Epidermal Growth Factor Receptor-Dependent ... - Europe PMC

Molecular Biology of the Cell Vol. 5, 339-347, March 1994

Epidermal Growth Factor Receptor-Dependent Stimulation of Amphiregulin Expression in Androgen-Stimulated Human Prostate Cancer Cells Inder Sehgal,* Jane Bailey,t Karen Hitzemann4 Mark R. Pittelkow4* and Nita J. Maihle*§ Departments of *Molecular Biology and Biochemistry, tGastroenterology, and Dermatology, Mayo Foundation, Rochester, Minnesota 55905 Submitted August 18, 1993; Accepted January 27, 1994 Monitoring Editor: Guido Guidotti

Amphiregulin is a heparin-binding epidermal growth factor (EGF)-related peptide that binds to the EGF receptor (EGF-R) with high affinity. In this study, we report a role for amphiregulin in androgen-stimulated regulation of prostate cancer cell growth. Androgen is known to enhance EGF-R expression in the androgen-sensitive LNCaP human prostate carcinoma cell line, and it has been suggested that androgenic stimuli may regulate proliferation, in part, through autocrine mechanisms involving the EGF-R. In this study, we demonstrate that LNCaP cells express amphiregulin mRNA and peptide and that this expression is elevated by androgenic stimulation. We also show that ligand-dependent EGF-R stimulation induces amphiregulin expression and that androgenic effects on amphiregulin synthesis are mediated through this EGF-R pathway. Parallel studies using the estrogen-responsive breast carcinoma cell line, MCF-7, suggest that regulation of amphiregulin by estrogen may also be mediated via an EGF-R pathway. In addition, heparin treatment of LNCaP cells inhibits androgen-stimulated cell growth further suggesting that amphiregulin can mediate androgen-stimulated LNCaP proliferation. Together, these results implicate an androgen-regulated autocrine loop composed of amphiregulin and its receptor in prostate cancer cell growth and suggest that the mechanism of steroid hormone regulation of amphiregulin synthesis may occur through androgen upregulation of the EGF-R and subsequent receptor-dependent pathways. INTRODUCTION Amphiregulin is an epidermal growth factor' (EGF) family member initially purified from conditioned media of MCF-7 cells treated with 12-0-tetradecanoylphorbol13-acetate (Plowman et al., 1990). This peptide induces variable effects on growth in different cell types. It is an autocrine factor for normal human mammary epithelial cells (Li et al., 1992) and promotes growth of normal fibroblasts and keratinocytes as well as some ovarian and pituitary tumor cell lines. However, amphiregulin has no growth effects on other tumor cell § Corresponding author.

1 Abbreviations used: EGF, epidermal growth factor; EGF-R, epidermal growth factor receptor; TGF-a, transforming growth factora; TPA, 1 2-0-tetradecanoylphorbol- 13-acetate. © 1994 by The American Society for Cell Biology

lines, including MCF-7 breast carcinoma cells (Shoyab et al., 1988) whereas it inhibits the growth of cancer cells lines such as the A431 human vulvar epidermoid carcinoma-derived cell line, breast, cervical, and ovarian carcinoma-derived cell lines (Shoyab et al., 1988), and normal ovarian surface epithelial cells (Johnson et al., 1991). The growth effects of amphiregulin may be concentration dependent because picomolar doses inhibit, whereas nanomolar doses stimulate, ovarian surface epithelial cells (Johnson et al., 1991). In addition to these various growth effects, amphiregulin also demonstrates variability in properties shared with EGF family members. In murine keratinocytes, amphiregulin is equally as potent as EGF in inducing cell proliferation (Shoyab et al., 1989); however it does not induce anchorageindependence in transforming growth factor (TGF-3)339

I. Sehgal et al.

treated normal rat kidney (NRK-SA6) cells, whereas both EGF and TGF-a support this activity (Shoyab et al., 1989; Todaro et al., 1990). In addition, amphiregulin does not fully compete with or displace EGF binding to the EGF receptor (R) (75% maximal displacement) (Shoyab et al., 1989). Differences between amphiregulin and EGF/TGF-a in receptor binding and growth effects may be due to an amino-terminal region consisting of several basic, hydrophilic residues (Lys, Arg) and to the lack of two carboxyl-terminal residues, an Asp and Leu, conserved between other members of the EGF family (Plowman et al., 1990). Substitutions at these positions have been shown to decrease receptor binding and biological activity of TGF-a (Saloman et al., 1990). Amphiregulin has not been shown to interact directly with the gene products of other members of the c-erb B family of receptors including c-erb B 2, 3, or 4 (Plowman et al., 1993), and thus, the EGF-R (c-erb B1/HER 1) is believed to be the sole cell surface receptor for amphiregulin in epithelial cells (Kannan et al., 1992). Heparin sulfate, a sulfated glycosaminoglycan, has been demonstrated to inhibit amphiregulin binding to the EGF-R in human keratinocytes (Cook et al., 1991). Amphiregulin is the only EGF family member inhibited by heparin; this inhibition is probably the result of heparin binding to the peptide's amino terminal hydrophilic residues (Cook et al., 1991; Li et al., 1992). Although amphiregulin expression has been reported in several human tumors, it has not been documented previously to play a role in prostate malignancy. Prostate cancer, the most commonly diagnosed cancer and the second leading cause of cancer death among males in the United States (Boring et al., 1993), is stimulated by androgens, and hormone deprivation is the primary treatment for advanced prostate cancer. Often, the clinical response to androgen withdrawal is temporary and these cancers ultimately recur after which survival is usually less than one year (Gibbons, 1987). It has been suggested that growth factor-mediated autocrine loops, induced by androgens, may contribute to hormone responsive growth and that these autocrine mechanisms may provide the means for cells to survive and continue growth after steroid deprivation (Knabbe et al., 1991). One of the most studied autocrine loops in prostate cancer is the interaction between EGF-R and its ligands. Although relative EGF-R levels are somewhat controversial in human prostate cancer tissues and have been reported higher or lower than in benign prostatic hyperplasia (Ibrahim et al., 1993), important EGF-R autocrine loops have been defined in cultured prostate cancer cells. In the hormoine-responsive tumor cell line LNCaP (Horoszewicz et al., 1983), androgen withdrawal results in the downregulation of the EGF-R (Schuurmans et al., 1988). Although this cell line synthesizes both EGF and TGF-a, these peptides are not regulated by androgen stimulation (Connolly and Rose, 1990). Two androgen-independent prostate cancer lines, the 340

DU-145 and PC-3, also secrete EGF/TGF-a, and express the EGF-R (Connolly and Rose, 1989; Hofer et al., 1991). Several heparin-binding growth factors have been identified in normal and malignant prostate cells (Matuo et al., 1987; Mansson et al., 1989, Shain et al., 1992); however, the 1.4-kilobase (kb) amphiregulin mRNA transcript has been reported to be undetectable in normal prostate tissues (Plowman et al., 1990). Here, we report the expression of amphiregulin message and peptide in both androgen-sensitive and androgen-independent prostate cancer cell lines and demonstrate that amphiregulin expression can be induced by androgenic stimulation. Furthermore, we demonstrate that amphiregulin is a crucial mitogen for maintaining androgen-stimulated tumor growth. Finally, we demonstrate that the mechanism of androgen-induced amphiregulin expression is dependent on an EGF-R-stimulated pathway. MATERIALS AND METHODS

Cell Lines and Culture The androgen-sensitive human prostate cancer cell line, LNCaP, was obtained from Drs. Donald Tindall and Charles Young, Mayo Clinic (Rochester, MN). The human breast cancer cell line, MCF-7, and two androgen-independent prostate cancer cell lines, DU-145 and PC-3, were obtained from American Type Culture Collection (Rockville, MD). LNCaP cells were maintained in RPMI-1640, 2 mM L-glutamine, 100 U/ml penicillin G, 100 jig/ml streptomycin sulfate, 7% heatinactivated fetal bovine serum (FBS), and 10 nM testosterone. PC-3 cells were maintained in Ham's F-12K media, 10% FBS, 100 U/ml penicillin G, 100 Ag/mi streptomycin sulfate, and 2 mM L-glutamine. MCF-7 and DU-145 cells were maintained in Eagle's minimum essential medium (MEM) plus Earle's salts, 10% FBS, 100 U/ml penicillin G, 100 ug/ml streptomycin sulfate, and 2 mM L-glutamine. For experimental assays, all cells were subcultured and allowed to attach for 2 d after which serum-/steroid-free defined media were added (phenol red-free RPMI-1640, 5 ,ug/ml insulin, 10 Ag/ml holo-transferrin, 30 nM Na selenite, 100 U/ml penicillin G, 100 Ag/ml streptomycin sulfate, 2 mM L-glutamine, 1.25 Mg/ml amphotericin B, ±1 nM of the synthetic, nonmetabolized androgen R1881 [New England Nuclear-Dupont, Boston, MA] [Bems et al., 1986], or ±1 nM estradiol 17-fl). TGF-a was used at a concentration of 20 ng/ml, and anti-EGFR monoclonal antibody was used at a concentration of 2 ,ug/ml. Cells were routinely verified free of mycoplasma. Growth factors were obtained from Upstate Biotechnology (Lake Placid, NY), and the EGFR monoclonal antibody 225 (Sato et al., 1983) was obtained from Dr. Hideo Masui, Memorial Sloan Kettering Cancer Center.

Northern (RNA) Analysis Poly A' RNA was isolated from cells grown to various time points in T-175 or T-75 tissue culture flasks as described (Stemnfeld et al., 1988). mRNA was denatured and electrophoretically separated on a 1.2% formaldehyde agarose gel, and samples were transferred to a Zetaprobe GT membrane (Bio-Rad, Richmond, CA) by capillary transfer as described (Ausubel et al., 1989), baked at 80°C for 2 h, prehybridized, hybridized, and washed according to the manufacturer's instructions (Bio-Rad, Zeta probe manual). An 874-basepair (bp) human amphiregulin cDNA fragment was freed by digestion of a cDNA clone (pSelect/AR) provided by Dr. Robert Coffey, Vanderbilt University Medical Center and Dr. Gary Shipley, Oregon Health Sciences University. The 874-bp fragment was purified (Geneclean, Bio 101, La Jolla, CA), used to construct random primed probes (Random Primers Molecular Biology of the Cell

Androgen and EGF-R Induce Amphiregulin DNA Labeling System, Bethesda Research Laboratories, Gaithersburg, MD), and radiolabeled using [a32-P] dATP (specific activity 3000 Ci/ mmol). Final radionucleotide concentration was 5 X 106 cpm/ml hybridization solution. After hybridization, blots were stripped (0.1% SSC, 0.1% sodium dodecyl sulfate SDS at 90°C), and the quantity of RNA was verified using a glyceraldehyde-3-phosphate-dehydrogenase (GAPD) probe. A 780-bp fragment of a cDNA clone encoding the human fetal liver GAPD was liberated from clone pHcGAP (obtained from the American Type Culture Collection) by double digestion with Pst I/Xba I, and this insert was used to develop a radiolabeled probe by random priming with [a-32P] dATP. Final radionucleotide concentration was 8 X 106 cpm/ml hybridization solution. Blots were exposed at -70°C on Kodak X-OMAT AR film (Rochester, NY).

Cell Growth Determinations To determine effects of Na heparin on the growth of LNCaP cells, cells were plated in T-25 flasks, allowed to attach for 48 h, then supplemented with serum-free media containing Na Heparin at 0, 0.1, 1.0, 10, 100, or 1000 gg/ml. Cells were refed every 3 d, and cell number determinations were made with a Coulter Counter (Coulter Electronics, Hialeah, FL) after 6 and 10 d.

Radioimmunoassay (RIA) for Amphiregulin: Antibody Preparation A fragment corresponding to residues 21-46 of the mature 84 aa amphiregulin sequence (AMP 21-46) was linked to bovine serum albumin (BSA) (Imject-activated immunogen conjugation kit, Pierce Chemical, Rockford, IL) and was used to generate two rabbit polyclonal antibody preparations (Cocalico Biologicals, Reamstown, PA) designated BE-7 and BE-9. The sequence of this fragment, 21Ser-Asp-LysPro-Lys-Arg-Lys-Lys-Lys-Gly3 -Gly-Lys-Asn-Gly-Lys-Asn-Arg-Arg-AsnArg40-Lys-Lys-Lys-Asn-Pro-CyS41, contains both putative nuclear localization signals (underlined). Serum from both rabbits recognized the peptide by enzyme-linked immunoadsorbent assay, with the BE-9 antibody showing higher titers. Serum from the BE-9 rabbit was precipitated with ammonium sulfate and affinity purified (Sulfolink coupling gel, Pierce) using high-pressure liquid chromatography (HPLC)purified fragment. Western blot analysis demonstrated that this affinity-purified amphiregulin polyclonal recognized the amphiregulin fragment, but not EGF, TGF-a, or BSA.

0.01% NaN3 pH 7.4. Iodinated fragment was diluted to 7500 cpm/ 100 ul, and affinity-purified antibody was diluted 1:2000 in RIA buffer. Amphiregulin fragment (AMP 21-46, molecular weight 2563) was used as a standard and was reconstituted in ethanol, diluted in RIA buffer, and stored frozen in aliquots of 10 ng/ml. The RIA consisted of 100 Ml of standard/sample, 100 ,u of label, and 100 Al of antibody. Assays were allowed to incubate at 4°C for 24 h and were then precipitated by sequential addition of 100 Ml of 4% normal rabbit serum, 100 ,ul of 25% goat antirabbit immunoglobin (both diluted in RIA buffer), and 200 Al of 7.5% polyethylene glycol 8000 in RIA buffer without gelatin (separation buffer). Precipitates were allowed to form for 1.5 h at 4°C after which sample tubes were centrifuged at 2800 X g for 20 min. The displacement range of 10-90% corresponded to a working assay sensitivity of 10-400 pg/ml of fragment (3.9-156 pM). The nonspecific and total binding for this RIA were -10 and 40%, respectively. EGF, TGF-a, unconditioned media (with or without steroid hormone), and the anti-EGF-R antibody did not interfere with assay conditions. Although there is overlap between the amphiregulin sequences detected by the polyclonal antibody BE-9 and the basic amino acid sequences to which heparin binds, affinity-purified BE-9 IgG precipitated labeled amphiregulin in heparin-containing media at levels equivalent to those precipitated in heparin-free media, indicating that the polyclonal antibody, and thus the RIA, can be used to assay conditioned media that contains heparin. This RIA was used to measure quantities of amphiregulin secreted by LNCaP, DU-145, PC-3, and MCF-7 cells grown in T-25 flasks with 2.5 ml serum-free media. LNCaP cells were plated at 10 000 cells/ cm2 and were grown with or without 1 nM of R1881 in serum-free medium for 6 d; then medium was changed and collected every 3 d under various conditions. Medium for control samples was collected from the 3-d period before initiation of the experimental treatments. MCF-7 cells were plated at 10 000 cells/cm2 and grown in serumfree media for 3 d under various conditions. Replicate flasks were counted to obtain cell density at each time point, and results are expressed as pg amphiregulin/106 cells. -

RESULTS Northern Analysis A single 1.4-kb amphiregulin transcript was detected in androgen-stimulated LNCaP cells (Figure 1). The

RIA for Amphiregulin: Iodination of Amphiregulin A distinct amphiregulin peptide was synthesized with the addition of a tyrosine residue penultimate to the C' terminus, termed AMP 2146-YC. This fragment was labeled with '25I by the chloramine T method. Briefly, 1 nmol of lyophilized AMP 21-46-YC was reconstituted in 40 MA of 0.25 M Na2HPO4 pH 7.6, (buffer A) and 10 M1 of 1251I (1 mCi) in a polyethylene tube. Chloramine T (10 Ml of a 1 mg/ml solution in buffer A) was added to the fragment mixture that was then vortexed 20 s. The oxidation reaction was stopped by addition of 25 Al of Na metabisulfite (2.5 mg/ml in buffer A), and the solution was applied to a G-10 column equilibrated with 3% acetic acid/0.2% BSA. Labeled peptide was separated from unreacted 125I by elution with 3% acetic acid/0.2% BSA and further purified by application to a C-18 reverse phase HPLC column. Two hundred microliters of the G-10-purified labeled fragment was eluted from the reverse phase column with a 10-60% gradient of acetonitrile/0.1% morpholine (in 0.1 M Na acetate/0.1% morpholine pH 4.0) at a flow rate of 0.5 ml/ min, and 0.5 ml fractions were collected. The radioactive peak was recovered, and the affinity-purified BE-9 antibody was used for subsequent RIA analysis of amphiregulin immunoreactivity in conditioned media.

RIA for Amphiregulin: Design of Immunoassays Immunoassays were performed in polyethylene tubes, using the RIA buffer consisting of 0.05 M KH2PO4, 0.15 M NaCl, 0.1% gelatin, and Vol. 5, March 1994

Amphiregulin 1 2 3 4 5 6

kL

2.37

-i

1.35 .-

GAPD

2.371.35Figure 1. Presence of amphiregulin mRNA in androgen-stimulated and -deprived LNCaP cells. PolyA+ RNA from LNCaP prostate carcinoma cells was fractionated on agarose-formaldehyde gels, transferred to a nylon membrane, and hybridized with human amphiregulin and GAPD probes as described in MATERIALS AND METHODS. The conditions shown are as follows: lane 1, androgen-(R1881) stimulated for 6 d, lanes 2-4, androgen-stimulated for 6 d followed by androgen-deprivation for 3, 6, and 9 d, respectively; lanes 5-6, androgen deprived for 6 d, then restimulated for 3 and 6 d, respectively.

341

I. Sehgal et al. h..JJ

androgen (R1881) - androgen

+

0

._20, -4

1500

\00~

bO

50

0 +6

-3

-6

-9 Re+3 Re+6 Re+9

Days of treatment Figure 2. Presence of amphiregulin peptide in conditioned media from androgen-stimulated and -deprived LNCaP cells. Media conditioned over 3 d was collected and immunoassayed from cells that had been androgen-stimulated for 6 d (+6) with R1881 and then androgen deprived for 3, 6, and 9 d or androgen-deprived for 6 d followed by androgen-restimulation for 3, 6, and 9 d. -, androgendeprived time points; Re +, androgen-restimulated time points.

within 24 h and blocked the decline in amphiregulin mRNA levels associated with hormone withdrawal (Figure 3A). Similar results were obtained using EGF. TGF-a stimulation of androgen-treated cells did not appear to increase the level of amphiregulin mRNA expression above the levels observed with androgen treatment alone (Figure 3B). However, blocking the EGF-R with a monoclonal antibody that competitively antagonizes EGF binding (Gill et al., 1984; Kames et al., 1992) reduced amphiregulin message levels in 24 h and through 6 d (Figure 3B). Antibody treatment did not, however, further reduce the already low levels of amphiregulin expression observed in androgen-deprived cells (Figure 3B). Amphiregulin peptide levels in the conditioned medium of hormone-deprived cells stimulated with TGFa reflected the mRNA data (Figure 4). TGF-a stimulated high levels of secreted amphiregulin, whereas treatment with the anti-EGF-R antibody had little effect on detected levels of peptide from these cells. However, although TGF-a stimulation of androgen-stimulated cells did not appear to increase amphiregulin mRNA levels, B.

A.

Amphiregulin

mobility of this transcript is consistent with the size of amphiregulin mRNA reported previously (Plowman et al., 1990; Li et al., 1992). The levels of expression of amphiregulin mRNA in LNCaP cells is reduced by androgen deprivation in a time-dependent fashion (Figure 1, lanes 2-4). Cells restimulated with androgen for 3 d expressed amphiregulin mRNA at levels equivalent to those seen in cells androgen deprived for 9 d (Figure 1, lanes 4-5). After 6 d of androgen restimulation, however, amphiregulin message levels were reinduced to initial levels (Figure 1, lane 6). Secreted levels of amphiregulin peptide are also reduced by androgen withdrawal (Figure 2), reaching a base-line level after 6 d hormone deprivation; androgen restimulation results in a gradual increase in amphiregulin secretion and initial levels are reached in medium collected from 6-9 d after hormone restimulation. The relatively lengthy time period required for downregulation of amphiregulin message and secreted peptide (6-9 d) after androgen deprivation and full reinduction with androgen treatment (6-9 d) suggested that androgen might not directly regulate amphiregulin expression. Because EGF-R number is upregulated by androgen stimulation within 3 d in LNCaP cells (Wilding et al., 1989), enhanced autocrine signaling through this receptor could provide an indirect mechanism of androgen-dependent regulation of amphiregulin expression. To test this hypothesis, LNCaP cells were stimulated with TGF-a either after or during androgen withdrawal. Results showed that TGF-a stimulation restored amphiregulin mRNA levels in androgen-deprived cells 342

kL-

1

2

3 4

1

2 3 4

5

6

2.371.35

X X-

GAPD

2.37 1.35 Figure 3. Amphiregulin mRNA modulation by EGF-R stimulation and inhibition. Effects of EGF-R stimulation on amphiregulin message levels in LNCaP cells were assayed by activating the EGF-R with TGF-a (20 ng/ml) or by inhibiting autocrine receptor stimulation with an anti-EGF-R monoclonal antibody. (A) Amphiregulin mRNA levels from androgen-treated LNCaP cells were compared with levels from androgen-deprived cells stimulated by TGF-a. Conditions shown are as follows: lane 1, androgen-stimulated for 6 d; lane 2, androgendeprived for 6 d; lane 3, androgen-deprived for 6 d followed by 24h stimulation with TGF-a; lane 4, androgen-deprived for 6 d in the continuous presence of TGF-a. (B) Amphiregulin mRNA levels were determined in androgen-stimulated cells treated with TGFa and antiEGF-R antibody, and in androgen-deprived cells treated with antiEGF-R antibody. Conditions shown are as follows: lane 1, 6-d androgen-stimulated LNCaP cells; lane 2, androgen-stimulated cells grown in the presence of TGF-a for 6 d; lane 3, androgen-stimulated for 6 d followed by exposure to anti-EGF-R antibody in androgen containing medium for 24 h; lane 4, androgen-stimulated for 6 d in the continuous presence of anti-EGF-R antibody; lane 5, androgen deprived for 6 d; lane 6, androgen deprived for 6 d in the presence of anti-EGF-R antibody. Molecular Biology of the Cell

Androgen and EGF-R Induce Amphiregulin

500 4

.

breast carcinoma derived MCF-7 cells that are estrogen

androgen

Dickson et al., responsive for growth (Page et al., 1983; with 1 nM estradiol-

androgen

1986), MCF-7 cells were stimulated

androgen

17-fl. Results at both the mRNA and peptide levels re-

400-

-a -300

T

E 200 100

0^ Day of o treatment +TGF-a

_

_ 3

6

+

+

+Anti EGF-R

3

+

6

+

0

3

6

+

+

3

6

+

+

4. Amphiregulin-secreted peptide modulation by EGF-R stimulation and inhibition. LNCaP cell media conditioned over 3 d was collected and immunoassayed from cultures grown with or without androgen (R1881) for 6 d (Day 0). Some cultures were then stimulated with TGF-at (20 ng/ml) or treated with an anti-EGF-R monoclonal antibody (2 Mg/ml) and collected at 3- and 6-d time points.

IFigure

ithe levels of amphiregulin in the conditioned media of

cells was elevated by TGF-a addition (Figure 4). Addition of anti-EGF-R antibody to cultures of androgen-stimulated cells reduced levels of iamphiregulin in the media of these cells by