Expression of N-Acetylglucosaminyltransferase I11 in Hepatic Nodules during Rat Liver Carcinogenesis Promoted by Orotic Acid*

The activity of N-acetylglucosaminyltransferase 111, which adds a “bisecting” GlcNAc in 81,4 linkage to the &linked Man of the core of Asn-linked oligosaccha- rides (Narasimhan, S. (1982) J. Biol. Chem. 257, 10235-10242), was determined in hepatic nodules of rats initiated by administration of a single dose of carcinogen 1,2-dimethylhydrazine. 2HC1 (100 mg/kg, intraperitoneal) 18 h after partial hepatectomy and promoted by feeding a diet supplemented with 1% or- otic acid for 32-40 weeks. N-Acetylglucosaminyl-transferase I11 was assayed using glycopeptide GlcNAc~1,2Mancu1,6(GlcNAcfi1,2MancY1,3)Manfil, 4GlcNAc~1,4GlcNAc-Asn as substrate and, as enzyme sources, microsomal membranes of the hepatic nodules, surrounding liver, regenerating liver, and age- and sex-matched control liver. The nodules had significant N-acetylglucosaminyltransferase I11 activity (0.78- 2.18 nmol GlcNAc transferred/h/mg of protein), while

The presence of bisecting GlcNAc residues in the carbohydrate moieties of y-glutamyltranspeptidase from the hepatomas of rats and humans and their absence from y-glutamyltranspeptidase from normal liver in both species (2, [28][29][30] suggests the induction of N-acetylglucosaminyltransferase I11 during carcinogenesis in liver. We have investigated this hypothesis in rats using a previously described experimental model for liver carcinogenesis (31)(32)(33). In this model, hepatic nodules are induced in male Fisher rats by administration of a single dose of the liver carcinogen 1,2-dimethylhydrazine -2HC1 intraperitoneally (100 mg/kg) 18 h after partial hepatectomy followed after a week by feeding a diet with 1% orotic acid (OA)' as a promoter. At 52 weeks on OA diet, there is a 100% incidence of hepatocellular carcinoma in the initiated rats. The model is thus suited to study the induction of N-acetylglucosaminyltransferase I11 at different stages of liver carcinogenesis, namely, initiated, promoted, progressive, and metastatic stages.
We show the presence of N-acetylglucosaminyltransferase I11 activity in the microsomal membranes of hepatic nodules The abbreviations used are: OA, orotic acid Bisecting GlcNAc-(GN), an N-acetylglucosaminyl residue linked in @1-4 linkage to the @-linked Man of the trimannosyl core of asparagine-linked oligosaccharides; ConA, Concanavalin A; Asn-linked or N-linked, Asparagine-linked DSS, Sodium 4,4-dimethyl-4-silapentane-l-sulfonate. of initiated rats on 1% OA diet for 32-40 weeks. Microsomal preparations from surrounding liver, regenerating liver (24 h after partial hepatectomy), and age-and sex-matched control liver had negligible activity. Induction of N-acetylglucosaminyltransferase I11 is thus evident in the preneoplastic stage of carcinogenesis. Preliminary reports have appeared (34-36).

RESULTS
Two series of experiments on N-acetylglucosaminyltransferase I11 in nodules were carried out. Series 1 assays (Table  I) lacked ATP in the incubation while Series 2 assays (Table  111) included 6.25 mM ATP in the incubations.

N-Acetylglucosaminyltransferase IZZ in Hepatic Nodules,
Surrounding Liver, and Control Liver (Series 1) Bio-Gel P2 Chromatography-N-Acetylglucosaminyltransferase I11 assays were performed both at pH 5.7 and 6.5 using nodules, surrounding liver, and control liver ( Table I). The yield of total products was assessed both by the AG-1 X 8 and P2 assays. The assessment of total 14C products by the AG-1 X 8 assay is subject to error because of the high "endogenous" value (no acceptor glycopeptide) due to free ['4C]GlcNAc formed by the breakdown of UDP-N-[1-'4C]acetyl-~-glucosamine (labeled donor) (Fig. 2, a and b, Table 11). Therefore, the total 14C products formed were also estimated by the P2 assay. In this method, free [14C]GlcNAc is well separated from the total 14C products (Fig. 2b).
ConA-Sepharose Chromatography-The void fractions from the Bio-Gel P2 column (above) were pooled and applied to a ConA-Sepharose column. Fig. 3a shows the elution profile of N-14C-acetylated substrate glycopeptide GnGn. Most of the radioactivity (97%) recovered from the ConA-Sepharose column bound to the column and eluted in a retarded manner characteristic of GnGn (1,42,43) in buffer containing 0.2 M methyl a-D-glucoside. In contrast, the elution profile of the purified total I4C products formed by hepatic nodules (Experiment 1, pH 6.5, Table I) showed that 84% of the radioactivity recovered from the column eluted in a retarded manner in buffer lacking methyl a- ). The products from hen oviduct membrane (l), and from B-lymphocyte lines: which eluted in a retarded manner identical to the product from hepatic nodules, have previously been characterized by 360 MHz high resolution proton NMR spectroscopy as the product of N-acetylglucosaminyltransferase I11 activity, i.e. GnGn(Gn) (see Introduction). We conclude that the major product formed by hepatic nodules (Fig. 3b) is due to Nacetylglucosaminyltransferase I11 activity (82-93% of total products). This has been confirmed by preparing a large scale incubation product and identifying it by 500 MHz proton NMR spectroscopy (see below).
The surrounding liver had negligible activity. Very small amounts of 14C products were obtained by pooling products from several identical incubations. The 14C products resolved on ConA-Sepharose chromatography (Fig. 3c) into three fractions. Only 12% of the radioactivity represented product due to N-acetylglucosaminyltransferase I11 activity. Control liver did not exhibit any detectable enzyme activity since radioactivity in the presence of GnGn was less than that in the absence of substrate (Table I). Application of the AG-1 x 8 eluates from tubes containing GnGn to a Bio-Gel P2 column did not yield any radioactivity in the void fraction of the column (data not shown).
N-Acetylglucosaminyltransferase IIZ Activity-N-Acetylglucosaminyltransferase 111 activities were calculated from the above data for microsomal membranes from hepatic nodules (32-35 weeks on OA diet), surrounding liver, and ageand sex-matched control liver (Table I). There was reasonable agreement between the AG-1 X 8, P2, and ConA-Sepharose assays ( Table I). The hepatic nodules had significant activity (0.78-1.28 nmol/h/mg of protein), the activity at pH 6.5 being higher than that at 5.7; surrounding liver had negligible activity, and control liver showed no activity under the assay conditions used.

Effect of ATP and AMP Addition to Incubation Mixtures to
Prevent Degradation of Labeled Donor Incubations without exogeneous GnGn had considerable radioactivity in the AG-1 X 8 eluates due to breakdown of the labeled donor GlcNAc. The degradation at pH 5.7 was less than that at pH 6.5. This high endogenous activity in tubes without substrate may lead to errors in the assessment of total 14C products formed when using the AG-1 x 8 assay. Several compounds were tested as potential inhibitors of UDP-GlcNAc breakdown, i.e. ATP, AMP, N&P2O7, EDTA, 5'-HgUDP, mercaptoethanol, and 2,3-dimercapto-propanol. Only ATP and AMP were effective in reducing the degradation. Na4P207 and EDTA inhibited N-acetylglucosaminyltransferase 111 while mercaptoethanol, 2,3-dimercapto-propanol, and 5'-HgUDP inhibited neither UDP-GlcNAc breakdown nor N-acetylglucosaminyltransferase I11 at the concentrations tested (data not shown). The results with ATP and AMP are shown in Table 11. ATP (5.55 and 11.1 mM) added to the incubation mixtures greatly inhibited UDP-GlcNAc breakdown with minimal inhibition of N-acetylglucosaminyltransferase I11 activity. Addition of 2.78 mM AMP along with 11.1 mM ATP was found to be better than ATP alone at that concentration. In the second series of experiments (see below), AMP and ATP were added at concentrations of 3.13 and 6.25 mM, respectively.

Effect of Protein Concentration and Time of Incubation on
Formation of Total 14C Products by Hepatic Nodule Microsomal Preparations The effect of increasing protein concentration (40-380 pg) and of time (0.5-4 h) on the formation of total "C products is shown in Fig. 4, a and b, respectively. Total 14C products formed are proportional to protein concentration up to 380 pg and to time up to 4 h.

Effect of pH on N-Acetylglucosaminyltransferase ZII Activity
in Hepatic Nodule Microsomal Membranes The pH activity curve (Fig. 4c) shows a broad optimum (pH 6-7.0) for N-acetylglucosaminyltransferase 111 activity. The membranes were from rats killed after 40 weeks on OA diet, and ATP (6.25 mM) and AMP (3.13 mM) were present in the incubation medium to inhibit degradation of the labeled donor. As in the absence of ATP (Table I), the activity at pH 6.5 (2.36 nmol/h/mg of protein) was higher than that at pH 5.7 (1.71 nmol/h/mg of protein). In all further studies, Nacetylglucosaminyltransferase I11 activity was assayed at pH 6.5.

Liver'
Total products Fractionation of total products on ConA-Sepharoaed In experiments 1 and 2, rats on OA diet were killed at 32 and 35 weeks, respectively.
' Total products formed were assessed by the difference in radioactivity in the AG-1 X 8 eluates of incubation E Total products formed were assessed by the radioactivity found in the void volume of a Bio-Gel P2 column on tubes with and without exogenous substrate GnGn.
chromatography of AG-1 X 8 eluates from incubations with exogenous substrate GnGn. Refer to Fig. 1; total products were measured by the P2 assay. e Specific activity of UDP-iV-[l-~4C]acetyl-D-glucosamine was 3945 DPM/nmol in this experiment. ' -= No activity. No difference between tubes with and without exogenous substrate GnGn. in total volume to 45 pl, and the absence of AMP in tubes 1-3.
Incubations were at pH 6.5 under conditions as described under "Materials and Methods" except for the change The microsomal membranes were the same preparation used in Experiment 1 of Table I. N-Acetylglucosaminyltransferase I11 assayed by ConA-Sepharose chromatography of total products assayed by e Total products assessed by AG-1 X 8 assays.

N-Acetylglucosaminyltransferase 111 Activity in Hepatic
Nodules, Surrounding Liver, Control Liver, and 24-H

Regenerating Liver (Series 2)
Incubations were at pH 6.5 and included 6.25 mM ATP and 3.13 mM AMP. The Bio-Gel P2 chromatography profiles (compare Figs. 2 and 5) show that the breakdown of donor substrate is very much reduced under these conditions.
ConA-Sepharose elution profiles of total 14C products from the four types of microsomal membranes are shown in Fig. 6, a-d. Only the profile from the nodule showed a major retarded peak due to N-acetylglucosaminyltransferase I11 activity. Surrounding liver, control liver, and 24-h regenerating liver showed negligible radioactivity retarded on the ConA column. The activities in these other three membranes were about 100-fold less than that of the nodules (Table 111). The nodules had activities of 1.75 and 2.19 nmol of GlcNAc transferred/h/mg of protein while the other membranes had negligible activity. The activities of N-acetylglucosaminyltransferase (I + 11) observed by fractionation of total 14C products on ConA-Sepharose (Fig. 6, Table 111) are much lower than the Vmax values since they are due to side reactions ( Fig. l), and substrate concentrations were below saturation values. N-Acetylglucosaminyltransferase (IV + V) activities may also be elevated in nodules (Tables I and 111), but the difference between nodules and controls is neither as marked nor as consistent as for N-acetylglucosaminyltransferase 111; further work is required to establish the significance of this observation.

N-Acetylglucosaminyltransferases I and 11 in the Four
Microsomal Membranes N-Acetylglucosaminyltransferase I and I1 were assayed under optimal conditions (see "Materials and Methods") in the microsomal membranes from the nodules, surrounding liver, control liver, and 24-h regenerating liver. The nodules were active for both N-acetylglucosaminyltransferase I and I1 activites (Table 111). However, both showed reduced activities when compared to the other three membranes. The reason for the decreased activities of N-acetylglucosaminyltransferase I and I1 in nodules is unclear at present.  cm) of N-acetylated total 14C products from the large scale incubation of hepatic nodules (analytical run). dpm applied = 3430. Recovery was 100%. b, Bio-Gel P6 chromatography of the major peak obtained on ConA-Sepharose chromatography of N-acetylated total "C products from the large scale incubation of hepatic nodules. Product 1, 54%; product 2,46%. 7b). Product 1 eluted near the void volume of the column and product 2 was retarded. The proportions of products 1 and 2 were 54 and 46%, respectively. Fig. 8 shows NMR spectra at 300 K representing the anomeric hydrogens of Man and "Acetyl Anomeric Protons Man H-2 Atoms "Triplet" -CH3 Protons

2.0794
e Not visible due to shift of the anomeric proton of Man al-3. 'There is no signal for bisecting GlcNAc in the substrate (GnGn) and the 4 signals observed are numbered 1-4, as shown in Fig. 8a. In the two products, there are 5 signals observed, numbered 1-5, as shown in Fig. 8, b and c. -. indicates no chemical shift observed.
GlcNAc, H-2 of Man, and the N-acetyl protons of GlcNAc, for substrate and for products 1 and 2. The chemical shifts are shown in Table IV. Spectra were also taken at 343 K in order to see the chemical shifts of the anomeric protons of plinked Man (not shown).
The chemical shifts for substrate glycopeptide GnGn (Fig.  Sa) are the same as those reported previously for a standard glycopeptide (1, 41, 47). The chemical shift of the anomeric proton of &linked Man, which was not observed at 300 K, was clearly seen in the 343 K spectrum at 4.7617 ppm (spectrum not shown). The NMR spectrum of product 1 (Fig. 8b) shows two new chemical shifts (a doublet at 4.4668 ppm and an additional -CH3 proton chemical shift a t 2.0660 ppm), as well as changes in the anomeric and H-2 protons of the three mannoses, and the appearance of a multiplet at 3.27 ppm; these are all typical of bisected oligosaccharides (1, 47, 48) i e . , product 1 can be identified as bisected glycopeptide GnGn(Gn). The spectrum of product 1 at 343 K showed the presence of the anomeric proton of P-linked Man at 4.713 ppm. The chemical shifts typical of a bisecting GlcNAc are also evident in product 2 (Fig. 8c). In addition there is the appearance of a resonance at 5.1875 ppm due to the a anomer of reducing GlcNAc. The N-acetyl region for product 2 shows resonances due to 5 GlcNAc residues but the downfield shift of resonance 1 to 2.0374 ppm clearly indicates the absence of an Asn-residue in the product. Product 2 can therefore be identified as bisected oligosaccharide GnGn(Gn). A similar observation was made earlier (41) when hen oviduct membranes were incubated with nonfucosylated transferrin glycopeptide GnGn; the product of N-acetylglucosaminyltransferase IV, after purification on ConA-Sepharose, was fractionated on Bio-Gel P10 chromatography into a glycopeptide product (50%) in the void volume and an oligosaccharide product (50%) which entered the column. The NMR spectra of both products were similar except for the absence of asparagine in the latter. This splitting of the product was attributed to the presence in hen oviduct of a glycosyl-asparaginase which cleaved the oligosaccharide from the asparagine moiety of the glycopeptide (44) thereby giving rise to the oligosaccharide product. Glycosyl-asparaginase has been isolated from rat liver (49) and kidney (cited in 44) besides hen oviduct. A similar mechanism is probably responsible for cleaving the Asn residue from the N-acetylglucosaminyltransferase I11 product of nodules.
The resonances a t 5.04, 4.9, and 4.5 ppm in the spectrum of product 1 were not identified and may be due to contaminants. The chemical shift around 4.08 ppm of both the products has not been assigned but has been observed with bisected structures (50).

DISCUSSION
We have demonstrated that hepatic nodules of rats initiated with the chemical carcinogen 1,2-dimethylhydrazine and promoted with a 1% orotic acid diet express N-acetylglucosaminyltransferase I11 activity. The product of N-acetylglucosamiayltransferase I11 activity has been characterized by 500 MHz proton NMR spectroscopy as bisected GnGn i e . , GnGn(Gn) (see Introduction for structure). Under the conditions of the enzyme assay, surrounding liver, control liver, and 24-h regenerating liver had negligible activity. Since the tissue around the nodules has undergone the same treatments (partial hepatectomy, exposure to chemical carcinogen, and prolonged influence of orotic acid) as the nodules themselves, the low N-acetylglucosaminyltransferase I11 activity in this tissue indicates that the expression of N-acetylglucosaminyltransferase I11 is associated with the precancerous stage of hepatocarcinogenesis. Our observations on N-acetylglucosa-minyltransferase I11 activity in the nodules are consistent with the studies of  who showed the presence of bisected oligosaccharides in the y-glutamyltranspeptidase of rat hepatoma and of serum from human hepatoma patients but not in the y-glutamyltranspeptidase of normal liver of both rat and humans. Our earlier studies (31) showed that feeding of orotic acid for 5 weeks to rats enhances the incidence of y-glutamyltranspeptidase positive foci in rat liver induced by chemical carcinogens; others (51, 52) have shown increased activity of y-glutamyltranspeptidase in liver of rat and mouse at the preneoplastic stage of chemical carcinogenesis. These findings suggest that y-glutamyltranspeptidase isolated from hepatic nodules may also show the presence of bisected N-linked oligosaccharides.
The carbohydrate structures of normal liver glycoproteins (2) do not have bisecting GlcNAc residues in their Asn-linked oligosaccharides. But, conflicting reports on the presence or absence of bisecting GlcNAc in liver cancer have appeared. a-Fetoprotein purified from the ascites fluid of hepatoma patients is devoid of bisecting GlcNAc (53), but the same protein in the cystic fluid of human yolk sac tumor (29) has bisecting GlcNAc in its N-linked oligosaccharides. Earlier studies by  failed to detect bisecting GlcNAc in serum y-glutamyltranspeptidase from colon cancer patients with liver metastases, but a recent study by Yamashita et al. (54) showed that 50% of the oligosaccharide chains of carcinoembryonic antigen purified from liver metastases of primary colon cancer patients had bisecting GlcNAc. This latter observation is contrary to that of Chandrasekaran et al. (55) who did not find bisecting GlcNAc residues in carcinoembryonic antigen from liver metastases of colon cancer patients. Chandrasekaran et al. (56) also did not find bisecting GlcNAc in the oligosaccharide chains of cy1 acid glycoprotein isolated from liver metastases of colon cancer patients. The lack of bisecting GlcNAc in y-glutamyltranspeptidase in human primary liver cancer is noteworthy (57). Because the foregoing results were obtained from human patients, it may reflect variation from patient to patient or it may reflect the variation in the cell type from which the malignant growth arose. It may also be possible that not all glycoproteins in liver cancer can be acted on by N-acetylglucosaminyltransferase I11 since studies by Kobata and coworkers (58) showed that only 40% of the oligosaccharide chains of y-glutamyltranspeptidase in hepatoma and 50% of the oligosaccharide chains of carcinoembryonic antigen in liver metastases from colon cancer (54) had bisecting GlcNAc. The significance of these observations remains to be elucidated.
Bisecting GlcNAc residues appear to be restricted to glycoproteins from tissues such as hematopoietic cells, kidney, oviduct, malignant tissues (hepatoma and yolk sac tumor), and abnormal skin fibroblasts. Our recent studies (59,6013 showed that N-acetylglucosaminyltransferase I11 is expressed in human B-lymphocyte lines but not in T-lymphocyte lines. N-Acetylglucosaminyltransferase I11 is induced in a dominant ricin-resistant mutant Chinese hamster ovary cell line named LEC 10; the parental line does not express any N-acetylglucosaminyltransferase I11 activity (61). The induction of Nacetylglucosaminyltransferase I11 observed in LEC 10 Chinese hamster ovary cells and in hepatic nodules (this study) may be related to the resistance to toxic agents such as CCl,, dimethylnitrosamine, aflatoxin B1,2-acetylaminofluorine, and diethylnitrosamine observed in preneoplastic hepatocyte nodules generated by different models of carcinogenesis in rat (52). Just as the drug resistance phenotype is common to all hepatic nodules (irrespective of how they are generated), it is possible that induction of N-acetylglucosaminyltransferase I11 may also be observed in all hepatic nodules. The possible relationship between expression of N-acetylglucosaminyltransferase I11 and resistance to toxic agents will be the subject of future investigations. The mechanism by which OA promotes carcinogenesis has yet to be understood. OA is a normal cellular constituent and is an intermediate in the biosynthesis of pyrimidine nucleotides. Pyrimidine nucleotides are involved both in DNA and sugar nucleotide biosynthesis. OA feeding causes imbalance in nucleotide pools (62-64) e.g., an increase in uridine nucleotides and a decrease in adenine nucleotides. Damage to liver DNA (65) and altered glycosylation (30% decrease in mannosyltransferase) in liver microsomes (66) of OA fed rats have been reported. Our future studies are aimed at three aspects: (i) whether or not N-acetylglucosaminyltransferase I11 is expressed at the initiated and the progressive stages of carcinogenesis, (ii) whether induction of N-acetylglucosaminyltransferase I11 is common to other models of liver carcinogenesis, and (iii) how the imbalance of nucleotide pools leads to altered expression of N-acetylglucosaminyltransferase 111.