Erythropoietin Promotes MCF-7 Breast Cancer Cell Migration by an ERK/Mitogen-activated Protein Kinase-dependent Pathway and Is Primarily Responsible for the Increase in Migration Observed in Hypoxia*

Recent studies indicate that cancer cells express erythropoietin receptor (EpoR). In this study, we have shown that erythropoietin (Epo) activates the mitogen-activated protein kinase, extracellular signal-regulated kinase (ERK), and promotes migration in MCF-7 breast cancer cells. Epo-stimulated MCF-7 cell migration was blocked by the MEK inhibitor PD098059 and by dominant negative MEK-1, indicating an essential role for ERK. When MCF-7 cells were exposed to hypoxia (1.0% O2) for 3 h, the Epo mRNA level increased 2.4 ± 0.5-fold, the basal level of ERK activation increased, and cell migration increased 2.0 ± 0.1-fold. Soluble EpoR and Epo-neutralizing antibody significantly inhibited hypoxia-induced MCF-7 cell migration, suggesting a major role for autocrine EpoR cell signaling. MCF-7 cell migration under hypoxic conditions was also inhibited by PD098059. These experiments identify a novel pathway by which exogenously administered Epo, and Epo that is produced locally by cancer cells under hypoxic conditions, may stimulate cancer cell migration.

Erythropoietin (Epo) 2 is a 34-kDa member of the class I cytokine family that initiates cell signaling by forming a trimeric complex with two molecules of the Epo receptor (EpoR) (1). Epo binding induces a conformational change in EpoR so that receptor-associated Janus kinase-2 is activated (2). This leads to phosphorylation of tyrosine residues in EpoR and recruitment of Src homology 2 domain-containing proteins. Signaling proteins activated downstream of EpoR and Janus kinase-2 include phosphatidylinositol 3-kinase, protein kinase C, and the mitogen-activated protein kinase, extracellular signal-regulated kinase (ERK) (3,4).
In adults, Epo is produced mainly by peritubular fibroblasts in the kidney; however, diverse cells in multiple organs also express Epo (5). Epo expression is stimulated by hypoxia as a result of activation of the transcription factor, hypoxia-inducible factor-1 (HIF-1␣) (6). Epo is well characterized for its ability to promote proliferation and differentiation of erythroid progenitor cells and also inhibit apoptosis in these cells (7); however, Epo is active outside the hematopoietic system as well. In the nervous system, Epo inhibits neuronal apoptosis resulting from hypoxia or ischemia (8 -10) and stimulates Schwann cell proliferation (11). Epo also regulates angiogenesis by multiple mechanisms, including stimulation of endothelial cell proliferation and migration (12).
The function of Epo in cancer is not well understood. EpoR is expressed by several human tumor types, including breast carcinoma, renal cell carcinoma, melanoma, endometrial carcinoma, and gastric carcinoma (13)(14)(15)(16)(17). In breast cancer, EpoR is observed at high levels by immunohistochemistry, whereas normal mammary tissue is typically EpoR negative (18). Many tumors are also strongly Epo positive (13). Furthermore, in at least one study, Epo and EpoR were observed at increased levels in hypoxic regions of breast cancers (13). Lai et al. (19) demonstrated increased Matrigel invasion by squamous cell carcinoma cells after treatment with Epo and attributed this increase to activation of the Janus kinase-2-STAT signaling pathway. In a separate study, reagents that inhibit Epo binding to EpoR, or that block Janus kinase-2 signaling, reduced growth of breast cancer cells in tumor Z-chambers implanted subcutaneously in rats (20). Cancer cells may be exposed to Epo in patients not only due to local synthesis but also as a result of Epo therapy, which is used to treat anemia secondary to chemotherapy and radiation therapy (21). Possible problems associated with Epo therapy in cancer were highlighted by a recent clinical trial in which patients with metastatic breast cancer received chemotherapy, with or without Epo. The trial was terminated early because of increased mortality in the Epo-treated patients (22).
In this study, we have demonstrated that Epo activates ERK in MCF-7 breast cancer cells in vitro. Epo also promotes MCF-7 cell migration, and this effect is ERK dependent. Under hypoxic conditions, the Epo mRNA level and the basal level of ERK activation are increased. These changes are accompanied by an increase in the rate of MCF-7 cell migration, which is significantly inhibited by Epo-neutralizing antibody, soluble EpoR, and by the MEK inhibitor, PD098059. Our results demonstrate that Epo may promote breast cancer cell motility by activating ERK under normoxic or hypoxic conditions. In hypoxia, EpoR autocrine signaling assumes a dominant role and determines the basal level of MCF-7 cell migration.
Immunoblot Analysis of ERK Activation-MCF-7 cells were seeded into 6-well plates (3 ϫ 10 5 cells/well) and cultured in phenol red-free, serum-free medium for 4 h before treatment with Epo (1.7 nM) for up to 120 min. In some experiments, the Janus kinase-2 inhibitor, AG490 (50 M), or vehicle was added to cultures for 2 h prior to adding Epo for 15 min. Cell extracts were prepared in 50 mM Hepes, 100 mM NaCl, 2 mM EDTA, and 1% (v/v) Triton X-100, pH 7.5, containing protease inhibitor mixture and 1 mM sodium orthovanadate. To assess the effects of hypoxia on the basal level of MCF-7 cell ERK activation, cells were transferred to an incubator equilibrated at 1% O 2 for 3 h or kept under normoxic conditions as a control. The protein concentration in each extract was determined by bicinchoninic acid assay (Sigma). Equal amounts of cell extracts were subjected to SDS-PAGE on 10% slabs, electrotransferred to polyvinylidene difluoride membranes (Bio-Rad), and probed with antibodies specific for phosphorylated ERK and total ERK.
EpoR Phosphorylation-MCF-7 cells (5 ϫ 10 5 ) were seeded into 6-cm plates and cultured in phenol red-free, serum-free medium for 4 h prior to treatment with Epo (1.7 nM) for 1 min. Cell extracts were prepared in radioimmune precipitation assay buffer containing 0.1% SDS, 0.5% sodium deoxycholate, 1% (v/v) Triton X-100, protease inhibitor mixture, and 1 mM sodium orthovanadate. Equal amounts of cell extracts were subjected to immunoblot analysis, using phospho-EpoR-specific antibody. Membranes were reprobed for total ERK as a loading control.
Cell Migration Assays-Migration of MCF-7 cells was studied using 6.5-mm Transwell chambers with 8-m pores (Corning Costar, Corning, NY) as previously described (25). The bottom surface of each membrane was coated with 20% fetal bovine serum for 2 h. Under these conditions, the major protein that adsorbs to the membrane is vitronectin (26). MCF-7 cells in serum-free, phenol red-free medium were treated with Epo, Epo-neutralizing antibody (10 g/ml), nonspecific murine IgG (10 g/ml), soluble EpoR (2.5 g/ml), or PD098059 (50 M) for 15 min at 37°C. Reagents were added at the same concentration to the medium in the Transwells. The bottom chamber contained 10% fetal bovine serum. Cells (10 5 ) were transferred to the top chamber of each Transwell and allowed to migrate under normoxic conditions or in an incubator equilibrated at 1% O 2 . After 6 h, membranes were stained with Diff-Quik (Dade-Behring, Deerfield, IL). Membranes were cut from each Transwell and transferred to microscope slides. Cells that migrated through the membrane to the lower surface were counted by light microscopy.
To study migration of GFP-expressing cells, translucent Biocoat cell culture inserts (BD Biosciences) were used. Cells were transfected to express either GFP and dn-MEK or GFP alone. After allowing cells to migrate for 6 h, cells were fixed in 4% formaldehyde for 20 min. The number of GFP-expressing cells that transferred to the underside of each membrane was counted by fluorescent microscopy.
Taqman Real-time qPCR-MCF-7 cells (5 ϫ 10 5 ) were seeded into 6-cm plates and cultured for 4 h in serum-free medium. The cultures were transferred to 1% O 2 for 3 h or kept under normoxic conditions. RNA was extracted using the RNeasy mini-kit (Qiagen) and treated with DNase1 (Qiagen). cDNA libraries were generated using oligo(dT) primers and the Superscript First Strand Synthesis kit (Invitrogen). Epo mRNA was measured by real-time qPCR (ABI 7300; Applied BioSystems) using a one-step program: 50°C, 2 min; 95°C, 10 min; 95°C, 15 s, 60°C, 1 min for 50 cycles. Hypoxanthine-guanine phosphoribosyltransferase mRNA was measured as a normalization control. The hypoxanthine-guanine phosphoribosyltransferase probe, primers, and Taqman 2ϫ qPCR master mix were purchased from Applied BioSystems (Foster City, CA). Epo primer and probes were previously described (11). Samples without cDNA were analyzed as "no template" controls. Samples were also studied without adding reverse transcriptase to confirm the absence of contamination with genomic DNA.

Epo Activates ERK in Breast Cancer
Cells-MCF-7 breast cancer cells typically demonstrate limited motility; however, agents that have been implicated in breast cancer progression, including urokinase-type plasminogen activator (uPA), epidermal growth factor, and heregulin activate ERK in MCF-7 cells and promote MCF-7 cell migration (25,27,28). MCF-7 cells express EpoR (29). Thus, to begin, we examined cell signaling in response to exogenously added Epo in MCF-7 cells.
Phosphorylated EpoR was increased within 1 min of exposure to Epo (1.7 nM), as determined by immunoblot analysis (Fig. 1A). Within 2 min, ERK also was phosphorylated (Fig. 1B). Both isoforms of ERK (p42 and p44) were observed with phospho-ERK-specific antibody. The largest signal was observed at 15-60 min; however, ERK activation was transient. Phosphorylated ERK returned to baseline by 120 min.
The Janus kinase-2 inhibitor, AG490, inhibited ERK phosphorylation in response to Epo in MCF-7 cells by 63 Ϯ 12% (p Ͻ 0.05, n ϭ 5). These results suggest that ERK activation occurs, in part, downstream of Janus kinase-2 (Fig. 1C). Equivalent results were reported previously by us in studies with primary Schwann cell cultures (11). Epo Promotes Cell Migration-MCF-7 cells that were treated with 1.7 nM Epo migrated across serum-coated Transwell membranes at a rate that was increased by 2.1 Ϯ 0.1-fold (p Ͻ 0.005, n ϭ 7) compared with untreated cells (Fig. 2A). We previously reported a similar increase in MCF-7 cell migration in response to uPA (25). The effects of uPA on MCF-7 cell migration reflected activation of ERK and, downstream of ERK, myosin light chain kinase (24,25).
To test the role of ERK in Epo-stimulated MCF-7 cell migration, we treated cells with the highly specific MEK inhibitor, PD098059. In the absence of Epo, PD098059 had no effect on MCF-7 cell migration ( Fig.  2A); however, PD098059 inhibited the Epo-induced increase in migration by 80 Ϯ 10% (p Ͻ 0.005, n ϭ 7).
To confirm the role of ERK as a critical effector of Epo-stimulated MCF-7 cell migration, cells were transfected to express dn-MEK1 and GFP or GFP alone. Cell migration was studied using Biocoat inserts, which allowed selective monitoring of fluorescent cells. As shown in Fig.  2B, dn-MEK1 had no effect on the basal level of MCF-7 cell migration but inhibited the pro-migratory activity of Epo by 90 Ϯ 18% (p Ͻ 0.05, n ϭ 9), confirming the results obtained with PD098059.
The K D for Epo binding to EpoR dimer is ϳ0.1 nM in erythrocyte precursors (30), although binding affinities for Epo are different in various cell types. In concentration dependence experiments, 13 pM Epo was sufficient to significantly increase MCF-7 cell migration (Fig. 2C); however, the curve was bimodal. Epo concentrations exceeding 10 nM were less efficacious. The reason for the bimodal response is unclear; however, similar results have been observed with uPA receptor-derived chemokinetic peptides, which also stimulate MCF-7 cell migration by activating ERK (31).
Hypoxia Activates ERK and Promotes MCF-7 Cell Migration-When MCF-7 cells were exposed to 1.0% O 2 for 3 h, Epo mRNA increased 2.4 Ϯ 0.5-fold, as determined by real-time qPCR (p Ͻ 0.05, n ϭ 9) (Fig.  3A). A modest but statistically significant increase in the basal level of phosphorylated ERK was also observed (p Ͻ 0.005, n ϭ 6) (Fig. 3B). Based on these results, we examined the effects of hypoxia on MCF-7 cell migration. MCF-7 cells were added to Transwell chambers, which were immediately transferred to a 1.0% O 2 environment. Migration was measured 6 h later. Fig. 4 shows that cell migration was increased by 2.0 Ϯ 0.1-fold (p Ͻ 0.001, n ϭ 9) in the low oxygen atmosphere compared with cells that were maintained under normoxic conditions. The increase in cell migration was inhibited by Ͼ70% by PD098059 (p Ͻ 0.001, n ϭ 11), again suggesting an essential role for ERK.
The hypoxia-induced increase in endogenously produced Epo suggested that autocrine EpoR signaling may be responsible for the increase in the basal level of MCF-7 cell ERK activation and migration in 1% O 2 . Hypoxia also has been reported to increase autocrine signaling though other receptors capable of activating ERK and promoting migration in cancer cells, including uPA receptor and c-Met (32,33). To test the role of autocrine EpoR signaling, MCF-7 cell migration was studied in 1% O 2 in the presence of soluble EpoR (2.5 g/ml). The increase in cell migration associated with hypoxia was inhibited by 53 Ϯ 5% (p Ͻ 0.005, n ϭ 9) (Fig. 5). A number of experiments were performed with higher concentrations of EpoR (up to 8 g/ml); however, the degree of inhibition was not increased. Soluble EpoR had no effect on MCF-7 cell migration under normoxic conditions.
Epo-neutralizing antibody also blocked the hypoxia-induced increase in MCF-7 cell migration. In this case, the degree of inhibition was 98 Ϯ 18% (p Ͻ 0.005, n ϭ 5). Epo-neutralizing antibody did not affect MCF-7 cell migration under normoxic conditions. Furthermore, nonspecific  , n ϭ 7). B, MCF-7 cells were co-transfected to express dn-MEK1 and GFP or GFP alone and then treated with Epo (1.7 nM) or vehicle as indicated. Cells were allowed to migrate for 6 h on Biocoat cell culture inserts. Cell migration is expressed as a percentage of that observed in cells transfected with pEGFP alone (mean Ϯ S.E., n ϭ 9). C, MCF-7 cells were treated with Epo at the indicated concentrations and allowed to migrate for 6 h. Cell migration is expressed as a percentage of that observed in the absence of Epo (mean Ϯ S.E., n ϭ 3; *, p Ͻ 0.05 by Student's t-test) murine IgG, at the same concentration, did not affect cell migration under normoxic or hypoxic conditions.

DISCUSSION
Under normal physiologic conditions, mammalian cells in vivo are exposed to PO 2 levels of 40 -100 torr (6 -16%), well below the standard PO 2 used in cell culture (150 torr) (34). There is considerable evidence that PO 2 levels may be much lower in poorly perfused regions of cancers. The resulting hypoxia has an incompletely understood effect on cancer cell physiology; however, there are studies that correlate cancer hypoxia with poor prognosis (35)(36)(37).
At the molecular level, hypoxia is associated with substantial changes in the pattern of gene expression (38). Many changes occur downstream of HIF-1 (reviewed in Ref. 39). HIF-1␣ is localized primarily in the cytoplasm, where it undergoes ubiquitination and proteasomal degradation; however, in hypoxia, these processes are inhibited and HIF-1␣ enters the nucleus where it is available to dimerize with HIF-1␤ to form the active transcription factor (40). Epo is among the many gene products that are transcriptionally regulated by HIF-1 (41), providing one explanation for the increase in Epo mRNA observed when MCF-7 cells are exposed to 1.0% O 2 , as reported here and elsewhere (29).
Epo may be viewed as orchestrating the response to hypoxia. In addition to its effects on erythroid development, Epo has pro-angiogenic activity (42) that may be disadvantageous in cancer. In this study, we focused on the signaling activity of Epo and its effects on cell migration, which represents a critical component of the cancer metastasis cascade (43). In MCF-7 cells, ERK activation is associated with increased cell migration, and this is at least partially due to activation of the down-stream kinase, myosin light chain kinase (24). Myosin light chain kinase plays an important role in focal adhesion turnover in the migrating cell (44).
In previous studies, we demonstrated that uPA induces highly transient activation of ERK in MCF-7 cells, accompanied by an increase in cell motility that is sustained for over 24 h (25). In the present study, we show that Epo has similar effects on MCF-7 cells. Under normoxic conditions, exogenously added Epo activated ERK and promoted cell migration by a pathway that was dependent on ERK, as demonstrated in experiments with PD098059 and dn-MEK1. ERK activation by Epo was transient (Ͻ2 h); however, the response was actually sustained for a longer time than what was previously observed with uPA (25).
Under hypoxic conditions, the basal level of MCF-7 cell ERK activation and cell migration was increased, suggesting the establishment of autocrine signaling pathways that support these processes. To identify the responsible autocrine signaling pathway, it was necessary to consider a number of receptor systems. Hypoxia increases uPAR expression in cancer cells (32). Furthermore, the Met tyrosine kinase is up-regulated and activated in cancer (33). Nevertheless, our results with Eponeutralizing antibody indicate that EpoR autocrine signaling plays a principal role in stimulating MCF-7 cell migration in hypoxia. Supporting results were obtained in experiments with soluble EpoR. The degree of inhibition of cell migration achieved with soluble EpoR was Ͼ50% but still less than that observed with Epo-neutralizing antibody. The reason for this is unclear but under investigation.
From these studies, a model emerges in which multiple extracellular factors contribute to activation of common signaling pathways involved in cancer cell migration and perhaps other processes that support cancer cell invasion and metastasis. Our results demonstrate that Epo is a potentially important signal for promoting cell migration. The relative importance of Epo, compared with other factors such as uPA and EGF receptor ligands, may depend on the PO 2 and thus vary within different locations of a single tumor.