J Cancer: 52,978-986 (1 992) 1992 Wiley-Liss, Inc.
Itit.
C
r
Publication of the InternationalUnion Against Cancer Publicationde I'Union InternationaleContre le Cancer
PROLIFERATION OF HUMAN COLON CANCER CELLS: ROLE OF EPIDERMAL GROWTH FACTOR AND TRANSFORMING GROWTH FACTORa Shuang HUANG,Jose M. TRUJILLO and Subhas CHAKRABARTY' Division of Laboratory Medicine, The Univeruty of Texas M.D. Anderson Cancer Center, Houston, TX 77030, USA. Human colon cancer cells produce and secrete a variety of polypeptide growth factors. The functional role of these growth factors, however, is poorly understood. Though the secretion of epidermal growth factor (EGF)-like activity and EGF-related molecules by human colon cancer cells in culture has been reported, it is not known whether colon cancer cells produce and secrete EGF, and the functional role of EGF in the growth control of these cells i s also unknown. We have shown that EGF acts as a potent growth stimulator on the moderately differentiated Moser colon cancer cell line and as an inhibitor on the highly metastatic KMl2SM cell line. In the present study, we show that EGF is produced by human colon cancer cells and characterize the levels of EGF mRNA expression and EGF protein secretion from 8 human colon cancer cell lines. The cell-surface EGF receptors on these cell lines were also characterized by radiolabeled ligand binding and Scatchard analyses. All the cell lines expressed EGF mRNA and secreted EGF. Both high- and low-affinity subtypes of EGF receptor were detected on 7 of the cell lines. These lines also secreted transforming growth factor (TGF)(I. Some cell lines exhibited a proliferative response t o treatment with either exogenous EGF or TGFa, while others did not respond t o treatment with these growth factors. Antibody-blocking experiments, using anti-EGF or antiEGF receptor antibody, suggested that these cell lines could be broadly classified into 2 groups in terms of their autocrine or paracrine growth regulation via the cell-surface EGF receptor: (I)cells that utilized EGF and/or TGFa; and (2) cells that did not utilize EGF or TGFa (via the cell-surface receptor), even though they secreted abundant amounts of these growth factors.
o 1992 Wilq-Liss, Inc.
Polypeptide growth factors play a critical role in regulating cell growth and differentiation (Sporn and Roberts, 1988; Smith et al., 1989; Bogler et al., 1990; Barnard et al., 1990). Abnormal expression of growth factors and their receptors or abnormal responses to growth factors, or both. may be involved in cellular transformation and in the maintenancc of the transformed phenotype (Laiho and Keski-Oja, 1989; Cross and Dexter, 1991). Human colon-cancer cells in culture produce and secrete molecules with transforming growth factor (TGF)a-like and epidermal growth factor (EGF)-like activity into the media (Coffeyet al., 1986; Hanauske et ~ l . 1987; , Anzano et al., 1989). It has also been reportcd that human colon cancer cell lines express T G F a mRNA (Coffey et al., 1987). T G F a is a growth-stirnulatory factor in colon cancer and colon epitheliumderived cells (Mulder and Brattain, 1989; Watkins and Levine, 1991; Markowitz et al., 1990). The physiological properties of E G F are nearly identical to those of T G F a (Moses et al.. 1989: Donaldson et al., 1989). E G F acts as a mitogen on many cell types, and most cell types possess cell-surface E G F receptors (Donaldson et al., 1989). Both E G F and T G F a bind to the same cell-surface receptor, through which they mediate their biological action. We have shown that E G F acts as a potent growth stimulator in the moderately differentiated human colon-cancer Moser cell line and is a potent growth inhibitor of the metastatic human colon cancer KM12SM cell line (Huang et a/., 1991). Thus, E G F may function as an important growth-control molecule in human colon-cancer cells. Human colon-cancer cells have been reported to respond to treatment with exoge-
nous E G F (Wan etal., 1988; Mulder and Brattain, 1989; Gross et al., 1991). Endogenous production and secretion of E G F by human colon-cancer cells, however. have not been reported. In this communication, we report the characterization of E G F expression and secretion in 8 human colon-cancer cell lines. Cell-surface E G F receptors were also characterized by radiolabeled ligand binding and Scatchard analyses. All the cell lines expressed E G F mRNA and secreted E G F protein (1.0 to 167.5 ng/mg cell protein) and T G F a (0.8 to 9.4 ng/mg). The proliferative responses of some of these cell lines to treatment with exogenous EGF, TGFa, anti-EGF antibody and anti-EGF receptor antibody were also characterized. The results of antibody-blocking experiments suggested that, in some cell lines, both E G F and T G F a played a growthstimulatory role via the cell-surface E G F receptor, while other cell lines were incapable of responding to these growth factors via the cell-surface E G F receptor even though they secreted E G F and TGFa. MATERIAL AND METHODS
Material Human recombinant EGF, T G F a and monoclonal antibody (MAb) to human E G F receptor were purchased from Upstate Biotechnology (Lake Placid, NY). The anti-EGF receptor antibody recognizes an antigenic determinant located on the extracellular domain of the receptor, immunoprecipitates a 175-kDa protein from A431 cell lysates, inhibits A431 cell growth, and competes with E G F for the ligand binding site (information supplied by manufacturer). MAb to human E G F was purchased from Oncogene Science (Manhasset, NY). T G F a radioimmunoassay kits were purchased from Biomedical Technologies (Stoughton, MA). Both the anti-EGF and anti-EGF receptor antibodies blocked the binding of '251-EGF to the human coloncarcinoma Moser cells cultured in monolayer (data not shown). W e have previously shown that the anti-EGF antibody blocks the proliferative effect of both exogenous and endogenously produced EGF, and the anti-EGF receptor antibody blocks the proliferative effect of endogenously produced E G F and T G F a in human colon cancer cells (Huang et af., 1991). Cell lines Human colon-cancer cell lines FET, JVC, RCA, CBS, C E O and HCT116 were obtained from Dr. M.G. Brattain, Baylor College of Medicine. The biological properties of these cells have been well characterized (Brattain et al., 1984: Mulder and Brattain, 1989). SW480 and HT29 cell lines were obtained from the ATCC (Rockville, MD). All cells were cultured in supplemental McCoy's medium containing 5% FBS at 37°C in a humidified COz incubator.
'To whom correspondence and reprint requests should be addressed, at the Division of Laboratory Medicine, Box 073, M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030, USA. Received: August 3, 1992.
979
EGF, TGFa AND COLON CELL PROLIFERATION
Qitantificaiion of EGF and TGFa secretion To measure secreted EGF and TGFa, the media of exponentially growing cells in 25-cm’ culture flasks were removed. The cells were then washed thoroughly several times with serumfree medium and cultured for another 3 days in serum-free medium. The serum-free conditioned media were then harvested for the determination of E G F and T G F a concentrations. E G F was quantified by an enzyme-linked immunosorbent assay (ELISA) inhibition method as described previously (Huang et al., 1991). Briefly, serum-free conditioned media wcre used to inhibit binding of anti-EGF antibody to insolubilized E G F in the ELISA wells. The amounts of E G F in the media were estimated from a calibration curve using varying amounts of E G F to inhibit binding of anti-EGF antibody to insolubilized EGF. The amounts of E G F in media were then expressed as ng EGF/mg cell lysate protein. T G F a was quantified with a T G F a radioimmunoassay kit (Biomedical Technologies) according to the manufacturer’s instructions. RNA analyses by Northern hybridization Northcrn hybridization was performed as previously described (Huang et al., 1991) using a 32P-labeled synthetic anti-sense oligonucleotide (20 bp corresponding to the 5’ end of the coding sequence of the mature human E G F protein) as a probe. A 32P-labeled GAPDH cDNA was used as a control probe (Huanget al., 1991).
culture plates. Cells were briefly washed with an acid-salt solution (0.2 M acetic acid, 0.5 M NaCI) to remove endogenous E G F bound to its cell-surface receptors before the addition of radiolabeled E G F in the binding assay. Data were analyzed by the Scatchard method using a microcomputer and the LIGAND program, and were evaluated for best fit in a 2-site ligand-binding model. RESULTS
EGF and TGFa secretion The amounts of E G F and T G F a secreted into 3-day conditioned media by 8 human colon-cancer cell lines are shown in Table I. Three of the cell lines (JVC, SW480 and C E O ) secreted negligible amounts of E G F (1.0-1.7 ng/mg cell protein) by comparison with FET, RCA, CBS, HT29 and HCTl16 cells, which secreted 81.8-167.5 ng EGFimg cell protein. The HCT116 cell line secreted the most E G F (Table I). Northern hybridization analyses showed that all the cells expressed E G F mRNA (Fig. la). The levels of E G F mRNAs were quantified by densitometric scanning of the autoradiograph and expressed as ratios of the amount of E G F to GAPDH probes bound (Fig. 1b). FET cells expressed the greatest amount of E G F mRNA and HT29 the next greatest. The amount of TGFu secreted by these cell lines was considerably lower than the amount of E G F secreted (Table I). The concentrations of T G F a secreted ranged from 0.8 to 9.4 ngimg cell protein, with the JVC cell line secreting the most.
Responses to treatment with exogenous EGF and TGFa Proliferative responses The proliferative responses of the cell lines to treatment Proliferative responses to treatment with EGF, TGFa, with E G F and T G F a were examined next. Figure 2 shows that anti-EGF antibody and anti-EGF receptor antibody were measured as described previously (Huang et a/., 1991) with one the cells could be broadly classified into 3 groups, based on modification. The assays were performed in 96-well tissue- their proliferative responses to EGF: (1) responsive (FET and culture plates, and the number of cells in the wells was JVC); (2) intermediately responsive (RCA and CBS); and (3) estimated by the colorimetric 3-(4,5-dimethylthiazoI-2-yl)-2,5- relatively unresponsive (SW480, HT29, G E O and HCTll6). diphenyltetrazolium bromide (MTT) assay (Carmichael et al., Both the F E T and JVC cell lines responded to E G F treatment 1987; Fan et al., 1989; Heo ef al., 1990). Briefly, 2,500 cells in a dose-dependent manner. Most responsive was the FET (counted in a hemocytometer) per well were seeded into cell line, which showed a maximum increase in growth of over tissue-culture plates and allowed to attach overnight. The 60% compared to untreated control cells over a 4-day culture media (McCoy’s medium containing 5% FBS) were then period (Fig. 2a). Figure 3 shows the proliferative responses of these cell lines replenished with fresh media containing EGF, TGFa, antiE G F antibody, or anti-EGF-receptor antibody in the concen- to TGFa. These cells could be broadly classified into 2 groups: trations indicated in “Results”. The media in control wells (1) responsive (FET, JVC, and RCA) and (2) relatively were replenished with fresh media without growth factors or unresponsive (CBS, SW480, HT29, G E O and HCTl16). A antibodies. Preliminary experiments showed that, under iden- maximum increase in growth of 30% relative to that of tical conditions using saturating concentration of the anti-EGF untreated control cells over a 4-day culture period was obor anti-EGF receptor antibodies (indicated in Figs. 4 and 5 ) , served in the most responsive FET and JVC cells (Fig. 3). the binding of I2’I-EGF to the Moser cells was completely Overall, the cell lines that responded to exogenous E G F also blocked (data not shown). An irrelevant MAb to nucleolar responded to exogenous TGFa. protein B23 (Chan et al., 1989) was also used as control. This antibody had no effect on the cells’ proliferation. TABLE I - EGF AND TGFa SECRETION BY HUMAN COLON-CANCER After 4 days of culture, 40 WI MTT solution (2 mg/ml in CELL LINES PBS) wcre added to each well. The cells were then incubated EGF TGFa Cell line in a humidified COz incubator for 3 hr for development of blue (na/mr cell protein) (nr:mr cell protein) formazan crystals. The solution in the wells was then carefully FET 109.9 f 20.2 3.0 t 0.8 removed by aspiration, and 200 p1 dimethylsulfoxide were JVC 1.7 f 1.1 9.4 f 0.5 added to each well. The culture plates were then gently shaken 7.1 2 1.2 RCA 81.8 f 12.4 on a horizontal shaker until the formazan crystals were CBS 85.4 f 8.6 7.1 2 2.1 completely dissolved. The optical density (OD) of each well SW480 1.6 2 0.3 0.8 2 0.2 was determined with an automatic multi-well plate reader at HT29 116.5 f 21.1 1.1 2 0.3 570 nm. The percentage stimulation or inhibition of growth GEO 1.0 t 0.4 1.3 r 1.1 was determined by the following formula: 5.7 t 2.2 HCT116 167.5 f 39.7 ~~~
[(OD of experimental wells-OD control wells)/ OD control wells] x 100
‘”I-EGF binding This assay was performed exactly as previously described (Huangef al., 1991) using cells in monolayer growth in 24-well
EGF and TGFa secretion into 3-day serum-free conditioned media was quantified as described in “Methods”. EGF (sample volume of 150 pl) was quantified by an ELISA method as described previously (Huang et a/., 1991). TGFa (sample volume of 100 PI) was quantified by a commercially available TGFa RIA kit. This assay was performed according to the instructions provided by the manufacturer.
980
HUANG ET AL.
6
45 1
W
0
cn
FIGURE 1 - Northern-blot analysis of EGF mRNA expression in human colon-cancer cell lines using a ["]P EGF oligonucleotide probe. (a) Twenty micrograms of cellular RNA from each cell line were fractionated by agarose-formaldehyde gel electrophoresis and blotted onto a nitrocellulose membrane. Top panel: autoradiograph of the membrane hybridized with EGF probe. Lower panel: autoradiograph of the same membrane rehybridized with ['?]P GAPDH cDNA probe after the first probe was stripped off with 50% formamide in 0.9 M NaCl and 0.09 M sodium citrate at 65°C for 30 min and the membrane was washed with this salt solution without formamide. Lanes 1 to 8: SW480. RCA, JVC, GEO, HT29, CBS, FET and HCTll6, respectively. (b) Ratios of EGF mRNA to GAPDH mRNA expression. The ratios were determined by densitometric tracing of the autoradiograph.
Proliferafive responses to treatment with anti-EGF and anti-EGF receptor antibodies Since 5 of the cell lines (FET, RCA, CBS, HT29 and HCTll6) produced relatively high levcls of EGF, anti-EGF and anti-EGF-receptor antibody-blocking experiments were performed to assess the role of E G F in autocrine or paracrine stimulation of these cells. Figure 4 shows that anti-EGF antibody blocked the proliferation of FET, RCA and CBS cells in a dose-dependent manner, while the other cell types were relatively unresponsive to treatment with the anti-EGF antibody. The JVC cells, which produced only minimal amounts of EGF, responded poorly to treatment with anti-EGF antibody (Fig. 4). We next determined the effect of blocking the binding of both E G F and T G F a to cell-surface E G F receptors on the growth of these cells by using anti-EGF receptor antibody. The FET, RCA and CBS cells (which were the most responsive to anti-EGF antibody) were also the most responsive to the growth-inhibitory effect of anti-EGF receptor antibody (Fig. 5). A slightly higher growth-inhibitory effect was observed with anti-EGF receptor antibody (48, 35 and 50% inhibition, respectively) than with anti-EGF antibody (30, 28, and 459% inhibition. respectively) at saturating antibody concentrations (Figs. 4 and 5). Since anti-EGF receptor antibody blocked the action of both E G F and TGFa, our data suggest that these cells may utilize both E G F and T G F a as growth stimulators via the cell-surface E G F receptor. The HT29 cells, which secreted little TGFa, were relatively unresponsive to anti-EGF receptor antibody. The JVC cells, which produced a negligible amount of E G F (1.7 ngimg protein) but the highest amount of TGF-a (9.4 ng/mg protein), were relatively unresponsive to the antiproliferative effect of anti-EGF receptor antibody (Fig. 5 ) . Both RCA and HCTl16, which secreted high levels of EGF, also responded poorly to anti-EGF receptor antibody (Fig. 5).
EGF receptors Since extracellular E G F and T G F a mediate their action through binding to the cell-surface E G F receptors, we analyzed E G F receptor expression on the surfaces of these colon-cancer cells by radiolabeled ligand binding assays and Scatchard plots. Figures 6 and 7 show the Iz51-EGFbinding profiles and Scatchard plots of these cells respectively. Table I1 summarizes the results of these E G F binding studies. Both high- and low-affinity E G F receptor subtypes were detected in 7 of the cell lines. In the HT29 line, only the low-affinity subtype of E G F receptor was detected. DISCUSSION
EGF-like and T G F a activities were originally identified in human colon-cancer cells by a combination of biological and radioreceptor assays (Coffey et al., 1986; Hanauske et al., 1987; Anzanzo et al., 1989). Since both E G F and T G F a bind to the same cell-surface E G F receptor (Donaldson et a/., 1989), the radioreceptor assay cannot distinguish E G F from TGFa. The secretion of a novel class of EGF-related molecules, cripto and amphiregulin, by human colon cancer cells has recently been reported (Ciardiello et al., 1991). The cripto and amphiregulin mRNAs (2.2 and 1.4 kb respectively) are much smaller than E G F mRNA, with a size of 4.6 to 4.8 kb (Ciccodicola et al., 1989; Plowman et al.. 1990). In this study, we used an EGF-specific ELISA assay and Northern hybridization employing an E G F oligonucleotide probe to demonstrate that human colon-cancer cells expressed EGF mRNA and secreted EGF. However, a wide range of E G F concentrations was secreted by these cells, ranging from 1 to 167 ng/mg cell protein over a 3-day culture period. These cells also secreted TGFa, albeit much less than E G F (1 to 9 ng/mg cell protein). Not all cells that secreted high levels of E G F expressed high levels of E G F mRNA. Most notable was
98 1
EGF. TGFa AND COLON CELL PROLIFERATION
A
JVC
B
FET
T
0 001
0.01
0.1
1
5
10
0.001
50
0.01
0
1
5
1
lo
50
100
D
RCA
C
'=er
U
0.001
5
z
0.01
0.1
1
5
10
0.001
50 100
E
SW480
oo