Adrenomedullin promotes formation of xenografted endometrial ...

Oncogene (2002) 21, 2815 ± 2821 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc

Adrenomedullin promotes formation of xenografted endometrial tumors by stimulation of autocrine growth and angiogenesis Martin K Oehler1,2, Stephen Hague1,2, Margaret CP Rees2 and Roy Bicknell*,1 1

Molecular Angiogenesis Laboratory, Cancer Research UK, Institute of Molecular Medicine, John Radclie Hospital, University of Oxford, Oxford OX3 9DS, UK; 2Nueld Department of Obstetrics and Gynaecology, John Radclie Hospital, University of Oxford, Oxford OX3 9DS, UK

The angiogenic peptide adrenomedullin (ADM) has been implicated as a mediator of the increased risk of endometrial hyperplasia and cancer resulting from the use of tamoxifen for the treatment and prevention of breast cancer. ADM has been shown to be induced by tamoxifen in the endometrium and to be a growth factor for endometrial endothelial cells in vitro. We have now shown ADM to be strongly angiogenic in the mouse subcutaneous sponge angiogenesis assay. To examine the role of ADM in tumor growth, the ADM cDNA was transfected into endometrial carcinoma cells followed by xenografting into athymic mice. Two endometrial cancer cell lines were employed, those in which transfection and expression of ADM resulted in no eect on growth in vitro (Ishikawa cells) and those in which expression of exogenous ADM stimulated in vitro growth (RL95.2 cells). A clear enhancement of tumor growth was seen with both cell lines but the eect was far greater with the RL95.2 cells. We conclude that ADM is pro-tumorigenic by stimulating either angiogenesis alone or by stimulating angiogenesis and carcinoma cell growth directly. The combined activities lead to a striking increase in tumor growth. These results provide the ®rst direct evidence of tumorigenic activity of ADM and provide further support for ADMs involvement in tamoxifen induced endometrial neoplasia. Oncogene (2002) 21, 2815 ± 2821. DOI: 10.1038/sj/ onc/1205374 Keywords: adrenomedullin; angiogenesis; tumorigenesis; tamoxifen; endometrial cancer

Introduction Adrenomedullin (ADM) is a 52-amino-acid peptide which belongs to the calcitonin gene peptide superfamily based on its slight homology with calcitonin gene-related peptide (CGRP) and amylin (Wimalawansa, 1997). It acts through the G protein-coupled receptor calcitonin receptor-like receptor (CRLR), with

*Correspondence: R Bicknell; E-mail: [email protected] Received 31 July 2001; revised 22 January 2002; accepted 31 January 2002

speci®city for ADM being conferred by the receptor associated modifying protein 2 (RAMP2) (McLatchie et al., 1998; Kamitani et al., 1999). Since its identi®cation (Kitamura et al., 1993) ADM has been shown to be a pluripotent peptide with properties ranging from inducing vasorelaxation to acting as a regulator of cellular growth (Hinson et al., 2000). ADMs special role in vascular physiology was demonstrated by an ADM gene knockout mouse which showed extreme hydrops fetalis and cardiovascular abnormalities (Caron and Smithies, 2001). We have previously shown that ADM is an endothelial cell growth factor and angiogenic in the chick chorioallantoic membrane (CAM) assay (Zhao et al., 1998). ADM is known to be a hypoxia regulated peptide (Cormier-Regard et al., 1998; Nakayama et al., 1998) which is under hypoxia-inducible transcription factor-1 (HIF-1) control similar to that of other angiogenic factors like vascular endothelial growth factor (VEGF) (Garayoa et al., 2000; Semenza, 1999). We have shown that ADM when upregulated under hypoxia acts as an antiapoptotic factor antagonizing hypoxia induced cell death of endometrial cancer cells by up-regulation of Bcl-2 (Oehler et al., 2001). ADM is expressed in a variety of malignant tissues (Lal et al., 1999; Hata et al., 2000; Rocchi et al., 2001) and was shown to be mitogenic for human cancer cell lines including lung, breast and colon lineages in vitro (Rocchi et al., 2001; Miller et al., 1996; Pio et al., 2000). Its role in tumorigenesis, however, is yet to be established. Tamoxifen is currently the drug of choice for the treatment of hormone dependent breast cancer and is being evaluated as a preventative agent in women at high risk of developing the disease (Cuzick, 2000; Dalton and Kallab, 2001). A concern, however, is the proliferative eect of TAM on the endometrium, which is associated with a twofold to sevenfold increased risk of endometrial cancer for long-term tamoxifen users (White, 1999; Bergman et al., 2000). In an attempt to understand why tamoxifen has the aforementioned eect on the endometrium we initiated a study to identify tamoxifen induced gene expression in endometrial isolates. PCR dierential display was used to identify tamoxifen-induced genes. One such gene was ADM for which we then showed increased expression in the endometrium of patients receiving

Adrenomedullin is tumorigenic MK Oehler et al

2816

tamoxifen (Zhao et al., 1998). We subsequently showed that ADM is an autocrine regulator of endothelial cell growth in the endometrium, being both produced by endometrial endothelial cells and stimulating their growth (Nikitenko et al., 2000). Although these observations implicate an involvement of ADM in tamoxifen induced proliferative changes of the endometrium, evidence for a direct tumorigenic activity of ADM is still missing. We here report for the ®rst time that ADM is angiogenic in the mouse subcutaneous sponge angiogenesis assay and that constitutive expression of ADM enhances tumor take and endometrial carcinoma cell growth in vivo. These results provide direct evidence for the tumorigenic activity of ADM and supports ADM as a mediator of tamoxifen induced endometrial hyperand neoplasia.

Results Angiogenic activity of ADM in the mouse subcutaneous sponge assay We have previously shown that ADM is angiogenic in the CAM, an avian in vivo assay. Here we examined the eect of ADM in a mammalian in vivo assay. ADM was shown to be strongly angiogenic in the mouse subcutaneous sponge angiogenesis assay (P50.05 (n=4, mean+s.e.m.) Wilcoxon matched pairs test) (Figure 1). Characterization of ADM expressing Ishikawa and RL95.2 clones The eect of ADM on endometrial cancer cell growth was examined by experiments with the endometrial carcinoma cell lines Ishikawa and RL95.2. These cell lines were chosen to represent major types of endometrial carcinoma cells. Ishikawa cells are estrogen receptor (ER) positive cells whereas late passage RL95.2 cells do not express the ER (Sundareshan and Hendrix, 1992; Zhang et al., 1995a; Bilimoria et al., 1996). The RL95.2 cells used in this study were of passage 4100. ADM under control of the CMV promoter in the mammalian expression vector pcDNAI/Neo was transfected into Ishikawa and RL95.2 cells by electroporation followed by selection with G418. After 16 weeks, surviving clones were isolated and expanded. Two Ishikawa cell clones (IshADM1 and Ish-ADM2) and two RL95.2 cell clones (RL-ADM1 and RL-ADM2) highly expressing ADM mRNA and peptide were identi®ed by immunohistochemistry of cytospins and Western analysis of conditioned media (Figure 2). The secreted ADM peptide was shown to be active by stimulation of growth of human dermal microvascular endothelial cells (HDMEC) by transfectant conditioned media (Figure 3). Finally RL95.2 cells were shown to express CRLR and RAMP2, that together function as the ADM receptor, by RT ± PCR (data not shown). Oncogene

Figure 1 (a) Histology of control and adrenomedullin treated sponge implants at day 20 after implantation (hematoxylin/eosin stained). (b) Vascular density of sponge implants determined by Chalkley counting. (n=4+s.e.m.). Note the greater penetration of ®brotic tissue into and the presence of many more erythrocyte containing vessels in the sponge that received ADM

Growth of transfectants in vitro In vitro growth of ADM transfectants and controls was determined by MTS-assay. Constitutive expression of ADM had no eect on Ishikawa growth but markedly stimulated that of RL95.2 cells (P50.05) (Figure 4). Growth of transfectants in vivo Transfected clones were examined for their ability to form tumors in vivo. Ishikawa cells are estrogendependent and very weakly tumorigenic in athymic mice. In view of this they were co-xenografted with an MDA-435S carrier line and estrogen supplemented as previously described (Zhang et al., 1995b; Ali et al., 2000). Figure 5 shows that while expression of ADM had a comparatively small eect on Ishikawa/MDA435S tumor growth (P50.05) it had a marked eect on growth of RL95.2 tumors (P50.01). Similarly striking was the eect of ADM expression on tumor take where ADM expression greatly facilitated tumorigenicity in RL95.2 cells (P50.05) (Table 1). Vascular density of xenografted tumors Frozen sections of the xenografted tumors were immunostained for the endothelial marker CD31 and


2817

Figure 2 Characterization of two Ishikawa and two RL95.2 transfected clones that overexpress adrenomedullin. The cells were shown to overexpress ADM by immunohistochemistry (top) and Western analysis (bottom)

Figure 3 Stimulation of human dermal microvascular endothelial cell growth by conditioned media from control and adrenomedullin transfected Ishikawa cells (*P50.05, n=5, mean+s.e.m.). O.D. is from the MTS assay and is proportional to the cell density (see Materials and methods)

the vessel density quantitated by Chalkley counting. A signi®cantly increased vascular density of tumors overexpressing ADM could be found in both Ishikawa/MDA-435S (data not shown) and RL95.2 tumors (P50.05 (n=3. Mean+s.e.m.) Wilcoxon matched pairs test) (Figure 6). Discussion The purpose of this study was to examine a role for the angiogenic peptide ADM in tumor formation. To this end, we have examined the eect of expression of ADM on the tumorigenesis of human endometrial carcinoma

cells in athymic mice. It was shown that ADM increases both tumor take and growth rate, but that the latter varies considerably on whether the carcinoma cell line itself is or is not growth stimulated by ADM. Two ADM expressing Ishikawa clones were isolated and their growth properties compared to that of a control empty vector transfectant. Expression of ADM had no eect on in vitro growth of those cells but caused a small but statistically signi®cant enhancement of in vivo growth. Ishikawa cells are known to express ADM receptors in vitro (Oehler et al., 2001), this however, appeared not to lead to a growth eect in the presence of ADM. Despite the lack of a growth eect of ADM on Ishikawa cells, ADM clearly enhances their survival in culture especially under hypoxia (Oehler et al., 2001). In the case of RL95.2 cells, transfectants grew faster than controls in vitro (Figure 4). Far more striking, however, was the dramatic increase in tumor growth seen in xenografts (Figure 5). The tumorigenic activity of ADM was supported by the tumor take data shown in Table 1. Thus while only 6 out of 10 implants of control transfectants gave tumors, 9 and 10 out of 10 of the ADM transfectant implants gave tumors. This eect was highly reproducible, indeed two separate experiments gave identical results. Examination of the vasculature in sections of the xenografted tumors immunostained with CD31 showed that in both Ishikawa/MDA-435S (data not shown) and RL95.2 tumors (Figure 6) there was a marked increase in the vascular density in those expressing ADM compared to controls. Vascular density was quantitated by Chalkley counting. Oncogene


2818

Figure 5 Eect of ADM expression on (a) Ishikawa/MDA-435S (P50.05, n=10 mean+s.e.m.) and (b) RL95.2 (P50.01, n=10+s.e.m.) cell growth in vivo

Figure 4 Eect of ADM expression on (a) Ishikawa and (b) RL95.2 cell growth in vitro. O.D. is from the MTS assay and is proportional to the cell number (see Materials and methods)

Dierent growth patterns of Ishikawa and RL95.2 cells in vivo may partly be explained by cell line speci®c properties. Ishikawa cells are known to be ER positive estrogen dependent cells and in this sense they are closer to endometrial epithelium than ER negative late passage RL95.2 cells. ER expression is well documented to inhibit in vitro and in vivo growth of endometrial cancer cells and was recently shown to inhibit VEGF induced angiogenesis (Ali et al., 2000). While xenografted RL95.2 cells readily formed tumors in athymic mice, Ishikawa did not. For this reason it was necessary to co-implant Ishikawa cells with carrier cells (in this case MDA-435S cells) and estrogen supplementation to observe growth in vivo. This widely used strategy (Kamitani et al., 1999) permitted a direct comparison of the growth of transfectants in vivo. We conclude that ADM is pro-tumorigenic in endometrial cancer cells as a result of its mitogenic and angiogenic properties. With Ishikawa cells enhanced in vivo growth due to increased angiogenesis was seen whereas in RL95.2 cells a direct mitogenic eect of ADM on the carcinoma also occurred. This resulted in a greater enhancement of tumor growth after transfection of ADM into RL95.2 cells compared to the same experiment with Ishikawa cells. Oncogene

Table 1 Eect of ADM-transfection on tumor take of RL95.2 endometrial carcinoma cells. The experiment was performed twice with identical results Clone RL-ADM1 RL-ADM2 Control

Tumor takea 10/10 (100%) 9/10 (90%)* 6/10 (60%)

a

Number of positive mice/number of injected mice. *P50.05

Of the many angiogenic polypeptides now identi®ed it appears that those induced by hypoxia are key players in tumor angiogenesis. ADM up-regulation under hypoxia is controlled by the hypoxia-inducible transcription factor-1 (HIF-1) as is VEGF (Semenza, 1999). HIF-1 is a heterodimer composed of the oxygenregulated HIF-1a and the constitutively expressed HIF-1b/ARNT subunits, both representing members of the PAS (Per, ARNT, Sim) basic-helix-loop helix family (Semenza, 1999). HIF-1 binds to hypoxia response elements (HREs) several of which are located in the ADM promoter (Nakayama et al., 1998). The importance of the HIF-1 response pathway in human tumorigenesis is underscored by the ®nding that HIF-1 is overexpressed in many cancers (Zhong et al., 1999) and that patients with strong expression of HIF-1 show a signi®cantly shorter overall survival (Birner et al.,


al., 1993). Human ADM-(1-52) was synthesized by the ICRFPeptide-Unit, Clare Hall, London, UK. All reagents were purchased from Sigma, Poole, UK, unless stated otherwise. The tumor cell line RL95.2 (CRL-1671) was obtained from the American Type Culture Collection, Bethesda, USA, the cell lines Ishikawa and MDA-435S from the Clare Hall Laboratories, Imperial Cancer Research Fund, London, UK and human dermal microvascular endothelial cells (HDMEC) from Clonetics/Biowhittaker, Wokingham, UK. Ishikawa cells were routinely cultured in DMEM supplemented with 10% FCS and 4 mM glutamine, RL95.2 cells in DMEM/Nutrient Mixture F-12 K Kaighn's Modi®cation ± 1 : 1 supplemented with 10% FCS and HDMEC in EGM-2MV Bullet kit medium (Clonetics/Biowhittaker, Wokingham, UK). All cells were cultured at 378C and 5% CO2.

2819

Vector construction and stable transfection

Figure 6 Immunohistochemistry of xenografted tumors that are formed on implantation of ADM-transfected RL95.2 cells and controls. (a) Vessels were visualized by staining with anti-CD31. (b) Vascular density quantitated by Chalkley counting. *P50.05 (n=3, mean+s.e.m.)

2000). More signi®cantly, disruption of hypoxiainducible transcription severely abrogates tumor growth (Kung et al., 2000). The critical eectors by which HIF1 modulates tumor vascularization and subsequent growth are, however, still unknown. One is VEGF and we propose that ADM may be another in that ADM is a mitogen for both endothelial and carcinoma cells as well as a survival factor for malignant cells under hypoxia (Zhao et al., 1998; Nikitenko et al., 2000; Oehler et al., 2001). ADMs role in the vasculature has been recently reviewed (Nikitenko et al., 2002). In contrast to a hypoxic stimulus, the mechanism by which TAM stimulates ADM expression is not clear as the promoter lacks palindromic oestrogen response elements (Ishimitsu et al., 1994) and it is not induced by oestrogen. The ADM promoter does, however, have two AP-1 binding sites at positions ± 1549 and ± 1801 respectively (Ishimitsu et al., 1994). As TAM has been demonstrated to initiate gene transcription via AP-1 sites (Webb et al., 1995) this could be a possible mechanism of transcription. Overall, this study presents evidence of a tumorigenic activity of ADM that further supports a role for ADM in tamoxifen induced endometrial hyper- and neoplasia. Materials and methods The human ADM cDNA was a gift from Dr Kitamura, First Department Internal Medicine, Kihara, Japan (Kitamura et

The full-length cDNA for human ADM was cloned into the XhoI/BamHI-restriction sites of pcDNAI/Neo to give plasmid pCMV-ADMneo in which expression of ADM is constitutively driven by the CMV-promoter. Ishikawa and RL95.2cells were transfected with pCMV-ADMneo or pcDNAI/Neo (empty vector control) by electroporation at 400 V with a capacitance of 25 mF in transfection buer (272 mM sucrose, 7 mM sodium phosphate pH 7.4, 1 mM MgCl2). Transfected cells were then selected for neomycin resistance by treatment with G418 (500 mg/ml) for 16 weeks. Isolated clones were expanded for further characterization. Reverse transcription (RT)-polymerase chain reaction (PCR) CRLR and RAMP2 gene expression in RL95.2 cells were examined by RT ± PCR using the Reverse-iT-kit (1st strand synthesis kit) (Advanced Biotechnologies, Epsom, UK). For the RT-reaction anchored oligo-dT primers included in the kit were used. Primers and PCR conditions were chosen as previously described (Oehler et al., 2001). Immunoblot analysis Preparation of samples Serum-free cell culture medium was conditioned by con¯uent cultures for 48 h. The medium was concentrated (1006) with a Centricon-3-Concentrator followed by a Microcon-5-Concentrator (Amicon, Beverley, USA). Protein concentrations were quanti®ed using the BCA protein assay (Pierce, Rockford, USA). The samples were stored at 7808C until assay. Immunoblotting For immunoblotting, samples were run on a gradient 10 ± 20% tricine, SDS ± PAGE (BioRad, Hercules, USA). The proteins were electrophoretically transferred onto a polyvinylidene di¯uoride membrane (Immobilon PVDF ± Millipore, Watford, UK). The PVDF membrane was blocked for 2 h in PBS with 0.1% Tween-20, 5% bovine serum albumin and incubated in a 1 : 1000 dilution of a polyclononal rabbit anti-human ADM antibody (Peninsula Laboratories, Merseyside, UK). Binding was visualized by a horseradish peroxidase-conjugated pig anti-rabbit IgG and developed with ECL Western blotting reagents (Amersham, Little Chalfont, UK). Quanti®cation of protein expression was performed by densitometry (Fluorchem, Alpha Innotech Corp, USA). Determination of in vitro cell growth Ishikawa and RL95.2 cell clones were quiescent for 24 h in their respective medium supplemented with 0.5% FCS before growth assays. Cells were seeded at 26103 cells/well in 96-well plates. The culture Oncogene


2820

medium was changed every 48 h and cell growth measured by MTS-assay (Promega, Madison, USA) every 24 h for 6 days. In the assay viable cells reduce MTS tetrazolium to a coloured formazan which is measured at 490 nm with a 96well plate reader (Dynex Technologies, Billingshurst, UK). For cell growth experiments with HDMECs, cells were seeded at 46103 cells/well in gelatin-coated 96-well plates. The cells were quiesced in endothelial growth medium (EGM2MV; Clonetics/Biowhittaker, Wokingham, UK) supplemented with 1% FCS for 24 h. Cells were then cultured in endothelial growth medium (EGM)/1% FCS with various concentrations of ADM or in EGM/1% FCS with conditioned media of ADM transfected Ishikawa clones (1 : 1). The cell culture medium was changed after 24 h and cell growth measured by MTS-assay after 48 h. Mouse subcutaneous sponge angiogenesis assay The sponge implantation model was set up as described previously (Walsh et al., 1996) with some modi®cations. C57 black mice were anesthetized using ¯uothane and received a subcutaneous sterile polyether sponge disc (868 mm) (Caligen Foam Ltd., Accrington, UK) under the dorsal skin. Rat adrenomedullin (rADM) (Peninsula Laboratories, Merseyside, UK) at concentrations 1078 ± 10712 M in a total volume of 90 ml physiological saline was injected daily through the skin into the sponges from day 1 to 20. The control mice received physiological saline alone. At the end of the experiment, the animals were sacri®ced by cervical dislocation and the sponges rapidly excised together with overlaying skin and underlying connective tissue. The sponges were ®xed in 10% formalin at 48C for 1 h and then immersed in 75% ethanol for 30 min and ®nally kept in 90% ethanol. For analysis of ®brovascular growth areas samples were embedded in paran wax and 10 mM thick sections evenly distributed throughout each sponge were stained with hematoxylin/eosin and CD31. Assessment of sponge vascularity The vascularity of the explanted sponges was determined by Chalkley counting (Fox et al., 1995). The three most vascular areas with the highest number of discreet vessels were identi®ed by scanning at low power (640 and 6100). Vascular density was determined using a 25-point Chalkley eyepiece graticule at 6250 magni®cation (the graticule covers an area of 0.155 mm72 at this magni®cation). The graticule was rotated in the eyepiece to where the maximum number of vessels was overlaid by graticule dots. Individual density was then obtained by taking the mean of three graticule counts. Xenograft experiments in BALB/c nu/nu mice BALB/c nu/nu mice were anesthesized using ¯uothane and RL95.2 cells injected (106 cells in 0.2 ml cell suspension) subcutaneously into one ¯ank. In a second setting mice received two subcutaneous implants: Ishikawa (106 cells) and MDA-435S cells (106) in 0.2 ml cells suspension and a 60-day release

pellet containing 1.5 mg of 17b-estradiol (Innovative Research of America, Toledo, Ohio, USA). Tumors were measured twice a week from ®rst appearance. When tumors reached 1.44 cm2 animals were sacri®ced by cervical dislocation. The tumor volume was calculated using the formula: Tumor volume (cm_)=(length6(width)2_)60.4 (Attia and Weiss, 1966). All animal experiments were performed in accordance with the British Home Oce Animals (Scienti®c Procedures) Act of 1986 under license number 70/4949. Immunohistochemistry of mouse tumor tissue samples Cryostat sections (8 mm) of tumors were air dried for 24 h, ®xed in acetone (10 min), air-dried for 40 min, and then stored at 7208C until use. The mouse tumor vasculature was examined immunohistochemically. Tissue sections were incubated with PECAM-1 (rat anti-mouse CD31) (Pharmingen, San Diego, USA) for 1 h at a 1 : 200 dilution, rinsed in PBS for 10 min, incubated with biotinylated rabbit anti-rat IgG at a 1 : 300 dilution for 30 min, rinsed again in PBS for 10 min, incubated with avidin ± biotin ± peroxidase complex, (Dako, Ltd.) for 30 min, with a ®nal rinse in PBS. The chromogen was developed for 10 min using diaminobenzidine tetrahydrochloride, 0.6 mg/ml in PBS containing 3 ml/ml of hydrogen peroxide. All sections were lightly counterstained with haematoxylin-eosin. Statistical analysis Statistics were performed using the SPSS software package. Comparisons between groups were made applying the Wilcoxon matched pairs test or the Fisher exact test. Abbreviations adrenomedullin, ADM; calcitonin receptor-like receptor, CRLR; receptor activity modifying protein 2, RAMP-2; tamoxifen, TAM; hypoxia-inducible transcription factor-1, HIF-1; vascular endothelial growth factor, VEGF; erythropoietin, EPO; ER, estrogen receptor; human dermal microvascular endothelial cells, HDMEC; endothelial growth medium, EGM; chick chorioallantoic membrane assay, CAM assay; optical density, O.D. Acknowledgments The authors acknowledge the excellent technical assistance of Sandra Peak, ICRF Clare Hall Laboratories, London in performing the xenograft and sponge assays, Helen Turley, Department of Cellular Science, University of Oxford for help with immunohistochemistry and cytospins and Mr Daryl Harman, Caligen Foam Ltd, Accrington for help in obtaining sponge implants. Financial support was provided by the Imperial Cancer Research Fund, The Deutsche Forschungsgemeinschaft and The Sir Jules Thorn Charitable Trust.

References Ali SH, O'Donnell AL, Balu D, Pohl MB, Seyler MJ, Mohamed S, Mousa S and Dandona P. (2000). Cancer Res., 60, 7094 ± 7098. Attia M and Weiss D. (1966). Cancer Res., 26, 1787 ± 1800. Bilimoria MM, Assikis VJ, Muenzner HD, Wolf DM, Satyaswaroop PG and Jordan VC. (1996). J. Steroid Biochem Mol. Biol., 58, 479 ± 488. Oncogene

Birner P, Schindl M, Obermair A, Plank C, Breitenecker G and Oberhuber G. (2000). Cancer Res., 60, 4693 ± 4696. Bergman L, Beelen M, Gallee M, Hollema H, Benraadt J and van Leeuwen F. (2000). Lancet, 356, 881 ± 887. Caron KM and Smithies O. (2001). Proc. Natl. Acad. Sci. USA, 98, 615 ± 619.


Cormier-Regard S, Nguyen SV and Claycomb WC. (1998). J. Biol. Chem., 273, 17787 ± 17792. Cuzick JA. (2000). Eur. J. Cancer, 36, 1298 ± 1302. Dalton RR and Kallab AM. (2001). South Med. J., 94, 7 ± 15. Fox SB, Leek RD, Weekes MP, Whitehouse RM, Gatter KC and Harris AL. (1995). J. Pathol., 177, 275 ± 283. Garayoa M, Martinez A, Lee S, Pio R, An WG, Neckers L, Trepel J, Montuenga LM, Ryan H, Johnson R, Gassmann M and Cuttitta F. (2000). Mol. Endocrinol., 14, 848 ± 862. Hata K, Takebayashi Y, Akiba S, Fujiwaki R, Lida K, Nakayama K, Nakayama S, Fukumoto M and Miyazaki K. (2000). Mol. Hum. Reprod., 6, 867 ± 872. Hinson JP, Kapas S and Smith DM. (2000). Endocr. Rev., 21, 138 ± 167. Ishimitsu T, Kojima M, Kangawa K, Hino J, Matsuoka H, Kitamura K, Eto T and Matsuo H. (1994). Biochem. Biophys. Res. Commun., 203, 631 ± 639. Kamitani S, Asakawa M, Shimekake Y, Kuwasako K, Nakahara K and Sakata T. (1999). FEBS Lett., 448, 111 ± 114. Kitamura K, Kangawa K, Kawamoto M, Ichiki Y, Nakamura S, Matsuo H and Eto T. (1993). Biochem. Biophys. Res. Commun., 192, 553 ± 560. Kung AL, Wang S, Klco JM, Kaelin WG and Livingston DM. (2000). Nat. Med., 6, 1335 ± 1340. Lal A, Lash AE, Altschul SF, Velculescu V, Zhang L, McLendon RE, Marra MA, Prange C, Morin PJ, Polyak K, Papadopoulos N, Vogelstein B, Kinzler KW, Strausberg RL and Riggins GJA. (1999). Cancer Res., 59, 5403 ± 5407. McLatchie LM, Fraser NJ, Main MJ, Wise A, Brown J, Thompson N, Solari R, Lee MG and Foord SM. (1998). Nature, 393, 333 ± 339. Miller MJ, Martinez A, Unsworth EJ, Thiele CJ, Moody TW, Elsasser T and Cuttitta F. (1996). Biol. Chem., 271, 23345 ± 23351.

Nakayama M, Takahashi K, Murakami O, Shirato K and Shibahara S. (1998). Biochem. Biophys. Res. Commun., 243, 514 ± 517. Nikitenko LL, Smith DM, Hague S, Wilson CR, Bicknell R and Rees MCP. (2002). Trends in Pharm. Sci., in press. Nikitenko LL, MacKenzie IZ, Rees MC and Bicknell R. (2000). Mol. Hum. Reprod., 6, 811 ± 819. Oehler MK, Norbury C, Hague S, Rees MCP and Bicknell R. (2001). Oncogene, 20, 2937 ± 2945. Pio R, Martinez A, Unsworth EJ, Kowalak JA, Bengoechea JA, Zipfel PF, Elsasser TH and Cuttitta FJ. (2000). Biol. Chem., 486, 3. Rocchi P, Boudouresque F, Zamora AJ, Muracciole X, Lechevallier E, Martin PM and Oua®k L. (2001). Cancer Res., 61, 1196 ± 1206. Semenza GL. (1999). Annu. Rev. Cell Dev. Biol., 15, 551 ± 578. Sundareshan P and Hendrix MJC. (1992). In Vitro, 28A, 544 ± 552. Walsh DA, Hu DE, Mapp PI, Polak JM, Blake DR and Fan TP. (1996). Histochemical Journal, 28, 759 ± 769. Webb P, Lopez GN, Uht RM and Kushner PJ. (1995). Mol. Endocrinol., 9, 443 ± 456. White IN. (1999). Carcinogenesis, 20, 1153 ± 1160. Wimalawansa SJ. (1997). Crit. Rev. Neurobiol., 11, 167 ± 239. Zhang L, Rees MCP and Bicknell R. (1995a). J. Cell Sci., 108, 323 ± 331. Zhang HT, Craft P, Scott PA, Ziche M, Weich HA, Harris AL and Bicknell R. (1995b). J. Natl. Cancer Inst., 87, 213 ± 219. Zhao Y, Hague S, Manek S, Zhang L, Bicknell R and Rees MC. (1998). Oncogene, 16, 409 ± 415. Zhong H, De Marzo AM, Laughner E, Lim M, Hilton DA, Zagzag D, Buechler P, Isaacs WB, Semenza GL and Simons JW. (1999). Cancer Res., 59, 5830 ± 5835.

2821

Oncogene