Characteristics of Cell Lines Established from Human Gastric CarcinomaJae-Gahb Park,1 Harold Frucht, Renato V. LaRocca, David P. Bliss, Jr., Yukio Kurita, Tchaw-Ren Chen,Jerry G. Henslee, Jane B. Trepel, Robert T. Jensen, Bruce E. Johnson, Young-Jue Bang, Jin-Pok Kim,and Adi F. GazdarCancer Research Institute, College of Medicine, Seoul National University, Seoul Ì10-744,Korea fJ-G. P., Y-J. B., J-P. K.J; NCI-Navy Medical Oncology Branch¡J-G.P., D. P. B., Y. K., J. B. T., B. E. J., A. F. G.J and Medicine Branch [R. V. L.J, National Cancer Institute, and National Institute of Diabetes, Digestive andKidney Diseases (H. F., R. T. J.¡,Bethesda, Maryland 20892; American Type Culture Collection. Rockville, Maryland 20853 [T-R. C.]; and Abbati Laboratories, NorthChicago, Illinois 60064 ¡J.G. H.J
ABSTRACT
We report the establishment and characterization of four continuouscell lines derived from human primary and metastatic gastric carcinomas,and we compare their properties with a panel of colorectal carcinoma celllines previously established and reported by us.
Our success rate in culturing gastric carcinomas was relatively low,especially from primary tumors, compared to colorectal carcinoma. Theseobservations may reflect the relatively modest number of gastric carcinoma cell lines established (mainly from Japan), compared to the abundance of colorectal carcinoma lines established worldwide.
All four gastric lines expressed the surface glycoproteins carcinoem-bryonic antigen and TAG-72 and three lines expressed CA 19-9. Two ofthe lines expressed aromatic amino acid decarboxylase but lacked othermarkers for neuroendocrine differentiation. All four lines were positivefor vasoactive intestinal peptide receptors but lacked gastrin receptors.In addition, two lines expressed receptors for muscarinic/cholinergicreceptors but not /9-adrenergic receptors.
Cytogenetic evidence for gene amplification was present in the celllines. All four lines contained varying numbers of double-minute chromosomes. One line, SNU-16, was amplified for the c-myc proto-oncogeneand contained four homogeneously staining regions. While c-myc and c-erb-B-2 RNA were expressed by all lines, there was no evidence ofamplification or overexpression of several other proto-oncogenes andgrowth factors.
The multiple properties we have described in our gastric carcinomacell lines are remarkably similar to those found in the panel of colorectalcarcinoma cell lines. These properties include morphology, growth characteristics, expression of surface glycoproteins, partial expression ofneuroendocrine cell markers, frequent chromosomal evidence of geneamplification, and occasional amplification of the c-myc proto-oncogene.Our four well characterized cell lines should provide useful additions tothe modest number currently available for in vitro studies of gastriccarcinoma.
INTRODUCTION
Cancer of the stomach is still a leading malignant disease inmany countries, including Korea, Japan, eastern Europe, Iceland, and South Africa (1,2). While the incidence in the UnitedStates is declining, stomach cancer is the sixth most commoncause of cancer deaths and still remains a major health problem(3). Thus far, the only treatment that results in significant cureof gastric cancer is well planned, aggressive, subtotal or totalgastrectomy (4). Because one third of patients will present withunresectable disease and another one third of patients willdevelop local or distant recurrence following curative surgery,other modalities of therapy have been studied in attempts toimprove treatment of widespread cancer and to prevent relapsein patients with resectable disease.
Cell lines established from human gastric cancers may pro-
Received8/8/89; revised11/30/89.The costs of publication of this article were defrayed in part by the payment
of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1To whom requests for reprints should be addressed, at Department of
Surgery, College of Medicine, Seoul National University and Seoul NationalUniversity Hospital. Seoul 110-744, Korea.
vide useful tools to study the biology of the disease and todevelop and test new therapeutic approaches. A large bank ofwell characterized cell lines should reflect the diversity of tumorphenotypes and provide adequate models for the study of tumorheterogeneity. While many well characterized cell lines havebeen established from colorectal carcinomas by several laboratories, including ours (5-7), gastric carcinoma has been muchmore difficult to culture. A modest number of gastric carcinomalines have been described, mostly from Japan (8-12). A fewlines have been established in other countries, including theUnited States (13-19). Herein we describe the establishmentand characterization of four gastric carcinoma cell lines, threefrom Korean patients and one from an American, and comparethese properties to a panel of colorectal carcinoma cell linespreviously described by us (6).
MATERIALS AND METHODS
Cell Culture. Cell lines were established from pathologically provengastric tumors, either directly or after heterotransplantation in athymicnude mice. Solid tumors were finely minced with scissors and dissociated into small aggregates by pipetting. For culture of primary tumors,invasive areas were selected from the serosal surface, whenever possible,to decrease the chance of microbial contamination. Ascitic fluids werecollected, pelleted, washed, and resuspended in growth medium. Approximately 1-5 x 10' cells were seeded into 25-cm2 flasks. Tumorswere cultured in RPMI 1640 supplemented with 10% heat-inactivatedfetal bovine serum (RIO). Initially, cultures were passaged whenevervigorous tumor cell growth was observed. Subsequent passages wereperformed weekly. Nonadherent cultures were passed by transfer orfloating multicellular aggregates. Adherent cultures were passaged atsubconfluence after trypsinization. If stromal cell growth was noted ininitial cultures, differential trypsinization (5) was used to obtain a puretumor cell population. Media and sera were obtained from Grand IslandBiological Co. (Grand island, NY). Cultures were maintained in humidified incubators at 37°Cin an atmosphere of 5% CO2 and 95% air.
Growth Characteristics. Population doubling times were determinedby seeding 0.5-3 x 10' viable cells into replicate 25-cm2 flasks and
performing counts daily for 14 or more days. Cultures were fed every3 or 4 days and 24 h prior to counting. For plating efficiencies, 1 x IO3cells were plated in five replicate 100-mm dishes, and colonies consisting of more than 50 cells were enumerated 21-30 days later, afterstaining with 0.5% crystal violet. For determining colony-formingefficiency in semisolid medium, 2 x 10s viable single cells were plated
in 3 ml of RIO medium containing 0.3% agarose, over a base layer ofRIO medium containing 0.5% agarose, in five duplicate 60-mm dishes.Tumorigenicity was tested by inoculating 5-10 x IO6cells s.c. into each
of five male athymic nude mice, BALB/c background, and observingthem twice weekly for progressive tumor growth. Histological examination was performed on tumors so obtained.
Cell Line Characterization. Saccomanno fluid-fixed Cytospin preparations of floating cultures and trypsinized adherent cultures and paraffin-embedded sections of xenografts were stained with hematoxylin-eosin, alcian blue, and mucicarmine.
For ultrastructural studies, cell pellets were fixed in 2.5% glutaral-dehyde, postfixed in I % osmium tetroxide, and stained with 1% uranylacetate, and thin sections were examined by an electron microscope.
Chromosome slides were prepared by the standard air-dried methodfrom exponentially growing cultures, after colcemid treatment at approximately 0.1 Mg/ml final concentration for 45 min. They werestained with the fast G-banding (20) for an overall karyotypic analysisand with Q- and C-bandings (21) to obtain complementary evidencefor the identification of specific chromosomes and chromosome segments. Chromosome number distribution was evaluated from 50 me-taphases, and the rate of higher ploidies (as compared with the modalchromosome number) was estimated from 500 metaphases.
Aromatic amino acid decarboxylase (EC 4.1.1.28), also known as L-DOPA decarboxylase, was assayed as previously described (22, 23).Values are reported in units (1 unit = 1 nmol CO2 released/h/mgsoluble protein).
All cell lines were immunophenotyped by indirect immunofluores-cence and flow microfluorometry, using a FACS II flow cytometer(Becton Dickinson, Mountain View, CA) equipped with an argon laser,utilizing the 488-nm line at 0.5-mW power. Monclonal antibodies usedwere Bl (CD20, pan-B-cell; Coulter, Hialeah, FL), B2 (CD21, CR2receptor; Coulter), B4 (CD 19, pan-B; Coulter), Tl l (CD2, pan-T cell;Ortho, Raritan, NJ), Leu-9 (CD7, pan-T-cell; Becton Dickinson), T9(transferrin receptor; Ortho), and the natural killer cell markers Leu-7,Leu-15 (CDllb) (Becton Dickinson), and NKH1 (Coulter). A viablesingle-cell suspension was prepared by incubation of the cells with amixture of collagenase I (140 mg/100 ml; Calbiochem, La Jolla, CA)and DNAse I (30 mg/100 ml; Calbiochem) in phosphate-bufferedsaline, for 30 min at 37°C,with gentle stirring. The cells were washed
in saline and resuspended at a concentration of Id nil in phosphate-buffered saline with I % bovine serum albumin and 0.1 % sodium azide(PBA), for incubation with primary antibody, using 0.5 x IO6 cells/
marker. Following incubation for 30 min with primary antibody, thecells were washed with PBA. incubated with fluorescein-conjugatedgoat anti-mouse immunoglobulin (Coulter) for 30 min at 4°C,washed,
and resuspended in PBA for analysis.Mycoplasma contamination was tested for by direct culture in agar,
the Hoechst stain method, and the use of a ribosomal RNA hybridization method (Gen-Probe, San Diego, CA). All lines were tested formurine virus contamination by the mouse antibody production tests(24). Cell homogenates were tested for the human forms of the followingenzymes by starch gel (25), using the Authentikit system (CorningScience Products, East Walpole, MA): purine nucleoside phosphorylase(EC 2.4.2.1), glucose-6-phosphate dehydrogenase (EC 1.1.1.49), pep-
tidase B (EC 3.4.11.4), and láclatedehydrogenase (EC 1.1.1.27).Antigen Expression and Secretion. Expression and secretion of CEA2
(26, 27), CA 19-9 (28), and TAG-72 (29, 30) surface antigens weremeasured as follows. Cells (1 x IO6)were seeded into replicate 75-cm2
flasks (13 ml total volume) and washed 3 times when semiconfluent.Two days later, cells from one flask were enumerated after trypsiniza-tion and discarded. Cells from the other flask were harvested using arubber policeman, washed 3 times, resuspended in 1% Nonidet P-40with 100 M phenylmethylsulfonyl fluoride (approximately 5-10 ngprotein/ml), sonicated twice on ice for 15 s, and stored at —70°Cuntil
assayed. Supernatant fluids from the same flask were collected, clarified, and stored frozen until assayed.
CEA was measured by the Abbott CEA-EIA monoclonal assay(Abbott Laboratories, North Chicago, IL). CA 19-9, a sialylated LewisA mucin antigen, was determined by the Centocor CA 19-9 radio-immunoassay (Centocor, Malvern, PA). TAG-72 was measured in ahomologous sandwich enzyme immunoassay, using the B72.3 monoclonal antibody in a ink-rot her plate assay configuration.
Probes. The probes used for DNA and RNA blots were c-myc, aCla\-EcoRl fragment containing the third exon (31); N-m>'c, an EcoRl-BamHl fragment containing the second exon (32): L-myc, a Sma\-£coRIfragment 3' to the gene containing the homology region to the
second exon of c-myc and N-myc (33); c-myb, an £coRI fragment
(probe 11,821-3350 nucleotides, from W. M. Kuehl) (34); c-sis (obtainedfrom ONCOR); IGF-2, 780-base paircDNA clone of rat IGF-2 insertedinto the Pstl site of pUC 12 (ampicillin resistant, 2.7 kilobases) (obtained from M. Rechler, National Institutes of Health); EGF-R, 2.4-kilobase cDNA fragment of the human EGF receptor (obtained fromI. Pastan, National Cancer Institute); neu, cDNA fragment pHER2-436-3 cloned into the EcoRl site of pUC 12 (obtained from A. Ullrich);chromogranin A, full length cDNA fragment (obtained from L. Hel-man, National Cancer Institute); and gastric releasing peptide, 900-base pair insert cloned into pBR 322 (obtained from J. Battey, NationalCancer Institute).
DNA Blot Hybridization. The cell lines were examined for rearrangements or amplifications of the c-myc, N-myc, L-myc, c-myb,and EGF receptor genes by Southern blot analysis (35). Cells wereharvested during log phase growth and DNA was prepared (36), digested with the £coRI restriction endonuclease. electrophoresed in0.8% agarose gels, and transferred to nitrocellulose filters. The DNAwas hybridized to "P-labeled purified DNA fragments, using the probes
described above. The blots were washed and exposed to radiographiefilm for periods varying from 2 to 10 days.
RNA Expression. Thirty (¿gof total RNA from the four gastriccarcinoma cell lines as well as appropriate controls were electrophoresed in a 1% agarose-formaldehyde gel and transferred to nylon filters(Nytran). The probes used were labeled with "P using nick translation.
Hybridization of the Northern blots with these probes was performedin 50% formamide, 5x standard saline citrate, 2x Denhardts, 0.1%sodium dodecyl sulfate, at 42°Cfor periods varying from 24 to 48 h.Washings were done in 0.1 x standard saline citrate at 65'C.
Detection of Surface Receptors by Radioligand Binding. ' I HII ( ( K8(2200Ci/mmol). '"I-BH-SP (2200 Ci/mmol), '"I-gastrin 17-1 (2200Ci/mmol), '"I-VIP (2200 Ci/mmol), '"I-OH-BZP (2200 Ci/mmol),[3H]-NMS (70-87 Ci/mmol), and tyrosyl-[3,5-3H]enkephalin (30-50
Ci/mmol) were obtained from New England Nuclear (Boston, MA).125I-Tyr4-bombesin (2200 Ci/mmol), '"I-BH-NMB (2200 Ci/mmol),and '"I-CGRP (2200 Ci/mmol) were prepared using the methodsdescribed previously (37-39). Cholecystokinin octapeptide was obtained from Research Plus (Bayonne, NJ); Tyr4-bombesin was obtainedfrom Bachern Inc. (Terranee, CA); gastrin-I, neuromedin B, VIP,secretin, substance P, and CGRP were obtained from Peninsula Laboratories (Belmont, CA); pindolol, naloxone hydrochloride, and atropinewere obtained from Sigma (St. Louis, MO).
Cells were washed and resuspended in 50 mM Tris buffer, consistingof 130 mM NaCl, 7.7 HIM KC1, 5 HIM MgCl2, 0.1% bovine serumalbumin, 1 min ethylene glycol bis(/3-aminoethyl etìter)-NJVJV'JV'-
tetraacetic acid, 0.1% bacitracin, 4 ng/ml leupeptin, and 2 ng/mlchymostatin, at pH 7.4. Cells at a concentration of 15 x IO6cells/ml(range, 5-25 x IO6cells/ml), in a volume of 0.5 ml were incubated withradioligand at a concentration of 50 pM for '"I-BH-CCK-8, 125I-Tyr4-bombesin, '"I-BH-NMB, '"I-VIP, and '"I-OH-BZP; 100 pM for '"I-gastrin 17-1; 125 pM for '"I-BH-SP and '"I-CGRP; 0.6 nM for [3H]-NMS; and 1 nM for tyrosyl-j'HJenkephalin, for 30 min at 22°C.After
incubation with iodinated radioligand, 0.133 ml of cells were sampled,spun down through cell buffer containing 4% bovine serum albumin ina Beckman Microfuge, and then washed twice, and the pellet wascounted in an Auto-Gamma 5000 Series gamma counter (PackardInstrument Corp., Sterling, VA). When tritiated radioligand was used,cells were sampled, washed over glass microfiber filters (WhatmanInternational Ltd., Maidstone, England), and then counted in a liquidscintillation spectrometer (Packard Instruments Corp., Sterling, VA).Nonsaturable binding was the amount of radioactivity associated withthe cells when the incubation contained radioligand plus the unlabeledpeptide in concentrations of 1-10 MM.
RESULTS
Establishment of Cell Lines. We attempted to culture 30primary and 12 metastatic gastric carcinomas. With one exception (see below), all specimens were from Korean patients. Fourgastric carcinoma cell lines were established from one primaryand three metastatic tumors. Thus, our overall success rates
were 3% from primary tumors and 25% from metastatic tumors.
One cell line, NCI-N87, was established from a liver metastasis of a gastric carcinoma arising in an American patient. Itwas passaged as a xenograft in athymic nude mice for threepassages before establishing it as a cell line. The remaininglines were established from Korean patients (Table 1). Three ofthe lines were from patients who had not received prior cyto-toxic therapy, while one line was from a patient who hadpreviously received chemotherapy including 5-fluorouracil,doxorubicin, and mitomycin-C.
Cell Line Characteristics. Cell line NCI-N87 grew as anadherent monolayer consisting of tightly knit epithelial cells(Table 2, Fig. 1). The other three cultures displayed bothadherent and floating subpopulations (Table 1), and the cellsappeared more rounded (Fig. 1).
Cell lines NCI-N87 and SNU-16 were tumorigenic inathymic nude mice (the other two lines were not tested). Population doubling times ranged from 26 to 47 h. Cloning efficiencies ranged from 1.9 to 27% (Table 2). All lines expressedhuman forms of the four enzymes tested and were free ofcontamination with Mycoplasma or murine viruses.
Based on culture, xenograft, and ultrastructural morphology,cell line NCI-N87 was classified as well differentiated and theother three lines as poorly differentiated (Table 2, Figs. 1 and2). These morphologies were identical to the degree of differentiation of the original tumor. However, one poorly differentiated cell line (SNU-16) formed small numbers of goblet cellsin culture (Fig. ID).
Expression and Secretion of Antigens. We studied expressionand secretion of antigens associated with gastrointestinal cellsand their tumors. As displayed in Fig. 3, all three antigens werefound in cell pellets of the gastric lines at levels comparable tothose present in colorectal carcinoma lines, although the levelsof CA 19-9 demonstrated considerable variation between gastriccell lines. Except for TAG-72, which is not generally secreted,the antigens were also detected in supernatant fluids (data notshown). Only one gastric carcinoma cell line (SNU-1) haddetectable concentrations of TAG-72 antigen in its supernatantfluid (900 ng/106 cells). As with colorectal cell lines, there was
no correlation between degree of morphological expression bythe gastric lines and antigen concentrations.
NE Cell Features. We assayed for DDC enzymatic activity,
" A, adherent; F/A, floating and adherent subpopulations.* The partially floating SNU cell lines were tested in semisolid medium, while
the plating efficiency of the adherent cell line (NCI-N87) was determined.c NT, not tested.
an enzyme characteristic of NE cells. Relatively high levels ofenzyme activity (87-89 units/mg protein) were detected in twolines (SNU-5 and SNU-16), both initiated from ascitic fluids.However, expression of neuroendocrine markers by these lineswas partial, because all lines lacked other evidence of neuroendocrine differentiation (dense core granules, chromogranin ARNA, and the natural killer cell antigens Leu-7 and NKH1).
Receptor Binding Studies. Saturable binding of I25I-VIP wasseen in SNU-1, SNU-5, and SNU-16 but was minimally detectable in NCI-N87 (Table 3). With 50 pM 125I-VIP, the
saturable binding as a percentage of the total binding was 8, 54,and 36% in SNU-1, SNU-5, and SNU-16, respectively. A smallbut appreciable amount of saturable binding of [3H]-NMS wasseen in SNU-5 and NCI-N87 but not in SNU-1 or SNU-16(Table 3). With 0.6 nM [3H]-NMS, the saturable binding as a
percentage of the total binding was 26% and 29% in cell linesSNU-5 and NCI-N87, respectively. There was no saturablebinding of the other radioligands (Table 3).
To further examine the interaction of 125I-VIPwith receptorson one of these cell lines, SNU-1, we studied the ability of VIPand secretin to inhibit binding of 125I-VIP. VIP caused detectable inhibition at 0.3 nM, half-maximal inhibition at 6.3 nM,and complete inhibition at 0.3 MM(Fig. 4). Secretin, althoughstructurally related to VIP, failed to inhibit binding of 125I-VIP
at concentrations of 0.1 /¿M,demonstrating it was at least 3000times less potent than VIP and that I25I-VIP was in fact binding
to a VIP receptor and not to a secretin receptor. When the datawere analyzed using a nonlinear least squares curve-fittingprogram, LIGAND (40), the inhibition of binding of 125I-VIPby VIP was best fit with a one-receptor model having a highaffinity for VIP (KA= 2.5 nM).
Gene Amplification and Expression. In one gastric cell line,SNU-16, an intense signal corresponding to the 12.7-kilobaseEcoRl c-myc fragment was detected (Fig. 5). We estimated thatthe signal from this line was about 50-fold greater than thegerm line level. No evidence of amplification or rearrangementswas noted with the N-myc, L-myc, myb, and EGF receptorgenes.
All of the gastric cell lines expressed levels of c-myc and erb-B RNA that were comparable to the levels detected in a panelof 11 colorectal carcinoma cell lines (data not shown). Ofinterest, the level of c-myc RNA expressed by the SNU-16 cellline was comparable to those expressed by the other three lines.No expression of the following genes by any of the cell lineswas detected: N-wyc, L-myc, c-sis, IGF-2, and gastrin releasingpeptide.
Cytogenetic Studies. Each cell line had unique modal kary-otypic characteristics, indicating its independent origin. Numerical data on the chromosome characteristics of four celllines are summarized in Table 4. SNU-1 and NCI-N87 hadnear-diploid and the other two lines had near-tetraploid modalchromosome numbers. Of interest, both near-tetraploid lineswere initiated from malignant ascites fluids. The rate of cellswith higher ploidies, as compared with the modal stem linechromosome numbers, was high in all but SNU-16. Althoughthe bimodal distribution within the modal chromosome numberranges (i.e., excluding the higher ploidies) was obvious only inSNU-16, two distinct sublines (in each cell line) could beidentified by chromosome band analysis (Fig. 6). Each sublinehad a unique and consistent subset of chromosomes that distinguished it. Generally, these differences totaled less than 10%of the chromosome complements. DMs were found in 12-64%of the cells of all four cell lines (Table 4). However, only oneor two DMs/cell were seen in all lines except SNU-16, and
Fig. 1. Morphology of gastric carcinoma cell cultures. The well differentiated cell line HCI-N87 (II demonstrates substrate adherence and grows as islands oftightly cohesive epithelial cells. The poorly differentiated cell lines SNU-1 (A), SNU-5 (C), and SNU-16 (D) demonstrate only partial substrate attachment and havefloating or loosely attached subpopulations. Phase contrast micrographs.
only a portion of cells in all lines possessed this aberration. InSNU-16, 4 to over 20 DMs were detected in some cells in bothsublines (Fig. 6).
HSRs were present in the minor subline of SNU-16. Thissubline had HSRs occurring at four separate chromosomallocations (6q, 1Ip, and two unidentified marker chromosomes).Up to three HSRs were present in individual cells (Fig. 6). Thelip HSR was the most common of the HSR chromosomes,whereas 9q HSR occurred only once. A more detailed accountof the HSRs will be reported elsewhere.
DISCUSSION
In this report we describe the establishment and characterization of four continuous cell lines derived from human primaryand metastatic gastric carcinomas. Our success rate was relatively low for gastric carcinomas, especially from primary tumors, compared to our success with colorectal carcinomas(about 40% from both primary and metastatic tumors) (6).3
These findings reflect the relative paucity of gastric carcinomalines established worldwide, compared to the abundance ofcolorectal carcinoma lines. Of interest, we have recently reported that the success rate of culturing non-small cell carcinomas is much higher with metastatic tumors than with primarytumors and that successful culture may be a negative prognosticfactor (41).
3A. F. Gazdar, J-G. Park, and H. Oie, unpublished data.
While a modest number of gastric carcinoma lines have beeninitiated by other investigators, mostly from Japan, our celllines express several interesting or unique features, including(a) expression and secretion of three gastrointestinal cell associated antigens; (b) frequent expression of the NE cell markerDDC; and (c) expression of vasoactive peptide and muscarinic/cholinergic receptors. We have also characterized a possiblyunique cell line with c-myc gene amplification and having asubline with HSRs at four separate chromosomel locations.Because we are not aware of any comprehensive review of theproperties of gastric carcinoma cell lines established from Oriental or Western patients, we cannot directly compare thecharacteristics of our panel of cell lines with those from othersources.
We investigated production and secretion of three membrane-bound glycoproteins associated with gastrointestinal cells (26-30). CEA is present on normal and malignant colonie tissuesas well as other cancers. CA 19-9, originally identified from acolon carcinoma line, is a sialylated lacto-/V-fucopentaose II, anoligosaccharide related to Lewis blood group substances. TAG-72, a high molecular weight glycoprotein, is expressed by manyepithelial malignancies including colon, breast, and ovary. Itsexpression appears to be influenced by the spatial configurationof the tumor cells, and it is reported to be rarely expressed inmonolayer cultures. Of interest, all our gastric lines expressedTAG-72 antigen, although the concentration was lowest in thewell differentiated NCI-N87 cell line demonstrating the greatestdegree of substrate adherence. All four lines also expressed
Fig. 2. Ultrastructural morphology of gastric carcinoma cell lines. A, well differentiated cell line NCI-N87 (x 3000). Tight junctions (arrows) and short microvilliare present on the free surfaces. Lysosomal granules are present in the cytoplasm. These features are suggestive of the intestinal type of gastric carcinoma. B-D,poorly differentiated cell lines SNU-1 (B) (x 3500). SNU-5 (C) (x 6000). and SNU-16 (D) (x 8000). The cells of these lines are loosely coherent and containmicrovilli which vary in number and length. Bundles of intermediate filaments are present in SNU-5. While most of the cells of SNU-16 were relatively undifferentiated,varying amounts of mucin were present. Occasional cells demonstrated a goblet cell morphology (as illustrated), with eccentric nuclei and cytoplasm filled with mucindroplets.
CEA, and three expressed CA 19-9. Lines expressing CEA andCA 19-9 actively secreted these antigens into the supernatantfluids, while only modest concentrations of TAG-72 antigenwere present in the supernatant fluids of one cell line. TAG-72is usually not secreted by expressor cells, while the other twoantigens are usually actively secreted. As previously reported(6) and displayed herein, the results of secretion of these antigens by the colorectal cell panel were similar, although thelevels of TAG-72 expression in three of four gastric lines werehigher than in the colorectal lines.
Two of our gastric cell lines expressed varying concentrationsof DDC, a key NE cell marker (42). DDC is essential forformation of biogenic amines, both serotonin and catechol-amines. But other NE cell markers (dense core granules, chro-mogranin A, Leu-7, and NKH1 antigens) were not expressed.These findings suggest that gastric tumors, as with colorectalcarcinomas, may frequently express part of the program of NEcell differentiation (6, 23). Because three of our four patientsfrom whom cell lines were established were not treated with
systemic chemotherapy, we cannot determine whether NE differentiation affects the response to chemotherapy or survival.
As with our colorectal cell lines (6), DMs appeared to occurfrequently in stomach lines. In two of the gastric lines, theappearance of DMs was obvious, since they were either presentin abundance (NCI-N87) or present in the majority of cells(SNU-16). In the other two lines, only a single DM was detectedin a minority of the cells. DM-like artifacts may occur in someinstances (43). Thus, our identification of DMs in these twocases needs to be interpreted with caution, although we believeour observations to be correct, based on examination of at least50 metaphases photographed for karyotypic analysis.
The presence of both DMs and HSRs or only DMs in somecells from cell line SNU-16 are of special interest. The colorectal carcinoma line, COLO 320 (44), from which the currentCOLO 320DM and COLO 320HSR sublines were isolated(44), probably had a similar situation (i.e., tumors with DMsevolved into culture sublines having one or more HSRs). Thepresence of only DMs or HSRs in different sublines of the same
Fig. 3. Antigen expression in gastric and colorectal cell lines. The relativeconcentrations of CEA, CA 19-9, and TAG-72 present in cell pellets of gastricand colorectal carcinoma cell lines are illustrated on a logarithmic scale.
Table 3 Binding of radiolabeled peplities to human gastric carcinoma cell lines
PEPTIDE (LOG MlFig. 4. Specific binding of 125I-V1Pto SNU-1 gastric carcinoma cells in the
presence of varying concentrations of unlabeled VIP and secretin as shown.
cell line suggests a probable causal relationship between theseaberrations. Of interest, the c-myc proto-oncogene is amplifiedin both COLO 320 and SNU-16 cell lines.
All four human gastric cell lines studied were positive forVIP receptors and two were positive for muscarinic/cholinergicreceptors in low numbers. These results are in agreement withprior studies which have revealed the presence of VIP receptorson occasional human cell lines of both gastric and colonieorigin (45, 46). Although gastrin receptors have also been
I t/j c/3 (/j z
12.7kb —
Fig. 5. Hybridization comparison of EcoR\ digests of gastric and small celllung (SCLC) carcinoma cell line DNA with human c-myc probe. The position ofthe germ line (not shown) and all cell line c-myc genes is 12.7 kilobases. TheDNAs of small cell lung cancer line NCI-N417 and gastric line SNU-16 showapproximately 50-fold amplification of the c-myc germ line, relative to the othercell lines.
Table 4 Summary ofcytogenetic data of the gastric carcinoma cell lines
" In addition, SNU-16 contained two subpopulations with modal chromosome
numbers of 86 and 92 (see text).
described on human cell lines of gastric and colonie origin (47,48) we were unable to demonstrate gastrin binding to the fourcell lines studied here. This may be due to differences inmethods of cell growth or binding studies but is more likely dueto the heterogeneity of gastric neoplasms. The evidence thatpeptide receptors exist on these cell lines suggests the possibilitythat growth and signal transduction may be modulated by eitherreceptor agonists or antagonists, as has been shown for gastrinreceptors on colon carcinoma cells and VIP receptors on gastriccarcinoma cells (45, 47).
Of interest, we found two gastric cell lines, one differentiatedand one poorly differentiated, that expressed receptors formuscarinic cholinergic agents. Receptors of this type have beenfound in pancreatic acini, where they play a role in amylaserelease (49). However, to the best of our knowledge, they havenot been described previously on gastric carcinoma cells. Wedid not detect evidence for expression of several other peptidereceptors or /3-adrenergic receptors on our gastric carcinomacell lines.
The myc genes are proto-oncogenes frequently associatedwith cytogenetic evidence of gene amplification (32, 50). Onegastric line, SNU-16, showed c-myc gene amplification, HSRs,and DMs. Similar findings were present in colorectal cell linesNCI-H716 (6) and COLO 320 (44). Previously, c-myc geneamplification in human gastric cancers has only been reportedin xenografts transplanted into nude mice (51, 52). All of thegastric lines, including line SNU-16, expressed relatively low
Fig. 6. Metaphase spreads of gastric cell line SNU-16. A metaphase spreadfrom the major subline (with lower chromosome number) (A) contains 17 DMs(some DMs are marked by arrowheads). No HSRs are present. A metaphasespread from the minor subline (with higher total chromosome number) (B)contains both DMs and HSRs. There are three HSR chromosomes: short arm ofNil (lip HSR), long arm of N6 (6q HSR), and an unidentified chromosome(HSR).
levels of c-myc RNA. Other oncogenes occasionally describedto be amplified or overexpressed in gastric tumors and cell linesinclude c-erb-B-2 (53), transforming gene hst (also known asHSTFl) (54, 55), and the ras family (56-58).
The multiple properties we have described in gastric carcinoma cell lines are remarkably similar to those found in a panelof colorectal carcinoma cell lines initiated by us (6). Theseproperties include morphology, growth characteristics, expression of surface glycoproteins, partial expression of NE cellmarkers, frequent chromosomal evidence of gene amplification,and occasional amplification of the c-myc proto-oncogene.
However, clinical studies indicate that there are major differences in the biology of these two tumor types. The in vitrochemosensitivity patterns to cytotoxic drugs and expression of
2779
the multidrug resistance-associated (MDRl) gene are very different in the gastric and colorectal cell lines (59). In addition,colorectal carcinoma cell lines express receptors for peptidehormones more frequently than gastric carcinoma cell lines(60).
Our four well characterized cell lines should provide usefuladditions to the modest number currently available for the invitro study of gastric carcinoma.
REFERENCES
1. Anonymous. Ministry of Health and Social Affairs. One year's report forcancer registry programme in the Republic of Korea. July 1, 1985-June 30,1986. J. Korean Cancer Assoc., 19: 131-258, 1987.
2. Mettlin, C. Epidemiologie studies in gastric adenocarcinoma. In: H. O.Douglass (ed.), Gastric Cancer, pp. 1-25. New York: Churchill Livingstone,1988.
3. Macdonald, J. S., Cohn, I., and Gunderson, L. L. Cancer of the stomach. In:V. T. DeVita, S. Hellman, and S. A. Rosenberg (eds.). Cancer, Principlesand Practice of Oncology, pp. 659-690. Philadelphia: J. B. Lippincott Co.,1985.
4. Romond, E. H., and Macdonald, J. S. Chemotherapy of gastric cancer. In:H. O. Douglass (ed.). Gastric Cancer, pp. 205-220. New York: ChurchillLivingston, 1988.
5. Brattain, M. G., Brattain, D. E., Fine, W. D., Khaled, F. M., Marks, M. E.,Kimball, P. M., Arcolano, L. A., and Danbury, B. H. Initiation and characterization of cultures of human colonie carcinoma with different biologicalcharacteristics utilizing feeder layers of confluent fibroblasts. Oncodev. Biol.Med., 2:355-366, 1982.
6. Park, J. G., Ole, H. K., Sugarbaker, P. H., Henslee, J. G., Chen, T. R.,Johnson, B. E., and Gazdar, A. F. Characteristics of cell lines establishedfrom human colorectal carcinomas. Cancer Res., 47: 6710-6718, 1987.
7. Willson, J. K. V., Bittner, G. N., Oberley, T. D., Meisner, L. F., and Weese,J. L. Cell culture of human colon adenomas and carcinomas. Cancer Res.,47:2704-2713, 1987.
8. Hojo, H. Establishment of cultured cell lines of human stomach cancer originand their morphological characteristics. Niigata Igakukai Zassi, 91: 737-763, 1977.
9. Sekiguchi, M., and Suzuki, T. List of human cell lines established andmaintained in Japan. Tissue Culture, 6: 527-548, 1980.
10. Motoyama, T., Hojo, H., and Watanabe, H. Comparison of seven cell linesderived from human gastric carcinomas. Acta Pathol. Jpn., 36: 65-83, 1986.
11. Kubota, H., Harada, T., Morikawa, S., and Nakamura, T. In vivo selectionof tumorigenic subline from non-tumorigenic human gastric carcinoma cells:in relation to proliferative properties in vivoand in nude mice. Gastroenterol.Jpn., 23:371-383, 1988.
12. Whelan, R., Gibby, E., Sheer, D., Povey, S., and Hill, B. T. Characterizationof a continuous cell line in culture established from a Krukenberg tumor ofthe ovary arising from a primary gastric adenocarcinoma. Eur. J. CancerClin. Oncol., 24: 1397-1408, 1988.
13. Dobrynin, Y. V. Establishment and characteristics of cell strains from someepithelial tumors of human origin. J. Nati. Cancer Inst., 31: 1173-1195,1963.
15. Smith, H. S. In vitro properties of epithelial cell lines established from humancarcinomas and nonmalignant tissue. J. Nati. Cancer Inst., 62: 225-230,1979.
16. Laboisse, C. L., Augeron, C., Couturier-Turpin, M-H., Gespach, C., Cheret,A-M., and Potet, F. Characterization of a newly established human gastriccancer cell line GHT-1 bearing histamine H2-receptors. Cancer Res., 42:1541-1548, 1982.
17. Barranco, S. C., Townsend, C. M., Jr., Casartelli, C., Macik, B. G., Burger,N. L., Boerwinkle, W. R., and Gourley, W. K. Establishment and characterization of an in vitro system for human adenocarcinoma of the stomach.Clin. Chest Med., 43: 1703-1709, 1983.
18. Lin, C-H., Fu, Z-M., Liu, Y-I., Jang, J-L., Xu, J-F., Chen, O-S., and Chen,H-M. Investigation of SGC-7901 cell line established from human gastriccarcinoma cells. Chin. Med. J. (Engl. Ed.), 97: 831-834, 1984.
19. Dippold, W. G., Krön,G., Boosfeld, E., Dienes, H-P., Klingel, R., Knuth,A., Wagner, R., and Buschenfelde, K-H. M. Z. Signet ring stomach cancer:morphological characterization and antigenic profile of newly establishedcell line (Mz-Sto-1). Eur. J. Cancer Clin. Oncol., 23:697-706, 1987.
20. Chen, T. R. Rapid trypsin G-band method. Sun el al.'s modification. Mamm.Chrom. Newsl., 22: 70-71, 1981.
21. Chen, T-R., and Shaw, M. W. Stable chromosome changes in a humanmalignant melanoma. Cancer Res., 33: 2042-2047, 1973.
22. Baylin, S. B., Stevens, S. A., and Shakir, K. M. Association of diamineoxidase and ornithine decarboxylase with maturing cells in rapidly proliferating epithelium. Biochim. Biophys. Acta, 541: 415-419, 1978.
23. Gazdar, A. F., Helman, L., Israel, M. A., Russell, E. K., Linnolla, I.,Mulshine, J., Schuller, H., and Park, J. G. Expression of neuroendocrinecell markers L-DOPA decarboxylase, chromogranin A, and dense core gran-
ules in human tumors of endocrine and non-endocrine origin. Cancer Res.,48: 4078-4082, 1988.
24. Collins, M. J., Jr., and Parker, J. C. Murine virus contaminants of leukemiaviruses and transplantable tumors. J. Nati. Cancer Inst., 49: 1139-1143,1972.
25. Harris, H., and Hopkinson, D. A. Handbook of Enzyme Electrophoresis inHuman Genetics. New York: North-Holland Publishing Co., 1978.
26. Fletcher, R. H. Carcinoembryonic antigen. Ann. Int. Med., 104:66-73. 1986.27. Novis, B. H., Gluck, E., Thomas, P., Steele. G. D., Zurawski, V. R., Jr.,
Stewart, R.. Lavin, P. T., and Zamcheck, N. Serial levels of CA 19-9 andCEA in colon cancer. J. Clin. Oncol., 4: 987-993, 1986.
28. Magnani, J. L.. Brockhaus, M., Smith, D. F., Ginsburg, V., Blaszczyk, M.,Mitchell, K. F., Steplewski, Z., and Koprowski. H. A monosialogangliosideis a monoclonal antibody-defined antigen of colon carcinoma. Science (Wash.DC), 212: 55-56. 1981.
29. Thor, A., Ohuchi, N., and Schlom, J. Distribution of oncofetal antigentumor-associated glycoprotein-72 defined by monoclonal antibody B72.3.Cancer Res., 46: 3118-3124, 1986.
30. Horan Hand, P., Colcher, D., Salomon, D., Ridge, J., Noguchi, P., andSchlom. J. Influence of spatial configuration of carcinoma cell populationson the expression of a tumor-associated glycoprotein. Cancer Res., 45: 833-840, 1985.
31. Battey, J., Moulding, C, Taub, R.. Murphy. W., Stewart, T., Potter, H.,Lenoir, G., and Leder, P. The human c-myc oncogene: structural consequences of translocation in the IgH locus in Burkitt lymphoma. Cell, 34:779-787, 1983.
32. Schwab. M., Alitalo. K., Klempnauer. K. H.. Varmus, H. E., Bishop, J. M.,Gilbert, F., Brodeur, G., Goldstein, M., and Trent, J. Amplified DNA withlimited homology to myc cellular oncogene is shared by human neuroblastoma cell lines and a neuroblastoma tumor. Nature (Lond.), 305: 245-248, 1983.
33. Nau, M. M., Brooks, B. J., Battey, J., Sausville, E. A., Gazdar, A. F., andMinna, J. D. L-myc: a new myc-related gene amplified and expressed inhuman small cell lung cancer. Nature (Lond.), 3IS: 69-73, 1985.
34. Bender, T. P., and Kuehl, W. M. Murine myb protooncogene mRNA: cDNAsequence and evidence for 5' heterogeneity. Proc. Nati. Acad. Sci. USA, 83:3204-3208, 1986.
35. Southern, E. M. Detection of specific sequences among DNA fragmentsseparated by gel electrophoresis. J. Mol. Biol., 98: 503-517, 1975.
36. Mieter, P. A., Hollis. G. F., Korsmeyer, S. J., Waldmann, T. A., and Leder,P. Clustered arrangement of immunoglobulin lambda constant region genesin man. Nature (Lond.), 294: 536-540, 1981.
37. Jensen, R. T., Moody, T., Pert, C., Rivier, J. E., and Gardner, J. D.Interaction of bombi-sili and litorin with specific membrane receptors onpancreatic acinar cells. Proc. Nati. Acad. Sci. USA, 75: 6139-6143, 1978.
38. Zhou, Z. C., Villanueva, M. L., Noguchi, M., Jones, S. W., Gardner, J. D.,and Jensen, R. T. Mechanism of action of calcitonin gene-related peptide instimulating pancreatic enzyme secretion. Am. J. Physiol., 251: G391-G397,1986.
39. von Schrenck, T., Heinz-Erian, P., Moran. T., Mantey, S. A.. Gardner, J.D., and Jensen, R. T. Characterization of a neuromedin B receptor inesophagus: evidence for subtypes of bombesin receptors. Am. J. Physiol.,256: 6747-6758, 1989.
40. Munson, P. J., and Rodbard. D. Ligand characterization of binding systems:a versatile, computerized approach. Anal. Biochem., 707: 220-229, 1980.
41. Stevenson, H., Gazdar, A., Phelps, R., Linnolla, R. I., Ihde, D., Ghosh, B.,Woods, E., Ole, H., O'Connor, T., Makuch, R., Kramer, B., and Mulshine,
J. Establishment of tumor cell lines in vitro is a negative prognostic factorfor survival in non-small cell lung cancer patients. Proc. Am. Soc. Clin.Oncol.. 8: 247, 1989.
42. Gazdar, A. F., Carney, D. N., Guccion, J. G., and Baylin, S. B. Small cell
carcinoma of the lung: cellular origin and relationship to other pulmonarytumors. In: F. A. Greco. R. K. Oldham. and P. A. Bunn (eds.). Small CellLung Cancer, pp. 145-176. New York: Gruñeand Stratton. 1981.
43. Barker, P. E. DMs in human tumor cells. Cancer Genet. Cytogenet., 5: 81-94, 1982.
44. Quinn. L. A., Moore, G. E.. Morgan, R. T., and Woods, L. K. Cell linesfrom human colon carcinoma with unusual cell products, double minutes,and homogeneously staining regions. Cancer Res., 39: 4914-4924, 1979.
45. Emami, S., Chaske, E., Bodierie, H., Gespach, C., Bataille, D., and Rosselin,G. Functional receptors for VIP. GIP, glucagon-29 and -37 in the HGT-1human gastric cancer cell line. Peptides, 7(Suppl. 1): 121-127, 1986.
46. Bishr Omary, M., and Kagnoff, M. F. Identification of nuclear receptors forVIP on a human colonie adenocarcinoma cell line. Science (Wash. DC), 238:1578-1581, 1987.
47. Singh, P., Rae-Venter, B., Townsend, C. M., Jr., Kahalil, T., and Thompson,J. C. Gastrin receptors in normal and malignant gastrointestinal mucosa:age associated changes. Am. J. Physiol., 249: G761-G769, 1985.
48. Weinstock, J., and Baldwin. G. S. Binding of gastrin (17) to human gastriccarcinoma cell lines. Cancer Res., 48: 932-937, 1988.
49. Gardner, J. D., and Jensen, R. T. Secretagogue receptors on pancreatic acinarcells. In: L. R. Johnson (ed.), Physiology of the Gastrointestinal Tract, pp.11090-1127. New York: Raven Press, 1987.
50. Little, C. D., Nau, M. M.. Carney, D. N., Gazdar, A. F., and Minna, J. D.Amplification and expression of the c-myc oncogene in human lung cancercell lines. Nature (Lond.), 306: 194-196, 1983.
51. Nakasato, F., Sakamoto. H., Mori, M., Hayashi, K., Shimosato, Y., Nishi,M., Takao, S., Nakatani, K., Terada, M., and Sugimura, T. Amplification ofthe c-myc oncogene in human stomach cancers. Gann, 75: 737-742, 1984.
52. Shibuya, M., Yokota, J., and Ueyama, Y. Amplification and expression of acellular oncogene (c-myc) in human gastric adenocarcinoma cells. Mol. Cell.Biol., 5:414-418, 1985.
53. Yamamoto, T., Ikawa, S., Akiyama, T., Semba, K., Nomura, N., Miyajima,N., Saito, T., and Toyoshima, K. Similarity of protein encoded by the humanc-erbB-2 gene to epidermal growth factor receptor. Nature (Lond.), 319:230-234, 1986.
54. Sakamoto, H., Mori, M., Taira, M., Yoshida, T., Matsukawa, S., Shimizu,K., Sekiguchi, M., Terada, M., and Sugimura, T. Transforming gene fromhuman stomach cancers and a noncancerous portion of stomach mucosa.Proc. Nati. Acad. Sci. USA, 83: 3997-4001, 1986.
55. Yoshida, M. C., Wada, M., Satoh, H., and Yoshida, T. Human HST1(HSTFI) gene maps to chromosome band Ilql3 and coamplifies with theINT2 gene in human cancer. Proc. Nati. Acad. Sci. USA, 85: 4861-4864,1988.
56. Ohuchi, N., Horan Hand, P., Merlo, G., Fujita, J., Mariani-Constantini, R.,Thor, A., Nose, M., Callahan, R., and Schlom, J. Enhanced expression of c-Ha-rai p21 in human stomach adenocarcinomas defined by immunoassaysusing monoclonal antibodies and in situ hybridization. Cancer Res., 47:1413-1420, 1987.
57. O'Hara, B. M., Oskarsson, M., Tainsky, M., and Blair, D. G. Mechanism of
activation of human ras genes cloned from a gastric adenocarcinoma and apancreatic carcinoma cell line. Cancer Res., 46: 4695-4700, 1986.
58. Bos, J. L., Veriaan-de Vries, M., Marshall, C. J., Veeneman, G. H., VanBoom, J. H., and Van der Eb, A. J. A human gastric carcinoma contains asingle mutated and an amplified normal alÃeleof the Ki-rai oncogene. NucleicAcids Res.. 12: 1209-1217, 1986.
59. Park, J. G., Kramer, B. S., Lai, S. L., Goldstein, L. J., and Gazdar, A. F.Chemosensitivity patterns and expression of human multidrug resistance(MDR1) gene by human gastric and colorectal cell lines. J. Nati. CancerInst., 82: 193-198, 1990.
60. Frucht, H., Oie, H. K., Park, J. G., Gazdar, A. F., Gardner, J. D., andJensen, R. T. Presence of peptide receptors on newly established humancolon cancer cell lines. Gastroenterology, 96: A161, 1989.
1990;50:2773-2780. Cancer Res Jae-Gahb Park, Harold Frucht, Renato V. LaRocca, et al. CarcinomaCharacteristics of Cell Lines Established from Human Gastric
