The Prostate 16:209-218 (1990)
Production of Epidermal Growth Factor and Transforming Growth Factor-a by the Androgen-Responsive LNCaP Human Prostate Cancer Cell Line Jeanne M. Connolly and David P. Rose Division of Nutrition and Endocrinology, Naylor Dana Institute for Disease Prevention, American Health Foundation, Valhalla, New York Epidermal growth factor (EGF)-related polypeptides may be involved in the growth of human prostate cancer cells, and in the androgen stimulation of hormone-responsive prostatic carcinomas. We have shown that androgen-responsive LNCaP cells, like the autonomous DU 145 human prostate cancer cell line, synthesize and secrete EGF and related polypeptides, including immunoreactive transforming growth factor-a (TGF-a). As determined by radioimmunoassay , intracellular EGF levels were approximately I 0 0 times those of TGF-a, but together these accounted for less than half of the total EGF-like polypeptides detected in a radioreceptor assay. Although LNCaP cell growth was stimulated by dihydrotestosterone (DHT), there was no evident effect on immunoreactive EGF levels in the medium after correction for cell number. Moreover, metabolic labeling experiments showed no effect of the androgen on EGF synthesis by LNCaP cells. Gel filtration chromatography of conditioned medium showed both high molecular weight species and the mature 6,000 dalton form of immunoreactive EGF. We conclude that although LNCaP prostate cancer cell growth is stimulated by DHT, it is unlikely that this is mediated directly via increased EGF synthesis by the tumor cells. Key words: prostatic carcinoma, dihydrotestosterone, autocrine
A variety of tumor types have been shown to secrete epidermal growth factor (EGF)-related polypeptides, including cell lines derived from carcinomas of the breast [ 1-41, colon [5 -71, ovary , lung , and prostate [ lo]. Although, in general the secreted polypeptides appear to be various molecular forms of transforming growth factor-a (TGF-a), and not immunoreactive EGF [ 1,7], we have reported previously that the DU145 human prostate cancer cell line is an exception in that it secretes both immunoreactive TGF-a and EGF into serum-free medium [ 101. Production of EGF-related polypeptides by the estrogen-responsive MCF-7 breast cancer cell line is stimulated in culture by the presence of estradiol [2,3]; and in ovariectomized athymic mice the requirement of these cells for estrogens is at least partially circumvented by infusion of culture medium into which the same cell line has secreted these growth factors . Thus, it appears that secretion of TGF-a may
Received for publication September 8, 1989; accepted December 28, 1989. Address reprint requests to D.P. Rose, Division of Nutrition and Endocrinology, Naylor Dana Institute for Disease Prevention, Valhalla, NY 10595. 0 1990 Wiley-Liss, Inc.
Connolly and Rose
provide an autocrine mechanism for growth regulation of breast cancer cells, and that the growth response to estrogens may be mediated, at least in part, by way of TGF-a mitogenicity . While growth of the DU 145 prostate cancer cell line is unaffected by androgens [ 111, the LNCaP cell line is androgen-responsive [ 12,131. We have, therefore, compared production of immunoreactive TGF-a and EGF by the DU 145 and LNCaP prostate cancer cell lines, and examined the androgen-responsive line for evidence that dihydrotestosterone (DHT) stimulates production of the EGF-related polypeptides.
MATERIALS AND METHODS Reagents Unlabeled human EGF (hEGF) and I2'I-hEGF for the radioreceptor assay (RRA) of EGF-like polypeptides and hEGF radioimmunoassay (RIA) and anti-hEGF antiserum were purchased from Biomedical Technologies (Stoughton, MA) and 'Hthymidine from ICN Radiochemicals (Irvine, CA). The reagents for the TGF-a RIA were obtained from Peninsula Laboratories (Belmont, CA). The 3'S-methionine, > I ,OOO mCi/mmole, was purchased from Amersham Corp. (Arlington Heights, IL), and the Ab-1 EGF antibody with protein A agarose for immunoprecipitation after metabolic labeling was obtained from Oncogene Science (Manhasset, NY). All of the antibodies used were monoclonal in origin. Cell Culture
The LNCaP and DU 145 cell lines were obtained from the American Type Culture Collection and cultured routinely in phenol red-containing-RPMI- 1640 medium (Gibco, Grand Island, NY) with 5% fetal bovine serum (FBS; Gibco) plus 100,OOO u n i t d l penicillin and 100 mg/L streptomycin as previously described [lo]. An experiment was performed to validate the stimulatory effect of DHT on LNCaP cell growth in the serum-free medium employed to obtain secreted EGF-like growth factors. The cells were plated at an initial density of 1 X lo5 cells/well (3 ml/well) into 6-well plates (Costar, Cambridge, MA), cultured for 24 h with FBSsupplemented RPMI-1640 medium, washed with RPMI- 1640, and refed with RPMI1640 containing ITS + Premix (Collaborative Research, Lexington, MA). This preparation provides 6.25 pg/ml of insulin and transferrin, 6.25 ng/ml of selenous acid, 1.25 mg/ml of bovine serum albumin (BSA), and 5.35 pg/ml of linoleic acid. After 24 h, the serum-free medium was replaced as before, but the DHT (2-80 ng/ml), dissolved in ethanol, was added to triplicate wells. The final ethanol concentration in all wells, including the controls, was 0.01%. After 2 days of incubation at 37"C, the cells were harvested by trypsinization and counted in a Coulter Counter [lo]. Preparation of Conditioned Medium (CM)
The LNCaP or DU 145 cells were plated into T-150 flasks containing 25 ml of FBS-supplemented medium, and grown to approximately 60% confluence. The medium was then replaced with RPMI-1640, penicillin, and streptomycin and ITS+ Premix. After a further 48 h incubation, the medium was replaced by fresh RPMI1640 plus ITS + Premix. Forty-eight hours later, at confluence, the media from all flasks were pooled, aprotonin (Sigma) was added to a final concentration of 0.2 TIU/ml, and the CM was centrifuged at approximately 600g for 15 min to remove
Growth Factors and LNCaP Tumor Cells
cellular debris. It was then passed aseptically through a Nalgene (Rochester, NY) 0.22 p m filter unit and stored at 4°C. The CM was subsequently dialyzed for 24 h, twice with 3 L of 1 M acetic acid and once with distilled water, prior to concentration by Amicon filtration (YM5 membrane, Amicon Corp., Lexington, MA; cut-off Mr 5 ,OOO), followed by lyophilization. Preparation of Homogenates for Growth Factor Assays
When the conditioned medium had been decanted, 2 ml of phosphate-buffered saline (PBS), without calcium or magnesium, was added to each flask. A disposable cell scraper and a 5 ml pipette were used to harvest the cells into a 50 ml centrifuge tube, and an aliquot set aside for counting in a Coulter Counter. After centrifugation for 15 min at 600g and 4"C, the cell pellet was stored at -70°C. Prior to assay, the cells were homogenized in a Teflon-glass homogenizer at 4°C with 4 volumes of chilled buffer (0.1 M NaCl, 5 mM MgCl,, 1% Nonidet P-40,0.5% Na deoxycholate, 2 KIU/ml bovine aprotonin, 20 mM Tris-HCI, pH 7.4). The homogenates were centrifuged at 10,OOOg for 20 min at 4"C, and the supernatants removed for the EGF-like polypeptide RRA, EGF, and TGF-a RIAs, and determination of total protein by the method of Lowry et al. . Radioassays
The RRAs and RIAs were performed as previously described [ 101. The TGF-a RIA has cross-reactivity with hEGF of less than 1%, with a sensitivity of 20 pg/assay tube. The hEGF RIA has a sensitivity of 0.1 ng/ml, an interassay coefficient of variation of 14%, and no demonstrable cross-reactivity with TGF-a. Gel Filtration Chromatography Medium conditioned by LNCaP cells was fractionated on a 2.6 x 70 cm Bio-Gel P-60 column (Bio-Rad Laboratories, Richmond, CA). A 2.0 ml volume of 100-fold concentrated, dialyzed CM was chromatographed with 1 M acetic acid at a flow rate of 10 ml/h. Fractions of 3.3 ml were collected, lyophilized, and each resuspended in 0.5 ml PBS for hEGF RIA. Metabolic Labeling of EGF
LNCaP cells were plated in 75 cm2 culture flasks in RPMI-1640 plus 5% FBS medium. After 24 h of culture, and a 1 h wash with methionine-free RPMI-1640 containing 3.15 pg/ml of insulin and 1.25 mg/ml of delipidized BSA (Collaborative Research), the cells were incubated in the same medium plus 700 pCi 35S-methionine/flask for 20 h. Parallel cultures were maintained with and without 40 ng/ml of DHT throughout the incubation procedures. Cell number was determined by counting per 100 x microscopic field. The cells were then washed, lysed, and immunoprecipitated with Ab-1 EGF antibody and protein A agarose according to the manufacturer's suggested procedure. The precipitates were analyzed by SDS-polyacrylamide (7.5%) gel electrophoresis (SDS-PAGE) followed by autoradiography with a 4 day exposure at -75°C. Molecular weight markers for PAGE (BRL, Gaithersburg, MD) were run with the immunoprecipitates and visualized with Coomassie blue. The 35S-labeled immunoprecipitates in the gel were subsequently quantified utilizing an Ambis Radioanalytic Imaging System (Ambis Systems, San Diego, CA).
TABLE 1. Effect of Dihydrotestosterone(DHT) on the Growth of LNCaP Prostate Cancer Cells Cultured in Serum-Free Medium* DHT concentration (ndml) 0" 2 4
10 20 40 60 80
Cell No. x lo4
22.8 2.8 34.0 2 0.5 36.0 f 0.2 37.1 f 1.4 25.7 f 2. I 41.0 2 1.7 36.0 & 1.1 37.6 2 1.8
*RPMI-1640 plus ITS+ (insulin 6.25 pg/ml, transferrin 6.25 pg/ml, selenous acid 6.25 nglml. BSA 1.25 mg/ml, linoleic acid 5.25 pg/ml). "With addition of 0.01% ethanol (DHT vehicle).
RESULTS Effect of DHT on LNCaP Cell Growth
The results of an experiment to confirm the stimulatory effect of DHT on the growth of LNCaP cells when cultured in serum-free medium supplemented with ITS are shown in Table I. The androgen increased cell number over the entire range of concentrations evaluated, with a peak effect at 40 ng/ml. This concentration was, therefore, selected for the subsequent experiments.
lntracellular and Secreted EGF and Related Proteins Studied by RRA and RIA
The intracellular content of EGF-like polypeptides determined by RRA and the two RIAs was similar for the two cell lines. The levels of EGF, expressed in terms of total protein, were approximately 100-fold those of TGF-a, but more than 50% of the EGF receptor-binding activity detected by the RRA was not accounted for in the RIAs (Table 11). In contrast to the similarity in the intracellular levels, medium conditioned by DU 145 cells had a much higher concentration of immunoreactive EGF than that conditioned by LNCaP cells (Table 11). The difference was even more pronounced when the results were expressed in terms of cell number, and when this was done the amounts of TGF-a secreted by DU 145 cells also appeared to be greater. The LNCaP cell line was also cultured in the presence of 40 ng/ml of DHT. As is shown in Table 111, the level of EGF secreted appeared to increase in the presence of DHT, but this was a consequence of the high cell density after the initial stages of medium conditioning, and was not apparent when the RIA results were expressed in terms of cell number instead of volume of culture medium. Metabolic Labeling of EGF Synthesized by LNCaP Cells
Metabolic labeling followed by immunoprecipitation was also used to examine the effect of DHT on the synthesis of EGF by LNCaP cells. The 35S-methioninelabeled polypeptides present in lysates prepared from cultures selected for a similar final low density, having a final cell number ratio of 8 5 in the absence and presence of DHT, respectively, were precipitated with hEGF antiserum and separated by SDS-PAGE, and are shown in Figure 1. Three major bands of immunoreactive
Growth Factors and LNCaP Tumor Cells
TABLE 11. Levels (Mean f SD) of EGF-Related Polypeptides in LNCaP and DU 145 Prostate Cancer Cell Homogenates Polypeptide (ng/mg total protein) EGF-related polypeptides determined by RRA lmmunoreactive EGF Immunoreactive TGF-a
DU 145 cells
* 0.74 * 0.47 * 0.003
6.69 1.95 2.27 2 1.15 0.02 2 0.001
5.75 2.56 0.03
TABLE 111. EGF-Related Polypeptides Secreted Into the Culture Medium by LNCaP and DU 145 Prostate Cancer Cells (Mean f SD) Polypeptide Polypeptide (pg/ml culture medium) lmmunoreactive EGF lmmunoreactive TGF-a Polypeptide (pg/106 cells) Immunoreactive EGF Immunoreactive TGF-a
2 1.2 2 0.6
* 1.1 * 0.5
LNCaP cells with DHT"
DU 145 cells
10.4 f 1.8 5 . 5 2 1.5
38.0 5.6 4.4 2 1.0
* 1.0 * 0.9
70.4 f 10.5 8.2 2 1.9
"The LNCaP cells were cultured in serum-free medium with 40 ng/ml of DHT.
polypeptide, with molecular weights above 200,000 daltons, and of approximately 65,000 and 55,000 daltons, were present in both the lysates from cells cultured with and without DHT. The total 35S-labeledEGF immunoprecipitates prepared from cells cultured in the absence and presence of DHT had a net cpm ratio of 8 5 as quantified by radioanalytic imaging. This ratio reflected exactly the ratio of cell number per treatment used for immunoprecipitation, again indicating that the androgen had no effect on the total amount of EGF synthesized by LNCaP prostate cancer cells, although a shift in distribution of species size was observed (Table IV). Immunoprecipitates from cultures containing DHT showed a lower proportion of immunoreactive EGF at 90,000 and 160,000 daltons than did those without DHT, with a converse increase in relative quantities of 55,000 and >200,000 molecular weight species (14% shifts in production). No polypeptide corresponding to the mature 6,000 dalton EGF was detected in this experiment. Gel Filtration Chromatography
The EGF RIA of fractions of LNCaP-conditioned medium separated by Bio-Gel P-60 gel filtration chromatography showed two peaks of immunoreactive polypeptide, one of approximately 6,000 daltons, and the other in the 10-25,OOO dalton molecular weight range (Fig. 2). DISCUSSION
Others have reported previously that the LNCaP human prostate cancer cell line possesses specific binding sites for EGF, and that its growth in culture is stimulated by this growth factor , TGF-a , and also by androgens [12,13]. The present study confirms the stimulation of LNCaP cell growth by DHT, which, under the culture conditions used for our experiments was optimal at a concentration of 40
Connolly and Rose
200 , 97, 68,
Fig. 1. Separation by SDS-PAGE of metabolically labeled immunoprecipitable EGF molecular species present in lysates from LNCaP cells after culture without added DHT (lane A), and in the presence of DHT, 40 ng/ml (lane B). The numbered arrows indicate the location and molecular weight (in kilodaltons) of standard markers.
TABLE IV. Distribution of "S-Labeled Immunoprecipitates From Cells Cultured in the Absence and Presence of Dihydrotestosterone (DHT) Without DHT Cell No. lmmunoprecipitate quantitation Net cpm/lane Major MW species (kD) 2200 160
5.8 x lo5
8 of total imrnunoprecipitate cpm
23 9 9 8 12
21 8 1 9 IS
ng/ml. As in the case of the hormone-unresponsive DU 145 cell line [lo], we now report that LNCaP cells secrete both immunoreactive hEGF and TGF-a.Moreover, both cell lines secrete the 6,000 dalton monomeric form of EGF, as well as immu-
Growth Factors and LNCaP Tumor Cells Albumin (67000)
Chyrnotrypsinogen Trypsin Insulin Inhibitor (WOO) (6000) A (25000)
Fraction Number Fig. 2. Bio-Gel P-60chromatography of LNCaP cell-conditioned medium. A 2.0 ml volume of 100fold concentrated, dialyzed material was applied to the column and eluted with 1 M acetic acid.
noreactive polypeptides of higher molecular weight into the culture medium. When treated with dimercaptoethanol, approximately two-thirds of the high molecular weight EGF secreted into the medium by DU 145 cells was converted to the 6,000 dalton form, indicating that it was formed by interpolypeptide disulfide bonds . Similar large polypeptides constitute the predominate form of hEGF in plasma [ 171. The concentrations of EGF-related polypeptides in tumor cell homogenates, as determined by RRA, were considerably higher than the combined levels of immunoreactive EGF and TGF-a. Our TGF-a RIAs were performed using rat TGF-a and rat TGF-a antiserum. Although rat and human TGF-a are structurally very similar, and the few differences in amino acids do not appear to be critical for antigenantibody interaction , it is possible that some human prostate cancer cell-produced molecular forms, while capable of binding to the EGF receptor, were not included in the assay. Alternatively, the cells may secrete another, unrecognized, species of EGF receptor-binding polypeptide. A striking difference between the two cell lines was in the level of EGF secreted into the culture medium; when expressed in terms of cell number, the concentration in medium conditioned by DU 145 cells was more than ten times that in LNCaP CM.
Connolly and Rose
Despite this difference, the levels in the cells themselves were similar, suggesting that, at least under the culture conditions existing in our experiments, DU 145 cells do not retain synthesized EGF to the same extent as LNCaP cells, but rather secrete larger amounts into the surrounding environment. These relatively high levels of tumor cell-produced EGF, and to a lesser extent TGF-a, in the medium may explain why we previously observed that exogenous EGF has little or no effect on DU 145 cell proliferation [ 101, whereas LNCaP cell growth is stimulated when the polypeptide is added to the culture medium . In the case of the DU 145 cells, maximal stimulation of growth by the EGF-related polypeptides may be taking place by an autocrine mechanism. A similar distinction applies to human breast cancer cell lines; growth of MCF7, and other estrogen-responsive cells, is stimulated in vitro by EGF, whereas the growth factor has no effect on estrogen-independent cell lines [ 191. The secretion of EGF-related polypeptides by cultured MCF-7 cells is increased in the presence of estradiol [2,3]. Estrogen supplementation is necessary for the establishment of MCF7 cells as solid tumors in the athymic nude mouse , but estrogen-free medium conditioned in vitro by estradiol-stimulated MCF-7 cells, when infused into ovariectomized athymic mice, will permit the development of small, but non-proliferative, tumors . However, MCF-7 cells transfected with a TGF-(r cDNA expression vector, secreting levels of a polypeptide greater than those produced by hormoneindependent breast cancer cell lines, still retain a proliferative response to estradiol both in vitro and in vivo . It appears, therefore, that an autocrine loop involving TGF-a alone is not sufficient to account fully for the mitogenic effect of estradiol on MCF-7 cells. In the present study, we used two approaches to examine the effect of DHT on the production and secretion of EGF and TGF-a by the androgen-responsive LNCaP prostate cancer cell line. First, we compared the levels of EGF determined by RIA in CM prepared without added DHT, or with an optimal concentration of the androgen for cell proliferation. When correction was made for the higher cell number resulting from stimulation of growth by the DHT, there was no difference in the EGF levels secreted per cell in medium conditioned with or without the androgen. Second, we performed metabolic labeling experiments followed by immunoprecipitation with an antibody specific for hEGF. Again, there was no apparent difference between total amounts of EGF detected after PAGE of lysates from cells cultured with or without DHT, although minor internal shifts in the proportion of high molecular weight forms of EGF isolated were observed. We conclude from these observations that although growth of the LNCaP cell line is stimulated by both exogenous EGF and androgens, it does not synthesize or secrete this growth factor in significant response to DHT. In consequence, it is unlikely that androgenic growth stimulation is mediated by way of enhanced EGF production. This does not, of course, exclude the possibility that androgen-regulated secretion of other growth factors, including those unidentified but recognized in the RRA, is involved in autocrine stimulation of LNCaP cell growth. Moreover, Schuurmans et al. [ 131 showed that the number of EGF receptor sites on LNCaP cells is increased when these are cultured in the presence of an androgen, and so a stimulatory effect of DHT on growth might be mediated by way of this mechanism. Although the 6,000 dalton monomeric form of EGF was present in medium conditioned by LNCaP cells, none was detected in lysates of cells after metabolic
Growth Factors and LNCaP Tumor Cells
labeling with 35S-methionine. Most likely, this was because the mature form of EGF contains only one methionine per molecule, which would permit labeling with one 3sS atom. Alternatively, the final step in processing the larger molecular precursor forms may occur immediately prior to secretion, so that the 6,000 dalton polypeptide does not accumulate within the cell. Further experiments are being performed with labeled cysteine and methionine, and immunoprecipitation of both lysates and CM, to distinguish between these possibilities. ACKNOWLEDGMENTS We are most grateful to Dr. Joanne Tillotson for her help with the metabolic labeling experiment. This work was supported by a SIC grant from the American Cancer Society. REFERENCES I . Perroteau I, Salomon D, DeBartoli M, Kidwell W, Hazarika P, Pardue R, Dedman J , Tam J: Immunological detection and quantitation of alpha transforming growth factors in human breast carcinoma cells. Breast Cancer Res Treat 7:201-210, 1986. 2. Lippman ME, Dickson RB, Gelmann EP, Rosen N, Knabbe C, Bates S, Brozert D, Huff K, Kasid A: Growth regulation of human breast carcinoma occurs through regulated growth factor secretion. J Cell Biochem 35:1-16, 1987. 3. Dickson RB, MacManaway ME, Lippman ME: Estrogen-induced factors of breast cancer cells partially replace estrogen to promote tumor growth. Science 232: 1540-1543. 1986. 4. Mori K, Kurobe M, Furukawa S, Kubo K, Hayashi K: Human breast cancer cells synthesize and secrete an EGF-like immunoreactive factor in culture. Biochem Biophys Res Commun 136:300305, 1986. 5 . Coffey RJ, Jr, Goustin AS, Soderquist AM, Shipley GD, Wolfshohl J, Carpenter G , Moses HL: Transforming growth factor a and p expression in human colon cancer lines: Implications for an autocrine model. Cancer Res 47:4590-4594, 1987. 6. Watkins LF, Brattain MG, Levine AE: Modulation of a high molecular weight form of transforming growth factor-a in human colon carcinoma cell lines. Cancer Lett 40:59-70, 1988. 7. Anzano MA, Rieman D, Prichett W, Bowen-Pope DF, Greig R: Growth factor production by human colon carcinoma cell lines. Cancer Res 49:2898-2904, 1989. 8. Bauknecht T, Kiechle M, Bauer G, Siebers JW: Characterization of growth factors in human ovarian carcinomas. Cancer Res 46:2614-2618, 1986. 9. lmanishi K, Yamaguchi K, Kuranami M, Kyo E, Hozumi T, Abe K: Inhibition of growth of human lung adenocarcinoma cell lines by anti-transforming growth factor-a monoclonal antibody. JNCI 81~220-223, 1989. 10. Connolly JM, Rose DP: Secretion of epidermal growth factor and related polypeptides by the DU 145 human prostate cancer cell line. The Prostate 15:177-186, 1989. 11. Hartley-Asp B, Gunnarsson PO: Growth and cell survival following treatment with estramustine, nor-nitrogen mustard, estradiol and testosterone of a human prostatic cancer cell line (DU 145). J Urol 127:818-822, 1982. 12. Horoszewicz JS, Leong SS, Kawinski E, Karr JP, Rosenthal H, Chu TM, Mirand EA. Murphy GP: LNCaP model of human prostatic carcinoma. Cancer Res 43:1809-1818, 1983. 13. Schuurmans ALG, Bolt J , Mulder E: Androgens stimulate both growth rate and epidermal growth factor receptor activity of the human prostate tumor cell LNCaP. The Prostate 12:55-63, 1988. 14. Lowry OH, Rosebrough NJ, Fan AL, Randall RJ: Protein measurement with the Fohn phenol reagent. J Biol Chem 193:265-275, 1951. 15. Wilding G, Zugmeier G, Knabbe C, Valverius E, Flanders K, Gelmann E P The role of transforming growth factors a and p in human prostate cancer cell growth. Proc Am Assoc Cancer Res 29:241, 1988. 16. Connolly JM, Rose DP: Secretion of epidermal growth factor (EGF) and related polypeptides by
COMOIIYand Rose androgen-responsive and unresponsive human prostate cancer cell lines. Endocrinology [Suppl] 125:280 (abstr.), 1989. Kurobe M, Furukawa S, Hayashi K: Molecular nature of human epidermal growth factor (hEGF)-like immunoreactivity in human plasma. Biochern Int 12:677-683, 1986. Marquardt H,Hunkapiller MW, H a d LE, Twardzik DR, DeLarco JE, Stephenson JR, Todaro GJ: Transforming growth factors produced by retrovirus-transformed rodent fibroblasts and human melanoma cells: Amino acid sequence homology with epidermal growth factor. Proc Natl Acad Sci USA 80:4684-4688, 1983. Davidson NE, Gelmann EP. Lipprnan ME, Dickson RB: Epidermal growth factor receptor gene expression in estrogen receptor-positive and negative human breast cancer cell lines. Mol Cell Endocrinol 1:216-223. 1987. Shafie SM, Grantham FH: Role of hormones in the growth and regression of human breast cancer cells (MCF-7) transplanted into athymic nude mice. JNCl 6 7 5 - 5 6 , 1981. Clarke R, Briinner N , Katz D, Glanz P, Dickson RB, Lippman ME, Kern FG: The effects of a constitutive expression of transforming growth factor-a on the growth of MCF-7 human breast cancer cells in vitro and in vivo. Mol Cell Endocrinol 3:372-380. 1989.