Peptide:N-Glycosidase Activity Found in the Early Embryos of Oryxias W i p e s (Medaka Fish) THE FIRST DEMONSTRATION OF THE OCCURRENCE OF PEPTIDE:N-GLYCOSIDASE IN ANIMAL CELLS AND ITS IMPLICATION FOR THE PRESENCE OF A DE-N-GLYCOSYLATION SYSTEM IN LIVING ORGANISMS*

The recent discovery of free oligosaccharides typical for the complex type of glycan chains terminating with a free di-N-acetylchitobiosyl structure in certain fish eggs and early embryos us to find an enzyme responsible for detachment of N-linked glycan chains from glycoproteins by hydrolyzing the &aspartyl-glucosylamine linkage in Oryzias latipes embryos. The enzyme,

8 To whom correspondence should be addressed. antennary chains. Relatively large amounts of free oligosaccharides accumulated in mature fish eggs, suggesting that they may possibly be derived from vitellogenins, abundant glycolipophosphoproteins incorporated and processed by the oocytes during oogenesis. As the second example, we have shown the presence of free penta-antennary glycan chains in the embryos (4-11 h postinsemination) of a flounder (Paralichthys oliuaceus) (3). In this case, the free glycan was derived from the unique penta-antennary glycan structure of the cortical alveolus glycoprotein (hyosophorin) of this fish. Because the hyosophorin-derived free oligosaccharide is not found in unfertilized eggs or inseminated eggs that fail to continue embryonic development, de-N-glycosylation appears to have some biological significance during embryogenesis of the fish. The most important common structural feature of the free oligosaccharides found in fish eggs is the presence of a di-N-acetylchitobiosyl structure at their reducing termini. This fact indicates that 1) these oligosaccharides are detached from glycoproteins and 2) the enzyme catalyzing the scission must be peptide-N4-(N-acetyl-P-glucosaminyl)asparagine amidase or peptide:N-glycosidase (PNGase). ' In this study, we demonstrate the presence of PNGase activity in the embryos of Medaka fish (Oryzias latipes) by partial purification of the enzyme and the structural analysis of the reaction products. PNGases have so far been obtained from almond emulsin (4), Flavobacterium meningosepticurn (5), and eight plant seeds (6), but never from animal sources. Our present findings are thus important as the first indication of the presence of PNGase in animal cells and may suggest the importance of a protein N-glycosylation and de-N-glycosylation system in the metabolism and possibly the function of glycoproteins.
Chemical Analysis-Methods of carbohydrate analysis were described previously (9). Amino acid analysis was carried out after hydrolysis in 6 N HC1 at 110 "C for 24 h under N, and precolumn derivatization with phenylisothiocyanate (10). Automated peptide sequence analysis was performed on an Applied Biosystem model 477A protein sequencer (3).
Assay of PNGase Actiuity-The enzyme solution (6 pl) containing 10-20 mM sodium acetate (pH 6.0), 0.25 M sucrose, 0.6 pg of soybean trypsin inhibitor (Sigma) and ~-['~C]hyosophorin (30,000 cpm) was incubated in a polypropylene microtube for 16 h a t 25 "C. The reaction mixture was clarified by centrifugation and applied as a spot on Whatman 3-MM paper. The chromatogram was developed ascendingly in 1-butanol/ethanol/water (4:2:3 v/v) at room temperature (11) until the solvent front migrated up to 7.5 cm. The dried chromatogram was cut in horizontal sections for the measurement of radioactivity. Under these conditions, "C-peptide released from ~-['~C]hyosophorin incubated with PNGase F (Takara Shuzo, Co., Kyoto) migrated 1.2-1.8 cm, whereas intact ~-['~C]hyosophorin remained at the origin. PNGase activity in fish egg preparations was determined from the radioactivity found in the area of paper chromatogram between 1.2 and 1.8 cm from the origin. One unit was defined as the amount of the enzyme releasing 1 pmol of I4C-peptide from ~-['~C]hyosophorin upon incubation for 16 h at 25 "C under the above-described conditions.
Partial Purification of PNGase from Medaka Embryos-All purification procedures were done at 4 'C. 233 g of Medaka embryos were homogenized in 370 ml of sodium acetate buffer (pH 6) containing 0.25 M sucrose, 3 mM EDTA, and 0.1 mg/ml soybean trypsin inhibitor with a Waring blender and filtered through Tetoron gauze. The filtrate was centrifuged first at 6,000 X g for 30 min and then at 100,000 X g for 1 h, and the supernatant was dialyzed against 10 mM sodium acetate buffer (pH 6), 0.25 M sucrose. The dialyzed solution was applied on a CM-Sephadex C-25 column (1.75 X 31 cm) equilibrated with 10 mM sodium acetate buffer (pH 6), 0.25 M sucrose. To the breakthrough fraction (836 ml) was slowly added 1020 ml of saturated ammonium sulfate solution (pH 7.1), and the mixture was left overnight. The precipitate collected by centrifugation at 20,000 X g for 30 min was dissolved in 10 ml of 10 mM Tris-HC1 ( Another three 0.4ml portions of the enzyme were added at 12-h intervals. The reaction mixture was applied to a Sephadex G-50 column (1.2 X 72 cm, eluted with 0.1 M NaC1). The released radioactive peptide eluted in peak 2 ( Fig. 3) was further purified on a Bio-Gel P-2 column (1.2 X 53 cm, eluted with 10 mM acetic acid), desalted on a Sephadex G-25 column, and subjected to composition and sequence analyses.
For the analysis of the released oligosaccharide from fetuin glycopeptide 11, the material eluted in peak l from the Sephadex G-50 column ( Fig. 3) was desalted on a Sephadex G-25 column and digested with 23 milliunits of Arthrobacter ureafaciens sialidase (Nacalai Tesque, Kyoto) under toluene in 0.5 ml of 50 mM sodium acetate buffer (pH 5.5) for 21 h at 37 "C. The reaction mixture was applied to a Sephadex G-50 column, and the fraction containing the free asialooligosaccharide and the asialoglycopeptide was desalted by passage through a Sephadex G-25 column. Separation of the asialooligosaccharide and the asialoglycopeptide was effected by anion-exchange chromatography on a Bio-Rad AG 1-X2 column (0.5 X 6 cm, C1form). The asialooligosaccharide was eluted as a breakthrough peak. One third of the asialooligosaccharide was reduced with 2 mg of NaBH, in 0.5 ml of 50 mM sodium borate buffer (pH 9.4) for 17 h at 25 "C before composition analysis.
For amino acid sequence analysis, 70 nmol of unmodified fetuin glycopeptide was digested with the PNGase fraction, and the released peptide was purified by a method similar to that described above for radiolabeled fetuin glycopeptide.

RESULTS AND DISCUSSION
Detection of PNGase Activity in the Fertilized Medaka Eggs and Partial Purification of the Enzyme-We have found the presence of two types of free sialooligosaccharide chains in the Medaka embryos (12). The results of structural analysis of the free oligosaccharides indicated that one type is derived from hyosophorin' and the other is from phosvitin (13). The fact that these oligosaccharides retain the di-N-acetylchitobiosyl structure at their reducing ends was considered to be an indication of the presence of PNGase in Medaka embryos.
In the present work, we could detect in the extract of Medaka embryos the enzyme activity that catalyzes the release of N-glycan chains from glycopeptides. The enzyme in the soluble fraction of the extract was partially purified by ammonium sulfate precipitation and a series of anion-exchange column chromatography (Figs. 1 and 2), and the purification scheme is summarized in Table I. Ammonium sulfate precipitation was effective in separating the enzyme protein from endogenous glycopeptides, and repeated anionexchange chromatography was necessary to remove yolk proteins that are abundant in fish embryos. An overall 2090-fold purification (specific activity, 1.57 units/mg of protein) and about 6% yield of PNGase was obtained from the crude extract of Medaka embryos. As judged from the results shown in Fig. 2, the PNGase fraction is still contaminated with proteinaceous material, and our preparation of the enzyme was not yet biochemically homogeneous. Nevertheless, the present study resulted in a 2090-fold purification with a 5.8% yield of activity, and the present PNGase is indeed the first A. Seko, unpublished observation. animal-derived enzyme that has been identified and purified to this degree of purity.
Digestion of Fetuin Glycopeptide ZZ with Medaka Embryo PNGase and the Analysis of the Products-Fetuin glycopeptide II was digested with partially purified Medaka PNGase, and the products were separated on a Sephadex G-50 column (Fig. 3a). High percentages of radioactivity (88%) were eluted in more retarded fractions (peak 2) than the intact glycopeptide (peak 1). The released peptide was recovered from peak 2, purified by Bio-Gel P-2 chromatography, and desalted on Sephadex G-25. Peak 1 contained the unreacted glycopeptide and the released sialooligosaccharide.
To separate the oligosaccharide from the intact glycopeptide, this fraction was subjected to Bio-Rad AG l-X2 anion-exchange chromatography after sialidase treatment.
Material positive in the phenol/sulfuric acid assay (15) and having no radioactivity was eluted in the breakthrough fractions, whereas the radioactive glycopeptide was retained on the column. Table II  summarized  Medaka embryo PNGase was identical with that of the fetuin glycopeptide II, except for the disappearance in the chromatogram of NH*-terminal leucine that had been N,N-dimethylated. No glucosamine was detected in the hydrolysate of the peptide fraction. The ratios of Gal and GlcNAc to 3 Man residues in the oligosaccharide fraction were identical with those in the parent glycopeptide.
The reduction of the oligosaccharide resulted in conversion of 1 mol of GlcNAc to GlcNAcol.
No amino acid was detected in the hydrolysate of the oligosaccharide fraction.
These results clearly indicate that oligosaccharide with di-N-acetylchitobiose at the reducing end was released by the enzyme. The amino acid sequence of fetuin glycopeptide II was determined to be Leu-Ala-Asn(CHO)-CmCys-Ser, whereas that of the released peptide was Leu-Ala-Asp-CmCys-Ser (Table III). Note that the third glycosylated Asn was converted into an Asp residue after de-N-glycosylation with the Medaka enzyme. All of these results support identification of the enzyme activity derived from Medaka embryos as PNGase. was separated from the parent glycopeptide on Sephadex G-50 (Fig. 3b, peak 2), purified on Bio-Gel P-2, and analyzed for amino acid and amino sugars after hydrolysis. As shown in Table II, amino acid composition of the released peptide was identical with that of the parent hyosophorin, except for the modified NHz-terminal aspartic acid residue.
No glucosamine was detected in the hydrolysate.
The results show that the N-glycan chain of L-hyosophorin is indeed released by the action of the PNGase fraction from Medaka embryos, and thus the accumulation of free N-glycan chains in Medaka and most probably in certain other fish embryos can be ascribed to the presence of this type of enzyme activity.
General Discussion-To our knowledge, no report for PNGase from an animal source has so far been found, although PNGase activity has been detected in the extracts of    a variety of plant seeds and bacteria (4)(5)(6). The occurrence of endo-j3-N-acetylglucosaminidase in animal has been reported (16). However, the products formed by the action of endo-@-N-acetylglucosaminidase do not retain a di-N-acetylchitobiosyl structure. Our previous findings of the accumulation at the late stage of oogenesis (1,2) and the blastulation stage of embryogenesis (3)* of certain fish species of free N-glycan chains that retain a di-N-acetylchitobiosyl structure at their reducing termini have strongly supported the supposition that PNGase activity is expressed also in animal cells, and its expression may be significant for the metabolism and function of certain glycoproteins. In other animal cells, however, the accumulation of glycoprotein-derived free oligosaccharides would not be observed since they are transported to lysosomes immediately after liberation from the parent glycoproteins and undergo further degradation catalyzed by glycohydrolases to their constituent monosaccharides (16 20). The removal of glycan chains was shown to affect the transport and maturation of the proproteins (17,19). In this connection, in the amino acid sequence of ricin, Ricinus communis seed lectin, as determined by protein sequencing, the 236th amino acid residue of the A unit was indicated as Asp (Zl), whereas this residue was deduced from cDNA base sequencing as Asn (22). It has been reported that a "heavy" or "variant" form of ricin A chain contains two N-glycan chains at Asn-10 and Asn-236 (23). The structure of the Nglycan chain at Asn-236 has recently been reported (24). If we combine these results, a "light" form of ricin A chain is most likely formed from the heavy form by site-specific de-N-glycosylation catalyzed by PNGase, which converts the asparagine glycosylamine bond at position 236 to aspartic acid. Our limited knowledge of de-N-glycosylation of N-linked glycoproteins catalyzed by PNGase prevents any further considerations about the functional role of de-N-glycosylation and of the free glycans liberated. Nevertheless, de-N-glycosylation of a glycopeptide or glycoprotein, which converts the carbohydrate-attached Asn residue to the Asp, thereby introducing negative charge and altering the peptide or protein conformation, may be a possible means to produce a functional conformation. To assess the precise functional role of the detachment of a large N-glycan chain from L-hyosophorin at the early stages of embryogenesis of Medaka, the use of monoclonal anti-PNGase antibodies appears to be promising.

Amino acid and carbohydrate compositions of fetuin glycopeptide ZZ, L-hyosophorin, and products obtained therefrom after digestion with the Medaka embryo PNGase
We have not systematically examined how the developmental expression of the presently found Medaka embryo's PNGase changes during embryogenesis, differentiation, and ontogenesis. However, our preliminary results using 10 embryos in each experiment showed that during Medaka embryogenesis (from the 8-32-stage to gastrula via morula and blastula stages), the PNGase activity appeared to rise rather progressively to a maximum at the late blastula stage, followed a decay (we have not examined embryos later than gastrulation stage for PNGase activity). We detected PNGase activity even in unfertilized eggs or ovary, but no information is available if the PNGase activity present in the unfertilized eggs is originated from the PNGase identified and partially purified here. The cellular localization and developmental profile of the PNGase identified in this study would have to be substantiated once an antibody to this enzyme becomes available. Attempts to further purify the Medaka embryo PNGase are currently underway.