INTRODUCTION

The trafficking of lysosomal hydrolases to the lysosome in higher eucaryotes depends on the specific modification of asparagine-linked oligosaccharides to contain a mannose 6-phosphate recognition marker. The initial and determining step in the generation of the mannose 6-phosphate recognition marker is catalyzed by the enzyme UDP-N-acetylglucosamine:lysosomal-enzyme N-acetylglucosamine-1-phosphotransferase (GlcNAc-phosphotransferase).¹ In the previous paper (1), we described the 488,000-fold purification of GlcNAc-phosphotransferase from lactating bovine mammary glands. The purified enzyme was found to be a complex of six subunits and is composed of homodimers of 166-, 56-, and 51-kDa subunits. The identification of GlcNAc-phosphotransferase as a multisubunit enzyme (1) suggests that it may be possible to link specific properties of the enzyme to specific protein subunits.

Previous studies have shown partially purified rat liver GlcNAc-phosphotransferase phosphorylates lysosomal enzymes at least 500-fold better than non-lysosomal glycoproteins with similar high mannose oligosaccharides (2). Isolated high mannose oligosaccharides were poor substrates, as were heat-denatured lysosomal enzymes (2). These studies have resulted in a model that proposes that GlcNAc-phosphotransferase recognizes a conformationally sensitive protein determinant found on lysosomal enzymes. The structure of this determinant in the lysosomal enzyme cathepsin D has been characterized (3, 4, 5). GlcNAc-phosphotransferase partially purified from the soil amoeba Acanthamoeba castellanii is also selective for a protein determinant on lysosomal enzymes (6).

In this paper, we have investigated the enzymatic properties and kinetics of GlcNAc phosphate transfer to lysosomal and non-lysosomal glycoproteins by a homogeneous preparation of bovine GlcNAc-phosphotransferase. Our results demonstrate that bovine GlcNAc-phosphotransferase selectively phosphorylates lysosomal enzymes. That this selectivity, a property previously observed with impure preparations, is also found with isolated GlcNAc-phosphotransferase demonstrates that it is a property of the enzyme itself and not an accessory factor. We also demonstrate, using photoaffinity labeling with 5-N₃-[-³²P]UDP-Glc, that the 166-kDa subunit contains the nucleotide sugar-binding site.

EXPERIMENTAL PROCEDURES

Materials

Scintiverse BD was from Fisher (Pittsburgh, PA). ConA-Sepharose and HiTrap NHS-activated columns were obtained from Pharmacia Biotech Inc. Ribonuclease B and -methylmannoside were obtained from Sigma. Porcine uteroferrin was the kind gift of Dr. R. Michael Roberts (University of Missouri, Columbia, MO). All other reagents were reagent grade or better and were from standard suppliers.

Methods

Most of the experiments described were performed with GlcNAc-phosphotransferase that had been purified on PT18-3M-Emphaze according to Method II (1). In some cases, the purified enzyme was further concentrated by chromatography on Mono Q. All these preparations had a specific activity of 10-12 µmol/mg/h and were homogeneous when examined on silver-stained SDS-polyacrylamide gels. The photoaffinity labeling experiments were performed with GlcNAc-phosphotransferase purified according to Method I (1). This preparation had a specific activity of 200 nmol/mg/h and was heterogeneous on silver-stained SDS-polyacrylamide gels.

UDP-GlcNAc (75 nmol) was incubated with 15 µmol of -methylmannoside in the presence of 50 mM Tris-HCl, pH 7.4, 5 mM MgCl₂, and 5 mM MnCl₂. GlcNAc phosphotransferase (10 µg, 100,000 units) was added, and the reaction was incubated at 37 °C for 18 h. A second reaction containing, in addition to the other components, 1 µCi of [-³²P]UDP-GlcNAc was processed in parallel to calibrate the columns used for product isolation. The reaction was stopped by the addition of 1.5 ml of 5 mM EDTA, pH 8.0, and applied to a 1-ml column of QAE-Sephadex A-25 equilibrated with 2 mM Tris base. The column was washed with 2 volumes of equilibration buffer, and the product was eluted with 2 ml of buffer containing 30 mM NaCl. The product was made 1 M in NH₄COOCH₃ and desalted on a 1.6 × 50-cm column of Bio-Gel P-2 equilibrated with water. The pooled product was concentrated by rotary evaporation and subjected to mass spectroscopy.

The mass of the enzymatic product was determined by MALDI-TOF-MS on a Hewlett-Packard LDI 1700XP mass spectrometer, which was operated at an accelerating voltage of 30 kV, an extractor voltage of 9 kV, and a pressure of 1.7 × 10⁶ torr. Polarity was positive. The dried product was dissolved in 100 mM 2,5-dihydroxybenzoic acid, 90% (v/v) methanol. Approximately 1 µl of the sample/matrix was placed on the probe tip and vacuum-crystallized. Samples were desorbed/ionized from the probe tip with a nitrogen laser ( = 337 nm) with a pulse width of 3 ns and delivering ~10.5 µJ of energy/laser pulse. The mass spectrum was recorded over a mass to charge (m/z) range of 1,000-2,500. The spectrum was averaged over 27 laser shots and plotted as arbitrary intensity versus m/z.

Experiments using -methylmannoside as acceptor were performed as described (7), except that the [-³²P]UDP-GlcNAc was isolated by HPLC as described (1), and the radioactivity in each assay was increased to 0.5 µCi. In experiments utilizing glycoprotein acceptors, reaction mixtures containing 0.5-1.0 µCi of [-³²P]UDP-GlcNAc were applied to a 1.0-ml column of ConA-Sepharose previously equilibrated with Tris-HCl, pH 7.6, 1 mM CaCl₂, 1 mM MgCl₂, and 0.3% Lubrol. Following washing with 20 ml of buffer, the resin was extruded into a scintillation vial and counted with 12 ml of Scintiverse BD. To avoid exceeding the capacity of the ConA-Sepharose columns in experiments using high concentrations of glycoprotein acceptors, only a fraction of the reaction mixture was applied to the ConA-Sepharose column.

For experiments using UDP-Glc as donor, [-³²P]UDP-Glc was synthesized similarly to [-³²P]UDP-GlcNAc (1) and purified by HPLC. The product had a specific activity of 4 mCi/µmol and was stored in 50% ethanol at 20 °C.

Reaction mixtures contained 5 mM MgCl₂, 5 mM MnCl₂, 1 mg/ml bovine serum albumin, 1 mM DTT, and 150 µM [-³²P]UDP-GlcNAc. Each reaction was buffered to the indicated pH with a 250 mM concentration of one of the following buffers: sodium acetate, pH 3.9-4.25; MES/NaOH, pH 4.9-6.4; BisTris-HCl, pH 6.0-6.9; and Tris-HCl, pH 6.9-9.4.

Assay of N-Acetylglucosamine-1-phosphodiester -N-Acetylglucosaminidase (Uncovering Enyzme)

[³H]GlcNAc-1-phospho-6-mannose -methyl was prepared at a specific activity of 2.1 µCi/µmol as described (8). Uncovering enzyme activity was determined using 1 µg of purified GlcNAc-phosphotransferase as described (8).

Uteroferrin was used without further purification. Bovine pancreatic ribonuclease B was obtained commercially and chromatographed on ConA-Sepharose to remove contaminating ribonuclease A as described previously (2). Porcine cathepsin D was purified from frozen spleens as described previously (9). The specific activity was 30 units/mg, and the protein was >90% cathepsin D on silver-stained SDS-polyacrylamide gels.

Proteins were quantitated by absorbance at 280 nm using the following molar extinction coefficients: ribonuclease B, 8,756 (10); cathepsin D, 20,991 (2); and uteroferrin, 31,000 (at 545 nm) (11). Protein concentrations for all other proteins including GlcNAc-phosphotransferase were estimated assuming E_1 cm^1% = 10.0.

Photoaffinity Labeling of GlcNAc-phosphotransferase with [-³²P]UDP-Glc

5-N₃-[-³²P]UDP-Glc was synthesized at a specific activity of 10 mCi/µmol as described previously (12, 13), purified by chromatography of the reaction mixture on a column of DEAE-cellulose, and stored in the dark at 20 °C in methanol. Samples were prepared as indicated in the legend to Fig. 5. Following a 15-s incubation, the indicated samples were irradiated with a hand-held UV lamp with the glass face removed (5,000 microwatts/cm²; Model UVS-11, Ultra-Violet Products) at a distance of 4 cm for 45 s. Reactions were terminated by the addition of 0.1 volume of 100% trichloroacetic acid and incubation for 30 min at 0 °C. The precipitated protein was collected by centrifugation at 15,000 × g for 10 min, and the pellets were dissolved in 25 µl of SDS-PAGE sample buffer by boiling for 5 min at 100 °C and analyzed on an SDS-6% polyacrylamide gel (14). The dye front containing the unreacted 5-N₃-[-³²P]UDP-Glc was cut off and discarded. The remaining gel was stained with Coomassie Blue, dried, and analyzed by autoradiography.

Photolabeling of GlcNAc-phosphotransferase with 5-N₃-[-³²P]UDP-Glc.

Affinity-purified polyclonal antibodies were prepared against 488,000-fold purified bovine GlcNAc-phosphotransferase. A New Zealand White rabbit was primed by intradermal injection of 20 µg of GlcNAc-phosphotransferase emulsified in Freund's complete adjuvant. Boosting was at 2-week intervals with 20 µg of GlcNAc-phosphotransferase emulsified in Freund's incomplete adjuvant. Blood was obtained from the central artery of the ear; antiserum was prepared; and immunoglobulins were isolated by precipitation with 50% saturation of ammonium sulfate. The pellet was dissolved in a minimal volume of phosphate-buffered saline and dialyzed against phosphate-buffered saline. GlcNAc-phosphotransferase-specific antibodies were isolated by chromatography of the crude immunoglobulin fraction on a HiTrap NHS-activated column (1 ml) upon which 200 µg of GlcNAc-phosphotransferase had been immobilized according to the manufacturer's instructions. The column was then washed with phosphate-buffered saline followed by water and eluted with 1 mM HCl. The eluted affinity-purified antibody was immediately neutralized with 1 M Tris-HCl, pH 9.0, and concentrated to 1 mg/ml in a Centriprep 30 concentrator, made 1 mM in NaN₃, and stored at 4 °C.

RESULTS

The structure of the product of the transfer reaction catalyzed by GlcNAc-phosphotransferase between the donor UDP-GlcNAc and the acceptor -methylmannoside was determined by MALDI-TOF-MS. A preparative scale reaction containing 75 nmol of UDP-GlcNAc and 15 mM -methylmannoside was incubated with 100,000 units of GlcNAc-phosphotransferase, and the reaction product was isolated as described under "Experimental Procedures." MALDI-TOF-MS of the enzymatic product identified four major species at m/z ratios between 390 and 550 (Fig. 1). The m/z ratios of the identified peaks are 498.3, 520.1, 535.8, and 552.4. These masses correspond to the following species, where M corresponds to the molecular ion: [M + Na]⁺, [M + 2Na  H]⁺, [M + Na + K  H]⁺, and [M + 2K  H]⁺. From these data, the mass of the molecular ion was determined to be 474-475, corresponding to the mass of the expected product, GlcNAc-1-phospho-6-mannose -methyl.

Purified GlcNAc-phosphotransferase was stable when stored at 4 °C or frozen at 80 °C in Tris buffer at pH 7-8 containing MgCl₂ and Lubrol. More than 80% of the activity remained after storage for 2 months at 4 °C.

Bovine GlcNAc-phosphotransferase was active between pH 5.7 and 9.4 (Fig. 2). With -methylmannoside as acceptor, the enzyme demonstrated peak activity between pH 6.7 and 7.5. Activity was independent of buffer composition for the buffers tested. Remarkably, the enzyme activity increased 3-fold in the narrow range between pH 6.0 and 6.6, an effect observed in both MES and BisTris buffers. Although the enzyme was only active between pH 5.7 and 9.4, the enzyme was stable within the pH range of 5-11. Assays were routinely performed at pH 7.4, where the enzyme demonstrates near maximal activity.

Table I shows the effects of potential activators or inhibitors on the activity of the enzyme measured with -methylmannoside as acceptor. GlcNAc-phosphotransferase required divalent cations for activity and was inactive in the presence of 10 mM EDTA. Mn²⁺ was ~20% more effective than Mg²⁺. The enzyme was minimally active in the presence of Ca²⁺. The affinity of the enzyme for divalent cations appeared to be low since no activity was observed upon dilution of the enzyme in buffer containing 5 mM MgCl₂ into an assay buffer lacking added divalent cations. The K_m for Mn²⁺ was determined to be 185 µM in reactions containing 0.05 M Tris-HCl, pH 7.4, 150 µM UDP-GlcNAc, 100 mM -methylmannoside, 2 mM DTT, and 0.3% Lubrol. Reducing agents and iodoacetamide had little effect on the enzyme activity. The enzyme was inhibited 39-74% by 5 mM ATP, ADP, UTP, or UDP, an effect that was not reversed by increasing the divalent cation concentration. Although the enzyme was inhibited 59% by 5 mM ATP, the enzyme was not inhibited by the 2 mM ATP present in the standard assay. UMP, a product of the enzymatic reaction, was without effect at 5 mM. UDP-galactose was without effect, but activity was inhibited 64% by 5 mM UDP-glucose. Mannose 6-phosphate or glucose 6-phosphate was individually modestly stimulatory. In contrast to the A. castellanii GlcNAc-phosphotransferase (15), the bovine enzyme was inhibited by ATP and ADP, effects not observed for the amoeba enzyme. The bovine enzyme was also significantly more inhibited by UDP than the amoeba enzyme.

Table I. Effects of potential activators and inhibitors on GlcNAc-phosphotransferase Enzyme assays were performed in triplicate using 100 mM -methylmannoside and 100 ng (1000 units) of purified bovine GlcNAc-phosphotransferase. Each reaction mixture contained 50 mM Tris-HCl, pH 7.4, 2 mg/ml bovine serum albumin, 0.15 mM UDP-GlcNAc, and 1 µCi of [-32P]UDP-GlcNAc. In addition, compounds 1-7 were assayed in the presence of 2 mM ATP, 1 mM dithiothreitol, and 50 mM N-acetylglucosamine. Compounds 8-10 were assayed in the presence of 5 mM ATP, 10 mM MgCl2, 10 mM MnCl2, and 50 mM N-acetylglucosamine. Compounds 11-20 were assayed in the presence of 10 mM MnCl2, 10 mM MgCl2, 1 mM dithiothreitol, and 50 mM N-acetylglucosamine. Compound 22 was assayed in the presence of 5 mM ATP, 10 mM MgCl2, 10 mM MnCl2, and 1 mM dithiothreitol. Activities are expressed as percent of the appropriate controls. Compound Concentration Activity mM % control  1. No additions 0  2. MgCl2/MnCl2 5 100  3. MgCl2 10 85  4. MnCl2 10 100  5. MgSO4 10 79  6. CaCl2 5 0.8  7. EDTA 10 0  8. DTT 2.5 123  9. 2-Mercaptoethanol 10 96 10. Iodoacetamide 1.5 114 11. ATP 5 41 12. ADP 5 61 13. AMP 5 118 14. UTP 5 66 15. UDP 5 26 16. UMP 5 100 17. UDP-galactose 5 95 18. UDP-glucose 5 34 19. Mannose 6-phosphate 5 134 20. Glucose 6-phosphate 5 126 21. Sodium phosphate, pH 7.0 5 45 22. GlcNAc 100 92

The kinetic properties of bovine GlcNAc-phosphotransferase with UDP-GlcNAc and -methylmannoside have been examined in detail. The K_m of the enzyme for UDP-GlcNAc was 30 µM (Fig. 3A), while the K_m for -methylmannoside was 64 mM (Fig. 3B). The K_m of the bovine enzyme for UDP-GlcNAc (30 µM) is similar to the values obtained with the rat liver enzyme (38 µM) (7) and the amoeba enzyme (43 µM) (6). The K_m value of the bovine enzyme for -methylmannoside (63 mM) is somewhat lower than the corresponding values for the rat liver (158 mM) and amoeba (568 mM) enzymes. The functional significance of this higher affinity for the synthetic acceptor is unclear.

Dependence of GlcNAc-phosphotransferase activity on the concentration of UDP-GlcNAc (A) and -methylmannoside (B).

Optimal GlcNAc-phosphotransferase activity was observed in reactions containing 0.05 M Tris-HCl, pH 7.4, 150 µM UDP-GlcNAc, 100 mM -methylmannoside, 5 mM MnCl₂, 5 mM MgCl₂, 2 mM DTT, and 1 mg/ml bovine serum albumin.

The inhibitory effects of UDP-glucose were also examined in detail (Fig. 4). Analysis of the kinetics of the inhibition of GlcNAc-phosphotransferase by UDP-Glc demonstrated that inhibition was strictly competitive, with a K_i of 733 µM. In reactions containing [-³²P]UDP-Glc instead of [-³²P]UDP-GlcNAc, the transfer of Glc 1-phosphate to -methylmannoside was demonstrated. Attempts to define the enzyme kinetics with UDP-Glc as donor were not possible because the [-³²P]UDP-GlcNAc was unstable under the assay conditions. Since UDP-Glc is utilized by the enzyme for a productive transfer, the "inhibition" is somewhat illusionary. The demonstration of strictly competitive inhibition of GlcNAc-phosphotransferase by UDP-Glc indicates that the two nucleotide sugars compete for a single nucleotide sugar-binding site.

With the identification of GlcNAc-phosphotransferase as a multisubunit enzyme, it was of interest to determine if the complex also contained the other enzyme involved in the biosynthesis of the phosphomannosyl recognition marker, uncovering enzyme. Uncovering enzyme activity was not detectable in the purified GlcNAc-phosphotransferase.

To identify the subunit(s) of GlcNAc-phosphotransferase containing the nucleotide sugar-binding site, photoaffinity labeling experiments were performed. Attempts to synthesize azido-UDP-GlcNAc using 5-N₃-UTP, glucosamine 1-phosphate, and yeast UDP-glucose pyrophosphorylase were unsuccessful (data not shown). Since kinetic analysis indicated that, in addition to functioning as a competitive inhibitor, UDP-Glc could also function as a substrate with a K_m of 110 µM, ~2-fold higher than the K_m for UDP-GlcNAc, 5-N₃-[-³²P]UDP-Glc was used to identify the nucleotide sugar-binding site. A protein of ~170 kDa was photolabeled on reduced SDS-polyacrylamide gels in a reaction that was UV-dependent and blocked by UDP-Glc and UDP-GlcNAc (Fig. 5). The effective competition by UDP-GlcNAc or UDP-Glc suggests that the binding site in the photolabeled protein has properties similar to the substrate specificity determined for GlcNAc-phosphotransferase. The photolabeling was not affected by the presence of -methylmannoside in the reaction. Increasing the concentration of GlcNAc-phosphotransferase 3-fold resulted in a comparable increase in the photolabeled product. Additional minor bands of unknown significance were identified at the highest concentration of GlcNAc-phosphotransferase. The specific absence of photoaffinity labeling of the other GlcNAc-phosphotransferase subunits (56 and 51 kDa) was noted, implying that these subunits do not contain nucleotide sugar-binding sites. On nonreduced SDS-polyacrylamide gels, a protein with a molecular mass of >212 kDa was labeled, consistent with the disulfide-linked homodimer composed of 166-kDa subunits. Again, the reaction was inhibited by excess UDP-GlcNAc or UDP-Glc. These results indicate that the nucleotide sugar-binding site in GlcNAc-phosphotransferase is located within the 166-kDa subunit and suggest that this is the catalytic subunit.

Bovine GlcNAc-phosphotransferase was assayed during purification by monitoring the transfer of GlcNAc-1-P to the synthetic acceptor -methylmannoside. Since the endogenous acceptor is the high mannose oligosaccharide on lysosomal hydrolases, it was of interest to determine if the purified bovine GlcNAc-phosphotransferase was also able to utilize glycoprotein acceptors. The lysosomal glycoproteins uteroferrin and cathepsin D and the non-lysosomal glycoprotein ribonuclease B were investigated as substrates for the purified bovine GlcNAc-phosphotransferase. Uteroferrin was an effective substrate and was highly phosphorylated even with acceptor concentrations in the low micromolar range. The extent of phosphorylation was also high, with ~30% of the molecules phosphorylated at 23 µM uteroferrin (the K_m). Cathepsin D was also an acceptor, with a K_m of 18 µM, but the extent of transfer was low, suggesting that many of the molecules did not contain suitable oligosaccharide structures to function as substrates (19). In contrast, ribonuclease B, although containing oligosaccharides capable of being phosphorylated (predominantly Man₈GlcNAc₂), was a poor substrate even at very high protein concentrations (Fig. 6).

The kinetic properties of the various acceptors are summarized in Table II. The lysosomal enzymes uteroferrin and cathepsin D were both excellent acceptors for bovine GlcNAc-phosphotransferase, with K_m values of 23 and 18 µM, respectively. The catalytic efficiency (V_max/K_m) of uteroferrin was 162-fold better than that of ribonuclease B. This is in spite of the fact that ribonuclease B contains three -1,2-linked mannoses, while uteroferrin contains at most a single -1,2-linked mannose.

Table II. Kinetic parameters of GlcNAc-phosphotransferase activity on various acceptors Acceptor Km Vmax Vmax/Km Relative catalytic efficiencya µM pmol/h/mg  -Methylmannoside 63,700 12    × 106 188 1 RNase B 1,244 1.5  × 106 1,210 6.4 Uteroferrin 22 4.3  × 106 195,000 1,040 Cathepsin D 18 NDb ND ND a  Catalytic efficiencies (Vmax/Km) were calculated relative to -methylmannoside = 1. b  ND, not determined.

When GlcNAc-phosphotransferase was incubated with excess affinity-purified rabbit anti-bovine GlcNAc-phosphotransferase, a differential effect on the transfer of GlcNAc phosphate to -methylmannoside and cathepsin D was observed. Transfer to -methylmannoside was inhibited 35%. In contrast, transfer to cathepsin D was completely blocked even at high concentrations of cathepsin D (Fig. 7). These results suggest that a subpopulation of the antibodies tested were able to selectively inhibit the phosphorylation of cathepsin D without interfering with transfer to -methylmannoside. These findings suggest that it may be possible to generate either monoclonal antibodies or subunit-specific polyclonal antibodies with similar properties to identify subunit(s) that interact with glycoprotein acceptors.

DISCUSSION

In this study, we used a homogeneous preparation of GlcNAc-phosphotransferase isolated from the lactating bovine mammary gland to investigate the enzymatic properties toward various acceptors. The data presented in this paper demonstrate that the bovine enzyme phosphorylates mammalian lysosomal enzymes better than non-lysosomal enzymes with similar oligosaccharide structures, similar to previous findings with partially purified rat (2, 16) and A. castellanii (6) enzymes. These data extend the previous findings by demonstrating that this selectivity for lysosomal enzymes is a property of the purified GlcNAc-phosphotransferase complex and not the result of other factors or proteins.

The selectivity of bovine GlcNAc-phosphotransferase for lysosomal enzymes results from two factors. First, the K_m for uteroferrin is ~56-fold lower than for ribonuclease B. Second, the V_max for uteroferrin is ~3-fold greater than for ribonuclease B. Together, these two factors result in a calculated catalytic efficiency ~162-fold greater for the lysosomal enzyme acceptor. The use of lysosomal enzymes purified from lysosomes as acceptors is associated with inherent difficulties since the oligosaccharides have frequently been truncated by glycosidases present in the lysosome. The single oligosaccharide of uteroferrin is composed predominantly of Man₅GlcNAc₂ structures, with lesser amounts of Man₆GlcNAc₂ and Man₄GlcNAc₂ (17). Since -1,2-linked mannoses are absolutely required for an oligosaccharide to function as an acceptor for GlcNAc-phosphotransferase (18), only the uteroferrin molecules bearing the Man₆GlcNAc₂ structures can function as acceptors, while the remaining molecules are competitive inhibitors. Because of these limitations in uteroferrin as an acceptor, the true difference between uteroferrin and ribonuclease B is likely to be at least 4-fold greater than determined. Ribonuclease B contains predominantly Man₈GlcNAc₂ structures and should provide an optimal oligosaccharide to function as an acceptor. The improved catalytic efficiency of ribonuclease B compared with that of -methylmannoside likely results from the presence of the preferred -1,2-linked mannose acceptor. Cathepsin D was also an effective substrate as indicated by a K_m of 18 µM; however, the extent of labeling was low, likely as a result of the highly truncated oligosaccharides present on most molecules (19). The amount of cathepsin D available prevented fractionation into specific glycoforms that would be expected to function as more efficient acceptors.

With the identification of bovine GlcNAc-phosphotransferase as a multiple subunit enzyme, the association of specific functions with specific protein subunit(s) becomes possible. UDP-Glc was found to be a competitive inhibitor/alternate substrate for bovine GlcNAc-phosphotransferase, similar to results previously reported for the rat enzyme (7). The finding of strictly competitive inhibition indicates that UDP-Glc and UDP-GlcNAc compete for the same nucleotide sugar-binding site. This observation allowed the identification of the subunit containing the nucleotide sugar-binding site with the photoprobe 5-N₃-UDP-Glc. The 166-kDa subunit is specifically photoaffinity-labeled with the photoprobe 5-N₃-[-³²P]UDP-Glc, indicating that the nucleotide sugar-binding site is localized to this subunit.