Unraveling Unique Structure and Biosynthesis Pathway of N-Linked Glycans in Human Fungal Pathogen Cryptococcus neoformans by Glycomics Analysis*

Background: N-Glycan structures of the human pathogenic yeast C. neoformans have not yet been elucidated. Results: Cryptococcal N-glycans were composed of mostly mannoses with addition of xylose and xylose phosphate residues. Conclusion: Cryptococcal N-glycans show serotype-specific structures with variation in the length and in the presence of xylose. Significance: This is the first survey on the structure and biosynthesis pathway of C. neoformans N-glycans. The encapsulated fungal pathogen Cryptococcus neoformans causes cryptococcosis in immunocompromised individuals. Although cell surface mannoproteins have been implicated in C. neoformans pathogenicity, the structure of N-linked glycans assembled on mannoproteins has not yet been elucidated. By analyzing oligosaccharide profiles combined with exoglycosidase treatment, we report here that C. neoformans has serotype-specific high mannose-type N-glycans with or without a β1,2-xylose residue, which is attached to the trimannosyl core of N-glycans. Interestingly, the neutral N-glycans of serotypes A and D were shown to contain a xylose residue, whereas those of serotype B appeared to be much shorter and devoid of a xylose residue. Moreover, analysis of the C. neoformans uxs1Δ mutant demonstrated that UDP-xylose is utilized as a donor sugar in N-glycan biosynthesis. We also constructed and analyzed a set of C. neoformans mutant strains lacking genes putatively assigned to the reconstructed N-glycan biosynthesis pathway. It was shown that the outer chain of N-glycan is initiated by CnOch1p with addition of an α1,6-mannose residue and then subsequently extended by CnMnn2p with multiple additions of α1,2-mannose residues. Finally, comparative analysis of acidic N-glycans from wild-type, Cnoch1Δ, Cnmnn2Δ, and Cnuxs1Δ strains strongly indicated the presence of xylose phosphate attached to mannose residues in the core and outer region of N-glycans. Our data present the first report on the unique structure and biosynthesis pathway of N-glycans in C. neoformans.

ScMnn4p/ScMnn6p, generating acidic glycans with negative charge (14). S. cerevisiae mutants lacking the OCH1 gene (Scoch1⌬) exhibit hypersensitivity to high temperature and cell wall perturbation reagents (15). Candida albicans, an ascomycetous yeast and opportunistic human pathogen, has an N-glycan structure different from that of S. cerevisiae, because Candida N-glycans are modified with both ␣and ␤-linked mannoses (16,17). Deletion of Candida OCH1 results in hypersensitivity to cell wall-perturbing agents and an attenuation of virulence in a murine model of systemic candidiasis (18). The OCH1 deletion mutants of other ascomycetous yeast species, such as Kluyveromyces lactis, Hansenula polymorpha, Yarrowia lipolytica, Schizosaccharomyces pombe, and Pichia pastoris, and an ascomycetous filamentous fungus Neurospora crassa also exhibit altered morphological phenotypes under stress conditions, suggesting that outer chain N-glycans are important for cell wall integrity in yeast and fungal species (19 -24). The only exception was observed in the ascomycetous filamentous fungus and opportunistic human pathogen Aspergillus fumigatus, where deletion of OCH1 does not affect normal growth even under various stress conditions (25).
Based on the Cryptococcus genome database, C. neoformans is predicted to have more than 50 putative mannoproteins, which typically contain potential Asn (N)-glycosylation sites, putative Ser/Thr (S/T)-rich regions for O-glycosylation, and a glycosylphosphatidylinositol anchor (6,26). However, only a small number of C. neoformans glycoproteins have been characterized, and detailed investigation on oligosaccharide structures has not been carried out except for carbohydrate composition (27)(28)(29). Previous work involving bioinformatics and radioactive N-glycan analyses indicated that the structure of the dolichol-linked N-glycosylation precursor in the ER of C. neoformans is Man 9 GlcNAc 2 -PP-Dol with no addition of glucose residues due to a lack of Alg glucosyltransferases (Alg6p, Alg8p, and Alg10p), implying different processing of cryptococcal N-glycans in the ER (30). In addition, C. neoformans was predicted to lack most of the genes encoding M-pol I and II subunits, suggesting N-glycans have less elongated ␣1,6-mannan chains compared with those of S. cerevisiae (31). However, only limited information is available on the structural characteristics of the N-linked glycans in Cryptococcus species. In this study, we performed a comparative N-glycan profile analysis of C. neoformans species complex using MALDI-TOF mass spectrometry and HPLC, and we present the first report on the serotype-specific presence of a ␤1,2-xylose residue in high mannose type N-glycans in C. neoformans. In addition, we showed that the OCH1 and MNN2 genes play major roles in cryptococcal N-glycan processing in the Golgi. Furthermore, we provide a line of data supporting the presence of xylose phosphate residues on the core and outer regions of N-glycans in Cryptococcus.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-The yeast strains used in this study are listed in supplemental Table 1 and were generally cultured in YPD broth medium (1% w/v yeast extract, 2% w/v peptone, 2% w/v glucose) with shaking (200 rpm) at 30°C. YPD NAT (YPD solid medium containing 100 g/ml nourseothricin, Werner BioAgents; where NAT is nourseothricin acetyltransferase) or YPD G418 (YPD solid medium containing 200 g/ml geneticin, Sigma) was used for selection of C. neoformans transformants.
DNA Manipulation-The primer sets used for disruption and reintegration of genes in this study are listed in supplemental Table 2. Each gene was disrupted in the C. neoformans serotype A H99 (MAT␣) strain background using overlap PCR or double joint-PCR strategies followed by biolistic transformation as described previously (32). Genomic DNA of C. neoformans transformants grown on YPD NAT was isolated from cell lysate using acid-washed glass beads (425-600 m, Sigma), and deletion of CnOCH1, CnMNN2, and CnKTR3 was confirmed by PCR (supplemental Fig. 1). For reintegration of wild-type genes into the corresponding mutant strains, genomic DNA fragments of CnOCH1 and CnMNN2 were obtained and introduced into the original genomic locus using pJAF1 (33) with the G418 resistance marker (supplemental Fig. 1).
Preparation of Glycoproteins-To obtain cell wall mannoproteins (cwMPs), C. neoformans cells freshly grown on YPD plates for 2 days were inoculated in 200 ml of YPD and incubated at 30°C for 24 h with shaking at 200 rpm. The cells at the stationary phase (A 600 ϭ 40 -50) were harvested and washed with water. To reduce the possibility of capsular polysaccharide contamination, the washed cells were incubated twice for 30 min at room temperature in an equal volume of DMSO as described previously (34). After centrifugation, the supernatants were decanted, and the cell pellets were washed with water. The washed cells were then resuspended in 0.1 M citrate buffer, pH 7.0, autoclaved at 121°C for 120 min, and centrifuged at 4,000 ϫ g for 10 min at 4°C (35). The supernatants (mixture of crude cell wall proteins and capsular polysaccharides) were recovered, and capsular polysaccharides were precipitated from the supernatants by slowly adding an equal volume of cold ethanol to the supernatants and removed by filtration with a syringe filter (0.45-m, Sartorius Stedim Biotech). After addition of 3 volumes of ethanol to the filtrate and incubation at 4°C overnight, crude cell wall proteins were collected by centrifugation at 4,000 ϫ g for 30 min at 4°C. The dried cell wall proteins were then dissolved with 10 ml of concanavalin A (ConA) binding buffer (20 mM Tris-HCl, pH 7.4, 0.5 M NaCl, 1 mM CaCl 2 , 1 mM MnCl 2 ) and incubated with 1 ml of ConA-Sepharose beads (GE Healthcare) in a column for 2 h with slow rotation. The beads were then washed with 10 ml of ConA binding buffer, and cwMPs were eluted by addition of 5 ml of 1 M methyl-␣-D-mannopyranoside. For isolation of secretory mannoproteins (sMPs), supernatants from 500 ml of YPD cultures incubated for 48 h at 30°C were obtained by centrifugation and filtration using a 0.45-m filter membrane, concentrated, and then exchanged with PBS buffer by tangential flow filtration using a 30-kDa cassette (PXC030C50, Millipore). The supernatants were further concentrated by using an Amicon-15 (30,000 molecular weight cutoff, Millipore). sMPs were purified from the concentrated culture supernatants using ConA, as described for the purification of cwMPs, except that PBS buffer was used instead of Tris-HCl buffer. The eluted glycoproteins were dialyzed with water for 2 days and dried using a vacuum evaporator (Hanil Scientific). The dried glyco-proteins were dissolved in water and quantified by spectrophotometer (NanoDrop, Thermoscientific).
Profiling of N-Linked Oligosaccharides by MALDI-TOF-From purified cwMPs or sMPs (100 -200 g), N-linked glycans were isolated using 3 l of peptide:N-glycanase F (500 units/l, New England Biolabs) and then purified by Carbograph Extract-Clean TM column (150 mg, Alltech). For MALDI-TOF analysis, a matrix solution consisting of 6-aza-2-thiothymine and 2,5-dihydroxybenzoic acid (Bruker Daltonics Inc.) (v/v, 1:1) in 0.25% acetonitrile (Burdick & Jackson) and 0.075% trifluoroacetic acid was mixed with samples of equal volume. The glycan samples were dried and then analyzed using Microflex TM mass spectrometer (Bruker Daltonics Inc.) operated in the reflective positive mode for neutral glycan analysis or in the linear negative mode for acidic glycan analysis. Alternatively, isolated N-glycans were dried, resuspended in 50 l of fresh 1% (w/v) sodium acetate⅐3H 2 O, and then labeled with 100 l of 2-aminobenzoic acid (2-AA) solution (30 mg of 2-AA and 30 mg of NaBH 3 CN in 1 ml of 4% sodium acetate⅐3H 2 O, 2% boric acid, in methanol) at 80°C for 45 min. Labeled N-glycans were purified using a SampliQ Cyano cartridge (100 mg, Agilent) to remove excess 2-AA.
Exoglycosidase Treatment-Purified N-glycans were reacted with 1 l of ␣1,2-mannosidase (␣1,2-MNS, 0.1 milliunit/l, Prozyme) in 20 mM ammonium acetate buffer, pH 5.0, for 12 h at 37°C, and half of the mixture with an additional 1 l of ␣1,2-MNS was further incubated for 12 h. The other half of the ␣1,2-MNS-treated mixture was subsequently reacted with 1 l of ␣1,6-mannosidase (␣1,6-MNS, 40 units/l, New England Biolabs) for 12 h at 37°C. Enzymes were removed using a 10K Microcon (Millipore), dried in a vacuum evaporator (Hanil Scientific), and reconstituted with 3-5 l of water for mass analysis. To determine the presence of xylose residue, 2-AA labeled N-glycans in 20 mM ammonium acetate buffer, pH 5.0, were mixed with 2 l of jack bean ␣-mannosidase (JBM, 150 milliunits/l, Prozyme), incubated for 48 h at 25°C, and then successively treated with 1 l of ␤1,2-xylosidase (from Xanthomonas sp., 20 microunits/l, Calbiochem) for 24 h at 37°C. After removal of exoglycosidases by filtration through a 10K Microcon, glycan profiles were analyzed by normal phase HPLC. Two peaks generated after JBM treatment were fractionated and further analyzed by MALDI-TOF. ␣1,6-Mannosyltransferase Activity Assay-Membrane fractions were obtained as described previously with slight modification (21). Pre-cultured C. neoformans in YPD was inoculated in 500 ml of YPD medium at initial optical density (A 600 ϭ 0.5), and cultured to mid-log phase (A 600 ϭ 5). The cells were harvested by centrifugation (3,000 ϫ g, 10 min), washed with water, and resuspended in 5 ml of PMS buffer (50 mM Tris-HCl, pH 7.5, 5% glycerol, and 2 l/ml protease inhibitor mixture (Sigma)). Aliquots (500 l) were transferred to 1.5-ml microcentrifuge tubes; glass beads were added at half the volume of the cell suspension, and cell lysis was then carried out through 5-10 cycles of alternating 1-min vortexing and 1-min cooling on ice. After ϳ50% of the cells were disrupted (as assessed by microscopy), lysates were centrifuged at 10,000 ϫ g for 20 min. The supernatant was separated, and an equal volume of PMS buffer was added to the pellets. After a second round of cell lysis, the supernatant was removed and added to the first round supernatant. Total supernatant (S1) was further centrifuged at 100,000 ϫ g for 1 h. High speed pellets were collected, resuspended in 100 l of 50 mM Tris-HCl buffer, pH 7.5, plus 5% glycerol, and stored at 4°C. Protein concentrations of high speed pellets were determined using the protein assay reagent (Bio-Rad). ␣1,6-Mannosyltransferase activity was assayed as described previously with slight modification (21). High speed pellets (500 g) were incubated in 100 l of 50 mM Tris-HCl, pH 7.5, buffer containing 2 mM MnCl 2 , 1 mM GDP-mannose, 0.5 mM 1-deoxymannojirimycin, and 0.1 g of Man 8 GlcNAc 2 -AA (Prozyme) as an acceptor at 30°C overnight. The reaction mixture was filtered through a Microcon (YM-10, Millipore), and the filtrate was analyzed by HPLC. To identify linkage, the Golgi reaction mixture was treated with ␣1,2-MNS.
N-Glycan Analysis by HPLC-Normal phase HPLC was conducted using an Asahipak NH2P-50 4E column (0.46 ϫ 25 cm, 5 m, Shodex) at a rate of 1.0 ml/min with Solvent A (100% acetonitrile) and Solvent B (50 mM ammonium formate in water, pH 4.4). The column was equilibrated with a solution containing 50% Solvent A and 50% Solvent B. After sample injection, the proportion of Solvent B was increased in a linear fashion to 68% for 50 min. 2-AA-oligosaccharides were analyzed with a Waters HPLC system composed of 515 dual pumps, a 717 plus autosampler, and a 2475 fluorescence detector with excitation and emission wavelengths of 360 and 425 nm, respectively. Data were collected using Empower TM 2 chromatography data software (Waters).
For acidic glycan analysis, the column was equilibrated with a solution containing 90% Solvent C (2% acetic acid, 1% tetrahydrofuran in acetonitrile) and 10% Solvent D (5% acetic acid, 3% triethylamine, 1% tetrahydrofuran in water). After sample injection (50 l), the proportion of Solvent D was increased in a linear fashion up to 90% for 60 min. 2-AA-oligosaccharides were detected using the same Waters HPLC system described above. After fractionation, each sample was dried and dissolved in water. The peaks separated on HPLC were further identified by MALDI-TOF mass spectrometry using the linear negative mode. All chemicals were purchased from Sigma unless mentioned otherwise.

RESULTS
Neutral N-Glycan Structures of C. neoformans-To obtain general information on the structure of neutral N-glycans of the Cryptococcus species complex, MALDI-TOF analysis was carried out in the positive mode with N-linked oligosaccharides assembled on cwMPs derived from serotype A strain H99 (␣-mating type) (Fig. 1A). A series of major peaks observed at m/z of 1258.1, 1420.1, 1582.1, 1744.1, 1906.0, 2067.9, 2230.1, and 2392.1 were matched well to high mannose type N-glycans ranging from Hex 5 HexNAc 2 to Hex 12 HexNAc 2 (H 5-12 ) , originating from the ascomycetous yeast and fungal species (glycan structure database of the Consortium for Functional Glycomics). In contrast, other major peaks with a mass value 30.0 m/z smaller than those of Hex n HexNAc 2 could not be assigned to any reported glycans. Considering that the mass difference of each peak corresponds to the molecular weight of C 1 H 2 O 1 , it was speculated that the unassigned glycan structures might be oligosaccharides containing a pentose residue (P 1 H 5-20 , Pen 1 Hex 5-20 HexNAc 2 ).
Given that most fungal N-glycans are of high mannose type, we predicted that N-linked oligosaccharides of C. neoformans would also be mostly constituted of mannose residues. As expected, digestion of N-glycans from serotype A strain H99 with ␣1,2-MNS resulted in the convergence of most peaks to H 5-6 and P 1 H 6 -8 , indicating that the outer chains of cryptococcal N-glycans were mostly extended by ␣1,2-mannose residues without a terminal ␣1,3-mannose cap (Fig. 1B). Subsequent digestion with ␣1,6-MNS resulted in the shift of H 6 to H 5 and P 1 H 6 to P 1 H 5 , although the peak of Pen 1 Hex 8 GlcNAc 2 (P 1 H 8 ) was not shifted (Fig. 1C). The P 1 H 8 glycan seems to be an incompletely digested product that was derived from a fraction of glycans lacking ␣1,6-mannose extension, because it was converted to P 1 H 5 by additional prolonged digestion with ␣1,2mannosidase (data not shown). These results indicated that the outer chains contained a single ␣1,6-mannose extension. The profiles of N-linked oligosaccharides obtained from sMPs of the H99 strain also showed the same pattern of H 5-12 glycans containing one residue of pentose as observed in those from cwMPs (Fig. 1D). We also analyzed glycan profiles of the C. neoformans cac1⌬ mutant, which has a deletion of the adenylyl cyclase gene and is deficient in capsule production (36), and confirmed the same profile pattern (Fig. 1E) thus excluding the possibility of capsule oligosaccharide contamination in the preparation of N-linked glycans. Exoglycosidase treatment of glycan samples from the Cncac1⌬ mutant also showed the same result as the wild type (data not shown). The findings strongly suggested that the N-glycans of serotype A strain were extended with mostly ␣1,2-linked mannose residues and a single ␣1,6-linked mannose residue and were further modified with a pentose residue bound to the inner region of the N-glycan.
Comparison of Neutral N-Glycan Profiles among Different Serotype Strains-Next, we examined whether the neutral N-glycan profiles of C. neoformans vary depending on mating type or serotypes. It was previously reported that phenotypic variants of one C. neoformans strain differ with respect to virulence and the arrangement of xylose moieties within the GXM of the capsule (2). Mating type has also been implicated as a virulence factor in C. neoformans. Epidemiological studies have shown that clinical isolates mostly have the ␣-mating type (96% average) (37). In addition, ␣ strains were reported as more virulent than congenic a strains in serotype D but not in serotype A (38).
There were no apparent differences found when the N-glycan profiles of strain H99 were compared with mating type ␣ or when congenic strain KN99 was compared with mating type a (Fig. 2, A and B), indicating that glycan profiles were not affected by mating type. The N-glycans of serotype D strains, JEC21 (mating type ␣) and JEC20 (mating type a), were shown to be shorter than those of serotype A; however, they also contained a pentose residue (H 5-10 and P 1 H 5-14 ) (Fig. 2, C and D). In contrast, the N-glycans of C. gattii serotype B strains, R265 and WM276 (both mating type ␣ but different molecular types), consisted of H 5-14 without a pentose residue, similar to those of other yeast species in general (Fig. 2, E and F). The N-glycans of JEC21 and CgR265 were shifted to P 1 H 5 and H 5 , respectively, by serial treatment with ␣1,2and 1,6-MNSs, indicating their core N-glycans were also extended by addition of ␣1,2and ␣1,6-linked mannose residues even though the extension is quite limited in serotypes B and D strains (supplemental Fig. 2, A and B). These results strongly suggested that cryptococcal N-glycans displayed serotype-specific differences not only in length but also in the presence of pentose.
Identification of a Pentose Residue as Xylose-In Cryptococcus, xylose is a sole pentose residue that is incorporated as a component of GXM, glucuronoxylomannogalactan, glucosylinositol phosphorylceramides, and O-glycans (39). The possible presence of xylose in N-glycans of C. neoformans has been implicated by a previous report showing that the cryptococcal laccase 1 protein (Lac1) from serotype D strain contains 4 mol of glucosamine and 22 mol of mannose/xylose/mol of protein (27). To determine whether the pentose attached to the inner region of the N-glycans was a xylose residue, 2-AA-labeled N-glycans fractionated in HPLC using an amine column were digested by treatment with JBM, which cleaves ␣1,2/3/6-linked mannoses. After JBM treatment, the N-glycans of serotypes A and D (Fig. 3A, panels a and d) converged to two peaks (X and Y, Fig. 3A, panels b and e). Then the subsequent ␤1,2-xylosidase treatment of the JBM-digested N-glycans resulted in an increased intensity of peak Y (Fig. 3A, panels c and f). In the case of the serotype B, N-glycans were already shifted to a single peak (Y, Fig. 3A, panels g and h) by treatment with JBM only. Peaks X and Y from JMB-treated N-glycans of three serotype strains were fractionated and analyzed for mass by MALDI-TOF. As shown in Fig. 3B, peaks X and Y corresponding to m/z of 862.6 and 730.3, respectively, were assigned as sodium adducts of P 1 H 1 -AA and H 1 -AA, respectively. Based on the data showing that a pentose residue was still retained even after removal of ␣1,2/3-mannoses from the trimannosyl core of N-glycans and that the pentose residue was removed by ␤1,2xylosidase treatment, we speculated that P 1 H 1 and H 1 correspond to Xyl 1 Man 1 GlcNAc 2 and Man 1 GlcNAc 2 , respectively. Thus, it is highly likely that a single xylose residue is attached to the first mannose residue of the trimannosyl core of N-glycans.
Moreover, we constructed a Cnuxs1⌬ mutant strain with a defect in the synthesis of UPD-xylose (UDP-Xyl) and compared its N-glycan profile to that of the wild-type strain. It was previously reported that the addition of xylose to the capsular polysaccharides GXM and glucosylinositol phosphorylceramides is defective in the Cnuxs1⌬ mutant (40). As seen in Fig. 3C, the N-glycans from cwMPs of the Cnuxs1⌬ mutant showed only high mannose-type N-glycans without pentose residues, strongly supporting the idea that the pentose residue was xylose. These results also indicated that UDP-Xyl was utilized as a sugar donor not only for capsule biosynthesis but also for N-glycan biosynthesis of C. neoformans through the classical protein secretory pathway.
In Silico Analysis of C. neoformans N-Glycan Biosynthesis Pathway-To reconstruct the N-glycan outer chain biosynthesis pathway of C. neoformans, we aimed to identify C. neoformans genes orthologous to S. cerevisiae genes participating in the N-glycan biosynthesis pathway ( Table 1). The in silico analysis based on the genome database of C. neoformans serotype A (H99) and serotype D (JEC21) suggested that several Golgi glycosyltransferase genes involved in processing the outer chain of N-glycans in S. cerevisiae were missing in C. neoformans. For example, C. neoformans does not have orthologs for M-pol I and II complex formation, such as VAN1, ANP1, MNN10, or MNN11, which are responsible for elongation of the ␣1,6-mannose backbone of outer chains. Furthermore, C. neoformans did not appear to contain orthologs of MNN4 responsible for mannosyl phosphorylation and MNN1 for addition of the terminal ␣1,3-mannose residue in S. cerevisiae glycans, indicating a different structure of the cryptococcal N-glycan outer chains.
Yeast and fungus-specific outer chain biosyntheses are initiated by Och1p having an ␣1,6-mannosyltransferase activity in the Golgi. In S. cerevisiae, Hoc1p (homologous to Och1p) resides in the M-pol II complex, although its function has not been defined (11). We identified three Cryptococcus genes encoding proteins homologous to yeast Och1p (CNAG_00744.2, CNAG_05836.2, and CNAG_01214.2) from the H99 genome database, designated CnOCH1, CnHOC1, and CnHOC2, respectively. We also found a single gene (CNAG_06782.2, designated CnMNN2) homologous to yeast MNN2/MNN5 genes encoding ␣1,2-mannosyltransferase and a single gene (CNAG_03832.2, designated CnKTR3) homologous to the S. cerevisiae KRE2/MNT1 family containing nine members of Golgi mannosyltransferases involved in both N-and O-linked glycan synthesis (13). The bioinformatics analysis indicated that the glycosylation pathway of C. neoformans might be simpler with a fewer number of components than those of other yeast species, which have larger protein families playing functionally redundant or inactive roles in glycosylation.

Functional Characterization of C. neoformans Och1p
Homologs-To analyze the function of the C. neoformans Och1p homologs, we constructed gene deletion mutants for each homolog (Cnoch1⌬, Cnhoc1⌬, and Cnhoc2⌬) and examined their neutral N-glycans structures by MALDI-TOF mass spectrometry. As seen Fig. 4A, N-glycans from the Cnoch1⌬ mutant (Fig. 4A, panel b) were noticeably much shorter than those of the H99 wild-type strain (Fig. 4A, panel  a), although no apparent changes were detected in N-glycan profiles from the mutant Cnhoc1⌬ and Cnhoc2⌬ strains (Fig.  4A, panels c and d). In the Cnoch1⌬ mutant, N-glycans larger than Xyl 1 Man 10 GlcNAc 2 (X 1 M 10 ) were hardly detected, but reintroduction of the wild-type CnOCH1 gene into the Cnoch1⌬ strain restored its N-glycan profile to that of the H99 strain (Fig. 4A, panel e), thus verifying the involvement of CnOch1p in the N-glycan processing of C. neoformans.
To examine which step of the outer chain biosynthesis pathway was affected by disruption of the CnOCH1 gene, N-glycans obtained from the Cnoch1⌬ mutant strain were sequentially treated with ␣1,2and ␣1,6-MNSs (Fig. 4B). After treatment with ␣1,2-MNS, most glycan peaks were converted to Man 5-6 GlcNAc 2 (M 5-6 ) or Xyl 1 Man 5-6 GlcNAc 2 (X 1 M 5-6 ) with M 5 and X 1 M 5 as major peaks (Fig. 4B, panel b). However, subsequent treatment with ␣1,6-MNS in the Cnoch1⌬ mutant did not generate any further changes in the N-glycan profiles (Fig. 4B, panel c), whereas the treatment with ␣1,6-MNS in the  wild-type strain did result in further shifts of X 1 M 6 and M 6 to X 1 M 5 and M 5 , respectively (Fig. 1C). The exoglycosidase ␣1,6-MNS can remove only linear ␣1,6-mannose residues at the nonreducing end. Because the presence of a branched ␣1,3mannose in the core N-glycans inhibits the removal of the 1,6mannose, the ␣1,6-MNS treatment of N-glycans from the wildtype strain would convert X 1 M 6 and M 6 to X 1 M 5 and M 5 , respectively. Thus, no shift of any N-glycan peak after the ␣1,6-MNS treatment of N-glycans from the Cnoch1⌬ mutant strain strongly supported the idea that the outer chains of N-glycans in the Cnoch1⌬ mutant do not possess ␣1,6-linked mannose residues because of the loss of CnOch1p function. We further confirmed that Cryptococcus Och1p is a functional homolog of S. cerevisiae Och1p by analyzing its complementation capacity to recover the defective phenotype of the S. cerevisiae och1⌬ mutant (supplemental Fig. 3A). Heterologous expression of CnOCH1, but not of CnHOC1 or CnHOC2, recovered the N-glycosylation defect of secreted invertase as well as the decreased resistance to hygromycin B in the Scoch1⌬ mutant (supplemental Fig. 3B). These results strongly suggest that CnHoc1p and CnHoc2p are not functional homologs of ScOch1p, despite sequence similarities between them. Further investigation will be necessary to determine whether they are functional orthologs of ScHoc1p or play Cryptococcus-specific functions. Moreover, we also observed decreased additional activity of an ␣1,6-mannose residue to the M 8 core form oligosaccharide in the och1⌬ mutant of C. neoformans (Fig. 4C).
These results further support that the Cryptococcus OCH1 gene is a functional ortholog of the S. cerevisiae OCH1 gene encoding ␣1,6-mannosyltransferase, which initiates outer chain branching of N-glycans in the Golgi by addition of a single ␣1,6-linked mannose residue to the Man 8 GlcNAc 2 core. Glycan Profile Analysis of Cnmnn2⌬ and Cnktr3⌬ Strains-To examine the function of the CnMNN2 and CnKTR3 genes in cryptococcal N-glycan biosynthesis, N-glycan profiles of the Cnmnn2⌬ and Cnktr3⌬ mutants were analyzed (Fig. 5A, panels  b and c). N-Glycans of the Cnmnn2⌬ strain were much shorter than those of the wild-type strain, consisting mostly of M 6 -9 and X 1 M 6 -10 , but were slightly longer than those of the Cnoch1⌬ mutant (Fig. 5A, panels a and b). Reintroduction of the CnMNN2 gene into the Cnmnn2⌬ strain recovered its glycan profile to that of the wild-type CnH99 strain (Fig. 5A, panel  d). When treated with ␣1,2-MNS, N-glycans of the Cnmnn2⌬ mutant were shifted to M 5-6 and X 1 M 5-7 (Fig. 5B, panel b), and subsequent treatment with ␣1,6-MNS generated peaks of M 5 and X 1 M 5 as major forms (Fig. 5B, panel c). Compared with the ␣1,2-MNS-treated N-glycans from the Cnoch1⌬ mutant, those from the Cnmnn2⌬ mutant were larger by a single hexose residue, which was removed by subsequent ␣1,6-MNS treatment. These results suggest that CnMnn2p mediates the addi-tion of ␣1,2-mannose residues after the first step of ␣1,6-mannose addition to the core N-glycan by CnOch1p in C. neoformans. In contrast to the notable defects in the elongation of outer chains in Cnoch1⌬ and Cnmnn2⌬ mutants, the N-glycan profile of the Cnktr3⌬ mutant was almost identical to that of the wild-type strain (Fig. 5A, panel c), indicating that CnKtr3p is not involved in the processing of N-glycans in C. neoformans. Interestingly, however, we observed a significant defect in O-glycan biosynthesis in the Cnktr3⌬ mutant (supplemental Fig. 4), suggesting a possible role of CnKTR3 in O-glycan elongation but not in N-glycan processing in C. neoformans.
Presence of Acidic N-Linked Oligosaccharides Containing Xylose Phosphate-A recent study on xylosylphosphotransferase 1 (Xpt1p) of C. neoformans reported that this enzyme catalyzes O-linked glycosylation of proteins by adding xylose phosphate to O-glycans using UDP-Xyl as a reaction donor (39). To investigate the possible presence of xylose phosphate residues in cryptococcal N-glycans, total N-glycan profiles, including both neutral and acidic glycans, were analyzed by HPLC using an amine column at an acidic pH, which can separate oligosaccharides based on their charges and sizes. Interestingly, all N-glycans of the wild-type H99 strain (serotype A) could be separated into four major peaks (groups 1-4) (Fig. 6A,  panel a). Notably, Cnoch1⌬, Cnmnn2⌬, and Cnuxs1⌬ mutants showed distinctive profiles compared with the wild-type strain. The peak assigned to group 3 was missing in Cnoch1⌬ and Cnmnn2⌬ mutants (Fig. 6A, panels b and c), and peaks corresponding to groups 2-4 were not detected in Cnuxs1⌬ mutant (Fig. 6A, panel d). Each group exhibited a similar HPLC profile with an amide column separating oligosaccharides based mainly on sizes (supplemental Fig. 5).
Further analysis using MALDI-TOF in the negative reflector mode revealed that group 1 and groups 2-4 in H99 were composed of neutral and acidic N-glycans, respectively (Fig. 6B). Group 1 consisted of neutral N-glycans with or without a xylose residue. In contrast, N-glycans from group 2 in the wild-type strain were composed of Ph 1 P 1-2 H 5-20 glycans containing an additional pentose phosphate residue with or without a xylose residue (Fig. 6B, panel b). Group 3 consisted of large glycans with a pentose phosphate residue (Ph 1 P 1 H 12-20 ) and group 4 mostly consisted of a core form of N-glycans containing a pentose phosphate residue (Ph 1 P 1 H 5-10 ) as the major portion (Fig.  6B, panels c and d). In particular, the disappearance of only group 3 in both Cnoch1⌬ and Cnmnn2⌬ mutants indicated that the glycans in group 3 were large N-glycan species containing a single pentose phosphate on the outer chains extended by addition of ␣1,6and ␣1,2-mannoses. Therefore, the absence of all peaks of groups 2-4 corresponding to acidic glycans in the Cnuxs1⌬ mutant (Fig. 6D, panel d) strongly supported that the pentose phosphate residues found in the acidic glycans were xylose phosphate, which appeared to be added to both the core and outer regions of N-glycans in C. neoformans. The absence of xylose phosphate in N-glycans from the Cnuxs1⌬ mutant also suggested that UDP-Xyl is also used as a donor to add xylose phosphate to N-glycans assembled on proteins.
We observed very small peaks corresponding to groups 2 and 4, and no peak for group 3, in the total N-glycan profiles of serotypes B and D (supplemental Fig. 6A), implying the presence of acidic N-glycans in the core form of N-glycans of sero-types B and D. However, the proportion of acidic glycans in serotypes B and D was much smaller (less than 10%) compared with those of serotype A, which contained more than 50%. The lack of a group 3 peak was consistent with the much shorter length of outer chains of N-glycans from serotypes B and D. Acidic N-glycans from serotypes B and D were also shown to consist of Ph 1 P 1 H 5-11 containing an additional pentose phosphate residue (supplemental Fig. 6, B and C), suggesting that C. neoformans N-glycans are modified by addition of a xylose phosphate regardless of the serotypes. Notably, an interesting observation is the presence of acidic glycans carrying a phosphate moiety without xylose, particularly in N-glycans from serotype B.
Growth Phenotypes of C. neoformans Mutant Strains Defective in Glycan Biosynthesis-In general, yeast mutant strains with defects in glycosylation show alteration in cell wall integrity and thus exhibit increased sensitivity to cell wall-disturbing reagents and high temperatures (41). We assessed the growth phenotypes of the C. neoformans mutants Cnoch1⌬, Cnhoc1⌬, Cnhoc2⌬, Cnmnn2⌬, Cnktr3⌬, and Cnuxs1⌬ in the presence of diverse cell wall-and membrane-disturbing reagents, including SDS, Calcofluor white, Congo red, sodium orthovanadate, and hygromycin B, and high temperature. All mutants grew as well as the wild-type strain under normal unstressed conditions on YPD at 30°C (Fig. 7). Unexpectedly, however, despite a significant alteration in the structure of outer chain N-glycans, the Cnoch1⌬ and Cnmnn2⌬ mutants showed no apparent change. In contrast, the Cnktr3⌬ mutant, which appeared to have a defect in O-glycosylation but not in N-glycosylation, and the Cnuxs1⌬ mutant with a defect in UDP-Xyl synthesis, displayed increased sensitivity to SDS and high temperature at 39°C, indicating that CnKtr3p and CnUxs1p are required for maintenance of cell wall stability.
The capsule of C. neoformans is intimately associated with the cell wall, which underlies the capsule and provides yeast mechanical strength under stressful conditions. However, we did not observe any defects in capsule formation in Cnoch1⌬, Cnhoc1⌬, Cnhoc2⌬, Cnmnn2⌬, or Cnktr3⌬ mutant strains (data not shown). These results imply that the truncated outer chain structure in N-glycans might be tolerable for maintaining the cell wall integrity of C. neoformans, different from other yeast species such as S. cerevisiae and C. albicans with hypermannosylated outer chains (8). Similarly, it was also reported that the outer chain structure of N-glycans is not important for maintenance of cell wall integrity in the filamentous fungi A. fumigatus (25).

DISCUSSION
N-Glycosylation, the most common type of eukaryotic protein glycosylation, involves the linkage of an oligosaccharide core to Asn residues in the ER, which is well conserved from yeast to mammals. However, subsequent processing of N-glycans is significantly different among different organisms and even in yeast species, generating diversity in N-glycan structures. Although N-glycans in yeasts are generally extended by addition of mannose to the core oligosaccharide in the Golgi, additional monosaccharide units such as galactose and N-acetylglucosamine (GlcNAc) are added in some species (11). N-Glycans of several yeast species also contain acidic sugars, which are composed of mannosyl phosphorylated sugars in most yeast species such as S. cerevisiae, C. albicans, P. pastoris, and Y. lipolytica. However, in S. pombe, the addition of pyruvate generates acidic glycans (11,43).
Historically, C. neoformans and its related species, such as C. gattii, have been further categorized by serotypes based on a defined set of capsular-reactive immune sera (44). C. gattii includes serotypes B and C, whereas strains classified as serotypes A, D, or AD hybrids make up C. neoformans. Serotypes A and D have been also classified as varieties of C. neoformans, var. grubii and var. neoformans, respectively (45). Serotypes A and D generally associated with diseases in immunocompromised individuals are distributed worldwide, whereas serotypes B and C known to infect immunocompetent persons are typically found in, but not limited to, tropical and subtropical regions. The number of xylose residues on the major repeat unit of capsular polysaccharides exhibits a serotype-specific pattern: 2:3:4:1 ϭ serotype A, serotype B, serotype C, and serotype D (46).
In this study, we performed comparative N-glycan profile analysis of several C. neoformans serotype strains, including C. neoformans var. grubii H99 and KN99 (serotype A), C. neoformans var. neoformans JEC21 and JEC20 (serotype D), and C. gattii R265 and WM276 (serotype B). We report, for the first time, the serotype-specific presence of a ␤1,2-xylose residue in high mannose type N-glycans of cryptococcal glycoproteins. Unexpectedly, serotype B strains containing the higher number of xylose residues in their capsule compared with serotypes A and D showed only high mannose-type N-glycans without xylose (Figs. 2 and 3). It is speculated that the differences in N-glycan structures among the serotypes tested in this study might reflect evolutionary divergence.
Interestingly, we also present data strongly indicating that C. neoformans N-glycans also contain xylose phosphate residues as acidic sugars ( Fig. 6 and supplemental Fig. 6). The relative proportion of acidic sugars among total N-glycans was more than 50% in serotype A, much higher than the reported proportion of less than 1% of total acidic O-glycans (39). It is noticeable that the portion of acidic sugars was much lower in serotypes B and D, compared with that of serotype A. N-Glycans from cwMPs of the Cnuxs1⌬ mutant contained high mannose-type oligosaccharides without any xylose and xylose phosphate residues. Based on our data, it is conceivable that UDP-xylose is used as a substrate not only for xylosylation but also for xylosyl phosphorylation of various glycoconjugates, including N-glycans. The physiological role of acidic glycans in yeast species has been implicated in the stress response to environmental changes. However, it has been reported the mannosyl phosphorylation is not required for macrophage interactions or for virulence in C. albicans, despite significant loss of ␤1,2-mannose oligosaccharides (47).
The in silico analysis indicated that C. neoformans might have a simpler N-glycan outer chain biosynthetic pathway. As expected from the lack of genes encoding M-pol I and M-pol II subunits, our data confirmed that the outer chains of N-glycans from C. neoformans have a very short ␣1,6-mannose extension, consisting mostly of a single ␣1,6-mannose residue, and are extended mainly by ␣1,2-mannose residues. Based on the structural information of N-glycans in C. neoformans mutants constructed in this study, we propose the N-linked outer chain biosynthetic pathway in C. neoformans as shown in Fig. 8. Sim- ilar to S. cerevisiae, the Man 8 GlcNAc 2 core glycan attached to the cryptococcal proteins in the ER is also elongated via the addition of an ␣1,6-mannose unit by CnOch1p, an initiating ␣1,6-mannosyltransferase in the Golgi. However, the ␣1,6 outer chain backbone is not further elongated in C. neoformans. Instead, the outer chain and core N-glycans are further elongated mainly via ␣1,2-mannose addition mediated by CnMnn2p. Moreover, although the exact order is not yet clear, a single ␤1,2-xylose residue is added to the first mannose at the trimannosyl core of N-glycan during the early processing stages in the Golgi. As in the case of the xylose addition to C. neoformans N-glycans, the modification of N-glycans by bisecting GlcNAc at the ␤-mannose of the N-glycan core was previously reported, and the corresponding transferase has been characterized in the basidiomycete Coprinopsis cinerea (10). Then, similar to the addition of mannosyl phosphate in S. cerevisiae, the xylose phosphate might be added to some N-glycans at the core and outer chain regions, generating acidic N-glycans with negative charges during the late processing stages. Considering that the addition of a xylose phosphate residue occurs at the late stage of N-glycan processing in C. neoformans, it is highly likely that a transporter for UDP-Xyl would be present at the Golgi membrane. There has been a report on the Golgi localization of a human transporter for UPD-Xyl, which can transport UPD-Xyl over the Golgi membrane (48). However, we could not exclude the possibility that a transporter for UPD-xylose could be also present in the ER. As indicated by the absence of a MNN1 ortholog in C. neoformans, our structural analysis data supported the idea that C. neoformans N-glycans do not undergo the final modification step to add terminal ␣1,3-linked mannoses to the outer chains.
In Cryptococcus, Cxt1p, a ␤1,2 xylosyltransferase, uses UDP-Xyl as a donor for the incorporation of xylose as a component of capsular polysaccharides and glucosylinositol phosphorylceramides (49). However, we observed that the Cncxt1⌬ mutant displayed the same N-glycan profile as that of the wild-type strain (data not shown), excluding the possibility that Cxt1p may act solely to transfer xylose to mannose residues during N-glycan biosynthesis. In fact, we identified several Cxt1 homologs from the C. neoformans genome database. Moreover, in our preliminary analysis of N-glycan profiles in the xpt1⌬ strain, which is defective in adding xylose phosphate to O-glycans due to lack of Xpt1p, we could still detect the presence of acidic glycans. 4 This indicates the presence of unknown xylosyltransferase(s) and xylosylphosphotransferase(s) specific for N-glycan biosynthesis or the involvement of other redundant enzymes in the addition of xylose and xylose phosphate residues to N-glycans in C. neoformans. In this study, we report the comprehensive information on the structure and biosynthesis pathway of C. neoformans N-glycome. However, our experimental approach based on the enrichment of the mannosecontaining structures by using ConA might miss a fraction of glycans that are not bound by ConA. Also, the release of N-glycans using peptide:N-glycanase F from cryptococcal mannoproteins might impose a bias on the structures analyzed, because the modification of the core GlcNAc by ␣1,3-linked fucose prevents the action of peptide:N-glycanase F. Thus, the possibility still remains that cryptococcal N-glycans with yet uncharacterized structures might be identified by further systematic analysis.
Cell surface mannoproteins contain both O-and N-linked oligosaccharides. Most information about glycosylation of proteins has been generated through studies of the model yeast S. cerevisiae. Very recently, information assessing the relevance of glycosylation for virulence has emerged for medically important fungal pathogens, including C. albicans, A. fumigatus, and C. neoformans. Previous studies have demonstrated that O-glycosylation is required for host-fungus interactions and virulence in all strains of C. albicans, A. fumigatus, and C. neoformans (50). However, the roles of N-glycan outer chains for host-fungus interactions and virulence were shown to be quite different between C. albicans and A. fumigatus. The Caoch1 null mutant with loss of the ␣1,6-linked polymannose backbone was shown be attenuated in virulence. However, similar infection experiments revealed no difference between the afoch1 null mutant and control strains, suggesting that N-glycan outer chains of A. fumigatus are not associated with virulence (25). To address whether the modification of N-glycan outer chains affects the virulence of C. neoformans, we tested the virulence of the Cnoch1⌬ mutant, which is defective in initiating extension of N-glycan outer chains, using a murine model of systemic cryptococcosis. Although the Cnoch1⌬ mutant showed slightly attenuated virulence compared with the wild-type strain, the Cnoch1⌬/CnOCH1 complemented strain was as virulent as the Cnoch1⌬ mutant (data not shown), suggesting that the function of CnOch1p is not critical for virulence in C. neoformans. Our data also strongly suggest that perturbation of outer chain processing of N-glycans in C. neoformans will not significantly affect virulence. For example, the N-glycans of serotype B appeared to be mostly core forms without extended outer chains or the addition of xylose. Moreover, they were shown to be much less modified with the addition of xylose phosphate compared with those of serotype A. Despite such noticeable differences in the outer chain N-glycans, serotype B can infect even immunocompetent persons.
In conclusion, we present evidence supporting the idea that C. neoformans N-glycans are high mannose type modified with addition of a ␤1,2-xylose in a serotype-specific way. Moreover, our data strongly indicate the presence of differential modification by addition of xylose phosphate. It remains a challenge to identify the glycosyltransferases responsible for these unique modifications of N-glycans with xylose and xylose phosphate in C. neoformans. It would be intriguing to elucidate regulatory mechanisms underlying the serotype-specific processing of cryptococcal N-glycans. Furthermore, we expect that the glycosylation-defective mutant strains developed in this study will be useful for systematic investigation on how structural alterations of N-/O-glycans affect the intensity of virulence and the extent of host immunological interactions in C. neoformans.