Level and Function of Epidermal Growth Factor Receptor Predict the Metastatic Potential of Human Colon Carcinoma Cells1

The purpose of this study was to determine whether production of liver metastasis by human colon carcinoma (HCC) cells depends on the response of tumor cells to organderived growth factors. HCC cells were isolated from several surgical specimens whose malignant potential differed (Dukes’ stage B or D tumors), adapted to grow in culture, and assessed for expression of the epidermal growth factor receptor (EGF-R). Northern blot analyses revealed that highly metastatic HCC cells expressed >5-fold the number of EGF-R mRNA transcripts as low metastatic cells. The level of mRNA correlated with the amount of EGF-R protein as detected by Western blotting, immunohistochemistry, and Scatchard analyses. HCC growth response in vitro to picograms of transforming growth factor a was associated with functional cell surface EGF-Rs as determined by receptor tyrosine kinase activity assays. The EGF-R gene was not amplified or rearranged in highly metastatic cells. However, fluorescence in situ hybridization analysis showed that the copy number of chromosome 7 was higher in the highly metastatic cells. HCC cells were selected in vitro for low or high expression of EGF-R. Subsequent to injection into nude mice, only cells with high expression of EGF-R produced a high incidence of liver metastasis. These data demonstrate that expression of EGF-R by HCC cells directly correlates with their ability to produce hepatic metastasis. INTRODUCTION Clinical observations and studies with experimental tumors have led to the conclusion that certain tumors produce metastasis to specific organs independent of vascular anatomy, blood flow, and number of tumor cells delivered to each organ (I, 2). The search for the mechanisms that regulate the production of organ-specific metastasis began in 1889, when Paget (3) analyzed the autopsy records of women who died of breast cancer and concluded that their pattern of metastasis was predictable. Paget (3) then suggested that metastasis occurred only when certain tumor cells (the ‘ ‘seed’ ‘) interacted with certain organs (the ‘ ‘soil’ ‘). To produce a clinically relevant metastasis, tumor cells must complete all of the steps of the process and then proliferate (1). A modern definition of the seed and soil hypothesis would therefore include evidence that organ-derived panacnine growth factors stimulate the growth of tumor cells expressing the appropriate receptors. We have reported that the implantation of HCC3 cells isolated from surgical specimens of Dukes’ stage D neoplasms into orthotopic organs of nude mice produces hepatic metastases, whereas cells from Dukes’ stage A or B tumors produce only local neoplasms (4, 5). The highly metastatic cells also respond to mitogens associated with liver regeneration induced by hepatectomy (1, 6). TGF-a has recently been shown to be a regulator of liver regeneration (7-9) and proliferation of normal colonic epithelial cells (10, 1 1). TGF-a exerts its effect through interaction with the EGF-R, a plasma membrane glycoprotein that contains within its cytoplasmic domain a tyrosine-specific PTK activity. The binding of TGF-a to the EGF-R stimulates a series of rapid responses, including phosphonylation of tyrosine residues within the EGF-R itself and within many other cellular proteins, hydrolysis of phosphatidyl inositol, release of Ca2 from intracellular stores, elevation of cytoplasmic pH, and morphological changes (12). After 10-12 h in the continuous presence of EGF or TGF-a, cells are committed to synthesize DNA and to divide (13). EGF-Rs are present on many normal and tumor cells (12, 13). Increased levels and/or amplification of EGF-R have been found in many human tumors and cell lines, including breast cancer (14), gliomas (15, i6), lung cancer (17), bladder cancer (18, 19), tumors of the female genital tract (20), the A431 epidermoid carcinoma (21), and colon carcinoma (22). These results suggest a physiological significance of inappropriate expression of the EGF-R tyrosine kinase in abnormal cell growth control. Whether TGF-a can also regulate the proliferation of metastatic HCC cells in the liver or lymph nodes is unclean. Received 7/1/94; accepted 8/23/94. t This work was supported in part by the Josef Steiner Foundation, Cancer Center Support Core Grant CA 16672, and grant R35-CA 42107 [I. J. F.] from the National Cancer Institute, NIH, and Grant PF-3442 from the American Cancer Society [R. R.]. 2 To whom requests for reprints should be addressed, at Department of Cell Biology, Box 173, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030. 3 The abbreviations used are: HCC, human colon carcinoma; EGF-R, epidermal growth factor receptor; SDS, sodium dodecyl sulfate: FBS, fetal bovine serum; eDNA, complementary DNA; FACS, fluorescenceactivated cell sorting; FISH, fluorescence in situ hybridization; PBS, phosphate-buffered saline; DTF, dithiothreitol; BSA, bovine serum albumin; MU, 3-(4,5-dimethlythiazol-2-yl)-2,5-diphenyltetrazolium bromide; FITC, fluorescein isothiocyanate; mAb, monoclonal antibody; TGF-a, transforming growth factor a; DAPI, 2,4-diamidine-2-phenylindole; TUBS, 0.1% Tween 20 (v/v) in Tnis-buffered saline; HEPES, 4-(2-hydnoxyethyl)-1-piperazineethane sulfonic acid; HA, high affinity; LA, low affinity. Research. on July 16, 2017. © 1995 American Association for Cancer clincancerres.aacrjournals.org Downloaded from 20 EGF-R and Colon Carcinoma Liver Metastasis Table 1 Production of experimental liver metastases of HCC cell lines subsequent to implantation into the spleen of nude mice” Incidence of Incidence of Median no. Cell line Origin spleen tumors liver metastasis liver metastasis KMI2C Dukes’ B2 24/25 3/25 0 KM12L1 Variant line 32/35 23/25 10 KM12L2 Variant line 20/20 15/20 50 KMI2L3 Variant line 9/9 9/9 80 KMI2L4 Variant line li/il il/il 100 KM12SM Variant line 14/14 13/14 88 MHC 1410 Dukes’ D 6/6 6/6 101 KM2O Dukes’ D 6/6 5/6 50 KM23 Dukes’ D 6/6 6/6 60 HT-29 Dukes’ C 6/6 5/6 14 a One X 106 viable cells were injected into the spleen of nude mice. Mice were killed when moribund or after 180 days (data from Refs. 4, 5, 23-25). In this article, we examined the expression and function of EGF-R in a series of HCC lines whose metastatic potential differed. The data show that the expression level (and function) of EGF-R directly correlates with the ability of HCC cells to produce hepatic metastasis. MATERIALS AND METHODS Cells and in Vitro Culture Conditions. A431 human epidenmoid carcinoma cells were obtained from the American Type Culture Collection (Rockville, MD). The human colon cancer line KM12C was established in culture from a Dukes’ stage B2 surgical specimen (4, 5). KM12C cells were injected into the spleen of nude mice and liver metastases were isolated and established in culture, and cells were then injected into the spleen of different nude mice. Cell lines KM12LI, KM12L2, KM12L3, and KM12L4 were established after one, two, three, or four cycles of in vivo selection (4). The KM2O and KM23 lines were established in culture from Dukes’ stage D surgical specimens (5). The highly metastatic MHC 1410 line (23, 24) and HT-29 line (25) were isolated from metastatic lesions of HCC. The metastatic potentials of these HCC cell lines had been established previously (Table 1). All tumor cell lines were maintained on plastic in Dulbecco’s modified Eagle’s medium supplemented with 10% FBS, sodium pyruvate, nonessential amino acids, L-glutamine, and 2-fold vitamin solution (GIBCO, Grand Island, NY), incubated in 5% C02-95% air at 37#{176}C. The cultures were free of Mvcoplasma and the following pathogenic murine viruses: reovirus type 3, pneumonia virus, K virus, Theiler’s encephalitis virus, Sendai virus, minute virus, mouse adenovirus, mouse hepatitis virus, lymphocytic choriomeningitis virus, ectromelia virus, and lactate dehydrogenase virus (as assayed by M. A. Bioproducts, Walkersville, MD). The cultures were maintained for no longer than 6 weeks after recovery from frozen stocks. Animals and Production of Tumors. Male athymic nude mice were obtained from the animal production area of the National Cancer Institute-Frederick Cancer Research Facility (Frederick, MD). The mice were maintained under specific pathogen-free conditions and were used for experiments at 8 weeks of age. Animals were maintained in facilities approved by the American Association for Accreditation of Laboratory Animal Care and in accordance with current regulations and standards of the United States Department of Agriculture, Department of Health and Human Services, and NIH. To produce tumors, tumor cells were harvested from subconfluent cultures by 2-3-mm treatment with 0.25% trypsin and 0.02% EDTA. Tnypsinization was stopped with medium contaming 10% FBS, and the cells were washed once in serum-free medium and resuspended in Hanks’ balanced salt solution. Only single-cell suspensions (1 x 106 cells/0.5 ml of Hanks’ balanced salt solution) with >90% viability were used for injections into the spleens of nude mice. The mice were killed 8-12 weeks thereafter, and the size and the number of liven nodules were determined. Histopathology studies confirmed the nature of the disease. DNA Analyses. DNA was extracted either directly from tumor tissue or from 1-3 X 106 tumor cells growing in culture (26, 27). Briefly, the cell pellet was resuspended in 630 p.1 buffer [55 mM Tnis (pH 8.0), 1 10 msi EDTA (pH 8.0), and 110 mM NaC1]. Proteinase K (0.5 mg/mI) and SDS (0.5% w/v) were added to the cell suspension for incubation at 50#{176}C overnight. The suspension was extracted first with phenol and then with a phenol chloroform solution several times and dialyzed against 50 mM Tnis (pH 8.0), iO msi EDTA (pH 8.0), and 10 mr i NaCI overnight at 4#{176}C. RNase A (70 p.g/ml) was added for 1 h at 37#{176}C. The digest was extracted first with phenol and then several times with a phenol-chloroform solution, and a final dialysis was performed against 10 msi Tnis (pH 8.0)-i msi EDTA (pH 8.0) for 18 h at 4#{176}C. Southern blot analyses were performed as described (27, 28). Blots were washed 2 or 3 times at 60#{176}C with 30 mM NaC1-3 msi sodium citrate (pH 7.2)-0.1% NaDodSO4 (w/v). mRNA Analyses Total cellular RNA was extracted from 1 X i07 tumor cells growing in culture using a guanidinium thiocyanate-hot phenol method as described previously (28) on using the FastTrack mRNA isolation system (Invitrogen, Inc., San Diego, CA). For Northern blot analyses, poly(A ) RNA was prepared by oligo(dT)-cellulose chromatography, fractionated on a 1% denaturing formaldehyde-aganose gel, electrotransfenred at 0.6 amp to a GeneScreen nylon membrane (DuPont Co., Boston, MA), and UV cross-linked with 120,000 p3/cm2 using a UV Stratalinker 1800 (Stnatagene, La Jolla, CA). Filters were washed 2 or 3 times at 60#{176}C with 30 msi NaC1-3 mM sodium citrate (pH 7.2)-0.1% NaDodSO4 (w/v). Hybridization Probes. The cDNA probes used were a 3.8-kilobase XhoI restriction endonuclease fragment from the plasmid pCHCEGFn corresponding to the full-length EGF-R cDNA (courtesy of Dns. F. Kern and M. Lippman, Lombandi Research. on July 16, 2017. © 1995 American Association for Cancer clincancerres.aacrjournals.org Downloaded from Clinical Cancer Research 21 Cancer Research Center, Washington, DC), a 0.9-kilobase EcoRI restriction endonuclease cDNA fragment from the plasmid pMT-TGF1 corresponding to human TGF a (29), and a 1 .3-kilobase PstI gene fragment corresponding to rat glyceraldehyde-3-phosphate dehydnogenase (30). Each cDNA fragment was purified by aganose gel electrophoresis, recovered using GeneClean (BlO 101, Inc., La Jolla, CA), and radiolabeled by the random primer technique using [a-’ 2P]deoxynbonucleotide triphosphates (31). Western Blotting. Tumor cells were washed and scraped into PBS containing 5 m i EDTA and 1 mr i sodium-O-vanadate and centrifuged, and the pellet was resuspended in lysis buffer without Triton X-iOO [1% Triton X-100, 20 mM Tnis-HC1 (pH 8.0), 137 mM NaCl, 10% glycerol (v/v), 2 mst EDTA, 1 msi phenylmethylsulfonyl fluoride, 20 p.M leupeptin, 0.15 units/ml aprotinin], sonicated, and centrifuged to recover insoluble protein. Lysis buffer was added to the pellet, and Triton X-100soluble protein was separated by centrifugation and diluted in sample buffer (62.5 mM Tnis-HC1, pH 6.8, 2.3% SDS, 100 mM DTF, and 0.005% bromphenol blue) and boiled. The proteins (20 p.g/lane) were resolved on 7.5% SDS-polyacnylamide gel electrophoresis and transferred onto 0.45-p.m nitnocellulose membranes. The filters were blocked with 3% BSA in Tnisbuffered saline (20 mM Tnis-HC1, pH 7.5, 150 msi NaCl), probed with polyclonal sheep anti-human EGF-R (1:1000) (UBI, Lake Placid, NY) in TFBS, incubated with horseradish peroxidaseconjugated donkey anti-sheep IgG (1:2000) (Sigma Immunochemicals, St. Louis, MO) in TT’BS. The blots were reprobed with rabbit anti-a actin 1:100 in T’FBS (Sigma Immunochemicals) followed with donkey anti-rabbit 1 :2000 in TTBS (Amensham Corp., Arlington Heights, IL). Protein bands were visualized using the ECL detection system (Amersham Corp.) Densitometry. The EGF-R mRNA or protein levels were quantified in the linear range of the film on a personal densitometer (Molecular Dynamics, Sunnyvale, CA) using the ImageQuant software program. Each sample measurement was calculated as the ratio of the average areas of the specific EGF-R gene transcripts or protein bands under analysis and the control 1.3-kilobase glyceraldehyde-3-phosphate dehydrogenase transcript or actin protein band, respectively. Immunohistochemical Analyses. Tumor cells were immunolabeled with monoclonal mouse anti-EGF-R antibody clone 528 (Oncogene Science, Inc., Uniondale, NY) using an immunogold silver staining technique (32). Cells were plated on 4-chamber slides, fixed with cold acetone, washed, and incubated with the blocking antibody for 20 mm. The samples were then incubated with either the primary on control antibody overnight in a humidified chamber at 4#{176}C. The samples were rinsed 3 times with PBS and incubated with gold-conjugated goat anti-mouse antibody for 1 h. The samples were rinsed, fixed with 2% gluteraldehyde:PBS (v/v) for iO mm, and rinsed 4 times with distilled water. The samples were treated with Silver Intense, washed, and examined microscopically using epipolanization optics (32). (The amount of polarized light reflected by the sample directly correlates with the amount of silver on the surface of the sample.) In Vitro Growth Response to TGF-a. Tumor cells were plated into 96-well tissue culture plates at 1500 cells/well and allowed to attach overnight in medium containing 5% FBS. The cells were rinsed and refed with fresh medium containing 0.5% FBS and human recombinant TGF-a (Collaborative Research, Inc., Bedford, MA) at concentrations ranging from 0.001 to 1 ng/ml. For neutralization experiments, cells were pretreated with an anti-EGF-R polyclonal antibody (Oncogene Science, Inc., Uniondale, NY) (0.01 p.g anti-EGF-R antibody/well) for 30 mm at 4#{176}C or an anti-TGF-a monoclonal antibody (Oncogene Science, Inc.) was added to recombinant TGF-a (1 p.g anti-TGF-a antibody per 0.5 ng TGF-a) for 30 mm at 4#{176}C and then added to the cells. The cells were incubated for an additional 96 h at 37#{176}C and the proliferative activity was determined by the MU assay (33). Briefly, after incubation for 2 h in medium containing MIT at 0.42 mg/ml, the medium was removed and the cells were lysed in dimethyl sulfoxide. The conversion of MT1’ to formazan by metabolically viable cells was monitored by a 96-well microtiter plate reader at 570 nm (Dynatech, Inc., Chantilly, VA). Percentage of stimulation was calculated by the formula [B-A/A] X 100, where A is the absorbance of the control cells and B is the absonbance of the treated cells. Quantitative Binding Analyses Cells were plated into 48-well tissue culture plates at 200,000 cells/well and allowed to attach overnight in medium containing 5% FBS. The cells were washed with PBS and 3-[ I]iodotyrosyl EGF (Amensham Corp.) was added at the indicated concentrations (0-8 nM) in binding buffer (HEPES buffer containing 0.1% BSA). The reaction was carried out at room temperature for 1 h, the cells were washed extensively with binding buffer and lysed in 1% NaOH, and the cell-associated radioactivity was measured in a gamma counter. Nonspecific binding was determined using a 100-fold molar excess of unlabeled ligand. The data were analyzed according to the method of Scatchard (34). The number of receptors and the apparent Kd were determined by linear regression analyses using the program MacLigand (courtesy of Robert E. Williams, UCLA, Los Angeles, CA). PTK Assay. Cell membranes were extracted according to a modification of a described procedure (35). Approximately S X 106 cells were scraped into 15 ml of ice-cold lysis buffer [50 msi Tnis-HC1 (pH 7.4), 1 msi EGTA, 1.5 msi MgC12, S msi Dli’, 250 mM sucrose, 1 mM phenylmethylsulfonyl fluoride, and iO p.g/ml aprotinin]. After a 10-mm sedimentation at 300 X g, the cells were resuspended in S ml of lysis buffer and disrupted twice for 10 s at 60% maximal power using a Soniprep 150 sonicaton (Cuntin Matheson Scientific, Inc., Houston, TX). The sonicated cells were centrifuged for S mm at 1000 X g to remove nuclei. The supernatant was recovered and sedimented by centrifugation at 100,000 X g for 90 mm at 4#{176}C. The pellet was washed once with lysis buffer, resuspended in 20 mM HEPES buffer (pH 7.4), divided into aliquots, and stored at -70#{176}C. EGF-R-associated PTK activity was measured with a PTK assay kit (GIBCO-BRL Life Technologies, Inc., Grand Island, NY). Briefly, 50 mg of membrane protein was incubated for 30 mm at 30#{176}Cin a 20-ml reaction mixture containing 30 msi HEPES (pH 7.4), 10 mM MgC12, 0.1 mst DTF, 25 mg/ml BSA, 0.15% Nonidet P-40 (v/v), 70 p.M sodium O-vanadate, 60 p.M Al?, 0.5 mM PTK substrate peptide RR-SRC, 1 p.Ci [-y-32P]ATP (specific activity, 5000 Ci/mmol), and 1.25 p.g/ml Research. on July 16, 2017. © 1995 American Association for Cancer clincancerres.aacrjournals.org Downloaded from