17β-Estradiol Up-regulates the Insulin-like Growth Factor Receptor through a Nongenotropic Pathway in Prostate Cancer Cells

Prostate carcinomas are usually androgen dependent in their initial stages. However, although initially responsive to antiandrogen treatments, these carcinomas eventually progress to an androgen-independent phenotype, for which no efficacious treatment is currently available. Deregulated growth and survival signaling is a characteristic of androgen-independent tumors, implying that factors other than androgens contribute to prostate cancer progression. Accumulating evidence suggests that also estrogens can have an effect on prostate cancer. Most primary prostate cancers express a recently discovered subtype of the estrogen receptor (the β subtype or ERβ; ref. 1) whereas the classic α subtype (ERα), which is predominant in breast cancer, is silenced by DNA hypermethylation. Interestingly, the ERβ subtype is expressed in most metastases, suggesting that this receptor may be a target in devising new treatments for late-stage prostate cancer (2, 3). Moreover, ER antagonists may inhibit growth and/or induce apoptosis in prostate cancer cell lines that express either only ERβ or both ER subtypes (1, 4). Taken together, these data suggest that estrogens could have a direct effect on prostate cancer.

The insulin-like growth factor (IGF) system is another strong candidate among the factors implicated in prostate cancer progression. This system plays a key role in regulating growth, resistance to apoptosis, and invasion in a variety of human malignancies (5, 6), and various lines of evidence suggest a role for the IGF system also in prostate cancer. IGF-I may increase proliferation of prostate cancer cells and protect them from apoptosis (5, 7), whereas antisense-mediated IGF-I receptor (IGF-IR) down-regulation suppresses rat tumor growth in vivo and prevents prostate cancer cell invasiveness (8). Similarly, IGF-IR blockade by monoclonal antibodies induces growth inhibition in human prostate cancer transplanted in immunodeficient mice (9). Moreover, in human prostate cancer cell xenografts, progression to androgen independence is associated with increased expression of both IGF-IR and IGF-I (10, 11) and increased responsiveness to IGF-I (12). Transgenic mice expressing human IGF-I in the prostate basal epithelium have activated IGF-IR and are prone to spontaneous prostate tumorigenesis (13). Finally, epidemiologic studies found an association between borderline to high IGF-I serum levels and prostate cancer risk (14). These data support the hypothesis that the IGF system is involved in prostate cancer, particularly in progression to androgen independence.

A number of studies have shown that estrogens and the IGF system may functionally interact. Most of these studies regard the ERα subtype and have been carried out in breast cancer MCF-7 cells. In these cells, estrogens are able to up-regulate insulin receptor substrate-1 expression, leading to increased signaling through the insulin receptor substrate-1/phosphatidylinositol 3-kinase (PI3K)/Akt pathway in response to IGF-I (15, 16). Moreover, estrogens increase IGF-I binding and IGF-IR mRNA expression in MCF-7 cells (17), whereas the antiestrogen ICI 182,780 decreases IGF-IR mRNA levels (18). The two ER subtypes, however, affect in a partially different way the IGF system; ERα, but not ERβ, transactivates the IGF-I promoter (19). Down-regulation of IGF-binding proteins is another mechanism through which estrogens may increase IGF-I response (20). The functional interactions between estrogens and the IGF system in prostate cancer are largely unknown.

We now report that estrogens markedly up-regulate the IGF-IR in prostate cancer cells. This effect is specific for IGF-IR because it does not occur for the cognate insulin receptor and can be mediated not only by ERα but also by ERβ. IGF-IR up-regulation by estrogens occurs also in androgen receptor (AR)–negative prostate cancer cells and does not require ER binding to DNA but it requires the activation of the Src-extracellular signal–regulated kinase (ERK)-1/2 pathway. Estrogens may therefore enhance the biological effects of IGFs in prostate cancer cells through a “nongenotropic” pathway (i.e., by a kinase-initiated effect that does not involve steroid receptor binding to canonical steroid response elements on DNA but may eventually affect gene transcription; ref. 21).

Cell media and all chemicals, unless otherwise stated, were obtained from Sigma. FCS and geneticin (G418) were from Invitrogen Laboratories; IGF-I, LY294002, PD98059, and PP2 were from Calbiochem; synthetic androgen R1881 was from NEN Life Science Products; Fugene6 transfection reagent from Roche Diagnostics; luciferase assay system from Promega Corp.; anti–IGF-IR (clone αIR3) and anti–v-Src (clone 327) monoclonal antibodies from Merck Chemicals Ltd.; polyclonal anti–IGF-IR antibody and anti-AR (clone 441), anti-ERα (clone D12), and anti-ERβ (H150) antibodies from Santa Cruz Biotechnology, Inc.; anti-phosphotyrosine antibody (4G10) from UBI; and anti–phospho-ERK1/2 and anti-ERK1/2 antibodies from Cell Signaling Technology, Inc. The cDNA encoding the human AR cloned into the expression vector pSV0 was provided by Dr. A.O. Brinkmann (Department of Reproduction and Development, Erasmus University, Rotterdam, the Netherlands). The cDNA encoding the mutated human AR ART877A was provided by Dr. M. Marcelli (Department of Medicine, Baylor College of Medicine and VA Medical Center, Houston, TX). The cDNA encoding the human ERα cloned into the expression vector pSG5 and the cDNA encoding the human ERβ cloned into the expression vector pSG5/puromycin were provided by Prof. P. Chambon (Institut de Génétique et de Biologie Moléculaire et Cellulaire, CNRS/INSERM/ULP, Collège de France, Illkirch, France). The cDNAs encoding the kinase-inactive mitogen-activated protein kinase/ERK kinase (MEK)-1 (Ser²²¹Ala), the kinase-inactive form of Src (Lys²⁵⁹Met), and the Myc-tagged wild-type p85 were provided by Dr. G. Castoria (Dipartimento di Patologia Generale della II Università di Napoli, Naples, Italy). The cDNA encoding the dominant-negative CCAAT/enhancer binding protein (C/EBP) was provided by Dr. C. Vinson (Laboratory of Metabolism, National Cancer Institute, Center for Cancer Research/NIH, Bethesda, MD). Rat IGF-IR gene promoter sequences corresponding to the full-length fragment (−476/+640), the 5′-flanking fragment (−476/+41), and the 5′ untranslated region (5′-UTR; +41/+640) fragment ligated upstream of the firefly luciferase reporter cDNA in the pGL3 vector (22) were provided by Dr. C.T. Roberts, Jr. (Department of Physiology and Pharmacology, Oregon Health and Science University, Portland, Oregon). The cDNAs encoding the various mutants of the ligand binding domain of the human ERα, E-CFP (fused to nontargeted CFP), E-Mem-CFP (with a membrane localization sequence), and E-Nuc-CFP (with a nuclear localization sequence); the plasmid encoding the full-length human ERα fused to the nontargeted cyan fluorescent protein (ERα-CFP) as well as the plasmid encoding the serum response element (SRE) ligated to the secreted alkaline phosphatase (SEAP) reporter gene; and the pyrazole compound were all provided by Drs. S. Kousteni and S. Manolagas (Division of Endocrinology and Metabolism, Department of Internal Medicine, University of Arkansas for Medical Sciences, Little Rock, AR).

Cells. AR-positive LNCaP and AR-negative PC-3 human prostate cancer cells, human kidney 293 cells (HEK293; AR and ER negative), and Cos-1 cells were obtained from the American Type Culture Collection (Manassas, VA) and were cultured as follows: LNCaP cells in RPMI, and PC-3, HEK293, and Cos-1 cells in DMEM, supplemented with 10% fetal bovine serum and 1% glutamine. AR-transfected PC-3 cells and PC-3-Neo cells were provided by Dr. E. Baldi (Department of Clinical Physiopathology, University of Florence, Florence, Italy).

Transient transfection and reporter assays. A transfection mixture containing 1 μg of DNA, 3 μL Fugene6 in 40 μL of medium without serum was added to each well. After 18 h, the medium was changed to serum-containing medium for 30 h. Cells were then serum starved overnight and incubated with 10 nmol/L 17β-estradiol (E₂), R1881, or vehicle for 24 h. For luciferase assay, cells were lysed and processed according to the manufacturer's instructions (Promega Corp.). Luciferase activity was normalized for transfection efficiency using a vector coding for the H2B-GFP reporter gene (pBOS H2B-GFP-N1; provided by Dr. J. Wang, Division of Biological Sciences and the Cancer Center, University of California, San Diego, CA).

For SRE-SEAP activity, supernatant was collected and SEAP activity measured using the Great EscAPe SEAP Chemiluminescence Kit (Clontech Laboratories, Inc.). The activity of each sample was measured by a multilabel counter Wallac 1420 VICTOR 3 (Perkin-Elmer).

IGF-IR and insulin receptor measurement and IGF-I binding studies. Cell lysate preparation and IGF-IR and insulin receptor measurements were carried out by Western blot analysis as previously described (23, 24). For binding studies, LNCaP cells grown to ∼60% confluence were serum starved and further cultured in the presence or absence of 10 nmol/L E₂, R1881, or vehicle for 24 h. Cells (3 × 10⁶) were then incubated with ¹²⁵I-IGF-I (10 pmol/L) for further 16 h at 4°C in the presence of increasing concentrations of cold IGF-I and cell-associated radioactivity was then measured in a gamma counter. Scatchard analysis was done with GraphPad Prism 4 software.

IGF-IR and ERK1/2 activation. For IGF-IR activation, cells were serum starved in medium without phenol red 24 h before stimulation with 10 nmol/L IGF-I. Cells were lysed in cold radioimmunoprecipitation assay buffer containing 50 mmol/L Tris (pH 7.4), 150 mmol/L NaCl, 1% Triton X-100, 0.25% sodium deoxycolate, 10 mmol/L sodium pyrophosphate, 1 mmol/L NaF, 1 mmol/L sodium orthovanadate, 2 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin, 10 μg/mL pepstatin, and 10 μg/mL leupeptin. The insoluble material was separated by centrifugation and the supernatants were incubated at 4°C for 2 h with 4 μg of the anti–IGF-IR αIR3 antibody coated with protein G-Sepharose. Immunoprecipitates were subjected to SDS-PAGE. The resolved proteins were transferred onto nitrocellulose membranes, immunoblotted with anti-phosphotyrosine 4G10 antibody, and detected by enhanced chemiluminescence (ECL). The filter was then stripped with buffer Restore (Pierce) and reprobed with an anti–IGF-IR antibody.

For ERK1/2 activation, cells were stimulated with 10 nmol/L E₂ and lysed in Laemmli buffer containing 62.5 mmol/L Tris-HCl (pH 6.8), 2% SDS, 10% glycerol, 50 mmol/L DTT, 0.01% bromophenol blue. Cell lysates were subjected to reducing SDS-PAGE on 10% polyacrylamide gel. The resolved proteins were transferred onto nitrocellulose membranes and immunoblotted with anti–phospho-specific ERK1/2 polyclonal antibody. The filters were then stripped and reprobed with anti-ERK1/2 polyclonal antibody.

Analysis of protein-to-protein interactions. Cells were serum starved in medium without phenol red for 24 h and stimulated with E₂ or R1881 for 2 min. Cells were lysed in cold RIPA buffer without sodium deoxycolate. The insoluble material was separated by centrifugation and the supernatants were incubated at 4°C under rotation for 18 h with anti–v-Src antibody or anti-ER antibodies. At the end of the incubation, immunocomplexes were separated by adding 30 μL of mix protein A/G-Sepharose for additional 30 min. The pellets were washed with lysis buffer four times and the protein was reduced in 40 μL of Laemmli buffer and subjected to SDS-PAGE. The resolved proteins were transferred onto nitrocellulose membranes, immunoblotted with specific antibody, and detected by ECL.

Reverse transcription-PCR. Total RNA (5 μg) was reverse transcribed with ThermoScript RT (Invitrogen) and Oligo dT primers. Synthesized cDNA (50 ng) was then combined in a PCR reaction using primers 5′-TTTCTGACAACGCCAAGGA-3′ (forward) and 5′-CAGGGTAGACGGCAGTTCAA-3′ (reverse) specific for AR (fragment size, 341 bp); 5′-GGCTCCGCAAATGCTACGAA-3′ (forward) and 5′-AGCGCCAGACGAGACCAATC-3′ (reverse) specific for ERα (fragment size, 462 bp); and 5′-ATACTTGCCCACGAATCTTT-3′ (forward) and 5′-TGTGATAACTGGCGATGGAC-3′ (reverse) specific for ERβ (fragment size, 375 bp). ELE-1 (housekeeping gene) amplification was done with the following primers: 5′-ATTGAAGAAATTGCAGGCTC-3′ (forward) and 5′-TGGAGAAGAGAGGCTGTATCT-3′ (reverse; fragment size, 280 bp). PCR amplification was carried out for 35 cycles of 30 s at 95°C, 30 s at 60°C, and 30 s at 72°C. The PCR products were analyzed by 2% agarose gel electrophoresis and stained with ethidium bromide.

Real-time PCR. Total RNA (5 μg) was reverse transcribed with ThermoScript RT (Invitrogen) and Oligo dT primers. Synthesized cDNA (25 ng) was then combined in a PCR reaction using primers 5′-GGGCCATCAGGATTGAGAAA-3′ (forward) and 5′-CACAGGCCGTGTCGTTGTCA-3′ (reverse) specific for the IGF-IR (fragment size, 330 bp). ELE-1 amplification was done as described earlier. Quantitative real-time PCR was done on an ABI Prism 7500 (PE Applied Biosystems) using SYBR Green PCR Master Mix (PE Applied Biosystems) following the manufacturer's instructions. Amplification reactions were checked for the presence of nonspecific products by dissociation curve analysis and agarose gel electrophoresis. Relative quantitative determination of target gene levels was done by comparing ΔC_t (25).

Cell cycle, apoptosis evaluation, and cell migration. Cells were grown to ∼60% confluence, serum starved in medium without phenol red for 24 h, and further cultured in the presence or absence of 10 nmol/L E₂ for 24 h. The medium was then replaced with medium containing 1% stripped serum and, after 4 h, cells were incubated in the presence or absence of 10 nmol/L IGF-I for 8 h.

For cell cycle analysis, cells were harvested and permeabilized in ethanol 70% for 18 h at −20°C, then centrifuged and resuspended in PBS containing 16 μg/mL propidium iodide plus 160 μg/mL RNase, incubated for 30 min in the dark, and then subjected to fluorescence-activated cell sorting (FACS; Coulter Elite flow cytometer, Beckman Coulter).

For apoptosis evaluation, cells were treated as indicated in Annexin V-FITC Apoptosis detection kit I (BD Biosciences). Annexin-positive cells were scored by FACS analysis. Values obtained were expressed as percent of Annexin-positive cells over the total cell population. For cell migration assay, cells were treated as previous reported (26).

E₂ induces IGF-IR up-regulation in LNCaP prostate cancer cells. LNCaP cells expressed the β, but not the α, isoform of the estrogen receptor (ERβ), as evaluated by reverse transcription-PCR (RT-PCR; data not shown). When exposed to E₂ for 24 h, these cells showed a dose-dependent increase in the expression of IGF-IR protein, whereas its close homologue insulin receptor was not affected (Fig. 1A). E₂ effect on IGF-IR expression was evident at a dose as low as 0.01 nmol/L and reached a maximum at 1 to 100 nmol/L (Fig. 1A). Time-course experiments indicated that IGF-IR expression started to increase after 4-h cell exposure to 10 nmol/L E₂ and reached a maximum after 24 h (Fig. 1B). A 24-h incubation was used in all subsequent studies.

Functional characteristics of IGF-IR in LNCaP cells preincubated with E₂. To evaluate possible changes in ligand binding characteristics associated with IGF-IR up-regulation, we carried out a Scatchard plot analysis of ¹²⁵I-IGF-I binding to LNCaP cells cultured in the presence or absence of 10 nmol/L E₂ for 24 h. Cells treated with E₂ showed a 4-fold increase of specific IGF-I binding (from 5.5 to 22.7 pmol/L/10⁶ cells) as compared with untreated cells. The dissociation constant (K_d), however, remained similar (0.32 versus 0.40 nmol/L in estrogen-treated or untreated cells, respectively) showing that E₂ increases IGF-I binding sites but only minimally affects the receptor affinity for IGF-I (Fig. 1C). In the same experiment, R1881 increased specific IGF-I binding by 7-fold (from 5.5 to 38.7 pmol/L/10⁶ cells; Fig. 1C). To evaluate whether the increased IGF-I binding capacity caused an enhanced IGF-I signaling, we measured IGF-IR autophosphorylation in response to IGF-I in LNCaP cells either preincubated or not with 10 nmol/L E₂ for 24 h. Time-course experiments indicated that IGF-IR autophosphorylation in response to IGF-I was markedly enhanced by preincubation with E₂ (Fig. 1D).

IGF-IR up-regulation is not due to E₂ binding to AR-T877A of LNCaP cells and can be mediated by both ERα and ERβ. The AR expressed by LNCaP cells bears a point mutation, AR-T877A, which affects AR binding specificity (27). To ascertain whether IGF-IR up-regulation could be mediated, at least in part, by E₂ binding to AR-T877A, we stably transfected HEK293 cells with either AR-T877A or wild-type AR. As shown in Fig. 2A, E₂ was unable to up-regulate IGF-IR in HEK293 cells transfected with either AR-T877A or wild-type AR, whereas R1881 was effective in both cases.

To evaluate whether, in addition to ERβ (the only ER subtype expressed by LNCaP cells), ERα is also able to up-regulate IGF-IR, we transfected HEK293 cells with either ERα or ERβ cDNA (Fig. 2B). Exposure of these cells to E₂ resulted in IGF-IR up-regulation in all cases (Fig. 2B), whereas R1881 was ineffective. Taken together, these data indicate that E₂ is able to up-regulate IGF-IR by binding to both ERα and ERβ and not by cross-reaction with AR-T877A. We then evaluated whether E₂ is able to up-regulate IGF-IR also in AR-negative prostate cancer cells and whether ectopic AR expression has any influence on the effect of E₂ in transfected cells. To this aim, we used androgen-independent PC-3 prostate cancer cells that express both ER subtypes (α and β) but not AR (confirmed by RT-PCR; data not shown). In addition to wild-type PC-3, we also studied two PC-3 cell clones transfected with AR cDNA and expressing AR to different degrees: the PC-3-AR6 clone, with an AR expression only slightly lower than LNCaP cells, and the PC-3-AR13 clone, with an AR content ∼6- to 8-fold lower than LNCaP cells (Fig. 2C). IGF-IR up-regulation by R1881 was strictly related to the AR content in these cell clones and it did not occur in wild-type PC-3. In contrast, E₂ up-regulated IGF-IR also in wild-type PC-3. The presence of AR, however, slightly enhanced the effect of E₂ (Fig. 2C). To confirm immunoblotting data, we carried out Scatchard plot analysis of ¹²⁵I-IGF-I binding data in PC-3 cells cultured in the presence or absence of 10 nmol/L E₂ for 24 h. E₂ treatment increased specific IGF-I binding from 13.1 to 26.3 pmol/L/10⁶ cells. The K_d was similar (0.30 versus 0.22 nmol/L in estrogen-treated or untreated cells, respectively; Fig. 2D). As expected, R1881 was ineffective (Fig. 2D).

Estrogen-induced IGF-IR up-regulation is mediated by a nongenotropic signaling pathway and requires Src/ERK1/2 activity. In previous studies, we found that androgen-mediated IGF-IR up-regulation is due to the activation of a nongenotropic pathway (26). We evaluated, therefore, whether E₂-induced IGF-IR up-regulation also depends on a similar mechanism. We first evaluated the effect of two synthetic estrogen-like compounds, estren (4-estren-3α,17β-diol), which induces only nongenotropic activities of ER, and a pyrazole compound, which induces the transcriptional activities of ER with minimal effects on the nongenotropic pathway (28). To confirm the specific activity of these two ligands in our system, we used a construct encoding for SRE-SEAP that is efficiently induced by both ERα and ERβ when located in the cytoplasm but is repressed by ERs present in the nucleus (21). As expected, estren, but not pyrazole, stimulated SEAP activity (Fig. 3A). In LNCaP cells, estren stimulated IGF-IR up-regulation to a similar degree as E₂, whereas pyrazole was ineffective (Fig. 3A), suggesting that nongenotropic pathways mediate IGF-IR up-regulation.

We next evaluated the possible involvement of the Src/Raf-1/ERK pathway. Indeed, exposure of LNCaP cells to E₂ increased ERK1/2 phosphorylation (Fig. 3B,, top). LNCaP cells were then incubated with E₂ in the presence or absence of various kinase inhibitors, including PD98059, a MEK-1 inhibitor, PP2, a Src inhibitor, and LY294002, a PI3K inhibitor. IGF-IR up-regulation by E₂ was almost completely blocked by both PD98059 (50 μmol/L) and PP2 (10 μmol/L), whereas LY294002 (20 μmol/L) was less effective (Fig. 3B,, middle). Similar results were obtained in HEK293 cells transfected with either ERα or ERβ, although the effect of LY294002 was only evident in cells transfected with ERα (Fig. 3B,, bottom). The antiestrogen ICI 182,780, which is an inhibitor of ER transcription, only slightly inhibited E₂-induced IGF-IR up-regulation in LNCaP cells (Fig. 3C).

To confirm the involvement of Src, LNCaP cells were stimulated with either E₂ or R1881 and then lysed and immunoprecipitated with anti-Src antibody. Immunoblotting with anti-ERβ antibodies confirmed that E₂ stimulation and, to a much lower extent, R1881 induced ERβ association to Src. R1881, but not E₂, induced AR association to Src (Fig. 3D). Both R1881 and E₂ induced recruitment to Src of p85, the regulatory subunit of PI3K. Moreover, immunoprecipitation with anti-ERβ antibody showed that ERβ associates with AR following stimulation with either E₂ or R1881 (Fig. 3D). Exposure to E₂ also stimulated the association of ERβ with the p85 subunit of PI3K (Fig. 3D). We also studied the ability of E₂ to induce association of either ERα or ERβ with p85 and Src in the absence of AR. To this aim, we cotransfected HEK293 cells with a myc-tagged p85 construct together with either ERα or ERβ cDNA. Cells incubated in the presence or absence of E₂ were immunoprecipitated with anti-ER antibodies and blotted with anti-myc or anti-Src antibodies. Cell exposure to E₂ induced association of both ERα and ERβ to p85 and Src (Fig. 3D).

Taken together, these data indicate that E₂-induced IGF-IR up-regulation occurs through the activation of nongenotropic signaling pathways involving Src and ERK1/2 activation, and that E₂ stimulates the association of both ERα and ERβ with Src and the p85 subunit of PI3K. In LNCaP cells, E₂ also induces AR association with Src and ER.

E₂ stimulates IGF-IR mRNA expression and IGF-IR promoter activity. To evaluate whether IGF-IR protein up-regulation after cell exposure to E₂ was due to reduced degradation or increased synthesis, LNCaP cells were incubated for 24 h with 10 nmol/L E₂ in the absence or presence of either actinomycin D or cycloheximide. Both compounds completely inhibited IGF-IR up-regulation by E₂, suggesting that both de novo mRNA and protein synthesis are required for this effect (Fig. 4A). IGF-IR mRNA expression, measured by quantitative real-time PCR, increased by ∼4-fold after 10 nmol/L E₂ for 24 h; this increase was also blocked by either actinomycin D or cycloheximide (Fig. 4B). Dose-response experiments indicated that IGF-IR mRNA started to increase with cell exposure to 0.01 nmol/L E₂ and reached maximum levels with 10 nmol/L E₂ (Fig. 4C). IGF-IR mRNA started to increase at 2 to 4 h after exposure to 10 nmol/L E₂ and reached maximum levels at 24 h (Fig. 4D).

To evaluate whether the increased mRNA expression in response to E₂ was reflected by an increased activity of the IGF-IR promoter, we transiently transfected HEK293 cells with a plasmid encoding the wild-type ER together with one of the three following luciferase constructs, each containing sequences of the IGF-IR promoter: the full-length promoter region (bp −476/+640), the 5′-flanking fragment (bp −41/+640), and the 5′-UTR fragment (−476/+41). Cotransfection with a green fluorescent protein (GFP) vector was used to normalize for transfection efficiency. After exposure to E₂, an increase in luciferase activity was observed in cells transfected with either the full-length fragment or the 5′-UTR promoter fragment (142 ± 12% and 138 ± 16% of unstimulated cells, respectively). In contrast, no luciferase increase was seen in cells transfected with the 5′-flanking fragment, indicating that the sequences responsible for E₂ stimulation are between bp −41 and +640 of the IGF-IR promoter region (not shown). We then evaluated the full-length IGF-IR promoter activity after cotransfection with either ERα or ERβ or AR-T877A. As expected, the promoter activity was increased by E₂ in the presence of ERα or ERβ but not AR-T877A (Fig. 5A).

We then evaluated whether IGF-IR promoter activity stimulation by E₂ also involved a nongenotropic signal. To this aim, we transfected HEK293 cells with plasmids encoding either an ER mutant targeted to the plasma membrane (E-Mem-CFP) or an ER mutant targeted to the nucleus (E-Nuc-CFP). The first ER form is unable to bind to DNA but retains the ability to activate the ERK pathway, whereas the second ER form is entrapped in the nucleus and unable to activate the ERK pathway. Receptor localization in transfected cells was confirmed by immunofluorescence (Fig. 5B,, top). To confirm the specific activity of the two mutant receptors, we used a SRE-SEAP construct that is induced by the ERs located in the cytoplasm but is repressed by ERs present in the nucleus (21). As expected, E₂ induced SRE-SEAP activity in the presence of E-Mem-CFP but not in the presence of E-Nuc-CFP (Fig. 5B,, bottom). HEK293 cells were then transiently cotransfected with plasmids encoding each of the mutant ER constructs and luciferase constructs containing the full-length IGF-IR promoter. Both E₂ and estren were able to stimulate the IGF-IR promoter activity in cells transfected with the DNA binding–defective E-Mem-CFP but not in cells transfected with the ER mutant targeted to the nucleus (E-Nuc-CFP; Fig. 5C). To gain additional insight into IGF-IR promoter regulation by E₂, we cotransfected HEK293 cells with a luciferase construct containing the full-length IGF-IR promoter together with either ERα or ERβ cDNA and with plasmids encoding either a dominant-negative MEK-1 or a dominant-negative C/EBP transcription factor. Both dominant-negatives completely blocked IGF-IR promoter activity in response to E₂ (Fig. 5D).

Biological effects of estrogen-induced IGF-IR up-regulation. We aimed to evaluate whether E₂-induced IGF-IR up-regulation enhanced the biological effects of IGF-I. To ascertain whether AR was required for E₂ effects, we carried out parallel experiments in both LNCaP and PC-3 cells. IGF-I was a weak stimulator of LNCaP and PC-3 cell proliferation. However, preincubation with E₂ significantly enhanced cell proliferation in response to IGF-I. The proportion of cells in S phase increased, on average, from ∼8% to 14% in LNCaP cells and from ∼34% to 40% in PC-3 cells (Fig. 6A and B). These differences were statistically significant (P < 0.01). A similar enhancing effect of E₂ preincubation was observed with regard to the antiapoptotic effect of IGF-I. IGF-I was barely protective from starvation-induced apoptosis (Annexin staining) in untreated LNCaP cells and virtually ineffective in PC-3 cells, which were less sensitive to serum starvation. However, IGF-I significantly reduced apoptosis in E₂-preincubated cells both in LNCaP and PC-3 (Fig. 6C). We also studied migration in cells seeded in Boyden chambers, coated at the lower side with 250 μg/mL collagen, after stimulation with 10 nmol/L IGF-I. Also in this case, the effect of IGF-I was very small or absent in cells not preincubated with E₂. In contrast, IGF-I consistently stimulated migration in E₂-preincubated cells, although this effect did not reach statistical significance (data not shown).

The IGF system plays a key role in the development and maintenance of the transformed phenotype in a variety of experimental models (5, 6, 29) and human malignancies, including prostate cancer (5, 22, 30–33). In this tumor, deregulation of various components of the IGF system has been reported at different tumor stages (11, 34, 35) and may play a role in prostate cancer progression to androgen-independence (10). We previously reported up-regulation of IGF-IR in LNCaP prostate cancer cells by androgens (26). In other models, such as breast cancer, several components of the IGF system may be up-regulated by estrogens through the classic ERα (17, 36, 37).

In the present study, therefore, we investigated whether E₂ up-regulates IGF-IR also in prostate cancer cells, and found that both IGF-IR content and autophosphorylation are increased after exposure to E₂ in both AR-positive and AR-negative prostate cancer cells. Moreover, E₂ sensitizes prostate cancer cells to the biological effects of IGF-I. These E₂ effects do not require ER binding to specific DNA sequences (estrogen response elements) but rather involve the activation of cytosolic kinases such as Src and ERK1/2 that subsequently induce an increase of IGF-IR promoter activity and gene transcription. This E₂ effect on IGF-IR expression must be considered, therefore, as a nongenotropic action of estrogens. Various lines of experimental evidence support this conclusion. First, the Src inhibitor PP2 and the MEK-1 inhibitor PD98059 completely block the IGF-IR protein up-regulation by E₂, both in LNCaP cells and in HEK293 transfected with either ERα or ERβ. The PI3K inhibitor LY294002 partially blocks the E₂ effect in LNCaP cells, suggesting also an involvement of PI3K. Moreover, E₂ exposure stimulates the association of both ERα and ERβ with Src and p85, the regulatory subunit of PI3K. Second, estren, a synthetic ER ligand that induces ERK1/2 and Elk-1 activation without affecting ER binding to DNA (38), reproduces this E₂ effect. In contrast, pyrazole, a compound that activates only the genotropic actions of E₂, is ineffective. Third, E₂ activates the IGF-IR promoter in cells transfected with a mutant ER devoid of DNA binding activity while retaining the nongenotropic activity; in contrast, a mutant ER that localizes in the nucleus and is devoid of nongenotropic activity is without effect.

The present results showing a major role of Src in mediating E₂ effects are supported by previous studies indicating that E₂ activates a Src-dependent pathway by inducing an interaction between the ER phosphotyrosine 537 and the SH2 domain of Src (39, 40). Other studies have also shown that ER, Src, and p85 form a ternary complex, whose assembly is stimulated by E₂ and which induces the activation of both the Src and the PI3K/Akt pathways (39). Activation of these kinase cascades will eventually affect gene expression by affecting multiple transcription factors, including Elk-1, which in turn activates expression of c-fos and down-regulation of C/EBPβ and c-Jun (21). Our preliminary results show that inhibition of C/EBPβ actually blocks E₂-induced IGF-IR promoter activity. This is a novel observation, and we are currently investigating the exact mechanism involving C/EBPβ in this regulation.

Because in LNCaP cells, both DHT and the synthetic androgen R1881 are potent inducers of IGF-IR (26) via the mutated AR (AR-T877A) expressed in these cells, we evaluated the possibility that the effect of E₂ could occur by cross-reaction with AR-T877A. Experiments in transfected HEK293 cells indicated that IGF-IR up-regulation after E₂ occurred via both ER subtypes but neither via wild-type AR nor via AR-T877A.

Data about ER subtype expression and functions in prostate epithelial cancer cells are limited. The classic ERα subtype is not expressed in normal prostate epithelium but is expressed in prostate stromal cells (1, 41), suggesting that estrogen effects on the prostate were mediated by factors produced by stromal cells. Then the ERβ subtype was discovered (42) and found expressed by the prostate malignant epithelium as well as ERα (41). However, the precise roles of ER subtypes in prostate cancer are unknown. In breast cancer, ERα is the major modulator of estrogen tumor-promoting effects whereas the role of ERβ is still controversial (43–46). In the prostate, ERβ is frequently expressed in dysplastic and cancerous prostate tissues and also in metastases (3, 47) whereas the ERα gene is often silenced. When expressed as the only ER subtype, ERβ may stimulate cell proliferation (47). Therefore, the relative abundance of the two ER subtypes may be important in determining the final biological effect. Our study suggests that, as far as nongenotropic effects are concerned, both receptor subtypes behave similarly.

In summary, our data show that E₂ specifically up-regulates IGF-IR in prostate cancer cells and sensitizes cancer cells to the biological effects of IGF-I. The E₂ effect can occur through both ERα and ERβ and does not involve ER binding to DNA but rather the activation of kinase cascades initiated by the association between ER-Src and PI3K and followed by ERK1/2 phosphorylation. AR expression is not required, although AR coexpression may potentiate the E₂ effect by associating with ER. These data raise several potential implications in prostate cancer development and treatment. First, estrogens may enhance the biological effects of IGFs by up-regulating IGF-IR both in AR-positive and AR-negative prostate cancers. Second, because ERβ is expressed in both normal and precancerous prostate epithelium, it is possible that environmental xenoestrogens may mimic the nongenotropic effects of E₂ and represent a risk factor for prostate cancer. Third, the modest results observed with the antiestrogen tamoxifen in prostate cancer (48) may be partially explained by the fact that tamoxifen is a partial ER agonist. Fourth, we now show that also the pure antiestrogen ICI 182,780 is only partially effective in antagonizing E₂-induced IGF-IR up-regulation, whereas Src and ERK1/2 inhibitors can better block this effect. In light of these observations, inhibitors of the IGF-IR or the Src-ERK pathway should be considered as an adjuvant therapy in prostate cancer.

Grant support: Associazione Italiana per la Ricerca sul Cancro (A. Belfiore and R. Vigneri) and PRIN-MIUR (Ministero Italiano Università e Ricerca) grants 2005055874-002 and 2005063915-004. G. Pandini is a recipient of a fellowship from the American Italian Cancer Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank all researchers who have provided materials that have made this work possible: Dr. P. Chambon for the ERα and ERβ expression plasmids, Dr. G. Castoria for MEK-1 and Src mutant and the Myc-tagged wild-type p85 constructs, Dr. C.T. Roberts, Jr. for IGF-IR gene promoter constructs, Drs. S. Kousteni and S. Manolagas for targeted and nontargeted constructs encoding the ER ligand binding domain, Dr. M. Marcelli for the ART877A expression vector, Dr. A.O. Brinkmann for the AR expression vector, Dr. E. Baldi for AR-transfected PC-3 cells, and Dr. C. Vinson for the dominant-negative C/EBP expression vector.