Alkaline phosphatase retained in HepG2 hepatocarcinoma cells vs. alkaline phosphatase released to culture medium: Difference of aberrant glycosylation

BBRC Biochemical and Biophysical Research Communications 330 (2005) 400–409 www.elsevier.com/locate/ybbrc

Alkaline phosphatase retained in HepG2 hepatocarcinoma cells vs. alkaline phosphatase released to culture medium: Difference of aberrant glycosylation Azin Nowrouzi, Razieh Yazdanparast * Institute of Biochemistry and Biophysics, University of Tehran, P.O. Box 13145-1384, Tehran, Iran Received 21 February 2005 Available online 16 March 2005

Abstract Liver tissue is the source of 90% of serum alkaline phosphatase (AP). The serum levels and structures of tumor marker proteins change under many disease conditions as well as cancer. The study was aimed at determining the type of alkaline phosphatase (AP) present in HepG2 hepatocellular carcinoma cell line. Alkaline phosphatase rich extracts of healthy human liver, HepG2 hepatocarcinoma cells, as well as the condition medium of HepG2 cells were prepared by extraction with 40% n-butanol and 30–50% acetone precipitation, and subjected to various chromatographic procedures. Lectin affinity chromatography of the samples with concanavalin A–Sepharose 4B showed considerable differences in the elution patterns. Non-denaturing polyacrylamide gel electrophoresis of the culture medium yielded a relatively slow migrating band of activity that coincided with none of the three bands of activity produced by the normal liver extract, nor with the bands of the cell pellet extract. Inhibition patterns were established by measuring the enzyme activities in the presence of varying concentrations of L-phenylalanine, L-leucine, L-homoarginine, and levamisole. The APs from the cell line were neuraminidase sensitive. According to the results the main AP produced and released to the medium by HepG2 cell line is an aberrantly glycosylated tissue non-specific AP. In addition, the differences between the cell-pellet AP and the culture medium AP seemed to stem from different sugar moieties in their structures.
2005 Elsevier Inc. All rights reserved. Keywords: Alkaline phosphatase; Isoenzyme; HepG2; Hepatocarcinoma; Culture medium; Amino acid; Aberrant glycosylation

Alkaline phosphatase (3.1.3.1) (AP) is a zinc and magnesium containing enzyme that is widespread in all living organisms. It is a glycoprotein attached to the membrane of mammalian cells by means of a specific glycan phosphatidylinositol moiety (GPI-anchor) covalently linked to the carboxyl terminus of the polypeptide chain [1–3]. The number of alkaline phosphatase isozymes depends on the species [4]. In humans, four different isoenzymes have been identified because they are encoded by four different gene loci [5,6]. Tissue non-specific alkaline phosphatase (TNAP) includes liver, bone, and kidney isoforms, and constitutes alkaline phosphatases found in *

Corresponding author. Fax: +9821 6404680. E-mail address: [email protected] (R. Yazdanparast).

0006-291X/$ - see front matter
2005 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2005.02.171

other tissues of the human body. While coded by the same genes, isoforms differ in their post-translational modifications such as their glycosylation pattern, which are organ-specific. The tissue-specific isozymes of alkaline phosphatase are placental (PAP), intestinal (IAP), and germ cell (GCAP) alkaline phosphatases. The placental isoenzyme (PAP) is characterized by its heat stability and susceptibility to inhibition by L-phenylalanine. Intestinal isoenzyme (IAP) is also inhibited by L-phenylalanine but is relatively heat-labile. Germ-cell alkaline phosphatase (GCAP) that is also known as Nagao or placental-like isoenzyme shows greater susceptibility toward inhibition by L-leucine when compared with PAP [6]. The liver/bone/kidney isoenzyme (TNAP) is heat-labile and resistant to phenylalanine,

A. Nowrouzi, R. Yazdanparast / Biochemical and Biophysical Research Communications 330 (2005) 400–409

but markedly inhibited by L-homoarginine and levamisole. These differences are often the basis for discriminating isoenzymes from each other. Placental alkaline phosphatase (PAP) increases in serum of healthy women during pregnancy [7] and in case of certain malignancies where it is known as Regan isozyme [8]. Intestinal alkaline phosphatase (IAP) is found in fetal and adult forms, the fetal being considered a heterodimer of adult intestinal and placental isozymes [9]. Germ cell alkaline phosphate (GCAP) is present in small amounts in testis and thymus [6]. As with Regan isozyme, ectopic expression of intestinal isozyme [10,11] and germ cell or Nagao isozyme [12] has been demonstrated in sera of a variety of cancer patients as well as cancerous tissues. Ectopic expression as opposed to eutopic expression refers to the production of an isozyme outside its normal location in terms of the organ involved and is attributed to genetic dedifferentiation during malignant cellular transformation. For example, Warnock and Reisman [13] reported the presence of a variant alkaline phosphatase in extracts of tumors from patients with hepatocellular cancer and intestinal-type alkaline phosphatase was found to exist in the medium of human hepatoblastoma cell line HUH-6 clone 5 by Yamamoto et al. [14]. Aside from ectopic APs, even the eutopically expressed APs in cancer seem to differ from their native counterparts with respect to their electrophoretic migration and heat stability or susceptibility to inhibitory actions of various amino acids. Previous studies comparing normal rat liver-AP and hepatoma AP have demonstrated significant differences in their carbohydrate chains, including high amounts of fucosylated high mannose chains in the hepatoma AP [15,16]. Liver, the main supplier of serum alkaline phosphatase, consists of nonparenchymal and parenchymal cells (hepatocytes) in a ratio of approximately 0.5:1 [17]. Diseases affecting the parenchyma (cirrhosis, hepatitis) usually lead to little change in AP levels of serum with concomitantly large rise in transaminases, however, diseases such as biliary obstruction, that afflict the hepatobiliary system (non-parenchymal cells), may in some instances cause AP elevations of up to 10 times normal with minimal rise in the transaminases; therefore, it is presumed that nonparenchymal cells, i.e., the bile canalicular surface and the hepatocytes of the biliary epithelium, are the origin of alkaline phosphatase, and parenchymal cells produce relatively less of this enzyme. Since 90% of liver cancers, including HepG2 cell line, have parenchymal origin, alkaline phosphatase is valued only second to some parenchymal proteins, as transferrin and a-fetoprotein, in being considered a useful tumor marker. Like other major serum proteins such as transferrin and a-fetoprotein, alkaline phosphatase is a tumor marker. Although the serum levels of alkaline phosphatase

401

increase in many disease conditions including cancer as well as a number of normal circumstances as cigarette smoking, aberrantly glycosylated isoforms are an indication of malignancy. Therefore, the study was aimed at finding how many different AP activities were expressed in hepatocarcinoma cell line HepG2 and to partially characterize them. We also compared the culture medium AP with the AP from the cell membranes of HepG2 cells. Our results indicated one major alkaline phosphatase activity in the HepG2 culture medium and a minor one that was usually observed only in very concentrated samples. The major AP was characterized as normal liver-AP that had undergone further aberrant glycosylation. Our results also indicated that alkaline phosphatase retained in the cell membrane was structurally different from the enzyme released by these cells to the medium and the difference lay in the carbohydrate moiety rather than the primary structures.

Materials and methods Reagents. All routine chemicals were of reagent grade and from Sigma Chemical (St. Louis, USA) or Merck (Germany). Concanavalin A–Sepharose 4B and the acrylamide gel reagents were obtained from Pharmacia Fine Chemicals (Sweden). Diethanolamine, D(+)-glucose, anhydrous, and p-nitrophenyl phosphate disodium salt (PNPP) were from Merck (Germany). 5-Bromo-3-indolyl phosphate, p-toluidine salt (BCIP), nitro blue tetrazolium salt (NBT), L-phenylalanine, L-leucine, L-homoarginine, levamisole, neuraminidase (from Clostridium perfringens, EC 3.2.2.18), human placental alkaline phosphatase, bovine intestinal alkaline phosphatase, and phenylmethylsulfonyl fluoride (PMSF) were purchased from Sigma (St. Louis, USA). 3-[(3-Chloroamidopropyl)dimethylammonio]-1-propanesulfonate (Chaps) was obtained from Duchefa (Netherlands). RPMI 1640-Medium was from Applichem (Darmstadt, Germany). a-Methyl mannopyranoside was purchased from Roche Applied Science (Germany). Fetal bovine serum (FBS) was purchased from University of Tehran, Veterinary School of Medicine. Human hepatocarcinoma cell line HepG2 was obtained from Pasteur Institute in Tehran, Iran. Liver sample preparation. Healthy human liver belonging to a 17year-old Iranian girl was obtained, within 9 h of death, from the Forensic Medicine Department of Tehran University of Medical Sciences with legal permission from this department. Following perfusion with cold normal saline, one piece was sent for pathological examination, a few pieces were dried in the lyophilizer and ground in a mortar, and the remaining tissue was stored at 70 C. Culture medium and cell pellet collection. HepG2 cells (1.5 · 106 cells/ml) were plated in large 75-cm2 flasks and grown at 37 C under 5% CO2–95% air in culture medium consisting of RPMI supplemented with 5% FBS, 1 · 105 U/L penicillin, and 100 mg/L streptomycin. Twenty-four hours after seeding, the culture media were collected, the adherent cells were gently washed with PBS, and fresh complete media were added. After another 24 h, the culture media were again collected. The cells were dislodged using a solution of 5 mM EDTA containing 1 mg/ml heat-inactivated bovine serum albumin and washed twice with PBS; the pellet was saved at 70 C. The collected culture media related to days 1 and 2, from all flasks, were pooled and concentrated by lyophilization and used for the initial analysis explained in Fig. 1. Preparation of partially crude AP extracts. Samples of lyophilized liver tissue (0.03 g) in triplicate were stirred for 4 h in extraction buffers A, B, and C containing 10 mM Tris–HCl, pH 8.5, and 7.6, 0.1 mM

402

A. Nowrouzi, R. Yazdanparast / Biochemical and Biophysical Research Communications 330 (2005) 400–409

Fig. 1. Comparison of relative amounts of medium-AP at day 1 and day 2, by (A) Sephadex G-200 gel chromatography. Samples (1 ml each) of day 1 and day 2 culture medium were independently applied to a column (1.4 · 72 cm) and the enzyme activity was eluted with 100 mM Tris–HCl, pH 8.0, 100 mM NaCl, 0.02 mM ZnCl2, and 0.1 mM MgCl2 at a rate of 17 ml/h with a fraction volume of 1 ml. Activity (solid lines) was measured as explained in Materials and methods and protein (no lines) was monitored by recording A280 for each fraction; (B) Agarose gel electrophoresis. A 0.3% agarose gel was used for analysis with the buffer as above. Lanes I and II, correspond to liver-AP, and lanes 1 and 2, correspond to medium-AP day 1 and day 2, respectively. MgCl2, 0.02 mM ZnCl2, and 5 mM PMSF at 4 C; the buffers differed in the type of detergent they contained. Buffer A contained 10 or 1 ml/ L Triton X-100, buffer B 50 mM Chaps, and buffer C 2%(v/v) Triton X-114. A 200 ml sample of spent culture medium, not related to any specific timing or treatment, was collected and lyophilized to reduce volume. It was then subjected to 40%(v/v) n-butanol extraction followed by 30% and 50%(v/v) acetone precipitation procedures. The same precipitation procedures were performed on the pooled liver tissue homogenates from buffer A. The acetone-precipitated materials of both the culture medium and the liver were dissolved in minimum volumes (1 ml) of buffer containing no detergent and stored at 20 C. The above-mentioned pellet of cells was lyophilized to dryness; three samples (0.03 g each) of pellet were stirred in 3 ml of the above extraction buffers A and B (pH 8.5) for 4 h in the cold room. The cell pellet crude extracts were mixed, passed through a small Sephadex G10 column (1.2 · 4 cm), concentrated by lyophilization, and used for initial analysis by electrophoresis. The rest of the pellet extract was butanol extracted and acetone precipitated as above. To all the samples NaN3 was added to a concentration of 0.01% as preservative. Chromatography. Gel filtration with Sephadex G-200 was performed with 10 mM Tris–HCl, pH 8.0, containing 100 mM NaCl, 0.02 mM ZnCl2, and 0.1 mM MgCl2 as the eluting buffer. Concanavalin A–Sepharose 4B lectin affinity chromatography was performed using both glucose and a-methyl mannopyranoside step gradients as described in detail in Fig. 7. Polyacrylamide gel electrophoresis. The electrophoretic analyses were carried out according to Sambrook and Russel [18], and Wong and Low [19]. In the former, the separating gel consisted of 1.5 mM

Tris–HCl, pH 8.8, and the stacking gel of 1.0 mM Tris–HCl, pH 6.8, with Tris–glycine, pH 8.3, as the buffer. In the other method, the separating gel (6% acrylamide) contained 0.375 M Tris–HCl, pH 8.8, and 5 ml/L Triton X-100, and the stacking gel (5%) contained 0.125 M Tris– HCl, pH 6.8, and 5 ml/L Triton X-100. Tank buffer consisted of 89 mM Tris–borate, pH 8.6, and 5 ml/L Triton X-100. In both methods, electrophoresis was carried out at 20 mA and the AP activity was visualized using 5 mg/ml solution of 5-bromo-3-indolyl phosphate, p-toluidine salt (BCIP) in 100 mM Tris–HCl, pH 9.5, 5 mM MgCl2, and 100 mM NaCl, in the presence or absence of NBT. The reaction was stopped in appropriate time using 20 mM Tris–HCl, 1 mM CaCl2, pH 2.9. Inhibition studies. In three independent trials, 20 ll of each sample (acetone precipitated samples of liver and the cell culture medium, cell pellet, as well as the human placental and the bovine intestinal alkaline phosphatase standards), in duplicate, was measured for residual activity in the presence of increasing concentrations of several inhibitors of alkaline phosphatase isozymes such as L-phenylalanine, L-leucine, L-homoarginine, and levamisole. The varying concentrations of each inhibitor were prepared in the appropriate buffer from 500 mM stock solutions. Bicarbonate buffer (50 mM H2 CO3 –HCO3 , 0.5 mM MgCl2, pH 10.0) was used for levamisole and L-homoarginine inhibition studies. In the case of phenylalanine, 1 M diethanolamine, pH 9.0, and for leucine, 1 M diethanolamine, pH 10.3, were used because both amino acids could not be dissolved in bicarbonate buffer and because the maximum inhibitory capacities of L-Phe and L-Leu have been suggested to be around pHs 9.0 and 10.3, respectively [20,21]. The inhibitor solutions (1 ml) were added to each sample tube followed by incubation of the mixtures at room temperature for 10 min prior to addition of 50 ll substrate (p-nitrophenyl phosphate) solution to a final concentration of 5 mM. After color development in 3 h at 37 C, the reaction was stopped with 0.4 M NaOH and the absorption at 405 nm was recorded; the percent residual activities were calculated against untreated samples. Heat inactivation. Heat treatments were carried out at 37, 56, 65, 70, and 85 C: aliquots of each sample (20 ll) were incubated in a water bath at those temperatures for 15 min. After heating, the samples were chilled quickly on ice and the residual activities were assayed at 37 C. Activity assay. Enzyme activity was measured using 5 mM pnitrophenyl phosphate as the substrate in bicarbonate buffer (50 mM H2 CO3 –HCO3 , 0.5 mM MgCl2, pH 10.0) throughout the paper [22] with the exception of L-phenylalanine and L-leucine in which 1 M diethanolamine buffer containing 0.5 M MgCl2 at pH 9.0 and pH 10.3 was used, respectively. After addition of substrate solution to the enzyme, the mixture was incubated at 37 C. The reaction was stopped by addition of 1 ml of 0.4 M NaOH at an appropriate time. Activity was always expressed as micromoles of p-nitrophenol produced per minute. Neuraminidase treatment. The acetone extracted preparations of liver tissue and the culture medium, as well as both placental and intestinal standards, were further treated with neuraminidase to eliminate sialic acid content of APs and analyzed by PAGE with regard to electrophoretic mobility before and after treatments. A volume of 40 ll of each sample was added to a 50 ll solution of 20 mM Tris–HCl, 60 mM NaCl, 0.2 mM MgCl2, and 2 mM CaCl2, pH 7.0, which contained 0.04%(w/v) NaN3 as a preservative. An equivalent of 0.1 U of neuraminidase (10 ll) was added to each sample [23]. The samples were mixed and incubated at 37 C for up to 24 h. Polyacrylamide gel electrophoresis was performed and the activity bands were visualized with NBT/BCIP.

Results The quantity of AP activities and protein concentrations in the collected day1 and day 2 culture media were determined according to enzyme activity assay given in Materials and methods and Lowry et al. [24], and

A. Nowrouzi, R. Yazdanparast / Biochemical and Biophysical Research Communications 330 (2005) 400–409

semi-quantitatively by G-200 Sephadex gel chromatography. As shown in Fig. 1A, protein concentrations in both days were similar. However, there was a substantial (11.2 ± 1.10-fold) increase of AP activity in day 2 (from 1.0 in day 1 to 11.7 lmol/min/ml in day 2). Similar finding with regard to activity was reported by De Broe et al. [25] who observed a 12-fold increase of the AP activity, between 30 min and 24 h, in the culture medium of HeLa cells; the observed difference in the timing might be due to different experimental conditions or the difference in the cell type. HepG2 cell line was established by Knowles et al. [26]; however, alkaline phosphatase was not listed among the 17 major plasma proteins secreted to cell culture medium by this cell line. We evaluated the relative amounts of AP in both samples also by agarose gel electrophoresis (Fig. 1B). The low levels of AP activity in day-1 medium sample indicate that the FBS used to prepare culture medium contained negligible amounts of alkaline phosphatase activity. Preliminary comparative experiments on healthy human liver extracted in Tris buffer solutions A, B, and C containing different detergents, namely, 1 ml/L Triton X-100, 50 mM Chaps, and 2%(v/v) Triton X-114, respectively, showed similar activity patterns (Fig. 2A, lanes 1–3) that remained unaltered after 40% butanol and 30–50% acetone precipitation procedures (Fig. 2B, lanes 1 and 2). Quantitatively, there were slight differences in the total AP released to the homogenate with each buffer (Table 1). In order to compare the AP of HepG2 cells with the AP of healthy human liver, acetone precipitates of liver tissue and cell pellet powders were prepared as explained in Materials and methods. The same was done to a large volume of the culture medium collected from HepG2 cell culture flasks that had been concentrated by lyophilization. The medium produced 114 lmol/min of total activity and a total protein concentration of 89 mg. Total AP activity and protein from 0.15 g of the liver sample were 44.9 lmol/min and 1.8 mg, respectively; and the cell pellet yielded 0.54 lmol/min of total activity and a total protein concentration of 1.4 mg. It should be mentioned that since the liver powder was exposed to different extraction buffers without prior separation of parenchymal and nonparenchymal cells, the electrophoretic pattern of the healthy human liver would provide a complete picture of AP activities corresponding to all types of liver cells. Fig. 2A shows the relative mobilities of APs from liver extracted in buffers A, B, and C, compared with culture medium, and normal human serum in polyacrylamide gel electrophoresis. Fig. 2B compares the polyacrylamide gel electrophoretic mobility of culture medium alkaline phosphatase with those of normal liver and the cell pellet extracted in buffers A and B. Of the two faint and intense bands of AP activities that were visible in electrophoretic patterns of the concentrated culture media (not shown), usually the

403

Fig. 2. Alkaline phosphatase activity stains of Hep-G2 cell culture medium compared with those of: (A) normal human liver extracted in buffers A (lane 1), B (lane 2), and C (lane 3), culture medium (lane 4), and healthy human serum (lane 5); and (B) healthy human liver in buffer A before acetone precipitation (lane 1), after acetone precipitation (lane 2), culture medium (lane 3), cell pellet AP extracted in buffer A (lane 4), and cell pellet AP extracted in buffer B (lane 5). A 6% separating and a 3.5% stacking gel were used according to Sambrook and Russel with Tris–glycine buffer, pH 8.3. Electrophoresis was performed at 20 mA per gel. Activity bands are stained with BCIP alone in (A) and with NBT/BCIP in (B).

faint band was not seen in dilute samples. The healthy serum contained some of the activity bands of the healthy liver extract in addition to the bone type alkaline phosphatase that travels immediately behind the liver isoenzyme (marked as B and L in Fig. 2A) in non-denaturing electrophoresis. Interestingly, the cell pellet did not show the major activity band found in the culture medium (Fig. 2B, lanes 3–5); treatment of the cell pellet extract with phospholipase C did not reproduce the medium-AP either (data not shown).

404

A. Nowrouzi, R. Yazdanparast / Biochemical and Biophysical Research Communications 330 (2005) 400–409

Table 1 The total AP released by different modes of extraction from normal liver and HepG2 cell-pellet lyophilized powders Sample (g)

Method of extraction

pH

Liver (0.03)

50 mM Chaps 0.1% Triton X-100 0.1% Triton X-100 1% Triton X-100 1% Triton X-100

8.5 7.6 8.5 7.6 8.5

Cell pellet (0.03)

50 mM Chaps 0.1% Triton X-100

8.5 8.5

Fig. 3 indicates the elution pattern of liver, medium, and cell pellet samples using G-200 Sephadex gel chromatography. The medium-AP was found to be neuraminidase sensitive according to Fig. 4. To further characterize the AP of the medium and cell pellet, these samples and the standard human placenta and bovine intestinal APs were subjected to inhibition studies with inhibitors known to be able to discriminate between different AP isoenzymes and heat inactivation studies; Fig. 5 summarizes the effects of various inhibitors and the heat denaturation curves are displayed in Fig. 6. It has been reported that malignant transformations are usually accompanied by modified glycosylation of cell surface glycoproteins [15,16]. To address this issue the ConA elution profiles of the healthy liver, the culture medium, and the cell pellet samples were obtained. Fig. 7A shows the ConA elution profiles of the liver and medium APs in the presence and absence of a representative inhibitor, levamisole. The overall profiles showed

Activity (lmol/min total volume) 51 ± 6.5 43 ± 16.4 62 ± 4.6 73 ± 7.0 78 ± 4.0 4.00 ± 4.3 3.99

[Protein] (mg/total volume)

Specific activity

5.1 ± 1.4 5.6 ± 0.7 5.4 ± 0.8 11.7 ± 1.1 9.1 ± 1.8

9.9 7.5 11.6 7.2 7.2

2.18 ± 0.3 9.5

1.83 0.42

Fig. 4. Electrophoretic pattern of APs. (1) Before neuraminidase treatment and (2) After neuraminidase treatment. The separating gel (6% acrylamide) contained 0.375 M Tris–HCl, pH 8.8, and 5 ml/L Triton X-100, and the stacking gel (5%) contained 0.125 M Tris–HCl, pH 6.8, and 5 ml/L Triton X-100. Electrophoresis was carried out at 20 mA. Tank buffer consisted of 89 mM Tris–borate, pH 8.6, and 5 ml/ L Triton X-100. AP activity was detected with NBT/BCIP as explained in the text. P, human placental AP; I, bovine intestinal AP; M, medium-AP; and L, liver-AP.

a considerable amount of fraction II in the medium compared to the liver although they differed significantly with respect to the amount of AP activity originally present in the loaded samples. Similarly, the comparison of elution profiles of the alkaline phosphatase activity of the cell pellet (Fig. 7B) with that of the culture medium (Fig. 7A) showed that the former had much more affinity for ConA resin than the latter; a clear indication of difference of glycosylation patterns.

Fig. 3. Elution patterns of medium, cell pellet, and liver alkaline phosphatases from Sephadex G-200 column. A 200 ll of each sample was loaded on the column (70 · 1.4 cm) and eluted with Tris–HCl, pH 8.0, at a flow rate of 5 ml/1 h. Fraction volume was 1.7 ml. AP activities in the samples eluted after the void volume of 32 ml. The activity and protein contents of the loaded samples were: 4.5 lmol/min and 3.56 mg for medium-AP; 0.10 lmol/min and 0.28 mg protein for cell pellet AP; and 8.9 lmol/min and 0.39 mg of protein for liver-AP. Absorbance was recorded in the case of blue-dextran at 275 nm and in case of cell pellet at 405 nm.

Discussion A few of serum glycoproteins, such as a-fetoprotein, transferrin, and alkaline phosphatase, are considered tumor markers of hepatoma [27]. Increases in their serum concentrations, their ectopic production, and modifications of their glycosylation patterns are means of differentiating liver cancer from a variety of benign liver diseases.

A. Nowrouzi, R. Yazdanparast / Biochemical and Biophysical Research Communications 330 (2005) 400–409

405

Fig. 5. Effects of varying concentrations of several amino acids on AP activity. Each amino acid was incorporated in the assay mixture so as to give the final concentration indicated. After 10 min of exposure, the substrate was added and, after an incubation period of 3 h, the residual activity was expressed as the percentage of activity relative to the control sample. (A) Levamisole; (B) homoarginine; (C) phenylalanine; and (D) leucine.

Fig. 6. Heat sensitivity of different APs. Each measurement is the mean of three duplicate trials. Experimental details are provided in Materials and methods.

The culture medium produced two bands of enzyme activity upon polyacrylamide gel electrophoresis. The major band of enzyme activity did not coincide with any of the bands seen for liver extracted under various buffer conditions nor with the bands of a normal human serum; the activity band of the HepG2 culture medium did not coincide with the band produced by the cell pellet even after both the cell pellet and medium were exposed to the hydrolyzing action of phospholipase C (data not included). However after treatment with neuraminidase, the main band of the cell culture medium

coincided with one of the activity bands of healthy liver implying that the difference lies in the varying sialic acid content of liver alkaline phosphatase in health and in cancer. The many kinds of variant alkaline phosphatases found in the extracts of tumor of patients with hepatocellular cancer include fetal intestinal AP, adult intestinal AP, the Kasahara AP, and the Nagao or Placental D-variant AP [28–32]. To characterize the type of AP in HepG2 liver cancer cell line, the liver, medium, and cell pellet APs were exposed to the discriminating capabilities of amino acid inhibitors (Fig. 5). Under most conditions cell pellet APs behaved similar to medium AP. With regard to L-leucine, medium and cell pellet APs were the most resistant of all AP types but the inhibiting effects of L-phenylalanine were not significantly different for liver compared to culture medium and cell pellet. L-Phenylalanine and L-leucine are known for their distinguished inhibitory effects on the placental and germ cell isozymes, respectively [6,33]. Levamisole and L-homoarginine are known to effectively inhibit liver isozyme at very low concentrations [4]. Under the effect of levamisole which inhibits all but intestinal AP, the medium AP behaved like placental AP, while under homoarginine it behaved similar to liver-AP, at low inhibitor concentrations. At higher concentrations of these inhibitors these APs behaved

406

A. Nowrouzi, R. Yazdanparast / Biochemical and Biophysical Research Communications 330 (2005) 400–409

Fig. 7. (A) Elution profile of liver-AP (left) and medium-AP (right) during ConA–Sepharose 4B lectin affinity chromatography in the absence and the presence of a chosen concentration of levamisole. Each graph consists of two activity profiles, obtained simultaneously in the absence and in the presence of the inhibitor. In the case of other inhibitors, the profiles were similarly synchronous (data not included). (B) Elution profile of HepG2 cell-pellet AP during ConA–Sepharose 4B lectin affinity chromatography. A 500 ll of each sample was loaded on a 0.5 · 9.0 cm column and eluted with step gradients of glucose and a-methyl mannopyranoside (a-MM). The activity and protein contents of the loaded samples were 11.4 lmol/min and 8.9 mg (specific activity, 1.28 lmol/mg) for medium-AP, 22.4 lmol/min and 0.9 mg (specific activity, 24.9 lmol/mg) for liver-AP, and 0.27 lmol/ min and 0.7 mg (specific activity, 0.38 lmol/mg) for the cell-pellet AP.The highlighted solid circles on the profiles mark the starting point of each gradient: Fractions 1–5, Tris–HCl-buffered saline (TBS); 6–15, 0.1 M glucose; 16–25, 0.5 M glucose; 26–35, 0.01 M a-MM; and 36–50, 0.5 M a-MM; prepared in TBS. From the eluted fractions, 50-microliter samples were used to obtain the activity profiles after incubation at 37 C for 18–24 h prior to recording the absorbance at 405 nm.

individually unique. The results in Fig. 5 suggested that the major AP in the culture medium and the cell pellet APs were neither placental nor Nagao, not even a hybrid of intestine and placenta. Neuraminidase sensitive nature of the activity band of the culture medium AP shown in Fig. 4 ruled out the possibility of adult intestinal and fetal intestinal type APs as they both are neuraminidase resistant, i.e., their electrophoretic mobility does not change after treatment with neuraminidase since they exist in asialo form [34]. The reverse is true about the Kasahara isoenzyme, which resembles the intestinal isozyme in all properties except its electrophoretic retardation after treatment with neuraminidase [32]. Medium-AP was insensitive to inhibition by phenylalanine and therefore could not be considered Kasahara isozyme. Based on the display of similar behavior of the HepG2 AP (culture medium or cell pellet) and healthy liver AP under the effects of amino acid inhibitors, it was concluded that the major AP in the culture medium and the cell pellet AP were the same as healthy liver-AP with respect to their primary structures although the two enzyme forms, partially purified by Yamamoto et al. [14] from medium spent for the culture of HUH-6 clone 5 cells, originally developed from hepatoblastoma tissue, were both identified as intestinal in type, based on their behavior toward inhibition by various amino acids, reactivity with antiserum against intestinal AP, and the lack of neuraminidase sensitivity. The ConA elution profiles (Figs. 7A and B) differed considerably with respect to the quantities of fraction I (the unbound fraction), fraction II (the weakly bound fraction) as well as strongly bound fractions or fraction III that are absent in the medium profile but apparently abundant in the cell pellet profile. In the ConA affinity profile, when the column is eluted with 0.01 M a-methyl mannopyranoside followed by 0.5 M of the same sub-

stance, fraction I corresponds to multiantennary or bisected complex type carbohydrate, fraction II to biantennary complex type with or without fucose, and fraction III represents high mannose or hybrid type sugar [35]. The presence of more of both fraction I and fraction II in the cell culture medium AP was in accordance with the numerous reports concerning the type of altered glycosylation in hepatoma which includes the increase of highly branched sugar chains, and those with fucosylated trimannosyl core. Such studies have been performed mainly on transferrin, although other serum glycoproteins such as hemopexin, a1-antitrypsin, a2-HS glycoprotein, a-fetoprotein, and a1-acid glycoprotein too have been detected with altered glycosylation [36–38]. According to the effect of the chosen concentrations of amino acid inhibitors on ConA profile, a prototype pair is shown in Fig. 7A, medium-AP showed more resistance towards the inhibitory action of all amino acids except L-Phe. Both liver and medium APs seemed to act synchronously under the effects of inhibitors. The HepG2 APs (medium and membrane) could be the liver isoenzyme with increased carbohydrate content that had led to an increase in mass, as seen in Fig. 3, and a greater heat resistance of both medium-AP and cell pellet AP compared to the liver-AP. Several investigators have established a positive relationship between N-linked glycosylation and thermostability [39,40]. With regard to heat stability, placental isoenzyme is the most heat stable and the liver specific isozyme is found to be the least stable type of alkaline phosphatases. According to Fig. 6, the heat denaturation curve of the culture medium showed a point of inflection at 65 C, which was possibly due to the minor form of AP present in the sample because the heat denaturation curve of the purified major medium-AP lacked this inflection point (the correspond-

A. Nowrouzi, R. Yazdanparast / Biochemical and Biophysical Research Communications 330 (2005) 400–409

ing curve not included in Fig. 6). We did not determine the nature of the minor AP present in the culture medium due to insufficient amounts available; however, the heat resistance observed at 65 C could be an indication of an intestinal form of AP or yet another highly glycosylated isoform. Intestinal type alkaline phosphatase was previously reported to exist, as a minor component, in hepatocellular carcinoma tissue [30]. The differences between the healthy liver tissue alkaline phosphatase and that of hepatoma liver cell line, in terms of electrophoretic mobility, the elevated heat stability, and higher resistance to inhibitors, were established to be due to different glycosylation patterns. Although we did not establish the type of modification in the carbohydrate moieties, our results clearly signified the role of sialylation. When we turned our attention to the difference of electrophoretic mobility observed between the medium AP and the cell pellet AP, a further difference in glycosylation pattern was manifested by the ConA profiles for the released enzyme to the medium and the cell membrane anchored one, both distinct from that of healthy liver (compare the medium and liver profiles in Figs. 7A with that in B). This was contrary to our anticipation that the high molecular weight of the cell pellet AP, as shown in Fig. 3, most likely, stemmed from the assembly of anchor intact enzymes released from cell membranes which would in that case lead to a more pronounced fraction I in the ConA profile. Although we used a slightly different system of eluting solutions, our ConA profiles could still be compared with those obtained by other researchers; for example, the pattern for the cell pellet AP with a marked expansion of fraction III (Fig. 7B) resembled that established for cancerous liver tissue [41]. The elution profile of healthy human liver was also similar to that of other researchers except that in our case the third fraction was spread over three separate areas. The lectin affinity profile of cell culture medium AP differed from that of healthy human serum in lacking fraction III. In more concentrated medium samples, a small peak of activity was observed just past the second point of step gradient (data not shown). The cell culture medium might be considered a near appropriate match for hepatoma afflicted human serum with the added advantage that here the AP from other sources would not interfere with the analytical procedures aimed only at the AP involved in cancerous liver tissue. The accumulation of AP in the culture medium would also imitate the situation in real hepatoma in which excessive sialylation of alkaline phosphatase hinders its clearance by galactose receptors from circulation thereby increasing its levels in hepatic serum [42,43]. In summary, it seemed that substantial amounts of alkaline phosphatase are produced by HepG2 hepatocellular carcinoma which may be due to distinctive gene rearrangement that has occurred in chromosome No. 1

407

of HepG2 cells [4,26]. Such rearrangements may have moved the TNAP gene locus to a position adjacent to an enhancer region of a protein common to hepatocytes. Our study revealed that two electrophoretically distinct types of alkaline phosphatase activities in the bathing medium of HepG2 hepatocarcinoma cell lines, the major one being TNAP or liver-AP with altered modification of its glycosylated polysaccharide moieties resembling modifications previously reported for transferrin and a-fetoprotein. These modifications seemed to affect both heat stability and amino acid inhibition sensitivity. It has previously been shown that antennae flexibility and bulkiness may serve in protecting a glycoprotein from proteolysis or modifying protein conformation and function [44]. The weaker inhibitory action of inhibitors may be caused by either the slight conformational changes in the protein structure brought about by aberrant glycosylation or the bulkier nature of polysaccharide chains relative to normal liver that hinder accessibility of the primary structure to those inhibitors. It has been reported that most of the glycoproteins secreted from the normal hepatocytes are not fucosylated at the trimannosyl core of their sugar chains. In contrast, the sugar chains of glycoproteins that are retained within intracellular organelles of normal hepatocytes are fucosylated at their trimannosyl core [36,45]. Therefore, further investigations are required as to the extent of total sialic acid content of the cell pellet AP, the culture medium AP in hepatocellular carcinoma, and the healthy human liver-AP, as well as the amount of fucose in each fraction eluted from ConA to determine more precisely the underlying differences in the sugar moieties of these AP samples. The observation that cell membrane AP did not produce the activity band similar to the intense band of medium AP, even after treatment with phospholipase C, reminds the question about the role of carbohydrate modification in the release of the enzyme from the cell surface to the medium, especially in cancer.

Acknowledgments We are thankful to the research council of the University of Tehran for financial support of this investigation. We thank Dr. Fakhreddin Taghaddosi-nejad of the Forensic Medicine Department of Tehran University of Medical Sciences for providing us with a saline perfused healthy human liver.

References [1] M.G. Low, D.B. Zilversmit, Role of phosphatidylinositol in attachment of alkaline phosphatase to membranes, Biochemistry 19 (1980) 3913–3918.

408

A. Nowrouzi, R. Yazdanparast / Biochemical and Biophysical Research Communications 330 (2005) 400–409

[2] N.M. Hooper, Glycosyl-phosphatidylinositol anchored membrane enzymes, Clin. Chim. Acta 266 (1997) 3–12. [3] M.T. Lehto, F.J. Sharom, Proximity of the protein moiety of a GPI-anchored protein to the membrane surface: a fret study, Biochemistry 41 (2002) 8368–8376. [4] H. Harris, The human alkaline phosphatases: what we know and what we donÕt know, Clin. Chim. Acta 186 (1989) 133–150. [5] L.E. Seargeant, R.A. Stinson, Evidence that three structural genes code for human alkaline phosphatases, Nature 281 (1979) 152–154. [6] J.L. Millan, T. Manes, Seminoma-derived Nagao isozyme is encoded by a germ-cell alkaline phosphatase gene, Proc. Natl. Acad. Sci. USA 85 (1988) 3024–3028. [7] A. Sembaj, E. Sanz, I. Castro, A. Gonzalez, C. Carriazo, J.M. Barral, Alkaline phosphatase isoenzymes in plasma of chagasic and healthy pregnant women, Mem. Inst. Oswaldo Cruz 94 (1999) 785–786. [8] W.H. Fishman, N.R. Inglis, S. Green, C.L. Anstiss, N.K. Gosh, A.E. Reif, R. Rustigian, M.J. Krant, L.L. Stolbach, Immunology and biochemistry of Regan isoenzyme of alkaline phosphatase in human cancer, Nature 219 (1968) 697–699. [9] C.M. Behrens, C.A. Enns, H.H. Sussman, Characterization of human foetal intestinal alkaline phosphatase, Biochem. J. 211 (1983) 553–558. [10] T. Kuwana, S.B. Rosalki, Intestinal variant alkaline phosphatase in plasma in disease, Clin. Chem. (1990) 1918–1921. [11] L.-C. Tsai, M.-W. Hung, Y.-H. Chen, W.-C. Su, G.-G. Chang, T.-C. Chang, Expression and regulation of alkaline phosphatases in human breast cancer MCF-7 cells, Eur. J. Biochem. 267 (2000) 1330–1339. [12] S. Watanabe, T. Watanabe, W.B. Li, B.-W. Soong, J.Y. Chou, Expression of germ cell alkaline phosphatase gene in human choriocarcinoma cells, J. Biol. Chem. 264 (1989) 12611–12619. [13] M.L. Warnock, R. Reisman, Variant alkaline phosphatase in human hepatocellular cancers, Clin. Chim. Acta 24 (1968) 5–11. [14] H. Yamamoto, M. Tanaka, H. Nakabayashi, J. Sato, T. Okochi, S. Kishimoto, Intestinal-type alkaline phosphatase produced by human hepatoblastoma cell line HUH-6 Clone 5, Cancer Res. 44 (1984) 339–344. [15] Y. Ikehara, Y. Hayashi, S. Ogata, A. Miki, T. Kominami, Purification and characterization of a major glycoprotein in rat hepatoma plasma membranes, Biochem. J. 241 (1987) 63–70. [16] T. Endo, T. Fujiwara, Y. Ikehara, A. Kobata, Comparative study of the sugar chains of alkaline phosphatases purified from rat liver and rat AH-130 hepatoma cells. Occurrence of fucosylated high mannose-type and hybrid-type sugar chains, Eur. J. Biochem. 236 (1996) 579–590. [17] S.N. Bhatia, U.J. Balis, M.L. Yarmush, M. Toner, Effect of cell– cell interactions in preservation of cellular phenotype: cocultivation of hepatocytes and nonparenchymal cells, FASEB J. 13 (1999) 1883–1900. [18] J. Sambrook, D. Russel, Molecular Cloning, A Laboratory Handbook, third ed., Cold Spring Harbor Laboratory Press, New York, 2001. [19] Y.W. Wong, M.G. Low, Phospholipase resistance of the glycosylphosphatidylinositol membrane anchor on human alkaline phosphatase, Clin. Chem. 38 (1992) 2517–2525. [20] D.A. Byers, H.N. Fernley, P.G. Walker, Studies on alkaline phosphatase. Inhibition of human-placental phosphoryl phosphatase by L-phenylalanine, Eur. J. Biochem. 29 (1972) 197–204. [21] M.F. Hoylaerts, T. Manes, J.L. Millan, Molecular mechanism of uncompetitive inhibition of human placental and germ-cell alkaline phosphatase, Biochem. J. 286 (1992) 23–30. [22] I. Koyama, Y. Sakagishi, T. Komoda, Different lectin affinities in rat alkaline phosphatase isozymes: multiple forms of the isozyme isolated by heterogeneities of sugar moieties, J. Chromatogr. 374 (1986) 51–59.

[23] M.-C. Hofmann, W. Jeltsch, J. Brecher, H. Walt, Alkaline phosphatase isozymes in human testicular germ cell tumors, their precancerous stage, and three related cell lines, Cancer Res. 49 (1989) 4696–4700. [24] O.H. Lowry, N.J. Rosebrough, A.L. Farr, R.J. Randal, Protein measurement with the Folin Phenol reagent, J. Biol. Chem. 193 (1951) 265–275. [25] M.E. De Broe, R.J. Wieme, G.N. Logghe, F. Roels, Spontaneous shedding of plasma membrane fragments by human cells in vivo and in vitro, Clin. Chim. Acta 81 (1977) 237–245. [26] B.B. Knowles, C.C. Howe, D.P. Aden, Human hepatocellular carcinoma cell lines secrete the major plasma proteins and hepatitis B surface antigen, Science 209 (1980) 497–499. [27] J.L. Millan, Alkaline phosphatase as a reporter of cancerous transformation, Clin. Chim. Acta 209 (1992) 123–129. [28] K. Higashino, S. Kudo, Y. Yamamura, Further investigation of a variant of the placental alkaline phosphatase in human hepatic carcinoma, Cancer Res. 34 (1974) 3347–3351. [29] K. Higashino, S. Kudo, Y. Yamamura, T. Honda, J. Sakurai, Possible identity between the hepatoma alkaline phosphatase and an isozyme of human amniotic membrane (FL cells), Clin. Chim. Acta 60 (1975) 267–272. [30] K. Higashino, R. Otani, S. Kudo, Y. Yamamura, A fetal intestinal-type alkaline phosphatase in hepatocellular carcinoma tissue, Clin. Chem. 23 (1977) 1615–1623. [31] L. Wray, H. Harris, Demonstration using monoclonal antibodies of inter-locus heteromeric isozymes of human alkaline phosphatase, J. Immunol. Methods 55 (1982) 13–18. [32] K. Higashino, K. Muratani, T. Hada, H. Imanishi, Y. Amuro, Y. Yamamoto, J. Furuyama, K. Hirano, Y.-M. Hong, M. Shimokura, T. Hirano, T. Kishimoto, Purification and some properties of the fast migrating alkaline phosphatase in FLamnion cells (the Kasahara isoenzyme) and its cDNA cloning, Clin. Chim. Acta 186 (1989) 151–164. [33] T. Nakayama, M. Yoshida, M. Kitamura, L-Leucine sensitive, heat-stable alkaline-phosphatase isoenzyme detected in a patient with pleuritis carcinomatosa, Clin. Chim. Acta 30 (1970) 546– 548. [34] Y. Nishihara, Y. Hayashi, T. Adachi, I. Koyama, T. Stigbrand, K. Hirano, Chemical nature of intestinal-type alkaline phosphatase in human kidney, Clin. Chem. 38 (1992) 2539–2542. [35] S.R. Carlsson, T. Stigbrand, Partial characterization of the oligosaccharides of mouse thymocyte Thy-1 glycoprotein, Biochem. J. 221 (1984) 379–392. [36] K. Yamashita, N. Koide, T. Endo, Y. Iwaki, A. Kobata, Altered glycosylation of serum transferrin of patients with hepatocellular carcinoma, J. Biol. Chem. 264 (1989) 2415–2423. [37] K. Yamashita, K. Taketa, S. Nishi, K. Fukushima, T. Ohkura, Sugar chains of human cord serum a-fetoprotein: characteristics of N-linked sugar chains of glycoproteins produced in human liver and hepatocellular carcinomas, Cancer Res. 53 (1993) 2970– 2975. [38] B. Campion, D. Leger, J.-M. Wieruszeski, J. Montreuil, G. Spik, Presence of fucosylated triantennary, tetraantennary and pentaantennary glycans in transferrin synthesized by the human hepatocarcinoma cell line Hep G2, Eur. J. Biochem. 184 (1989) 405–413. [39] K. Higashino, M. Hashinotsume, Y. Yamamura, Effect of neuraminic acid removal on the properties of a variant alkaline phosphatase in hepatoma, Clin. Chim. Acta 40 (1972) 305–307. [40] D.N.P. Doan, G.B. Fincher, Differences in thermostabilities of barley (1 fi 3, 1 fi 4)-b-glucanases are only partly determined by N-glycosylation, FEBS Lett. 309 (1992) 265–271. [41] I. Koyama, M. Miura, H. Matsuzaki, Y. Sakagishi, T. Komoda, Sugar-chain heterogeneity of human alkaline phosphatases: differences between normal an tumor-associated isozymes, J. Chromatogr. 413 (1987) 65–78.

A. Nowrouzi, R. Yazdanparast / Biochemical and Biophysical Research Communications 330 (2005) 400–409 [42] X. Li, B. Mortensen, C. Rushfeldt, N.-E. Huseby, Serum cglutamyltransferase and alkaline phosphatase during experimental liver metastases. Detection of tumor-specific isoforms and factors affecting their serum levels, Eur. J. Cancer 34 (1998) 1935–1940. [43] E. Blom, M.M. Ali, B. Mortensen, N.-E. Huseby, Elimination of alkaline phosphatases from circulation by the galactose receptor.

409

Different isoforms are cleared at various rates, Clin. Chim. Acta 270 (1998) 125–137. [44] N. Sharon, Glycoproteins, TIBS 9 (1984) 198–202. [45] K. Matsumoto, Y. Maeda, S. Kato, H. Yuki, Alteration of asparagine-linked glycosylation in serum transferrin of patients with hepatocellular carcinoma, Clin. Chim. Acta 224 (1994) 1–8.