A Di-N-acetylchitobiase Activity Is Involved in the Lysosomal Catabolism of Asparagine-linked Glycoproteins in Rat Liver*

Transcription

1 THE JOURNAL OF BIOLOGICAL CHEMISTRY by The American Society of Biological Chemists, Inc. Vol. 261, No. 13, Issue of May 5, pp Printed in ~.s.a. A Di-N-acetylchitobiase Activity Is Involved in the Lysosomal Catabolism of Asparagine-linked Glycoproteins in Rat Liver* (Received for publication, August 29, 1985) Michael J. Kuranda and Nathan N. Aronson, Jr. From the Department of Molecular and Cell Biology, Althuse Laboratby, The Pennsylvania State University, University Park, Pennsylvania A perfused rat liver was used to study the effects of 5-diazo-4-oxo-~-norvaline on lysosomal glycoprotein catabolism. Addition of this compound (1.0 mm) to the perfusate reduced activity of 8-aspartyl-N-acetylglucosylamine amidohydrolase by 99% in 1 h. Treated livers were unable to completely degrade endocytosed N-acetyl[14C]glucosamine-labeled asialo-al-acid glycoprotein as evidenced by a 50% reduction in radiolabeled serum glycoprotein secretion compared to controls. This decreased degradation was matched by a lysosomal accumulation of glycopeptides with the structure: GlcNAcB(1-4)GlcNAc-Asn. The result suggested the presence of an unrecognized glycosidase in rat liver lysosomes, since this remnant was extended by one more GlcNAc residue than would have been expected after specific inactivation of the amidohydrolase. Such a novel enzyme would therefore catalyze cleavage of the N-acetylglucosamine residue at the reducing end of al-acid glycoprotein oligosaccharides only following removal of the linking Asn. The activity was then detected in lysosomal extracts by using intact asialo-biantennary oligosaccharides labeled with [ HI galactose or N-a~etylI~~C]glucosamine residues as a substrate. To prevent simultaneous digestion of the material from its nonreducing end, 8-D-galactosidase in the enzyme extract was first inactivated with the irreversible active site-directed inhibitor, 8-D-galactopyranosylmethyl-p-nitrophenyltriazene. The observed di-n-acetylchitobiose cleaving activity worked optimally at ph 3.4 and was uniquely associated with the lysosomal fraction of the liver homogenate. The enzyme also cleaved triantennary chains and di-nacetylchitobiose, but failed to hydrolyze substrates that had been reduced with NaBH4. The new glycosidase was well separated from N-acetyl-B-D-glucosaminidase (assayed with p-nitrophenyl-8-d-glucosaminide) by gel filtration chromatography and had an apparent molecular weight of 37,000. A similar enzyme that hydrolyzes di-n-acetylchitobiose had previously been found in extracts of human liver (Stirling, J. L. (1974) FEBS Lett. 39, ). *This research was supported by United States Public Health Service Grant AM from the National Institute of Arthritis, Diabetes, Digestive and Kidney Diseases and by a grant from the Pennsylvania State University Agricultural Experiments Station. Authorized for publication as Paper 7244 in the journal series of the, Pennsylvania Agricultural Experiment Station. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact Numerous pathological conditions which result from genetically defective lysosomal exoglycosidases have been characterized (1,2). A consequence of many of these disorders is the accumulation in tissues and urine of a set of oligosaccharide fragments which appear to be derived from the. aberrant catabolism of asparagine-linked glycoproteins. Most of these remnants contain a single N-acetylglucosamine residue at their reducing end in place of the di-n-acetylchitobiose moiety which is normally present at that position of the intact glycoprotein. Recently, we found similar structures accumulated in rat liver lysosomes following catabolism of asialo-alacid glycoprotein by tissue that had been pretreated with different glycosidase inhibitors (3). The occurrence of these structures has led to the hypothesis that an endo-n-acetyl-@- D-glucosaminidase is an important step in lysosomal glycoprotein catabolism. Although such an activity has been demonstrated in mammalian tissues, the evidence so far is that this activity is associated with the soluble portion of the cell and not with the lysosomal compartment (23, 24). We have now discovered that the enzyme responsible for hydrolysis of the di-n-acetylchitobiose moiety during asparagine-linked glycoprotein catabolism in rat liver is, in fact, a lysosomal di- N-acetylchitobiase. This unique glycosidase in addition to hydrolyzing di-n-acetylchitobiose also cleaves this disaccharide unit when it occurs at the reducing end of complex-type oligosaccharides. The experiments leading to the discovery of this new lysosomal activity are described in this report. In addition, the importance of this step in the overall scheme of glycoprotein degradation is discussed. EXPERIMENTAL PROCEDURES Preparation of Radiolabeled Glycoproteins and Oligosmchnrides- Serum glycoproteins were labeled with D-[l-14C]glucosamine (56 mci/ mmol, ICN) in the perfused liver system described by Aronson (4). q-acid glycoprotein with 14C-labeled sialic acid and N-acetylglucosamine residues was isolated from the perfusate using a modified procedure of Shibata et al. (5). Serum from turpentine-injected rats was used as a source of unlabeled w-acid glycoprotein and purifed as above. Both preparations were desialylated using insoluble neuraminidase (Sigma) (6). Asialo-cY1-acid glycoprotein with 3H-labeled terminal sugar residues was prepared by oxidizing the carbon 6 primary alcohol of galactose residues to an aldehyde with galactose oxidase followed by reduction back to an alcohol with sodium borotritide as described by Kuranda and Aronson (3). Radioactive oligosaccharides were produced by treating samples of sugar-labeled asialo-al-acid glycoprotein ( pg) with anhydrous hydrazine followed by re- N-acetylation with acetic anhydride according to the method of Takasaki et al. (7). The released sugar chains were treated with 0.05 N H2S04 for 1 h at 80 C to remove any remaining sialic acid residues not hydrolyzed by neuraminidase treatment. Purified bi- and triantennary complex-type oligosaccharides were then preparatively separated on a high resolution Bio-Gel P-4 column as described below. Inhibitors-8-D-Galactopyranosylmethyl-p-nitrophenyl triazene was synthesized from 8-D-galactopyranosylmethylamine as previously described (3). 5-Diazo-4-oxo-N-trifroacetyl-~-norvaline

2 5804 Di-N-acetylchitobiase Activity in Glycoprotein Degraclation methyl ester was synthesized according to the method of Liwschitz et al. (8). The protecting trifluoroacetyl and methyl ester groups were removed by hydrolysis with cold NaOH to yield the inhibitor, 5- diazo-4-oxo-~-norvaline, which was purified by chromatography on a column of celite and charcoal according to Handschumacher et az. (9). Perfusion of Rat Livers-Livers were surgically removed from fasted male Wistar rats weighing from g and perfused in a 100-ml cyclic system maintained at 37 "C as described previously (4). This same apparatus was used for preparing al-acid glycoprotein with "C-labeled sialic acid and N-acetylglucosamine residues and for studying the metabolism of radiolabeled asialoglycoproteins. Preparation of Lysosomal Extracts-Purified lysosomes were isolated,from the livers of rats treated by an intraperitoneal injection of Triton WR-1339 according to the procedure of Leighton et al. (10). In brief, this technique involved isolation of a lysosome-rich pellet by differential centrifugation followed by final purification of the detergent-filled lysosomes in a discontinuous sucrose gradient. Isolated lysosomes removed from the gradient were diluted with an equal volume of 0.25 M sucrose and centrifuged at 100,000 X g for 60 min. The resulting pellet was suspended in HzO to a final protein concentration of approximately 5 mg of protein per ml and stored frozen (-20 "C) in 0.5-ml aliquots. Protein was assayed as described by Stauffer (11) using bovine serum albumin as a standard. Bio-Gel P-4 Chromatography-Radioactive oligosaccharides from crude extracts were separated on a column (2.6 X 78 cm) of Bio-Gel P-4 ( mesh, Bio-Rad) by elution with distilled water. High resolution separations of oligosaccharides were performed on a column (1.5 X 100 cm) of Bio-Gel P-4 (-400 mesh) maintained at 55 'C and pumped with distilled and deaerated water at a flow rate of 18 ml/h (100 p.s.i.) using a "45 solvent delivery system (Waters Associates). Samples ( pl)were applied through a universal liquid chromatography injector (model UK6, Water Associates). Glucose oligomers prepared by the partial hydrolysis of dextran were added to samples as internal standards an detected with a differential refractometer (model R401, Waters Associates) (12). Correlations between collected fractions and recorded detector response were made by means of a Pharmacia fraction collector (Frac-100) equipped with an event,marker. Assay of the Oligosaccharide Di-N-acetylchitobiose Cleaving Activ- ity-five pl of BGalMNT' (0.1 pmol) was added to 100 pl of enzyme sample and incubated at 37 "C for 1 h. A second 5-pl aliquot of PGalMNT was then added followed by an additional 1 h of incubation. Twenty-five pl of this mixture was combined with 25 pl of McIlvaine citrate/phosphate buffer of varied ph (13) and 5 pl of radioactive oligosaccharides (10,000 cpm). Hydrolysis was allowed to proceed for 1-12 h. The reaction was terminated by heating to 100 "C for 5 min. Two hundred pl(4 mg) of standard glucose oligomers were added and the samples were centrifuged briefly to remove precipitated protein. The supernatants were then chromatographed directly on the high resolution Bio-Gel P-4 column described above. Radioactivity which was eluted at 11 glucose units was used to quantitate the amount of di-n-acetylchitobiose cleaving activity present. Enzyme units are expressed in counts/min/h that were eluted at that position. Assay of Di-N-acetylchitobiaseActivity-Di-N-acetylchitobiose (8.0 mm, 50 pl), McIlvaine's citrate/phosphate buffer (150 pl, ph 4.0) (13), and 50 pl of enzyme sample were incubated at 37 "C for 5 h. The reaction was terminated by the addition of 150 pl of 5% sodium borate, ph 8.8. Liberated N-acetylglucosamine was then quantitated by the Morgan-Elson reaction according to Reissig et al. (14). Samples were heated to 100 "C for 3 min and cooled, and 1.5 ml of freshly diluted p-dimethylaminobenzaldehyde reagent was added. The sam- ples were then incubated at 37 "C for 20 min and the optical densities were determined at 585 nm. N-Acetyl-P-D-glucosaminidase Digestion of Glycopeptides-Lyophilized glycopeptides were dissolved in 200 p1 of 0.05 M sodium citrate buffer, ph 5.0. Jack bean N-acetyl-j3-D-glucosaminidase (Sigma) was desalted by repeated concentration and reconstitution with citrate buffer using a Centricon disposable microconcentrator (Amicon). One unit (approximately 20 pl) of the glycosidase was added to the sample and the reaction was incubated at 37 "C. Ten pl of toluene were also added to retard bacterial growth. Eight h later, a second 1 unit aliquot of enzyme was added. The reaction was then allowed to proceed for The abbreviations used are: PGalMNT, P-D-galactopyranosylmethyl-p-nitrophenyltriazine; DONva, 5-diazo-4-oxo-~-norvaline; GMl, major brain monosialoganglioside, Le. I13NeuAc-GgOse4Cer. an additional 12 h. The enzyme was precipitated by heating at 100 "C for 5 min. After centrifugation, aliquots of the supernatant were analyzed by chromatography on Bio-Gel P-4. RESULTS Liver Catabolism of a1-acid Glycoprotein after Treatment with 5-Diazo-4-ozo-~-norualine (D0Nua)-Rat al-acid glycoprotein is a serum protein that contains an unusually high amount of carbohydrate distributed among a total of six biand triantennary asparagine-linked sugar chains (15, 16). Previously we have used this molecule as a model substrate in conjunction with different glycosidase or proteinase inhibitors to study the catabolism of asparagine-linked oligosaccharides in a perfused rat liver system. In this study, we have further extended this work to include the effects of DONva. This asparagine analog has been shown by Tarentino and Maley (17) to be a potent irreversible inhibitor N-acetylglucosylamine amidohydrolase (glycosyl asparaginase) from hen oviduct. Consistent with their observation, perfusate containing 1.0 mm DONva reduced rat liver glycosyl asparaginase by 99% in 1 h (Table I) while all glycosidases involved in glycoprotein catabolism remained unaffected. Normal liver metabolism of [14C]GlcNAc-labeled al-acid glycoprotein involves: 1) capture of these molecules by Gal/ GalNAc receptors on hepatocyte cell surfaces, 2) transport of the substrate to lysosomes, 3) hydrolysis and release of the radiolabeled amino sugar product, and 4) reutilization of a major portion of [14C]GlcNAc residues for serum glycoprotein biosynthesis (27). This process is experimentally illustrated in Fig. 1. Capture of radiolabeled ligand is indicated by the rapid loss of acid-precipitable radioactivity from the perfusate, while reincorporation of free [14C]GlcNAc into secreted serum glycoproteins is evidenced by the return of acid-precipitable radioactivity into the perfusate after approximately 60 min. Treatment with DONva (1.0 mm) for 1 h prior to addition of radiolabeled glycoprotein had no effect on the uptake phase of metabolism but reduced radioactive serum glycoprotein reappearance in the perfusate by approximately 50%. Characterization of the Radwacqiuity in Liver after DONua Treatment-To better assess the direct effects of DONva on lysosomal catabolism of the labeldd al-acid glycoprotein, the livers from DONva and control experiments were subjected to subcellular fractionation according to the method of de- TABLE I Effect of DONva on liver glycosidases Thirty min after the start of perfusion, a 2-g sample was excised from the liver and homogenized in 6 volumes of 0.2% Triton X-100. Sixteen mg of DONva were then added to the perfusate and 60 min later the remainder of the liver was homogenized in 0.2% Triton. The homogenates obtained before and after treatment with DONva were assayed for glycosidase activity and protein (11). 8-Aspartylglucosylamine amidohydrolase activity was assayed using GlcNAc-Asn as a substrate according to the procedure of Mahadevan and Tappel (25). Units are expressed in micromoles of GlcNAc produced per h. The remaining glycosidases were assayed using the correspondingp-nitrophenyl glycosides (26). Units are defined as micromoles of p-nitrophenol released per min. The results are from a single perfusion experiment. Enzyme -DONva +DONva unitslmg protein X 10' N-Acetyl-0-D-glucosaminidase P-D-Galactosidase a-d-mannosidase a-l-fucosidase Aspartylglucosylamine amidohydrolase

3 4 c Di-N-acetylchitobiase Activity in Glycoprotein Degradation d io $ I I I Time (mid FIG. 1. Effect ofdonva on liver metabolism of N-acetyl[ 4C]glucosamine-labeled asialo-cy1-acid glycoprotein. Thirty min after the start of perfusion, 16 mg of DONva was added to the circulation. After an additional 60 min, 350 /.cg of [14C]GlcNAc- labeled asialo-al-acid glycoprotein (1.5 X lo6 cpm) was added. At the indicated times, 0.2-ml samples of perfusate were precipitated with 1.0 ml of cold 4% phosphotungstic acid in 2 N HC1. Radioactivity in the precipitate was measured as described previously (27). 0, DONva added; A, control. 30!- I ::[- _, 2 VI FIG. 2. Bio-Gel P-4 chromatography of oligosaccharides accumulating in the heavy and light mitochondrial fraction ML after treatment with DONva. Livers from the experiments depicted in Fig. l were homogenized in 0.25 M sucrose and each homogenate was separated by centrifugation according to deduve et al. (18). The heavy and light mitochondrial fractions were repelleted by centrifugation, suspended in 10 ml of distilled water, and frozen and thawed three times. These suspensions were centrifuged at 100,000 X g for 75 min. The resulting supernatants contained greater than 90% of the radioactivity associated with the original pellets. Five-ml samples of each extract were chromatographed on Bio-Gel P-4 ( mesh). A, control; B, DONva-treated liver; C, peak I (panel B) digested with jack bean N-acetyl-8-D-glucosaminidase as described under Experimental Procedures. Arrows represent the elution volumes of standards: u,, bovine serum albumin; ut, glucose. Duve et al. (18). The lysosome-rich heavy and light mitochondrial fractions were then extracted and analyzed for radioactive glycopeptides by Bio-Gel P-4 chromatography (Fig. 2). The comparison (Fig. 2, A and B) revealed an additional symmetrical peak of included radioactivity in the DONva extract that was not present in the control. On treatment with jack bean N-acetyl-@-D-glucosaminidase, this extra material yielded two peaks of radioactivity (Fig. 2C). The late eluting peak from the digest represented 40% of the radioactivity and chromatographed identically with free GlcNAc. Fractions containing the early eluting peak were then pooled, lyophilized, and rechromatographed on paper. The radioactive fragment migrated identically with commercial GlcNAc-Asn (Fig. 3). The identity of this material was confirmed by its co-elution with standard GlcNAc-Asn on a high resolution Bio-Gel P-4 column (data not shown). From these, data, it appeared that inactivation of glycosyl asparaginase in situ by DONva resulted in the accumulation of a glycopeptide with the structure GlcNAc/3(1-4)GlcNAc- Asn. Our interpretation of this initial observation was: 1) a separate enzyme other than N-acetyl-@-D-glucosaminidase (see Table I) is responsible for hydrolyzing the di-n-acetylchitobiose component of al-acid glycoprotein in lysosomes, and 2) this reaction occurs from the reducing end of the oligosaccharides only after prior removal of the interfacing Asn by glycosyl asparaginase (see Discussion ). To test this hypothesis, we subjected the entire biantennary oligosaccharide (asialo-form) to lysosomal extracts that contained no 0- D-galactosidase activity. Detection of a Di-N-acetylchitobiose Cleaving Activity in Lysosomal Extracts-Complete asialo-biantennary chains were prepared from [3H]Gal or [14C]GlcNAc-labeled asialoal-acid glycoprotein by hydrazinolysis. These radiolabeled oligosaccharides were initially digested with tritosome extracts which had been preincubated triazene. Since this compound is an irreversible active site-directed inhibitor of rat its use was expected to prevent degradation of the oligosaccharide substrates from their nonreducing ends. Digestion of [3H]Gal biantennary chains with a purified ly- sosomal extract resulted in a limited hydrolysis of the oligosaccharides. Fig. 4A shows the elution profile of the digest on a high resolution Bio-Gel P-4 column. Two peaks of radioactivity are evident. Besides some starting [3H]Gal-oligosaccharide which was eluted at 13 glucose units, only a single digestion product was present at the position of 11 glucose units. No exoglycosidase cleavage of the oligosaccharide occurred during the reaction as was evidenced by the lack of any [3H]Gal release (free Gal is eluted as a single glucose unit). This result was in sharp contrasto the elution profile of a control digest done with tritosomes not pretreated with PGalMNT (Fig. 4B). In that experiment, nearly all radioactivity was eluted as [3H]Gal. Incubation of [ C] GlcNAc-labeled oligosaccharide with extract yielded a slightly different set of fragments (Fig. 4C). Besides the substrate, two peaks of degradation products were observed; one was eluted at 11 glucose units as before while the other was eluted at two glucose units. The IO Distance (cm) latter new peak co-chromatographed with free GlcNAc. When [14C]GlcNAc-labeled oligosaccharide was digested with the FIG. 3. Identification of N-acetyl-8-D-glucosaminidase digestion products by paper chromatography. Peak I1 (Fig. 2, panel C) was lyophilized and the residue was spotted on Whatman No. 1 paper and separated by descending chromatography using 1- butanol/acetic acid/water (41:5) (28). Indicated standards were run concurrently in adjacent lanes and detected with ninhydrin. Portions of the chromatogram from lanes containing radioactivity were cut into strips (2.0 X 0.5 cm) which were placed in 1 ml of Hz0 in the bottom of 20-ml scintillation vials. Radioactivity was measured after adding 10 ml of a commercial scintillation fluid.

4 5806 Di-N-acetylchitobiase Activity in Glycoprotein Degradation l. 5 ~ l I.o A D Assay of the Di-N-acetylchitobiose Cleaving Activity in Subcellular Fractions of Rat Li~er-[~H]Gal-labeled biantennary oligosaccharides were chosen as the substrate since both the di-n-acetylchitobiose cleaving activity and P-D-galactosidase could be followed simultaneously. The former reaction was quantitated by the amount of radioactivity which was eluted at 11 glucose units on Bio-Gel P-4 and the latter activity was monitored by the free [3H]Gal released. Using tritosome extracts treated with PGalMNT as an initial enzyme source, the velocity of the di-n-acetylchitobiose cleavage was found to be proportional to the reaction time up to at least 10% hydrolysis of the substrate (Fig. 5). In these experiments only a minimal amount of exoglycosidase activity ([3H]Gal release) was observed. The ph optimum for cleavage of the di-n-acetylchitobiose core was at 3.4 (Fig. 6). All subsequent assays were conducted at this ph. Using the above protocol to quantitate di-n-acetylchitobiose core cleaving activity, we assayed subcellular fractions of rat liver prepared according to the method of deduve et al. (18). Fig. 7 summarizes the results from the fractionation study. The highest relative specific activity of the new glycosidase was found in the light mitochondrial fraction. The pattern of distribution was essentially identical with that for N-acetyl-P-D-glucosaminidase, a known lysosomal marker en- Volume (mi) zyme. The sedimentation properties of the newly discovered FIG. 4. Digestion of biantennary oligosaccharides with B- di-n-acetylchitobiose hydrolyzing activity taken together D-galactosidase-deficient lysosomal extracts. Lysosomal extracts were treated with PGalMNT as described under Experimental with its very acidic ph optimum strongly indicated that this Procedures. Incubation with the inhibitor inactivated greater than hydrolase was located in the lysosomal compartment. 95% of the 8-D-galactosidase activity as assayed usingp-nitrophenyl- Based on the commercial availability of di-n-acetylchitop-d-galactopyranoside as a substrate. Radiolabeled biantennary chains (10,000 cpm) were incubated with control and 8-D-galactosidase-deficient lysosomal extracts for 6 h at ph 5.0. The digests were T-0 then analyzed by Bio-Gel P-4 chromatography (-400 mesh). A, [3H] galactose-labeled oligosaccharides plus PGalMNT; B, [3H]galactoselabeled oligosaccharides minus PGalMNT; C, N-a~etyl[ ~C]glucosamine oligosaccharides plus PGalMNT; D, N-acetyl[ 4C]glucosamine oligosaccharides minus PGalMNT. Numbered arrows represent peak elution volumes of standard glucose oligomers (12). The elution position of both labeled intact biantennary oligosaccharide substrates is indicated in panel A (0. complete tritosome extract (not treated with PGalMNT), a series of different degradation products resulted with the major peak being free [14C]GlcNAc(Fig. 40). The partial 40-50% hydrolysis of the biantennary substrate seen in the 6-h incubations (Fig. 4, A and C) probably was due to the presence of oligosaccharides that were chemically altered at their reducing end N-acetylglucosamine residue. Hydrazinolysis is known to cause isomerization and degradation reactions to the reducing termini (7) and such modified chains would not likely be acted upon by the di-n-acetylchitobiase activity. The above results indicated that when the normal sequence of lysosomal exoglycosidase cleavages was blocked by 0Gal- MNT, in vitro catabolism of the biantennary oligosaccharide was limited to removal of only a single N-acetylglucosamine residue at the reducing end of the chains. Thus, a major peak of degradation product was eluted from the Bio-Gel P-4 column 2 glucose units later than the intact chains (Fig. 4, A and C). In previous perfused rat liver experiments using BGalMNT in situ, a fragment with identical elution properties was characterized by us to be similar to a GM1 oligosaccharide which lacks the reducing end GlcNAc. This cleavage at the reducing end was further confirmed by the release of free [ C] GlcNAc following digestion of the [ CIGlcNAc-labeled chains with a 0-galactosidase-inhibited extract (Fig. 4C). Based on this in vitro phenomenon, we devised an assay to determine the subcellular localization of the unique glycosidase responsible for this cleavage in hepatocytes. Time ( h 1 FIG. 5. Lysosomal hydrolysis of reducing end N-acetylglucosamine and [3H]galactose residues from [SH]galactose-labeled biantennary oligosaccharide chains as a function of time. Lysosomal extracts were treated with PGalMNT and then mixed with [3H]Gal-biantennary oligosaccharides (see Experimental Procedures ). At the indicated times, the reactions were terminated by heating to 100 C for 5 min. The digests were separated on Bio- Gel P-4 (-400 mesh) and the amount of radioactivity which was eluted at 11 glucose units and 1 glucose unit ([3H]galactose) was quantitated by scintillation counting. All hydrolysis reactions were run at ph 5.0.0, radioactivity which was eluted at 11 glucose units; W, [3H]galactose PH FIG. 6. Assay of the di-n-acetylchitobiose cleaving activity as a function of ph. Lysosomal extracts treated with PGalMNT were buffered at the indicated ph values and mixed with [3H]galactose-labeled oligosaccharide substrate (see Experimental Procedures ). The digests were chromatographed on Bio-Gel P-4 (-400 mesh) and the amount of radioactivity which was eluted as 11 glucose units was quantitated.

5 Di-N-acetylchitobiase Activity in Glycoprotein Degradution 5807 a 1 1 ~ , Percent Total Protein I N I M I L I P I S I Subcellular Froction FIG. 7. Subcellular distribution of N-acetyl-8-D-glucosaminidase and di-n-aeetylchitobiose cleaving activity in rat liver. Rat liver was homogenized and fractionated according to the method of deduve et al. (18) into a nuclear (N), mitochondrial (M), light mitochondrial (L), microsomal (P), andsupernatant (5 ) fraction. These fractions were then assayed for protein (ll), N-acetyl-8-Dglucosaminidase (26), and di-n-acetylchitobiose cleaving activity (see Experimental Procedures ). The latter glycosidase was assayed at ph 3.4. A, N-acetyl-8-glucosaminidase; B, di-n-acetylchitobiose cleaving enzyme. Y a Volume (mi) FIG. 8. Separation of rat liver activities cleaving di-n-acetylchitobiose by molecular sieve chromatography. A rat liver was homogenized in 2 volumes of M acetic acid, 0.1 M NaCl. Precipitate was removed by centrifugation at 25,000 X g for 30 min and the supernatant was dialyzed overnight at 4 C against 20 volumes of 0.05 M sodium phosphate, 0.15 M NaC1, ph 7.0. Any additional precipitate that formed during dialysis was removed by centrifugation and a 2.0-ml sample of the supernatant was applied to a column (1.5 X 100 cm) of Ultrogel AcA 54 (LKB Laboratories) previously equilibrated with phosphate buffer. Fractions were then assayed for N-acetyl-8-D-glucosaminidase (absorbance nm) (26) and di-n-acetylchitobiase (absorbance nm) (see Experimental Procedures ). Arrows represent the elution volumes of standards: I, blue dextran (2 X lo3 kda); 2, bovine serum albumin (67 kda); 3, ovalbumin (43 kda); 4, carbonic anhydrase (29 kda); 5, cytochrome c (12.4 kda). biose and the potential simplicity of an assay system with it as substrate, we determined whether the same glycosidase that hydrolyzed the di-n-acetylchitobiose core region of the oligosaccharides would also hydrolyze this disaccharide. Since N-acetyl-p-D-glucosaminidase could also potentially hydrolyze this compound, it was necessary to develop a method to differentiate these two potential activities. Fig. 8 shows the elution profile of a crude liver extract on Ultro-Gel AcA-54. This gel filtration resolved two peaks of enzyme activity. In Fraction I, the activity assayed with di-n-acetylchitobiose coeluted with the N-acetyl-p-D-glucosaminidase activity assayed with p-nitrophenyl-n-acetyl-p-d-glucosaminide. In contrast, Fraction 11, which exhibited the majority of the activity assayed with di-n-acetylchitobiose, did not hydrolyze the p-nitrophenyl glycoside. The peak fractions from both Fractions I and I1 were then assayed with [3H]Gal-labeled oligosaccharide as a substrate. Only Fraction I1 was able to hydrolyze the reducing end GlcNAc from the biantennary complex chains. Based on its gel filtration properties, Fraction 11, which contains the di-n-acetylchitobiase activity, has an apparent molecular weight of 37,000. Several properties of the crude enzyme were studied using the PGalMNT-treated tritosome extracts. Chemical reduction of the N-acetylglucosamine to N-acetylglucosaminitol at the reducing end of either [3H]Gal biantennary chains or di-nacetylchitobiose made them resistant to cleavage by the oligosaccharide di-n-acetylchitobiase. Apparently, the N-acetylglucosamine residues must be in a pyranose form for efficient hydrolysis. Triantennary chains isolated from labeled al-acid glycoprotein were readily cleaved by the enzyme. The di-n-acetylchitobiase assayed in vitro with radiolabeled oligosaccharides was not inhibited by DONva at concentrations as high as 1.0 mm. DISCUSSION We have found that in situ inactivation of glycosyl asparaginase with DONva resulted in the incomplete catabolism of asialo-al-acid glycoprotein delivered to the lysosomal compartment by receptor-mediated endocytosis. Unexpectedly, this compound inhibited not only hydrolysis of the GlcNAc- Asn linking groups but also hydrolysis of the di-n-acetylchitobiose component of al-acid glycoprotein oligosaccharides. This was evidenced by liver accumulation of a glycopeptide having the structure: GlcNAcp(1-4)GlcNAc-Asn. In contrast, we previously had found in similar 3-h in situ experiments using BGalMNT to prevent degradation from the nonreducing ends of oligosaccharide chains that the di-n-acetylchitobiose unit of the core region was readily split (3). Intact oligosaccharides lacking only the reducing end GlcNAc residue accumulated in the lysosomes. Based solely on that previous work, we originally interpreted such structures to be products of an endo-n-acetyl-p-d-glucosaminidase activity. The present isolation of a degradation intermediate in glycosyl asparaginase-deficient tissue that contained an intact di-n-acetylchitobiose core has caused us to re-evaluate these earlier experiments. After a 1-h DONva treatment, the outer chain GlcNAc residues which are linked to a-d-mannose appeared to be effectively removed in 3 h as evidenced by their absence on the accumulating glycopeptide, GlcNAcp( 1-4)GlcNAc-Asn. This would suggest that GlcNAc is not hydrolyzed as efficiently from the nonreducing end of the di-nacetylchitobiose moiety by lysosomal N-acetyl-P-D-glucosaminidase which was still active in the treated liver (Table I). Therefore, a separate glycosidase in the lysosomes might normally be responsible for hydrolysis of the di-n-acetylchitobiose unit. Assuming that DONva did not directly inhibit such a glycosidase, it then seemed likely that the enzyme required prior removal of the linking Asn from its substrate in order to catalyze hydrolysis of the reducing end amino sugar. Based on this premise, we have used free oligosaccharides to detect a di-n-acetylchitobiose hydrolase that we believe is physiologically involved in the lysosomal catabolism of asparagine-linked glycoproteins in rat. Evidence already existed for such a di-n-acetylchitobiose cleaving activity in human tissues. Many structures have been isolated in human genetic glycosidase deficiencies which are marked at their reducing terminus by the presence of a p-dmannosyl-linked GlcNAc residue and at their nonreducing termini by the corresponding sugar which normally would

6 5808 Di-I?-acetylchitobiase Activity in Glycoprotein Degradation have been cleaved by the genetically defective hydrolase (2). For example, in Sandhoffs disease, a deficiency of both N- acetyl-p-d-ghcosaminidases A and B exists and fragments accumulate which contain nonreducing end GlcNAc in typical positions attached to a-d-mannose residues. However, the di- N-acetylchitobiose core region is still hydrolyzed, apparently by a hexosaminidase A and B independent mechanism, to yield a P-D-mannose-linked GlcNAc at the reducing terminus (19, 20). The discovery, isolation, and wide utilization of have found that treatment of the perfused rat liver with the bacterial proteinase inhibitor leupeptin results in the inability to remove fucose when the polypeptide of al-acid glycoprotein is inhibitor-protected.' We also assume that human and rat glycosyl asparaginases are similar to the glycosidase from hen bacterial enzymes such as endo-n-acetyl-p-d-glucosamini- oviduct, which requires its substrate to have an unsubstituted dases D and H were strong and logical experimental models aspartate moiety for hydrolysis (17). However, removal of for the idea that many such fragments isolated from human fucose would also appear to precede hydrolysis of the Asn genetic glycosidase deficiencies were produced by the action residue, since Tarentino et al. (33) noted hen oviduct glycosylof a similar endoglycosidase. Thus, using substrate glycopep- asparaginase cleavedasn from the IgMglycopeptide a-ltides labeled on their Asn with radioactive acetic anhydride, Fuc(1 + 6)-@-~-GlcNAc-Asn only after fucose had been elimseveral investigators have detected such an endoglycosidase activity in human and rat tissues (21-23). This enzyme, however, was unexpectedly found to have a neutral ph optimum and to be exclusively in the cytosolic fraction of rat liver (24, 34). Based on our results, such glycopeptides would not be expected to be substrates for the newly discovered lysosomal di-n-acetylchitobiose core cleaving activity. Thus, the acidic glycosidase actually involved in lysosomal glycoprotein catabolism would not have been detected using the previous assay techniques (neutral ph and Asn-linked oligosaccharides). The significance of the previously discovered endo-nacetyl-p-d-glucosaminidase is as yet unclear. However, in recent work, Anumula and Spiro (29) have reported detection of a neutral endo-8-n-acetylglucosaminidase in thyroid microsomes which suggests a possible role for such an enzyme in the regulation of biosynthetic N-glycosylation. In 1974, Stirling (32) found an enzyme in extracts of human liver which degrades di-n-acetylchitobiose and is very similar to the rat liver activity we have described here. The human enzyme had a ph optimum of 3.5, a molecular weight between 25,000 and 50,000, and did not cleave 4-methylumbelliferyl N-acetyl-P-D-glucosaminide. We therefore assume that a sim- ilar degradative system exists in rats and humans and propose that the overall catabolism of glycoproteins with complex asparagine-linked oligosaccharides is composed of two ordered degradation sequences which proceed in opposite directions (see Fig. 9). The first series includes a stepwise removal of sugars from the nonreducing ends of the chains. This series is initiated by removal of galactose residues (in the case of asialo-substrates) and terminates following hydrolysis of the mannose residue P(1-4) linked to the di-n-acetylchitobiose core. Interruption of these ordered hydrolytic events results in the inability to complete subsequent steps in the sequence as evidenced by the oligosaccharide fragments observed to accumulate in rat liver lysosomes following treatment with the appropriate glycosidase inhibitors (3) or in parallel human genetic glycosidase deficiencies (2). A second degradation sequence likely encompasses removal of the interfacing Asn and those sugars (di-n-acetylchitobiose moiety and fucose) in the immediate vicinity of the polypeptide. Extensive hydrolysis of the polypeptide appears to be a OLIGOSACCHARIDE (1-51 (I-IV) P ROTEIN-~~~~~EI FIG. 9. Proposed scheme for the lysosomal catabolism of complex asparagine-linked glycoproteins rats in and humans. prerequisite for two steps in this part of catabolism: removal of any fucose (if present) on the N-acetylglucosamine residue linked to the asparagine (step 11) and the separation of the asparagine and oligosaccharide (step 111). For example, we inated. Further support for this sequence is the fact that fucosylated oligosaccharides which still contain the linking Asn are the major degradation remnants found in human fucosidosis (30), a genetic disease in which only a-l-fucosidase is lacking. Based on the work presented in this report, hydrolysis of the di-n-acetylchitobiose moiety requires the prior loss of Asn and therefore would become the last event (step IV) in the series of reactions that occur at the protein linkage region in the order: 1) peptide hydrolysis, 2) fucose hydrolysis, 3) Asn removal, and 4) cleavage of the di-n-acetylchitobiose unit. One fact which does not appear to fit the scheme in Fig. 9 is the lack of accumulation of GlcNAc@(1-4)GlcNAc~-Asn in the Finnish patients that suffer from aspartylglucosaminuria (31). Instead, these individuals mainly exhibit steady state build-up of the smaller GlcNAc-Asn unit in their tissues. However, our results which showed that N-acetyl-P-D-glucosaminidase can slowly cleave the disaccharide di-n-acetylchitobiose (Fig. 8) can explain this occurrence in the diseased state. Thus, the latter exoglycosidase has an extended time period to catalyze removal of the nonreducing end GlcNAc from GlcNAc@( 1-4)GlcNAcp-Asn initially present as a degradative intermediate in the defective lysosomes. N-Acetyl-@- D-glucosaminidase therefore would not normally catalyze significant removal of GlcNAc from the nonreducing end of the di-n-acetylchitobiosyl moiety, but instead the importance of the newly discovered lysosomal di-n-acetylchitobiase in rat and humans is to more efficiently split the di-n-acetylglucosamine unit from the opposite direction and only after Asn has been removed as depicted in Fig. 9. Since the di-nacetylchitobiose cleaving activity of rat liver lysosomes can also hydrolyze di-n-acetylchitobiose, this reaction provides the basis for a convenient assay which can be exploited to eventually purify and further characterize this interesting new glycosidase. Acknowledgments-The help of Dr. Spencer Shames with the synthesis of 5-diazo-4-oxo-~-norvaline is gratefully appreciated. REFERENCES 1. Beaudet, A. L. (1983) The Metabolic Basis of Inherited Disease, pp. 7&3-802, McGraw-Hill, Publications, Minneapolis 2. Warner, T. G., and O'Brien, J. S. (1983) Annu. Reu. Genet. 17, Kuranda, M. J., and Aronson, N. N., Jr. (1985) J. Biol. Chem. 260, Aronson, N. N. (1982) Biochem. J. 203, Shibata, K., Okubo, H., Ishibashi, H., and Tsuda, K. (1977) Biochim. Biophys. Acta 495, Kuranda, M. J., and Aronson, N. N. (1983) Arch. Biochem. Biophys. 224, M. J. Kuranda and N. N. Aronson, Jr., unpublished results.

7 Di-N-acetylchitobiase Activity in Glycoprotein Degradation Takasaki, S., Mizuochi, T., and Kobata, A. (1982) Methods En- 21. Nishigaki, M., Muramatsu, T., and Kobata, A. (1974) Biochem. zymol. 83, Biophys. Res. Commun. 59, Liwschitz, Y., Irsay, R. D., and Vincze, A. I. (1959) J. Chem. SOC. 22. Overdijk, B., Van Der Kroef, W. M. J., Lisman, J. J. W., Pierce, R J., Montreuil, J., and Spik, G. (1981) FEBS Lett. 128, Handschumacher, R. E., Bates, C. J., Change, P. K., Andrews, A. 366 T., and Fischer, G. A. (1968) Science 161, Pierce, R. J., Spik, G., and Montreuil, J. (1979) Biochem. J. 180, 10. Leighton, F., Poole, B., Beaufay, H., Baudhin, P., Coffey, J. W., Fowler, S., and deduve, C. (1968) J. Cell Biol. 37, Pierce, R. J., Spik, G., and Montreuil, J. (1980) Biochem. J. 185, 11. Stauffer, C. E. (1975) Anal. Biochem. 69, Yamashita, K., Mizuochi, T., and Kobata, A. (1982) Methods 25. Mahadevan, S., and Tappel, A. L. (1967) J. Biol. Chem. 242, Enzymol. 83, McIlvaine, T. C. (1921) J. Biol. Chem. 49, Aronson, N. N., Jr., and de Duve, C. (1968) J. BWZ. Chem. 243, 14. Reissig, J. L., Strominger, J. L., and Leloir, L. F. (1955) J. Biol Chem. 217, Aronson, N. N., Jr., and Docherty, P. A. (1983) J. Biol. Chem. 15. Ricca, G. A., andtaylor, J. M. (1981) J. Biol. Chem. 256, , Arima, T., and Spiro, R. G. (1972) Chem. J. Biol. 247, Yoshima, H., Matsumoto, A., Mizuochi, T., Kawasaki, T., and 1848 Kobata, A. (1981) J. Biol. Chem. 256, Anumula, K. R., and Spiro, R. G. (1983) J. Biol. Chem. 258, 17. Tarentino, A. L., and Maley, F. (1969) Arch. Biochem. Biophys , Yamashita, K., Tachibana, Y., Takada, S., Matsuda, I., Arashima, 18. deduve, C., Pressman, B. C., Gianetto, R., Wattiaux, R., and S., and Kobata, A. (1979) J. Biol. Chem. 254, Applemans, F. (1955) Biochem. J. 60, Maury, C. P. J. (1982) J. Inherited Metub. Dis. 5, Ng Ying Kin, N. M. K., and Wolfe, L. S. (1974) Biochem. Biophys. 32. Stirling, J. L. (1974) FEBS Lett. 39, Res. Commun. 59, Tarentino, A. L., Plummer, T. H., Jr., and Maley, F. (1975) 20. Strecker, G., Herland-Peers, M.C., Fournet, B., Montreuil, J., Biochemistry 14, Dorland, L., Haverkamp, J., Vliegenthart, F. G., and Farriaux, 34. Lisman, J. J. W., Van der Wal, C. J., and Overdijk, B. (1985) J. P. (1977) Eur. J. Biochem. 811, Biochem. J. 229,