Erythropoietin regulates tumour growth of human ... - Carcinogenesis

8 downloads 60 Views 564KB Size Report
1a activates transcription of Epo (22). Most human primary and metastatic cancers express abundant HIF-1a, but benign tumours and normal tissues do not  ...
Carcinogenesis vol.24 no.6 pp.1021±1029, 2003 DOI: 10.1093/carcin/bgg060

Erythropoietin regulates tumour growth of human malignancies

Yoshiko Yasuda5, Yoshihiko Fujita, Takuya Matsuo, Satoshi Koinuma, Satoshi Hara1, Akira Tazaki, Mie Onozaki, Mitsuhiro Hashimoto2, Terunaga Musha3, Kazuhiro Ogawa4, Hiroyoshi Fujita4, Yukio Nakamura3, Hitoshi Shiozaki1 and Hiroshi Utsumi2 Department of Anatomy, Division 1 and 1Department of Surgery, Division 1, Kinki University School of Medicine, Osaka-Sayama, Osaka 589-8511, 2 Research Reactor Institute of Kyoto University, Osaka 590-0451, 3 Department of Obstetrics and Gynaecology, Kyorin University School of Medicine, Mitaka 181-8611 and 4Laboratory of Environmental Biology, Hokkaido University School of Medicine, Hokkaido 060-8638, Japan 5

To whom correspondence should be addressed Email: [email protected]

In addition to the chief function of erythropoietin (Epo) in promoting erythropoiesis, some other roles have been found in the brain and uterus. We have reported that signalling pathways of Epo and Epo receptor (EpoR) are involved in the tumourigenesis of ovarian and uterine cancers. To determine whether Epo plays a similar role in other malignancies, we studied the expression of Epo in several malignant human cell lines. We found that 24 malignant human cell lines examined express Epo and EpoR regardless of their origins, types, genetic characteristics and biological properties and secrete a very small amount of Epo individually and that most of them respond to hypoxic stimuli by enhanced secretion of Epo. To determine whether the Epo±EpoR pathway operates in tumours of these cell lines, we transplanted several cell lines into nude mice and confirmed the presence of Epo-responsive sites in xenografts in which the phosphorylation of the STAT5 (signal transducer and activator of transcription) is detectable. Furthermore, in nude mice we blocked the Epo signalling in xenografts of two representative cell lines, stomach choriocarcinoma and melanoma, by i.p. injections of EpoR antagonist and found inhibition of angiogenesis and survival of tumour cells leading to destruction of tumour masses and disturbances of phosphorylation of STAT5. In contrast, Epo mimetic peptide promotes angiogenesis and tumour cell survival. These findings suggest that Epo is indispensable for the growth and viability of malignant tumour and also that the deprivation of Epo signalling may be a promising therapy for human malignancy.

Abbreviations: ADC, adenocarcinoma; E2, estradiol-17b; EMP1, erythropoietin mimetic peptide 1; EMP9, erythropoietin mimetic peptide 9; Epo, erythropoietin; EpoR, erythropoietin receptor; HIF, hypoxia-inducible factor; IdU, 5-iodo-20 -deoxyuridine; STAT5, signal transducer and activator of transcription; TdT, terminal deoxy nucleotidyl transferase. #

Oxford University Press. All rights reserved

Downloaded from https://academic.oup.com/carcin/article-abstract/24/6/1021/2390450 by guest on 01 December 2017

Introduction Cancer is a disease initiated by the stepwise accumulation of genetic disorders affecting the balanced expression of oncogenes and suppressor genes, which leads to the promotion and progression of tumourigenesis. Until now, several cascades have been shown to be relevant to the cell cycle, replicative life span, apoptosis, proliferative signals and mobilization of resources in the development of cancers. However, we have never seen such a single molecule playing a key role in malignant tumourigenesis with the acquisition of properties of unlimited, self-sufficient growth and resistance to normal homeostatic regulatory mechanisms. We found that a molecule, erythropoietin (Epo), is substantially involved in the growth, viability and angiogenesis of malignant tumours. In erythropoiesis, Epo prevents apoptotic death of Epo-responsive erythroid precursor cells (1) and stimulates their proliferation and differentiation into erythrocytes (2±4). Binding Epo to its specific receptor (EpoR), which belongs to a family of cytokine receptors that have no tyrosine kinase domain, induces homodimerization of EpoR and the subsequent activation of a Janus kinase 2 (JAK2) through tyrosine phosphorylation, leading to activation of a transcriptional factor, STAT5 (signal transducer and activator of transcription), that induces mitosis of erythroid precursor cells (4,5). Moreover, during erythropoiesis, rigid regulation and control of EpoR is evident: the number of EpoR sites is 1000 per cell in colony forming uniterythroid and 200 per cell in erythroblasts, but there is no site in reticulocytes in human bone marrow (6); EpoR mRNA is stable with a mRNA half-life of only 90 min in human erythroid progenitor cells (7). The kidney and fetal liver produce Epo required for adult and fetal erythropoiesis, respectively. In addition to these organs, there is increasing evidence that many non-haematopoietic organs and tissues express Epo and EpoR; embryonic stem cells (8), the embryo proper including its developing nervous system (9), brain (10±13), uterus (14) and ovary (15,16) express Epo; and endothelial cells (17,18), embryonic stem cells (8), the embryo proper including its nervous system (9), decidual cells (9), uterus (14) and ovary (15,16) express EpoR. These widely dispersed Epo±EpoR production sites imply various physiological functions of Epo±EpoR signalling in our body in addition to erythropoiesis. In fact, Epo shows new physiological functions in the brain; neurons express EpoR (10,13) and astrocytes produce Epo (10,12). Epo prevents ischaemia-induced and toxic-stressed death of neurons in vivo (19,20), and apoptotic neuronal death in the developing brain (21). In the uterus, Epo plays an important role in angiogenesis via EpoR expressed in vascular endothelial cells of the uterine endometrium (14). Epo production in the kidney, liver and brain is hypoxia-inducible (11,22,23), but in the uterine endometrium, it depends on estradiol-17b (E2) (14). Furthermore, we have reported that the proliferation and survival of uterine and ovarian cancers depend on the 1021

Y.Yasuda et al.

Epo±EpoR signalling pathway; blockade of the Epo signal with a monoclonal antibody against Epo and soluble EpoR destroyed their xenografts in nude mice (24) and their tissue blocks in vitro (16) through death of both transformed cells themselves and intervening capillary endothelial cells. To extend our previous findings we studied whether the Epo± EpoR pathway operates in other malignancies. In this paper we report that 24 malignant human cell lines express Epo and EpoR mRNA regardless of their origins, types, genetic characteristics and biological properties, and respond to hypoxic stimuli by enhanced secretion of Epo individually. The Epo signalling pathway is detectable in xenografts of the cell lines, but not in the corresponding cell lines. Blockade of Epo signal by i.p. injections of EpoR antagonist demonstrated that Epo± EpoR signalling contributes substantially to the development of xenografts of the cell lines. Materials and methods Sampling clinical materials We obtained the patients' informed consent for their specimens to be used in the experiments. Immediately after resection of malignant tumours, we put resected tissues into liquid nitrogen or fixed them in Zamboni solution. Cell culture In all experiments, we used each cell line at 75% confluency. In normoxia, we cultured cell lines in 5% CO2 in air at 37 C, and in anoxia, in 5% CO2 , 10% H2 , and balanced N2 gases in an anaerobic box (ANX-1, Hirasawa Works, Tokyo, Japan), in which H2 gas reacts with O2 in the box to make water that is mediated by catalyst (AZX-250, Hirasawa Works). The anoxia was checked with a sheet of anaerobic indicator, oxoid (Oxid, Hante, UK), placed in the box. 5-Iodo-20 -deoxyuridine uptake assay We selected 12 cell lines to determine their proliferation by the colourimetric 5-iodo-20 -deoxyuridine (IdU) assay with the use of a DNA±IdU labelling and detection kit (MK420, Takara, Ohtsu, Japan) according to the manufacturer's protocols. Briefly, 4 days after culture, we added 100 ml of 20 mM IdU to each cell and incubated for 24 h. Each cell line was fixed with 0.1 N NaOH±ethanol, blocked with blocking solution, incubated with anti-IdU antibody at 37 C for 30 min, and then allowed to react with peroxidase-conjugated secondary antibody at 37 C for 30 min. The colour reaction was performed with tetramethyl benzidine. All processes were done in a dark room. Plates were read with a microplate reader (Bio-Rad, Hercules, CA, USA; Benchmark) at a wavelength of 450 nm. RT±PCR amplification We prepared total RNA of each sample with a total RNA isolation kit (Trizol, Gibco BRL, Carlsbad, CA, USA). We reverse-transcribed 1 mg of the total RNA with AMV reverse transcriptase and oligo (dT) primer followed by PCR using each pair of primers for Epo (sense 50 -CACCACTCTGCTTCGGGCTC30 , antisense 50 -ACCTGGAGAAGTCACAGCTT-30 ), EpoR (sense 50 -GAAGTAGTGCTCCTAGACGCC-30 , antisense 50 -CGGCTCCACTGCCTGCATCG-30 ) and b-actin (sense 50 -ATCCTCACCCTGAAGTACCC-30 , antisense 50 -ATTTCCCGCTCGGCCGTGGT-30 ) under annealing conditions of 65 C and 36 cycles for Epo and EpoR, 20 cycles for b-actin. Epo assay We determined the content of Epo in each culture medium containing 2  106 cells after 24 h incubation in triplicate with an enzyme-linked immunoassay using recombinant human Epo (gift from Snow Brand Milk Products, Tokyo, Japan) as the standard. This assay measures Epo levels as low as 1 pg/ml. We concentrated each culture medium with a concentrator (Vivaspin 20, Sartorius, Gaettingen, Germany) according to the protein content of each medium. We did not determine the medium of UT-7 because the medium itself contained 2.4 ng/ml of Epo (Figure 1F). Tumour formation and in vivo studies All animal experiments described in this paper were approved by the institution and performed under guidelines from Kinki University Animal Facilities. For transplantation, we injected 2±4  106 cells/0.1 ml of each corresponding medium subcutaneously into the interscapular region of each mouse (BALB/ cA Jcl-nu, CLEA, Tokyo, Japan) at 6±8 weeks of age. We measured the tumours 3 times a week. When the tumours were 6  6  5 mm in size, we

1022 Downloaded from https://academic.oup.com/carcin/article-abstract/24/6/1021/2390450 by guest on 01 December 2017

Fig. 1. Expression of Epo and EpoR in cell lines. (A±D) Immunoreactive Epo (A) and EpoR (C) in cytoplasm showing heterogenous expression among cells. Negative expression of Epo (B) and EpoR (D) in the cytoplasm stained with each antibody pre-absorbed with each antigen. Scale bar, 50 mm. (E) RT±PCR analysis of 24 malignant cell lines shows mRNAs for Epo, EpoR and b-actin. RT-, no reverse transcriptase in each RNA. (F) Content of Epo fg/ml (mean  SEM) in each number under normoxia (solid bars) and under anoxia (open bars). Anoxia was achieved in an anaerobic box (Hirasawa Works) as described in the Materials and methods. Epo content of the medium of UT-7 was not shown because the medium itself contained 2.4 ng/ml of Epo. Number in parentheses indicates the time of the highest of six determined values. Asterisk indicates significantly higher value under anoxia than normoxia by Student's t-test,  P 5 0.001,  P 5 0.01,  P 5 0.05. The number of each lane corresponds to that in Table I.

injected intraperitoneally 0.1 ml of erythropoietin mimetic peptide (EMP)9 or EMP1 dissolved in saline (Ohtsuka, Tokyo, Japan) or 0.1 ml of saline (Ohtsuka) four times at 1 h intervals/day for 3 days into the P39 and for 5 days into the SCH xenografts. One injection of 0.1 ml contained 1.0 or 0.5 mg/ml of EMP9, 0.06 or 0.03 mg/ml of EMP1 or saline. We used at least three mice for each treatment, i.e. 34 with xenografts and two with non-treated tumours. Seven days after the last injection, we killed the mice under deep anaesthesia and extirpated the tumours. After macroscopy, we fixed all tumours in Zamboni solution and processed them for immunohistochemical examinations and froze some tumours in liquid nitrogen. Immunostaining and histology We have described all the methods used for immunostaining of resected specimens (16), cell lines (9), cryosections of xenografts (24) and for staining specificity of each antibody (9). We used the following primary antibodies: anti-Epo (Genzyme, 1:500), anti-EpoR (1:250), anti-Ki67 (DAKO, 1:1), antiCD31 (Santa Cruz, 1:250) and anti-b-human chorionic gonadotropin (b-hCG) (ProGen, 1:10). To detect apoptosis, we used an in situ apoptosis detection kit (Intergen, Norcross, GA, USA). We analysed all sections stained with

Erythropoietin regulates tumour growth of human malignancies

anti-CD31 antibody and haematoxylin (Delafield, Merck, Darmstadt, Germany) for angiogenesis: We counted cross-sectioned and longitudinally sectioned capillaries and sprouts in 100 areas of 17.6  10 ÿ3 mm2 in serial sections separated by 60 mm. The magnification was 200 times for SCH, and 920 times for P39 tumours. To evaluate the proliferating and dying cells, we counted the number of Ki67-immunoreactive nuclei, and dying cells with pyknotic, chromatin-condensed and perforated nuclei in 100 fields of the same area we used to count vessels. Immunoprecipitation and western blotting We lysed tumours in buffer containing 60 mM Tris±HCl, 1 mM NaVO4 , 0.2% NP-40, 10% glycerol, pH 7.6 and a Protease Inhibitor Cocktail (Roche, Basel, Switzerland). After discarding the insoluble materials, we incubated 100 mg of protein lysates with polyclonal anti-STAT5 antibody (Santa Cruz, CA, USA) for 2 h. Then we collected the immunocomplexes with protein G± Sepharose (Amersham Biosciences, Piscataway, NJ, USA) for an additional hour and washed them three times in lysis buffer. For the membrane preparation, we homogenized tumours in a buffer containing 10 mM Tris±HCl (pH 7.6), 200 mM sucrose and a Protease Inhibitor Cocktail, and centrifuged it. After SDS±polyacrylamide gel electrophoresis and transfer onto Clear Blot membrane-P (Atto, Tokyo, Japan), we detected phospho-STAT5 in immunoprecipitates and EpoR in membrane preparations with a monoclonal anti-Tyr (P) antibody (Upstate, Lake Placid, NY) and polyclonal anti-EpoR antibody (Santa Cruz), respectively, and developed the blots with HRP-conjugated secondary antibodies and enhanced chemiluminescence (Amersham). Statistical analysis We used the Student's t-test to determine a significant difference in means of vascular density, and of uptake of IdU, and a c2 test to compare the incidence of cell populations.

Results Expression of Epo and EpoR in malignant human cell lines We examined cell lines established from patients' tumours and from commercial sources including cells with and without p53 or mitochondria and those resistant or sensitive to radiation (26) or chemotherapy. These are listed in Table I. First, we looked for Epo±EpoR pathways in these cell lines.

Surprisingly, all cell lines showed positive Epo and EpoR immunoreactivity in their cytoplasm (one example is shown in Figure 1A±D), indicating autocrine secretion of Epo. Then, we confirmed, with the use of RT±PCR analysis, the expression of Epo and EpoR mRNA in all cell lines including HepG2, which expresses Epo mRNA (27), and UT-7, which expresses EpoR mRNA (28). All cell lines expressed both transcripts differently (Figure 1E). Simultaneously, we examined their expression in normal human fibroblast cell lines, normal human fibroblast (NCF) and normal human dermal fibroblast (NHDF), and found positive expression. Then we examined primary cultures of mouse embryonic fibroblasts, Csa , Chinese hamster cell lines, CHO and V79-B310H, and confirmed negative expression (data not shown). As Epo is a secreted protein (3,4) with hypoxia stimulating its secretion (22,23), and as tumours are hypoxic (29), we measured with an enzyme-linked assay (30) Epo protein in each culture medium under normoxia and anoxia. Under normoxia, the Epo content of these cell lines ranged from 40.0 to 90.0 fg/ml except for that of SCH, which showed the highest value, 768.4  170.0 (mean  SEM) fg/ml (Figure 1F). Under anoxia, 18 of the 23 cell lines secreted significantly more than under normoxia whereas in the remaining five cell lines, A172, HMV1, 170, WiDr and MCF7, secretion was comparable (Figure 1F). The highest among the six values of most cell lines ranged from 80.0 to 170.0 fg/ml (Figure 1F). The rise in Epo level differed among the cell lines (Figure 1F); SCH, SBC3, DLD1, P22, C32TG and HepG2 significantly secreted at least twice as much as under normoxia. Furthermore, the Epo levels of G361, P39, 220, SBC3 and T47D cell lines reached their peaks within 2 h, those of P22 and AZ521 at 4 h and that of the remaining cell lines at 46 h (Figure 1F ). To compare the growth rate among these cell lines, we cultured 12 of the 24 cell

Table I. Origin and characteristics of cell lines Cell name 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

T98G A172 HMVI G361 P39 P22 C32TG 170 220 AZ521 SCH DLD-1 WiDr A549 SBC3 T47D MCF7 MIA PaCa Hela P0 9A PC-3 K562 HepG2 UT7

Tissues ÿ

Glioblastoma p53 Glioblastoma p53 ‡ Vagina, melanoma Melanoma radiation resistant Melanoma radiation sensitive Melanoma Amelanoic melanoma Eophageal ADC (CDDP resistant) Esophageal ADC (CDDP sensitive) Gastric cancer, epithelial type Stomach, choriocarcinoma Colon ADC type Colon ADC type Lung ADC Lung, small cell carcinoma Mammary cancer Breast carcinoma Pancreatic cancer Uterine ADC Mutant of Hela without mitochondria Prostate ADC Leukaemia Hepatoma Leukaemia

Mediuma used

Obtained from

1 1 2 2 2 2 3 2 2 3 4 4 3 3 3 2 3 3 2 20 5 4 1 6

ATCC (CRL1690) ATCC (CRL1620) Dr Hasumi, ref 25 ATCC (CRL1424) Ref. 26 Ref. 26 HSRRB (JCRB0227) Dr Miyaharab Dr Miyaharab HSRRB (JCRB0061) HSRRB (JCRB0251) HSRRB (JCRB9094) HSRRB (JCRB0224) HSRRB (JCRB0076) HSRRB (JCRB0067) Dainippon Pharm. Co. ATCC (HTB22) HSRRB (JCB0070) Dainippon Pharm. Co. Dr Banc HSRRB (JCRB9110) Dr Taketanid ATCC (HTB22) Ref. 29

a 1, Dulbecco MEM; 2, a-MEM; 3, MEM; 4, RPMI1640; 5, HamF12; 6, I's. MEM; 20 , a-MEM, 10% glucose and glutamine; each medium is supplemented with 10% FCS. b Dr Tokiharu Miyahara, Department of Surgery, Faculty of Medicine, Kyoto University, Japan. c Dr Sadayuki Ban, National Institute of Radiological Science, Chiba, Japan. d Dr Shigeru Taketani, Department of Biotechnology, Kyoto Institute of Technology, Japan.

1023 Downloaded from https://academic.oup.com/carcin/article-abstract/24/6/1021/2390450 by guest on 01 December 2017

Y.Yasuda et al.

lines in respective medium for 4 days and assayed the uptake of IdU. The uptake of the SCH was the highest among others (Figure 2). Taken together, cell lines from very malignant tumours appear to secrete very much Epo. These results indicate that the 24 malignant cell lines studied express Epo and EpoR mRNA, that they react to hypoxic stimuli with individual responsiveness, and that Epo functions in an autocrine manner in them. The reason that the normal cell lines NCF and NHDF expressed Epo and EpoR mRNA and the answer to whether there are any differences in expression between normal and malignant cell lines remain to be clarified. Identification of Epo-responsive sites in malignant tumours To determine whether Epo±EpoR signalling exists in malignant tumours that consist of tumour cells, capillary networks and other supporting cells we immunostained xenografts and their corresponding resected specimens with anti-EpoR antibody. We detected EpoR immunoreactivity in the tumour cells and endothelial cells in xenografts of SBC cell line and small cell carcinoma (Figure 3A and E), A549 cell line and lung adenocarcinoma (ADC) (Figure 3B and F) and PC-3 cell lines and prostate ADC (Figure 3C and G). No EpoR immunoreactivity was discernible in sections stained with the supernatant of a mixture of soluble EpoR and EpoR±antibody (Figure 3D and H). Next, we analysed the lysates of these specimens and some other resected specimens by western blotting. All lysates showed the band for EpoR as the positive control of UT-7 (28) (Figure 3I). These findings indicate that most malignant tumours have Epo-responsive sites, which are the transformed cells themselves and the intervening vascular endothelial cells. Signal transduction of Epo in cell lines and xenografts The question then arises: are these cell lines transplantable into nude mice, and will the resulting tumours respond to the withdrawal of Epo signalling by their destruction, as seen in the xenografts of ovarian and uterine cancers (24)? We injected subcutaneously HMV1, G361, P39, 220, SCH, DLD1, A549, SBC3, Hela, PC-3 and HepG2 cell lines into nude mice and confirmed their transplantability: tumour swelling within 1 week for HepG2, up to 8 weeks for G361.

We have detected previously the signal transduction of Epo by deprivation of Epo signal in blocks of uterine and ovarian ADC in vitro by injections of antibody to Epo or of soluble EpoR receptor (16). Then we determined the amount of phosphotyrosine (P-Tyr) in G361, P39, A549, SBC3 and HepG2 cell lines, and compared it with that in the corresponding xenografts. We recognized the strong band for P-Tyr in five xenografts, but in their cell lines the band was weak or not detectable (Figure 4). These findings suggest that although signal transduction of Epo operates in these cell lines, its expression is too weak to detect, but in the xenografts it is strong enough to detect, as noted by others (31). Withdrawal of Epo signalling in xenografts To deprive Epo signalling in xenografts, we first treated tumours of G361, P39, 220, SCH, DLD1, A549, SBC3, Hela, PC-3 and HepG2 with local intra-tumour injections of antibody to Epo or of an antagonist to EpoR (EMP9) (32) and found reduced vascularity and focal degradation in tumour masses 7 days after treatment (data not shown). Then we worked intensively with tumours of P39 and SCH cell lines in at least three mice per experiment with i.p. injections of EMP9 and EMP1 (32). As in our preliminary in vitro experiments, exposure to 0.5 mg/ml of EMP9 four times at 1 h intervals for 24 h caused growth reduction of the P39 and the SCH cell lines by 64.5 and 10.4% of each control level, respectively, we adopted different protocols; those with P39 tumours received four injections per day for 3 consecutive days, and those with the SCH cell line received four injections per day for 5 days. Seven days after the last injection, we extirpated the tumours under deep anaesthesia and simultaneously examined the interior of the mouse body. No mice showed any bleeding in the body cavity or haematoceles in the organs. We resected blocks of lung, liver, kidney, intestine, spleen and mesentery of EMP9-exposed and control mice, and processed them for EPON sections for microscopic examination. No abnormalities of the vessels or bleeding foci were detectable. The haematocrit values were 46.5  0.8, 49.8  1.2, 51.5  0.5, 48.8  0.3, 53.5  0.5, 50.1  0.5 and 48.2  0.5 (mean  SEM, n ˆ 10) in tumour-bearing mice exposed to

Fig. 2. Proliferating activity of cell lines in vitro. Twelve cell lines were cultured for 4 days and then analyzed by IdU uptake assay as described in the Materials and methods. Values indicate means  SEM. Asterisk indicates significant difference from the values of P39 cell line by Student's t-test,  P 5 0.001,  P 5 0.01. Columns with the same letter show significant differences (a, P 5 0.001; b, P 5 0.01).

1024 Downloaded from https://academic.oup.com/carcin/article-abstract/24/6/1021/2390450 by guest on 01 December 2017

Erythropoietin regulates tumour growth of human malignancies

Fig. 4. Tyrosine phosphorylation of STAT5 in cell lines and their corresponding xenografts. Lysates of cell lines and xenografts were immunoprecipitated with anti-STAT5. Western blotting with anti-P-STAT5 shows band in xenografts, but not in its cell lines.

Fig. 3. Expression of EpoR in xenografts and resected tumours. (A±D) Section of xenografts, SBC3 (A), A549 (B) and PC-3 (C) cells, and malignant tumours, small cell carcinoma (E), lung ADC (F) and prostate cancer (G) stained with anti-EpoR antibody show an EpoR-positive tumour (arrowheads) and endothelial cells (arrows). Negative expression in sections of PC-3 tumours (D) and lung ADC (H) stained with pre-absorbed EpoR antibody. Scale bar, 50 mm. (I) Western blotting of EpoR protein in xenografts and resected tumours. L-SCC, lung small cell carcinoma; L-ADC1, 2, two samples from lung ADC; P-ADC, prostate adenocarcinoma; S-ADC1, 2, two samples from stomach adenocarcinoma; UT-7, positive control for EpoR.

10 and 20 mg/ml of EMP9, 0.6 and 1.2 mg/ml of EMP1, saline and no injection, and no transplantation, respectively. These results indicate that in host animals EMP9 did not damage the vascular channels or red blood cell formation over the limited time scale of the experiment. With a dissecting microscope, we examined all tumours macroscopically. In two of the three P39 tumours exposed to 6 mg/ml of EMP9 there were balloon-like swellings with vacant holes inside containing small amounts of transparent liquid enclosed by thin tissue beneath the capsule (Figure 5A). The other EMP9-exposed tumours showed small or large pulverized tumour masses. No such drastic alterations were detectable in the controls (Figure 5B and C) or in non-treated

tumours (Figure 5D). When cutting SCH tumours we detected fluid leaking from the capsules of almost all tumours; markedly bloody in 50% of EMP9- and moderately bloody in 50% of EMP1-exposed tumours, and transparent in saline- and nontreated tumours. The EMP9-exposed tumours had many holes and large defects of dark-red tissue leaving thread-like structures (Figure 5E), whereas the control and non-treated tumours showed no tissue defects but had many capillaries in pink tissues (Figure 5F±H). The histopathological features of nontreated P39 and SCH tumours resembled those of the original tumour, containing EpoR immunoreactive transformed cells and capillary endothelium (Figure 5K and O) and many endothelial cells (Figure 5L and P); whereas those of treated P39 and SCH tumours showed destruction of the tissues with many dead cells (Figure 5I and M) and few capillaries (Figure 5J and N). Microscopic examination of all tumours revealed the causes of the drastic macroscopic changes seen in EMP9-exposed tumours. In P39 tumours, pulverization appeared as many foci of loosely arranged tissue, which contained many dead tumour cells with melanin granules in clusters or dispersed (Figure 6A) and few intact capillaries (Figure 6B) resulting in the loss of delicate connections between tumour cells and intervening capillary networks. In these regions there were many apoptotic cells (Figure 6C). EMP1-exposed tumours differed from EMP9-exposed tumours with their numerous compactly arranged cells, EpoR-immunoreactivity (Figure 6D) adjacent capillaries (Figure 6E) and few apoptotic cells (Figure 6F). SCH tumours showed a loss of epithelial structures under which many dying cells (Figure 6G) and fragmented endothelial cells (Figure 6H) were discernible. The dying cells were apoptotic (Figure 6I). These findings indicate the accumulation of bloody fluid in the capsule. In contrast, the EMP1-exposed tumours showed a well-developed epithelial structure with small cells (Figure 6J) and many vascular channels (Figure 6K); apoptosis was sporadic (Figure 6L). Moreover, the EpoR-immunoreactive SCH tumour cells also reacted to anti-b-hCG antibody (data not shown). To evaluate angiogenesis and the growth and death of tumours quantitatively, we counted the vessels in sections stained with anti-CD31 antibody and the cells with positive Ki67 nuclei in sections stained with anti-Ki67 antibody in all tumours. The total number of cells counted was over 13 000± 4000 per sample in P39 and SCH tumours; the results are summarized in Figure 6M±P. Both doses of EMP9 significantly suppressed angiogenesis (P 5 0.001; Figure 6M and 1025

Downloaded from https://academic.oup.com/carcin/article-abstract/24/6/1021/2390450 by guest on 01 December 2017

Y.Yasuda et al.

Fig. 5. Macroscopic section of xenografts of P39 and SCH cell line. (A±D) from P39 tumours, (E±H) from SCH tumour. (B±H) longitudinal half-cut and (A) additionally transverse-sectioned. (I±P) histopathology of treated (I,J,M,N) and non-treated tumours (K,L,O,P). (A) EMP9 (6 mg) injected tumour has a huge hole, leaving thin tissue and few capillaries. (B) EMP1 (0.4 mg) injected tumour shows compact tumour masses with many capillaries. (C) Saline injected tumour. (D) Tumour 5 weeks after transplantation shows brown- and light brown-pigmented mass with many capillaries. (E) EMP9 (10 mg) injected tumour shows many tissue defects (asterisk). (F) EMP1 (0.6 mg) injected tumour shows smooth surface with many capillaries. (G) Saline treated tumour. (H) Tumour 4 weeks after transplantation shows a pink mass with capillaries. (A±H) Scale, 1 mm. (I) and (J) Section of (A) shows degenerating EpoR-positive tumour cells (arrowheads) (I), and fragmented endothelial cells (arrows) (J). (K and L) Section of (D) shows EpoR-positive tumour cells (arrowheads) with pigment (K) alongside slender endothelial cells (arrows) (L). (M and N) Section of (E) shows degenerating tumour cells (arrowheads) (M) and endothelial cells (arrows) (N). (O and P) Section of (H) shows many capillaries and extending sprouts (arrows) (P). Scale bar, 25 mm.

O), reduced the proliferation of cells and increased the number of dying cells (P 5 0.001 versus controls; Figure 6N and P), whereas EMP1 stimulated angiogenesis and cell proliferation (P 5 0.001 or P 5 0.01; Figure 6M±P). Moreover, western blot analysis of lysate of P39 tumours showed very weak tyrosine phosphorylation of STAT5 immunoprecipitates in EMP9-exposed tumours and strong bands in EMP1-exposed tumours (Figure 6Q). These findings indicate that EMP9 induces death of tumour cells and proliferating capillary endothelial cells through disturbances in JAK2±STAT5 pathways, while EMP1 stimulates the growth of both. Discussion Here we provide evidence for a novel role of Epo; Epo contributes to the growth, viability and angiogenesis of almost all, 1026 Downloaded from https://academic.oup.com/carcin/article-abstract/24/6/1021/2390450 by guest on 01 December 2017

if not all, malignant tumours. Our results imply that most malignant tumours have Epo-responsive sites on transformed cells themselves and on capillary endothelial cells, so that Epo plays as substantial a role in tumourigenesis as it does in erythroid progenitor cells; Epo protects these cells from apoptosis and stimulates them to proliferate, but these cells have no regulation of the number of EpoR sites as in erythroid differentiation (6), so that Epo appears to cause unlimited proliferation. Although the autocrine Epo level is very much lower than the physiological Epo level (5±25 pg/ml) (2), it seems to be sufficient to maintain the microenvironment of the tumour mass, and the greater the tumour, the more Epo is secreted in response to hypoxic stimuli in the tumour mass leading to stepwise acquisition of self-sufficient growth independent of the normal environment of surrounding tissues.

Erythropoietin regulates tumour growth of human malignancies

Fig. 6. Destruction in xenografts of P39 and SCH cell lines. (A±F) are microscopic sections of P39, and (G±L) of SCH tumour. (A±C) are from tumour shown in Figure 5(A); (D±F) shown in Figure 5(B); (G±I) shown in Figure 5(E); (J±L) shown in Figure 5(F). Panels of first transverse lane are sections stained with anti-EpoR antibody; second lane, with CD31 antibody; third lane, demonstrated by terminal deoxy nucleotidyl transferase assay. Arrow points to capillary; asterisk, interior of vacant cavity. (A) Area pointed to in Figure 5(A). Note many dying tumour cells with aggregation of melanin (double arrows) and vacant spaces. (B) Adjacent to (A) showing few capillaries. (C) Apoptotic cells (arrowheads) and bodies. (D) Area pointed to in Figure 5B. Note densely packed cells and capillary with positive EpoR immunoreactivity. (E) Adjacent to (D) showing many capillaries alongside tumour cells. (F) Scattering of apoptotic bodies (arrowheads). (G) Area marked in Figure 5E. Note many dying cells and few cells with EpoR reactivity. (H) Adjacent to (G) showing fragmented capillaries (arrowheads). (I) Note many apoptotic cells (arrowheads). (J) Area marked in Figure 5F. Note ADC type structure. (K) Adjacent to (J) showing capillaries, sprouts and sinusoid-like lining (arrowheads). (L) Note a few apoptotic cells. Scale bar, 50 mm. (M and O) EMP9-reduced and EMP1-induced angiogenesis in P39 (M) and in SCH tumours (O). Each column expresses the number of vessels as mean  SEM of 100 fields of sections as described in the Materials and methods. Significantly different from the values of controls (none or saline).  P 5 0.001;  P 5 0.01. (N and P) EMP9 suppressed growth and increased death of cells in P39 (N) and SCH tumours (P). Each column shows ratio of cells alive, including proliferation (pro) and dead. Significantly different from the values of controls (none or saline) ( P 5 0.001,  P 5 0.01). (Q) Tyrosine phosphorylation of STAT5 in P39 tumours. Lysates of tumours were immunoprecipitated with anti-STAT5. Western blotting with anti-P-Tyr shows no band in lysate of P39-tumour exposed to EMP9, but a strong band in that exposed to EMP1.

1027 Downloaded from https://academic.oup.com/carcin/article-abstract/24/6/1021/2390450 by guest on 01 December 2017

Y.Yasuda et al.

During the past 30 years it has become clear that malignant haemopoietic organs produce Epo, and non-haemopoietic malignant organs such as cerebellar haemangioblastoma (32), uterine and ovarian cancers (16) also produce Epo. Furthermore, in renal ADC, the Epo production site is in the tubular cells (33), not in the peritubular interstitial cells that are the site of Epo synthesis in normal kidney (34). However, nobody had noted that Epo is involved in the development of kidney cancer until Westenfelder and Baranowski (36) noted this possibility, although it was known to be a cause of polycythaemia associated with autocrine Epo in tumours. We have reported that both normal and malignant ovary and uterus produce Epo (15,16), and that in normal tissues the Epo-responsive sites are vascular endothelial cells and cells with cyclic changes, such as follicular cells in the ovary and decidual cells in the uterine endometrium (15), whereas the sites in malignancies are intervening capillary endothelial cells and transformed cells of both organs (16). The low levels of Epo in normal kidney (37), ovary and uterus (16) and the high levels in their malignancies (16,37) suggest that Epo is involved in malignant tumourigenesis. Moreover, there is accumulating evidence that non-haemopoietic organs and tissues have transcripts for Epo and EpoR in the human body. In our present studies normal fibroblast cell lines also expressed the signal. Taken together, it is reasonable to consider that Epo plays physiological roles other than erythropoiesis in haemopoietic and non-haemopoietic organs and that in malignancies of these organs, an unknown mechanism appears to lead to the activation of transcription of Epo and EpoR. Thus, any factor that induces disorder in Epo signalling in target organs or tissues appears to cause their malignant transformation. However, the details of the nature of such factors remain to be clarified. Angiogenesis is essential for the development of malignant tumours: some angiogenesis inhibitors and inhibition of the Raf-1 gene expressed on capillary endothelium (38) suppress the proliferation of endothelial cells and lead to the regression of established tumours. On the contrary, Epo±EpoR signalling acts as a mitogen in the endothelium where EpoR mRNA is present (17,18) through tyrosine phosphorylation of proteins including phosphorylation of STAT5 (39). Previously, we demonstrated that the expression of Epo and vascular endothelial growth factor (VEGF) mRNA depends on the estrogen level in the uterine endometrium of mice, and that without Epo cyclic endometrial growth is reduced due to the inhibition of angiogenesis (14). In the present study, EMP9 reduced angiogenesis in SCH and P39 tumours, but did not damage the vessels in the host, indicating that EMP9 damages only uncontrollably proliferating capillaries, not static vessels. On the contrary, EMP1 stimulated angiogenesis through tyrosine phosphorylation of STAT5. A recent report of blockade of Raf-1 gene in the endothelium revealed inhibition of angiogenesis in xenografts of a melanoma cell line (38). Former reports described a relationship between Raf-1 and Epo signalling; Epo activates Raf-1 and activated Raf-1 is necessary for Epo and interleukin 3 growth signalling pathways (40); in a UT-7 cell line Epo and Raf-1 activate mitogen-activated protein (MAP)-kinase (41). These cascades described in the cell lines are not equally applicable to organized malignant tumours; however, Epo-induced angiogenesis in malignant tumours appears to be activated through the Raf-1±MAP kinase pathway. Under hypoxic stimulation, hypoxia-inducible factor (HIF)1a activates transcription of Epo (22). Most human primary 1028 Downloaded from https://academic.oup.com/carcin/article-abstract/24/6/1021/2390450 by guest on 01 December 2017

and metastatic cancers express abundant HIF-1a, but benign tumours and normal tissues do not always do so (42). Furthermore, studies of xenografts of ES cells with a loss of HIF-1a reveal attenuation of tumour growth and inhibition of angiogenesis (43,44); however, xenografts of ES cells with HIF1a ‡= ‡ show tumour growth with abundant vascularity (43). We have reported previously that expression of Epo and VEGF mRNA is dependent on the level of E2, not on hypoxia in the uterine endometrium (14). As ES cells express mRNAs for Epo and EpoR (8), as blastocysts express estrogen receptor mRNA (45) and as E2 has an angiogenic capability (46), Epo induced by E2 may operate in these embryoid bodies, and simultaneous induction of VEGF and E2 from host blood may co-operate in leading to their growth. Taken together, activation of Epo involved in tumourigenesis appears to occur through HIF-1a and E2. In conclusion, signal transduction of Epo takes part in both growth and angiogenesis of tumours and is a putative molecule contributing substantially to the development of almost all malignancies. Acknowledgements The authors thank Dr Alice S.Cary for reading and correcting the manuscripts, Dr Makiko Sakai, Hidetoshi Higashiguchi and Takashi Morimoto for technical help and Keiko Yamashita for preparing the paper.

References 1. Koury,M.J. and Bondurant,M.C. (1990) Erythropoietin retards DNA breakdown and prevents programmed death in erythroid progenitor cells. Science, 248, 378±381. 2. Kranz,S.B. (1991) Erythropoietin. Blood, 77, 419±434. 3. Jelkmann,W. (1992) Erythropoietin: structure, control of production, and function. Physiol. Rev., 72, 449±489. 4. Youssoufian,H., Longmore,G., Neumann,D., Yoshimura,A. and Lodish,H.F. (1993) Structure, function, and activation of the erythropoietin receptor. Blood, 81, 2223±2236. 5. Ihle,J.N. (1995) Cytokine receptor signalling. Nature, 377, 591±594. 6. Broudy,V.C., Lin,N., Brice,M., Nakamoto,B. and Papayannopoulou,T. (1991) Erythropoietin receptor characteristics on primary human erythroid cells. Blood, 77, 2583±2590. 7. Wickrema,A., Krantz,S.B., Winkelmann,J.C. and Bondurant,M.C. (1992) Differentiation and erythropoietin receptor gene expression in human erythroid progenitor cells. Blood, 80, 1940±1949. 8. Schmit,R.M., Bruyns,E. and Snodgrass,H. (1991) Haematopoietic development of embryonic stem cells in vitro; cytokine and receptor gene expression. Genes Dev., 5, 728±740. 9. Yasuda,Y., Nagao,M., Okano,M., Masuda,S., Sasaki,R., Konishi,H. and Tanimura,T. (1993) Localization of erythropoietin and erythropoietinreceptor in postimplantation mouse embryos. Dev. Growth Differ., 35, 711±722. 10. Masuda,S., Nagao,M., Takahata, K., Konishi,Y., Gallyas,F. Jr., Tabira,T. and Sasaki,R. (1993) Functional erythropoietin receptor of the cells with neural characteristics: comparison with receptor properties of erythroid cells. J. Biol. Chem., 268, 11208±11216. 11. Masuda,S., Okano,M., Yamagishi,K., Nagao,M., Ueda,M. and Sasaki,R. (1994) A novel site of erythropoietin production: oxygen-dependent production in cultured rat astrocytes. J. Biol. Chem., 269, 19488±19493. 12. Marti,H.H., Wenger,R.H., Rivas,L.A., Straumann,U., Digicaylioglu,M., Henn,V., Yonekawa,Y., Bauer,C. and Gassmann,M. (1996) Erythropoietin gene expression in human, monkey and murine brain. Eur. J. Neurosci., 8, 666±676. 13. Morishita,E., Masuda,S., Nagao,M., Yasuda,Y. and Sasaki,R. (1997) Erythropoietin receptor is expressed in rat hippocampal and cerebral cortical neurons, and erythropoietin prevents in vitro glutamate-induced neuronal death. Neuroscience, 76, 105±116. 14. Yasuda,Y., Masuda,S., Chikuma,M., Inoue,K., Nagao,M. and Sasaki,R. (1998) Estrogen-dependent production of erythropoietin in uterus and its implication in uterine angiogenesis. J. Biol. Chem., 273, 25381±25387.

Erythropoietin regulates tumour growth of human malignancies

15. Yasuda,Y., Fujita,Y., Musha,T., Tanaka,H., Shiokawa,S., Nakamatsu,K., Mori,S., Matsuo,T. and Nakamura,Y. (2001) Expression of erythropoietin in human female reproductive organs. Mol. Cell Dev., 106, 215±222. 16. Yasuda,Y., Fujita,Y., Masuda,S., Musha,T., Ueda,K., Tanaka,H., Fujita,H., Matsuo,T., Nagao,M., Sasaki,R. and Nakamura,Y. (2002) Erythropoietin is involved in growth and angiogenesis in malignant tumours of female reproductive organs. Carcinogenesis, 23, 1797±1805. 17. Anagnostou,A., Liu,Z., Steiner,M., Chin,K., Lee,E.S., Kessimian,N. and Noguchi,C.T. (1994) Erythropoietin receptor mRNA expression in human endothelial cells. Proc. Natl Acad. Sci. USA, 91, 3974±3978. 18. Yamaji,R., Okada,T., Moriya,M., Naito,M., Tsuruo,T., Miyatake,K. and Nakano,Y. (1996) Brain capillary endothelial cells express two forms of erythropoietin receptor mRNA. Eur. J. Biochem., 239, 494±500. 19. Sakanaka,M., Wen,T.-C., Matsuda,S., Masuda,S., Morishita,E., Nagao,M. and Sasaki,R. (1998) In vivo evidence that erythropoietin protects neurons from ischemic damage. Proc. Natl Acad. Sci. USA, 95, 4635±4640. 20. Sadamoto,Y., Igase,K., Sakanaka,M., Sato,K., Otsuka,H., Sakaki,S., Masuda,S. and Sasaki,R. (1998) Erythropoietin prevents place navigation disability and cortical infarction in rats with permanent occlusion of the middle cerebral artery. Biochem. Biophys. Res. Commun., 253, 26±32. 21. Yu,X., Shacka,J.J., Eells,J.B. et al. (2002) Erythropoietin receptor signalling is required for normal brain development. Development, 129, 505±516. 22. Smenza,G.L. and Wang,G.U. (1992) A nuclear factor induced by hypoxia via de novo protein synthesis binds to the human erythropoietin gene enhancer at a site required for transcriptional activation. Mol. Cell Biol., 12, 5447±5454. 23. Bunn,H.F. and Poyton,R.O. (1996) Oxygen sensing and molecular adaptation to hypoxia. Physiol. Rev., 76, 839±885. 24. Yasuda,Y., Musha,T., Tanaka,H. et al. (2001) Inhibition of erythropoietin signalling destroys xenografts of ovarian and uterine cancers in nude mice. Br. J. Cancer, 84, 836±843. 25. Hasumi,K., Sakamoto,G., Sugano,H., Kasuga,T. and Masubuchi,K. (1978) Primary malignant melanoma of the vagina: study of four autopsy cases ultrastructural findings. Cancer, 42, 2675±2686. 26. Utsumi,H. and Elkind,M.M. (1993) Human melanoma cells: relationship between their radiosensitivity and the repair of potentially lethal damage. Radiat. Oncol. Invest., 1, 29±33. 27. Goldberg,M.A., Allison,G.C., Cunningham,J.M. and Bunn,F. (1983) The regulated expression of erythropoietin by two hepatoma cell lines. Proc. Natl Acad. Sci. USA, 80, 6269±6293. 28. Komatsu,N. and Fujita,H. (1993) Induced megakaryocytic maturation of the human leukemic cell line UT-7 results in down-modulation of erythropoietic receptor gene expression. Cancer Res., 53, 1156±1161. 29. Hockel,M., Vorndran,B., Schlenger,K., Bauûmann,E. and Knapstein,P.G. (1993) Tumour oxygenation: a new predictive parameter in locally advanced cancer of the uterine cervix. Gynecol. Oncol., 51, 141±149. 30. Goto,M., Murakami,A., Akai,K., Kawanishi,G., Ueda,M., Chiba,H. and Sasaki,R. (1989) Characterization and use of monoclonal antibodies directed against human erythropoietin that recognize different antigenic determinants. Blood, 74, 1415±1423. 31. Buettner,R., Mora,L.B. and Jove,K. (2002) Activated STAT signaling in human tumours provides novel molecular targets for therapeutic intervention. Clin. Cancer Res., 8, 945±954.

32. Johnson,D.L., Farrell,F.X., Barbone,F.P. et al. (1998) Identification of a 13 amid acid peptide mimetic of erythropoietin and description of amino acids critical for the mimetic activity of EMP1. Biochemistry, 37, 3699±3710. 33. Trimble,M., Caro,J., Talalla,A. and Brain,M. (1991) Secondary erythrocytosis due to a cerebellar hemangioblastoma: demonstration of erythropoietin mRNA in the tumor. Blood, 78, 599±601. 34. Da Silva,J.-L., Lacombe,C., Bruneval,N., Casadavall,N., Leporrier,M., Camilleri,J-P., Bariety,J., Tambourin,P. and Varet,B. (1990) Tumor cells are the site of erythropoietin synthesis in human renal cancers associated with polycythemia. Blood, 75, 577±582. 35. Koury,S.T., Bondurant,M.C. and Koury,M.J. (1988) Localization of erythropoietin synthesizing cells in murine kidneys by in situ hybridization. Blood, 71, 524±527. 36. Westenfelder,C. and Baranowski,R.L. (2000) Erythropoietin stimulates proliferation of human renal carcinoma cells. Kidney Int., 58, 647±657. 37. Katsuoka,Y., McGonigle,R., Rege,A.B., Beckman,B. and Fisher,J.W. (1983) Erythropoietin production in human renal carcinoma cells passaged in nude mice and in tissue culture. Gann, 74, 534±541. 38. Hood,J.D., Bednarski,M., Frausto,R., Guccione,S., Reisfeld,R.A., Xiang,R. and Cheresh,D.A. (2002) Tumor regression by targeted gene delivery to the neovasculature. Science, 296, 2404±2407. 39. Haller,H., Christel,C., Dannenberg,L., Thiele,P., Lindschau,C. and Luft,F.C. (1996) Signal transduction of erythropoietin in endothelial cells. Kidney Int., 50, 481±488. 40. Carroll,M.P., Spivak,J.L., McMahon,M., Weich,N., Rapp,U.R. and May,W.S. (1991) Erythropoietin induces Raf-1 activation and Raf-1 is required for erythropoietin-mediated proliferation. J. Biol. Chem., 266, 14964±14969. 41. Gobert,S., Duprez,V., Lacombe,C., Gisselbrecht,S. and Mayeux,P. (1995) The signal transduction pathway of erythropoietin involves three forms of mitogen-activated protein (MAP) kinase in UT7 erythroleukemia cells. Eur. J. Biochem., 234, 75±83. 42. Zhong,H., De Marzo,A.M., Laughner,E., Lim,M., Hilton,D.A., Zagzag,D., Buechler,P., Isaacs,W.B., Semenza,G.L. and Simons,J.W. (1999) Overexpression of hypoxia-inducible factor 1a in common human cancers and their metastases. Cancer Res., 59, 5830±5835. 43. Ryan,H.E., Lo,J. and Johnson,R.S. (1998) HIF-1 is required for solid tumour formation and embryonic vascularization. EMBO J., 17, 3005± 3015. 44. Carmeliet,P., Dor,Y., Herbert,J.-M. et al. (1998) Role of HIF-1a in hypoxia-mediated apoptosis, cell proliferation and tumour angiogenesis. Nature, 394, 486±490. 45. Hou,Q. and Gorski,J. (1993) Estrogen receptor and progesterone receptor genes are expressed differentially in mouse embryos during preimplantation development. Proc. Natl Acad. Sci. USA, 90, 9460±9464. 46. Losordo,D.W. and Isner,J.M. (2001) Estrogen and angiogenesis: a review. Arterioscler. Thromb. Vasc. Biol., 21, 6±12.

Received October 30, 2002; revised January 31, 2003; accepted April 4, 2003

1029 Downloaded from https://academic.oup.com/carcin/article-abstract/24/6/1021/2390450 by guest on 01 December 2017