An Inherent Role of Integrin-Linked Kinase-Estrogen Receptor α Interaction in Cell Migration

Introduction

Cell migration is a critical process for cancer invasion and metastatization. It results from a dynamic interplay between the substrate and cytoskeleton proteins located at focal adhesion complexes (e.g., vinculin and actin): the sequential disruption of the focal contacts, the extention of membrane protrusions (i.e., lamellipodia and filopodia), and the formation of new focal points all contribute to the generation of traction forces that allow the cell to move. Cell migration is profoundly influenced by specific stimuli (e.g., growth factors) through the activation of biochemical pathways within the focal adhesion complex ( 1).

Integrin-linked kinase (ILK) has been implicated in diverse physiologic and pathophysiologic processes. It localizes to focal adhesion sites through the NH₂-terminal ankyrin domain, binds to phosphatidylinositol 3,4,5-trisphosphate via the central pleckstrin homology–like domain, and with its COOH-terminal kinase domain interacts with the cytoplasmic tail of β₁-integrin. Following growth factor stimulation, the phosphatidylinositol 3-kinase (PI3K)–produced second messenger phosphatidylinositol 3,4,5-trisphosphate activates ILK, which in turn directly phosphorylates AKT on the phosphoinositide-dependent kinase-2 site, thus driving cell proliferation and migration ( 2). Further, ILK plays a critical role in cancer progression and invasion of estrogen receptor α (ERα)–positive breast cancers through an as yet unclear mechanism ( 3). Nonetheless, previous studies have also reported a loss of ILK during tumor progression and an inhibitory effect of ILK on the growth of breast cancer cells ( 4). Thus, the role of ILK in breast cancer is not completely understood.

Cellular actions of the sex hormone 17β-estradiol (estrogen, E2) contribute to the uncontrolled growth of human ERα-positive breast cancers. Indeed, antiestrogen therapies have defined the importance of the estrogen-ERα signaling pathway in the progression of the ERα-positive tumors ( 5, 6). In recent years, it is increasingly evident that cellular actions of estrogen depend on the intracellular localization of ERα. In the nucleus, the estrogen-activated ERα regulates transcription (i.e., genomic mechanism) by binding to the estrogen-responsive sequence within the promoter of the related genes and recruiting transcriptional cofactors (i.e., coactivators and corepressors; ref. 6). Coactivators associate to ERα via the common LXXLL (L, leucine; X, any amino acid) nuclear receptor interacting motif ( 7), whereas corepressors bind ERα through a less conserved interaction domain (i.e., L/IXXI/VI; L, leucine; I, isoleucine; V, valine; X, any amino acid; ref. 8). In addition, receptor phosphorylation plays a critical role in ERα transcriptional activity, with phosphorylation of Ser¹¹⁸ required for full ERα activation ( 9). However, the identity of the E2-dependent signaling pathway regulating Ser¹¹⁸ phosphorylation is still controversial and likely context dependent ( 10– 13).

In addition to the nuclear transcriptional activity of ERα, estrogen rapidly activates signal transduction cascades (e.g., PI3K/AKT pathway) in the extranuclear compartment (i.e., nongenomic mechanism), which in turn can trigger cell proliferation ( 14). Recently, the nongenomic effects of estrogen have been linked to endothelial monolayer permeability and migration, as well as endometrial cancer cell migration ( 15– 17). However, despite the fact that E2 signaling could affect breast cancer cell adhesion and migration ( 18), the contribution of ILK signaling node in rapid estrogen responses and in breast cancer cell migration remains unknown. In this context, it is interesting to note that ERα and ILK have been shown to regulate overlapping physiologic processes and to share common interacting partners (i.e., heat shock protein 90 and caveolin-1; refs. 19– 22). These observations raise the possibility of crosstalk between ERα and ILK cellular networks that might be important in the estrogen-dependent cell migration. Here, we report an inherent role of ERα-ILK interaction in ERα signaling of breast cancer cells.

Materials and Methods

Cell culture and reagents. Cells were purchased and grown as previously described ( 16). Dextran-coated, charcoal-treated FCS (DCC serum), 17β-estradiol (estrogen, E2), and chemicals were purchased from Sigma Chemical Company (St. Louis, MO). 4-Estren-3α,17β-diol (estren) was purchased from Steraloids, Inc. (Newport, RI). Specific antibodies against T7-epitope (Novagen, Milwaukee, WI); actin, vinculin, cathepsin D (Sigma, St. Louis, MO); ERα (Chemicon, Temecula, CA); extracellular signal–regulated kinase (ERK)-1, ERK2 (Santa Cruz Biotechnology, Santa Cruz, CA); and cyclin D1 (NeoMarkers, Fremont, CA) were used. All others antibodies were purchased from Cell Signaling Technology (Beverly, MA). The pure antiestrogen ICI182,780 was purchased from Tocris (Ellisville, MO). All other inhibitors were purchased from Calbiochem (San Diego, CA). All the cell treatments were administrated after 48 hours starvation in 1% DCC medium.

Confocal microscopy. Confocal microscopy analyses were done as previously reported ( 16). Cells were fixed in 4% paraformaldehyde and immunofluorescently labeled for vinculin (green) or with fluorescently conjugated phalloidin (for actin; red) and ToPro3 (for DNA; blue). Microscopic analyses were done with an Olympus FV300 laser scanning confocal microscope in accordance with established methods, using sequential laser excitation to minimize the possibility of fluorescence emission bleed through. Each representative image is at the same cellular level and magnification.

Boyden chamber assays. Migration potential was done as previously reported ( 16). Variations were applied for siRNA transfection. For uncoated Boyden chamber assay, MCF-7 cells were plated in six-well plates. After 12 hours, cells were transfected for 24 hours with 0.2 nmol of ILK siRNA or with scrambled oligonucleotides. The selectivity of ILK knockdown was checked by Western blot compared with vinculin expression. Following 48 hours of serum starvation, cell migration assays were done by treating cells overnight with estrogen. Migrated cells were detected by counting them.

Stable clones. MCF-7 cells stably expressing ILK were generated by transfecting ILK by electroporation according to the standard methods. Stable clones were selected by G418 selection (1 mg/mL) as previously reported ( 23).

Glutathione S-transferase pull-down assay. In vitro transcription and translation of ERα were done with the T7-TNT kit (Promega Biosciences, San Luis Obispo, CA), in which 1 μg of cDNA in pcDNA 3.1 vector was translated in the presence of [³⁵S]methionine and diluted in glutathione S-transferase (GST) binding buffer [25 mmol/L Tris-HCl (pH 8.0), 50 mmol/L NaCl, 10% glycerol, 0.1% NP40] as previously reported ( 24). An equal aliquot was used for each GST pull-down. The GST pull-down assays were done by incubating equal amounts of GST, GST-tagged full-length proteins, and GST-tagged mutation constructs immobilized on glutathione-Sepharose beads (Amersham Biosciences, Piscataway, NJ) with in vitro translated ³⁵S-labeled protein. Bound proteins were isolated by incubating the mixture for 3 hours at 4°C in GST binding buffer. After washing, the proteins were eluted in 2× SDS buffer, separated by SDS-PAGE, and visualized by autoradiography.

Cell extracts, immunoblotting, and immunoprecipitation. Cells were grown in 1% DCC medium for 48 hours and then stimulated with estrogen (10 nmol/L). When indicated, different concentrations of the inhibitors were added 1 hour before ligand stimulation. Cell extracts and immunoblotting were done as previously reported ( 16). For immunoprecipitation, cells were lysed in 50 mmol/L Tris-HCl (pH 8.0), 500 mmol/L NaCl, 3 mmol/L MgCl₂. Equal amount of protein (1 mg, max 200 μL) was diluted to a final volume of 1 mL with 50 mmol/L Tris-HCl (pH 8.0), 3 mmol/L MgCl₂ to adjust the final concentration of NaCl to 100 mmol/L. Lysates were precleared with protein G beads for 4 hours at 4°C. Immunopercipitation was then done overnight at 4°C using 1 μg antibody/mg protein. Complexes were collected with protein G beads for 4 hours at 4°C. After extensive washing in 20 mmol/L Tris-HCl (pH 8.0), 50 mmol/L NaCl, 1 mmol/L EDTA, proteins were detected as described above.

ILK kinase assay. ILK enzymatic activity was assayed in MFC-7 cells lysed in NP40 buffer [Na-deoxycholate 0.5%, NP40 1%, HEPES 50 mmol/L (pH 7.4), NaCl 150 mmol/L] as previously reported ( 25). Briefly, ILK was immunoprecipitated with mouse ILK antibody (Upstate Biotechnology, Lake Placid, NY) overnight at 4°C from 250 μg of lysate. After immunoprecipitation, beads were resuspended in 30 μL of kinase buffer containing 1 μg of recombinant substrate [glycogen synthase kinase 3β (GSK3β) fusion protein; Cell Signaling Technology] and 200 μmol/L cold ATP, and the reaction was carried out for 30 minutes at 30°C. The phosphorylated substrate was visualized by Western blot with phospho-GSK3β antibody (Ser⁹). Total GSK3β was detected with the appropriate antibody (Cell Signaling Technology).

siRNA transfection. A pool of four individual RNA interference against ILK was purchased from Dharmacon (Lafayette, CA) and used according to the protocol of the manufacturer. Twenty-eight hours was allowed to elapse after transfection to allow efficient silencing of the gene. The selectivity of ILK knockdown was evaluated by analyzing the ILK expression levels.

Chromatin immunoprecipitation analysis. Approximately 10⁶ cells were treated with formaldehyde (1%, v/v) for 10 minutes at 37°C to cross-link histones to DNA. Chromatin immunoprecipitation was done as previously described ( 26) using the following PCR primers: pS2-FP, 5′-GAATTAGCTTAGGCCTAGACGGAATG-3′; pS2-RP, 5′-AGGATTTGCTGATAGGACAGAG-3′.

Results

Involvement of ILK in estrogen-dependent cell migration. To evaluate whether the rapid effects of estrogen (E2) might be involved in the regulation of breast cancer cell migration, changes in the morphology of a ductal carcinoma cell line (MCF-7) were first analyzed by staining the cytoskeletal adhesion proteins vinculin and actin. Thirty minutes of treatment of MCF-7 cells with estrogen induced the appearance of pseudopodia and filopodia and increased the number of cellular focal points ( Fig. 1A ). Interestingly, treatment with estren, a synthetic ERα ligand, also promoted cytoskeleton remodeling ( Fig. 1A). To delineate whether the phenotypic changes observed with estrogen and estren might be translated into long-term physiologic responses, MCF-7 cell motility and migration were analyzed using an established wound healing assay and a noncoated Boyden chamber assay, respectively. We found that estrogen or estren stimulated significant wound closure and increased migration ( Fig. 1B and data not shown). Because estrogen can activate the ERK/mitogen-activated protein kinase (MAPK) and PI3K/AKT pathways ( 14), two known stimulators of cell migration ( 27, 28), we tested the involvement of these signaling cascades in the estrogen-induced cell migration of MCF-7 cells. We found that the PI3K inhibitor Ly294002 prevented estrogen-evoked cell migration whereas no change in the effect of estrogen was detected when cells were exposed to the MAPK/ERK kinase (MEK) inhibitor PD98059 ( Fig. 1C).

Because ILK is a downstream effector of PI3K ( 2) and because PI3K, but not ERK/MAPK, is required for estrogen-mediated MCF-7 cell migration ( Fig. 1C), we reasoned that the blockade of ILK might prevent this estrogen-dependent effect. Therefore, the effect of selective reduction in the cellular ILK content was next examined by siRNA in MCF-7 cells ( Fig. 1D, inset). In scrambled siRNA–transfected cells, estrogen increased the number of migrating cells whereas transfection with ILK targeted siRNA prevented estrogen-dependent migration across the Boyden chamber membrane ( Fig. 1D) but did not affect the basal level of cell migration or cell viability (data not shown).

Furthermore, in MCF-7 cells, estrogen-stimulated wound closure was prevented by the pharmacologic inhibition of PI3K, ILK, or AKT inhibitor, as well as by the transfection of the ILK siRNA (data not shown). These findings reveal a critical role for the PI3K axis, particularly its downstream effector ILK, in the regulation of estrogen-mediated cell migration.

ILK binding to ERα. One of the better characterized means by which ERα regulates different physiologic processes is through direct interaction to the NR box located within the key molecule involved in that particular physiologic process. Because one canonical (i.e., LXXLL) and four noncanonical (e.g., VXXLI) consensus binding sites for ERα ( 7) are present within the ILK protein sequence (Supplementary Fig. S1A), we assessed ERα-ILK interactions both in vitro and in vivo. Pull-down experiments using GST-tagged full-length (FL) ILK, GST-ILK deletion constructs, and ³⁵S-labeled full-length ERα indicated that ERα binds to ILK in vitro ( Fig. 2A ). In particular, ERα strongly interacted with the NH₂-terminal domain of ILK upstream of its pleckstrin homology domain (amino acid 126-212, Δ2) whereas regions spanning amino acids 1-125 (Δ1) and 213-452 (Δ3-Δ4) exhibited no or weak binding ( Fig. 2B; Supplementary Fig. S1A). Substitutions of L144 and L148 to alanine in the FL-ILK canonical NR box prevented the interaction, suggesting that the NR motif mediates ERα-ILK association ( Fig. 2B). Next, we analyzed physiologic ERα-ILK interactions by coimmunoprecipitating endogenous ERα and ILK from cell lysates. ILK could be effectively coprecipitated along with ERα in both MCF-7 and endometrial adenocarcinoma (HEC1A) cells using an ERα antibody but not a control immunoglobulin G ( Fig. 2C). These data indicate that the ERα-ILK association occurs in diverse physiologic contexts.

Signaling-dependent regulation of ERα-ILK interaction. To evaluate a direct functional involvement of ILK in ERα signaling, we next tested the effect of pure antiestrogen ICI182,780 on ERα-ILK binding in exponentially growing MCF-7. Surprisingly, ICI182,780 increased the amount of ILK associated with ERα ( Fig. 3A ), raising the possibility that ILK selectively binds nonactivated ERα. Accordingly, estrogen induced a rapid (30-60 minutes) decrease in the amount of ILK that coimmunoprecipitated with ERα ( Fig. 3A). No effect on ERα ( Fig. 3A) or ILK (Supplementary Fig. S1B) levels were detected during the period of estrogen treatment. These observations indicate that, on activation, the ERα dissociates from ILK, thus supporting a functional role for ERα-ILK interactions in breast cancer cell biology.

Next, we sought to define the signaling mechanism underlying the effects of estrogen on ERα-ILK binding. In MCF-7 cells, estrogen-induced rapid and transient activation (5-30 minutes) of the ERK/MAPK and PI3K/AKT pathways ( 14) paralleled a rapid and sustained (60 minutes) estrogen-evoked increase in ERα Ser¹¹⁸ phosphorylation (Supplementary Fig. S1C). The overlapping kinetics of both signal transduction kinases and ERα Ser¹¹⁸ phosphorylation suggested the involvement of the rapid signaling pathways in the regulation of receptor phosphorylation and ERα-ILK complex dissociation. To test this hypothesis, the status of ERα Ser¹¹⁸ phosphorylation was evaluated with a set of pharmacologic inhibitors. The estrogen-induced increase (60 minutes) in ERα Ser¹¹⁸ phosphorylation was unaffected by pretreating cells with PD98059 whereas ICI182,780 completely blocked estrogen-induced ERα Ser¹¹⁸ phosphorylation. Pretreatment of cells with the PI3K inhibitor Ly294002 significantly reduced estrogen-induced ERα Ser¹¹⁸ phosphorylation compatible with the PI3K pathway mediating estrogen-dependent Ser¹¹⁸ phosphoryation ( Fig. 3B). As expected, the pretreatment of cells with PD98059 and Ly294002 effectively inhibited ERK1/2 and AKT phosphorylation in basal and estrogen-induced conditions, thus showing the efficiency of the kinase inhibitors (Supplementary Fig. S1D).

An involvement of the PI3K pathway in the estrogen-dependent ERα Ser¹¹⁸ phosphorylation was supported by the demonstration that the AKT1/2 inhibitor AI8 induced a dose-dependent reduction in estrogen-induced ERα Ser¹¹⁸ phosphorylation at concentrations known to inhibit AKT activation in MCF-7 cells (ref. 29; Fig. 3C). ERα protein levels were not modified by these inhibitors.

Because decreased ERα phosphorylation on Ser¹¹⁸ paralleled the increase in ERα-ILK association (i.e., ICI182,780 treatment) and because the estrogen-induced activation of the PI3K/AKT pathway was involved in the regulation of ERα Ser¹¹⁸ phosphorylation, the blockade of the PI3K pathway might prevent estrogen-dependent dissociation of the ERα-ILK complex. Consequently, we examined the ability of the PI3K pathway to interfere with the ERα-ILK interaction. Surprisingly, in coimmunoprecipitation experiments, Ly294002 alone mimicked the effect of estrogen in inducing the ERα-ILK complex dissociation whereas it blocked the estrogen-induced reduction in the amount of ILK coimmunoprecipitated with ERα ( Fig. 3D). Collectively, these observations strongly suggest that the PI3K pathway is involved in the estrogen-induced ERα Ser¹¹⁸ phosphorylation and in the regulation of the ERα-ILK interaction.

Estrogen activation of ILK. In MCF-7 cells, the estrogen-induced activation of PI3K results in phosphorylation of AKT on Ser⁴⁷³ ( 14). Moreover, ILK can also directly activate AKT by phosphorylating it on Ser⁴⁷³ ( 2). Because the present data indicate a functional role for the PI3K axis in the modulation of estrogen-dependent cell migration and in the regulation of the ERα-ILK interaction, it is possible that estrogen may also activate ILK.

Therefore, the activation of ILK and AKT was evaluated under a series of experimental conditions in MCF-7 cells. Estrogen treatment (30 minutes) induced a significant increase in the kinase activity of ILK immunoprecipitated from MCF-7 cells, as assessed by in vitro phosphorylation of the recombinant GSK3β on Ser⁹ (ref. 25; Fig. 4A ). In addition, pretreatment of cells with Ly294002 significantly prevented the estrogen-induced increase in ILK kinase activity ( Fig. 4A), indicating PI3K dependence of estrogen-mediated ILK activation. Estrogen administration (30 minutes) also induced AKT Ser⁴⁷³ phosphorylation, which could be blocked by the PI3K inhibitor Ly294002 (Supplementary Fig. S1D). Moreover, pretreatment of MCF-7 cells with a cell-permeable ILK inhibitor, QLT-0267, ( 30) prevented estrogen-induced AKT phosphorylation ( Fig. 4B), thus suggesting that estrogen may activate the PI3K/ILK/AKT pathway.

Estren, which specifically stimulates extranuclear ERα signaling ( 31), was used to determine whether the effects of ERα were primarily cytosolic. Estren administration (30 minutes) also increased ILK kinase activity and AKT phosphorylation ( Fig. 4A; Supplementary Fig. S2A and B). Furthermore, pretreatment of cells with either Ly294002 or QLT-0267 blocked the estren-induced AKT phosphorylation (Supplementary Fig. S2A and B). In addition, cell pretreatment with the palmitoyl-acil transferase inhibitor 2-bromo-hexadecanoid acid, which blocks ERα signaling by inhibiting palmitoylation-dependent ERα membrane localization ( 32), also prevented estren-induced AKT phosphorylation without significantly affecting the basal level of AKT activation (Supplementary Fig. S2B).

To further analyze the contribution of the cytoplasmic ERα signaling, we next examined the role of ILK in AKT activation both in a human hepatoma cell line (HepG2), which physiologically expresses cytoplasmic ERα ( 33), and in the mammary carcinoma cell line ZR-75 stably expressing the T7-MTA1s transgene, which sequesters the ERα in the cytoplasm ( 23). In both cell lines, estrogen-induced AKT phosphorylation was prevented by QLT-0267 (Supplementary Fig. S2C and D). Notably, the overexpression of MTA1s hyperstimulated AKT phosphorylation. Interestingly, QLT-0267 differentially affected basal AKT activation, thus suggesting a differential sensitivity of different cell lines to ILK inhibition. Overall, these findings suggest that estrogen induces the activation of the PI3K/ILK/AKT pathway through a nongenomic mechanism. We next examined whether ILK activation was involved in estrogen-induced changes of cell morphology. Pretreatment of cells with QLT-0267 prevented the ability of estrogen to induce the rapid appearance of pseudopodia and of new focal contacts ( Fig. 4C), thus suggesting a critical role for ILK in the early effects of estrogen on breast cancer cell motile phenotype.

ILK as a potential switch for estrogen-induced ERα Ser¹¹⁸ phosphorylation from the ERK/MAPK pathway to the PI3K/AKT pathway. Because estrogen activates the PI3K/ILK/AKT pathway and because PI3K contributes to ERα phosphorylation, we speculated that the inhibition of ILK might also compromise ERα Ser¹¹⁸ phosphorylation. Surprisingly, QLT-0267 pretreatment increased the levels of both basal and estrogen-evoked ERα Ser¹¹⁸ phosphorylation in a dose-dependent manner ( Fig. 5A ). Remarkably, in the same samples, QLT-0267 also enhanced the estrogen-evoked ERK1/2 phosphorylation in a dose-dependent manner ( Fig. 5B; lysates from Figs. 4B and 5A). Note that ILK inhibitor did not modify the basal levels of the signaling molecules. Similar results were obtained when endogenous ILK was efficiently knocked down with siRNA. In control siRNA–transfected MCF-7 cells, estrogen treatment increased ERK1/2 and ERα Ser¹¹⁸ phosphorylation. Remarkably, ILK siRNA transfection resulted in ERK1/2 hyperphosphorylation and in an enhancement of estrogen-induced ERα Ser¹¹⁸ phosphorylation ( Fig. 5C and data not shown). No significant changes in ERα cellular content were observed under these experimental conditions. These unexpected findings reveal a causative correlation between up-regulation of ERα Ser¹¹⁸ phosphorylation and ERK activation on ILK inhibition.

To test whether in the absence of ILK the ERK/MAPK pathway might trigger ERα Ser¹¹⁸ phosphorylation, scrambled siRNA–transfected or ILK siRNA–transfected cells were treated with estrogen in both the presence and absence of either PI3K or MEK inhibitors. Pretreatment with PD98059 did not affect ERα Ser¹¹⁸ phosphorylation in scrambled oligonucleotide–transfected cells but prevented it in ILK siRNA–transfected cells ( Fig. 6A ). On the contrary, the inhibition of PI3K reduced estrogen-induced ERα Ser¹¹⁸ phosphorylation in scrambled oligonucleotide–transfected cells but did not affect it in ILK siRNA–transfected cells ( Fig. 6B). As expected, no significant change in ERα expression was observed and ILK levels were reduced by ILK siRNA but unaffected by control siRNA ( Fig. 6A and B). These results support the notion that ILK retargets ERα Ser¹¹⁸ phosphorylation from the estrogen-induced ERK/MAPK pathway to the PI3K pathway.

ILK is a negative regulator of estrogen response. Because Ser¹¹⁸ phosphorylation is required for full ERα transcriptional activation ( 9), the ILK knockdown or inhibition-dependent ERα hyperphosphorylation suggested that ILK may negatively regulate ERα activity. Indeed, in MCF-7 cells, the estrogen-triggered antiestrogen-dependent recruitment of ERα to the pS2 promoter was further increased by the QLT-0267-dependent ILK inhibition ( Fig. 6C). In addition, basal and estrogen-induced expression levels of the endogenous estrogen-responsive genes cyclin D1 and cathepsin D were up-regulated in ILK siRNA–transfected cells with respect to scrambled oligonucleotide–transfected cells ( Fig. 6D). Finally, the role of ILK in ERα signaling was also evaluated in low- or high-expressing V5-ILK/MCF-7 stable clones (Supplementary Fig. S3A). Estrogen-evoked ERα Ser¹¹⁸ phosphorylation was lower in the stable clones than in MCF-7/pcDNA cells, with the reduction correlating with the amount of V5-ILK expression, thus indicating that ILK overexpression prevents estrogen-dependent ERα Ser¹¹⁸ phosphorylation. ERα protein levels were unchanged in the different cell lines (Supplementary Fig. S3B). Accordingly, the recruitment of ERα by estrogen to the pS2 promoter was substantially reduced in V5-ILK clone #21 compared with control cells. Notably, this effect was prevented by the ILK inhibitor QLT-0267 (Supplementary Fig. S3C). On the other hand, ILK overexpression enhanced the ability of the cells to migrate across the Boyden chamber with respect to MCF-7/pcDNA cells. Interestingly, estrogen did not further stimulate cell migration in MCF-7/V5-ILK clone #21, thus revealing a threshold role for ILK levels in cell migration (Supplementary Fig. S3D). In brief, these findings reveal that in breast cancer cells, ILK plays an inherent negative regulatory function in estrogen response but drives cell migration.

Discussion

The principal goals of the present work were to define the crosstalk between ILK and ERα signaling and to evaluate the role of ILK in the regulation of estrogen-mediated cell migration and ERα function. This was based on the observations that both ILK and ERα-driven intracellular networks ultimately result in the regulation of cell migration. In the present study, we provide evidence that ILK participates in ERα signaling by directly interacting with ERα in breast cancer cells. Accordingly, the rapid estrogen-modulated ERα-ILK dissociation synchronizes the estrogen-induced ERα phosphorylation and ILK activation. As a consequence, estrogen regulates ERα-mediated gene transcription and cell migration of breast cancer cells.

ILK is a major signaling node that links integrin and growth factor signaling to a variety of cellular responses. The cellular functions of ILK rely on its ability to function both as a scaffold molecule and as a signaling serine/threonine kinase. In particular, the COOH-terminal catalytic domain of ILK contains the β₁-integrin binding region and also mediates ILK interactions with actin-binding proteins. In addition, the NH₂-terminal ankyrin repeats precede the pleckstrin homology–like domain and mediate the ILK localization to focal adhesions ( 2). The in vitro interaction of ILK to ERα occurs through the ILK NR box (i.e., LXXLL) located between the ILK pleckstrin homology–like domain and the ankyrin repeats. Thus, ILK may function as a scaffold protein bridging ERα to integrins. Although no evidence exists for direct ERα-integrin interaction, crosstalk between nuclear receptor and integrin signaling has been suggested ( 34– 36). The results presented here show for the first time the association of ERα with a key functional component of integrin signal transduction and suggest that ERα might be tethered to the cell membrane by integrin-associated docking proteins.

Remarkably, estrogen induces the dissociation of the ERα-ILK complex and the PI3K pathway seems to be involved in the regulation of ERα-ILK interactions. However, we cannot rule out the potential effect of other estrogen-responsive pathways on ERα-ILK interaction. Nonetheless, a signaling scenario might be envisioned for the regulation of the ERα-ILK interaction. In particular, in the presence of estrogen, the dissociation of ERα-ILK complex may be due to the enhanced affinity of ERα with p85, the regulatory subunit of the PI3K ( 14), and a reduced affinity for ILK. On the other hand, the lack of ERα-ILK interaction observed when PI3K is inhibited may be attributed to the blockade of specific PI3K-dependent posttranslational modification(s) of ILK, which may be important for the basal ERα-ILK interaction. Finally, a possible explanation for the association of ERα with ILK observed in the presence of both PI3K inhibitor and estrogen might be because of an increased affinity of the active ERα-p85 complex to ILK. This three-party hypothesis is currently under investigation in our laboratory.

Interestingly, estrogen induces the rapid PI3K-dependent ERα phosphorylation on Ser¹¹⁸. Despite the overwhelming evidence that ERα Ser¹¹⁸ is phosphorylated in response to estrogen, the mechanism for Ser¹¹⁸ phosphorylation is still controversial. The Ser¹¹⁸ residue is located within a consensus sequence for ERK/MAPK ( 10), and ERK/MAPK can phosphorylate ERα Ser¹¹⁸ in vitro ( 9, 11). However, in MCF-7 cells, estrogen-induced ERα Ser¹¹⁸ phosphorylation is ERK/MAPK independent (refs. 10, 13; present study). This observation implicates other kinases in the estrogen-dependent regulation of Ser¹¹⁸ phosphorylation (e.g., cyclin-dependent kinase 7; ref. 11). Here, we present evidence that the PI3K pathway can mediate estrogen-induced ERα Ser¹¹⁸ phosphorylation when ILK is intact. GSK3β, which is downstream of the PI3K/ILK/AKT pathway, can phosphorylate ERα in vitro ( 13). Because GSK3β is inhibited by the PI3K pathway, GSK3β or other kinases (e.g., cyclin-dependent kinase 7) may be responsible for the modest ERα Ser¹¹⁸ phosphorylation detected in the presence of PI3K or AKT inhibitors.

Unexpectedly, in the absence of ILK, estrogen-mediated ERα Ser¹¹⁸ and ERK phosphorylation were up-regulated. Remarkably, the ERK/MAPK pathway was responsible for the estrogen-dependent hyperphosphorylation of ERα Ser¹¹⁸ in the absence of ILK. The mechanisms underlying this process are unclear as estrogen-dependent ERK/MAPK activation is intact in ILK-containing cells. However, the binding of ILK to ERα in basal conditions may prevent productive access of ERK/MAPK to ERα and, thus, receptor phosphorylation. Nonetheless, these findings indicate that ILK switches estrogen-induced ERα Ser¹¹⁸ phosphorylation from the ERK/MAPK pathway to the PI3K pathway.

On growth factor stimulation, the PI3K-produced phosphatidylinositol 3,4,5-trisphosphate triggers ILK activation, which results in direct AKT phosphorylation ( 2). Estrogen triggers the association of the PI3K regulatory subunit p85 with ERα, PI3K activation, and AKT phosphorylation in MCF-7 cells ( 14). Our findings show estrogen-dependent ILK activation in MCF-7 cells and define ILK as a downstream target of PI3K and as an upstream activator of estrogen-mediated AKT activation. Furthermore, the ability of estren to mimic the effect of estrogen indicates that estrogen activates the PI3K/ILK/AKT pathway by engaging extranuclear-localized ERα ( 31). Although concerns exist about the pool of ERα (i.e., membrane versus cytoplasmic) responsible for estrogen intracellular signals, the fact that the inhibition of palmitoyl-acil transferase activity prevents the estren-mediated AKT activation implicates the membrane-tethered ERα in the activation of estrogen-evoked signal transduction ( 31, 32). Nonetheless, as ILK inhibition blocks AKT activation in cytoplasmic ERα-expressing cell lines, the contribution of the cytoplasmic pool of ERα cannot be excluded ( 23, 33).

Although the mechanism for ERα plasma membrane localization is highly debatable, nongenomic ERα-mediated membrane starting activation of signaling cascades (e.g., PI3K/AKT) is required for breast cancer cell proliferation ( 14). Remarkably, accumulating evidence has now established that the rapid ERα-mediated effects of estrogen impinge also on cytoskeleton remodeling. Estrogen-mediated rearrangement of actin stress fibers and dissociation of adherens junctions ultimately result in the migration of vascular endothelial and endometrial cancer cells ( 15– 17). However, this mechanism has not been shown in breast cancer–derived cells. Our data indicate that estrogen regulates cell motility and migration of breast cancer cells. Cell motility is required for cancer cells to develop an aggressive phenotype. The resulting increase in migration potentially favors metastasis. The PI3K signaling axis is intimately associated with deregulated cancer cell growth and survival and is a driving force for metastasis. In this respect, both PI3K and ILK can promote cell motility and migration ( 2, 28). Interestingly, the role of AKT in cell migration is isoform dependent (i.e., AKT1 versus AKT2) and strictly regulated ( 37, 38). Although we could not discriminate between the two AKT isoforms, our data show that the estrogen-evoked nongenomic activation of the PI3K/ILK/AKT pathway regulates breast cancer cell migration.

Remarkably, different cancer cell lines show a different sensitivity to ILK inhibition. The ILK inhibitor QLT-0267 is an optimized second-generation compound, which selectively inhibits the ILK phosphotransferase activity 1,000-fold over other kinases ( 30). Cellular studies using QLT-0267 revealed the importance of ILK signaling in cancer. In particular, the sensitivity of ILK-dependent AKT phosphorylation increases drastically in the progression from normal breast epithelium to breast cancer cells; thus, breast cancer cells become “addicted” to ILK ( 3). Our results extend the concept of “ILK addiction” in other types of cancers and highlight the importance of targeting ILK signaling for cancer therapy.

Although commonly considered an oncogene ( 2), under some conditions, ILK might function as a tumor suppressor gene in breast cancer ( 4). In particular, whereas ILK overexpression results in tumor growth and invasion in mouse mammary epithelial cells ( 39, 40) and cyclin D1 up-regulation in human breast cancer cells ( 41), expression of ILK is down-regulated in metastatic breast cancers, and ILK reexpression suppresses the invasive phenotype of metastatic breast cancer cells ( 4). Although these discrepancies may be cell type specific, they define a dual function of ILK in breast cancer cells ( 4). However, in the present work, the analysis of both ILK knockdown and stable overexpression in MCF-7 cells revealed the critical importance of ILK expression levels in breast cancer cells. Indeed, ILK down-regulation or inhibition results in hyperstimulation of ERα Ser¹¹⁸ phosphorylation, enhances ERα association with target chromatin, and increases estrogen response while blocking cell migration. On the other hand, ILK overexpression negatively affects ERα Ser¹¹⁸ phosphorylation and association with chromatin but enhances cell migration. These contrasting results point to a critical role for ILK in the fine-tuning of the balance between the ERα-dependent cell migration and gene transcription.

In conclusion, our discoveries show an intrinsic role for ERα-ILK interaction in the regulation of breast cancer cell migration and suggest that the deregulation of ILK signaling in breast cancers may lead to ERα-independent–like or estrogen-hypersensitive phenotypes (i.e., ILK down-regulation) or to more aggressive tumors (i.e., ILK overexpression).