Biosynthesis ofO-N-Acetylglucosamine-linked Glycans inTrypanosoma cruzi

In this study, we have characterized the activity of a uridine diphospho-N-acetylglucosamine:polypeptide-α-N-acetylglucosaminyltransferase (O-α-GlcNAc-transferase) from Trypanosoma cruzi. The activity is present in microsomal membranes and is responsible for the addition of O-linked α-N-acetylglucosamine to cell surface proteins. This preparation adds N-acetylglucosamine to a synthetic peptide KPPTTTTTTTTKPP containing the consensus threonine-rich dodecapeptide encoded by T. cruzi MUC gene (Di Noia, J. M., Sánchez D. O., and Frasch, A. C. C. (1995) J. Biol. Chem. 270, 24146–24149). Incorporation ofN-[3H]acetylglucosamine is linearly dependent on incubation time and concentration of enzyme and substrate. The transferase activity has an optimal pH of 7.5- 8.5, requires Mn2+, is unaffected by tunicamycin or amphomycin, and is strongly inhibited by UDP. The optimized synthetic peptide acceptor for the cytosolic O-GlcNAc-transferase (YSDSPSTST) (Haltiwanger, R. S., Holt, G. D., and Hart, G. W. (1990)J. Biol. Chem. 265, 2563–2568) is not a substrate for this enzyme. The glycosylated KPPTTTTTTTTKPP product is susceptible to base-catalyzed β-elimination, and the presence ofN-acetylglucosamine α-linked to threonine is supported by enzymatic digestion and nuclear magnetic resonance data. These results describe a unique biosynthetic pathway for T. cruzi surface mucin-like molecules, with potential chemotherapeutic implications.

Trypanosoma cruzi is the causative agent of Chagas' disease, a multisystemic disorder endemic in much of Latin America. This protozoan has a complex life cycle involving morphologically distinct stages in mammalian and insect hosts. Bloodsucking triatomine bugs transmit infective trypomastigotes to the mammalian host, which multiply intracellularly as amastigotes prior to differentiation into trypomastigotes, which, after rupture of the host cell, enter the blood stream, enabling infection of fresh cells or ingestion by a feeding triatomine bug, thus completing the biological cycle of transmission (1). Antigenic glycoconjugates, including the highly O-glycosylated sialoglycoproteins, known as mucin-like molecules, have been strongly implicated in the molecular mechanism of attachment to and invasion of mammalian host cells (2). The sialic acid residues present in these molecules are derived from host sialoglycoconjugates (3) and are transferred to the T. cruzi surface glycoproteins by a unique trans-sialidase (4). The first evidence for O-glycosylation of serine and/or threonine in trypanosomal glycoproteins came from studies of T. cruzi GP-25 (5), a glycoprotein corresponding to the C-terminal domain of cruzipain (6). More recently, O-glycosylated mucinlike proteins were demonstrated in metacyclic (7) and cellderived trypomastigotes (8) and in epimastigotes (9,10). Structural analyses have shown that these O-glycans vary between strains and developmental stages (9 -11).
The striking feature of these O-glycans is that they are linked to the peptide backbone through an N-acetylglucosamine (GlcNAc) unit, with threonine (Thr) rather than serine (Ser) being the usual site of attachment (10,12). The GlcNAc-Thr core can be extended by addition of Galp, Galf, and sialic acid residues. These were the first O-GlcNAc-linked oligosaccharides reported in surface glycoproteins. Previously, single O-GlcNAc units linked to Ser and/or Thr have been described on nuclear and cytosolic glycoproteins (13). This post-translational modification on the nuclear and cytoplasmic proteins is catalyzed by a cytosolic O-GlcNAc-transferase (14).
The unusual addition of O-GlcNAc to T. cruzi surface glycoproteins prompted us to investigate this post-translational modification in more detail. In this paper, we describe a novel UDP-GlcNAc:polypeptide ␣-N-acetylglucosaminyltransferase from T. cruzi and specify the optimal conditions for its activity. We have also characterized the in vitro glycosylation products and have established the anomeric configuration of GlcNAc O-linked to Thr. (40 - 4 in water for 1 h at room temperature, followed by addition of 100 mol of unlabeled NaBH 4 . Borate salts were removed by repeated addition of methanol and evaporation, passage through a mixed-bed ion exchange resin, and gel filtration on Bio-Gel P-4 (extra fine). The radioactive monosaccharides [6-3 H]GlcN and [6-3 H]GalN were prepared by hydrolysis of radioactive sugar nucleotides in 3 M HCl at 100°C for 4 h. Unlabeled UDP-GlcNAc, SP Sephadex (SP-C25-120), and ␤-N-acetylglucosaminidase from jack beans were obtained from Sigma. The peptides KPPTTTTTTTTKPP, YSDSPSTST, and YSPTSPSK (with the O-␤-GlcNAc on the serine at position 5) used in this study were kindly supplied by A. C. C. Frasch, R. S. Haltiwanger, and G. W. Hart, respectively. All other reagents were of the highest available quality.

Materials-Radioactively labeled UDP-[6-3 H]GlcNAc
Parasites-The Y strain of T. cruzi was used in all enzymatic experiments. Native sialoglycoproteins were purified from epimastigotes of the Y, CL-Brener, and Dm28c strains. Epimastigotes were cultured at 28°C in brain-heart infusion medium supplemented with hemin and 5% of fetal calf serum, and harvested in the exponential phase of growth (15). Trypomastigotes were obtained from LLC-MK 2 cells infected with tissue culture-derived trypomastigotes for 6 days, maintained at 37°C in RPMI medium containing 10% fetal calf serum, under 5% CO 2 (8).
Microsomal Membrane Preparation-Pellets of 2 ϫ 10 11 epimastigotes or 1.12 ϫ 10 9 cell-derived trypomastigotes were ground in liquid nitrogen. The homogenate was diluted with 10 -20 ml of 250 mM sucrose and 25 mM Tris/HCl, pH 7.4 (Tris/sucrose buffer), and centrifuged for 10 min at 12,000 ϫ g. The supernatant was then ultracentrifuged for 1 h at 120,000 ϫ g, and the resulting pellet was resuspended in Tris/sucrose buffer with a glass-Teflon homogenizer and ultracentrifuged as above. The pellet was then resuspended in Tris/sucrose buffer and appropriately diluted for protein assay (16).
Enzymatic Assay: O-␣-GlcNAc-transferase Activity-The standard assay contained, in a final volume of 50 l, 25 mM Tris/HCl buffer (pH 7.4), 5 mM MgCl 2 , 5 mM MnCl 2 , 0.1% Triton X-100, 1.5 Ci of UDP-[ 3 H]GlcNAc (40 -60 Ci/mmol), and 6.8 nmol of synthetic peptide acceptor substrate (KPPTTTTTTTTKPP). The reaction was initiated by addition of microsomal membranes (250 g of protein). Control assays without the acceptor peptide were used to correct for endogenous activity. The mixture was incubated at 28°C for 30 min, and the reaction terminated by addition of 950 l of 50 mM formic acid. The reaction mixture was loaded onto a 1-ml sulfopropyl-Sephadex column (SP-C 25-120) equilibrated in 50 mM formic acid. The column was washed with 5 ml of 50 mM formic acid, and the peptide and labeled glycopeptide were eluted with 4 ml of 0.5 M NaCl. Incorporation of [ 3 H]GlcNAc into peptide was determined on aliquots of the eluate by liquid scintillation counting after addition of Bray solution (5 ml). In experiments to increase radiolabel incorporation, the reaction mixture (100 l) contained 25 mM Tris-HCl (pH 7.4), 5 mM MgCl 2 , 5 mM MnCl 2 , 15 Ci of UDP-[ 3 H]GlcNAc, 6.8 nmol of acceptor peptide, and microsomal membranes (1 mg of protein). Other conditions and the method for the isolation of the labeled peptide were as described above. For testing of trypomastigote microsomal fraction, the reaction mixture (50 l) contained 25 mM Tris-HCl (pH 7.4), 5 mM MgCl 2 , 5 mM MnCl 2 , 1.5 Ci of UDP-[ 3 H]GlcNAc, 3.4 nmol of acceptor peptide, and 25 g (protein) of the microsome preparation. Other conditions were as described above.
HPLC Purification of O-Glycosylated Peptide-Incorporation of [ 3 H]GlcNAc into synthetic peptide was confirmed by RP-HPLC on a C18 column (5 m, 4 ϫ 250 mm, Amersham Pharmacia Biotech) eluted with an H 2 O/acetonitrile/trifluoroacetic acid gradient. Solvent A was 0.1% aqueous trifluoroacetic acid, and solvent B was 0.089% trifluoroacetic acid in acetonitrile. The linear gradient was started 5 min after injection from 0 to 15% solvent B over 40 min (held until 50 min). The flow rate was 1 ml/min, and 0.8-or 1.0-ml fractions were collected. Aliquots were mixed with 5 ml of Bray solution, and the radioactivity was determined by liquid scintillation counting. Elution of unlabeled peptide was monitored by absorbance at 206 nm.
Analysis of the Reaction Product-[ 3 H]GlcNAc-labeled peptide fraction obtained by RP-HPLC was rechromatographed on Bio-Gel P-4 (extra fine) (120 cm ϫ 0.5 cm) at 0.6 ml/h with 0.2 M ammonium acetate as the eluent. The column was calibrated using bovine serum albumin and [ 3 H]glucose as marker for the void (v o ) and included volumes, respectively. The elution positions of authentic standards of [ 3 H]Glc-NAcO, Galp␤134[ 3 H]GlcNAcO, and Galp␤136(Galp␤134)[ 3 H] GlcNAcO were also determined. Fractions of 300 l were collected and assayed for radioactivity. The labeled peptide fraction was desalted, lyophilized, and dissolved in 10 mM NaOH, containing 0.3 M NaBH 4 , at 37°C for 48 h to ␤-eliminate and reduce saccharides linked to threonine (17). The solution was neutralized with 2 M acetic acid and passed through Dowex 50W-X8 (25-50 mesh H ϩ form). Boric acid was removed by repeated evaporation with methanol. The residue was dissolved in distilled water and analyzed by gel filtration on Bio-Gel P-4 column (as described above) and by descending paper chromatography in ethyl acetate:pyridine:water (8:2:1) for 48 h (18). The distribution of radioactivity on the paper chromatogram was determined by cutting the paper strips into 1-cm sections and liquid scintillation counting.
In the experiment using a microsome preparation from trypomastigote forms, the [ 3 H]GlcNAc-labeled peptide was submitted to acid hydrolysis in 3 M HCl, at 100°C for 4 h; the HCl evaporated in vacuo, and the radiolabeled product was analyzed by descending paper chromatography in butanol:pyridine:0.1 M HCl (5:3:2) for 18 h (19).
Digestion with Jack Bean ␤-N-Acetylglucosaminidase-The reaction mixture for jack bean ␤-N-acetylglucosaminidase digestion contained HPLC-purified [ 3 H]GlcNAc-labeled peptide (approximately 5 ϫ 10 4 cpm), 50 mM phosphate-citrate buffer, pH 5.0, and 0.8 units of enzyme in a final volume of 100 l. The mixture was incubated at 37°C for 18 h (20, 21), the reaction was terminated by addition of 900 l of 50 mM formic acid, and the solution was loaded onto a 1-ml SP-Sephadex (25-120) column equilibrated in 50 mM formic acid. Released [ 3 H]Glc-NAc was eluted with 50 mM formic acid (5 ml), lyophilized, and the radioactivity was determined by liquid scintillation counting. p-Nitrophenyl N-acetyl-␤-D-glucosaminide was used as substrate control. As a further positive control, 100 mol of the unlabeled synthetic glycopeptide YSPTSPSK, with the O-␤-GlcNAc on the serine at position 5 (kindly provided by G. W. Hart) was digested under identical conditions. The reaction was terminated by addition of 50 mM formic acid, and the liberated GlcNAc was recovered on SP-Sephadex as above and quantified using the Morgan-Elson reaction (22). A standard curve was constructed using D-GlcNAc similarly chromatographed on SP-Sephadex.
Preparation of GlcNAc-rich Glycopeptides from Native Sialoglycoproteins for NMR Spectroscopy-Sialoglycoproteins from epimastigote forms were purified as described by Previato et al. (4). The sialoglycoproteins from T. cruzi CL-Brener strain were subjected to partial acid hydrolysis (0.2 M trifluoroacetic acid for 2 h at 100°C). The GlcNAc-rich glycopeptides were recovered by gel filtration on a column of Sephadex G25 SF (1 ϫ 10 cm). Also, GlcNAc-rich glycopeptides were obtained from Y, CL-Brener, and Dm28c sialoglycoproteins by Smith degradation (23). Briefly, to 2 ml of a solution containing 30 mg of T. cruzi sialoglycoproteins in 0.1 M sodium acetate (pH 4.6), an equal volume of 0.2 M NaIO 4 was added. After 24 h at 4°C, oxidation was terminated by addition of glycerol. The oxidized products were recovered by gel filtration on Sephadex G25 SF (1 ϫ 10 cm) column, using water as eluent, at a flow rate of 1 ml/min. The oxidized glycoproteins were reduced with sodium borohydride for 3 h at room temperature. Boric acid was removed by repeated evaporation with methanol. The material, oxidized and reduced, was partially hydrolyzed with 20 mM trifluoroacetic acid for 30 min at 100°C. The resulting GlcNAc-rich glycopeptides were recovered by gel filtration on Sephadex G25 SF as above.
NMR Analysis of the GlcNAc-rich Glycopeptides-NMR spectra were acquired on a Varian Unity 500 NMR spectrometer equipped with a 5-mm PFG (pulsed field gradient) triple resonance probe and at indicated probe temperatures of 30 or 40°C. Standard pulse sequences were used except for the introduction of an echo sequence into the ROESY and TOCSY spectra and WHSQC pulse sequence, which is an implementation of the method of Wider and Wü thrich (24). The mixing time in the TOCSY spectra was 80 and 150 ms in the ROESY spectra. Presaturation of the residual water signal was achieved using a low power pulse from the transmitter. In the WHSQC spectra, obtained at 500 MHz, heteronuclear decoupling was achieved using the GARP sequence. 1 H and 13 C chemical shifts were referenced to internal 3-(trimethylsilyl)tetra-deuteropropionic acid at zero ppm ( 1 H) and Ϫ1.8 ppm ( 13 C, to tetramethylsilane at zero) (25).

Development of an Assay for T. cruzi O-␣-GlcNAc-transferase: Optimization of Assay Conditions-An assay for O-Glc-
NAc-transferase activity was developed using microsomal membranes prepared from epimastigotes of the Y strain of T. cruzi. To measure the transfer of GlcNAc in vitro, UDP-[6-3 H]GlcNAc was used as the GlcNAc donor and the synthetic peptide KPPTTTTTTTTKPP as acceptor. This peptide was chosen due to its similarity with a common motif in the peptide sequence of deglycosylated sialoglycoproteins from the Y strain 2 and the MUC gene family products from T. cruzi (26). KPPTTTTTTTTKPP proved to be successful as a substrate for T. cruzi GlcNAc-transferase activity. As shown in Fig. 1, the incorporation of [ 3 H]GlcNAc into the peptide was time dependent for a period up to 2 h (Fig. 1A) and was proportional to increasing amounts of protein (using a microsomal membrane preparation as the enzyme source) (Fig. 1B) and of the peptide acceptor (Fig. 1C). The standard assay was performed at 28°C, although a similar rate was observed at 20°C. At 37°C, the activity was slightly less than that observed at 4°C (Fig. 1A).
Effect of Inhibitors-Addition of 5 mM UDP reduced incorporation of [ 3 H]GlcNAc into the synthetic peptide by 90%. Transferase activity was not reduced in the presence of tunicamycin or amphomycin, even when added at 10 g/ml.
Ion Dependence and Optimum pH-The transferase was in-active in the absence of metal ions, but activity was regained by adding Mn 2ϩ . Table I shows that among seven divalent cations tested, Mn 2ϩ was the most effective in restoring activity. Co 2ϩ and Ca 2ϩ were able to restore 20 and 15% of the activity observed in the presence of Mn 2ϩ . The O-GlcNAc-transferase activity had a pH optimum between 7.5 and 8.5, with maximum activity at pH 8.5. Activity decreased gradually below pH 7.0 and above pH 8.5, with only 50% of the activity at pH 8.5 present at pH 9.0 (Fig. 1D).
Characterization of the Glycosylated Peptide Product-The radiolabeled material recovered from the SP-Sephadex column was characterized by several techniques. A single peak of radioactivity was obtained after reversed phase HPLC on a C18 column with an elution volume distinct from that of the unlabeled peptide acceptor substrate ( Fig. 2A). Bio-Gel P-4 chromatography of this radiolabeled material indicated that its apparent molecular mass was greater than that of the unlabeled peptide acceptor (Fig. 2B). After base-catalyzed ␤-elimination and reduction of the purified radiolabeled glycopeptide, radioactive material eluted at the same volume as authentic Nacetylhexosaminitol on the P-4 column (Fig. 2C), and its identity was confirmed as [ 3 H]GlcNAcO by descending paper chromatography (Fig. 2D). In assays using larger amounts of UDP-[ 3 H]GlcNAc, two radiolabeled fractions were observed in RP-HPLC (Fig. 3A) and on the P-4 column (Fig. 3B). One fraction showed a chromatographic profile identical with that of the glycopeptide obtained under standard conditions (Fig. 2,  A and B). The other fraction was assumed to be a glycopeptide substituted with more than one GlcNAc residue (Fig. 3, A and B), as both fractions liberated [ 3 H]GlcNAcO (identified by descending paper chromatography) on reductive ␤-elimination (Fig. 3C).
To investigate whether cell-derived trypomastigotes of T. cruzi also express the O-GlcNAc-transferase activity, a microsomal preparation from 10 9 trypomastigotes was tested. This fraction catalyzed the synthesis of SP-Sephadex-retained radioactive material, which coeluted with the "epimastigote" glycopeptide in RP-HPLC (Fig. 4A). After acid hydrolysis, the radioactive product was identified as [ 3 H]GlcN by descending paper chromatography (Fig. 4B).

FIG. 1. Effect of incubation time and substrate concentration on the activity of T. cruzi O-␣-GlcNAc-transferase. T. cruzi O-␣-
GlcNAc-transferase activity, in the standard assay, as a function of (A) incubation time at different temperatures (4, 20, 28, 37°C); B, enzyme concentration, reported as the amount of protein in the microsomal preparation; C, amount of synthetic peptide acceptor KPPTTTTTTT-KPP; D, transferase activity was assayed at different pH levels in Tris-MES buffer. The assays were performed as described under "Experimental Procedures." hydrolysis (CL-Brener strain) or Smith degradation (Y, CL-Brener, and Dm28c strains) was determined by high field NMR spectroscopy. The spectra from all these samples were effectively identical. Fig. 5 shows expansions of the TOCSY (Fig. 5A) and ROESY (Fig. 5B) of the GlcNAc-rich glycopeptide from Y strain. The resonances between 4.4 and 5.4 ppm (Fig. 5A) were assigned to saccharide anomeric protons. The most intense of these originate from the O-linked glycans. Those between 4.44 and 4.53 ppm displayed a pattern of cross-peaks in the TOCSY spectrum typical of ␤-Galp spin systems, ( 3 J H3,H4 small and H-4 at low field) (9). The anomeric protons from the GlcNAc O-linked to Thr resonated between 4.75 and 4.95 ppm, with the most intense at 4.75 ppm. This chemical shift dispersity is presumably attributable to heterogeneous substitution by Glc-NAc on the peptide backbone and to the presence of galactosylation of some GlcNAc residues.
Because glycopeptide from strain Dm28c was available in greatest quantity, WHSQC spectra with and without heteronuclear decoupling were recorded from this material (Fig. 6), which enabled partial assignment of the 13 C spectrum and determination of the value of one bond proton carbon-coupling constants ( 1 J C1,H1 ). N-Acetyl amino sugars are easily recognized by the high field positions of the C-2 resonance in their 13 C spectra, which are observed at approximately 51.4 and 56.1 ppm for the methyl glycosides of GlcNAc and GalNAc, respectively (27). In the WHSQC spectrum of the GlcNAc-rich glycopeptide from Dm28c, the resonance at 54.6 ppm was assigned as C-2 of GlcNAc and was correlated to an H-2 resonance at 3.85 ppm, which in turn was correlated to anomeric proton resonances between 4.75 and 4.95 ppm in the TOCSY (Fig. 5A) and COSY spectra, which were thus unambiguously assigned as H-1 of GlcNAc. The corresponding anomeric carbons resonated between 99.5 and 99.9 ppm (from the same WHSQC spectrum) (Fig. 6, A and B).
In the ROESY spectrum (Fig. 5B), NOEs were observed from the most intense GlcNAc H-1 resonance at 4.75 ppm and the N-acetyl methyl resonances to H␤ and methyl resonances of Thr, indicating linkage of the GlcNAc to Thr in the peptide backbone.
Some minor resonances, not attributable to the O-linked glycans, were also present. One set had chemical shifts for the H-1, C-1, and other structural reporter groups consistent with those reported for high mannose chains (28,29), although only a Man 2 (GlcNAc) 2 stub was present in the Smith-degraded samples (Fig. 5A).
The anomeric configuration of the GlcNAc residue was assigned from the values of the 3 J H1,H2 and 1 J H1,C1 coupling constants, the pattern of intra-residue NOEs, and the proton and the carbon shifts of the GlcNAc residues. The value of 3 J H1,H2 in the GlcNAc residue was estimated as 3-4 Hz from resolved cross-peaks observed in the ROESY spectrum (Fig.  7C). This is consistent with the ␣-configuration (H-1 and H-2 gauche) (30), as a value of 8 Hz would be expected for GlcNAc␤13OThr (in which H-1 and H-2 have a trans diaxial orientation), as was observed for the GlcNAc␤3 Asn present in the stub of the high mannose oligosaccharide chain (Fig. 7A). The value of 1 J H1,C1 for the GlcNAc13OThr was determined from the WHSQC spectrum, obtained at 600 MHz without heteronuclear decoupling (Fig. 6B). The magnitude of this coupling constant depends principally on the orientation of the anomeric proton and is typically 160 and 170 Hz for axial and equatorial H-1 values, respectively (i.e. for ␤ and ␣ linkages in the case of a D-hexopyrananose in the 4 C 1 conformation) (31,32). A value of 171 Hz was observed for 1 J H1,C1 of GlcNAc H-1 (Fig. 6A), confirming an ␣ linkage whereas for the ␤ -Galp resonances the measured value was 161 Hz.
The only significant NOE observed between the GlcNAc H-1 to other protons on the saccharide ring was to the H-2, which is characteristic of the ␣-glyco configuration. In contrast, ␤-glyco No data for GlcNAc␣13OThr containing model compounds have been reported, but in glycopeptides containing GlcNAc-␤13O-linked to Ser (33,34) or Thr (34,35), the chemical shift of the GlcNAc H-1 is in the range 4.40 to 4.57 ppm, clearly different from the values observed in the T. cruzi glycopeptides. The shifts for the ␣-GlcNAc H-1 resonances are unusually high field compared with ␣-GlcNAc in oligosaccharides (where the value is typically Ͼ5 ppm) but are consistent with the high field location of the GalNAc␣13OThr anomeric proton in model systems for mammalian mucin, where published values are in the range 4.87-4.92 ppm (36,37). The observed values for the GlcNAc residue in the T. cruzi mucin are thus in better agreement with an ␣rather than a ␤-linkage. Other proton chemical shifts were in agreement with this assignment. The carbon chemical shifts of the GlcNAc C-1 resonances were in the range 99.5-99.9 ppm (Fig. 6), as expected for the ␣-configuration. DISCUSSION We have previously characterized a novel series of O-Glc-NAc-linked oligosaccharides from surface sialoglycoproteins of T. cruzi (9,10). The first step in the biosynthesis of these O-glycan chains is attachment of GlcNAc to the peptide backbone. In the present study, we show that microsomal membrane preparations from epimastigotes and trypomastigotes of T. cruzi Y strain have an O-␣-GlcNAc-transferase that attaches GlcNAc to threonine in a suitable acceptor peptide. We used several approaches to define the activity of this enzyme and to assess its relationship to the cytosolic O-GlcNAc-transferase (14,38).
We demonstrate that the synthetic peptide KPPTTTTTTTT-KPP can function as acceptor for this novel transferase. This peptide incorporates a threonine-rich sequence, originally reported by DiNoia et al. (26), who showed that TTTTTTTTKPP is a common repeating motif in T. cruzi MUC gene products. The locations of the glycosylated threonines were not identified, but the peptide was a good substrate for in vitro glycosylation, at least two GlcNAc residues being incorporated in heavily labeled experiments. The nonapeptide YSDSPSTST, described by Haltiwanger et al. (38) as an optimum substrate for the cytosolic O-GlcNAc-transferase, was not glycosylated by the T. cruzi enzyme. UDP-[ 3 H]GlcNAc was the activated Glc-NAc donor in the T. cruzi system. Potentially, the mechanism of the reaction could be either direct transfer from UDP-Glc-NAc or via formation of activated dolichol donors. To distinguish between these possibilities, the transferase activity was assayed in the presence of excess UDP and with the antibiotics tunicamycin or amphomycin, which are potent inhibitors of Dol-P-dependent glycosylation (39,40). Only UDP abolished incorporation of [ 3 H]GlcNAc into the acceptor peptide, indicating that UDP-GlcNAc acts directly as the GlcNAc donor. The cytosolic O-GlcNAc-transferase (38) also uses sugar nucleotides directly as sugar donors.
The T. cruzi enzyme, in common with most glycosyltransferases (41), requires divalent metal cations for activity, with Mn 2ϩ being the most effective. This is in contrast to the cytosolic O-GlcNAc-transferase, which shows no metal ion dependence (38). Other differences between the T. cruzi microsomal enzyme and the ubiquitous cytosolic O-GlcNAc-transferase are that the former has an optimal pH range of 7.5-8.5, remains active at 37°C, and has increased activity when treated with Triton X-100. Most strikingly, the T. cruzi enzyme attaches GlcNAc to the hydroxylated amino acid via an ␣-linkage, whereas the anomeric specificity of the cytosolic enzyme is ␤. In support of this, the [ 3 H]GlcNAc peptide (produced in vitro) was not susceptible to digestion of with ␤-N-acetylglucosaminidase from jack beans, although under identical digestion conditions GlcNAc was readily liberated from unlabeled synthetic YSPTSPSK (with the O-␤-GlcNAc on the serine at position 5), corresponding to a sequence from the C-terminal repeat domain of the large subunit of RNA polymerase II (42), which in vivo is glycosylated by the cytosolic transferase. It is unlikely that the lack of susceptibility of the [ 3 H]GlcNAc-K 2 P 4 T 8 peptide to ␤-N-acetylglucosaminidase is attributable to its amino acid sequence, as the jack bean enzyme is able to deglycosylate a diverse range of structurally distinct glycoproteins and glycopeptides, including sequences from nuclear pore protein, human erythrocyte band 4.1 protein, and the 65-kDa erythrocyte cytosolic protein (38,42). More compellingly, NMR analysis of GlcNAc-rich glycopeptides from the native sialoglycoconjugates of T. cruzi showed unambiguously that the GlcNAc residue linked to Thr has the ␣-anomeric configuration, the 1 J C, H , NOE, and chemical shift data all being consistent with the ␣rather than the ␤-anomer.
FIG. 7. Expansion of selected cross-peaks from TOCSY and ROESY spectra of T. cruzi GlcNAc-rich glycopeptide. Partial 500 MHz proton NMR spectra of selected cross-peaks, showing the 8 Hz 1 J H1,H2 coupling for the 34GlcNAc␤13 NAsn residue of the high mannose chain (A). B, the unresolved 1 J H1,H2 coupling of Man␣13, typically 1.8 Hz; C, the 3 J H1,H2 coupling of the GlcNAc␣131OThr, estimated at 3-4 Hz. A and B are expansions of the TOCSY spectrum of GlcNAc-rich glycopeptides obtained from partial acid hydrolysis of CL-Brener strain sialoglycoproteins, whereas C is from the ROESY spectrum of the same compound. All spectra were collected at 30°C. NAc is either absent or present only at very low levels in P. falciparum (45) and that glucose rather than N-acetylglucosamine is the O-linked sugar in G. lamblia (46). Because the O-GlcNAc-transferase activity of T. cruzi is associated with the microsomal fraction, and because its known natural substrates are GPI-anchored N-linked surface glycoproteins (10), it seems likely that it is associated with some compartment of the secretory pathway, although confirmation of this must await completion of detailed localization studies. Because the enzyme differs in its anomeric specificity, kinetic properties, and possibly in cellular location from the cytosolic enzyme described in higher eukaryotes, it may prove to be unique to T. cruzi. If so, it constitutes an exciting potential target for the rational design of novel chemotherapeutic agents. Purification of the transferase is currently in progress and should soon enable these questions to be addressed.