Pittsburgh Cancer Institute (S. M. G.), University of Pittsburgh, Pittsburgh, Pennsylvania 15213. Received 2 September 1998; accepted 25 November 1998).
In VitroCell. Dev.Biol.~Anima135:190-197,April 1999 © 1999Societyfor In VitroBiology 1071-2690/99 $05.00+0.00
ISOLATION A N D CHARACTERIZATION OF A HUMAN HEPATIC EPITHELIAL-LIKE CELL LINE (AKN-1) FROM A NORMAL LIVER ANDREAS K. NUSSLER,1 GIANNA VERGANI, SUSANNE M. GOLLIN, KENNETH DORKO, SUSANNE GANSAUGE, SIDNEY M. MORRIS, JR., ANTONY J. DEMETRIS, MINORU NOMOTO, HANS G. BEGER, AND STEPHEN C. STROM
Departments of General Surgery (A. K. N., S. G., H. G. B.), and Anatomy (G. E), University of Ulm, 89075 Germany, and Departments of Pathology (A. J. D., M. N., K. D., S. C. S), Molecular Genetics and Biochemistry (S. M. M.), Human Genetics and the University of Pittsburgh Cancer Institute (S. M. G.), University of Pittsburgh, Pittsburgh, Pennsylvania 15213 Received 2 September 1998; accepted 25 November 1998)
SUMMARY The isolation and characterization of human liver cell lines are rather difficult due to limited material and poor growth in cell culture. In this report, we present the isolation, culture and characterization of a new epithelial-like liver cell line (AKN-1) with a heterogeneous cell population and many characteristics of the biliary epithelium. The AKN-1 cell line stained positively with antibodies to epithelial cytokeratin polypetides CK 8, 18, and 19. In addition, the cell line expressed the anti-human epithelial-related antigen (MOC-31), the human epithelial antigen (HEA), and the gamma-glutamyl transpeptidase, the hematopoietic growth factor, stem cell factor, and also its receptor, c-kit. The cell line failed to express albumin and factor 8 by immunohistochemistry. It did show, however, a twofold increase in 7-ethoxyresorufin-O-deethylase activity. Cytogenetic characterization revealed rare breakpoints in chromosome 2, which to our knowledge, have not yet been reported in liver cells.
Key words: biliar epithelial-like cells; cytochrome P450 activity; human liver; chromosome. isolate nonparenchymal cells. Using this method, we isolated a nonparenchymal cell fraction with epithelial-like cell characteristics of hepatocytes that, after culture, resulted in the AKN-1 cell line.
INTRODUCTION Over the last years, significant knowledge has been gained on various aspects of human liver diseases. Besides clinical studies, basic research studies have examined hepatocellular functions, such as drug hepatoxicity, regeneration, malignant transformation, and chemical carcinogenesis. However, the lack of human liver tissue available for research purposes makes these studies rather difficult, and investigations are often done either in other species or in human tumor cell lines derived from patients undergoing liver resection. Many laboratories are investigating interactions between different cell types within the liver; however, the isolation of epithelial or endothelial cells from normal human liver and the establishment of cell lines has been difficult due to limited biopsy material. Recently, it was demonstrated that human hepatocyte lines can be established by coculturing the hepatocytes with rat liver epithelial cells (27). Nevertheless, the purity of these hepatocyte lines was only 95% (27) and the remaining 5% of cells were of rat origin. However, the use of coculture models (hepatocytes plus nonparenchymal cells) to maintain hepatocyte functions is well established (8,19,27). We have recently shown that human liver tissue obtained from specimens that either were refused for transplantation due to medical reasons or from livers split for transplantation into a smaller recipient can provide large numbers of normal isolated hepatocytes (4). An adaptation of our hepatocyte isolation technique can be also used to
MATERIALS AND METHODS
Cell isolation. The human liver sample was obtained from a 10-yr-old male through the Keystone Skin Bank and cold-preserved by in situ infusion of University of Wisconsin (UW) solution. Although the liver was considered normal, no recipient was available and therefore the specimen was donated for research. The liver sample was obtained according to the institutional guidelines of the Keystone Skin Bank and the University of Pittsburgh and tested negative for HBV and HIV. Macroscopically, the specimen was considered normal. Hepatocytes were isolated by collagenase-P (Boehringer Mannheim, Indianapolis, IN) digestion as described recently (4,21). Briefly, after enzymatic digestion, hepatocytes were separated from the nonparenchymal cell fraction by differential centrifugation at 50 × g. The pellet contained mainly hepatocytes which we further purified to study various aspects of primary human hepatocyte functions (2,22,23). The supematant was used to isolate nonparenchymal cells. The supernatant was centrifuged in a 50-ml Falcon tube at 400 × g for 5 rain at 4 ° C, washed in phosphate-buffered saline (PBS), and then passed over a Percoll gradient (Pharmacia, Piscataway, NJ) with densities of 30, 20, and 10% Percoll at 50 X g for 20 min at 4 ° C. Ceils at the interface of 30 and 20% as well as 20 and 10% were collected separately, washed twice in PBS, and placed into culture. Using this technique, we isolated approximately 5 X 107 nonparenchymal cells from each density fraction. Viability was assessed by trypan blue exclusion and exceeded 90%. Cell culture. The freshly isolated cell fractions were plated onto gelatincoated culture plates (Coming Glass Works, Coming, NY). The initial culture medium consisted of Williams medium E supplemented with insulin (10 6 M), HEPES (N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid) (15 mM), L-glutamine (2 raM), penicillin (500 U/ml), streptomycin (500 gg/ml), 10% low endotoxin calf sermn (all from GIBCO, Eggenstein, Germany) and 25 ng
1To whom correspondence should be addressed at Humboldt-University Berlin, Campus Virchow-Klinikum. Department of General Surgery and Transplantation, Augustenburger Platz 1, 13353 Berlin, Germany. 190
191
HUMAN LIVER BILIAR EPITHELIAL-LIKE CELLS human recombinant epidermal growth factor (EGF, Pepro Tech Inc., Rocky Hill, NJ) per ml (2,4,21,22,23). Within l0 d, the cell fraction isolated from the interface of the 30 and 20% Percoll gradient showed uniform cell growth and was called AKN-1. Cells collected from the interface of the 20 and 10% Percoll gradient did not grow. From the 10th passage onwards, the cells were cultured and kept in serum-free, chemically defined medium (CDM) without EGF supplementation as described by Isom et al. (11). The cell line has now been cultured for more than 4 yr. The cells may be grown either directly on plastic petri dishes, or collagen- or gelatin-coated glass slides (Nunc Inc., Naperville, IL) in a humidified incubator with 95% air and 5% C02 at 37 ° C. Light microscopy (LM). Cells were grown on plastic culture dishes and examined under the phase-contrast microscope. Other cells grown on covex~ slips were stained by the periodic aci&Schiff reaction (PAS) in the presence of diastase. Transmission electron microscopy (TEM). For transmission electron microscopy cells were grown on petri dishes (Nunclon, 35 × 10 ram, Nunc, Wiesbaden, Germany) until semiconfluence, washed twice, fixed in 3% vol/ vol glutaraldehyde (0.1 M PBS, 2% vol/vol sucrose, pH 7.3), and postfixed with 2% osmium tetroxide. Ultrathin sections were analyzed in a Philips EM 301 electron microscope (Philips, Einthofen, The Netherlands) (33). Scanning electron microscopy (SEM). For SEM studies, ceils were grown on tissue culture coverslips (Thermanox, 10.5 × 22 mm, Lab-Tek Division, Miles Scientific Inc., Naperville, IL) rinsed with prewarmed PBS without Ca2+ and Mgz+. and fixed in 3.5% vol/vol glutaraldehyde (0.1 M phosphate buffer, 2% vol/vol sucrose, pH 7.3) for 2 h. After a brief rinse in distilled water, cells were dehydrated in alcohol and dimethoxypropane, and sputte1~eoated with 10 nm gold (MED 010 Balzers Union. Liechtenstein). SEM examination was carried out with a Zeiss DSM 962 microscope (Carl Zeiss, Oberkochen, Germany) (33). lmmunohistochemical analysis. Staining of AKN-1 cells was performed on collagen-coated four-well glass slides. Cells were seeded at a density of 20 000 cells/well in CDM as described before. The antibodies used for these investigations are all directed towards human proteins and include anti-c-kit (rabbit polyclonal from Collaborative Biomedical, Waltham, MA), CAM 5.2 (eytokeratins 8 + 18, mouse monoclonal from Bceton Dickinson, Bedford, MA), stem cell factor (SCF; rabbit polyelonal AB255NA, from R + D Systems, Minneapolis, MN), cytokeratin 19 (M772, mouse monoclonal), alpha-l-antitrypsin (cd-AT, rabbit polyclonal A012), and albumin (rabbit anti-human polyclonal, A001) all from the DAKO Corp. (Carpinteria, CA). Nuclear p53 protein accumulation was detected with the mouse monoclonal D01 antibody (Oncogene Science, Cambridge, MA). To substantiate the epithelial-like origin of the cell line, we stained for the following: epithelial-related antigen (anti-human mouse monoclonal MOC31), human epithelial antigen (HEA), mouse anti-human from Fisher Scientific Co., Pittsburgh, PA, and for the gamma-glutamyl transpeptidase (GGT), 1/5000 goat anti-rat polyclonal, a gift from Dr. Bryan Petterson, University of Pittsburgh, PA. The staining procedures used were from the Vectastin ABC/ DAB kit (Vector Labs Inc., Burlington, CA) and were carried out according to the manufacturer's instructions. Fluorescence-activated cell sorting (FACS) analysis. FACS-Scan (Becton Dickinson, Heidelberg, Germany) analysis was performed for the constitutive expression of the intracellular adhesion molecule-1 (ICAM-1/CD54), CD44std, and HLA-1. For detection of human antigens, mouse anti-human monoclonal antibodies directed against the antigens were used at a dilution of 1/100 according to the manufacturer's instructions. The adhesion molecules ICAM-1 and CD44std. were detected with the antibodies RR/1 and SSF2, respectively (Bender and Co., Vienna, Austria). HLA class I expression was examined with the W6/32 monoclonal antibody (DAKO, Hamburg, Germany). The Tip E monoclonal antibody (Oncogene Science, Cambridge, MA) served as a negative control. The binding of the monoclonal antibodies to constitutively expressed antigens was detected by an IgG purified goat antimouse fluorescein isothiocyanate-labeled antibody (Becton Dickinson, San Jose, CA). Statistical analysis of the binding capacity was performed with the Cellquest program (Becton Dickinson, Heidelberg, Germany). Isolation and analysis of RNA. Total RNA was isolated from cultured AKN1 ceils, cultured hmnan hepatocytes, freshly isolated human hepatocytes, and human and rat livers, and subjected to Northern blot analysis as described (2,22,23). Cloned cDNAs used for hybridization included those encoding for human albumin (2.2 kb mRNA; kindly provided by Dr. Joseph Locker, University of Pittsburgh, Pittsburgh, PA), rat argininosuccinate synthetase (1.67
kb mRNA) (20), rat tyrosine aminotransferase (2.4 kb mRNA) (7), and rat 18S ribosomal RNA (1.87 RNA) (13). Cytochrome P450 enzyme activit~: To investigate whether the AKN-1 cell line exerts some cytochrome P450 function, we performed the following experiments: Cells were grown to confluence in Williams' medium E, then incubated for 48 h with 1 gM dexamethasone and 1 mM phenobarbital as well as 1 ~g 3-methylcholanthrene per ml to measure CYP3A and CYP1A isoforms, respectively. Then we analyzed for 7-ethoxycoumarin-O-deethylase (ECOD) and 7-ethoxyresorufin-O-deethylase (EROD) activity in the supernatant (14) over the period of 1 h using a fluorescent Spectrophotometer (Lumistar, BMG, Offenburg, Germany). Results are expressed as picomoles of product formed/mg protein/rain. Cytogenetic analysis. Cell cultures between Passages 15 to 20 were treated with 0.02 gg Colcemid (Irvine Scientific, Santa Ana, CA) per ml for 16 h and removed from the flasks with trypsin/EDTA solution (0.25-0.1 g/L). Cells were incubated in 0.075 M KC1 solution for 30 rain at 37 ° C and harvested by conventional cytogenetic procedures. Chromosomes were prepared for light microscopy with standard methods, including those for trypsin-Giemsa banding. Twenty-two ceils were fully analyzed and karyotyped. Composite karyotypes were prepared according to ISCN (10). In addition, 600 cells were evaluated by microscopic observation to determine the frequency of tetraploidy. Dual color fluorescence in situ hybridization was carried out on the fixed cell preparations from the cytogenetic harvests with a biotin-labeled alphasatellite probe for chromosome 8 (D8Z1) and a digoxigenin-labeled alphasatellite probe for chromosome 20 (D20Z1, Oncor, Gaithersburg, MD). The probes were hybridized to the slides following the manufacturer's instructions. Briefly, slides were pretreated in 2 × SSC (saline sodium citrate), dehydrated in a graded series of alcohols, and denatured in 70% formamide/2 x SSC. The chromosome 8 and 20 alpha-satellite probes were mixed and denatured, applied to the slides, and hybridized overnight. The following day, the slides were wasbed in 0.25 X SSC for 5 minat 72 ° C and rinsed in 1 X PBD (phosphate-buffered detergent). Detection was completed by application of fluoreseein-labeled avidin and rhodamine-labeled anti-digoxin to the slides, followed by additional 1 × PBD washes. Slides were eounterstained with 4',6-diamidino-2-phenylindole and viewed with an Olympus fluorescence photomicroseope equipped with an appropriate triple band pass filter. Tumorigenicity test. Cells were harvested from cultures, and 100-gl aliquots containing 10~ cells suspended in serum-free medium were injected subcutaneously into the intrascapular fat pads of 10 athymic nude mice (Harlan Sprague Dawley, Indianapolis, IN). Fourteen d postinjeetion, tumors were harvested and the size was measured.
Single-strand conformationpolymorphism (SSCP) analysis of the p53 gene. High-molecular-weight DNA was extracted from the AKN-1 cell line according to the method of Sambrook et al. (29). Polymerase chain reaction (PCR) and subsequent SSCP analysis was performed by the methods and PCR primers described by Orita et al. (26) and Toguchida et al. (30), with minor modifications (5). Briefly, five sets of primers for exons 4-9 of the p53 gene were used to amplify these exons. Each fragment was amplified by PCR with Gene ATAQ Controller (Pharmacia LKB, Freiburg, Germany) in the presence of radiolabeled [a-32p]dCTP. The PCR products were electrophoresed at room temperature. The gel was dried and exposed to a Kodak X-Omat film at - 80 ° C for 12 h with an intensifying screen. RESULTS
Light microscopy (LM). Under the light microscope, the cell line appears to be epithelial-like and consist of a heterogeneous cell population. The monolayer contains several cells undergoing mitoses and the active cell division rate is about 28 min from prophase to metaphase. The monolayer is formed by flat, polygonal-shaped cells adhering to the plastic surface. The cells form filopodia to contact neighboring cells at low density (Fig. 1 a). Transmission electron microscopy (TEM). The TEM confirmed the LM results, showing epithelial-like cells with a polygonal shape and the presence of one or two central nuclei with numerous nuclear pores (Fig. 1 b). These structures resembled eanaliculi biliares (Fig. 1 b). The cells show some evidence of differentiation and cell contact with the formation of junctional complexes. In about 90% of the cells, focal contacts are linked by junctional complexes consisting of des-
192
NUSSLER ET AL.
a
C
'~
J
:: : ,1
i=
M
FIG. 1. a, Morphologyof AKN-1 cells in culture, illustrating their epithelial-like morphology. Note mitosis (M). Magnification, × 860. b, Transmission electron micrograph of AKN-1 cells. The cells show central nucleus (N) and formations resembling canaliculi biliares (CB). Magnification, × 4500. c, Focal contacts (desmosomes) and a junctional complex is present at the site of cell-cell contact (D). Magnification, × 25 000. d, Exocytosis between two cells (E) and on the cell surface (E). Magnification, X 25 000. e, Mitochondria, shelflike type (M). Note microfilaments (F) and glycogen (G). Magnification, X 25 000. f, Scanning electron micrograph of AKN-1 cells, confirming their epithelial-like morphology. Note the presence of numerous naicrovilli on the surface of the cell facing the medium and neighboring cells (V). Magnification, X 2000.
mosomes or macula adherens underlying the plasma membrane with sites of cytoskeletal attachment and intracellular filaments near the junctional complexes. Microvilli are present (Fig. 1 c) and occasionally exocytosis was observed (Fig. 1 d). The mitochondria are focally prominent and characteristically contain shelf-like cristae (Fig. 1 e). The lysosomes swell rudimentarily and peroxisomes are observed in some cells; the cytoplasm is primarily filled with free polyribosomes. In addition, the cells contain glycogen (Fig. 1 e). Scanning electron microscopy (SEM). SEM also revealed an epithelial-like cell surface in this cell line and short microvilli decorated areas of the cell surface in approximately 90% of the ceils (Fig. 1
f). Immunohistochemistry. Immunohistochemical analyses of the AKN-1 cells revealed that they express a mixture of parenchymal hepatocyte markers as well as markers usually expressed by biliary epithelial cells (Fig. 2 a-f). The AKN-1 cells reacted with antibodies to alpha-l-antitrypsin (otl-AT), stem cell factor (SCF) and its receptor c-kit, and cytokeratins 8, 18, and 19, while showing no detectable reactions to the antibodies directed towards Factor 8 or albumin.
Almost all cells of the AKN-1 cell line expressed on the cell surface the human epithelial antigen (HEA) (Table 1) or an epithelial-related antigen reacting with MOC-31 (Fig. 2 f ) . The latter is a 40-kDa transmembrane glycoprotein present on most normal and malignant epithelia (20). Furthermore, we found that about 50% of the cells scored positive for the biliary epithelial marker 7-glutamyl transpeptidase (GGT, Table 1). The presence of glycogen already detected by TEM was additionally confirmed with periodic acid-Schiff staining. Expression of liver-specific mRNAs. RNA derived from normal human and rat livers scored positively for tyrosine aminotransferase (TAT) whereas AKN-1 ceils and both freshly isolated and 24-h cultured human hepatocytes did not (Fig. 3). Albumin is a specific marker for liver parenchymal ceils. Human liver and 24-h cultured human hepatocytes showed strong albumin mRNA expression. In contrast, the AKN-i cell line did not express albumin mRNA. Levels of argininosuccinate synthetase mRNA in AKN-1 cells were approximately equivalent to those of normal human liver, where this enzyme is most abundantly expressed.
193
HUMAN LIVER BILIAR EPITHELIAL-LIKE CELLS
FIG. 2. Immunohistochemical characteristics of the AKN-1 cell line. AKN-1 cells reacted with a, Antibodies to cd-antitrypsin, magnification, X 20; b, Stem cell factor, magnification, X 20; c, c-kit, magnification X 20; d, Cytokeratins 8, 18 (CAM 5,2), magnification × 20; e, Cytokeratin 19, magnification, X 20;f, Epithelial-related antigen (MOC-31), magnification, × 20.
Cytochrome P450 enzyme activity. To substantiate our hypothesis that the AKN-1 cell line may be useful for studying the metabolism of various hormones and xenobiotics, we analyzed for ECOD and EROD activity. We found a nearly twofold increase in CYP 450dependent deethylation of ethoxyresorufin using dexamethasone (1.28 pmoles/mg protein/min), and phenobarbital (1.35 pmnles/mg protein/min) as compared to unstimulated control cultures (0.663 pmoles/mg protein/rain). Furthermore, we found no EROD enzyme activity using 3-methylcholanthrene and no ECOD activity using the same inducer.
I
2
3
Q
411
w
4
5
• AS
.,, ALB
TABLE 1 SUMMARY OF IMMUNOHISTOCHEMICAL CHARACTERISTICS OF THE AKN-1 CELL LINE Antibody
Results -
MOC-31 HEA GGT Stem cell factor c-kit CK 19 CK 8, 18 p53 alpha-l-antitrypsin Glycogen Albumin Factor VIII RAg
+ + + + + + + + + + + + + + + + + + +
~+
= weakly, + +
= moderately, + + +
= strongly e x p r e s s e d , - -
= not e x p r e s s e d .
TAT
i
i
4 18S
FIG. 3. Northern blot analysis of liver-specific gene expression. Total RNA (20 lag) from cultured AKN-1 cells (lane 1), 24-h cultured human hepatocytes (lane 2), freshly isolated human hepatocytes (lane 3), normal human liver (lane 4), and normal rat liver (lane 5) was applied to each lane as indicated and subjected to Northern blot analysis as described in Materials and Methods. The cloned DNA probes encoded human albumin (ALB; 2.2 kb mRNA), rat argininosuccinate synthetase (AS; 1.67 kb mRNA), rat tyrosine aminotransferase (TAT; 2.4 kb mRNA), and rat t8S ribosomal RNA (1.87 RNA).
194
NUSSLER ET AL.
11 tt
,, I~
I 1
it ~'
|1 15
t9
2u
2 1
? .'
~,
,'
FIG. 4. Trypsin-Giemsa banded metaphase cell from the AKN-1 cell line sideline with trisomy 8: 47,XY,inv(2)(p21p25) dup(2)(p2lp25), + 8[cp4].
Karyotype analysis. The cell culture used for kalTotype analysis appears to have a mosaic abnormal karyotype including a stemline, 46,XY, inv(2)(p21p25)dup(2)(p21p25) [cp13] and two sidelines, one with the prior finding plus trisomy 8: 47,XY, inv(2)(p21p25)dup(2) (p21p25), + 8[cp4] and a near-tetraploid subpopulation with clonal karyotypic findings, including one extra copy of chromosomes Y, 4, 12, 20, and 22, the structurally abnormal chromosome 2 seen in the stemline and other sideline, and only three, rather than four chromosomes 4: 89-100(4n),XXYY,+Y[2],inv(2)(p21p25)dup(2) (p21p25)x2,-413], + 1212], + 2013], + 2212][cp5] (Fig. 4). The abnormal chromosome 2 appears to be a complex rearrangement in which a paracentric inversion of the short arm from band 2p21 to 2p25 occurred, followed by a duplication of the same region, also in inverted orientation, resulting in partial trisomy 2 (p25~p21). Microscopic evaluation of 600 metaphase cells for ploidy indicated that 77/600 or approximately 13% were near-tetraploid. No karyotypically normal cells were observed. Molecular cytogenetic analysis by fluorescence in situ hybridization (FISH) with centromeric probes to chromosomes 8 and 20 showed 60% of cells with a disomic chromosome pattern for these chromosomes, 10% of cells with three copies of chromosome 8 and two copies of chromosome 20, and 13% of cells scattered around the tetrasomic chromosome pattern. Control peripheral blood lymphocytes expressed two signals for each chromosome in 88% of ceils and loss of one signal in about 7% of cells. Thus, the molecular cytogenetic analysis confirms the classical
karyotyping and confirms that this cell culture consists of three abnormal karyotypically related, but distinct cell populations, Tumorigenicity. The routine morphologic analysis suggested that a neoplastic transformation had occurred in this epithelial-like cell line. Karyotype analysis confirmed this suspicion, so the cells were injected into nude mice to investigate whether tumor growth would occur. The injection of 106 cells into nude mice resulted in tumor growth within 14 d. The mean tumor size in 10 injected mice was 1.4 _+ 0.3 cm3. FACS-scan-analysis. Since this epithelial-like cell line grew in nude mice and expressed karyotypic abnormalities, we investigated the expression of the adhesion molecules, ICAM-1/CD54 and CD44std./CD44v6, which may have an important role in tumor cell invasion and metastasis. We found that ICAM-1/CD54 is expressed in low levels on the cell surface. In contrast, we found a high expression of CD44std. and its splice variant CD44var6 (9) (Fig, 5). p53-gene expression. Significant nuclear accumulation of p53 protein was observed (Table 1), which is generally considered to correspond frequently to the mutation of the tumor suppressor gene. However, a recent study of hepatocellular carcinoma showed that immunohistochemically detectable overaccumulation of p53 protein may occur in the absence of mutation (1). The AKN-1 cell line appears to fall into this category of tumors which show significant accumulation of p53 protein in the nucleus even though no mutations in exons 4-9 of the p53 gene were observed by SSCP analysis (data not shown),
HUMAN LIVER BILIAR EPITHELIAL-LIKE CELLS
Trp E (negative control)
ICAM-1
Data.011
FIG. 5. FACS-analysis of adhesion molecules constitutively expressed by AKN-1 cells. Detection of ICAM-1 and CD44std. was achieved with the monoclonal antibodies RR/1 and SSF2, respectively. The Trp E monoclonal antibody served as a negative control. The binding of the monoclonal antibodies (dilution 1/100) to constitutively expressed antigens was detected by an IgG purified goat antimouse FITC-labeled antibody.
CD44std.
Data.O03
oo 5 ~
g=
( b
~3 10
10
10
103
104
o ,o'
DISCUSSION The AKN-1 human hepatic epithelial-like cell line was established in long-term cell culture, consisting of a heterogeneous cell population and can be maintained in vitro in the absence of serum or growth factors. In the literature, several criteria are required to identify cells of epithelial origin. LM characterizes the cell shape and size; TEM determines the overall morphology and the formation of desmosomes which are present in almost all epithelial tissues in vivo but frequently lost in epithelial-like cell lines in vitro (17). The cell line described here fulfills most of the requirements to be identified as a cell of epithelial origin: the mosaic-like shape, the presence of desmosomes, and the expression of cytokeratins. We believe the AKN-1 cells to be of epithelial origin from our microscopy studies, their lack of Factor 8 expression (an endothelial marker) and the positive staining for the biliary epithelial marker CK 19, HEA, GGT, and the epithelial marker MOC 31. However, additional studies also showed some evidence of parenchymal origin. Markers which are consistent with the parenchymal hepatocyte origin of the ceils include the positive reaction with antibodies to otl-AT, cytokeratins 8 and 18 (6) and the presence of glycogen in the cytoplasm. The choice of ECOD and EROD as a measure of proper liver function was made because the methods assess different families of CYP activity. EROD assesses CYP 1A activity, whereas ECOD activity is catalyzed by a broader number of CYP isoforms. The major isoforms in human livers are CYP 3A proteins (14); thus, the positive EROD activity (although weak) seen after incubation with dexamethasone and phenobarbital indicates that the AKN-1 cell line has some CYP450 activity. Although differentiated parenchymal cells express albumin and tyrosine aminotransferase, AKN-1 cells did not express these markers. Additionally, these cells reacted with antibodies for the hemopoietic growth factor, SCF and its receptor, c-kit. The transformed nature of these cells, their expression of cytokeratins 8, 18, and 19, SCF, and weak expression of c-kit are consistent with the classification of hepatoceliular-like carcinoma (6). In the same line of evidence falls the observation that all human hepatocellular carcinoma cell lines express high levels of SCF but show little to no expression of the SCF receptor, c-kit (31). Since the cell line stained positive for SCF and grows without the addition of growth factors and serum, it is conceivable that the cell line is derived from a 'facultative' stem cell of the terminal biliary ductules. The origin of the terminal ductule cells is a matter of debate. Some investigators believe that it is derived from ceils of the larger interlobular ducts, whereas others claim it is of parenchymal origin (6,12). Human embryonic liver cells express CK 8, 18, and
~0O 101
102
103
195
1,~4
19, as does the AKN-1 cell line, and expression of these markers is increased during development of periportal hepatocytes. This layer of hepatocytes is referred to as the ductal plate (32). Considering that these ceils were originally isolated from a 10-yr-old boy, it is possible that they are precursor ceils of the ductal plate. This could explain the biliary epithelial-like origin as well as weak hepatocyte characteristics of the AKN-1 cell line. Although the AKN-1 cell line was originally isolated from a healthy organ donor, it became invasive and showed characteristics of tumor growth in nude mice. However, it could not be clearly identified as a hepatocellular carcinoma though it was highly invasive. We cannot provide evidence as to whether the original isolated ceils were invasive or the culture conditions led to the transformation. Since mutations in the tumor suppressor genep53 are known to occur rather frequently in hepatocellular carcinomas, predominantly in the 'hot spots of mutations' in exons 4-9 (34), we analyzed genomic DNA of AKN-1 cells by SSCP analysis. In contrast to the immunohistological staining where an accumulation of apparently wt-p53 was observed in the nucleus, no mutation in the p53 gene was detected. A recent report from Bourdon and coworkers described an accumulation of wt-p53 in 50% of hepatocellular carcinomas staining positively by immunohistochemistry (1). In these cases, a stabilization of the wt-p53 gene by complexation to mdm2 protein may be one of the reasons for nuclear accumulation of wt-p53 (24). In addition, it is well known that wt-p53 is able to induce growth arrest in transformed cells of various origin, principally in the G1 phase (25). Therefore, it is conceivable that in tumor ceils, the normal balance of growth regulatory signals is disturbed by the loss or mutation of wt-p53 gene. Under these circumstances, mutant p53 protein could bind and inactivate normal proteins, possibly resulting in cell transformation and genetic changes occurring during the stage of tumor progression. Several adhesion molecules, such as ICAM-1/CD54 and CD44std./CD44v6, are linked to tumor invasion and metastasis (5,9). The AKN-1 cell line constitutively expresses these adhesion molecules, which may also explain the tumor growth in vivo. The karyotypic analysis revealed several chromosomal breakpoints in this epithelial-like cell line. Whether genes at the breakpoints are rearranged is not detectable by these methods. One gene that has been mapped to 2p21 is hMSH1 (2p22-21), which is involved in DNA mismatch repair. No other genes that appear to be intriguing have been mapped to 2p21 or 2p25 to date. To our knowledge, no chromosome 2 abnormalities with similar breakpoints in liver cells have been reported (18). Interestingly, trisomy 20 has been reported in several cases of hepatoblastoma (16) that could also be the case in this cell line.
196
NUSSLER ET AL.
Since this cell line expresses, to some extent, hepatocyte-specific traits, we recently used it to study the inducible nitric oxide synthase gene. The AKN-1 cell line is one of the rare human cell lines to express the inducible nitric oxide synthase gene in response to inflammatory cytokines. The levels of nitric oxide synthase gene expression and nitric oxide production are similar to those in h u m a n hepatocytes (15). In addition, we have used the AKN-1 cell line to further characterize the transcriptional regulation of the h u m a n inducible nitric oxide synthase gene (3). CONCLUSION We have developed a h u m a n epithelial-like liver cell line consisting of a heterogeneous cell population, most likely of biliary origin. Since the AKN-1 cell line, cultured in a defined medium without supplementation (e.g., growth factors), retains many liver-specific functions, the use of this novel tumor cell line may permit the control and examination of variables normally present in vivo. Therefore, it is a potentially attractive system that may be useful for examining the direct effects as well as the metabolism of hormones, drugs, xenobiotics, and liver carcinogens. In addition, this cell line could serve as a model for studying tumors in vivo after transplantation into nude mice, thus allowing the assessment of effectiveness of new anti-tumor therapies. ACKNOWLEDGMENTS The authors are grateful to Mr. W. Swaney for excellent technical assistance with the classical and molecular cytogenetic studies which were carried out in the UPCI Cytogenetics Facility. The present study was partially supported by NIH grants P30-CA 47904 (S. M. G.), GM 50897 (S. M. M.), GM-44100, and Klinikft~rderung of the University of Ulna P-229 and P-368 (A. K. N.)
REFERENCES 1. Bourdon, J. C.; D'En-ico, A.; Paterlini, E; Grigioni, W.; May, E.; Debuire, B. p53 protein accumulation in European hepatocellular carcinoma is not always dependent on p53 gene mutation. Gastroenterology 108:1176-1182; 1995. 2. Caraceni, P.; Gasbarrini, A.; Nussler, A. K.; Di Silvio, M.; Bartoli, F.; Borle, A. B.; Van Thiel, D. H. Human hepatoeytes are more resistant than rat hepatocytcs to anoxia-reoxygenation injury. Hepatology 20:1247-1255; 1994. 3. De Vera, M. E.; Shapiro, R. A.; Nussler, A. K.; Mudgett, J. S.; Morris, S. M., Jr.; Simmons, R. L.; Billiar, T. R.; Geller, D. A. Transcriptional regulation of the human inducible nitric oxide synthase (NOS2) gene by cytokines: initial analysis of human NOS2 promoter. Proc. Natl. Acad. Sci. USA 93:1054-1058; 1996. 4. Dorko, K.; Freeswick, E D.; Bartoli, F.; Ciealese, L.; Bardsley, B. A.; Tzakis, A.; Nussler, A. K. A new technique for isolating and culturing human hepatocytes from whole or split livers not used for transplantation. Cell Transplant. 3:387-395; 1994. 5. Gansauge, E; Gansauge, S.; Zobywalski, A.; Scharnweber, C.; Link, K. H.; Nussler, A. K.; Beger, H. G. Differential expression of CD44 splice variants in human pancreatic adenocarcinoma and normal pancreas. Cancer Res. 55:5499-5503; 1995. 6. Gerber, M. A.; Thung, S. N. Cell lineages in human liver development, regeneration, and transformation. In: Sirica, A. E., ed. The role of cell types in hepatocarcinogenesis. Boca Raton: CRC Press; 1992:209226. 7. Granger, T.; Guenet, C.; Dietrich, J. B.; Chasserot, S.; Fromont, M.; Befort, N.; Jami, J.; Beck, G.; Pictet, R. Complete complementary DNA of rat tyrosine aminotransferase messenger RNA. J. Mol. Biol. 184:347-350; 1985.
8. Guguen-GuiUouzo, C.; Clement, B.; Baffet, G.; Beaumont, C.; MorelChany, E.; Glaise, D. Maintenance and reversibility of active albumin secretion by adult rat hepatocytes co-cultured with another liver epithelial cell type. Exp. Cell Res. 143:47-54; 1983. 9. Gtinthert, U.; Hofman, M.; Rudy, W.; Rebel, S.; Z~ller, M.; Haussmann, I.; Matzku, S.; Wenzel, A.; Ponta, H.; Herrlich, P. A new variant of glycoprotein CD44 confers metastatic potential to rat carcinoma ceils. Cell 65:13-24; 1991. 10. ISCN Guidelines for Cancer Cytogenetics, Supplement to An International System for Human Cytogenetic Nomenclature. In: Mitelman, E, ed. Basel: S. Karger; 1983. 11. Isom, H. C.; Georgoff, I. Quantitative assay for albumin-producing liver cells after simian virus 40 transformation of rat hepatocytes maintained in chemically defined medium. Proc. Natl. Acad. Sci. USA 81:6378-6382; 1984. 12. Jones, A. L.; Spring-Mills, E. The liver and gallbladder. In: Weiss, L., ed. Cell and tissue biology. Berlin, New York: Urban und Schwarzenberg; 1988:687-714. 13. Katz, R. A.; Mitsialis, S. A.; Guntaka, R. V. Studies on methylation of avian sarcoma proviruses in permissive and non-permissive cells. J. Gen. Virol. 64:429-435; 1983. 14. Lubinski, J.; Flint, O. P.; Durham, S. K. In vitro induction by phenobarbital and nmthylcholanthrene measured in an automated 24-well plate assay for cytochrome P450-dependent activity (pentoxyresorufin Odepentylase and ethoxyresorufin O-deethylase. In Vitro Toxicol. 7:1323; 1994. 15. Luss, H.; Nussler, N. C.; Beget, H.-G.; Nussler, A. K. Expression and detection of inducible nitric oxide synthase in experimental models of inflammation. Methods (Orlando) 10:51-56; 1996. 16. Mascarello, J. T.; Jones, M. C.; Kadota, E R.; Krous, H. N. E Hepatoblastoma characterized by trisomy 20 and double minutes. Cancer Genet. Cytogenet. 47:243-247; 1990. 17. Mayer, D.; Schafer, B. Biochemical and morphological characterization of glycogen-storing epithelial liver cell lines. Exp. Cell Res. 138:114; 1982. 18. Mitelman, E Catalog of chromosome aberrations in cancer. 14th ed. New York: Wiley-Liss; 1991. 19. Morin, O.; Normand, C. Long-term maintenance of hepatocyte functional activity in co-culture: requirements for sinusoidal endothelial ceils and dexamethasone. J. Cell Physiol. 129:103-110; 1986. 20. Morris, S. M., Jr.; Moncman, C. L.; Holub. J. S.; Hod, Y. Nutritional and hormonal regulation of mRNA abundance for arginine biosynthetic enzymes in kidney. Arch. Biochem. Biophys. 273:230-237; 1989. 21. Nussler, A. K.; Di Silvio, M.; Billiar, T. R.; Hoffman, R. A.; Geller, D. A.; Selby, R.; Madariaga, J.; Simmons, R. L. Stimulation of the nitric oxide synthase pathway ira human hepatocytes by cytokines and endotoxin. J. Exp. Med. 176:261-264; 1992. 22. Nussler, A. K.; Di Silvio, M.; Liu, Z. Z.; Geller, D. A.; Freeswick, P.; Dorko, K.; Bartoli, E; Billiar, T. R. Further characterization and comparison of inducible nitric oxide synthase in mouse, rat and human hepatocytes. Hepatology 21:1552-1560; 1995. 23. Nussler, A. K.; Liu, Z. Z.; Di Silvio, M.; Sweetland, M. A.; Geller, D. A.; Lancaster, J. R., Jr.; Billiar, T. R.; Freeswick, P. D.; Lowenstein, C. L.; Simmons, R. L. Hepatocyte inducible nitric oxide synthesis is influenced in vitro by cell density. Am. J. Physiol. 267:C394-C401; 1994. 24. Oliner, J. D.; Pietenpol, J. A,; Thiagalingam, S.; Gyuris, J.; Kinzler, K. W.; Vogelstein, B. Oncoprotein MDM2 conceals the activation domain of tumour suppressor p53. Nature 362:857-860; 1993. 25. Oren, M. The involvement of oncogenes and tumor suppressor genes in the control of apoptosis. Cancer Metastasis Rev. 11:141-147; 1992. 26. Orita, M.; Suzuki, Y.: Sekiya, T.; Hayashi, K. Rapid and sensitive detection of point mutations and DNA polymorphism using the polymerase chain reaction. Genomics 5:874-879; 1989. 27. Roberts, E. A.; Letarte, M.: Squire, J.; Yang, S. Characterization of human hepatocyte lines derived from normal liver tissue. Hepatology 19:1390-1399; 1994. 28. Ruittenbeek, T. Immunocytology of body fluids. MOC-31, a monoclonal antibody discriminating between mesothelial and epithelial cells. Arch. Pathol. Lab. Med. 118:265-271: 1994.
HUMAN LIVER BILIAR EPITHELIAL-LIKE CELLS 29. Sambrook, J.; Fritsch, E. F.; Maniatis, T. Isolation of high-molecularweight DNA from mammalian cells. In: Molecular cloning. 2nd ed. Cold Spring Harbor, NY: Cold Spring Harbor Laborato~7 Press; 1989:9.14-9.23. 30. Toguchida, J.; Yamaguchi, T.; Meyers, P. A.; Little, J. B.; Sasaki, M. S.; Weichselbaum, R. R.; Yandell, D. W. Prevalence and spectrum of germline mutations of the p53 gene among patients with sarcoma. N. Engl. J. Med. 326:1301-1308; 1992. 31. Toyota, M.; Hinoda, Y.; Takaoka, A.; Makiguchi, Y.; Takahashi, T.; Itoh, R.; Imai, K.; Yachi, A. Expression of c-kit and kit ligand in human colon carcinoma cells. Tumor Biol. 14:295-302: 1993.
197
32. Van Eyken, P.; Sciot, R.; Callea, E; Van der Steenk, H.; Moerman, P.; Desmet, V. J. The development of the intrahepatic bile duct in man: a keratin-immunohistochemical study. Hepatology 8:1586-1595; 1988. 33. Vergani, G.; Fri~sch, D. Morphologie epitheloider Rattenleberzellen in Kuhur. Verh. Anat. Ges. 81:621-622; 1987. 34. Volkmann, M.; Mtillel, M.; Hofinann, W. J.; Meyer, M.; Hagelstein, J.; Rath, 04 Kommerell, B.; Zentgraf, H.: Galle, R R. The humoral immune response to p53 in patients with hepatocellular carcinoma is specific for malignancy and independent of the alpha-fetoprotein status. Hepatology 18:559-565; 1993.
