Erythropoietin induces cancer cell resistance to ionizing radiation and to cisplatin

Recombinant human erythropoietin (rhEPO) has been widely used in the clinic for treatment of anemia associated with cancer, with the potential for a possible concomitant effect to increase tumor oxygenation and thereby, radiocurability (1). Hypoxic cells are more resistant to inactivation by low-linear energy transfer X- and γ-rays than well-oxygenated cells (2, 3). Consequently, the radiosensitivity of malignant neoplasms is decreased by the presence of either chronically or transiently hypoxic neoplastic cells. Moreover, hypoxic neoplastic cells are clinically more aggressive and have a more metastatic potential than their well-oxygenated counterparts (4, 5).

Recently, experimental findings from various laboratories showed that the expression of erythropoietin and its receptor (EPOR) is not restricted to the hematopoietic system. Erythropoietin and EPOR are expressed in cancer cells from the kidney (6), brain (7), breast (8), and as well as other tissues (9–12). These findings suggest that the biological effects of erythropoietin are not limited to the regulation of erythropoiesis. Erythropoietin may have a broader role in regulating normal and neoplastic cell survival and in enhancing the process of angiogenesis. An autocrine erythropoietin signaling has been recently showed by Acs et al. (13) to inhibit hypoxia-induced apoptosis in human breast carcinoma cells.

Erythropoietin is a M_r ∼30,000 glycoprotein hormone produced by kidney and fetal liver. It acts via the EPOR to stimulate growth, prevent apoptosis, and induce differentiation of RBC precursors (14, 15). The interaction between erythropoietin and its receptors is in the ratio of one molecule of erythropoietin to two molecules of EPOR resulting in a triggering receptor-mediated signaling event with a subsequent receptor internalization and degradation (16). Erythropoietin receptor, although a member of the cytokine receptor family, has no intrinsic kinase activity. It binds and activates intracellular tyrosine kinases [Janus kinase (JAK)] to elicit its mitogenic effects (17). JAKs represent a family of cytoplasmic tyrosine kinases that play a major role in cellular signaling process (18, 19). Having bound to a ligand, JAKs are activated and subsequently phosphorylate a number of substrates, including the receptors themselves. The phosphorylated receptors provide docking sites for the SH2 domain containing the signal transducers and activators of transcription (STAT) transcription factor family. Subsequently, JAKs phosphorylate the STATs on a single tyrosine residue, causing STATs dimerization, migration to the nucleus, and regulation of gene transcription. Among different JAKs, the JAK2 is of particularly interest. It is activated upon binding of erythropoietin to its receptor (20). A carboxyl-truncated JAK2 mutation caused significant negative effects on erythropoietin-induced JAK2 activation and mitogenesis (21). JAK2 activation subsequently leads to the phosphorylation and activation of STAT5 (22). In addition to JAK2/STAT5, erythropoietin also activates other signal transduction pathways, such as the extracellular signal-regulated kinase pathway and signal transduction molecules, such as SHC/GRB-2, CRK-1, Vav, phosphatidylinositol 3-kinase, and the protein phosphatases SHP-1and SHP-2 (23–27). Activation of these pathways requires an intact form of the EPOR.

In this study, we investigated the in vitro effect of erythropoietin/EPOR-mediated JAK2/STAT5 signaling pathway in cancer cells exposed to ionizing radiation and to cisplatin.

U87 and K562 cells were obtained from American Type Culture Collection (Rockville, MD). HT100 primary cells and cervical carcinoma biopsy samples designated HT67, HT219, HT270, and HT280 were generously provided by Dr. Joan Turner (Cross Cancer Institute, University of Alberta). U87 and HT100 cells were cultured in DMEM and K562 cells were cultured in RPMI 1640 (Life Technologies, Inc., Gaithersburg, MD) supplemented with 10% (v/v) of fetal bovine serum (FBS), batch number 22524017 (BioMedia Canada, Drummondville, Quebec, Canada). FBS heat inactivation was conducted at 56°C for 30 minutes. Heat inactivation was done in order to inactivate any endogenous erythropoietin in FBS before its use in the experiments. rhEPO, EPREX epoetin alfa, was obtained from Ortho Biotech (Janssen-Ortho, Inc., Toronto, Ontario, Canada). Tyrphostin AG490 was obtained from Sigma Chemical Co. (St. Louis, MO). All cell lines were maintained in a humidified atmosphere of 5% CO₂ at 37°C.

Whole-cell extracts were prepared by a single freeze-thaw cycle (28). Briefly, cells were harvested, washed with ice-cold PBS and lysis-supporting buffer [25 mmol/L KCl, 10 mmol/L NaCl, 1 mmol/L MgCl₂, 10 mmol/L HEPES-KOH (pH 7.2), 0.1 mmol/L EDTA, 2 μg/mL leupeptin (Sigma Chemical), 2 μg/mL aprotinin (Sigma Chemical), 0.5 mmol/L DTT, 0.5 mmol/L phenylmethylsulfonyl fluoride], repelleted, and resuspended in a lysis-supporting buffer to a final volume of 2.5 times the original packed cell volume. Resuspended cells were incubated on ice for 10 minutes, frozen by plunging into liquid nitrogen, quickly placed at 37°C until slightly thawed, and placed on ice. NaCl and MgCl₂ concentrations of fully thawed samples were adjusted to 0.5 and 10 mmol/L, respectively, by adding high-salt buffer (5 mol/L NaCl, 100 mmol/L MgCl₂, and 5 mmol/L DTT). Cell extracts were incubated on ice for 5 minutes and microcentrifuged at 10,000 × g for 5 minutes at 4°C. Supernatants were collected, and protein concentrations were quantified using the method of Bradford (29).

Cell extracts containing 10 to 50 μg of protein were separated by SDS-PAGE and transferred by electrophoresis onto nitrocellulose membranes (Millipore, Bedford, MA) for 1 hour in 25 mmol/L Tris-HCl, 192 mmol/L glycine, and 20% methanol. Membranes were blocked in TBS (Tris-HCl) containing 5% nonfat milk and probed with anti-EPOR (C-20, Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and α-tubulin monoclonal antibodies (NeoMarkers, Fremont, CA). Antigen/antibody complexes were detected by horseradish peroxidase-conjugated anti-mouse antibody and enhanced chemiluminescence (Amersham Pharmacia Biotech, Piscataway, NJ), according to the manufacturer's instructions. The bands were quantified by densitometric analysis using Scion Image software (Scion Co., Frederick, MD).

U87 and HT100 cells were cultured for 24 hours on eight-well Lab-Tek II Chamber Slides (Nalge Nunc International Co., Naperville, IL). Cells were washed gently with PBS and fixed for 3 minutes with ice-cold 100% acetone. The cells were covered with 5% bovine serum albumin in PBS and incubated for 1 hour at room temperature. The cells were washed once with PBS and incubated with mouse monoclonal anti-human EPOR fluorescein-conjugated antibody (R&D Systems, Minneapolis, MN) overnight at 4°C, at 1:200 dilution in 1% bovine serum albumin in PBS. Following incubation, the cells were washed twice with PBS for 5 minutes and coverslips were mounted using a nail polish and examined by fluorescence microscope.

The in vitro biological activity of rhEPO was determined by measuring the hemoglobin production of the human erythroleukemia cell line K562. Fetal bovine serum (FBS) heat inactivation was done in order to inactivate any endogenous erythropoietin in FBS before its use in the experiments. Heat inactivation of FBS was a critical and necessary step in the experimental design of this study, in order to minimize any confounding effect of endogenous erythropoietin in FBS that may otherwise mask the effect of exogenous rhEPO when the latter is added. K562 cell suspension was diluted in culture media supplemented with heat-inactivated or regular FBS at 2.5%, 5.0%, 10.0%, and 20.0% (v/v) in the presence or absence of 0, 2.5, 5.0, and 10.0 units/mL of rhEPO to yield 1.0 × 10⁵ cells/mL. The cell suspension (100 μL) was transferred into a 96-well microtiter plate using a multichannel pipette. Following 96 hours of incubation, the cells were assayed for the presence of hemoglobin using the 2,7-diaminofluorene colorimetric reagent as previously described (30). Briefly, following 96 hours of incubation, the cells were centrifuged and the growth medium aspirated. The cells were lysed with ice-cold water containing 1% of NP40, and cell lysates were adjusted to protein concentration of 2.5 mg/mL in 0.2 mL volumes. Working 2,7-diaminofluorene solution (0.6 mL) was added to the samples and the mixture was incubated for 5 minutes at room temperature. Absorbance measurements were made at 610 nm within 10 minutes following the incubation period.

The effect of ionizing radiation or cisplatin (Sigma Chemical) treatment on the survival of U87 and HT100 cells cultured in 10% heat-inactivated FBS and stimulated with 30 units/mL of rhEPO, was measured using the 3-(4,5-dimethylthiazole-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and clonogenic survival assays. Tyrphostin B42 (AG490, Calbiochem, La Jolla, CA) was used as a specific inhibitor of JAK2 tyrosine kinase activity. For the MTT assay, the cells were plated on 96-well plates at 5,000 cells/well. The cells were allowed to attach to the bottom of the wells for 6 hours after which the cells were exposed to various concentrations of cisplatin. Following 6 hours of incubation, the cisplatin-containing medium was aspirated. The cells were washed gently with 0.2 mL of serum-free medium and incubated with 0.2 mL of corresponding growth media for 72 hours. A solution of MTT (50 μL of a 5 mg/mL solution in PBS) was then added, and the cells were incubated for another 3 hours. The medium was removed and replaced by 100 μL of DMSO, and the plate absorbance was measured at 570 nm. For the clonogenic survival assay the cells were plated onto 25-cm² tissue culture flasks (Sarstedt, Newton, NC) at various concentrations (200–15,000 cells/well) and exposed to γ-irradiation from a ¹³⁷Cs source for time periods equivalent to 2.0, 4.0, 6.0, 8.0, and 10.0 Gy. After 14 days, cells were washed with PBS, fixed with gluteraldehyde, and stained with 1% crystal violet. Washed dishes were dried, and colonies containing >50 cells were scored. The surviving fraction was plotted on a log scale against drug concentration or radiation dose.

The two-tailed Student's t test was used to compare statistical significance among various groups.

Western blot analysis of U87, HT100 cells, and cervical carcinoma biopsy samples designated HT67, HT219, HT270, and HT280 showed that a specific immunoreactive band, of M_r ∼66,000 and corresponding to the EPOR protein, was present in all samples (Fig. 1A). Immunocytochemical labeling showed that both U87 and HT100 cells expressed EPOR on the cell surface (Fig. 1B).

Our initial results of U87 and HT100 cell proliferation and survival showed large variations between experiments done with different batches of FBS. We also noticed that addition of the rhEPO to the growth medium had little or no effect when the FBS used for preparation of the growth medium was not heat inactivated. It has been previously reported that heat inactivation of serum samples at 56°C destroyed erythropoietin in specimens tested by the mouse spleen cell assay (31). In the light of our initial results, we hypothesized that the effect of rhEPO on U87 and HT100 cell proliferation and survival might be masked by the levels of endogenous erythropoietin present in normal noninactivated FBS, and that this endogenous level of bovine erythropoietin varies from one batch to another. It has been previously reported that rhEPO induced differentiation in the K562 human erythroleukemia cell (32–34). Therefore, we used this method to test our hypothesis by studying the effect of FBS heat inactivation and the effect of rhEPO addition in K562 human erythroleukemia cells. Erythroid differentiation of K562 cells was induced by culturing the cells for 4 days in medium containing heat-inactivated or normal noninactivated FBS at 2.5%, 5.0%, 10.0%, and 20.0% (v/v) and in the presence or absence of 0, 2.5, 5.0, and 10.0 units/mL of rhEPO (Fig. 2A and B). Addition of rhEPO to heat-inactivated serum led to a significant increase in K562 differentiation in a dose-dependent manner (Fig. 3A), whereas no effect of rhEPO addition was seen in the samples where normal noninactivated FBS was used to prepare growth medium (Fig. 2B).

Colony survival and the MTT assays were used to determine the effect of rhEPO on the sensitivity of U87 and HT100 cells to ionizing radiation and to cisplatin, respectively. Both cell lines were cultured in growth medium containing heat-inactivated FBS. At first, we determined if rhEPO addition alone would affect cell sensitivity to ionizing radiation and cisplatin. Treatment of U87 and HT100 cells with 30 units/mL of rhEPO was associated with a decrease in radiation sensitivity compared to untreated controls (ID₅₀; 7.4 versus 10.1 Gy and 4.1 versus 5.0 Gy, respectively; P < 0.01; paired t test) (Fig. 3; Table 1 ). Similarly, treatment of U87 and HT100 cells with 30 units/mL of rhEPO was associated with decrease in sensitivity to cisplatin in both cell lines (ID₅₀; 14.0 versus 28.6 μmol/L and 6.7 versus 15.7 μmol/L, respectively; P< 0.01;paired t test; Fig. 4; Table 1). To investigate whether inhibition of JAK2 kinase, an effector of the EPOR,affects U87 and HT100 cell sensitivity to ionizing radiation and to cisplatin, we tested the effect of a specific JAK2 inhibitor tyrphostin (AG490) on the erythropoietin-induced cell survival. As illustrated in Figs. 3 and 4 and Table 1, the administration of AG490 alone did not affect U87 and HT100 cell sensitivity to ionizing radiation or to cisplatin, respectively (ID₅₀; 7.4 versus 6.9 Gy and 4.1 versus 4.0 Gy, respectively; P < 0.01; paired t test). However, when AG490 was present in the culture medium of cells treated with rhEPO, the rhEPO-induced cell survival of U87 and HT100 cells was abolished and the level of sensitivity returned back to control values. The result suggested that the observed effect of rhEPO-induced resistance to ionizing radiation and cisplatin was JAK2 dependent.

Anemia was shown to be a strong predictor of poor outcome for patients treated with radiotherapy (35). Attempts have been made in order to increase oxygen delivery to tumors and to improve treatment response and overall survival. One method that may improve oxygen delivery to hypoxic tumor tissues is increasing the hemoglobin levels. Indeed, patients with cervical carcinoma who received blood transfusions showed improved outcomes with radiation therapy (36). However, due to the risk of HIV and hepatitis transmission from contaminated blood, the patients are often not transfused unless their hemoglobin levels fall below 100 g/L. An alternative method for maintaining higher red cell count is the administration of rhEPO. Several studies have shown that rhEPO was effective in increasing hemoglobin levels in cancer patients and that the increase was also associated with an improvement in patient-reported quality of life (37). Whereas there is evidence of an erythropoietin enhancement of radiosensitivity of xenografted human tumor in anemic nude mice, the radiosensitivity of tumor in nonanemic animals' tumor could not fully be restored (38). However, recent experimental evidences from several laboratories (10–12) have raised the possibility of adverse effects related to rhEPO treatment in cancer patients (39, 40). Western blot analysis of U87, HT100, and cervical carcinoma biopsy samples designated HT67, HT219, HT270, and HT280 showed that a specific immunoreactive band corresponding to M_r ∼66,000 of the EPOR protein was present in all samples (Fig. 1A). Immunocytochemical labeling of both U87 and HT100 cells has also confirmed EPOR expression on the cell surface (Fig. 1B). Thus, our results are in agreement with recent studies showing that the EPOR is expressed in nonhematopoietic tissues (7–9).

One difficulty in studying the effects of exogenous rhEPO in cancer cells is inconsistency of the results. Whereas we could clearly show the existence of EPOR in the cancer cell lines studied in the current work, we could not consistently show an effect on the corresponding cellular response to rhEPO. The effect of exogenous rhEPO was observed to vary from one batch of FBS to another and that the level of erythropoietin as measured with the anti-erythropoietin antibody indicated the level of erythropoietin in FBS is low (100 milliunits/mL). The discrepancy in the finding indicates that although the bovine erythropoietin may be functionally active as the human erythropoietin with the cell line, it might not be immune-crossreact with the anti-erythropoietin antibody used in the erythropoietin detection method. A functional assay of the erythropoietin activity in the serum may be a more reliable measurement of the endogenous erythropoietin activity in the FBS. It has been previously shown that K562 cells cultured in growth medium supplemented with 10% FBS possess constitutively phosphorylated STAT5, whereas incubation of the cells in serum-free medium for 24 hours reduced the base level of STAT5 tyrosine phosphorylation (41–43). Moreover, treatment of serum-starved K562 cells with erythropoietin 30 units/mL of rhEPO lead to an increase in the tyrosine phosphorylation of STAT5, suggesting that presence of rhEPO in the cultured medium was required for STAT5 tyrosine phosphorylation. Although the STAT5 protein is a substrate for the JNK2, which becomes activated in response to erythropoietin binding to EPOR (44, 45), other extracellular signaling proteins were shown to be involved in JAK-STAT signaling (19). Nevertheless, because erythroid cell differentiation of K562 cells was shown to be erythropoietin dependent, it allowed us to utilize this cell line to determine the factor(s) behind the inconsistency of our early results. The effect of serum heat inactivation and addition of rhEPO to the growth medium on K562 cells erythroid differentiation were analyzed in different batches of the FBS. Without heat inactivation, the FBS is very capable of induce K562 cells erythroid differentiation, indicating of high level of erythropoietin activity in the serum sample. On the other hand, heat inactivation of the FBS leads to a lost of K562 cells erythroid differentiation ability (Fig. 2) and that the addition of rhEPO to heat-inactivated serum led to a significant increase in K562 differentiation in a dose-dependent manner. No effect of rhEPO addition was seen in the samples where nonheat-inactivated FBS was used to prepare growth medium, suggesting that the presence of endogenous erythropoietin in noninactivated serum masked the effect of rhEPO in cell proliferation studies (Fig. 2). The results of nonheat-inactivated and heat-inactivated fetal bovine serum also indicate that in vitro investigation of erythropoietin effect on nonhematopoietic cells should be proceed with conditions setup for cytokines such as IFN, where all assays are carried out using heat-inactivated FBS. It is important to reduce the level of endogenous erythropoietin in FBS before studying the effect of rhEPO in cell cultures.

We use two methods in measuring the cell survival, clonogenic assay for ionizing irradiation and MTT assay for cisplatin. The presence of rhEPO in the culture media increases the cell survival to cytotoxic agent cisplatin and to ionizing irradiation. As mentioned above, erythropoietin activates many signal transduction pathways. These include the JAK2/STAT5, the extracellular signal-regulated protein kinase pathway, and signal transduction molecules such as SHC/GRB-2, CRK-1, Vav, phosphatidylinositol 3-kinase, and the protein phosphatases SHP-1and SHP-2 (23–27). Among the pathways, the JAK2 kinase has been shown to be an early effector of the EPOR signal transduction pathway (17) and seems to be the most specific for erythropoietin. Hence, we tested the effect of the JAK2 kinase inhibitor, Tyrphostin (AG490), on the erythropoietin induced survival in U87 and HT100 cancer cells. Our experimental findings showed that the increased resistance of U87 and HT100 cells to ionizing radiation and to cisplatin after exposure to rhEPO can be abolished by the treatment with the specific JAK2 kinase inhibitor, tyrphostin (AG490; Figs. 3 and 4; Table 1). The effect of AG490 is specific to the erythropoietin-induced resistance, as the administration of AG490 alone did not affect U87 and HT100 cell sensitivity to ionizing radiation or to cisplatin, respectively. The result indicates that the signal transduction pathway is mediated by erythropoietin, although the signal transduction protein JAK2 in the human malignant glioma U87 and the primary cervical carcinoma cell lines HT100 is intact and functional in promoting cell growth and in inhibiting apoptosis induced by erythropoietin exposure. Recently, Liu et al. (46) tested the effect of both erythropoietin and granulocyte macrophage colony-stimulating factor in a series of cancer cells. They found that coculture with either the mitogen-activated protein kinase kinase inhibitor U0126 or the granulocyte macrophage colony-stimulating factor antagonist E21R negated the induced resistance to cytotoxic chemotherapeutic agent cisplatin by granulocyte macrophage colony-stimulating factor. The result indicates that the mitogen-activated protein kinase signal transduction pathway mediated by the granulocyte macrophage colony-stimulating factor is intact in the tested cancer cells. It is not known, however, whether or not the JAK2 kinase is also involved, as this pathway was not tested. In the light of the result by Liu et al. (46), it is possible that the mitogen-activated protein kinase signal transduction pathway mediated by erythropoietin/EPOR might also be active in our cancer cell lines to provide resistance toward ionizing irradiation and to cisplatin. We are currently testing this possibility.

The presence of an intact erythropoietin pathway in cancer cells may help to explain the possible failure to show any benefit of disease control in the recent published clinical trials of erythropoietin in patients treated with radiation and/or chemotherapy (40, 47, 48). Our findings also indicate that the administration of rhEPO to cancer patients undergoing anticancer therapy should proceed with caution, especially when the cancer type has been shown to be positive for EPOR. Recently, Pinel et al. (49) showed that erythropoietin induces a reduction of hypoxia before and during fractionated irradiation. This resulted in an improvement of radioresponse in the human glioma xenografts (49). The status of the EPOR expression in the glioma cells used in the study, however, is unknown. The use of erythropoietin may be beneficial in improving the radiosensitivity and/or chemosenstitivity in EPOR negative tumors. Perhaps erythropoietin should be considered to be a more general cytokine in its biological effect rather than a specific cytokine for RBC precusors. In vivo studies are planned to determine the effect of erythropoietin administration on the radiosensitivity and/or the chemosensitivity of EPOR positive and negative neoplastic cells in animal models.

We thank Sigat Choudhury, Paul Kauler, and Jean-Pierre Paiement for their helpful discussions.