A β-1,2-Xylosyltransferase from Cryptococcus neoformans Defines a New Family of Glycosyltransferases*

Cryptococcus neoformans is an opportunistic fungal pathogen characterized by a prominent polysaccharide capsule that envelops the cell. Although this capsule is dispensable for in vitro growth, its presence is essential for virulence. The capsule is primarily made of two xylose-containing polysaccharides, glucuronoxylomannan and galactoxylomannan. There are likely to be multiple xylosyltransferases (XTs) involved in capsule synthesis, and the activities of these enzymes are potentially important for cryptococcal virulence. A β-1,2-xylosyltransferase with specificity appropriate for capsule synthesis was purified ∼3000-fold from C. neoformans, and the corresponding gene was identified and cloned. This sequence conferred XT activity when expressed in Saccharomyces cerevisiae, which lacks endogenous XT activity. The gene, termed CXT1 for cryptococcal xylosyltransferase 1, encodes a 79-kDa type II membrane protein with an N-linked glycosylation site and two DXD motifs. These latter motifs are believed to coordinate divalent cation binding in the activity of glycosyltransferases. Site-directed mutagenesis of one DXD motif abolished Cxt1p activity, even though this activity does not depend on the addition of a divalent cation. This may indicate a novel catalytic mechanism for glycosyl transfer. Five homologs of Cxt1p were found in the genome sequence of C. neoformans and 34 within the sequences of other fungi, although none were found in other organisms. Many of the homologous proteins are similar in size to Cxt1p, and all are conserved with respect to the essential DXD motif. These proteins represent a new family of glycosyltransferases, found exclusively within the fungal kingdom.

Our current understanding of capsule biosynthesis is limited. By screening mutagenized C. neoformans cells for defects in capsule formation, Kwon-Chung, Janbon, and their co-workers (16 -21) identified a number of genes that appear to be involved in capsule synthesis. Disruption of any one of the four CAP genes (CAP10, CAP59, CAP60, and CAP64) yields an acapsular phenotype, suggesting that these each play a central role in capsule synthesis (16 -19). Five homologs of CAP10 (CAP1-5) have also been identified, but the biochemical functions of the CAP genes and these homologs have remained undefined (22).
Glycosyltransferases catalyze the specific transfer of a monosaccharide from an activated donor, such as a nucleotide disphosphosugar, to an acceptor. These enzymes are typically 30 -50-kilodalton type II membrane proteins (23). Glycosyltransferases have been organized into families based on sequence similarity (24) and can be broadly divided into two superfamiles (GT-A and GT-B) based on structural patterns. Most members of the GT-A group contain a DXD motif that is involved in the binding of a divalent cation cofactor important for catalytic activity (25)(26)(27)(28)(29)(30)(31). Although a few GT-A enzymes lack the DXD motif and cation requirement, this is more typical of the GT-B glycosyltransferases.
The synthesis of large glycans such as GXM or GalXM generally requires the sequential action of several glycosyltransferases. Thus, it is likely that a number of these enzymes, including up to seven xylosyltransferases, are actively involved in capsule synthesis. It is also reasonable to suggest that some of the previously identified capsule synthesis associated genes may encode glycosyltransferases, especially as one of them, CAP59, encodes a homolog to a known mannosyltransferase (32). However, no glycosyltransferase with a defined role in capsule synthesis has been identified.
Capsule synthesis offers a potential target for antifungal chemotherapy. To understand this process, we must identify the glycosyltransferases involved in GXM and GalXM production. Because of the importance of xylose residues within these structures, we have focused on identifying xylosyltransferases (XTs) involved in capsule biosynthesis. As no homologs to known mammalian or plant XTs exist within the C. neoformans data base, we took a biochemical approach to this question. Here we describe the discovery, purification, characterization, and expression of Cxt1p, a ␤-1,2-XT from C. neoformans with activity appropriate for the synthesis of either GXM or GalXM. Cxt1p is a large, apparently cation-independent glycosyltransferase that defines a new family of glycosyltransferases. This family includes C. neoformans Cap10p and its homologs and is exclusive to fungi.

EXPERIMENTAL PROCEDURES
Materials-Unless otherwise noted, all chemicals were from Sigma.
Strains and Cell Growth-C. neoformans wild-type strain JEC21 (serotype D MAT␣) was provided by Joseph Heitman (Duke University). C. neoformans ags1⌬ strain was generated in our laboratory (serotype D MAT␣ ags1⌬) (33), and Saccharomyces cerevisiae strain TDY172 (RSY620; MATa ade2-1 trp1-1 leu2-3,112 ura3-1 his3-11,15 pep4::TRP1) was from Randy Schekman (University of California, Berkeley). For activity analysis, 50 ml of minimal medium minus uracil (URAϪ) (34) was inoculated with a single colony of JEC21 or ags1⌬ and incubated at 30°C with continuous shaking (200 rpm) until the A 600 was between 1 and 2. For purification, such a culture of ags1⌬ was then inoculated into 3 liters of the same medium and grown at 30°C with continuous shaking for 3-4 days until the A 600 was between 4 and 5. TDY172 was grown at 30°C with continuous shaking in 50 ml of minimal broth media supplemented with uracil to an A 600 of 1-2.
Xylosyltransferase (XT) Enzyme Assays-Activity was assayed by monitoring the transfer of [ 14 (15 nCi), and 100 mM Tris, pH 6.5. The reaction was incubated for 4 h at 20°C and terminated by applying the assay mixture to a small disposable column containing 0.6 ml of AG2X-50 resin (Bio-Rad), followed by 30 l of deionized water to ensure the sample was fully loaded into the resin. 600 l of deionized water was then applied to the column, and the eluate collected and spun at 13,000 ϫ g for 5 min to remove particulate materials, and the supernatant fraction removed to another tube. 14 C-Labeled trisaccharide product was detected either by scintillation counting or TLC. For the latter, all or part of the supernatant was dried under N 2 at 50°C, resuspended in 15 l of 40% n-propyl alcohol, and applied to a dried (70°C for 30 min) 20 ϫ 20-cm Silica Gel-60 TLC plate (EMD Chemicals, Gibbstown, NJ). The plate was developed in 5:4:2 n-propyl alcohol:acetone:water for 2 h, dried, and then developed for an additional 2 h in the same solvent. After drying, the standards were visualized by spraying with 0.2% orcinol in 75:15:10 ethanol:sulfuric acid:water and incubating for 5-10 min at 70°C. The sample lanes were sprayed with Enhance surface autoradiography spray (PerkinElmer Life Sciences) and allowed to dry, and the radioactive products were visualized by autoradiography.
Production and Purification of Trisaccharide Product for Analysis-To generate product for structural analysis, XT reactions were scaled up. Ten 70-l reactions (20 l of partially purified enzyme (post Sephacryl S-300, see below), 12.5 mM ␣-1,3-Man 2 (0.88 mol), 12.8 mM UDP-xylose (0.9 mol) (Carbosource Services, Athens, GA), 57 nmol of UDP-[U-14 C]xylose (0.015 Ci), and 147 mM Tris, pH 6.5) were incubated at 20°C for 30 h and applied to AG2X-50 columns as above. Each column was eluted with 700 l of deionized water, and the eluates were pooled, lyophilized, resuspended in 100 l of 40% n-propyl alcohol, streaked onto two Silica Gel 60 TLC plates, and developed as above. The radioactive trisaccharide was localized using a TLC plate scanner (System 200A imaging scanner; Bioscan Inc., Washington, D.C.), and the corresponding area of silica was recovered. The product was eluted from the silica powder by vortexing for 1 min with 5 ml of water and allowing the mixture to remain for 1 h at room temperature. The silica was sedimented (10,000 ϫ g, 10 min, room temperature), and the supernatant was lyophilized and resuspended in 1 ml of water. The product was then purified by solid phase extraction using a protocol adapted from Ref. 35. Briefly, the suspension was applied to a 0.25 g (1 ml) Envi-Carb solid phase extraction column (Supelco, Bellefonte, PA) that had been preconditioned with 3 ml of 80% (v/v) acetonitrile in 0.1% (v/v) trifluoroacetic acid followed by 2 ml of water. The loaded column was washed with 3 ml of water, and the product was eluted with 3 ml of 25% acetonitrile in water (v/v). The acetonitrile in the eluate was removed by evaporation under N 2 at 50°C, and the remainder of the sample was lyophilized.
␣-Mannosidase Treatment of the XT Product-XT product was partially purified by TLC as described above. The product was then incubated with jack bean ␣-mannosidase (Oxford Glycosystems, Oxford, UK) for 18 h at 37°C in buffer supplied by the manufacturer.
One-dimensional 1 H and Two-dimensional 1 H-1 H Nuclear Magnetic Resonance Spectroscopy-Samples of the XT product and the ␣-1,3-Man 2 substrate (ϳ0.3-1.0 mg) were deuteriumexchanged by repeated lyophilization from D 2 O and then dissolved in 0.5 ml of D 2 O for NMR analysis. One-dimensional 1 H NMR, two-dimensional 1 H-1 H-gCOSY, two-dimensional 1 H-1 H-TOCSY, and one-dimensional 1 H-1 H nuclear Overhauser effect spectra were acquired at 25°C on a Varian Unity Inova 500-MHz spectrometer (Department of Chemistry, University of New Hampshire), using standard acquisition software available in the Varian VNMR software package. Proton chemical shifts are referenced to internal acetone (␦ ϭ 2.225 ppm).
Permethylation and Linkage Analysis-A portion of the XT product was permethylated using the method of Ciucanu and Kerek (36) with the modification of Ciucanu and Costello (37). An aliquot of the permethylated product was further treated (by hydrolysis, reduction, and per-O-acetylation) for analysis of partially methylated alditol acetates (PMAAs), using the protocols described by Levery and Hakomori (38). The PMAAs were analyzed on an Rtx-5MS-bonded phase-fused silica capillary column (30 m ϫ 0.25 mm; 0.25 m phase thickness; Restek Corp., Bellefonte, PA) in the splitless mode using a Trace GC ultra gas chromatograph interfaced to a Polaris Q ion trap mass spectrometer (Finnigan MAT, San Jose, CA), with MS operated in electron ionization mode, and gas chromatography programmed as described previously (38). The derivatives were identified by retention times and characteristic electron impact mass spectra compared with standards (39 -41).
Positive Ion Mode Electrospray-Ionization Mass Spectrometry-Mass spectrometry was performed in the positive ion mode on a linear ion trap (LTQ, ThermoFinnigan San Jose, CA), with sample introduction via direct infusion in 50% MeOH-H 2 O (for the native XT product) or 100% MeOH (for the permethylated XT product), and a sample concentration of ϳ100 ng/l.
Xylosyltransferase Purification-All steps were carried out on ice or at 4°C, unless otherwise indicated. Q-Sepharose, DEAE, and Sephacryl S-300 chromatography were carried out on anÄKTA FPLC (GE Healthcare). Protein was measured with the Bio-Rad DC protein reagent (Bio-Rad) using IgG as the standard. All buffers used in the purification are listed in Table 1.
Crude Membrane Preparation-ags1⌬ cells grown in large scale as above were harvested by centrifugation (9,000 ϫ g, 15 min) at 4°C, resuspended in Buffer A, and recentrifuged. The cells were resuspended in 30 ml of Buffer A (total volume Ϸ 75 ml), divided into three 50-ml conical tubes, and mixed with 12 ml of 0.5-mm glass beads. Cells were broken by vortex mixing six times for 2 min, with 2 min on ice between each mixing, and cell lysates were transferred to fresh 50-ml conical tubes. The beads were washed with 15 ml of Buffer A, and this wash was pooled with the lysate. To sediment broken cells and debris, each pool was centrifuged (1,200 ϫ g, 20 min), and membranes were recovered from the supernatant fraction by ultracentrifugation (60,000 ϫ g, 45 min). The supernatants were discarded, and each membrane pellet was suspended in 50 ml of Buffer A, resedimented (60,000 ϫ g, 30 min), and resuspended in 2 ml of Buffer A. For smaller scale membrane preparations, glass bead lysis was performed, either in a Mini-Bead Beater 8 (Biospec Products, Bartlesville, OK) or by vortex mixing as in Ref. 42.
Solubilization-Pooled crude membranes were mixed with 20% Triton X-100 to a final concentration of 1% and incubated on ice for 30 min, with brief vortex mixing every 5 min. This solution was cleared by centrifugation (60,000 ϫ g, 30 min), and the supernatant fraction (SolZ) was recovered.
Q-Sepharose Chromatography-The SolZ fraction (2-3 ml) was filtered using 0.45-m spin filters and promptly applied to a 20-ml HiPrep 16/10 Q-Sepharose Fast Flow column (GE Healthcare) that had been pre-equilibrated with Buffer B. The column was then washed with 200 ml of Buffer B, and activity was eluted with a 200-ml gradient from Buffer B to Buffer C. 4-ml fractions were assayed as above, and the activity peak was pooled and concentrated to ϳ500 l (SolQ) using a 15-ml Amicon Ultra 10-kDa molecular mass cutoff spin concentrator.
Sephacryl S-300 Gel Filtration Chromatography-The Sol Q fraction was filtered and applied to a 316-ml HiPrep 26/60 Sephacryl S-300 gel filtration column (GE Healthcare) that was pre-equilibrated with Buffer D. The column was eluted with a 253-ml (0.8 column volume) isocratic gradient of Buffer D, with the first 95 ml going to waste (void volume). The remaining 158 ml were collected into 4-ml fractions and assayed, and the activity peak was concentrated to 1 ml (SolS).
Concanavalin A Chromatography-The SolS fraction was diluted to 10 ml with Buffer E, reconcentrated to 1 ml, and diluted back to 2.5 ml with Buffer E. This was then applied to a 2.5-ml column of concanavalin A-Sepharose 4B (GE Healthcare) that had been pre-equilibrated with Buffer E. The column was capped and allowed to rock at 4°C overnight. The flowthrough was then collected and the column washed with 60 ml of Buffer E. 50 ml of Buffer F was then added to the column. After 10 ml of Buffer F had passed through and been collected, the column was capped and allowed to rock for 30 min at 4°C. The remainder of the Buffer F was then allowed to elute, pooled with the first 10 ml, and concentrated to 1 ml (SolC). DEAE Chromatography-To desalt SolC, it was diluted to 10 ml with Buffer G, concentrated to 1 ml, rediluted to 10 ml, and concentrated to 0.8 ml. The final solution was filtered and applied to a 1-ml HiTrap DEAE-Fast Flow column (GE Healthcare). The column was washed with 15 ml of Buffer H and eluted with a 20-ml gradient from Buffer H to Buffer I. 1-ml fractions were assayed, and the activity peak was concentrated to 1 ml (SolD).
␣-1,3-Man 2 Affinity Chromatography-The SolD fraction was diluted to 15 ml with Buffer G and extensively desalted with three successive concentrations and dilutions, to a final volume of 1 ml. This was mixed with 1 ml of Buffer J and loaded onto a column containing 5 ml of a custom-synthesized ␣-1,3-Man 2agarose resin (Carbohydrate Synthesis, Oxford, UK) that had been pre-equilibrated with Buffer J. The column was capped, allowed to rock at 4°C for 30 min and then at 25°C for 90 min, and then washed with 30 ml of Buffer K at 25°C. The column was then moved to 4°C and eluted with a 44-ml gradient from Buffer K to Buffer L, collecting 0.85-ml fractions. The activity peak was combined into two pools that were each concentrated to 25 l; one pool that consisted of the shoulders of the peak and the other the middle of the peak.
Protein Analysis-The two pools from the ␣-1,3-Man 2 column were resolved by SDS-PAGE on a 12% gel, which was stained with SYPRO Ruby (Bio-Rad) as per the manufacturer's instructions and visualized on a Bio-Rad Molecular Imager FX. An ϳ90-kDa band of interest was excised and submitted to the Protein and Nucleic Acid Chemistry Laboratory at Washington University School of Medicine for trypsin digestion followed by HPLC separation and mass spectrometry analysis of the resulting fragments. The mass spectra were compared with the C. neoformans protein data base, and the protein sequence corresponding to the band of interest was then analyzed using the Prosite domain prediction server and the TMpred transmembrane domain prediction server.
Expression in S. cerevisiae-CXT1 was amplified from JEC21 cDNA by PCR using primers Exp-sense and Exp-antisense ( Table 2) to incorporate a His 6 tag at the C terminus, as well as HindIII and BamHI sites at the 5Ј and 3Ј ends, respectively. The product was cloned into a TOPO pCR2.1 vector (Invitrogen) and sequenced (Protein and Nucleic Acid Chemistry Laboratory, Washington University School of Medicine). A Topo-CXT1 clone with the correct sequence was then digested with HindIII and BamHI, and the released fragment was cloned into a 2-m yeast expression vector between the promoter and terminator of phosphoglycerate kinase (plasmid pPGK (43); from K. Blumer, Washington University School of Medicine). The resulting plasmid and empty vector were transformed into TDY172 by electroporation, and transformants containing the URA3 marked plasmid were selected by plating onto URAϪ medium. To assess XT activity, two independent transformants, bearing either pPGK-CXT1 or empty pPGKm were grown overnight in 50 ml of URAϪ medium, and membranes were prepared and assayed as above. Expression was confirmed by immunoblotting with anti-His antibody (1:5000) using standard methods. Expressed, solubilized Cxt1p-His 6 was partially purified using a Talon Superflow metal affinity resin (Clontech). Briefly, 500 l of solubilized membranes were applied to 2 ml of resin that was equilibrated with 50 mM Tris 7.5, 0.01% Triton X-100, 200 mM NaCl. The column was washed with 5 ml of the same buffer, and then with 15 ml of the same buffer containing 10 mM imidazole. Protein was eluted with 15 ml of the same buffer containing 250 mM imidazole, and the eluate was collected and concentrated.
Site-directed Mutagenesis-The CXT1 sequence in pPGK-CXT1 was mutated using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) with the primers listed in Table 2. The mutagenized sequences were confirmed by DNA sequencing, and correct plasmids were electroporated into TDY172 and analyzed as above.

Development of a Xylosyltransferase Activity Assay-To
detect an XT with activity appropriate for capsule synthesis, we developed an assay for xylose transfer to an ␣-1,3-linked disaccharide of mannose (␣-1,3-Man 2 ). This substrate resembles both the backbone of GXM and the dimannose moiety of GalXM (see Figs. 1 and 2). Cryptococcal membranes were incubated with this substrate in the presence of UDP-[ 14 C]xylose; unutilized UDP-[ 14 C]xylose was removed by anion exchange, and the radiolabeled products formed were separated by thin layer chromatography and detected by autoradiography. As  shown in Fig. 3, a radiolabeled product, which migrated more slowly than the ␣-1,3-Man 2 standard, was detected. Appearance of this product was dependent on ␣-1,3-Man 2 addition, time, and membrane protein concentration (data not shown). Notably, the activity was not dependent on the addition of divalent cations, nor could activity be reduced with up to 10 mM EDTA (data not shown).
Analysis of the Oligosaccharide Product of the XT Assay-We hypothesized that the XT product was a trisaccharide, as it migrated in this TLC solvent system similar to other trisaccharide standards (data not shown). To confirm this, the product was digested with jack bean ␣-mannosidase. This resulted in its quantitative conversion to a radiolabeled product that migrated close to the ␣-1,3-Man 2 standard (Fig. 3). Because this ␣-mannosidase enzyme only cleaves terminal, nonreducing mannose residues, this suggested that the product of the XT assay was a trisaccharide with the radioactive xylose moiety linked to the reducing mannose of the ␣-1,3-Man 2 .
To verify our conclusions about the assay product and to determine the linkage of xylose attachment, purified product was subjected to structural characterization by MS, NMR, and methylation linkage analysis. In positive mode electrospray ionization-MS in a quadrupole ion trap (ThermoFinnigan LCQ), we observed a molecular salt/adduct at nominal, monoisotopic m/z 497, consistent with [M ϩ Na] ϩ for a trisaccharide consisting of one pentose and two hexose residues (spectrum not shown). Following permethylation, the molecular adduct [M ϩ Na] ϩ was observed strongly in electrospray ionization MS at a nominal monoisotopic m/z 637 consistent with the same composition. MS 2 fragmentation in the ion trap yielded a trio of major product ions consistent with loss of either the nonreducing pentose residue (m/z 463, m/z 431, [m/z 463 Ϫ MeOH]) or one hexose residue (m/z 419) (spectrum not shown). Loss of a permethylated hexose residue from either the reducing or nonreducing end may produce product ions of the same m/z, so further MS n analysis would not determine to which hexose residue the pentose was attached. To address this question, NMR spectroscopic analysis was performed. Fig. 4 compares the one-dimensional 1 H-NMR spectra of the ␣-1,3-Man 2 substrate (panel A) and XT product (panel B). The latter clearly shows changes consistent with glycosylation by another sugar residue, including a pair of additional ␤-linked H-1 resonances of unequal amplitude at 4.543 ppm ( 3 J 1,2 ϭ 7.7 Hz) and 4.412 ppm ( 3 J 1,2 ϭ 7.8 Hz). This spectrum also shows significant chemical shift changes for the four downfield resonances corresponding to H-1 of both mutarotatotory forms of the original Man␣-1,3Man␣/␤ disaccharide acceptor, which are in the same relative proportions as the two additional ␤-configured H-1 signals. Assignment of the product spectrum by two-dimensional NMR methods showed both of the new ␤ H-1 signals to be coupled into two respective 6-proton spin systems, as shown in the TOCSY spectrum (Fig. 4, panel C); sequential analysis of their J-coupling patterns confirmed that they belong to two distinct ␤-Xyl spin systems, consistent with their incorporation into a product undergoing slow mutarotatory interconversion. Furthermore, the large magnitude of the chemical shift differences corresponding to the two ␤ H-1 resonances, along with the large magnitude of the shift changes for the two H-1 signals corresponding to the ␣ and ␤ forms of the reducing end Man residue (5.245 ppm, 3 J 1,2 ϭ 1.7 Hz, and 4.981 ppm, 3 J 1,2 ϭ 0.9 Hz, respectively), are consistent with    2,3Man␣ H-1 and 3Man␤ H-1 or 2,3Man␤ H-1, respectively (panels A and B). Assignments of H-1 and other protons of nonreducing residues in the two forms of each oligosaccharide are denoted by superscripts ␣ and ␤. The TOCSY spectrum (panel C) shows only those correlations arising directly from the monosaccharide anomeric protons (4.3-5.4 ppm in F1 dimension). Assignment strategy also incorporated TOCSY spectra with lower mixing times, as well as gCOSY spectra giving only 3-bond correlations (not shown). In the TOCSY spectrum, the prefix H is omitted from proton designations for clarity, and complete spin systems for the ␤-Xyl residue(s) are visible and marked; for Man residues, additional correlations are visible in other parts of the spectrum. glycosylation of the interconverting reducing end Man residue. The large chemical shift difference between the two ␤ H-1 resonances also suggests glycosylation at a site very near to the anomeric carbon, i.e. at O-2. By contrast, the other pair of H-1 resonances, at 5.149 ppm ( 3 J 1,2 ϭ 1.8 Hz) and 5.136 ppm ( 3 J 1,2 ϭ 1.5 Hz) experience little change with respect to their positions in the spectrum of the acceptor substrate; these only differ from each other by a small increment because of long range effects of mutarotation of the reducing end monosaccharide.
Our structural assessments were confirmed by one-dimensional 1 H-1 H nuclear Overhauser effect difference spectra acquired following selective pre-irradiation of the two ␤-Xyl H-1 signals (supplemental Fig. 1). The magnitude of signals appearing in nuclear Overhauser effect difference spectra are indicative of spatial proximity (less than 5 Å) between the irradiated and nearby nuclei, because of cross-relaxation highly dependent on internuclear distance (and other parameters, including relative geometry and molecular motion). Pre-irradiation of the major ␤-Xyl H-1 resonance yielded selective time-dependent enhancements, not only for other resonances within the same sugar ring (␤-Xyl H-2, H-3, and H-5ax), but also strong inter-residue enhancements for the reducing Man ␣ H-1 and H-2 signals (supplemental Fig. 1, panel B). Upon preirradiation of the minor ␤-Xyl H-1, only two enhancements could be clearly observed, probably due to the lower abundance of this configuration. These were an intra-residue enhancement of ␤-Xyl H-3 and an inter-residue enhancement of reducing Man ␤ H-2 signal (supplemental Fig. 1, panel C). Construction of molecular models for both forms of the trisaccharide product show that these patterns of enhancements are consistent only with linkage of the ␤-Xyl residue to O-2 of the reducing Man.
Additional confirmation of xylose linkage to O-2 of the reducing mannose was provided by linkage analysis of the XT product following permethylation, depolymerization, reduction, and peracetylation to produce partially methylated PMAAs (data not shown). In the subsequent gas chromatography-MS analysis, PMAA derivatives were detected corresponding to nonreducing terminal Xyl, nonreducing terminal Man, and 32(33)-linked Man. Because the Man residues in the acceptor substrate were joined by a 3-linkage, it follows that the added Xyl residue occupied the 2-position of the di-substituted reducing Man in the intact XT product.
Purification and Identification of the ␤-1,2-Xylosyltransferase -We were interested that the product formed in our XT assay corresponded to structures present in GalXM (Fig. 2) and GXM of all serotypes (Fig. 1). We therefore proceeded to purify the protein responsible for this activity. We were aided by the fact that the product of interest was the dominant radiolabeled species in our assay (Fig. 3, Ϫ lane). This allowed us to use scintillation counting as a method of detection, rendering the assay more rapid, less laborious, and appropriate for use in XT    Crude membranes  3  1  100  1% Triton X-100 extract  9  3  97  QFF anion exchange  200  67  82  S-300 gel filtration  260  87  40  Concanavalin A affinity/DEAE  315  105 17 Man-␣-1,3-Man affinity 8900 2967 3 a A unit of activity was defined as the amount of enzyme that transfers 100 cpm of ͓ 14 C͔Xyl from UDP-͓ 14 C͔Xyl to ␣ -1,3-dimannoside per min.
purification. As described under "Experimental Procedures," we prepared membranes from the serotype D ags1⌬ strain (33). XT activity in this strain is equivalent to wild type (data not shown), but the cells were considerably easier to mechanically disrupt. We then used both conventional and affinity resins to enrich the ␤-1,2-XT ϳ3,000-fold from washed cryptococcal membranes (Table 3). Notably, the XT activity eluted from the S-300 gel filtration column at the position of a 90-kDa protein.
Final enrichment of the XT enzyme was not to homogeneity but did lead to a significant reduction in the number of bands on an SDS-polyacrylamide gel. Chromatography using ␣-1,3-Man 2 -agarose affinity resin was a particularly effective step. Fractions from the XT activity peak observed upon elution of this resin were combined in two pools, with the middle of the activity peak separated from the combined shoulders of the peak. Each pool was concentrated and assayed for XT activity, which indicated that there was 2.5 times more activity in the middle "peak" pool. We next analyzed these two fractions by SDS-PAGE, seeking a protein band that was severalfold more abundant in the peak fraction and of roughly 90 kDa. One band (Fig. 5, arrow) displayed those characteristics. This band was excised and submitted for trypsin/HPLC/MALDI analysis, which yielded peptide masses matching a 694amino acid (79 kDa) predicted protein from C. neoformans serotype D genome sequence. This sequence had been previously named CAP3 based on homology to CAP10, but it had no known function.
Because we had purified the XT from membranes and the activity bound to a concanavalin A resin (Table 3), we expected that the identified protein would have a transmembrane domain and N-linked glycosylation site. As shown in Fig.  6, the predicted protein did have these features, in addition to two DXD motifs commonly found in glycosyltransferases.
Expression of the Putative Xylosyltransferase in Yeast-To confirm that the sequence we identified encoded the enzyme catalyzing the XT activity, we expressed this hypothetical protein in S. cerevisiae, which lacks any XT activity (Fig. 7, left lanes). Sequence encoding a C-terminal His 6 -tagged version of the open reading frame was cloned into a 2-m plasmid, and expression was confirmed by immunoblotting with anti-His antibodies (data not shown). Expression of the protein did indeed lead to the appearance of substrate-dependent XT activity (Fig. 7, middle lanes), and the product co-migrated with authentic product produced by C. neoformans (Fig. 7, right lanes). This confirmed that the gene we had  identified was correct. Furthermore, XT activity could be partially purified from solubilized membranes of S. cerevisiae expressing this protein using a Co 2ϩ metal affinity column (data not shown). Upon this confirmation of activity, we renamed the gene CXT1 for cryptococcal xylosyltransferase 1 (GenBank TM accession number 905015).
Site-directed Mutagenesis of CXT1-We wondered whether N-glycosylation and both DXD motifs were important for the XT activity. To address this, we separately mutated each of these loci of the pPGK-CXT1 expression plasmid, and we expressed the mutated proteins in S. cerevisiae. None of the mutations altered protein expression (immunoblotting data not shown). However, when the N-glycosylation site was mutated (N141Q), the apparent gel mobility of Cxt1p did shift to the molecular weight predicted for the unmodified protein sequence, suggesting that the protein is normally N-glycosylated at that site. Despite this mutation, we saw little change in XT activity (Fig. 8). In contrast, mutation of either DXD motif to AXD led to a dramatic reduction in activity, with complete loss of activity upon mutagenesis of the first DXD motif, at position 550 (Fig. 8).
Identification of Cxt1p Homologs-We were intrigued by the size and apparent cation independence of Cxt1p and searched for homologous proteins in C. neoformans and other organisms. Within C. neoformans (serotype D), we found five homologs with identities ranging from 20 to 30% and similarities from 30 to 42%. This group consisted of CAP10 and its four homologs (supplemental Table S-1). We also discovered 34 additional homologs in other fungal organisms, with identities ranging from 12 to 23% and similarities from 21 to 32%. The homologs were generally large in size, with only five being less than 60 kDa (supplemental Table S-1). Although neither the DXD motif at position 659 nor the N-glycosylation site of Cxt1p was conserved (data not shown), the activity-dependent DXD motif at position 550 of Cxt1p is conserved in all of these homologs (Fig. 9).
Finally, Cxt1p was also present in the C. neoformans serotype A (H99) genome sequence. This sequence had 93% identity and 95% similarity to the serotype D sequence and was conserved with respect to both DXD motifs and to the N-glycosylation site (not shown). Furthermore, a robust, apparently identical XT activity was also detected in the H99 serotype A strain (data not shown).

DISCUSSION
In the course of our investigations of capsule synthesis in C. neoformans, we have detected and purified a ␤-1,2-XT that transfers xylose to ␣-1,3-Man 2 . We have also cloned the corresponding gene, now named CXT1 for cryptococcal xylosyltransferase 1.
Cxt1p has the type II membrane protein structure typical of glycosyltransferases, although it is large for this class of enzymes (23).Interestingly,theXTactivityisnotdivalentcationdependent, even though it is reduced dramatically by mutagenesis of the Cxt1p DXD motifs. An Arabidopsis thaliana XT that is not dependent on any divalent cation has also been described (44), but this enzyme does not have a DXD motif. It is thus unclear how the DXD motifs are functioning in Cxt1p, and this may indicate a novel catalytic mechanism. Alternatively, it is possible that a divalent cation is bound so tightly to one of the Cxt1p DXD motifs that it cannot be removed with EDTA, or that mutation of the DXD motif causes a folding defect deleterious to activity even though that motif is not directly involved in catalysis. Future structural studies of this enzyme should clarify this issue.
Glycosyltransferases are grouped into families based on sequence homology, and maintained in the CAZy (carbohydrate active enzymes) data base. Although homology is a convenient and useful means for glycosyltransferase classification, it is insufficient to predict the precise transferase functions of these enzymes (24). Thus, we cannot conclude that the Cxt1p homologs are all XTs. However, this group of sequences does represent a new family of glycosyltransferases that has been submitted for inclusion into the CAZy data base. 3 Interestingly, the members of this novel family of glycosyltransferases are found only in fungal organisms, including a number of human, animal, and plant pathogens. If these enzymes play a role in the virulence or survival of such pathogens, they would be intriguing prospects as targets for future development of anti-mycotic compounds.
Elucidating the biochemical pathways of cryptococcal capsule synthesis has proven challenging. The CAP genes were identified a number of years ago as required for capsule synthesis, but we still do not know the role that their products play in capsule synthesis. Here we show that one of the CAP gene products, Cap10p, has homology to Cxt1p and is conserved with regard to a DXD motif vital to Cxt1p activity. Cxt1p represents the first CAP gene homolog whose product has a defined biochemical function, and this homology suggests that CAP10 and its homologs (CAP1, -2, -4, and -5) encode GTs that are likely to be involved in capsule synthesis.
We hypothesize that there are multiple XTs in C. neoformans involved in capsule biosynthesis. It is notable that the Xyl-␤-1,2-Man-␣-1,3-Man motif formed by CXT1 is present in the GXM of all serotypes, even the simplest serotype D strains (Fig.  1). Furthermore, this structure is also present in GalXM (Fig. 2). Thus, the activity of Cxt1p is consistent with a role in the synthesis of either GXM or GalXM, and we would expect this enzyme to be present and active in all serotypes. So far, we have shown activity in serotypes D and A.
Although our overall interests focus on cryptococcal capsule synthesis, it is possible that CXT1 is involved in the synthesis of other glycans. For example, the Xyl-␤-1,2-Man-␣-1,3-Man motif is also found in glycolipids (45) and O-glycans (46). Furthermore, a ␤-1,2-xylosyltransferase activity with similar characteristics has also been described in the nonpathogenic fungus Cryptococcus laurentii (47). This C. laurentii XT activity was believed to be involved in O-glycan synthesis based on the structure of the product formed in vitro (48). However, this enzyme was not purified, and thus its role in protein glycosylation has not been definitively established. Defining the precise role of Cxt1p in cryptococcal biology must await the generation and analysis of mutant strains, work now in progress.