Int. J. Cancer: 67,668-675 (1996) 0 1996 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de I’Union Internationale Contre le Cancer CHARACTERIZATION OF TUMORIGENIC SUB-LINES FROM A POORLY TUMORIGENIC HUMAN COLON-ADENOCARCINOMA CELL LINE Teresa LOPEZ-CONEJO, Nieves OLMO,Javier TURNAY, Juana NAVARROand Antonia LIZARBE’ Departamento de Bioqiiimica y Biologia Molecular, Facultad de Qiiimrcas, Universidad Complutense, 28040-Madrid, Spain. The interaction of tumor cells with extracellular-matrix components is suspected to play an important role in tumorigenesis induction. The tumorigenicity of a poorly tumorigenic human colon-adenocarcinomacell line (BCS-TC2) was induced by co-injection with Matrigel. A new cell sub-line, BCS-TC2. I, was isolated and establishedfrom these tumors. Implantationof these cells in nude mice in the absence of Matrigel-generated tumors which allowed the establishment of another tumorigenic cell sub-line, BCS-TC2.2. Matrigel and laminin, but not collagens, promote the tumorigenicity of BCS-TC2 cells, probably due to specific interactions of a pre-existing minor cell sub-population with laminin, which facilitate the initial growth of these cells in vivo. Cytogenetic analysis reveals that both sub-lines originate from the parental one, but a new marker in chromosome 9 is observed. These sub-lines present a lower degree of differentiation, as deduced from the lower CEA content, 5‘-nucleotidase and alkaline-phosphatase activities. No variation is observed in the mRNA and protein expression of the 67-kDa laminin-bindingprotein. However, an increase in PI integrins and a parallel decrease in p4 integrin were detected. Thus, the new sub-lines, compared to the parental cells, present karyotypic and phenotypic differences such as the expression of a distinctive integrin pattern. This system represents a useful model for understanding the development and progression of tumorigenicity in cancer cells. o I996 Wiley-Liss,Inc. Data from the American Cancer Society reveal the high incidence and mortality of colorectal cancer. Lung, prostate, breast and colorectal cancers are the most common causes of cancer death. Diagnosis, surgical techniques and therapies have been greatly improved, but to date most deaths from colorectal and other cancers are caused by metastasis. Knowledge of the regulatory mechanisms of tumor growth, local infiltration and metastasis could help to find more effective therapy to deal with disseminated cancer cells; this is dependent on the availability of relevant in vivo models for this disease. The extracellular matrix (ECM) plays an active role during healing and development as well as in tumor growth and progression, modifying tumor-cell adhesion, migration and differentiation (Ruoslahti, 1992; Passaniti et al., 1992; Noel et al., 1993; Grant et al., 1994). In the carcinogenic process, the tumor cell must migrate through the surrounding connective tissue and disrupt basement membranes, a specialized ECM, allowing the cells to cross these barriers and disseminate through the organism. Individual ECM components appear to enhance the malignant phenotype in vitro and promote tumorcell growth in vwo (Passaniti et al., 1992). In fact, the malignancy of tumor cells has also been associated with adhesion to specific ECM components (Ruoslahti, 1992; Kim etal., 1995). The use of Matrigel, a reconstituted basement membrane derived from the EHS murine tumor, has greatly improved understanding of the role of ECM in tumor growth. Matrigel promotes tumor-cell growth in vitro (Kleinman et al., 1986; Fridman et al., 1990, 1992; Albini et al.. 1992; Vukicevic et al., 1992u; Noel et al., 1993) and enhances in vivo tumorigenicity of transformed cells when co-injected into nude mice (Fridman et al., 1990, 1992; Albini et ab, 1992; Vukicevic et al., 19926; Topley rt a/., 1993). The interaction of cell-adhesion receptors with ECM components can affect the biological behavior of neoplastic cells by influencing the process of tumor invasion and metastasis (Castronovo, 1993; Kim et al., 1995). Supporting this notion, a 67-kDa high-affinity laminin-binding protein has been directly related to the invasive and metastatic potential of several tumor cell lines (Rao et al., 1989; Castronovo, 1993). It has also been suggested that the modulation of the levels of specific integrins is associated with these processes. There is increasing evidence that integrin expression is altered following malignant transformation (Dedhar and Saulnier, 1990; Castronovo, 1993; Kim et al., 1995; Fujita et al., 1995). However, no uniform pattern of changes with this process has been described up to date. BCS-TC2 cells, derived from a primary and poorly differentiated human colon adenocarcinoma, provide an interesting model for studying the influence of the ECM on the malignant behavior of epithelial transformed cells. In the present study we demonstrate that Matrigel modifies the tumorigenicity of these human transformed cells. BCS-TC2 cells show very low tumorigenicity after S.C. injection into nude mice, either alone or in the presence of different types of collagens. However, the malignant potential is greatly enhanced when the cells are co-injected with Matrigel or laminin, the major basementmembrane glycoprotein. We show that interaction of Matrigel with these adenocarcinoma cells selects a sub-set of neoplastic cells which, after propagation and establishment in culture, show tumorigenic potential by themselves in the absence of Matrigel. While the expression of the 67-kDa laminin-binding protein is apparently not modified, changes in several characteristics and alteration in the expression pattern of different integrins with affinity to laminin have been detected in these cells. MATERIAL AND METHODS Matrigel, ECM protein preparations and antibodies The “reconstituted basement membrane” gel, Matrigel, containing laminin, type-IV collagen, nidogen and heparansulfate proteoglycan, was extracted from the Engelbreth-HolmSwarm (EHS) tumor as described by Kleinman et al. (1986). Low Matrigel, from which low-molecular-weight components were partially removed, was prepared according to Vukicevic et al. (1992b). Nidogen-free laminin, type-I, type-IV and type-V collagens were prepared as described by Olmo et al. (1992). Mouse anti+, chain (Alex 1/4), anti-al chain (TS 217) and anti-as chain (HP 2/1) monoclonal antibodies (MAbs) were kindly provided by Dr. Sanchez-Madrid (Hospital de la Princesa, Madrid, Spain). Rat MAb (GoH3) specific for an extracellular epitope on sub-unit, was a generous gift from Dr. A. Sonnenberg (Blood Transfusion Service, Amsterdam, The Netherlands). Mouse anti-67-kDa laminin-binding protein MAb (MLUCS) was a gift of Dr. S. MCnard (Istituto Nazionale Tumori, Milan, Italy). MAbs against the a2(PlE6). a3 (PlB5) and pa (3El) integrin sub-units were purchased from Chemicon (Temecula, CA); mouse antibodies to Ki-67 nuclear antigen (MIB-1) and carcinoembryonic antigen (CEA) were ‘To whom correspondence and reprint requests should be addressed, at fax: +-34-1-3944159. Received: April 4, 1996 and in revised form May 8, 1996 CHARACTERIZATION OF TUMORICENIC SUB-LINES from Immunotech (Marseille, France). The murine anticytokeratin (AEI and AE3) and anti-leukocyte common antigen (LCA) antibodies were obtained from Signet (Dedham, MA). Tumor cells and culture conditions BCS-TC2 human colon-adenocarcinoma cells were cultured in standard conditions; the growth and characteristics of these cells have been reported (Turnay et al., 1990). Standard culture medium was Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FCS, penicillin (50 IU/ml), streptomycin (SO pgiml) and L-glutamine (300 pg/ml). Cell harvesting for assays was performed by treatment of nearconfluent cullures with 0.05% (wiv) trypsin, 0.02% (w/v) EDTA. After centrifugation, cells were allowed to recover the cell surface for at least 1 hr in supplemented medium, washed with PBS (50 mM sodium phosphate, p H 7.4, 0.15 M NaCI) and re-suspended at the required cell density. The cell viability was measured by the Trypan-blue exclusion test. 669 incubated with the secondary peroxidase-conjugated goat anti-mouse antibody (Pierce, Rockford, IL). After further washing, sections were stained using 3,3'-diaminobenzidine tetrahydrochloride as chromogen, counterstained with hematoxylin, dehydrated and mounted. Establishment of in vivo-passaged cell lines from BCS- TC2 parental cells Tumor specimens were dissected aseptically, washed with PBS and minced into small portions with a surgical blade. Tumor tissue was dissociated by collagenase treatment (200 units/ml of collagenase Type-I and 270 unitsiml DNase, both from Sigma, St. Louis, MO) for 4 hr at 37°C. After centrifugation, the resulting material was re-suspended in standard culture medium and plated onto cell culture flasks. Flattened cells, mainly exhibiting a polygonal morphology, were obtained, and they grew firmly attached to the culture support adopting a greatly organized monolayer. These cells (BCSTC2.1), cultured and passaged under standard culture conditions, were characterized and used for a secondary in vivo Cell-adhesion and (3H]thymidine-incoiporation studies passage. The whole procedure was repeated with the tumors Cell-adhesion assays on different substrates were performed obtained after inoculation of BCS-TC2.1 cells, thus allowing as described (Turnay et al., 1994). In order to check the effect the establishment of a new cell sub-line (BCS-TC2.2 cells). of ECM components and serum on cell proliferation, 5 x lo4 Studies with these variant sub-lines were performed with cells cells were seeded into 96-well plates in the presence of 5% between the 10th and 15th passages. Tumorigenicity assays of FCS. After 24 hr the serum content was reduced to 0.5%. One the sub-lines were performed preparing the cells as indicated day later, the medium was replaced by serum-free standard for the parental ones. culture medium containing [ 3H]thymidine (Amersham, Aylesbury, UK; 35 Ci/mmol; 1 kCi per well) and the ECM factor to FACS Quantification of cell-surface receptors was performed by be tested: laminin, collagen type-I or collagen type-IV, each at a final concentration of 20 pg/ml. The effect of FCS was also flow-cytometry analysis (FACS) using a FACScan (Becton studied at different concentrations (0, 5 and 10%) in the Dickinson, San Jose, CA) and data were analyzed on a H P 310 culture medium. After 18-hr incubation at 37"C, the medium computer. Briefly, growing cultures were ttypsinized and was removed and the cells were incubated 10 min at room allowed to recover the cell surface. Tumor cells (loh) were temperature with 5% tricloroacetic acid, followed by 3 washes stained by incubation with saturating concentrations of the with ice-cold ethanol. Radioactivity was measured after cell different MAbs, followed by treatment with fluoresceinlysis in 0.1 N NaOH containing 0.1% SDS. Protein content was isothiocyanate(F1TC)-conjugated goat anti-mouse or anti-rat determined in parallel wells. IgG (Chemicon). Cells were then rinsed in PBS and resuspended in PBS containing 1 kgiml propidium iodide. In In vivo tiimor growth studies order to compare the phenotype of the different cell preparaMale athymic BALBic nude mice (8 weeks old) were tions, all main operative settings were kept constant throughpurchased from IFFA-CREDO (L'Arbresle, France) and out the study. were housed in a laminar-flow cabinet under specific-pathogenfree conditions. In vivo tumorigenicity of BCS-TC2 cells was Northern -blot hybridization tested by S.C. inoculation in the lumbar region of 0.2 ml of a Total cellular R N A was extracted from tumor biopsies or mixture of tumor cells in DMEM alone or in the presence of cell cultures using the guanidine-thiocyanate method, electrothe studied factor. The mice were surveyed daily to determine phoresed on a 1% formaldehyde-denaturing agarose gel, and when tumors first became apparent upon visual inspection blotted onto nylon membrane filters. Hybridization was carand, from then on, tumor sizes were measured weekly. Latency ried out at 42°C in 50% formamide buffer with a 32P-laminintime was estimated as the number of days needed to obtain binding protein oligonucleotide probe (complementary to tumors of 0.2 to 0.3 mm diameter. Tumor size was calculated nucleotides +230 to +259, according to Raoet al., 1989) for 16 by measuring the dimensions of the tumor mass (average of the hr. The stringency wash was performed at 55°C in 2 x SSC 2 right-angle diameters) with a vernier caliper at various times (SSC: 20 mM sodium citrate, pH 7.0, containing 150 mM after injection. At the end of the experiment, the mice were sodium chloride), 0.1% SDS for 30 min followed by 0.1 x SSC, killed and the tumors were excised and used in the different 0.1% SDS at room temperature for 15 min. Afterwards, R N A experiments. For each data point, 6 to 8 mice were given bands were visualized by autoradiography. Hybridization to injection per group, and each experiment was repeated at least 32P-28S R N A probe was used to test R N A transfer and to normalize densitometric signals. 3 times. Immurioliistociiemist~studies At selected times after S.C. injection of the tumor cells, animals were killed and autopsied. The tumor at the primary site of the inoculum was examined, also lungs, kidneys, spleen, liver and pancreas for gross metastases. Tumor specimens were excised, fixed in 10% formaldehyde in PBS, and routinely processed for paraffin embedding. For histological examination, 4-km thick sections were stained with hematoxylin and eosin. Immunohistochemical staining was performed by incubation of the tumor specimens for 1 hr at 37°C with the primary antibody (MIB-1, CEA, LCA, AE1 or A E 3 ) at optimal concentrations. After washing with PBS, the slides were Other characterizations and procedures Karyotype studies were performed according to standard procedures (Turnay et al., 1990). 5'-nucleotidase activity was measured as described (Olmo et al., 1992). CEA content was determined in the culture medium and in the cell pellet by using the commercially available Roche-CEA kit. For this purpose, 3 x lo6 cells were re-suspended in 15 ml of culture medium and seeded on 75-cm2 culture flasks. Four days after plating the medium was changed to another containing only 2% FCS. Seven days after seeding, culture media were collected and concentrated by lyophilization; cell layers were rinsed and scraped off with PBS. Alkaline-phosphatase activity 670 LOPEZ-CONEJO ETAL. was assayed using the Sigma diagnostics kit (procedure 245); 3.5 X lo5 cells were seeded into 60-mm plastic culture plates and were maintained in continuous culture for 20 days, with medium changes every 2 days. After washing, the cell monolayers were homogenized, and parallel determinations of alkalinephosphatase activity and protein concentration were performed. A 2.0 - 0 Control Matrigel 0.6 mg Matrigel 1.2 mg A E0 Y 1.5 - RESULTS The BCS-TC2 cell line was established in culture from a primary human colon adenocarcinoma. The morphological, immunological and ultrastructural features of these cells agree with their epithelial origin (Turnay et al., 1990). They have poor ability to form colonies (2% of colony-forming efficiency in semi-solid media and 22% of plating efficiency) thus showing very low in vitro tumorigenicity. Xenograft studies indicated low or non-tumorigenic potential in vivo, even when these cells were injected in large numbers into athymic mice (Turnay et al., 1990). We have reported that the in vitro behavior of BCS-TC2 cells is modulated by ECM proteins (Turnay et al., 1994). Effect of Matrigel on the tumorigenicity of BCS-TC2 cells The effect of Matrigel on the ability of BCS-TC2 cells to grow in vivo was examined by S.C. injection into nude mice of different numbers of cells mixed with Matrigel. The coinjection of 10h cells with Matrigel resulted in a significant increase in the incidence, tumor-growth rates and final tumor sizes compared with cells injected only with medium. The latency time was also greatly reduced when cells were coinjected with Matrigel (Fig. 1).Tumors grew in all the mice (18 of 18) given a co-injection with Matrigel, whereas only 5 of 42 nude mice given an injection of tumor cells alone produced viable tumors. When the number of cells was reduced to 2.5 x lo5, no tumors at all were induced in animals injected with BCS-TC2 cells alone, while the incidence was maintained when they were co-injected with Matrigel. Tumor size in experimental and control groups differed significantly 12 weeks after injection as a result of enhanced growth of tumor cells in the presence of basement membrane (2.19 2 0.70 vs. 1.12 & 0.21 cm), and the latency period was shorter (5-7 1’s. 35-40 days). Figure 1 shows the dose-dependent effects of Matrigel on tumor growth. When 0.6 or 1.2 mg of Matrigel were S.C. co-injected with BCS-TC2 cells into athymic mice, tumor growth was significantly higher than in animals receiving an equal number of cells without Matrigel. Figure 1 also shows the evolution of the mean tumor diameter as a function of time after the injection. At 10 weeks, doses as low as 0.6 mg/mouse cause a 2-fold increase in tumor size. A reduction in the latency period of the tumors is also observed when the amount of Matrigel increases. Histological studies Cells formed solid tumors surrounded by a limiting membrane when injected S.C.in nude mice. Visual observation of the tumors showed extensive vascularization. In large tumors, the central area was replaced by edema, showing necrotic tissue where blood vessels were not present. No apparent metastases were detected in the mice bearing S.C.tumors when different organs were examined. The histological appearance, morphology and immunohistochemical data of tumors developed in mice in the presence of Matrigel were indistinguishable from those obtained after injection of BCS-TC2 cells in culture medium alone. In general, 2 different layers could be observed in the tumors, an external one, highly cellular and compacted, and an internal area with less and more disseminated cells. In the external area of these tumors, the cells are closely associated cn 1.o 0.5 0.0 2 4 6 8 1 0 1 2 TIME (weeks) FIGURE 1 - Effect of Matrigel on the in vivo growth of BCS-TC2 cells. The effect of Matrigel on the tumorigenicity of BCS-TC2 cells was examined by S.C.injection into nude mice. Tumor growth rate was analyzed after injection of 1 x loh BCS-TC2 cells alone (A)or in the presence of 0.6 mg (0)or 1.2 mg (0)of Matrigel. Tumor size measurements were performed with a caliper at designated time points. Studies were done in triplicate with n = 6-8; data represent mean values. and proliferate actively, as deduced from the positive staining obtained with anti-67-Ki antibody. They are positively stained with anti-AEl cytokeratin antibody, but no reaction is detected with anti-AE3 or anti-CEA antibodies, suggesting a poor degree of differentiation. In contrast, cells from the inner area of the tumors appear less proliferative (negative staining with anti-67-Ki antibodies) but apparently more differentiated, as deduced from the positive reaction with anti-CEA and anti-AE3 antibodies. Effect of Matrigel components on in vivo tumorigenicity of BCS-TC2 cells Type-I collagen gel is not able to support tumor growth (Table I). This indicates that the physical nature of Matrigel (gel structure at 37°C) does not appear to be a requirement for cell growth. Growth factors present in complete Matrigel could also be at least partially responsible for the mitogenic effects. However, low-Matrigel, in which the possible content in growth factors is significantly reduced (Vukicevic et al., 1992b), is also able to induce tumor growth (Table I, Fig. 20). The incidence as well as the tumor size is almost the same when Matrigel or low Matrigel is used; only a small increase in latency time is observed with the latter (Table I). Latency time is also dependent on the low Matrigel doses, decreasing from 3 to 2 weeks when the amount used in the inoculation is increased from 0.2 mg to 0.4 mg. BCS-TC2 cells were also co-injected S.C.with purified ECM proteins. W e have observed that when these cells were co-injected with type-IV or type-V collagen, tumor growth was not promoted (Table I). However, we have found that the co-injection of cells with purified laminin significantly increases the growth of BCS-TC2 tumors (Table I, Fig. 2b), suggesting that this basement-membrane protein has a specific 671 CHARACTERIZATION OF TUMORIGENIC SUB-LINES role in tumor dcvelopment. Thus, injection of lo6 cells with 0.4 mg of laminin resulted in tumor formation in the 80% of the mice with a latency time of less than 2 weeks (Table I). Similar results were obtained using only as little as 0.14 mg of laminin per injection. In vitro efect of laminin and collagens on BCS-TC2 cell brhavror Wc havc checked the in vitro response of BCS-TC2 cells to laminin in comparison with type-I and type-IV collagens, both of them with no effect on in vivo tumorigenicity. Cell attachment is greatly enhanced when the plastic surface is coated with laminin, but collagens also promote a similar effect (Fig. 3a). The incorporation of [ ’Hlthymidine is activated by laminin in the absence of serum, as observed in Figure 3b. However, this effect appears not to be directly linked to the in vivo induction of tumorigenicity by laminin since type-I and TABLE I - EFFECT OF MATRIGEL AND EXTRACELLULAR-MATRIX COMPONENTS ON THE IN V N O GROWTH OF BCS-TC2 CELLS Treatment’ None Matrigel Low Matrigel Type-I collagen gel Type-IV collagen Type-V collagen Laminin Tumor size (cm)’ Latency time (days) - - 1.98 t 0.57 1.58 ? 0.61 10 12 - - 1.19 ? 0.28 10 ‘lo6cells were injected S.C. in nude mice in the absence (none) or in the presence of Matrigel or different extracellular-matrix components. Doses of 0.4 mg of Matrigel, low Matrigel and laminin were used, whereas collagens were each used at 2 different amounts, 1.25 and 2 mg.-?Tumor size was calculated at the end of the experiment (12 weeks), as described in “Material and Methods”. A type-IV collagens activate the thymidine incorporation almost twice as effectively as laminin and similarly to 5% FCS. Isolation and characterization of tumor cellsfrom S.C. Matrigel-induced t~imors BCS-TC2.1 cells were isolated from tumors obtained after S.C. injection of BCS-TC2 cells with Matrigel, and were established and maintained in culture through successive in vitro passages. Cell morphology and organization of the culture is similar to that of the parental cell line, showing a well-spread epitheloid morphology, with some small spindleshaped cells. When BCS-TC2.1 cells were injected alone into athymic mice, the obtained tumors were histologically identical to those obtained when the parental cells were co-injected with Matrigel. These new tumors were subjected to the same procedure as the initial ones. A new cell sub-population, BCS-TC2.2 cells, was obtained in this second in vivo passage and the cells were also established in culture. Whereas the parental BCS-TC2 cell line required Matrigel to form tumors, the new BCS-TC2.1 and BCS-TC2.2 sub-lines were fully tumorigenic in its absence (Fig. 4). A significant increase in tumor growth was observed when the tumor size was analyzed 12 weeks after implantation of loh cells in the absence of Matrigel: values of 1.12 & 0.21,2.37 -t 0.58 and 2.57 & 0.61 cm were obtained for parental cells and for BCS-TC2.1 and BCS-TC2.2 cells respectively. Differences were also evident in tumor latency period, which was reduced from 5 to 6 weeks (parental cells) to 2 weeks (BCS-TC2.2 cells), and incidence data, increased from 12% in parental cells to 67% and 100% in BCS-TC2.1 and BCS-TC2.2 cells respectively. Some of the characteristics of the parental BCS-TC2 cells and the 2 in vivo-obtained sub-lines are summarized in Table 11. Chromosome analysis confirms the human origin of the studied cells. Karyotype studies indicate that the 2 chromosomic abnormalities detected in BCS-TC2 cells (a +der (15) marker with an inclusion of unknown origin and an altered B Low Matrigel 0.2 mg 0 Low Matrigel 0.4 mg 2.0 0 Laminin 0.2 mg Laminin 0.4 mg n E 0 W 1.5 N v) PL 0 1.0 E 3 0.5 0.0 - I I I I I I 2 4 6 8 10 12 I I I I I I I 2 4 6 8 10 12 TIME (weeks) FIGURE 2 -In vivo growth of BCS-TC2 cells in the presence of low Matrigel and laminin. Cells were pre-mixed with 0.2 or 0.4 mg of (a) low Matrigel or (b) laminin, and injected S.C.into nude mice. Tumor size was measured as described in “Material and Methods”. Studies were done in triplicate with n = 6-8; data represent mean value. LOPEZ-CONEJOETA. 672 0.3 A 02.5 x l o 5cells 4.0 1.O x 1o6 cells T 3.5 0.2 -E z BCS-TC2 3.0 BCS-TC2.1 T 0 BCS-TC2.2 T T Y w 2.5 p! 2.0 0 0.1 5I- 1.5 1.o T 0.5 15 10 5 FIGURE 3 - Effect of lamipin, collagens and serum on BCS-TC2 cell adhesion and [ 3H]thymidine incorporation. (a) BCS-TC2 cell adhesion to plastic surfaces coated with BSA, laminin and collagens type I and type IV. After 30-min incubation, attached cells were quantitated by staining with 0.5% crystal violet and analysis in an ELISA reader. (b) Effect of ECM components and serum on the incorporation of [3H]thymidine by BCS-TC2 cells. Mitogenic activity was measured by incubation of the cells in serum-free media supplemented with the correspondin factor or at different serum concentrations, in the presence of [!H]thymidine. Studies were done in triplicate with 4 wells per point; data represent mean value ? SD. chromosome 16) were maintained in the studied metaphases from the sub-lines. However, BCS-TC2.1 and BCS-TC2.2 cells present a new chromosomic marker, an alteration in chromosome 9. CEA production was also considered in this comparative analysis, and was measured in the culture medium as well as in t h e cell monolayer. Values detected in both sub-lines are lower than those obtained for the parental cells. In addition, a reduction of alkaline-phosphatase and 5'-nucleotidase activities were observed in the sub-lines (Table 11). Cell adhesion to different sub-strata (BSA, laminin and collagen types I and IV) is similar in the sub-lines and in the parental cells. However, ["]thymidine incorporation is greatly enhanced in the sub-lines in comparison with the parental cells (2.8- and 12-fold for BCS-TC2.1 and BCS-TC2.2 cells respec- IN LT 3 2 FIGURE 4 - Tumorigenicity of BCS-TC2.1 and BCS-TC2.2 sublines. Tumor size, incidence (IN), and latency time (LT) of the in vivo growth of the parental cells and BCS-TC2.1 and BCS-TC2.2 sub-lines, were determined after S.C.injection of 2.5 X 10' or lo6 cells into nude mice. Other conditions of the experiments are as stated in Figure 1. Data represent mean value 2 SD. tively). When the effect of ECM components on the [?H]thymidine uptake of the sub-lines was analyzed, no significant alterations were observed, indicating in vitro growth independent of the presence of ECM components in the culture medium for these cells. Expression of the 67-kDa laminin-bindingprotein mRNA Northern-blot analysis of total R N A using an specific oligonucleotide probe shows the expression of the 67-kDa laminin-binding protein mRNA in tumors induced by BCSTC2 cells alone, in Matrigel-induced tumors and in those obtained from the BCS-TC2.1 cells. In agreement with other reports (Rao et al., 1989; Castronovo, 1993). a single mRNA species of 1.2 kb is detected (Fig. 5 ) . Quantification of the Northern-blot hybridization was performed by densitometric analysis. Comparison of the steady-state of this mRNA with "P-28S rRNA hybridization indicates a similar level of 67-kDalaminin-binding-protein-mRNA expression in the 3 types of tumors. A similar result was obtained when R N A extracted from the 3 cell lines was analyzed (Table 11). These results are in agreement with the cell-surface expression of this protein. Table I11 shows the FACS analysis of the 3 cell lines using a MAb specific for the 67-kDa laminin-binding protein. The percentage of cells expressing this antigen and the mean fluorescence intensity (MFI) are similar when the 3 cell lines are considered. Itiregrin expression on BCS-TC2 cells and sub-lines The cell-surface expression of integrin chains was examined by flow-cytometric analysis of the parental cells and in vivoobtained sub-lines. All the integrin chains considered may be involved in heterodimers, with affinity towards laminin. Table 111 shows the percentage of cells expressing each antigen as well as the MFI. Data indicate that all 3 cell lines express a diverse array of integrins on their cell surface. Several observations can be drawn from the comparison of the percentage of positive cells and the MFI values corresponding to the different antigens and cell lines: (i) PI, aZ,a3and (Y6 integrin chains CHARACTERIZATION OF TUMORIGENIC SUB-LINES 673 TABLE 11 - CHARACTERISTICS OF BCS-TC2 CELLS AND THE IN VIVO-OBTAINED SUB-LINES BCS-TC2 BCS-TC2.1 BCS-TCZ.? KaVotyPe 46XX, +der(l5), 46XX, +der(l5), 46XX, +der(l5), +der( 16) +der(l6), +der(9) +der(16), +der(9) CEA cell culture monolayer 5 ' -nucleotidase3 Alkaline phosphatase4 67-kDa lamininbinding protein5 100%' loo%* 24 x lo-' 25.7 1 43% 53% 15 x 18.1 38% 55% 12 x 10-5 6.4 1 1 HT-2Y nd 333% 140% nd nd 7 CEA content data are given as percentage values, 100% being the value obtained for BCS-TC2 cells. activity is expressed as nmol '0.89 ng CEA/mg protein.-?33.19 ng CEA1 lo6 ~elIs.-~5'-nucleotidase activity of cells maintained 20 days in culture AMP hydrolizedil5 min/~ell.-~Alkaline-phosphatase is expressed as mU/mg protein. Values are the mean of 3 independent experiments with triplicate samples; SD was always lower than lO%.rData were obtained from the Northern-blot analysis of total RNA extracted from the different cell lines using a 32P-67-kDalaminin-binding protein probe normalized to 32P-28SrRNA probe hybridization.-All values are referred to that of BCS-TC2 cells. The CEA content and expression of the 67-kDa laminin-binding protein are also compared with HT-29 cells, a more differentiated and tumorigenic cell line. Antigen (antihndvl 6i-kDa IMLUCS) F~GURE 5 - Northern-blot analysis of total RNA using a specific probe for the 67-kDa laminin-binding protein mRNA. Tumors obtained when parental cells were injected S.C. in nude mice in cell-culture medium alone (a), or in the presence of Matrigel (b), and tumors induced after injection of BCS-TC2.1 cells in culture medium (c) were used for mRNA expression. Total RNA extracted from the tumor tissues (10 pg) was separated on a formaldehyde-agarose gel, blotted onto a nylon filter, and hybridized with a 32P-67-kDalaminin-binding protein probe. Densitometric analysis of the 67-kDa laminin-binding protein mRNA signals, normalized to the hybridization using a 32P-28S rRNA probe, indicate no significant differences in the expression of this protein. are expressed by almost 100% of the cells, but an increase in the MFI is observed from the parental cells to the sub-lines; (ii) in contrast, a reduction in both parameters for the expression of p4and aIintegrin chains is detected; and (iii) a4 integrin chain is not expressed by parental cells or by the sub-lines. DISCUSSION Matrigel has been successfully used in in vitro and in vivo studies contributing to the knowledge of the role of tumor-cell ECM interactions on the stimulation of tumor growth, promotion of cell differentiation, invasion and metastasis, as well as the angiogenic process (Fridman et al., 1990,1992; Albini et al., 1992; Passaniti et al., 1992; Vukicevic et al., 1992u, b; Noel et al., 1993; Grant et al., 1994; Topley et al., 1993). BCS-TC2 cells, a human colon-adenocarcinoma cell line, show very low tumorigenic potential after S.C. injection in nude mice. Here, BCS-TC2 90 (466) BCS-TC2.1 BCS-TC2.2 23 (14) 100 228) 100 t318) 8 (10) 100 420) 100 i303) 2 (0) 99 (98) 100 366) 56 !90, 97 (447) 0 0) 98 t139) 100 485) 37 $98, 99 (475) Cells were pre-incubated with hybridoma-culture supernatants, followed by washing and labeling with fluorescein-isothiocyanatelabeled goat anti-mouse or goat anti-rat antibodies. Data were collected in a logarithmic scale and the percentage of positive cells was obtained by subtracting the background given by the negative control. MFI values are given in parentheses. a and p, integrin chains; 67-kDa, 67-kDa laminin-binding protein. we have observed that these cells greatly increase their tumorigenicity when co-injected with Matrigel. Even though specific tumor-cell interactions with Matrigel are important for the stimulation of tumor-cell proliferation, it is not yet clear how Matrigel promotes this process. At 37"C, Matrigel forms a multi-layered basement-membrane-like gel structure which could facilitate tumor growth by holding the cells together (Kleinman et al., 1986). However, Matrigel appears to be not only a passive structural support, since we have observed that the co-injection of BCS-TC2 cells in a type-I collagen gel does not result in any improvement in tumor growth over cells injected with medium only, as reported for other cell lines (Fridman et al., 1990; Passaniti et al., 1992). Since Matrigel is rich in growth factors (Vukicevic et nl., 1992u, b), we have analyzed the influence of a significant reduction of such agents on tumor growth using low Matrigel. We have observed that BCS-TC2 tumor-cell growth is also supported by low Matrigel. In contrast, we have found that the co-injection of BCS-TC2 cells with type-IV collagen or type-V collagen does not promote tumor formation. These results suggest specific roles for other basement-membrane components in tumor growth. In fact, laminin is able to promote tumor growth in a similar way to Matrigel, even at very low concentrations. Thus, the induction of tumorigenicity in BCSTC2 cells by Matrigel seems to be triggered mainly by laminin, even though the influence of other components such as growth 674 LOPEZ-CONEJOE T A L factors or proteases cannot be excluded and may enhance laminin effects. The role of the intact laminin molecule as inducer of in vivo tumor growth has been described in only a few other cases, such as human colon-adenocarcinoma WiDr cells and A253 or A549 cells (Topley et al., 1993; Grant et al., 1994). In most studies, when cells have been co-injected with laminin, tumor growth was not promoted (Fridman et al.. 1990, 1992; Albini et al., 1992; Vukicevic et al., 1992b). The mechanisms through which Matrigel or laminin exert their effects remain to be elucidated. Several possibilities have been suggested, such as the induction of cell growth in the initial stages of tumor progression, the promotion of angiogenesis, or the release of host factors that may enhance the viability of tumor cells (Fridman et al., 1990, 1992; Passaniti et al., 1992; Vukicevic et al., 199%; Grant et al., 1994). Laminin increases in vitro proliferation and adhesion of BCS-TC2 cells; however, this mechanism does not appear to be the only one involved in the induction of tumorigenicity, since other ECM molecules such as collagens type I and type IV promote similar in vitro effects, but do not induce tumorigenesis. Tumor grafts or cells obtained from Matrigel-induced tumors (A549 cells, SW480 colon-adenocarcinoma cells, SCLC cells or retinoblastoma cells) show the same characteristics as the corresponding parental cells, being unable to induce tumors in the absence of Matrigel (Albini er al., 1992; Topley et al., 1993). BCS-TC2 cells behave differently, since cells isolated from Matrigel-induced tumors are able to form tumors in the absence of Matrigel. The characteristics of the new tumorigenic cell sub-line BCS-TC2.1 are maintained after a new in vzvo passage (BCS-TC2.2 cells), thus suggesting that a process of selection and adaptation has taken place in which the cells acquire a tumorigenic phenotype, as suggested for a few other tumoral cell lines (Fridman et al., 1992). The obtained sub-lines keep the specific markers on chromosomes 15 and 16, but they present an additional marker (chromosome 9). They also show higher in vitro [3H]thymidine uptake, which, in contrast to parental cells, is independent of the presence of ECM components. Higher tumorigenicity is normally associated with a lower degree of differentiation. Our results regarding CEA content, alkaline-phosphatase and 5'-nucleotidase activities in the parental cells and in the sub-lines could agree with this hypothesis. All these results point to a decrease in the differentiation degree, even though the Farental cell line is already poorly differentiated. From our data, it seems that laminin or Matrigel induces selection of a less differentiated, more resistant and tumorigenic sub-population within the heterogeneous original cell population. The process by which ECM molecules initiate the cell response is known to involve specific cell-surface receptors. Data from co-injection of BCS-TC2 cells with laminin point out to a potential importance of this Matrigel component as a modulator of BCS-TC2 cell function through the interaction with cellular receptors. Multiple laminin-binding proteins have been described, including members of the integrin family and other surface proteins grouped as non-integrin lamininbinding molecules (Castronovo, 1993). A correlation has been established between cell malignancy and mRNA levels or protein expression of the high-affinity 67-kDa laminin-binding protein (Rao et al., 1989; Castronovo, 1993). However, in our study the enhancement of tumorigenicity is not associated with an increase of the mRNA levels or of the expression of this protein in the cell surface. It has been suggested that the expression of a tumorigenic phenotype may require certain genetic changes, which may be different from those required for the development of a metastatic phenotype (Ess et al., 1995). Thus, the expression of high levels of the 67-kDa laminin-binding protein may be helpful for the appearance of an invasive phenotype, facilitating migration and basement- membrane degradation, but not for the promotion of tumorigenicity. The role of integrins in the signal transduction of cell ECM interactions is well known. Changes in the expression of these receptors have also been described in association with neoplastic transformation (Dedhar and Saulnier, 1990; Castronovo, 1993; Kim er al., 1995; Fujita er al., 1995). The biological relevance of these changes remains to be elucidated, but a potential role in the appearance of the tumorigenic or malignant phenotype has been suggested. We have examined integrin expression in BCS-TC2 cells and the changes that take place in the tumorigenic sub-lines, focussing mainly on those integrins with affinity to laminin (a&, and a6P4integrins, and other integrins such as alp1,a2P1and a3PI,which also possess affinity with collagens and fibronectin). We have observed a decrease of P4 integrin expression in the tumorigenic BCS-TC2.1 and BCS-TC2.2 sublines. The a& integrin is an essential constituent of hemidesmosomes in differentiated epithelial cells, contributing to the formation of tight interactions with the underlaying basement membrane (Nagle et al., 1995). This integrin appears to be involved in the endocytotic cycle, and the decrease in its expression could therefore be related to higher motility of the tumorigenic cells, as was also suggested for prostate and breast carcinoma cells, and colorectal tumors (Bretscher, 1992; Natali et al., 1992; Nagle et al., 1995). The expression of a6P4integrin has also been described as inversely correlated with the differentiation degree of epithelial cells (Nagle et al., 1995). O n the other hand, a decrease of PI integrins in aggressive poorly differentiated tumors has been described as a general process. However, some of the reported data are still controversial. We have observed an increase in pI-integrin expression in the tumorigenic BCS-TC2.1 and BCS-TC2.2 sub-lines. Higher levels of integrins have been reported in other cell types when comparing highly tumorigenic and invasive cells with their nontumorigenic and poorly invasive counterparts (Dedhar and Saulnier, 1990). Increased expression of PI integrins in colorectal-cancer tissues and in poorly differentiated gastric adenocarcinomas has also been described (Fujita et al., 1995). This apparent disagreement concerning PI integrin has been discussed by Kim et al. (1995), who argue that the controversy may be due to different causes, such as the many biological activities of integrins or the use of different antibodies. In summary. the use of the nude-mouse-xenograft model in the presence of Matrigel has allowed us to obtain and establish in culture, from poorly tumorigenic colon-adenocarcinoma cells, sub-lines which show tumorigenicity by themselves. Our results suggest that the interaction of specific ECM components, such as laminin, with a particular cell sub-population present in the heterogeneous parental cells appears to trigger specific cellular mechanisms which promote their growth and allow the selection or survival of this genetically different cell sub-population. The cell sub-lines present an altered phenotype that affects at least the degree of differentiation and the expression of ECM receptors which could be result from or be related to the generation of tumorigenicity. Thus, the system herein described may prove useful in defining the events that take place during the progression of tumor cells to a more malignant status, and could contribute to knowledge of the function of ECM components in tumorigenesis. ACKNOWLEDGEMENTS This work was supported by grant PB92-0552 from the DGCYT (Spain) and by a grant from the Fundacidn Cientifica de la A.E.C.C. (Spain). Flow-cytometry analyses were performed at the Flow Cytometry Service of the Complutense University. CHARACTERIZATION OF TUMORIGENIC SUB-LINES 675 REFERENCES ALBINI, A.. MELCHIORI, A., GAROFALO, A., NOONAN, D.M., BASOLO, proteins by normal and malignant human prostate tissue. Amer. J. F.. TARABOLETTI, G., CHADER.G.J. and GIAVAZZI,R., Matrigel Pathol., 146,1498-1507 (1995). promotes retinoblastoma cell growth in vifro and in vivo. Int. J. Cancer, NATALI.P.G., NICOTTA,M.R., BOTTI,C., MOTTOLESE, M., BOGOTTI, 52,234-240 (1992). A. and SEGATTO,O., Changes in the expression of 4 3 4 integrin BRETSCHER, M.S., Circulating integrins: cusp1, a&. and Mac-1, but not heterodimer in primary and metastatic breast cancer. Brit. J. Cancer, a,$,,41 or LFA-1. EMBO J., 11,405410 (1992). 66,318-322 (1992). CASTRONOVO. V., Laminin receptors and laminin-binding proteins NOEL,A,, NUSGENS, B., LAPIERRE, CH. and FROIDART, J.M., Interacduring tumor invasion and metastasis. Invasion Metastasis, 13, 1-30 tions between tumoral MCF7 cells and fibroblasts on Matrigel and (1993). purified laminin. Matrix, 13,267-273 (1993). DEDHAR, S. and SAULNIER, R., Alterations in integrin-receptor expres- OLMO.N., TURNAY, J., RISSE.G.. DEUTZMAN, R., VON DER MARK,K. sion on chemically transformed human cells: specific enhancement of and LIZARBE,M.A., Modulation of 5'-nucleotidase activity in plasma laminin and collagen-receptor complexes. J. Cell B i d , 110, 481-489 membranes and intact cells by extracellular-matrix proteins laminin ( 1990). and fibronectin. Biochem. J., 282,181-188 (1992). Ess, K., CHEN,H., KIER,A. and BRACKENBURY, R., Suppression of PASSANITI, A., ISAACS, J.T., HANEY,J.A., ADLER,S.W., CUJDIK.T.J.. tumorigenicity, but not invasion, in glioblastoma/HeLa cell hybrids. J. LONG,P.V. and KLEINMAN, H.K., Stimulation of human prostaticcell. Physiol., 162,341-347 (1995). carcinoma tumor growth in athymic mice and control of migration in FRIDMAN, R., GIACCONE, G., KANEMOTO, T., MARTIN,G.R., GAZDAR, culture by extracellular matrix. Znf. J. Cancer, 51,318-324 (1992). A.F. and MULSHINE, J.L.. Reconstituted basement membrane (matri- RAO,N.C., CASTRONOVO, V., SCHMITT, M.CH., WEWER,U.M., CLAYgel) and laminin can enhance the tumorigenicity and the drug SMITH,A.P., LIOTTA,L.A. and SOBEL,M.E., Evidence for a precursor resistance of small-cell-lung-cancer cell lines. Boc. nat. Acad. Sci. of the high-affinity metastasis-associated murine laminin receptor. (Wash.), 87,6698-6702 (1990). Biochemistry, 28,7476-7486 (1989). FRIDMAN, R., SWEENEY, T.M., ZAIN,M., MARTIN,G.R. and KLEIN- RUOSLAHTI,E., Control of cell motility and tumor invasion by MAN, H.K.. Malignant transformation of NIH-3T3 cells after subcuta- extracellular-matrix interactions. Brit. J. Cancer, 66,239-242 (1992). neous co-injection with a reconstituted basement membrane (matri- TOPLEY,P., JENKINS, D.C., JESSUP,E.A. and STABLES, J.N., Effect of gel). Int. J. Cancer. 51,740-744 (1992). reconstituted basement-membrane components on the growth of a FUJITA,S., WATANABE. M., KUBOTA, T.. TERAMOTO, T. and KITMIMA. panel of human tumour cell lines in nude mice. Brit. J. Cancer, 67, M., Alteration of the expression in integrin-P1 suh-units correlates 953-958 (1993). with invasion and metastasis in colorectal cancer. Cancer Left., 91, TURNAY, J., OLMO,N.. GAVILANES, J.G., BENITEZ,J. and LIZARBE, 145-149 (1995). M.A., Establishment and characterization of a new human colonGRANT, D.S., KIBBEY, M.C., KINSELLA, J.L., CID, M.C. and KLEINMAN.adenocarcinoma cell line: BCS-TC2. Cytofechnologv, 3,75-88 (1990). H.K., The role of basement membrane in angiogenesis and tumor TURNAY,J., OLMO,N., LOPEZ-CONEJO, M.T. and LIZARBE,M.A., growth. Pafhol. Res. Pracf., 190,854-863 (1994). Matrix components and behaviour of human adenocarcinoma cells. Zn KIM.W.H.. JUN.S.H.. KIBBEY. M.C.. THOMPSON. E.W. and KLEINMAN. vitro cell. dev. Biol.,30,643-647 (1994). H.K.. Expression of f31 integrin in laminin-adhesion-selected human VUKICEVIC, S., KLEINMAN. H.K., LUYTEN, F.P., ROBERTS, A.B., ROCHE, colon-cancer cell lines of varying tumorigenicity. Invasion Metastasis. N.S. and REDDI.A.H., Identification of multiple active growth factors 14,147-155 (1995). in basement-membrane Matrigel suggests caution in interpretation of KLEINMAN, H.K., MCGARVEY,M.L., HASSELL,J.R., STAR, V.L., cellular activity related to extracellular-matrix components. Exp. Cell CANNON, F.B., LAURIE,G.W. and MARTIN,G.R.. Basement-mem- Res., 202,l-8 (1992a). brane complexes with biological activity. Biochemistry. 25, 312-318 VUKICEVIC, S., SOMOGYI, L., MARTINOVIC, I., ZIC,R., KLEINMAN, H.K. (1986). and MARUSIC.M., Reconstituted basement membrane (Matrigel) NAGLE.R.B., HAO, J.S., KNOX,J.D., DALKIN,B.L.. CLARK,V. and promotes the survival and influences the growth of murine tumors. Znt. CROSS,A.E., Expression of hemidesmosomal and extracellular-matrix J. Cancer, 50,791-795 (199%).