J Biol Chem, Vol. 274, Issue 51, 36392-36402, December 17, 1999

Identification of a UDP-GlcNAc:Skp1-Hydroxyproline GlcNAc-transferase in the Cytoplasm of Dictyostelium^*

Patana Teng-umnuay, Hanke van der Wel, and Christopher M. West^§

From the Department of Anatomy and Cell Biology, University of Florida College of Medicine, Gainesville, Florida 32610-0235

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Skp1 is a cytoplasmic and nuclear protein required for the ubiquitination of cell cycle regulatory proteins and transcriptional factors. In Dictyostelium, Skp1 is modified by a linear pentasaccharide, Gal1-6Gal1-Fuc1-2Gal1-3GlcNAc, attached to a hydroxyproline (HyPro) residue at position 143. To study the formation of the GlcNAc-HyPro linkage, an assay was developed for the transfer of [³H]GlcNAc from UDP-[³H]GlcNAc to Skp1-HyPro-143 or a synthetic Skp1 4-HyPro peptide. The cytosolic but not the particulate fraction of the cell mediated transfer in a time-, concentration-, and HyPro-dependent fashion. Incorporated radioactivity was alkali-resistant and was recovered as GlcNH₂ after acid hydrolysis, consistent with linkage of GlcNAc to HyPro. The GlcNAc-transferase activity was purified 130,000-fold as a single component with a recovery of 5%. Key to the purification was the synthesis of a novel affinity resin linking UDP-GlcNAc at its 5-uridyl position. The purified activity had an apparent M_r of ~45,000 by gel filtration, required dithiothreitol and a divalent cation, and consisted predominantly of a M_r 51,000 band after SDS-polyacrylamide gel electrophoresis that was photoaffinity labeled with 5-¹²⁵I-[3-(p-azidosalicylamido)-1-propenyl-UDP-GlcNAc in a UDP-GlcNAc-sensitive fashion. Its apparent K_m values for UDP-GlcNAc and Skp1 were submicromolar. The presence of the enzyme in the cytosolic fraction, its dependence on a reducing environment, and its high affinity for UDP-GlcNAc strongly suggest that Skp1 is glycosylated by a HyPro GlcNAc-transferase that resides in the cytoplasm.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Skp1 belongs to the SCF complex that is involved in the ubiquitination of cell cycle and other regulatory proteins including transcriptional factors (1-4). In Dictyostelium discoideum, Skp1 is encoded by two similar genes, fpa1 and fpa2 (5), and is found in both the cytoplasm and the nucleus based on immunofluorescence localization.¹ Dictyostelium Skp1 is modified by an unusual linear pentasaccharide, Gal1-6Gal1-Fuc1-2Gal1-3GlcNAc, attached to a HyPro² residue at amino acid position 143 (6). As the first glycosyltransferase in the modification of Pro-143, the hypothetical Skp1-HyPro GlcNAc-transferase is poised to control the glycosylation of Skp1 at this amino acid position. In support of such a regulatory role, a Skp1 isoform that is hydroxylated at Pro-143 but is not glycosylated can be detected when the fpa1 gene is overexpressed (6).

The Skp1-HyPro GlcNAc-transferase is likely to be distinct from previously described GlcNAc-transferases based on the novelty of the GlcNAc-HyPro linkage formed. Most known protein GlcNAc-transferases are compartmentalized within the lumen of the Golgi apparatus (7-10). However, a Ser/Thr GlcNAc-transferase has been described that functions in the cytoplasmic compartment (11-14). Although glycosylation of Skp1 in a compartment of its function, the cytoplasm, might be simplest, there is precedence for bidirectional trafficking of proteins across the membrane of the rER for secretion or degradation (15), which might also be utilized by Skp1. This more complex scenario would accommodate the hydroxylation of Pro-143, as all known prolyl hydroxylases are also located within the rER (16). Determining whether Skp1 is modified in the cytoplasm or a compartment of the secretory pathway is important for understanding the regulation and significance of this pathway.

Studies on another glycosyltransferase in the Skp1-HyPro modification pathway, a GDP-Fuc:Gal1-3HexNAc 1,2Fuc-transferase, strongly suggested that it is compartmentalized in the cytoplasm (17). This Skp1 1,2Fuc-transferase activity was purified to near-homogeneity from a cytosolic extract of Dictyostelium and was found to be absolutely dependent on DTT and Mg²⁺. Negligible activity could be detected in the vesicle fraction of the cell. The requirement for DTT implied that the enzyme normally functions in a reducing environment, i.e. the cytoplasm rather than the lumen of the secretory pathway. Furthermore, the 1,2Fuc-transferase exhibited submicromolar K_m values for GDP-Fuc and Skp1. These exceptionally high affinities, relative to those of Golgi enzymes, reinforced the cytoplasmic localization model, where substrates would not be as concentrated as they are in the Golgi. A similar high affinity for its donor substrate was reported for the cytoplasmic Ser/Thr GlcNAc-transferase (11). Although the compartmentalization of the Skp1 1,2Fuc-transferase and its role in modifying Skp1 remains to be confirmed by genetic studies, the findings suggest that Skp1 is modified by a novel, sequentially acting, six-enzyme pathway residing in the cytoplasm.

To investigate the properties and compartmentalization of the Skp1-HyPro GlcNAc-transferase, an assay was developed to guide its identification and purification. An activity has been detected in a cytosolic extract of Dictyostelium that appears to transfer GlcNAc from UDP-GlcNAc to Skp1-HyPro-143. By using a combination of conventional chromatography and affinity chromatography based on a resin containing a novel UDP-GlcNAc linkage, a candidate M_r 51,000 protein has been purified to near-homogeneity. The kinetic properties of the purified Skp1-HyPro GlcNAc-transferase activity and its absolute dependence on a reducing environment suggest, as for the Skp1 1,2Fuc-transferase activity described above, that the first glycosyltransferase in the Skp1 HyPro modification pathway also resides in the cytoplasm.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

GlcNAc-transferase Assay

Substrates-- Skp1A-Myc was purified from strain HW120 through the mAb 3F9 affinity column step as described (6). Recombinant Skp1A-His₁₀ was purified on a nickel chelating column as described (17) and further on a mAb 3F9 affinity column. Purified proteins were concentrated in Centriplus-10 ultrafiltration concentrators (Amicon) to about 10 µM, and aliquots were stored at 80 °C. Synthetic peptides corresponding to amino acids 133-155 of Skp1, KIFNIKDFTPEEEEQIRKENEW and KIFNIKNDFT(4-OH)PEEEQIRKENEW, were synthesized by the ICBR Protein Core at the University of Florida and verified by amino acid composition and matrix-assisted laser desorption-time of flight-mass spectrometry analyses. Peptides were purified by reversed-phase HPLC, dried, and dissolved in 50 mM HEPES-NaOH, pH 7.4, at 20 mg/ml. UDP-[³H-6]GlcNAc (NEN Life Science Products) had a specific activity of 34.8 Ci/mmol.

Assay-- GlcNAc-transferase activity was assayed by the transfer of [³H]GlcNAc from UDP-[³H]GlcNAc to Skp1A-Myc or peptide-(133-155). Typically, assays contained 1 µM Skp1A-Myc, 50 mM HEPES-NaOH, pH 7.8, 5 mM MgCl₂, 5 mM DTT, 0.5 mg/ml bovine serum albumin, and 0.3-0.66 µM UDP-[6-³H]GlcNAc in a final volume of 35 µl. 1 mM ATP and 6 mM NaF were included in assays of S100 (where S100 is the cytosolic cell fraction isolated as the supernatant after centrifugation at 100,000 × g for 1 h) and DEAE column fractions. Following addition of enzyme, reactions were incubated at 30 °C for 1 h. Reactions were stopped by addition of either 500 µl of ice-cold 10 mM sodium pyrophosphate in 10% (w/v) trichloroacetic acid, or 35 µl of 2× Laemmli electrophoresis sample buffer containing 60 µg/ml soybean trypsin inhibitor (Sigma). Reactions stopped with 2× sample buffer were boiled for 3 min and resolved on a 7-20% SDS-PAGE gel. The gel was stained with 0.25% (w/v) Coomassie Blue R-250 in 45% (v/v) methanol, 10% (v/v) acetic acid for 1 h, destained in 5% (v/v) methanol, 7.5% (v/v) acetic acid overnight, and rinsed in H₂O for 0.5-1 h. Gel slices corresponding to the position of soybean trypsin inhibitor (M_r 20,100), which comigrated with Skp1A-Myc, or to the peptide, were excised and incubated in 10% TS-2 (Research Products International, Mt. Prospect, IL), 0.6% 2,5-diphenyloxazole, 0.015% 1,4-bis[2-(4-methyl-5-phenyloxazolyl)]benzene in toluene (Scintanalysis grade). After 72 h, ³H incorporation was determined by scintillation counting (Beckman LS6500). Negative control samples were quenched at zero time or lacked added Skp1, and values (15-35 dpm) were subtracted. The SDS-PAGE assay was more sensitive than the trichloroacetic acid assay and was used for the kinetic results shown.

Reactions quenched with trichloroacetic acid were incubated on ice and vacuum-filtered over a GF/C (Whatman) glass fiber filter prewetted with ice-cold 10% (w/v) trichloroacetic acid. The reaction tube was rinsed with 10 mM sodium pyrophosphate in 10% trichloroacetic acid which was also transferred to the filter, and the filter was rinsed three more times with ice-cold 10% trichloroacetic acid. The filter was then rinsed 4× with 1 ml each of ice-cold acetone. Filters were counted in 10 ml of Scintiverse BioHP mixture (Fisher) as above. Negative control values, ranging from 200 to 500 dpm, were derived from reactions lacking added enzyme and were subtracted from the results given.

In the assay of some particulate fractions, the reaction mixture was clarified at 100,000 × g for 70 min prior to analysis by the SDS-PAGE method. In addition, some supernatants were subjected to immunoprecipitation with mAb 9E10 conjugated to activated Sepharose CL-4B beads at 15 mg/ml (18); the washed precipitates were then analyzed by the SDS-PAGE method.

Preparation of Affinity Resins

Sepharose 6B-Thiopropyl-5-Hg-dUMP-- 5-Hg-dUMP was synthesized as described (19). Ethanol-precipitated 5-Hg-dUMP was washed with ethyl acetate:ethanol (1:4), dissolved in 50 ml of H₂O, and loaded onto a 10-ml Chelex 100 resin (Bio-Rad) to remove residual Hg²⁺ as described (20). The flow-through fraction contained 5-Hg-dUMP to be used for the coupling reaction. The yield of mercuration, measured by the shift of absorbance spectrum from 260 to 267 nm was 71%.

Thiopropyl-Sepharose 6B was prepared according to the manufacturer's protocol (Amersham Pharmacia Biotech) as follows. The dried material was swollen in 50 mM Tris-HCl, 0.1 M NaCl, pH 7.0, at room temperature overnight and washed on a Buchner funnel with a large excess of H₂O. Free thiol groups were released by suspending the gel in 1% (w/v) DTT in 0.3 M NaHCO₃, 1 mM Na₂EDTA, pH 8.4, for 1 h. The gel was washed with a large excess of 0.5 M NaCl, 0.1 M acetic acid, 1 mM Na₂EDTA, followed by H₂O. The gel was then resuspended in the Chelex 100 flow-through fraction of 5-Hg-dUMP with gentle mixing on a shaking platform for 2 h at room temperature. The coupling yield was determined from the decrease of absorbance at 267 nm.

Sepharose 4B-NHS-5-(3-Amino)allyl-UDP-GlcNAc-- Synthesis of 5-Hg-UDP-GlcNAc was performed as above for 5-Hg-dUMP except the Chelex 100 resin step was omitted, with a 70% yield. 5-(3-Amino)allyl-UDP-GlcNAc was synthesized from 5-Hg-UDP-GlcNAc and purified over a DEAE-Sepharose Fast Flow column as described (19). The product exhibited absorbance maxima at 240 and 287 nm and a minimum at 262 nm, typical for a compound with an exocyclic double bond with a pyrimidine ring. Fractions containing 5-(3-amino)allyl-UDP-GlcNAc were pooled giving a 63% yield, dried under vacuum centrifugation, dissolved in anhydrous dimethylformamide, and adjusted with triethylamine to pH 7.5, measured with pH paper. Two ml of NHS-activated Sepharose 4 Fast Flow gel (Amersham Pharmacia Biotech) was washed with 10 ml of dimethylformamide on a Buchner funnel, and the gel was resuspended in the 5-(3-amino)allyl-UDP-GlcNAc solution. The coupling reaction was performed for 24 h with gentle mixing on a shaking platform at room temperature; however, maximal coupling, monitored by the decrease of absorbance at 287 nm, occurred in 2 h. The dimethylformamide was then replaced by 1 M aminoethanol in 50 mM Tris-HCl, pH 7.8, for 2 h. One ml of gel was packed in a HR5/5 column (Amersham Pharmacia Biotech), washed with 50 mM Tris-HCl, pH 7.8, followed by 0.1 M boric acid, pH 8.0, and stored in 2 M NaCl, 0.005% thimerosol in 50 mM HEPES-NaOH, pH 7.8.

GlcNAc-transferase Purification

Buffers-- Buffer pH values were adjusted at room temperature, and solutions were degassed and chilled, and filtered 1 M DTT was added just prior to use. Protease inhibitors were added to buffers A-C to the following concentrations immediately prior to use: 1 mM phenylmethylsulfonyl fluoride, 10 µg/ml leupeptin, and 10 µg/ml aprotinin. Buffer A is as follows: 50 mM HEPES-NaOH, pH 7.4, 1 mM DTT, 5 mM MgCl₂, 0.1 mM Na₂EDTA, 15% (v/v) glycerol, protease inhibitors; buffer B is as follows: 50 mM HEPES-NaOH, pH 7.8, 5 mM DTT, 5 mM MgCl₂, 0.1 mM Na₂EDTA, 15% (v/v) glycerol, protease inhibitors; buffer C is as follows: 50 mM HEPES-NaOH, pH 7.8, 5 mM MgCl₂, 0.1 mM Na₂EDTA, 15% (v/v) glycerol, protease inhibitors; buffer D is as follows: 50 mM HEPES-NaOH, pH 7.8, 5 mM DTT, 0.1% (v/v) Tween 80; 5 mM MgCl₂, 0.1 mM Na₂EDTA, 15% (v/v) glycerol; buffer E is as follows: 50 mM HEPES-NaOH, pH 7.8, 5 mM DTT, 5 mM MgCl₂, 0.1 mM Na₂EDTA; and buffer F is as follows: 50 mM HEPES-NaOH, pH 7.8, 5 mM DTT, 0.01% (v/v) Tween 80; 5 mM MgCl₂, 0.1 mM Na₂EDTA, 15% (v/v) glycerol.

Cell Lysis and DEAE-Sepharose Chromatography-- Stationary phase strain Ax3 or HW120 cells grown in HL-5 were filter-lysed and successively centrifuged at 3000 × g for 2 min and 100,000 × g for 70 min to generate the cytosolic S100 supernatant fraction as described previously (17). The pellets were resuspended in lysis buffer and re-centrifuged, yielding the P3 and P100 (where P100 is the particulate cell fraction equivalent to the pellet formed by centrifugation at 100,000 × g for 1 h) fractions, respectively. Nuclei were prepared using either Nonidet P-40 or digitonin as described (21). The S100 supernatant (800 ml) was pumped onto a 450-ml DEAE-Sepharose Fast Flow column equilibrated in buffer A and eluted with the same buffer (17). Skp1 was quantitatively retained on the column, whereas GlcNAc-transferase activity appeared in the wash fractions, which were frozen at 80 °C. The column was washed with 2 M NaCl in 50 mM HEPES-NaOH, pH 7.4, prior to reuse.

Ammonium Sulfate Precipitation-- (NH₄)₂SO₄ (ultrapure, ICN) was added to the pooled active fractions from the DEAE-Sepharose column to a final concentration of 25% saturation with stirring at 4 °C. After centrifugation at 20,000 × g for 20 min, the supernatant was brought to 60% (NH₄)₂SO₄ saturation. The pellet was collected by centrifugation, dissolved in 300 ml of 20 or 25% saturated (NH₄)₂SO₄ in buffer B, and recentrifuged to clarify.

Phenyl-Sepharose Chromatography-- The 25-60% (NH₄)₂SO₄ cut was applied to a phenyl-Sepharose Fast Flow (low-sub) column (2.6 cm x 20 cm) pre-equilibrated with 20 or 25% saturated (NH₄)₂SO₄ in buffer B at 4 °C, washed with the same buffer until the A₂₈₀ dropped to below 1% of its maximum, and eluted with a linear decreasing 600-ml gradient to buffer B. Fractions were stored at 80 °C. The column was washed in 6 M urea in 50 mM HEPES-NaOH, pH 7.8, followed by 1 M NaOH, before reuse.

Reactive Red-120 Chromatography-- The phenyl-Sepharose activity pool was loaded onto a Reactive Red-120 Fast Flow column (1.6 × 5 cm), equilibrated in buffer B, at 4 °C at 80 ml/min. After washing with buffer B until the A₂₈₀ dropped to 10% of maximum, the column was eluted with a 150-ml 0-2 M NaCl linear or step gradient prepared in buffer B. Fractions were assayed after desalting on 9 ml of Sephadex G-25 columns (PD-10 columns, Amersham Pharmacia Biotech) pre-equilibrated in buffer D, and stored at 80 °C. The Reactive Red-120 column was precycled according to manufacturer's instructions (Sigma) prior to reuse.

Superdex 200 Chromatography-- The Reactive Red column activity pool was concentrated in a Centriplus-10 ultrafiltration device (Amicon) to <2 ml and was loaded onto a Superdex 200 (16/60) column (Amersham Pharmacia Biotech) equilibrated in buffer C. The column was isocratically eluted at 0.8 ml/min at 22 °C, and fractions were stored at 80 °C. The column was calibrated with the following M_r standards: horse spleen apoferritin (443,000), yeast alcohol dehydrogenase (150,000), bovine serum albumin (66,000), ovalbumin (45,000), and soybean trypsin inhibitor (20,100).

5-Hg-dUMP Chromatography-- The Superdex 200 activity pool was loaded onto a 1-ml 5-Hg-dUMP column equilibrated in buffer C. DTT was added to the flow-through fraction to 5 mM, which was concentrated in a Centriplus-10 ultrafiltration device to <2 ml. After use, the column was cleaned with 5 mM dUMP and stored in 2 M NaCl, 0.005% thimerosol in 50 mM HEPES-NaOH, pH 7.8.

Aminoallyl-UDP-GlcNAc Affinity Chromatography-- The concentrated flow-through fraction from the 5-Hg-dUMP column was loaded at 100 µl/min onto the aminoallyl-UDP-GlcNAc-Sepharose column described above. This column was mounted on a Smart HPLC system (Amersham Pharmacia Biotech) and equilibrated in buffer D, which contained 0.1% (v/v) Tween 80 to stabilize enzyme activity. The column was washed with buffer F until recovery of the original base line and was eluted with a 7.4-ml gradient of 0-2 mM UMP in buffer F at 22 °C and 250 µl/min. To detect enzyme activity, fractions were either desalted on PD-10 columns equilibrated in buffer D or cyclically concentrated and diluted in Microcon-10 ultrafiltration devices (Amicon) with buffer F to reduce UMP to <20 µM. For kinetic studies, active fractions were pooled and desalted on a Fast Desalting PC3.2/10 column mounted on the Smart System HPLC that was equilibrated with buffer E and run at 100 µl/min at 22 °C. The column was washed with 8 M urea, 50 mM DTT, 50 mM Tris-HCl, pH 7.5, prior to reuse.

Superdex 75 Chromatography-- For SDS-PAGE and photoaffinity labeling, the UDP-GlcNAc affinity pool was concentrated to <50 µl in a Microcon-10 ultrafiltration device and applied to a Superdex 75 PC3.2/30 column (Amersham Pharmacia Biotech) pre-equilibrated in buffer F, which contained 0.01% Tween 80 (v/v). The column was eluted at 0.05 ml/min at 22 °C, collecting 25-µl fractions. The column was calibrated with bovine serum albumin (M_r 66,000), ovalbumin (M_r 43,000), carbonic anhydrase (M_r 29,000), and cytochrome C (M_r 12,400).

Protein Determination-- Protein concentration was determined by a commercial modification of a Coomassie Blue dye-binding method (22) according the manufacturer's protocol (Pierce), using bovine serum albumin as a standard. After UDP-GlcNAc affinity chromatography, protein was alternatively estimated from its A₂₈₀ value, assuming an extinction coefficient of 1 ml/mg for a 1-cm path length, or by analysis of phenylisothiocyanate-amino acid derivatives after acid hydrolysis of the sample at the ICBR Protein Chemistry Core Laboratory, using norleucine as an internal standard.

Characterization of the Purified GlcNAc-transferase

SDS-PAGE-- Column fractions were diluted with equal volumes of 2× Laemmli sample buffer containing 5 mM DTT and boiled for 2 min. Iodoacetamide was added to 40 mM, and samples were applied to a 7-20% (w/v) linear gradient SDS-PAGE gel (17). Gels were silver-stained (23).

Photoaffinity Labeling-- 0.5 µmol of 5-ASA-UDP-GlcNAc (19), a generous gift of Dr. A. E. Elbein, was reacted with 2 mCi of carrier-free Na¹²⁵I from NEN Life Science Products as described (24). 5-[¹²⁵I]ASA-UDP-GlcNAc was preincubated with Superdex 75 fractions at 30 µM in the presence or absence of 700 µM UDP-GlcNAc for 15 min at 22 °C. The reaction mixture was then exposed for 2 min to 254 nm radiation from a 30-watt mineralight at a distance of 1 cm. The reaction was diluted with 2× Laemmli electrophoresis buffer and processed as described in the SDS-PAGE section above. The stained gel was then autoradiographed against Kodak Bio-Max HP film at 80 °C.

Kinetic Analyses-- Kinetic studies were performed on the UDP-GlcNAc affinity pool after passage over a Fast Desalting PC3.2/10 column. Reactions were incubated for 1 h using the SDS-PAGE assay.

Reaction Product Characterization

Reductive Alkaline Degradation of [³H]GlcNAc-Skp1-- Gel slices containing [³H]Skp1 or [³H]4-HyPro peptide were rinsed in H₂O, minced, and diluted with an equal volume of 1 M NaBH₄ in 0.2 M NaOH. After incubation at 45 °C for 20 h, the supernatant was neutralized with an equal volume of 0.2 M acetic acid in MeOH and dried under vacuum centrifugation. The radioactive residue was dissolved in 2 ml of H₂O, adjusted to pH 9.5 with 2-amino-2-methyl-1 propanol as determined using pH paper, and loaded onto a 1-ml Dowex 1 column pretreated with 5 ml of 1 M HCl and 10 ml of H₂O. The flow-through was diluted in Scintiverse-HP scintillation mixture and counted as above.

Acid Hydrolysis of [³H]GlcNAc-Skp1-- Radioactive gel slices were hydrolyzed in 1 ml of 4 M trifluoroacetic acid at 100 °C for 4 h. The supernatant was dried by vacuum centrifugation, dissolved in H₂O, dried again, dissolved in 100 µl of H₂O containing 1 nmol of GlcNH₂, and chromatographed on a PA-10 column on a DX-500 HPLC system (Dionex) in 17 mM NaOH at 1 ml/min. The eluant was monitored by pulsed amperometric detection (17), and fractions were counted in a liquid scintillometer.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

A UDP-GlcNAc:Skp1-HyPro GlcNAc-transferase Activity Is Found in the S100 Fraction-- Strain HW120 expresses a c-Myc-tagged form of Skp1A under the control of the discoidin promoter (6). Skp1A-Myc isolated from the S100 fraction of this strain contains an abundant isoform that is hydroxylated but not glycosylated at Pro-143. This unglycosylated isoform is predicted to be an acceptor substrate for the hypothetical Skp1 GlcNAc-transferase. An assay for the Skp1 GlcNAc-transferase was developed based on transfer of ³H from UDP-[³H]GlcNAc to Skp1A-Myc, in the presence of 1 mM DTT, 5 mM MgCl₂, and 50 mM Tris-HCl at pH 7.5. The S100 fraction from stationary phase cells of the normal strain Ax3 was initially tested as a source of enzyme. Skp1 was purified after the reaction by SDS-PAGE, and incorporated radioactivity was measured by scintillation counting of the gel slice. Addition of Skp1A-Myc stimulated incorporation of ³H into the Skp1 band over 3-fold, whereas Skp1A-His₁₀, a recombinant form of Skp1A that is not hydroxylated at Pro-143 (6), did not (Fig. 1A, solid bars). No stimulation by Skp1A-Myc was seen at other M_r positions in the gel (data not shown). In addition, 10 µg/ml mAb 3F9, which is specific for Skp1 in Western blots (25), almost completely blocked stimulation of incorporation by added Skp1A-Myc (data not shown). The S100 fraction from strain HW120 exhibited a high level of spontaneous incorporation in the absence of added Skp1A-Myc (Fig. 1B), suggesting that endogenous Skp1A-Myc was a substrate after cell lysis. Consistent with this interpretation, incorporation was only slightly further stimulated by additional Skp1A-Myc, and the high level of spontaneous incorporation was markedly reduced by mAb 3F9 (Fig. 1B, solid bars). These results suggested the presence of a Skp1-HyPro GlcNAc-transferase in the S100 fraction of the cell.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Skp1-HyPro GlcNAc-transferase-like activities in various cell fractions. Cell extracts were incubated with 0.8 µM UDP-[3H]GlcNAc in the presence of the indicated substrate or mAb inhibitor. A and B, after 2 h, the reaction mixture was subjected to SDS-PAGE and incorporation of 3H was measured at the gel positions of Skp1 (black bars) and the peptide (open bars). A, an S100 preparation from the normal strain Ax3 (6.3 mg of protein/ml) was incubated either alone (none) or in the presence of 0.54 µM Skp1A-Myc, 0.54 µM Skp1A-His10, or 0.2 mM 4-HyPro peptide-(133-155). B, an S100 preparation from the Skp1A-Myc expression strain HW120 (6.0 mg of protein/ml) was incubated as in A or in the presence of 10 µg/ml mAb 3F9 IgG. C, P3 (3.0 mg of protein/ml) and P100 (5.0 mg of protein/ml) particulate fractions, and an S100 fraction (6.0 mg/ml), all from strain Ax3, were assayed in the absence (open bars) or presence (shaded bars) of 0.54 µM Skp1A-Myc for 1 h. The P3 and P100 fractions were permeabilized with 0.25% Tween 80 as indicated. The reaction mixture was recentrifuged after incubation to remove particulate material, and in some cases, the +Skp1 samples were also immunoprecipitated with mAb 9E10-Sepharose (black bars) prior to SDS-PAGE to determine incorporation into the Skp1A-Myc band.

To gain further evidence that the GlcNAc-transferase activity was specific for the HyPro residue, a synthetic peptide corresponding to residues 133-155 of Skp1 with 4-HyPro at the equivalent of residue 143 was tested as a substrate. The 4-HyPro peptide induced substantial increases of ³H into the peptide band in the S100 fractions of both strains Ax3 (Fig. 1A, open bars) and HW120 (Fig. 1B, open bars). A synthetic peptide containing Pro in place of 4-HyPro was inactive (not shown, see Fig. 6 below). The 4-HyPro peptide markedly reduced incorporation into endogenous Skp1 of strain HW120 (Fig. 1B). This indicated that the activity that labeled the 4-HyPro peptide was equivalent to the activity that labeled Skp1A-Myc and implied that labeling of Skp1A-Myc was associated with HyPro-143.

To verify that ³H was transferred as GlcNAc, [³H]Skp1 was subjected to acid hydrolysis and analyzed on a Dionex PA-10 column in 17 mM NaOH. All released ³H co-chromatographed with GlcNH₂, the expected acid hydrolysis product of GlcNAc (shown for purified enzyme below). The transferase activity was apparently specific for UDP-GlcNAc, as substitution with UDP-[³H]GalNAc also led to recovery of [³H]GlcNH₂ after acid hydrolysis. This indicated that UDP-[³H]GalNAc was converted to UDP-[³H]GlcNAc, prior to incorporation, by a UDP-HexNAc epimerase activity in the crude extract. For evidence that GlcNAc was linked to a HyPro residue, the alkali esistance of the sugar-protein linkage was examined. When in vivo labeled [³H]fucose-Skp1 was previously subjected to reductive alkaline degradation, the majority of radiolabeled product had a charge of 2 at pH 9.5 (26). Since the fucoglycan is now known to be neutral and attached to the polypeptide via an alkali-resistant GlcNAc-HyPro linkage (6), the negative charge must have derived from scission of the polypeptide backbone. After subjecting gel slices containing ³H-labeled Skp1A-Myc or Skp1-HyPro peptide to reductive alkaline degradation, only 16 and 12% of released ³H, respectively, emerged in the flow-through fractions from a Dowex-1 column. These observations were inconsistent with a linkage to Ser or Thr, as neutral [³H]N-acetylglucosaminitol would be expected to be released from either of these amino acids (27). Since retention of the alkali-degraded ³H on the Dowex-1 column was expected for a linkage to HyPro (6), and since HyPro was required in both the Skp1 and peptide substrates, it was concluded that incorporation of [³H]GlcNAc occurred via formation of a GlcNAc-HyPro linkage.

The enzyme activity was optimized with respect to temperature and pH. Activity increased as a function of temperature up to 30 °C, which was 3 °C above the maximum physiological temperature of Dictyostelium, and decreased at 33 °C. pH dependence was tested after gel filtration of the S100 extract in 50 mM Tris-HCl and MES-NaOH buffers ranging from pH 6.0 to 8.5. Highest activity was found at pH 7.5-8.0, and activity was <10% at pH 6.0 (data not shown). pH 7.8 was subsequently used for storage and assay buffers.

To address whether the Skp1 GlcNAc-transferase activity might be present in other subcellular fractions, the cell lysate was successively centrifuged at 3000 × g for 2 min and 100,000 × g for 70 min. After washing, these P3 and P100 fractions contained 17 and 27% of total cell protein. Addition of Skp1A-Myc to either the P3 or P100 fractions had a negligible effect on incorporation of ³H into the Skp1 band (data not shown). Similar negative results were seen when the particles were permeabilized with 0.25% Tween 80, a detergent that does not inhibit the GlcNAc-transferase (see below). However, background levels of incorporation were up to 4-fold higher than the maximum level of enzyme activity detected in the S100 fraction, potentially obscuring Skp1-dependent incorporation. Removal of particles after incubation by ultracentrifugation increased the sensitivity of the assay by reducing background incorporation. As shown in Fig. 1C, addition of Skp1A-Myc resulted in little or no stimulation of uptake into Skp1 in either of the fractions in the presence or absence of Tween 80. To determine if any of the incorporated ³H could be attributed to Skp1A-Myc, the post-assay supernatants were incubated with an excess of mAb 9E10 conjugated to Sepharose beads. Although prior immunoprecipitation recovered >70% of ³H incorporated into the Skp1A-Myc band in the S100 fraction (Fig. 1C), negligible ³H was recovered from the P100 or P3 reactions carried out in the absence or presence of Tween 80. In other experiments not shown, no activity was detected in a nuclear preparation purified from the P3 fraction, although the possibility of movement from the nucleus into the cytosolic fraction during cell lysis cannot be excluded. In addition, assays performed on mixtures of S100 and P3 or P100 fractions showed no evidence for an inhibitor in the particulate fractions (data not shown). At present, the lack of evidence for Skp1 GlcNAc-transferase activity in the particulate fractions suggests that the enzyme is cytosolic.

Early Purification Steps of the UDP-GlcNAc:Skp1-HyPro GlcNAc-transferase Activity-- The S100 GlcNAc-transferase activity from strain HW120 was purified by a combination of conventional and affinity chromatographic methods, starting with anion exchange chromatography. When applied to a DEAE Fast-Flow Sepharose column at pH 7.4, enzyme activity eluted in the wash fractions after the bulk of protein (Fig. 2A). Similar weak binding to the column occurred using either 50 mM Tris-HCl or 50 mM HEPES-NaOH buffers, or at pH 8.0, and no additional activity was eluted in a gradient of 0-1 M NaCl (data not shown). Activity did not adsorb to an SP-Sepharose column over the pH range of 6.0-7.4 (data not shown). In pilot experiments using strain Ax3 cells, activity in the DEAE wash pool was recovered at 100% of the S100 level (data not shown) and, since the presence of endogenous Skp1A-Myc precluded comparing activities between these two fractions in strain HW120, the recovery level shown for this experiment shown in Table I (lines 1 and 2) is based on the strain Ax3 results.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Early purification steps of the Skp1-HyPro GlcNAc-transferase activity from the S100 fraction. A, an S100 preparation from strain HW120 was pumped onto a DEAE-Sepharose Fast Flow column equilibrated in buffer A at pH 7.4. Skp1 GlcNAc-transferase activity was retarded on the column eluting primarily in the wash fractions behind the bulk of the protein flow-through. B, the active fractions were pooled, and material that precipitated between 25 and 60% saturated (NH4)2SO4 was dissolved in 25% saturated (NH4)2SO4 and loaded onto a phenyl-Sepharose Low-sub column pre-equilibrated with 25% saturated (NH4)2SO4 in buffer B. Activity was eluted as a single component after the bulk of the protein in a descending gradient down to 0% (NH4)2SO4 in buffer B. C, the phenyl-Sepharose activity pool was loaded onto a Reactive Red-120 column and eluted broadly in the range of 1-2 M NaCl (not shown). The activity pool was concentrated in a Centriplus-10 cartridge and fractionated on a Superdex 200 gel filtration column equilibrated in buffer C. Activity eluted as a single component near the Mr 45,000 position.

Susceptibility to precipitation by various concentrations of (NH₄)₂SO₄ was evaluated by comparing activities of precipitates and supernatants, after decreasing the (NH₄)₂SO₄ by repeated concentration and dilution in Microcon centrifugal ultrafiltration devices. Enzyme activity in the DEAE pool was resistant to precipitation by 25% (w/v) (NH₄)₂SO₄, but >85% of activity was precipitated by 60% (NH₄)₂SO₄. Enzyme recoveries in the 25% (NH₄)₂SO₄ supernatant and 60% (NH₄)₂SO₄ pellet were 81 and 50% (Table I, lines 2 and 3), but since the lost activity was recovered after the next chromatography step (see below), the low apparent recoveries were likely due to inhibition by residual salt (see below).

The 60% (NH₄)₂SO₄ precipitate was redissolved in 25% (NH₄)₂SO₄ and applied to a column of phenyl-Sepharose Low-sub. Enzyme activity eluted as a single peak near the end of a descending gradient of 25-0% (NH₄)₂SO₄ (Fig. 2B) and was recovered at 156% of the level in the DEAE pool (Table I, line 4). Elution with 50% ethylene glycol yielded no additional activity (data not shown).

The phenyl-Sepharose activity pool was applied to a Reactive Red-120 fast flow column. Activity bound to the column and eluted broadly in the range of 1-2 M NaCl with 10% yield, or 16% overall yield (Table I, line 5), in a gradient of 0-2.0 M NaCl. Loss of enzyme activity did not appear to result from potential subunit dissociation as the apparent M_r of the enzyme, determined by gel filtration (see below), did not change during the purification. In addition, reconstitution of the activity pool with the Reactive Red-120 flow-through pool, the phenyl-Sepharose activity pool, or the S100 fraction failed to yield more than additive activity levels. In other purifications, recovery from the dye column was better than 80%, but comparable losses occurred at either earlier steps or the subsequent concentration step, so that the overall yield was similar. The self-limiting character of the activity loss suggested a predisposition of the enzyme to down-regulation which did not involve loss of a subunit.

The Reactive Red-120 activity pool was concentrated in a Centriplus-10 ultrafiltration device and applied to a Superdex 200 gel filtration column, equilibrated in a buffer lacking DTT in preparation for the subsequent purification step. Activity eluted as a single symmetrical peak with an apparent M_r of 45,000 (Fig. 2C) and a recovery of 92% (Table I, line 6). Similar results were obtained when the column buffer contained 5 mM DTT (data not shown). The Superdex 200 activity pool was purified 220-fold relative to that of the S100 fraction, despite an overall recovery of only 15%. The best recovery obtained was 35% was in a separate, larger scale purification (data not shown).

Characterization of the Partially Purified GlcNAc-transferase Activity-- The concentrated DEAE activity pool was selectively inhibited by NaCl relative to KCl, with only 30% activity remaining in 100 mM NaCl (Fig. 3A). Inhibition at higher ionic strengths was also observed with (NH₄)₂SO₄ (data not shown). The Superdex 200 activity pool was stimulated 2.5-fold by 5-10 mM MgCl₂ relative to 1 mM, but a declining effect was seen at higher concentrations (Fig. 3B). Titration of MgCl₂ by equimolar (5 mM) EDTA abolished activity, indicating absolute dependence on a divalent cation. Addition of 1 mM MnCl₂ or CaCl₂, in the presence of 5 mM MgCl₂, to the DEAE pool inhibited activity 10-20%, increasing to 40-60% inhibition at 10 mM (Fig. 3C). Thus the standard assay contained 5 mM MgCl₂ buffered at pH 7.8 and contained no added NaCl or KCl.

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Characteristics of the partially purified Skp1-HyPro GlcNAc-transferase activity. A, effects of NaCl and KCl. The concentrated DEAE activity pool was assayed over a range of NaCl (solid bars) or KCl (open bars) concentrations, in the presence of 5 mM MgCl2. B, effects of MgCl2. The Superdex 200 activity pool was assayed over a range of MgCl2 concentrations. Approximately zero concentration was achieved by adding 5 mM Na2EDTA, pH 7.8, to chelate the 5 mM MgCl2. C, effects of CaCl2 and MnCl2. The concentrated DEAE activity pool was assayed over a range of concentrations of CaCl2 (open bars) and MnCl2 (solid bars), in the presence of 5 mM MgCl2. D, effect of DTT. The Superdex 200 activity pool from a column run with buffer C, which lacked DTT, was assayed over a range of DTT concentrations. E, time course. The dUMP-Sepharose activity pool was assayed in duplicate by starting the reaction with either the enzyme (open squares), Skp1A-Myc (open circles), or UDP-[3H]GlcNAc (data not shown). Similar results were obtained whether the reaction was started with Skp1A-Myc or UDP-GlcNAc. Straight lines represent linear best fits of the data. F, effect of detergents on activity and stability. The Superdex 200 activity pool was incubated at 4 °C in the presence of different detergents at 0.1% (v/v). At 0 h (open bars) and 24 h (solid bars), aliquots of the mixture were assayed. G, comparative effects of 0, 0.1, and 1 mM uracil on the activities of the Superdex 200 (open bars) and dUMP-Sepharose activity (solid bars) pools. N.D., not detected.

The dependence of activity on reducing potential was tested in an assay of the Superdex 200 activity pool. No activity was detected in the absence of DTT (Fig. 3D), and activity was highest at 5-10 mM DTT. Absence of DTT during Superdex 200 chromatography did not affect the stability, apparent M_r, or recovery of the enzyme (data not shown).

The Superdex 200 activity pool was assayed for time and concentration dependence of activity. Assays of 0.6 µg of protein yielded activity proportional to incubation time for at least 4 h, at 30 °C, when the reaction was started by the addition of enzyme (Fig. 3E). However, if the reaction was started by the addition of either UDP-[³H]GlcNAc or Skp1A-Myc, less activity was apparent at the early time points. Extrapolation of the results suggested there was a lag phase of 13 min before the initiation of linear kinetics. Possibly the enzyme is inhibited by the presence of one substrate in the absence of the other or interaction of the two substrates is rate-limiting for the reaction. At higher concentrations of enzyme, the reaction was linear only to 1-2 h.

Activity in the Superdex 200 pool was proportional to the amount of protein over the range of 0.6-1.8 µg/assay in assays conducted for 0.5-1.0 h. However, over longer time periods or at higher levels of total protein, activity levels plateaued. This was not due to enzyme inactivation, as addition of more enzyme at the end of a 4-h incubation failed to increase incorporation (data not shown). Addition of more substrate, either Skp1A-Myc or UDP-[³H]GlcNAc greatly stimulated incorporation,³ suggesting complex effects that are not understood.

The activity of the Superdex 200 pool was unstable at 4 °C (Fig. 3F). However, addition of 0.01-0.1% (v/v) Tween 20 or Tween 80 stabilized activity, and in addition stimulated activity up to 2-fold, with Tween 80 exerting a more positive effect (Fig. 3F). Although the other detergents tested could stabilize activity, they initially suppressed activities up to 50%. Inclusion of 0.1% (v/v) Tween 80 in the Superdex 200 column buffer did not affect its elution time (data not shown).

Preparation of Affinity Purification Columns-- To design affinity ligands related to UDP-GlcNAc which could be used for purifying and photolabeling the enzyme, the Superdex 200 pool was assayed in the presence of UDP-GlcNAc analogues to identify molecular determinants critical for inhibition. Using the standard assay containing 0.66 µM UDP-[³H]GlcNAc, the analogues UMP, UDP, UTP, 4-thio-UMP, Bio-11-UTP, and UDP-Glc inhibited activity greater than 50% at 1 mM in a concentration-dependent fashion (Table II). In contrast, uridine, dUDP, dUMP, UDP-hexanolamine, and CTP had little or no effect over this concentration range. 2',3'-o-Isopropylidene-uridine, ATP, UDP-GalNAc, and uracil showed complex effects, with stimulation at 0.1 mM and inhibition in some cases at 1 mM. Stimulation at lower concentrations probably resulted from inhibition of other activities that sequester or degrade UDP-GlcNAc, an effect which might then have been overcome at higher concentrations by direct inhibition of the GlcNAc-transferase activity.

UMP was the best inhibitor but UDP and UTP also inhibited substantially, suggesting that these nucleotides could bind the enzyme. Binding was specific for uridine nucleotides as CTP and ATP were weak inhibitors. Modifications of the uridine ring at the 4- or 5-positions (see Fig. 4A for diagram) were tolerated as 4-thio-UMP and Bio-11-UTP were also good inhibitors, suggesting that these might be favorable positions for attaching to a resin or a photoreactive group. Removal of the 2'-hydroxyl of the ribose moiety negated the inhibitory activity of UDP, indicating that this position should not be modified. UDP-hexanolamine and UDP-GalNAc were also not good inhibitors, suggesting that only certain moieties linked to the -PO₄ will fit within the active site of the enzyme. Thus UDP-GlcNAc with substitutions at the 4- or 5-position of the pyrimidine ring showed the greatest promise for being recognized by the enzyme.

View larger version (26K): [in this window] [in a new window]   Fig. 4.   Later purification and photoaffinity labeling of the purified GlcNAc-transferase activity. A, the Superdex 200 activity pool (Fig. 2C) was applied to the dUMP-Sepharose column in buffer C, and the flow-through fraction (not shown) was applied to the 5-(3-aminoallyl)-UDP-GlcNAc-Sepharose resin, whose structure is shown, equilibrated in buffer D. The enzyme activity was eluted with a 0-2 mM UMP gradient in buffer B. Eluted fractions were desalted in Microcon-10 cartridges before assaying GlcNAc-transferase activity. B, the UDP-GlcNAc affinity pool was concentrated to 35 µl and applied to a Superdex 75 column equilibrated in buffer F. Activity co-eluted with a base-line-resolved A280 peak. C, selected fractions from this peak were analyzed by SDS-PAGE and silver staining. Only the region of the gel showing detectable bands is shown. D, fractions from the activity peak were incubated in 30 µM 5-[125I]ASA-UDP-GlcNAc, in the presence or absence of 700 µM UDP-GlcNAc as indicated, and exposed to UV irradiation. Samples were subjected to SDS-PAGE and autoradiographed. The predominant protein band at Mr 51,000, seen by both silver staining and photoaffinity labeling, is referred to as Gnt51 in the text.

Two kinds of affinity columns were synthesized. The first was intended as a decoy for other nucleotide-binding proteins whose existence was suggested by the stimulatory activity of certain nucleotides. The second was intended to bind specifically the GlcNAc-transferase.

dUMP was chosen as the decoy nucleotide based on its inactivity as an inhibitor but with structural similarity to UMP, a potent inhibitor. It was linked to Sepharose via its 5-position to be more likely to attract other proteins also able to bind 5-linked UDP-GlcNAc. 5-Hg-dUMP was synthesized and conjugated to thiopropyl-Sepharose to a concentration of 53 µmol/ml resin.

To prepare the UDP-GlcNAc-Sepharose column, the 5-Hg-group of 5-Hg-UDP-GlcNAc was replaced by allylamine to yield 5-(3-amino)allyl-UDP-GlcNAc (Fig. 4A). The coupling reaction was performed by acylation of the terminal NH₂ group with NHS-activated Sepharose in dimethylformamide. The concentration of 5-(3-amino)allyl-UDP-GlcNAc, calculated from the decrease of A₂₈₇ after the reaction, was 14 µmol/ml gel.

Final Purification of the Skp1-HyPro GlcNAc-transferase Activity-- Since the Hg-S bond of the 5-Hg-dUMP column was sensitive to reducing reagents, DTT was removed during the prior Superdex 200 purification step. The Superdex 200 activity pool was applied to the 5-Hg-dUMP column, and GlcNAc-transferase activity was quantitatively recovered in the flow-through fraction with 99% yield (Table I, line 7). After purification over the dUMP column, stimulation by 1 mM uracil was reduced from the 40% value found for the Superdex 200 pool to 9% (Fig. 3G). These data suggested that activities that sequestered or degraded UDP-GlcNAc were removed by this column.

The 5-Hg-dUMP-Sepharose flow-through fraction was replenished with 5 mM DTT and applied to the 5-(3-amino)allyl-UDP-GlcNAc-Sepharose column. Fractions eluted with an ascending gradient of 0-2 mM UMP were assayed after dilution or desalting in Microcon-10 concentrators. Activity was recovered as a single peak (Fig. 4A) with 95% yield (Table I, line 8).

For the final purification step, the concentrated UDP-GlcNAc affinity pool was applied to a Superdex 75 PC3.2/30 column. Activity eluted as single symmetrical peak (Fig. 4B) with a yield of 30% (Table I, line 9) and an apparent M_r of 13,000. Activity eluted at M_r 45,000 in another trial that contained 5× greater activity, suggesting weak binding to the column at lower protein levels. The eluted activity profile corresponded closely to the profile of a base-line-resolved A₂₈₀ peak, suggesting that the enzyme may be homogenous. This was supported by SDS-PAGE analysis, which showed that the peak fractions contained predominantly a M_r 51,000 protein and small amounts of M_r 40,000 and 36,000 proteins (Fig. 4C). The M_r 51,000 value was similar to the M_r 45,000 value for the enzyme activity suggested by the gel filtration separations. To gain direct evidence that one of the proteins detected by SDS-PAGE corresponded to the enzyme protein, susceptibility to labeling with photoactivated 30 µM 5-[¹²⁵I]ASA-UDP-GlcNAc was tested. The photoactive aryl azide moiety of this photoaffinity probe, which has been used to label a Golgi GlcNAc-transferase (19), was linked at the same position as was the leash in the affinity resin. After exposure to UV irradiation in the presence or absence of a 23-fold excess of unlabeled UDP-GlcNAc, fractions were analyzed by SDS-PAGE and autoradiography. The M_r 51,000 protein was substantially labeled, and the protein was protected from labeling by excess UDP-GlcNAc (Fig. 4D). The M_r 40,000 and 36,000 bands were also labeled but not specifically as excess UDP-GlcNAc did not significantly inhibit labeling. Since photoinduced labeling of the M_r 51,000 protein by 5-[¹²⁵I]ASA-UDP-GlcNAc appeared to depend on the availability of a high affinity UDP-GlcNAc-binding site, this protein, referred to as Gnt51, was concluded to possess the Skp1-HyPro GlcNAc-transferase activity of the fractions.

Characterization of the Skp 1-HyPro GlcNAc Reaction Product-- The Skp1 GlcNAc-transferase assay measured the transfer of radioactivity from UDP-[³H]GlcNAc to Skp1 or the Skp1 peptide. To verify that the radioactivity was transferred as GlcNAc, gel slices containing [³H]Skp1A-Myc from the in vitro assay of the affinity purified GlcNAc-transferase pool were treated with 4 M trifluoroacetic acid for 4 h at 100 °C. Radioactivity was eluted from the gel at 20% yield and analyzed on a Dionex PA-10 column (Fig. 5). All eluted radioactivity co-chromatographed with GlcNH₂, consistent with the presence of GlcNAc that had been de-N-acetylated by the acid treatment, in the reaction product. Acid hydrolysis with 6 M HCl for 6 h at 120 °C gave similar results but with lower recovery of radioactivity (data not shown).

View larger version (18K): [in this window] [in a new window]   Fig. 5.   High pH anion exchange chromatography of acid-hydrolyzed [3H]GlcNAc-Skp1A-Myc. [3H]Skp1A-Myc from an assay of the UDP-GlcNAc-affinity activity pool was purified by SDS-PAGE, and the Coomassie Blue-stained gel slice containing [3H]Skp1A-Myc was hydrolyzed in 4 M trifluoroacetic acid at 100 °C for 4 h. The supernatant was dried, dissolved in water containing 1 nmol of GlcNH2, and chromatographed on a Dionex-PA10 column. A, elution profile of sugar standards detected by pulsed amperometry. B and C, elution profile of acid-hydrolyzed [3H]Skp1A-Myc, showing coelution of the GlcNH2 internal standard detected by pulsed amperometry (B) and radioactivity (C).

The unpurified GlcNAc-transferase activity catalyzed the formation of an alkali-resistant linkage between GlcNAc and Skp1 or the Skp1-HyPro peptide as described above. A similar test was carried out using Skp1A-Myc labeled by the purified GlcNAc-transferase. Sixty-four percent of the radioactivity released from reductive alkali-treated [³H]GlcNAc-Skp1 bound to Dowex-1 resin at pH 9.5, compared with 75% binding of the in vivo [³H]fucose-labeled product to SP-Sephadex at this pH (26). The similar binding of the in vitro and in vivo labeled products was consistent with linkage of the radioactive moiety to a HyPro residue.

Kinetic Properties of the Skp1-HyPro GlcNAc-transferase-- Activity of the UDP-GlcNAc affinity pool was measured over a range of substrate concentrations, using a 1-h incubation started by addition of the enzyme and quantitated using the SDS-PAGE assay. The dependence of activity on the concentration of UDP-GlcNAc, in the presence of 1 µM Skp1A-Myc, conformed to the Michaelis-Menten model as shown by the linear form of the Lineweaver-Burk double-reciprocal plot of the data (Fig. 6A). The estimated K_m value for UDP-GlcNAc was 0.16 µM with a V_max of 8.0 nmol/mg/h. This K_m value is similar to the relatively low values, compared with those of Golgi transferases (e.g. Ref. 10), seen for two other cytoplasmic glycosyltransferases, the Skp1 Fuc-transferase (17) and the general Ser/Thr GlcNAc-transferase (11). The low V_max value suggested that the enzyme was down-regulated in its purified state.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Kinetic analysis of the Skp1-HyPro GlcNAc-transferase with respect to its donor and acceptor substrates. A UDP-GlcNAc affinity activity pool was assayed over a range of concentrations of UDP-[3H]GlcNAc, Skp1A-Myc, or the 4-HyPro peptide-(133-155) for 1 h. Incorporation of 3H was determined by SDS-PAGE. Results of duplicate assays are plotted and manually fitted in the upper panels. Double-reciprocal plots of the averaged data, with least squares best fit lines and the derived Km and Vmax values, are given in the lower panels. A, activity was determined in the presence of 1 µM Skp1A-Myc and various concentrations of UDP-[3H]GlcNAc. B, activity was determined in the presence of 0.66 µM UDP-[3H]GlcNAc and varying concentrations of Skp1A-Myc or recombinant Skp1A-His10. C, activity was determined in the presence of 0.66 µM UDP-[3H]GlcNAc and varying concentrations of 4-HyPro peptide-(133-155) or Pro peptide-(133-155).

Activity with respect to the both the Skp1A-Myc and Skp1 peptide substrates also conformed to the Michaelis-Menten model (Fig. 6, B and C). In the presence of 0.66 µM UDP-[³H]GlcNAc, apparent K_m values with respect to Skp1A-Myc and the Skp1 peptide were 0.56 µM and 1.6 mM, respectively. Negligible activity was detected using substrate isoforms lacking the target HyPro residue, either Skp1A-His₁₀ (Fig. 6B) or the Skp1-Pro peptide (Fig. 6C). The K_m value for Skp1A-Myc was approximate because only a fraction of this protein preparation was the substrate glycoform, and the unhydroxylated isoform present in this preparation is an inhibitor (see "Discussion"). The V_max value with respect to Skp1A-Myc was 12.6 nmol/h/mg, which was slightly greater than that determined for UDP-GlcNAc probably because Skp1A-Myc was not saturating in that reaction. Although the V_max observed with respect to the synthetic peptide, 4.2 nmol/h/mg, was about 1/3 the value for full-length Skp1A-Myc, this small difference contrasted sharply with the 3000-fold difference in apparent K_m values between these two substrates. This suggests that although the peptide lacks information critical for efficient recognition by the GlcNAc-transferase, it is almost fully compatible for interacting with the catalytic site of the enzyme.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The reducing terminus of the Skp1 pentasaccharide is a GlcNAc residue that is linked to HyPro-143, based on prior mass spectrometry studies (6). In the present study, an assay was developed to search for and characterize an enzyme that might be responsible for catalyzing the synthesis of this linkage in vivo. This search turned up a protein in a cytosolic extract of Dictyostelium, Gnt51, that is capable of transferring [³H]GlcNAc from UDP-[³H]GlcNAc to Skp1-HyPro-143.

Specificity and Properties of the Skp1-HyPro GlcNAc-transferase-- The GlcNAc-transferase activity appeared to catalyze the formation of a GlcNAc-HyPro-143 linkage based on the following evidence. 1) Skp1A-Myc from strain HW120, but not Skp1A- His₁₀ expressed in Escherichia coli or Skp1 from the normal strain Ax3, was an acceptor substrate (Figs. 1, A and B, and 6B). Skp1A-Myc is the only one of these Skp1 preparations with an unsubstituted HyPro at position 143 (6). 2) A 23-amino acid synthetic peptide corresponding to the Pro-143 region of Skp1 was a competitive acceptor when it contained 4-HyPro-143 but not with a Pro at this position (Fig. 6C). 3) ³H could be recovered from the acceptor as [³H]GlcNH₂ by acid hydrolysis, consistent with the transfer of [³H]GlcNAc (Fig. 5). 4) The linkage of [³H]GlcNAc was alkali-resistant like that found in native Skp1 (26), unlike the alkali-labile linkage formed between GlcNAc and Ser or Thr (27). The ability of 4-HyPro peptide to serve as a substrate with a V_max only 3 times less than that of Skp1A-Myc (Fig. 6) suggests that Skp1 HyPro-143 is normally 4-OH-substituted, but this remains to be experimentally demonstrated.

The enzyme activity preferred a pH of 7.5-8.0, 5-10 mM Mg²⁺, and 5-10 mM DTT (Fig. 3). The enzyme was inhibited at higher ionic strengths, but less so by K⁺ than Na⁺. The purified enzyme was activated and stabilized by Tween 80 or Tween 20, which did not alter its apparent M_r on Superdex gel filtration columns. Although the enzyme bound phenyl (Low- sub)-Sepharose at high ionic strength, it was readily eluted at low ionic strength (Fig. 2B), thus not exhibiting a hydrophobic character that might mediate membrane association.

The activity exhibited unusually low K_m values, in the submicromolar range, for both its natural donor and acceptor substrates (Fig. 6, A and B). The K_m with respect to Skp1A-Myc was, however, approximate, because 1) only a fraction of the Skp1A-Myc pool consisted of the isoform containing unsubstituted HyPro-143, and 2) the isoform containing unhydroxylated Pro-143 was probably inhibitory, as a 9-fold excess of Skp1A-His₁₀, which is not hydroxylated at Pro-143 and was not a substrate (Fig. 6B), inhibited the enzyme by 50%.³ The peptide substrate exhibited a K_m value that was >3000-fold higher than the value for Skp1A-Myc but a V_max value that was only 3-fold lower. Thus it appears that the peptide contains necessary information for interacting with the enzyme-active site, but the other part of the protein is important for substrate recognition at low Skp1 concentrations.

Throughout its purification up until the final step, the enzyme activity exhibited an apparent M_r of 45,000 based on gel filtration compared with globular protein standards (Fig. 2C). The apparent M_r of Gnt51, determined by SDS-PAGE to be 51,000 (Fig. 4C), conformed with the gel filtration-derived value and suggests that the enzyme functions as a monomeric species with an M_r of 51,000.

The specificity of the enzyme for Skp1-HyPro-143 and its high affinity for this substrate suggests it is the GlcNAc-transferase that modifies Skp1 in vivo. During each step of the purification from the S100 fraction, the enzyme purified as a single species (Figs. 2 and 4), suggesting that there is only a single enzyme with this activity. No activity was detected in the P100 (microsomal) or P3 (nuclear) particulate fractions of the cell in the presence or absence of detergent (Fig. 1C).

There are about 4000 molecules of Gnt51 per cell, based on the recovery after purification, 1.3 µg from 4.4 g of starting material, after correction for the 95.2% loss of activity. With a V_max of about 21 nmol/mg/h for the fully purified enzyme or 1.1 molecules of substrate processed per h per molecule of enzyme, this corresponds to only about 4500 reaction events per h per cell. This is considerably less than the multi-µmol/mg/h V_max values typically seen for Golgi GlcNAc-transferases and the 550 nmol/mg/h value measured for the Ser/Thr GlcNAc-transferase found in the cytoplasmic compartment (11). Given the estimate of 50,000 copies of Skp1 per cell (26) and a cell doubling time of 8 h, this activity would be barely sufficient to support the glycosylation of Skp1 if it is the only substrate. However, the level of Skp1-HyPro GlcNAc-transferase activity is far less than that of two other enzymes in this pathway, the Skp1 prolyl hydroxylase⁴ or the Skp1 Fuc-transferase (17). Thus the enzyme may be down-regulated under the conditions of our assay. That the enzyme is regulated is suggested by the lag phase of 12-15 min when the assay was started by the addition of either of the substrates rather than the enzyme (Fig. 3E), and the ability of additions of either substrate to extend the activity of long term assays. If the loss of activity during purification were completely the result of down-regulation, i.e. there was 100% recovery of enzyme protein, this would yield a specific activity of 438 nmol/mg/h, which is comparable to the activity (550 nmol/mg/h) reported for the cytoplasmic Ser/Thr GlcNAc-transferase (11).

Purification of the Skp1-HyPro GlcNAc-transferase-- Key to the purification of the Skp1 GlcNAc-transferase activity was the synthesis of the 5-(3-aminoallyl)-UDP-GlcNAc-Sepharose fast-flow resin. The 5-position of the uracil ring was chosen for tethering the UDP-GlcNAc because analogues that were modified at this or the 4-position were the best inhibitors (Table II), indicating that the enzyme was most tolerant of modifications there. Previously, a 5-Hg-UDP-GlcNAc-resin had been used to purify a Golgi GlcNAc-transferase (10), but this resin substituted at 18 µmol/ml only retained 10% of the HyPro GlcNAc-transferase activity (data not shown). Possibly, binding to UDP-GlcNAc required a longer spacer for the resin linkage, or the presence of DTT, absent due to the lability of the Hg-linkage to reducing agent but required for enzyme activity. The synthesis of 5-(3-aminoallyl)-UDP-GlcNAc-Sepharose circumvented each of these potential problems.

The 5-Hg-dUMP-Sepharose purification step reduced activation of the partially purified GlcNAc-transferase activity by uracil (Fig. 3G) and related molecules. Presumably this was due to depletion of nucleotide-binding proteins that competed with the GlcNAc-transferase for access to UDP-GlcNAc.

Comparison with Other GlcNAc-transferases-- As described above, the Skp1-HyPro GlcNAc-transferase exhibits high affinities for its donor and acceptor substrates that are uncharacteristic of previously described Golgi GlcNAc-transferases (e.g. Ref. 10). Instead, the HyPro GlcNAc-transferase resembles the recently cloned cytoplasmic Ser/Thr GlcNAc-transferase, which also directly modifies hydroxyl amino acid side chains (14). However, these two enzymes are distinctive in terms of their acceptor amino acids, apparent M_r values, state of oligomerization, sensitivity to monovalent cations, dependence on divalent cation, and requirement for DTT.

Another cytoplasmic protein-Ser/Thr GlcNAc-transferase is the -toxin produced by Clostridia novyi (28). This high molecular weight protein translocates across the mammalian plasma membrane and modifies Thr-37 of Rho and related proteins. Based on comparison with related glucosyltransferase toxins, the catalytic domain has an M_r of <60,000 and is dependent on divalent cations for activity (29). The -toxin shares no obvious sequence homology with the endogenous Ser/Thr GlcNAc-transferase but does share the characteristic of divalent cation dependence with the Skp1-HyPro GlcNAc-transferase and apparently functions as a monomer with a catalytic domain of approximately the same size as Gnt51.

Compartmentalization of the Skp1-HyPro GlcNAc-transferase-- Skp1 functions in the cytoplasm and nucleus, yet complex protein glycosylation is generally thought to occur in vesicles of the secretory pathway, including the rER and Golgi. Previous biochemical studies on the Skp1 Fuc-transferase suggested that it resided in the cytoplasm, raising the possibility that the entire six-enzyme pathway required, on theoretical grounds, to modify Skp1-Pro-143 might also be found in the cytoplasm. The present study on the first glycosyltransferase in the pathway, the Skp1-HyPro GlcNAc-transferase, yielded results consistent with this prediction. The enzyme was readily detectable in the cytosolic fraction of the cell after cell lysis (Fig. 1). The enzyme exhibited an absolute dependence on DTT (Fig. 3D), which is likely incompatible with functioning in the oxidizing environment of the secretory pathway. Finally, the high affinity of the enzyme for its donor and acceptor substrates (Fig. 6) also seems more compatible with a cytoplasmic compartmentalization, where substrates are not as concentrated as they are in the secretory pathway. Confirmation of this model will depend on isolating the Skp1-HyPro GlcNAc-transferase gene, which will also allow verification that this is the enzyme that modifies Skp1 in vivo.

	ACKNOWLEDGEMENTS

The generous gift of 5-ASA-UDP-GlcNAc by Dr. A. E. Elbein, University of Arkansas, is gratefully acknowledged. Peptides were synthesized by Alfred Chung of the University of Florida ICBR Protein Chemistry Core Lab. The expertise of Brandon Parker of the University of Florida ICBR Glycobiology Core Lab in operating the Dionex HPLC is appreciated, and Stephen Larner is thanked for help in the immunoprecipitations.

	FOOTNOTES

^* This work was supported in part by National Institutes of Health Grant RO1-GM37539 and funds from the University of Florida.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Present address: Dept. of Pathology and Laboratory Medicine, Emory University, 1364 Clifton Rd. N.E., H-185, Atlanta, GA 30322.

^§ To whom correspondence should be addressed: Box 100235, 1600 SW Archer Rd., Gainesville, FL 32610-0235. Tel.: 352-392-3329, Fax: 352-392-3305; E-mail: westcm@college.med.ufl.edu.

¹ S. Compton, K. Dobson, M. Sweetinburgh, and C. M. West, unpublished data.

³ H. van der Wel, P. Teng-umnuay, and C. M. West, unpublished data.

⁴ C. M. West and P. Teng-umnuay, unpublished data.

	ABBREVIATIONS

The abbreviations used are: HyPro, hydroxyproline; 5-ASA-UDP-GlcNAc, 5-[3-(p-azidosalicylamido)-1-propenyl]-uridine diphosphate GlcNAc; DTT, dithiothreitol; GlcNH₂, D-glucosamine; mAb, monoclonal antibody; HPLC, high pressure liquid chromatography; MES, 4-morpholineethanesulfonic acid; PAGE, polyacrylamide gel electrophoresis; rER, rough endoplasmic reticulum.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES