Int. J . Cancer: 46, 935-943 (1990) 0 1990 Wiley-Liss, Inc.
Publication of the International Union Against Cancer Publication de I‘Union lnternationale Contre le Cancer
EPIDERMAL GROWTH FACTOR RECEPTORS AND EGF-RESPONSIVENESS OF THE HUMAN BREAST-CARCINOMA CELL LINE PMC42 Paul MONAGHAN, Michael G. O R ~ ~ ; R and O D Michael J. O’HARE Institute of Cancer Research, Haddow Laboratories, Cotswold Road, Sutton, Surrey, SM2 5NG, U . K . PMC42 is an early-passage, well-differentiated, pleiomorphic human breast-cancer cell line. It expresses an average of 2.4 X lo5 EGF receptor (EGFR) sites per cell when cultures are assayed using a radioreceptor assay. This is a relatively high figure for breast-carcinoma cell lines. I t s responses t o epidermal growth factor (EGF) were analyzed morphologically and by flow cytometry and immunocytochemistry. PMC42 responded in vitro t o EGF w i t h morphological changes, inhibition of “doming” and an increase in DNA synthesis in serum-containing medium in confluent or nearconfluent cultures. When EGF receptor sites were localized immunocytochemically in PMC42, they were found t o be variably expressed on the membrane of cells grown in the absence of EGF, with discrete areas of strong and weak cells of different morphological appearances; flow cytometry indicated a 10-fold range in EGFR levels between individual cells. Cultures maintained in the presence of EGF were less heterogeneous in their morphology and had a predominantly cytoplasmic, relatively uniform localization of EGFR with downregulation of membrane EGFR Revels. EGF-stimulated cultures were stained simultaneously for the presence of EGFR and the proliferation-related antigen Ki-67 and analyzed by both flow cytometry and immunocytochemistry. No correlation was observed between Ki-67 positivity and level 6f EGFR expression by individual cells. Electron microscopic localization of receptor sites demonstrated that the receptors were concentrated predominantly on the lateral and basal membranes. These results show that the PMC42 line reflects a number of the properties of heterogeneous breast tumours in situ. It may also continue t o express growth-factor receptors in a manner that reflects their topographical distribution on normal luminal breast epithelial cells.
Epidermal growth factor (EGF) is a polypeptide whose ability to affect growth and differentiation in a variety of tissues is mediated by a specific cell-membrane receptor (EGFR) (Carpenter and Zendegui, 1986). As is often stated, the precise physiological functions of EGF in vivo remain to be determined, although evidence from several different tissues suggests that it may have a role in regulating both growth and differentiation. Thus. in the presence of EGF, DNA synthesis and cell division are enhanced in vivo in ectodermal (Carpenter, 1981) and endodermal cells (Oka et al., 1983; Goodlad et al., 1985), and proliferation of mesodermal and ectodermal cells is increased in vitro (Gospodarowicz et al., 1978). EGF is present in high concentrations in breast secretions including milk (Carpenter, 1980) and cyst fluid (Jaspar and Franchimont, 1985), where levels are over 100 times higher than in blood, suggesting that it may play a specific role in mammary biology. In cultures of normal rodent mammary epithelial cells, EGF increases cell proliferation (Imagawa et al., 1985) and influences the synthesis of milk proteins (Vonderhaar and Nakhasi, 1986). When implanted in slow-release capsules in the rodent mammary gland in vivo, EGF promotes local lobulo-alveolar development (Vonderhaar, 1987). As circulating levels of EGF in human plasma are very low (630 nm. The peeled away from the resin block, leaving the cells within the peak of the signal from red fluorescence was displayed against block, the space left being refilled with resin, and polymerizaits area and a region delineated which excluded any debris and tion continued for a further 24 hr at 60°C.
EGF RECEPTORS IN PMC42
Vertical sections of cells were cut using a diamond knife on a Reichert OMU4 ultramicrotome and examined without further staining in a Phillips EM400 electron microscope. RESULTS
Effects of EGF on cell morphology
Cultures of PMC42 grown to confluence at early-passage numbers consist of a heterogeneous monolayer of epithelioid cells, the majority of which have a tightly packed polygonal appearance (Type-2 cells). while others are more elongated (Type-3 cells) and a few present a flat, spread appearance (Type-7 cells). Although other morphological types have been
937
described in these cultures (Whitehead et al., 1983a), these are the most obvious. Once confluence had been reached, small “domes” of tightly-connected epithelial cells are formed over a 7- to 10-day period (Fig. l a ) , typically 20-50 domes/cm2. The addition of EGF to cultures results in the discrete areas of polygonal epithelial cells becoming converted within 48 hr into an irregular and more elongated shape, which results in the cultures assuming a more uniform appearance (Fig. lb). The doming seen in control cultures when they reach confluence does not occur in EGF-treated cultures of comparable age, (lo5 siteskell) and responds positively in vitro in serumcontaining medium had been identified, although other lines do express EGFR in widely differing levels between individual lines. Thus, Fitzpatrick et al. (19846) found 14 breast-tumour cell lines which had detectable levels of EGF binding. The number of receptor sites measured varied from 2 X lo2 to 1.9 X lo6 sitedcell (Imai et al., 1982; Fitzpatrick et al., 19846; Filmus et al., 1985). Furthermore, there was no correlation between the mean levels of receptors on the cells and the proliferative responses to the growth factor in vitro (Imai et al., 1982; Osborne et al., 1980; Fitzpatrick et al., 19846). The MDA-468 line, which has the highest mean EGFR levels of any breast-cancer line yet measured (1.9 X lo6 siteskell) is actually growth-inhibited by EGF (Filmus et al., 1985). The extent of specific binding of lZ5I-EGF to cultures of PMC42, as measured in this study, corresponds to 240,000 sites per cell (Fig. 2). This calculation gives the average number of binding sites over all the cells in culture and EGFR has been quantified in this manner in the majority of previous studies. The localization of EGF receptors in PMC42 with the antibody EGFRl indicates that a figure for the average number of sites per cell is misleading. Quantitative data from flow cytometric analysis indicted that the distribution of EGFR was continuous and heterogeneous, with a 10-fold range of reactivities (Fig. 4). The number of binding sites on some of the positive cells in cultures of PMC42 must, therefore, be considerably greater and may approach or exceed the levels of receptors (>106/cell), which is the average level observed in MDA-468 (Filmus et al., 1985) and some human squamouscarcinoma cell lines (Cowley et al., 1984). This observation raises the possibility of a corresponding heterogeneous proliferative response to the growth factor, as significant growth responses measured by flow cytometry were detected only in cultures at or near confluence (Tables I and 11). Immunocytochemistry showed that the EGFR-positive cells in PMC42 cultures were not distributed at random, but were clustered in areas of strong reactivity which were often sharply demarcated from adjacent cells (Fig. 5). However, the flow cytometric and immunocytochemical results with the Ki-67 proliferationrelated antigen provided no evidence that the levels of EGFR on individual cells influenced their proliferative response to the growth factor; all cells in such cultures appeared equally likely to divide in response to EGF. Levels of EGFR on normal human breast cells are low, in comparison with PMC42, and they have not been unequivocally localized to specific cell types or subcellular sites at the
light-microscopic level (Gusterson er al., 1984; Damjanov et al., 1986). However, some indication of a possible alternative role for the receptor in breast tissue comes from its ultrastructural localization in the tumour-cell cultures. Positive-staining PMC42 cells show a distinct epithelial morphology (Monaghan et al., 1985), and are similar to breast luminal epithelial cells in vivo (Smith et al., 1985). The localization of EGF receptors on the lateral and basal regions of these epithelial-like cells (Fig. 5) may reflect their localization in normal human breast tissue. This position would allow the epithelial cells to bind exogenous EGF arriving from the blood, possibly to concentrate it from the plasma for secretion into the milk (Gresik et al., 1984). Is the lack of correlation between growth responses and EGFR levels seen in the PMC42 line also representative of likely responses of breast tumours in vivo? Several studies have shown that a significant proportion (3650%)of human breasttumour biopsies specifically bind lZ5I-EGF(Fitzpatrick et al., 1984a; Skoog et al., 1986; Sainsbury et al., 1985; Perez et al., 1984). However, when the EGFRl MAb (Waterfield et al., 1982) is used in an immunochemical assay, only a small proportion (14%) of breast tumour biopsies show strong positive staining (B .A. Gusterson, personal communication), although a larger percentage (
Publication of the International Union Against Cancer Publication de I‘Union lnternationale Contre le Cancer
EPIDERMAL GROWTH FACTOR RECEPTORS AND EGF-RESPONSIVENESS OF THE HUMAN BREAST-CARCINOMA CELL LINE PMC42 Paul MONAGHAN, Michael G. O R ~ ~ ; R and O D Michael J. O’HARE Institute of Cancer Research, Haddow Laboratories, Cotswold Road, Sutton, Surrey, SM2 5NG, U . K . PMC42 is an early-passage, well-differentiated, pleiomorphic human breast-cancer cell line. It expresses an average of 2.4 X lo5 EGF receptor (EGFR) sites per cell when cultures are assayed using a radioreceptor assay. This is a relatively high figure for breast-carcinoma cell lines. I t s responses t o epidermal growth factor (EGF) were analyzed morphologically and by flow cytometry and immunocytochemistry. PMC42 responded in vitro t o EGF w i t h morphological changes, inhibition of “doming” and an increase in DNA synthesis in serum-containing medium in confluent or nearconfluent cultures. When EGF receptor sites were localized immunocytochemically in PMC42, they were found t o be variably expressed on the membrane of cells grown in the absence of EGF, with discrete areas of strong and weak cells of different morphological appearances; flow cytometry indicated a 10-fold range in EGFR levels between individual cells. Cultures maintained in the presence of EGF were less heterogeneous in their morphology and had a predominantly cytoplasmic, relatively uniform localization of EGFR with downregulation of membrane EGFR Revels. EGF-stimulated cultures were stained simultaneously for the presence of EGFR and the proliferation-related antigen Ki-67 and analyzed by both flow cytometry and immunocytochemistry. No correlation was observed between Ki-67 positivity and level 6f EGFR expression by individual cells. Electron microscopic localization of receptor sites demonstrated that the receptors were concentrated predominantly on the lateral and basal membranes. These results show that the PMC42 line reflects a number of the properties of heterogeneous breast tumours in situ. It may also continue t o express growth-factor receptors in a manner that reflects their topographical distribution on normal luminal breast epithelial cells.
Epidermal growth factor (EGF) is a polypeptide whose ability to affect growth and differentiation in a variety of tissues is mediated by a specific cell-membrane receptor (EGFR) (Carpenter and Zendegui, 1986). As is often stated, the precise physiological functions of EGF in vivo remain to be determined, although evidence from several different tissues suggests that it may have a role in regulating both growth and differentiation. Thus. in the presence of EGF, DNA synthesis and cell division are enhanced in vivo in ectodermal (Carpenter, 1981) and endodermal cells (Oka et al., 1983; Goodlad et al., 1985), and proliferation of mesodermal and ectodermal cells is increased in vitro (Gospodarowicz et al., 1978). EGF is present in high concentrations in breast secretions including milk (Carpenter, 1980) and cyst fluid (Jaspar and Franchimont, 1985), where levels are over 100 times higher than in blood, suggesting that it may play a specific role in mammary biology. In cultures of normal rodent mammary epithelial cells, EGF increases cell proliferation (Imagawa et al., 1985) and influences the synthesis of milk proteins (Vonderhaar and Nakhasi, 1986). When implanted in slow-release capsules in the rodent mammary gland in vivo, EGF promotes local lobulo-alveolar development (Vonderhaar, 1987). As circulating levels of EGF in human plasma are very low (630 nm. The peeled away from the resin block, leaving the cells within the peak of the signal from red fluorescence was displayed against block, the space left being refilled with resin, and polymerizaits area and a region delineated which excluded any debris and tion continued for a further 24 hr at 60°C.
EGF RECEPTORS IN PMC42
Vertical sections of cells were cut using a diamond knife on a Reichert OMU4 ultramicrotome and examined without further staining in a Phillips EM400 electron microscope. RESULTS
Effects of EGF on cell morphology
Cultures of PMC42 grown to confluence at early-passage numbers consist of a heterogeneous monolayer of epithelioid cells, the majority of which have a tightly packed polygonal appearance (Type-2 cells). while others are more elongated (Type-3 cells) and a few present a flat, spread appearance (Type-7 cells). Although other morphological types have been
937
described in these cultures (Whitehead et al., 1983a), these are the most obvious. Once confluence had been reached, small “domes” of tightly-connected epithelial cells are formed over a 7- to 10-day period (Fig. l a ) , typically 20-50 domes/cm2. The addition of EGF to cultures results in the discrete areas of polygonal epithelial cells becoming converted within 48 hr into an irregular and more elongated shape, which results in the cultures assuming a more uniform appearance (Fig. lb). The doming seen in control cultures when they reach confluence does not occur in EGF-treated cultures of comparable age, (lo5 siteskell) and responds positively in vitro in serumcontaining medium had been identified, although other lines do express EGFR in widely differing levels between individual lines. Thus, Fitzpatrick et al. (19846) found 14 breast-tumour cell lines which had detectable levels of EGF binding. The number of receptor sites measured varied from 2 X lo2 to 1.9 X lo6 sitedcell (Imai et al., 1982; Fitzpatrick et al., 19846; Filmus et al., 1985). Furthermore, there was no correlation between the mean levels of receptors on the cells and the proliferative responses to the growth factor in vitro (Imai et al., 1982; Osborne et al., 1980; Fitzpatrick et al., 19846). The MDA-468 line, which has the highest mean EGFR levels of any breast-cancer line yet measured (1.9 X lo6 siteskell) is actually growth-inhibited by EGF (Filmus et al., 1985). The extent of specific binding of lZ5I-EGF to cultures of PMC42, as measured in this study, corresponds to 240,000 sites per cell (Fig. 2). This calculation gives the average number of binding sites over all the cells in culture and EGFR has been quantified in this manner in the majority of previous studies. The localization of EGF receptors in PMC42 with the antibody EGFRl indicates that a figure for the average number of sites per cell is misleading. Quantitative data from flow cytometric analysis indicted that the distribution of EGFR was continuous and heterogeneous, with a 10-fold range of reactivities (Fig. 4). The number of binding sites on some of the positive cells in cultures of PMC42 must, therefore, be considerably greater and may approach or exceed the levels of receptors (>106/cell), which is the average level observed in MDA-468 (Filmus et al., 1985) and some human squamouscarcinoma cell lines (Cowley et al., 1984). This observation raises the possibility of a corresponding heterogeneous proliferative response to the growth factor, as significant growth responses measured by flow cytometry were detected only in cultures at or near confluence (Tables I and 11). Immunocytochemistry showed that the EGFR-positive cells in PMC42 cultures were not distributed at random, but were clustered in areas of strong reactivity which were often sharply demarcated from adjacent cells (Fig. 5). However, the flow cytometric and immunocytochemical results with the Ki-67 proliferationrelated antigen provided no evidence that the levels of EGFR on individual cells influenced their proliferative response to the growth factor; all cells in such cultures appeared equally likely to divide in response to EGF. Levels of EGFR on normal human breast cells are low, in comparison with PMC42, and they have not been unequivocally localized to specific cell types or subcellular sites at the
light-microscopic level (Gusterson er al., 1984; Damjanov et al., 1986). However, some indication of a possible alternative role for the receptor in breast tissue comes from its ultrastructural localization in the tumour-cell cultures. Positive-staining PMC42 cells show a distinct epithelial morphology (Monaghan et al., 1985), and are similar to breast luminal epithelial cells in vivo (Smith et al., 1985). The localization of EGF receptors on the lateral and basal regions of these epithelial-like cells (Fig. 5) may reflect their localization in normal human breast tissue. This position would allow the epithelial cells to bind exogenous EGF arriving from the blood, possibly to concentrate it from the plasma for secretion into the milk (Gresik et al., 1984). Is the lack of correlation between growth responses and EGFR levels seen in the PMC42 line also representative of likely responses of breast tumours in vivo? Several studies have shown that a significant proportion (3650%)of human breasttumour biopsies specifically bind lZ5I-EGF(Fitzpatrick et al., 1984a; Skoog et al., 1986; Sainsbury et al., 1985; Perez et al., 1984). However, when the EGFRl MAb (Waterfield et al., 1982) is used in an immunochemical assay, only a small proportion (14%) of breast tumour biopsies show strong positive staining (B .A. Gusterson, personal communication), although a larger percentage (
