence, and the assay was carried out in triplicate.
Virulence Assays-For virulence testing of strains, immunocompetent female BALB/c mice (Harlan Sera-Lab Ltd., Loughborough, UK) were challenged intravenously. Strains were grown with shaking for 18 -24 h in NGY medium at 30°C. Cells were washed twice with water and resuspended in physiological saline. Groups of five or six mice were inoculated intravenously with each strain at 2.1 ϫ 10 4 colony-forming units (cfu)/g of mouse body weight. Mice were monitored over 28 days; animals showing signs of distress or illness were humanely terminated; and deaths were recorded as occurring the following day. The kidneys and brains were removed post mortem and homogenized in 0.5 ml of water, and C. albicans tissue burdens were determined by viable counting.

Isolation and Analysis of CaPMR1-We initially identified
CaPMR1 before the completion of the C. albicans genome project by PCR with primers based on an unpublished sequence showing homology to S. cerevisiae PMR1 and a degenerate primer based on the P-loop found in all P-type ATPases. The regions surrounding CaPMR1 were then cloned by unidirectional Vectorette PCR (Sigma, Haverhill, UK) (48). The CaPMR1 open reading frame of 2754 bp (GenBank TM /EBI accession AJ277171) is predicted to encode a protein of 917 amino acids. Subsequently, the C. albicans genome sequence has been completed, enabling this open reading frame to be confirmed as the only homolog of S. cerevisiae PMR1. As in S. cerevisiae, CaPMR1 is linked to SUA5, which is involved in translation initiation. However, the orientation of the two genes is reversed, and the conservation of gene order does not extend further in either direction along the chromosome.
The deduced amino acid sequence of CaPMR1 demonstrated high homology to other secretory pathway P-type ATPases of subfamily IIA (68.7, 62.4, 62.1, 58.3, 57.6, 51.8, and 51.5% identity to Pmr1p of Pichia angusta, Kluyveromyces lactis, S. cerevisiae, and Yarrowia lipolytica; Schizosaccharomyces pombe Pgak2; and human and rat ATP2C1, respectively). Ten putative transmembrane regions are predicted by hydrophobicity profile analysis (available at www.cbs.dtu.dk/services/TM-HMM/) and by the similarity of this profile to other P-type Ca 2ϩ -ATPases. As with other P-type Ca 2ϩ -ATPases, the transmembrane domains are clustered into two groups separated by a central hydrophilic loop. All 10 conserved regions (regions a-j) that characterize the P-type ATPase family (49) are maintained in CaPMR1 (Fig. 1). In particular, region f contains the aspartic acid residue in the D 349 KTGTLT motif that is the phosphorylation site in P-type ATPases (34). Region c contains the T 189 GE motif believed to be important in removing the phosphate group from the enzyme intermediate. Regions g and i, responsible for ATP binding, are also well conserved.
Deletion of CaPMR1-CaPMR1 was disrupted in strain CAI-4 by the standard ura-blaster method (39). This involved the deletion of the central 1803-bp region containing all 10 regions characteristic of P-type ATPases. To avoid potential problems with ectopic expression of URA3, the Ura Ϫ derivatives were transformed with plasmid CIp10 so that URA3 was expressed from the neutral RPS1 locus (40,41). A re-integrant strain was also constructed in which CaPMR1 was introduced under the control of its own promoter into the Capmr1⌬ null mutant at the RPS1 locus. Strain CAI-4 transformed with CIp10 was used as a control in all experiments.
Deletion of CaPMR1 had no direct effect on growth rate in either YEPD medium or minimal medium (SD medium). However, the yeast cells in liquid medium tended to form small aggregates ( Fig. 2A). There was also no obvious defect in hyphal formation in response to serum, RPMI 1640 medium, or Lee's medium at pH 6.5. However, there was a delay in filament formation on solid Spider medium (Fig. 2B). Both the small cellular aggregates and the delay in filament formation on Spider medium were restored to those of the wild-type strain in the re-integrant strain. Hence, in vitro growth was not markedly affected in the Capmr1⌬ null mutant.
Calcium-related Growth Defects in Capmr1⌬-As mentioned above, deletion of CaPMR1 had no effect on the growth rate of the strain in laboratory media. Manganese ions are normally present only at trace concentrations below 10 M in laboratory media and cannot be depleted further without chelation. Therefore, growth parameters were initially assessed under Ca 2ϩdepleted conditions alone. Even when the Capmr1⌬ mutant was grown in Ca 2ϩ -depleted medium, there was no obvious effect on growth rate (Fig. 3A). The only source of Ca 2ϩ in this medium was from the calcium salt of pantothenoate (at 0.8 M) and contaminants. However, when viable cell counts were measured, it was clear that the null mutant started to lose viability as the cells entered stationary phase. After 72 h of growth in Ca 2ϩ -depleted medium, the viability of the null mutant dropped by over 1000-fold (Fig. 3B). When the null mutant was grown in standard SD medium ([Ca 2ϩ ] ϳ 1 mM), there was also a slight decrease in viability at 72 h. Loss of viability was not detected when the strain was grown in SD medium supple- mented with 10 mM CaCl 2 . The wild-type and re-integrant strains displayed no loss of viability when grown on SD medium at any of the Ca 2ϩ levels tested.
The Capmr1⌬ null mutant was also hypersensitive to EGTA. Growth of the null mutant was completely blocked by the addition of 15 mM EGTA, whereas the wild-type and re-integrant strains were unaffected (Fig. 4). This hypersensitivity could be reversed by the addition of 10 mM Ca 2ϩ or Mn 2ϩ , but not by an equivalent concentration of Mg 2ϩ , Na ϩ , or K ϩ . Unlike in S. cerevisiae, deletion of CaPMR1 did not alter sensitivity to the calmodulin inhibitor trifluoperazine or affect the level of manganese toxicity (data not shown).
Glycosylation Defects in the Capmr1⌬ Mutant-CaPMR1 provides the Golgi with manganese ions that are required as a cofactor for the Golgi-resident mannosyltransferases. We therefore analyzed the extent of N-and O-linked glycosylation in the Capmr1⌬ null mutant. We assessed the electrophoretic mobility of secreted acid phosphatase, which is known to be heavily N-glycosylated, by activity staining (50). The mobility of acid phosphatase present in protein extracts from the Capmr1⌬ null mutant was increased on the native gel (Fig. 5A). Hence, the null mutant had an N-glycosylation defect. After treatment with endoglycosidase H to remove the N-linked side chains, the acid phosphatase migrated faster through the native gel; however, it still appeared as a diffuse band, presumably due to variations in O-linked glycosylation. In this case, there was no difference among the null mutant and wild-type and re-integrant controls. However, because O-linked side chains are short, this method cannot be used to assess O-glycosylation defects.
N-Linked glycosylated side chains contain the acid-labile fraction phosphomannan, which is incorporated as the protein passes through the Golgi. This fraction provides the cell wall with its negative charge and can be easily detected with the cationic dye Alcian blue. The Capmr1⌬ null mutant demon-strated a severe reduction in specific Alcian blue binding from 143 Ϯ 6.7 units in the wild-type strain to 6.7 Ϯ 3.8 units in the null mutant (5% of the wild-type strain). The re-integrant strain had an intermediate Alcian blue binding phenotype of 102 Ϯ 7.5 units (74% of the wild-type strain). We also analyzed the acid-labile phosphomannan fraction directly by TLC. Cells were labeled with D-[2-3 H]mannose, and phosphomannan was isolated by HCl treatment of cell walls. The TLC autofluorogram clearly demonstrated that the Capmr1⌬ null mutant was almost completely devoid of phosphomannan (Fig. 5B). The wild-type and re-integrant strains displayed normal phosphomannan of one to eight ␤1,2-linked mannose residues (26,51).
The consequence of loss of CaPMR1 on the O-mannan structure was also assessed by TLC analysis. O-Mannan of C. albicans typically comprises one to five ␣1,2-linked mannose residues; the first is added in the endoplasmic reticulum, and subsequent residues are added in the Golgi. The TLC analysis clearly demonstrated that O-mannan was truncated in the Capmr1⌬ null mutant (Fig. 5C), with almost no detectable Man 3 to Man 5 . The fact that Man 2 was present on the Omannan isolated implies that some Golgi-based mannosyltransferase activity was retained in the Capmr1⌬ null mutant. Re-integration of CaPMR1 to the null mutant restored the normal O-mannan structure.
To confirm the gross glycosylation defect in the Capmr1⌬ null mutant and to determine the effect on the cell wall, we analyzed wall composition by high performance anion-exchange chromatography with pulsed amperometric detection. This method determines the carbohydrate composition of the wall by analyzing its monosaccharide composition, where glucose, mannose, and glucosamine are detected. These levels related to the presence of glucan, mannan, and chitin in the cell wall, respectively. This demonstrated a clear defect in mannan levels, which dropped from 39% of the cell wall in the wild-type strain to only 8% in the Capmr1⌬ null mutant (80% reduction). This decrease was reciprocated with an increase in glucan levels from 60% in the wild-type to 90% in the null mutant. The level of chitin in the strains was unchanged. Cell Wall Sensitivity and Cell Integrity Pathway Activation-To determine the effect of deleting CaPMR1 on the integrity of the cell wall, we tested the null mutant for sensitivity to a range of cell wall-perturbing agents and other agents whose effects have been associated with altered cell walls and glycosylation. The Capmr1⌬ null mutant was clearly hypersensitive to the cell wall-perturbing agents Calcofluor white and Congo red and displayed increased sensitivity to hygromycin B (Fig. 6), a phenotype commonly seen in glycosylation mutants (52). However, there was no change in the level of sensitivity to other agents such as SDS, caffeine, vanadate, tunicamycin, NaCl, and KCl. Antifungal susceptibility testing demonstrated that the null mutant showed no change in sensitivity to antifungal agents in clinical use, including the azoles (fluconazole, itraconazole, and ketoconazole), flucytosine, amphotericin B, terbinafine, and the ␤1,3-glucan synthase inhibitor caspofungin (data not shown).
The hypersensitivity of the Capmr1⌬ null mutant to cell wall-perturbing agents shows that the wall is altered and is more sensitive to stress. Therefore, we determined whether the cell integrity pathway was induced in the mutant. The cell integrity pathway signals through Pkc1p and results in the phosphorylation of the MAPK Mkc1p (43). We tested the activation of Mkc1p in our strains by Western blotting with a commercially available antibody that binds to phosphorylated Mkc1p. Mkc1p was activated in the Capmr1⌬ null mutant during exponential growth, whereas Mkc1p was not activated in the wild-type and re-integrant controls (Fig. 7). As a control, the strains were also stressed with 100 g/ml Calcofluor white for 2 h. This again demonstrated activation of the cell integrity pathway. The results of sensitivity testing and the constitutive activation of the cell wall integrity pathway show that the Capmr1⌬ null mutant has a defective cell wall.
Adhesion of the Capmr1⌬ Null Mutant-Mannoproteins have been implicated in C. albicans adhesion; hence, we tested the ability of the Capmr1⌬ mutant to adhere to BEC. Initially, it appeared that there was an increase in adherence from 0.38 Ϯ 0.03 adhered cells/BEC in the wild-type control compared with 1.19 Ϯ 0.22 adhered cells/BEC in the null mutant. However, the null mutant grew as small clumps of cells ( Fig. 2A), and this increase in adhesion may reflect the null mutant clumping phenotype. Indeed, the number of BEC with one or more adherent yeast cells did not increase (76 Ϯ 1.4% negative for the wild-type control and 74 Ϯ 4.5% negative for the null mutant), suggesting that adhesion was not significantly altered.
The Capmr1⌬ Null Mutant Is Attenuated in Virulence-The virulence of the Capmr1⌬ null mutant and relevant control strains was tested in a mouse model of systemic infection. The Capmr1⌬ null mutant was highly attenuated in virulence, with all mice surviving until the end of the experiment, whereas challenge with the wild-type control demonstrated a mean survival time of 7 days (log rank test; p Ͻ 0.05) (Fig. 8 and Table II). The re-integrant strain demonstrated an intermediate phenotype, with a mean survival time of 13.8 days, which was significantly different compared with the null mutant (log rank test; p Ͻ 0.05). The intermediate phenotype is suggestive of a gene dosage effect on virulence. Such effects are typical of other heterozygous and re-integrant strains containing a single wild-type copy of the target gene (53)(54)(55). In terms of tissue burdens determined for kidneys and brains of infected mice, the wild-type and re-integrant strains were not significantly different. The Capmr1⌬ null mutant demonstrated a clear reduction in tissue burdens, with a Ͼ2 log reduction in kidney cfu/g and a 1 log reduction in brain cfu/g (Table II). There were also a significant number of Candida-negative organs: 33% of kidneys and 50% of brains with the Capmr1⌬ mutant. Hence, although the Capmr1⌬ mutant was not affected in growth in vitro, this null mutant was extremely attenuated in virulence in a mouse model. DISCUSSION In this study, we have analyzed the effect of a gross defect in glycosylation in C. albicans by creating a null mutant in PMR1, which is required for glycosylation in the Golgi. Pmr1p is a P-type ATPase and supplies the Golgi body with calcium and manganese ions. In the Golgi, manganese ions are required as an essential cofactor for mannosyltransferases. Hence, by disrupting PMR1, we could determine the effect of a general defect in glycosylation both at the cellular level and in pathogenesis. The Capmr1⌬ null mutant clearly demonstrated a general glycosylation defect and associated cell wall changes and was severely attenuated in virulence in a murine model of systemic infection.
Previous studies have also examined gross glycosylation defects by studying the synthesis (29) and transport (27) of GDPmannose into the Golgi, where it acts as the mannose donor for mannosyltransferases. Total lack of GDP-mannose would block not only glycosylation in the Golgi, but also the essential processes of O-glycosylation and biosynthesis of the glycosylphosphatidylinositol anchor in the endoplasmic reticulum, as these steps require dolichylphosphomannose, which is synthesized from GDP-mannose. Hence, CaSRB1, which synthesizes GDPmannose, was found to be essential, and its depletion was found to have pleiotropic effects (29). The Golgi GDP-mannose transporter VRG4 is essential in C. albicans (27) and S. cerevisiae (56,57). This suggests that a complete lack of glycosylation in the Golgi results in non-viability. The essential nature of both CaSRB1 and CaVRG4, while emphasizing the importance of glycosylation, means that the role of the glycans in host-fungus interactions and virulence cannot be assessed. The Golgi GDPase CaGDA1 has also been studied (30); this enzyme converts GDP to GMP, which is then exported from the Golgi by Vrg4p in an antiport process with GDP-mannose. However, CaGDA1 is nonessential, presumably due to functional redundancy, and does not exhibit a gross glycosylation defect. It does demonstrate a partial defect in O-glycosylation and, as visualized by Alcian blue binding, has a slight reduction in phosphomannan; however, it displays no defect in the acid-stable N-glycans.
In S. cerevisiae, loss of PMR1 results in a strain displaying a slow growth phenotype, which can be alleviated by adding exogenous calcium ions to the medium; however, higher Ca 2ϩ concentrations in the medium result in the strain losing viability more quickly upon entering stationary phase (31). In our studies, deletion of CaPMR1 had no effect on the growth rate of the strain, even when grown on calcium-depleted medium. Additionally, the viability phenotype was opposite of that seen in S. cerevisiae, in that the Capmr1⌬ null mutant lost viability upon entering stationary phase after growth on Ca 2ϩ -depleted medium. Therefore, there must be subtle differences in the homeostasis of calcium ions in S. cerevisiae and C. albicans. Unlike in S. cerevisiae, we also saw no change in the level of manganese toxicity to the Capmr1⌬ null mutant. Therefore, there are also differences in the tolerance of S. cerevisiae and C. albicans to manganese ions.
The observed calcium-dependent drop in viability of the null mutant after entering stationary phase may affect pathogenicity. Calcium ions are present in serum at 95 g/ml, 48 g/ml (ϳ1.2 mM) of which is available in an accessible ionized form (58). This concentration approximates that present in standard SD medium, where only a slight decrease in long-term viability was seen (ϳ10% drop at 72 h). Therefore, although we cannot rule out that this may have influenced the virulence study, it is likely to be only a minor effect at physiological calcium and manganese ion levels and, as such, could not account for the almost complete loss of virulence in the Capmr1⌬ mutant.
Work in Y. lipolytica has shown that loss of Pmr1p can affect secretion. Depending on the marker protein analyzed, secretion is unchanged, enhanced, or decreased (59). Loss of Pmr1p can also affect the secretion of proteins expressed heterologously in S. cerevisiae and K. lactis (35,36,60). We tested the level of secretion in the Capmr1⌬ null mutant by analyzing the activity of cell-associated acid phosphatase and secreted aspartyl proteinase (data not shown). We detected a slight decrease in the level of cell-associated acid phosphatase activity (70 Ϯ 2.4% of the wild-type strain) and a slight increase in the level of secreted aspartyl proteinase activity (163 Ϯ 3.7% of the wild-type strain). Therefore, the changes in secretion detected in the Capmr1⌬ null mutant were not as high as those reported in other organisms. However, it was clear from these published reports that the levels of change detected are dependent on the protein that is secreted. Therefore, we cannot rule out that proteins other than cell-associated acid phosphatase and secreted aspartyl proteinase may be more severely affected in the Capmr1⌬ null mutant.
The Capmr1⌬ null mutant had a clumpy phenotype, with most cells present in small aggregates. This may be due to changes in the cell wall composition resulting in flocculation. However, examination of the clumps suggests that they may be the result of a cell separation defect. This suggests that daughter cells are delayed in cell wall separation, perhaps due to a decrease in the activity of cell wall-hydrolyzing enzymes. A similar clumping phenotype has been reported for other C. albicans glycosylation mutants such as the Camnt1⌬/ Camnt2⌬ double mutant (19). A more severe clumping phenotype was apparent in the Camnn9⌬ and Cavrg4⌬ mutants (25,27). A severe cytokinesis defect in an S. pombe pmr1⌬ mutant has also been reported recently (61). The clumping phenotype posed technical difficulties with in vitro assays of adhesion. However, it was clear that the overall number of epithelial cells adhered to was unaffected in the mutant, suggesting that loss of Pmr1p does not have a marked effect on adhesion.
Evidence was found for a gross defect in glycosylation in the Capmr1⌬ null mutant, with N-glycosylation severely affected as demonstrated by the under-glycosylation of acid phosphatase and the almost complete absence of phosphomannan (Fig.  5). O-Glycosylation was also affected such that O-mannan was truncated with a marked reduction in Man 3 to Man 5 (Fig. 5). The observation that Man 2 was still present in O-mannan isolated from the Capmr1⌬ null mutant implies that some mannosyltransferases are still active in the Golgi. This may reflect the localization of specific mannosyltransferases within the Golgi, where enzymes of the cis-Golgi may receive manganese from vesicles derived from the endoplasmic reticulum. Therefore, in the Capmr1⌬ null mutant, Man 2 may have been present because it was added in an early Golgi compartment that received manganese ions from endoplasmic reticulumderived vesicles. This also suggests that phosphomannan, which is almost completely absent in the Capmr1⌬ null mutant, is a component added later in the Golgi network and, as such, is more sensitive to the loss of Pmr1p. Alternatively, the presence of Man 2 in the O-mannan of the Capmr1⌬ null mu- tant may reflect the affinity of specific mannosyltransferases for manganese ions. Those that have the highest cation affinity may remain most active in a manganese-limited compartment. The cell wall was significantly altered in the Capmr1⌬ null mutant, exemplified by the 80% drop in mannose content correlating with the gross glycosylation defect. The consequences of the loss of Pmr1p were significant enough to result in the constitutive activation of the cell integrity pathway acting to compensate for the loss of mannan. The null mutant was hypersensitive to the cell wall-perturbing agents Calcofluor white and Congo red, which bind chitin and glucan and interfere with their synthesis and cross-linking. Increased sensitivity to hygromycin B, a phenotype characteristic of N-glycosylation mutants in S. cerevisiae (52), was also seen. However, we detected no change in sensitivity to vanadate, resistance to which is linked with N-glycosylation mutants in S. cerevisiae (57,62). There was also no change in sensitivity to SDS, which would target both cell wall proteins and the plasma membrane, or to high salt conditions, indicating that the mutant was not osmotically fragile.
The Capmr1⌬ null mutant displayed no significant growth defect in vitro, yet had an altered cell wall due to the gross glycosylation defect. This allowed us to assess the importance to virulence of glycosylation events that occur in the Golgi. The null mutant was highly attenuated in virulence, with all mice surviving the course of the experiment and displaying a significant drop in organ burdens. The low organ burdens and the number of Candida-negative organs suggest either that the host was able to effectively prevent colonization or that the mutant was unable to reach deep-seated infection sites. These results underline the importance of glycosylation to both the cell wall architecture and the pathogenesis of C. albicans disease. Questions still remain regarding the identification of specific epitopes involved in host-pathogen interactions, although, in terms of carbohydrate epitopes, recent work has confirmed the importance of O-glycans (19) and down-played the importance of phosphomannan (26). Continued research into the construction, modification, and presentation of specific mannan epitopes should ultimately provide answers to these questions.