Background
Tumor cells within a growing lesion often need to adapt and survive in hypoxic conditions. One-way tumor cells are known to respond to hypoxia is to up-regulate the transcription factor hypoxia inducible factor (HIF). HIF has two subunits, HIF-1α and HIF-1β [
1], and intracellular oxygen levels can modulate HIF-1α levels, while HIF-1β is constitutively expressed [
2]. In normoxic conditions, it has been shown that a complex including functional von Hippel-Lindau (pVHL), a key tumor suppressor gene in clear cell renal cell carcinoma (RCC) is able to rapidly degrade HIF-1α [
3]. However, in the absence of a functional pVHL, HIF-1α can accumulate, in hypoxic or normoxic conditions [
4,
5]. When the HIF complex translocates to the nucleus it binds to hypoxia-response elements of DNA leading to the regulation of multiple hypoxia-inducible genes [
6,
7]. One of the lesser-known hypoxia-inducible genes encodes the glycoprotein, erythropoietin (EPO), which is in fact a hormone, produced by the kidneys and to a lesser extent the liver [
8]. EPO stimulates the production of red blood cells in the bone marrow [
9]. Accordingly, one of the key indications for its use is in the management of severe anemia [
10], a situation that can often occur during the administration of cytotoxic chemotherapy in the treatment of malignancies.
Recently, concerns have arisen over the potential of recombinant human erythropoietin (rhEPO) treatment and an association with tumor growth [
11,
12]. The effect may be induced through interaction with tumor cell EPO receptors (EPOR), which when activated promote the tumor vascularization required for adequate oxygenation [
13,
14]. An understanding of the mechanism of EPO in tumor biology and when EPO treatment is likely to be efficacious is an important goal at this juncture. In this study, we performed a series of
in vitro and
in vivo analyses to test whether EPO can stimulate the growth of renal cells. We found that rhEPO administration stimulated cellular proliferation, and the effect was enhanced in a hypoxic state, which we report for the first time. Mechanistic investigations revealed that EPO stimulates the expression of cyclin D1 while inhibiting the expression of p21
cip1 and p27
kip1 through the phosphorylation of JAK2 (JAK-Stat pathway) and ERK1/2 (MAPK pathway), leading to a more rapid progression through the cell cycle. We were also able to demonstrate that the growth of renal cell carcinoma xenograft tumors was increased in tumors with increased hypoxia when systemic rhEPO was administered. These investigations provide some insight into the mechanism of EPO in tumor cell stimulus, and show that the effects are significantly enhanced in association with hypoxic conditions.
Materials and method
Immunohistochemistry
Commercial tissue microarrays (TMA) (MC5003a, US Biomax, Inc., Rockville, MD) constructed from clinical samples obtained from a cohort of 500 patients (400 malignant tissues and 100 benign tissues from 20 different organs) were examined by immunohistochemical staining. The clinicopathologic variables of the study cohort are available at
http://www.biomax.us/tissue-arrays/Multiple_Organ/MC5003a. TMAs were examined by H&E for histological verification of disease status. TMAs were deparaffinized followed by antigen retrieval using citric acid buffer (pH 6.0, 95°C for 20 mins). Slides were treated with 1% hydrogen peroxide in methanol to block endogenous peroxidase activity. After 20 mins of blocking in 1% bovine serum albumin (BSA), the TMAs were incubated overnight at 4°C with anti-human EPO antibody (sc-7956; rabbit polyclonal, dilution 1/200 in 1% BSA) and anti-human EPOR antibody (sc-695; rabbit polyclonal, dilution 1/100 in 1% BSA) from Santa Cruz Biotechnology (Santa Cruz, CA). Next, the slides were incubated with 2 μg/mL of biotinylated anti-rabbit IgG secondary antibody (Vector Laboratories, Burlingame, CA) for 30 mins at room temperature. Subsequently, the sections were stained using Standard Ultra-Sensitive ABC Peroxidase Staining kit (Pierce/Thermo Fisher Scientific, San Jose, CA) and 3, 3'- diaminobenzidine (DAB; Vector Laboratories), counterstained by hematoxyline, dehydrated, and mounted with a cover slide. Mouse xenograft tumors from the human renal cancer cell line Caki-1, known to stain strongly for EPO and EPOR were used as a positive control.
The proportion of positive cells was scored by two investigators (AL, MM) in four grades and represented the estimated proportion of immunoreactive cells (0 = 0% of cells; 1 = 1% to 40%; 2 = 41% to 75% and 3 = 76% to 100%). The intensity was scored and represented the average intensity of immunopositive cells (0 = none; 1 = weak; 2 = intermediate and 3 = strong). The proportion and intensity scores were combined to obtain a total EPO or EPOR staining score, which ranged from 0 to 6. The EPO or EPOR expression level was determined based on the total EPO or EPOR staining score as follows: none = 0, low = 1 or 2, moderate = 3 or 4, high = 5 or 6 [
15]. A third investigator (CJR) reviewed discrepancies and rendered a final score. The comparison between EPO and EPOR expression in human tumors and benign tissues was calculated using Mann–Whitney U test.
Cells, reagents and equipment
Human renal cancer cell lines; Caki-1, 786-O, 769-P (ATCC, Manassas, VA), and the normal primary human renal tubule epithelial cells (RPTEC; Lonza, Walkersville, MD) were available for analysis. Cancer cell lines were maintained in RPMI1640 medium supplemented with 10% fetal bovine serum, 50 units/ml penicillin and 50 mg/ml streptomycin (Invitrogen Corporation, Carlsbad, CA). RPTEC was maintained in renal epithelial cell basal medium (REBM) supplemented with REGM complex (Lonza CC-3190). All cells were incubated in humidified atmosphere at 37°C in air with 5% CO2 (normoxic conditions). For hypoxic conditions, cells were incubated at 37°C containing 1% O2, 5% CO2, and balance N2 in a humidified incubator. The oxygen level was automatically maintained with an oxygen controller (ProOx P110; Biospherix, Redfield, NY) supplied with compressed nitrogen gas. Recombinant human EPO (rhEPO) was purchased from R&D Systems, Inc. (Minneapolis, MN).
Immunoblotting
Whole cell lysates were prepared using RIPA buffer with Halt Protease Inhibitor Cocktail (Thermo Fisher Scientific) as previously reported [
16]. Twenty micrograms of total protein (assessed using BCA protein assay) were subjected to SDS-PAGE using Mini-PROTEAN TGX precast gels (Bio-Rad Laboratories, Richmond, CA). Proteins were transferred to polyvinylidene difluoride (PVDF) membrane (Bio-Rad). Anti-human pVHL (#2738, dilution 1:1 000), HIF-2α (#7096, dilution 1:1 000), p-Jak2 (#4406, dilution 1:1 000), total Jak2 (#3230, dilution 1:1 000), p-Stat5 (#9359, dilution 1:1 000), total Stat5 (#9363, dilution 1:1 000), p-Akt (#4060, dilution 1:1 000), total Akt (#9272, dilution 1:1 000), p-ERK1/2 (#4370, dilution 1:1 000), total ERK1/2 (#9102, dilution 1:1 000), cyclin D1 (#2978, dilution 1:1000), cyclin D3 (#2936, dilution 1:1 000), CDK4 (#2906, dilution 1:1 000), CDK6 (#3136, dilution 1:1 000), p21
cip1 (#2947, dilution 1:1000), p27
kip1 (#3686, dilution 1:1 000) and p15 (#4822, dilution 1:1 000) were purchased from Cell Signaling Technology. Anti-human HIF-1α (sc-53546, dilution 1:200), VEGF (sc-152, dilution 1:200), EPO (sc-7956, dilution 1:1 000), total EPOR (sc-697, dilution 1:1 000) and p-EPOR (sc-20236, dilution 1:1 000) antibodies were purchased from Santa Cruz Biotechnology. Equal loading was confirmed with β-actin (AC-15, dilution 1:10 000, Sigma-Aldrich) [
17]. Stained proteins were detected using the ECL Plus Western Blotting Detection System (GE Healthcare).
Proliferation and viability assay
Human renal cells Caki-1, 786-O, 769-P and RPTEC were plated in 96 well dishes in triplicate (103 cells/well) and incubated in normoxic condition. Cells were then subjected to increasing doses of rhEPO (0–50 units/mL) and incubated in normoxic or hypoxic conditions. After 48 hrs, cell proliferation was determined by CellTiter-Glo Luminescent cell viability assay (Promega, Madison, WI) according to manufacturer’s instructions. Luminescence was measured using a FLUOstar Optima Reader (BMG LABTECH, Ortenberg, Germany). Three independent experiments were performed in triplicate.
Cell cycle analysis
Human renal cells were seeded in 6-well plates at a density of 2 × 105 cells per well and incubated for 24 hrs. Cells were starved for 18 hrs in serum/growth factors-free media containing 0.1% BSA in normoxic or hypoxic condition. After starvation, media were replaced with fresh media containing 2% FBS with or without 2 units/mL of rhEPO and incubated for 10 hrs in normoxic or hypoxic condition. Cells were harvested and fixed with 70% ethanol overnight at -20°C. Next, cells were suspended in propidium iodide (PI) staining buffer containing 50 μg/ml PI and 200 μg/ml RNase A and incubated in 37°C for 15 min. PI fluorescence was determined by flow cytometry using a FACSCalibur and CellQuest software for acquisition (BD Biosciences, San Jose, California). Cell cycle phase distribution was analyzed and reported by using FlowJo software (TreeStar Inc., Ashland, OR). Three independent experiments were performed in triplicate.
Cell synchronization and measurement of DNA synthesis using EdU labeling
To obtain populations of cells in G
0/G
1 phase, all human renal cells were arrested by double thymidine block as described previously [
18]. Briefly, human renal cells were seeded at 5 × 10
4 cells per well in a 6-well plate. Cells were blocked for 18 hrs with 2.5 mM thymidine (Sigma-Aldrich), released for 6 hrs, washed to remove the thymidine, and then exposed again to 2.5 mM thymidine this time for 16 hrs in normoxia or hypoxia. The cells were then released from the double thymidine block by culturing in 2% FBS-containing fresh media with or without 2 units/mL of rhEPO and allowed to progress through G1 and into S-phase. The percentage of proliferating cells was determined at 0, 2, 4, 6, 9 and 12 hrs after release from the double thymidine block using the Click-iT® EdU Alexa Fluor® 647 Flow Cytometry Assay Kit (Life Technologies, Carlsbad, CA) according to the manufacturer's instructions. EdU (5-ethynyl-2´-deoxyuridine) is a thymidine analog that becomes incorporated into DNA during active cellular DNA synthesis. Detection is determined via a copper catalyzed covalent reaction between an azide (conjugated to Alexa Fluor 647) and an alkyne. EdU (10 μM) was added to each well 2 hrs prior to harvesting. Cells were trypsinized and fixed in 4% formaldehyde. Cell Quest Pro Software determined cellular DNA synthesis using FlowJo Software. Three independent experiments were performed in triplicate.
In vivo tumorigenicity
Animal care was in compliance with the recommendations of
The Guide for Care and Use of Laboratory Animals (National Research Council) and approved by our local IACUC. The subcutaneous tumorigenicity assay was performed in athymic BALB/c (nu/nu) mice, 6 to 8 weeks old purchased from Harlan Laboratories (Indianapolis, IN). Procrit (epoetin α; Amgen Inc, Thousand Oaks, CA) was used for the
in vivo treatment of EPO. The properties of rhEPO were tested
in vivo using a subcutaneous xenograft model by inoculating 10
6 Caki-1, 786-O and 769-P cells as described previously [
16,
19]. Since RPTEC cells are benign and not known to produce xenograft tumors, this cell line was not tested
in vivo. After 24 hrs, mice were divided randomly into two groups (control or 200 international units (IU)/kg of rhEPO) and treatment was initiated. RhEPO was administered subcutaneously once weekly. Control mice received vehicle alone (PBS) on the same schedule. At least 10 animals were in each group. Tumor volumes were measured twice weekly with digital calipers and calculated by V (mm
3) = length × (width)
2 × 0.5236. After 10 wks of treatment, the mice were sacrificed. However, 30 mins before being sacrificed, each mouse was intraperitoneally injected with 0.1 mL (60 mg/kg of body weight) of pimonidazole hydrochloride (Hypoxyprobe-1 Plus Kit; Hypoxyprobe Inc., Burlington, MA), according to the manufacturer's instructions [
20]. Subsequently, the mice were sacrificed and xenografts resected. The excised tumors were placed in 10% buffered formaldehyde solution and embedded in paraffin. Paraffin blocks were sectioned for H&E staining and immunohistochemical (IHC) staining.
Immunohistochemical (IHC) analysis of xenograft tumors
Paraffin embedded tumors were sectioned (4 μm), deparaffinized in xylene and rehydrated using graded percentages of ethanol. Slides were treated with 1% hydrogen peroxide in methanol to block endogenous peroxidase activity. Staining was conducted using anti-human EPO antibody (sc-7956, dilution 1:200), anti-human EPOR antibody (sc-695, dilution 1:100), HIF-1α (sc-53546, dilution 1:100), VEGF (sc-152, dilution 1:200), cyclin D1 (#2978, dilution 1:50), p21
cip1 (#2947, dilution 1:100), p27
kip1 (#3686, dilution 1:200), anti-human Ki-67 (MIB-1, dilution, 1:200; Dako). Biotin-labeled horse anti-mouse IgG or rabbit IgG (2 μg/ml in 1% BSA blocking buffer) was used as secondary antibody. Immunoreactive signals were amplified by formation of avidin-biotin peroxidase complexes and visualized using 3, 3'- diaminobenzidine (DAB). Nuclear counterstaining was conducted with hematoxylin. Proliferative index analysis was determined as previously described [
16]. In addition, slides were immunostained with fluorescein isothiocyanate (FITC)–conjugated primary antibody against pimonidazole (1:50) and horseradish peroxidase–labeled secondary anti-FITC monoclonal antibody (1:50) supplied with the hypoxia detection kit (Hypoxyprobe-1 Plus Kit), according to a modification of the manufacturer's instructions as described previously [
20].
Statistical analyses
All data are expressed as mean ± standard deviation (SD) and mean ± standard error of the mean (SEM). Statistical analyses were conducted using GraphPad Prism 5.0 (GraphPad Software, Inc.). The comparison between EPO and EPOR expression in cancer vs. benign tissue was calculated using Mann–Whitney U test. For most in vitro and in vivo comparisons, a 2-tailed unpaired Student t test or Mann–Whitney U test was conducted. Differences were considered statistically significant at p < 0.05.
Discussion
Questions were first raised about the possible exacerbating influence of EPO on human tumors after a landmark study was published in 2003 [
12]. Specifically, Heinke
et al. reported significantly shorter progression-free survival and overall survival in a cohort of head and neck cancer patients who were receiving radiation therapy and rhEPO, the latter presumably administered to overcome therapy-induced anemia. In a comparable cohort, Overgaard and colleagues subsequently reported a similar reduction in survival of head and neck patients undergoing tumor therapy while receiving rhEPO [
27]. Table
1 illustrates the meta-analysis results of Glaspy
et al. that examined EPO affects on disease progression in cancer patients receiving chemotherapy [
28]. When outcomes were analyzed ‘per protocol’, there was no significant effect of rhEPO on disease progression. However, a post-hoc analysis reported by Henke
et al. including erythropoietin receptor (EPOR) expression suggested that loco-regional progression-free survival was poorer in patients with EPOR-positive tumors receiving rhEPO [
29]. Unfortunately, additional studies using this EPOR antibody revealed problems of non-specific binding of the antibody thus reducing the validity of these results [
30]. In the genitourinary literature, only limited reports have commented on RCC disease progression in patients receiving rhEPO [
31‐
33]. Thus, the equivocal data does not allow one to draw definitive conclusions. Consequently, we are confronted with conflicting results when assessing the affects of rhEPO administration in cancer patients.
Table 1
Meta-analysis results of oncology trials that examined erythopoietin’s affect on disease progression in patients receiving chemotherapy
Osterborg et al. 1996[ 34] | Hematologic | 144 | 1.20 (0.60-2.40) |
Littlewood et al. 2001[ 35] | Solid (non-hematologic) | 375 | 0.64 (0.40-1.02) |
| Breast | 223 | 1.02 (0.46-2.26) |
Vansteenkiste et al. 2002[ 37] | SCLC and NSCLC | 314 | 0.58 (0.30-1.11) |
| Hematologic | 344 | 1.08 (0.66-1.76) |
Vadhan-Raj et al. 2003[ 39] | Gastric and rectal | 60 | 1.01 (0.35-2.94) |
| Breast | 354 | 0.82 (0.39-1.72) |
| SCLC | 224 | 0.85 (0.50-1.44) |
Leyland-Jones et al. 2005[ 42] | Breast | 939 | 0.84 (0.64-1.08) |
Osterborg et al. 2005[ 43] | Hematologic | 343 | 0.74 (0.44-1.25) |
| Mixed | 344 | 1.20 (0.75-1.91) |
Wilkinson et al. 2006[ 11] | Ovarian | 181 | 7.47 (0.95-58.54) |
| Hodgkin’s lymphoma | 1303 | 0.86 (0.33-2.24) |
| Breast | 463 | 1.07 (0.82-1.40) |
| SCLC | 596 | 0.87 (0.52-1.46) |
| Cervical | 74 | 0.87 (0.32-2.33) |
| Cervical | 109 | 1.02 (0.48-2.15) |
Similarly,
in vivo model studies on the topic are contradictory. In a Lewis lung carcinoma xenograft model, rhEPO was noted to increase primary tumor growth [
50]. However in ovarian and other xenograft models, systemic administration of rhEPO did not result in growth of primary tumors [
51,
52]. Our results demonstrate the importance of assessing more than one cell line
in vitro and
in vivo. Though all of the cells in our study possessed EPOR, we demonstrated that the administration of rhEPO resulted in the stimulation of growth of 786-O xenograft tumors, but not of Caki-1 xenografts. The only significant difference in the composition of these xenograft tumors was that 786-O possessed more regions of hypoxia; a state in which significantly exacerbates the effects of rhEPO
in vitro. It was critical to assess these cell lines in an
in vivo model, because similar to Fujisue and others [
53], we noted in
in vitro that Caki-1 cells had an increase in proliferation when exposed to rhEPO in the normoxic or the hypoxic state. However, this was not reproduce in the xenograft model thus we were able to postulate that tumors with a reduced oxygen tension (
e.g., large, expansive tumors) are more likely to be stimulated when exposed to EPO. Regarding our
in vivo experiments, we noted a failure of 769-P cells to grow as subcutaneous tumors in nude mice. Though reported as tumorigenic by ATCC, limited studies have reported on this aspect [
54,
55]. However, our
in vitro results of 769-P cells are similar to previously published 769-P
in vitro results [
53].
In our IHC tissue arrays in which tissue hypoxic status was unknown, EPO expression score was significantly elevated in lung cancer (
p = 0.003) and lymphoma (
p = 0.018), but not in RCC (
p = 0.91). Furthermore, EPOR expression score was significantly elevated in lung (
p = 0.011), lymphoma (
p = 0.007), thyroid (
p = 0.032), uterine (
p = 0.038) and prostate cancers (
p = 0.011), however it was not elevated in RCC (
p = 0.17). The lack of EPO or EPOR correlation by IHC in RCC
vs. benign samples substantiates a previous large cohort (n = 195) reported by Papworth
et al.[
21], but is contradictory to two small studies from Asia (combine n = 129) [
56,
57]. Interestingly a recent study noted that EPO levels were elevated in high stage RCC compared to low stage RCC [
58]. Thus further investigation into this, and correlating the tumor hypoxic status to EPO/EPOR expression may be warranted.
Our results provide evidence that EPO exposure leads to stimulation of JAK2 and ERK1/2 signaling, which in turn positively regulates progression through the cell cycle by inducing cyclin D1 and inhibiting p21
cip1 and p27
kip1 expression (Figure
4). The progression through the cell cycle is further potentiated under hypoxic conditions. Tumor hypoxia is noted in approximately 30% of RCC [
59] and is known to increase in all lesions as tumor burden increases. In this study, we present clear evidence that rhEPO is a potent mitogen, especially under hypoxia. Through pharmacologic stimulation, we also show that active JAK2 and ERK1/2 signaling tightly controls cyclin D1 expression in a panel of human cell lines (Figure
5). We have also found that exposure to rhEPO resulted in significant growth of 786-O xenografts (which contained many regions of hypoxia), with concomitant increased expression of cyclin D1 (Figure
6).
It is known that active EPOR can stimulate JAK2 kinase [
23] and cause subsequent activation of multiple signaling pathways, including the MAPK-ERK-1/2 pathway [
24]. For example, Jeong
et al. treated human ovarian cells with rhEPO (50,000 mU/ml) and noted an increase in the phosphorylation of extracellular signal related kinase (ERK)-1/2, but no change in cellular growth or survival [
25]. Similarly, treatment of lung cancer cells resulted in an increase in ERK-1/2 levels [
50]. We were able to confirm that rhEPO can induce JAK2 and ERK1/2 expression in renal cell lines. Furthermore, the increase in cellular proliferation seen with rhEPO could be abrogated with the addition of the JAK2 or ERK1/2 inhibitor (Additional file
1: Figure S1). Thus, cells can circumvent JAK2-dependent pathway for the JAK2-independent pathway (ERK1/2). Mannello and other previously reported about a JAK2-independent pathway [
60].
After synchronizing cells with a double thymidine block strategy, exposure to rhEPO was noted to more rapidly advance the cells through the cell cycle. Cursory studies have described how EPO may affect molecules related to cell cycle. For example, STAT5 is an intracellular protein associated with the cytoplasmic portion of EPOR with a noted interplay between the phosphorylation of JAK2 and STAT5 [
61]. Phosphorylated JAK2 forms homodimers and translocates to the nucleus where it directly binds to the DNA and activates cyclin D1 [
22]. We showed that EPO stimulation of two renal cell lines, RPTEC (normal primary human renal tubule epithelial cells with wild-type VHL) and Caki-1 (clear cell RCC with wild-type VHL), under normoxic conditions resulted in cyclin D1 overexpression. But in hypoxic conditions, rhEPO stimulation resulted in cyclin D1 upregulation in all four renal cell lines tested (Figure
3D), and this induction was accompanied by unabated progression through G1-phase of the cell cycle. Furthermore, rhEPO treatment, both in normoxic and hypoxic conditions, resulted in a down regulation of p21
cip1 and p27
kip1. Downregulation of these molecules was more pronounced during hypoxia, shedding light on molecular mechanisms involved and further confirming that EPO effects are exacerbated by hypoxia. The re-evaluation of large cohorts with respect to EPO and hypoxic state of the tumor could shed light on this phenomenon and help direct future clinical trials. These data presented herein suggest that rhEPO treatment may have adverse effects in specific scenarios and thus the use of rhEPO in the cancer patient should be considered carefully weighing the benefits and risks.
Competing interests
Makito Miyake, Adrienne Lawton, Ge Zhang and Evan Gomes Giacoia declare that they have no competing interests, while S. Goodison and C.J. Rosser are officers of Nonagen Bioscience Corporation.
Authors’ contributions
MM carried out in vitro and in vivo experiments, performed and analyzed IHC. AL analyzed IHC. GZ assisted in in vitro experiments. EGG assisted in in vivo experiments. SG assisted with drafting/revising manuscript. CJR conceived of the study, and participated in its design and coordination and helped to draft the manuscript and secured funding. All authors read and approved the final manuscript.