Lysosomal enzyme targeting. N-Acetylglucosaminylphosphotransferase selectively phosphorylates native lysosomal enzymes.

Lysosomal enzymes contain 6-phosphomannosyl moieties which mediate their translocation to lysosomes. This recognition marker is synthesized by the sequential action of UDP-N-acetylglucosamine:lysosomal enzyme N-acetylglucosamine-1-phosphotransferase and alpha-N-acetylglucosaminyl phosphodiesterase. A new assay for the N-acetylglucosaminylphosphotransferase, using alpha-methylmannoside as acceptor, is presented. Using this assay, we partially purified the transferase and examined its substrate specificity. The transferase exhibited a very high affinity toward lysosomal enzymes (apparent Km values of less than 20 microM) and was greater than 100-fold more efficient (Vmax/Km) when using lysosomal enzymes as acceptors as compared to nonlysosomal glycoproteins that contain high mannose oligosaccharide units. Heat denaturation of the lysosomal enzymes resulted in the loss of acceptor activity. The model compounds alpha-methylmannoside and Man5--8GlcNAc were poor acceptors. We propose that this enzyme catalyzes the initial, determining step by which synthesized acid hydrolases are distinguished from other newly synthesized glycoproteins and thus are eventually targeted to lysosomes.

the phosphomannosyl residues is now established, little information is available pertaining to the specificity of the process for lysosomal enzymes. In this report, we examine the substrate specificity of partidy purified rat liver N-acetylglucosaminylphosphotransferase. The transferase exhibits a striking preference for acid hydrolases over other potential glycoprotein acceptors. We propose that N-acetylglucosaminylphosphotransferase is responsible for the initial recognition of lysosomal enzymes as distinct from other newly synthesized glycoproteins.
General Methods-Protein concentrations were determined by a modification of the method of Lowry et al. (19) or by absorbance a t 280 nm. Descending paper chromatography was performed using Solvent D as previously described (10).
Assay of N-Acetylglucosaminylphosphotransferase Using a-Methylmannoside as Acceptor-Each incubation contained 150 ~L M [P-"'P] UDP-GlcNAc (usually 300,000 cpm), 100 mM a-methylmannoside, 5 mM ATP, 10 mM MnC12, 10 mM MgC12, 250 p~ dithiothreitol, 100 pg of bovine serum albumin (preincubated 30 min at 56 "C), 0.3% (w/v) Lubrol PX, 50 mM Tris-HC1, pH 7.4, and enzyme in a final volume of 0.05 ml in a 1.5-ml polypropylene tube. Following incubation at 37 "C for 30 min, the reactions were stopped by placement in a boiling water bath for 5 min. Next, 1.3 ml of 2 mM Tris base was added and the tubes were centrifuged a t 12,ooO X g for 5 min. The supernatant was applied to a column of QAE-Sephadex (0.5 X 3.0 crn) and washed with 2 ml of 2 mM Tris base. The Gl~NAc-["~P]a-methylmannoside was selectively eluted with four 1-ml aliquots of 30 mM NaCl in 2 rnM Tris base. The four fractions were pooled and counted after the addition of 12 ml of scintillation mixture. Control incubations contained either no enzyme, 40 mM EDTA, or a-methylgalactoside in place of the a-methylmannoside. All three controls gave similar background values of <0.01% of the radioactivity used. Evidence that this assay measures N-acetylglucosaminylphosphotransferase activity includes the fact that fibroblasts from patients with mucolipidosis I1 (I-cell disease) have <1% of the activity of normal controls (20). Further details of the validation of the assay and proof of the product will be presented in a subsequent publication. One unit of enzyme activity is defined as the transfer of 1 pmol of N-acetylglucosamine 1-[3zPJphosphate/h a t 37 "C using 100 mM n-methylmannoside.
Assay of N-Acetylglucosamine 1-Phosphate Transfer to Oligosaccharides-The isolation of high mannose-type oligosaccharides from immunoglobulin M (Wa) is described elsewhere (20). N-acetylglucosamine phosphate transfer to the oligosaccharides was assayed as described for transfer to a-methylmannoside except that the final assay volume was reduced to 0.01 ml and the product was eluted off the QAE-Sephadex with 20 mM NaC1/2 mM Tris base. The 'lP product bound to Con A-Sepharose and after hydrolysis (N HC1,4 h, 100 "C), co-migrated with mannose 6-phosphate upon paper chromatography.
' The abbreviation used is: Con A, concanavalin A.

Lysosomal Enzyme Phosphorylation
Assay of N-Acetylglucosamine 1 -Phosphate Transfer to Glycoproteins-The incubation mixtures contained 150 p [p-32P]lJDp-GlCNAC (1-2 X lo6 cpm), 5 mM ATP. 10 mM MnC12, 10 mM MgC12, 250 pM dithiothreitol, 0.1 to 1.0% (w/v) Lubrol PX, 50 mM N-acetylglucosamine, 50 mM Tric-HCI, pH 7.45, N-acetylglucosaminylphosphotransferase (usually 150 to 250 units), and acceptor protein in a volume of 0.05 ml. After incubation (37 "C, 30 min) the reactions were stopped by the addition of 0.1 ml of 1.5% (w/v) phosphotungstic acid/ 0.75 N HC1 at 0 OC. After 5 min at 0 "C, the tubes were centrifuged (5 min, 12,000 X g, 4 "C) and the supernatant was discarded. The pellets were washed with 1 ml of 1% phosphotungstic acid/0.5 N HC1, centrifuged as above, and resuspended by sonication in 0.5 ml of 100 mM glucose 6-phosphate/20 mM CaC1,/100 mM Tris-HC1, pH 8.0. One mg of pronase was added and the samples were incubated at 56 "C. After 1 h, the samples were heated at 100 "C for 10 min and then cooled. One ml of water was added and the samples were centrifuged (5 min, 12,000 X g, room temperature). The supernatant was applied to a pipette column (0.5 X 1.5 cm) of Con A-Sepharose and washed with 100-200 ml/column of 150 mM NaCL/10 mM Na-phosphate, pH 7.5, in 1 h. The Con A-Sepharose was then extruded from the column and counted in a scintillation spectrometer after the addition of 1 ml of water and 10 ml of 3a70. Care was taken so that the capacity of the Con A-Sepharose (-60 nmol of high mannose oligosaccharide chain/ ml of resin) was not exceeded. When the washed protein pellet was hydrolyzed (N HC1,4 h, 100 "C), desalted on Dowex 50, H' form (0.5 X 4.5 cm), and subjected to paper chromatography, the amount of 32P co-migrating with mannose 6-phosphate was the same as that binding to Con A-Sepharose in the standard assay.
Purification of Rat Liver N-AcetylglucosaminyZphosphotransferase-Details of the enzyme preparation will be presented elsewhere. Briefly, smooth membranes were prepared from rat liver (13) and extracted with 10 mM mannose 6-phosphate/fO mM Namaleate/O.l% (w/v) Lubrol PX followed by 5 nm MgCId25 mM Tris-HCI, pH 7.4/0.1% Lubrol PX. The membranes were then repeatedly extracted with 5 mM MgCld25 mM Tris-HC1, pH 7.4/0.3% Lubrol PX. The 0.3% Lubrol extracts were pooled and chromatographed on DEAE-cellulose. Fractions were assayed using a-methylmannoside as acceptor and the active fractions were pooled and concentrated by batch elution of a small DEAE-cellulose column. Typically, this pool was 75-110-fold purified with a 30% yield as compared to the initial homogenate which had a specific activity of 40-50 units/mg of protein. Some preparations of the N-acetylglucosaminylphosphotransferase were free of detectable endogenous acceptor activity whereas others had a small amount of endogenous acceptor activity.

RESULTS
Partial Purification of N-acetylglucosaminylphosphotransferase-Our previous assay for N-acetylglucosaminylphosphotransferase activity depended on the transfer of Nacetylglucosamine 1-["'Plphosphate to endogenous glycoprotein acceptors (10). Therefore, in order to purify the enzyme it was necessary to develop an assay which is independent of endogenous acceptors. We found that the enzyme can transfer N-acetylglucosamine l-[32P]phosphate to a-methylmannoside and this forms the basis for a sensitive, specific, and facile assay (see "Experimental Procedures"). Using this assay to follow activity, the transferase was solubilized from rat liver smooth membranes with 0.3% Lubrol PX and partially purified by DEAE-cellulose chromatography. The final preparations were 75-110-fold purified. The enzyme had a pH optimum of 7 to 8 with 50% activity at pH 6 or 9 and less than 1% activity at pH 4 to 5. The apparent K , for UDP-GlcNAc was 21 pM.
Activity toward Glycoprotein Acceptors-Using such partially purified preparations, we determined the ability of the enzyme to phosphorylate the oligosaccharide units of various exogenous glycoproteins. Table I shows the results of a typical experiment using a series of glycoprotein acceptors that have in common the presence of high mannose-type oligosaccharide units. It is evident that the lysosomal enzyme a-N-acetylglucosaminidase is the best acceptor, being active at low protein concentrations. In contrast, none of the five nonlysosomal glycoproteins are good acceptors. To facilitate comparison, the acceptor concentrations have been expressed in terms of the concentration of high mannose-type oligosaccharide units. The best nonlysosomal acceptor on a weight basis, ribonuclease B, has an unusually large oligosaccharide content. In other experiments, we found that human placental P-hexosaminidase A, human placental &hexosaminidase B, human hepatic &galactosidase, and partially purified porcine hepatic P-hexosaminidases A and B were all comparable in acceptor activity to the a-N-acetylglucosaminidase.
In order to quantitatively compare the acceptor activity of lysosomal enzymes with other possible acceptors, we measured the rate of N-acetylglucosamine 1-[32P]phosphate transfer as a fmction of acceptor concentration. Fig. 1 shows that a-N-acetylglucosaminidase is an excellent acceptor with an apparent K,,, of 8.9 p~. Ribonuclease B, on the other hand, is a much poorer acceptor with an apparent K, of 916 p~. The V,,, values for these two acceptors are similar. Studies with human placental /3-hexosaminidases A and B showed that the apparent K, values for these substrates are in the low micromolar range ( Table 11). The acceptor activity of the other nonlysosomal glycoproteins was so low that it was not possible to determine their apparent K,,, and V,,, values. Table I1 also presents the kinetic parameters for two carbohydrate acceptors, a-methylmannoside and Man5.aGlcNAc oligosaccharide.
These molelcules have very high apparent K , values, being 1O3-lO4-fold greater than those of lysosomal enzymes. Their Vmax values are 10-100-fold greater than those of the glycoprotein acceptors.
The relative catalytic efficiency ( V,,, divided by apparent K,) of the transferase toward the various acceptors is also shown in Table 11. The three lysosomal enzymes are phosphorylated at least 100-fold more efficiently than either ribonuclease B, Man5.aGlcNAc Oligosaccharide, or a-methylmannoside. Different preparations of the transferase exhibited vari-

TABLE I1
Kinetic parameters of N-acetylglucosaminylphosphotransferase activity toward various acceptors Enzyme assays were performed as detailed under "Experimental Procedures." Two enzyme preparations with the same specific activity were used in the emeriments. able activity toward a-N-acetylglucosaminidase relative to amethylmannoside, ranging from 2% (as in Table 11) to 0.7%. The reason for this variation is not clear.
To probe the substrate specificity of the N-acetylglucosaminylphosphotransferase further, lysosomal enzymes were heat denatured and then used as acceptors. Fig. 2 shows one such experiment. Incubation of a-N-acetylglucosaminidase at 67 "C resulted in progressive denaturation as indicated by the loss of hydrolase activity. The denatured enzyme also lost its ability to serve as an acceptor of N-acetylglucosamine 1-["P] phosphate even though the protein remained in solution.
Similar results were obtained in experiments using human placental /%hexosaminidase A and partidy purified porcine hepatic /?-hexosaminidases A and B. Although the loss or modification of some heat-labile moiety cannot be ruled out, these results strongly suggest that acceptor protein conformation is recognized by N-acetylglucosaminylphosphotransferase.

DISCUSSION
The data presented in this paper demonstrate that UDP-N-acetylg1ucosamine:lysosomal enzyme N-acetylglucosamine-1-phosphotransferase uses lysosomal enzymes as substrates greater than 100-fold more efficiently than it does other glycoproteins which also contain high mannose-type oligosaccharides. The apparent K,,, values of the transferase for lysosomal enzymes (t20 p~) are among the lowest known for the reaction of a glycosyltransferase with its glycoprotein substrate (31). It is currently felt that the sequence and/or conformation of the polypeptide portion of a glycoprotein influences the processing of its asparagine-linked oligosaccharide units. However, there are few studies assessing the influence of the polypeptide backbone on the action of a particular processing enzyme or glycosyltransferase. The preference of the N-acetylglucosaminylphosphotransferase for lysosomal enzymes in these in vitro assays is a demonstration of the remarkable specificity of a glycosyltransferase for the class of glycoproteins on which it acts in vivo.
The demonstration that heat denaturation of acid hydrolases destroys their ability to accept N-acetylglucosamine 1phosphate is intriguing. Although other interpretations are possible, the most plausible explanation is that the specificity of the N-acetylglucosaminylphosphotransferase is for a particular protein conformation that is unique to lysosomal enzymes. The nature of this conformational requirement i s obscure at this time. However, since both ribonuclease B (with Lysosomal Enzyme Phosphorylation an exposed oligosaccharide unit (17, 32)) and free high mannose oligosaccharides are poor acceptors, it is unlikely that accessibility of the carbohydrate chain to the transferase is the sole conformational requirement for recognition.
With the demonstration of its specificity, the N-acetylglucosaminylphosphotransferase reaction becomes the earliest known point at which newly synthesized acid hydrolases are distinguished from other newly synthesized glycoproteins. While it is possible that lysosomal enzymes undergo some other specific modification prior to exposure to the transferase and that the observed specificity of the phosphorylation reaction is secondary to this other (heat-labile) modification, there is no reason to postulate this more complex pathway. Similarly, the specificity of the transferase makes it unnecessary to postulate the existence of a mechanism for the specific segregation of newly synthesized acid hydrolases which would precede the exposure to the N-acetylglucosaminylphosphotransferase. Based on the data presented, we propose that the N-acetylglucosaminylphosphotransferase is the initial and determining enzyme for the pathway which eventually results in the segregation of acid hydrolases into lysosomes.