The G-Protein–Coupled Receptor CLR Is Upregulated in an Autocrine Loop with Adrenomedullin in Clear Cell Renal Cell Carcinoma and Associated with Poor Prognosis

Adrenomedullin (encoded by an ADM gene) is a multifunctional peptide involved in cellular proliferation, survival, and angiogenesis (1, 2). Whilst the role of adrenomedullin in tumor biology has been established from studies using in vitro and mouse models and although ADM gene expression is known to be upregulated by hypoxia (a typical feature of solid tumors; refs. 3–5), there is a general lack of information about the expression of adrenomedullin receptors in human cancers and its correlation with the disease outcome in particular (reviewed in ref. 4).

In vitro, adrenomedullin mediates its activities through binding to G-protein–coupled receptor (GPCR) calcitonin-like receptor (CLR, encoded by the CALCRL gene; refs. 6, 7). The GPCRs are the largest family of cell surface molecules involved in signal transduction and they control key physiologic functions, including regulation of blood pressure, immune system activity, and inflammation. These receptors are involved in some of the most prevalent human diseases, as reflected by the fact that GPCRs directly or indirectly represent the targets of 50% to 60% of all current therapeutic agents (8–10). For example, the analgesic drugs codeine and morphine target opioid receptors in the nervous system and the GPCR antagonist zibotentan (ZD4054) targets the endothelin A receptor in solid tumors (9, 11). Recently, some GPCRs, such as chemokine and prostaglandin receptors, have emerged as crucial players in tumor growth and metastasis, suggesting their importance in initiation and progression of cancer in humans (4, 11, 12).

In the present study, we tested the hypothesis that GPCR CLR may be important in tumor biology in man and could play a role in some human cancers. Our aims were to determine those tumor types that may rely on adrenomedullin, examine the pattern and level of expression of adrenomedullin receptor CLR in a prototypical example, and conduct a clinicopathologic analysis to assess its diagnostic and/or prognostic use. We therefore carried out a large-scale in silico gene expression (mRNA profiling) microarray study coupled with cancer profiling cDNA array hybridization and identified tumors that had upregulated ADM expression compared with patient-matched normal tissue counterparts. Our data showed that ADM expression was significantly upregulated only in clear cell renal cell carcinoma (RCC) and endometrial adenocarcinoma (P < 0.01) but not in other tumor types, suggesting that in human tissues/organs it is upregulated in cancer type–specific manner. We then focused on RCC as potentially an adrenomedullin signaling–driven tumor and, by using immunohistochemistry and a tissue microarray (TMA) approach, analyzed CLR expression in a large cohort of patient-matched paired carcinoma and normal kidney tissue samples. Our data showed that CLR was localized in tumor cells and tumor vessels and that its expression was upregulated in RCC, when compared with patient-matched controls. Moreover, CLR expression was higher in advanced stages and correlated with tumor grade and overall patient survival. Our findings suggest that GPCR CLR has clinical significance in RCC as a promising immunohistochemical biomarker of disease outcome and as a potential molecular target for therapeutic agents.

Please see details of the following methods in the Supplementary Data: antibody, cell culture, SDS–PAGE and immunoblotting, cancer profiling microarray hybridization and Northern blotting, RNA isolation, cDNA synthesis, and quantitative real-time PCR (qRT-PCR).

Public Affymetrix GeneChip HG-U133plus2 arrays data for 11 carcinoma types (n = 168 samples) were obtained from the Expression Project for Oncology expO (http://www.intgen.org/) via the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) repository (http://www.ncbi.nlm.nih.gov/projects/geo; accession GSE2109). Raw data were preprocessed and normalized using Bioconductor software for R (13), and the robust-multiarray algorithm (rma) as previously described (14–16).

The Cancer Profiling Microarray 7841 (cDNA microarray), consisting of normalized paired cDNA samples (n = 241) generated from the total RNA of 13 organs (breast, uterus, colon, stomach, ovary, lung, kidney, rectum, thyroid, prostate, cervix, small intestine, and pancreas) was obtained from Clontech (http://www.clontech.com). Each cDNA pair consisted of carcinoma and corresponding organ-matched normal tissue samples obtained from the same patient, represented the entire mRNA message expressed in given tissues, and was used to provide comprehensive ADM gene–disease correlation data.

ADM cDNA was cloned into TOPO vector (Promega) and ubiquitin cDNA probe was from BD Biosciences. Inserts were excised with restriction enzymes and labeled with ³²P-dCTP using the Megaprime DNA Labeling Kit (GE Healthcare). The specificity of the probes was confirmed by Northern blotting, which was conducted as previously described (17). Cancer profiling microarray was consequentially hybridized with each individual probe according to the manufacturer's instructions. The microarray was exposed to Hyperfilm (GE Healthcare) and subsequently to Phosphoscreen. The hybridization signals were further analyzed using ImageQuant software. Ratio of ADM:Ubiquitin was calculated to determine the relative mRNA expression levels and jittered dot-plot graphs were generated using GraphPad Prism software.

A total of 131 patients with histologically confirmed renal carcinoma from two centers [John Radcliffe Hospital, Oxford, United Kingdom (cohort 1; n = 113 patients) and Canterbury District Health Board, Christchurch, New Zealand (cohort 2; n = 18 patients)] were included in this study.

Baseline histologic characteristics on all analyzed renal tissue samples from cohort 1 (used for immunohistochemical analysis as described later) and the information on age and tumor characteristics (e.g., stage, grade, size, etc.) for patients with RCC are provided in Tables 1 and 2, respectively. The cause of death was obtained from the Oxford Cancer Registry from the Oxford Cancer Intelligence Unit (http://www.ociu.nhs.uk/; ref. 18). The median follow-up time after nephrectomy was 43 months (ranging from 1 to 296 months) and a mean overall survival was of 3.3 years. None of the patients received novel experimental therapies, such as tumor vaccines or molecular targeted medicine, nor had received any conventional treatment for at least 6 months before the surgery. Overall survival was defined as the time from nephrectomy and cancer diagnosis until death, the last time that the patient was known to be alive, or until the end of follow-up. All hematoxylin and eosin (H&E)–stained slides were reviewed by experienced genitourinary pathologists and histologically classified according to the World Health Organization (WHO) classification of tumors (19). Tumors were classified according to the Fuhrman grade (20) and tumor—node—metastasis (TNM) staging system (21) and data were deposited in the Cancer Registry. T stage data were available for all patients, whereas N and M staging data were not and therefore not included in analysis. The samples from the cohort 2 patients were collected for routine histologic evaluation and also for reverse transcriptase PCR (RT-PCR) and qRT-PCR analysis as described later.

All patients were confirmed to have sporadic disease based on their medical records (obtained from Oxford Cancer Registry), according to which no family history of RCC was elucidated. The Central Oxfordshire Research Ethics Committee (C00.147; C02.216) and Canterbury Ethics Committee (V2-4 02.06.98-01.05.2002) approved the use of all human tissues used in this study.

TMA of 1 mm cores (from formalin-fixed, paraffin-embedded specimens of renal carcinoma and normal tissues) with 2-fold redundancy from 113 patients (including 87 pairs of patient-matched samples) of cohort 1 was constructed (18) and 4 μm sections were cut. The presence of all analyzed patient-matched normal and tumor tissue samples on the same TMA section enabled the performance of immunostaining under standardized conditions and its quantitative comparison. Immunohistochemical analysis of CLR expression was conducted using previously characterized rabbit polyclonal anti-human CLR antibody LN1436 (22) and preimmune serum from the same rabbit, in which the antibody was raised, as a control. Additional control was conducted using primary antibody preincubated with 10 μg/mL peptide, against which it was raised, for 90 minutes at room temperature before immunostaining. Scoring of the intensity and proportion of tumor or normal epithelial cells was conducted by two independent pathologists and on two independently immunostained TMA using a semiquantitative analysis as described elsewhere (23). Briefly, the intensity of the staining [“no staining” (0), “weak staining” (1), “moderate staining” (2), or “strong staining” (3)] and the percentage of stained cells [0%–10% (1), 10%–50% (2), 51%–80% (3), or 81%–100% (4)] were determined. Intensity was multiplied by percentage to obtain “CLR intensity and percentage score” for each tissue sample on TMA. The scoring of the intensity (but not the percentage) of the CLR immunoreactivity in tumor vessels or normal vessels was conducted following the same approach.

RNA isolation, cDNA synthesis, RT-PCR, and qRT-PCR were conducted as previously described (17, 24) using tissue samples collected into the liquid nitrogen and stored at −80°C. For qRT-PCR, the following primer/probe kits were used: Hs00181605_m1 (ADM) and Hs99999903_m1 (ACTB; both from Applied Biosystems). Relative quantification of gene expression was conducted using previously described method (25), based on the mean value of qRT-PCR reactions carried out in triplicate. Human ACTB was used as a reference gene to normalize for differences in the amount of total RNA in each sample. The comparator for the clinical samples was the median from normal kidney samples.

All statistical analyses were conducted using SPSS Statistics software (version 16.0), GraphPad Prism or Microsoft Excel computer programs. mRNA expression levels in patient-matched normal and cancer tissues were analyzed using Wilcoxon signed-rank t test. Analysis between the level of CLR immunostaining in tumor cells on TMA and various clinicopathologic parameters (patient's age, tumor size, histology, T stage, and Fuhrman grade) was conducted using Spearman rank correlation or paired t tests depending on the dataset as previously conducted in other studies (26, 27). For all tests, two-sided analysis was conducted and actual P values are shown. P values of less than 0.05 were considered statistically significant.

For the survival analysis, all immunostained on the TMA RCC samples were divided into two categories according to CLR immunostaining in tumor cells—“CLR positive” (1–12) or “CLR negative” (0) based on intensity and percentage score; or divided at the median into two categories according to the staining in tumor/normal vessels—“CLR high” and “CLR low.” Kaplan–Meier curves and log-rank Mantel–Cox test was used to estimate the association of CLR expression levels in tumor cells or tumor vessels with an overall survival to obtain protein-disease outcome data based on available data for patients with RCC (total n = 87; Table 2). All survival analyses refer to overall survival times, where time to death from disease represents an event. Patients were censored in survival according to the date last seen by a doctor as indicated earlier. The median survival times and HRs within each subgroup were estimated from Kaplan–Meier curves.

There was a significant heterogeneity of ADM expression across 11 human tumor types studied here (Fig. 1A). The highest (mean log₂ value 11.5) levels were in RCC and the lowest (mean log₂ value 8) levels were in breast carcinoma, that is, approximately 10-fold variation. Furthermore, ADM expression was significantly altered only in three tumor types out of 13 analyzed carcinomas (upregulated in kidney and uterine tumors, whereas downregulated in breast carcinoma; Fig. 1B; Supplementary Fig. S1). The highest upregulation of ADM mRNA expression was found in RCC when compared with paired normal kidney (Fig. 1C and Supplementary Fig. S1) and this was confirmed by qRT-PCR using an independent cohort of 18 pairs of patient-matched renal carcinoma and normal control tissue samples (Fig. 1D).

CLR was variably expressed in tumor cells and tumor vessels in RCC and in normal patient-matched tissue samples (n = 87 pairs, including 69 RCC cases) and in RCC cell lines, as shown by immunohistochemistry and/or immunoblotting. CLR was localized in both tumor cells and tumor vessels (Fig. 2 and Supplementary Fig. S2A and S2B) and its expression was upregulated in renal tumors in general and in individual types (Supplementary Fig. S3A and S3B), including RCC (Fig. 3A), when compared with control patient-matched normal tissues.

CLR was terminally but not core-glycosylated when expressed in RCC tissues (Supplementary Fig. S4A). Fully processed terminally glycosylated form of CLR is only produced in the presence of the receptor activity modifying proteins (RAMPs) and only in this particular form this GPCR is expressed at the cell surface and can bind to the ligand, that is, to be “functional” (28). Therefore, our findings suggest that CLR is likely to be in its cell surface–bound/functional form (6, 22, 28, 29), when expressed in RCC tumor cells and tumor vessels in situ (Fig. 2 and Supplementary Fig. S2). In vitro, CLR was also expressed in some, but not all, renal tumor cell lines (Supplementary Fig. S4B). We confirmed that the presence of terminally glycosylated CLR in renal tumor tissues or cultured renal tumor cell lines and primary microvascular endothelium was associated with the expression of RAMP1, 2, and 3 mRNAs (Supplementary Fig. S5), which encode proteins that facilitate terminal glycosylation and transport of the CLR to the cell membrane (6).

CLR expression in tumor cells (but not in tumor vessels) was significantly higher (P < 0.05) in advanced stages (II and III; Supplementary Fig. S6) and correlated with high tumor grade (P < 0.01), but not with tumor size or patient age (Table 3). Furthermore, patients with RCC with CLR expressed in tumor cells had significantly shorter overall survival when compared with those with no CLR (with median survival times of 543 and 784 days, respectively and HR, 0.4182; P = 0.01; Fig. 3B).

Renal cancers account for 2% to 4% of cancers worldwide (http://www.cancer.org). The 5-year survival of patients with renal cancer is 40.5%, largely due to metastasis which accompany 75% of the cases (30, 31). In these patients, systemic or adjuvant treatments, such as conventional cytotoxic chemotherapy or cytokine immunotherapy, give a sustained response rate of only 10% and they are associated with considerable toxic effects (30–32). Recently, the unraveling of the hypoxic response in RCC through von Hippel-Lindau (VHL)/hypoxia-inducible factor (HIF)–mediated signaling has shown how its downstream pathways and processes, such as angiogenesis, are integral to the pathogenesis/progression of renal cancer (5). Indeed, tumor responsiveness to the newer anticancer and antiangiogenic drugs such as sorafenib, sunitinib, and bevacizumab, targeting vascular endothelial and/or platelet-derived growth factor (VEGF and PDGF, respectively; both products of the genes which are transcriptional targets of HIF) pathways, showed the benefit of exploiting them for targeted therapy in human cancers, including RCC (33–36). Nevertheless, a significant number of patients with RCC do not completely respond (or become resistant) to these antiangiogenic or current combination therapies, which often require long-term administration for continued disease control and have several adverse effects. This suggests that other molecular pathways may be activated in this type of cancer and play a significant role in its progression and also support the need for the continued exploration of these pathways and prognostic or predictive markers for this disease as well as for the development of novel agents and new treatment combinations (32).

Adrenomedullin signaling pathway(s) mediated by GPCR CLR could be one such potentially significant axis for therapeutic intervention in RCC due to the upregulation of ADM expression by hypoxia and HIF (similar to VEGF and PDGF genes), and also because the role for adrenomedullin in tumor growth and angiogenesis has been already established by studies using mouse xenografted tumor models and in vitro assays (reviewed in refs. 3, 4). A recent study suggested that signaling via CLR has potential role in bypassing the normal requirement for VEGF signaling in endothelial cells during embryogenesis (37). However, while these observations pointed to a potential role for adrenomedullin in tumor biology and angiogenesis, the clinical significance or functional contribution of adrenomedullin signaling to tumorigenesis in human tissues is as yet poorly defined (reviewed in ref. 4). To date, studies in man have been limited to measurements of ADM mRNA levels (38, 39) or adrenomedullin peptide concentration in blood and tissues (40) and to the use of few tumor cell lines or a limited number of neoplastic tissues; all only occasionally done with the comparison between the patient-matched normal and cancer samples. The data about CLR expression, including distribution/localization and presentation at the cell surface in vivo, in human tumor tissues are also very limited and have not been systematically studied (22, 38, 41). In particular, there is a general lack of information about the association of adrenomedullin and/or CLR expression with disease outcome and patient survival in various cancer types, including RCC. As a result, limited data are available not only for the evaluation of a possible role for adrenomedullin and CLR during initiation and progression of cancer in humans but also for an overall assessment of their value and potential in clinical oncology, for example, for diagnostics, prognosis, and/or targeted therapy (4, 29).

In the present study, by using large-scale in silico gene expression data screening coupled to the analysis of patient-matched carcinoma and normal tissue samples, we found that ADM mRNA expression in human organs and tissues is upregulated only in a small number of tumors and in RCC in particular; in keeping with two previous studies of ADM expression in renal cancer with combined numbers of 62 cases (38, 39). Furthermore, our TMA and immunoblotting data showed that adrenomedullin receptor CLR is localized both in tumor cells and tumor vessels and expressed in its functional (i.e., terminally glycosylated/cell surface–bound) form in renal carcinoma tissues. These findings suggest a potent autocrine loop or paracrine mode for adrenomedullin action and possibly biologically active in vivo roles for this CLR-mediated signaling in both angiogenesis and/or tumor cell biology within RCC microenvironment, especially upon upregulation of ADM expression. We therefore hypothesized (and tested) that CLR might be important in RCC progression.

In support of this hypothesis, we found that CLR expression in tumor cells is significantly upregulated compared with patient-matched normal epithelial cells and that its high levels strongly associate with advanced stages and correlate with tumor grade as well as with shorter survival of patients with RCC. To our knowledge, this is the first report showing that in human cancers the adrenomedullin receptor CLR has a prognostic value as the immunohistochemical biomarker of patient survival. Another important site for adrenomedullin action within the RCC microenvironment is tumor vessels, where CLR is also significantly upregulated when compared with patient-matched normal kidney vessels. We have previously shown that adrenomedullin interacts with endogenous CLR in cultured primary human microvascular endothelial cells, which express this GPCR in tissues, and induces their proliferation and migration in vitro (22), suggesting the role for adrenomedullin during angiogenesis and in vascular biology in general in human organs. In the present study, despite the lack of correlation of upregulated CLR expression in tumor vessels with various clinicopathologic parameters and patients survival, the increased levels of this GPCR might nevertheless have a direct impact on tumor angiogenesis—the hallmark of cancer progression in the kidney (5, 32, 35, 36, 42)—due to simultaneous upregulation in the expression of both the ligand and its receptor within the tumor microenvironment, as revealed in this study. Alternatively, because some of the RCC vessels may be lymphatics and because adrenomedullin may play a role in lymphatic endothelial cell biology (43–46), CLR can affect tumor lymphangiogenesis (and possibly tumor spread via lymphatic vessels) and therefore further vessel subtyping may provide additional prognostic information in the future.

It is widely acknowledged that the differential expression of a specific molecule in tumor compared with normal tissues is often crucial for the development of targeted therapies and strategies in cancer (47). Therefore, upregulated (in an autocrine loop with adrenomedullin) by RCC tumor cells and tumor vessels CLR has a potential as a cell surface–presented molecular target for future functional studies in renal cancer using already developed range of modulators of adrenomedullin-induced effects (and hence downstream signaling pathways mediated by this GPCR within human tumor tissue), including antibodies and peptide antagonists of adrenomedullin receptors (4, 48, 49).

In summary, the present study reveals previously unrecognized cancer type–specific upregulation of ADM expression in human tissues/organs and prognostic value for adrenomedullin receptor CLR in RCC. Our findings suggest the potential use for CLR as a novel target for future clinical studies and drug development for the therapy for RCC, and possibly other hypoxia-driven cancers. In particular, drug-resistant RCC could be an ideal model for further evaluating the prognostic use of CLR as a biomarker in first-line as well as refractory to antiangiogenic or combination treatments patients and for investigating the potential role for adrenomedullin- and CLR-mediated signaling in tumor cells and tumor vessels resistance to conventional chemotherapy and current targeted therapies.

L.L. Nikitenko is a consultant/advisory board member of Scientific Centre of the Family Health and Human Reproduction Problems, Siberian Branch of Russian Academy of Medical Sciences, Irkutsk, Russia. No potential conflicts of interest were disclosed by the other authors.

Conception and design: L.L. Nikitenko, M.C.P. Rees, A.L. Harris, S.B. Fox

Development of methodology: L.L. Nikitenko, D. Generali

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.L. Nikitenko, R. Leek, S. Gunningham, H.R. Morrin, S.B. Fox

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L.L. Nikitenko, R. Leek, S. Henderson, D. Generali, A. Pellagatti, A.L. Harris, S.B. Fox

Writing, review, and/or revision of the manuscript: L.L. Nikitenko, R. Leek, S. Henderson, D. Generali, S. Gunningham, H.R. Morrin, A. Pellagatti, M.C.P. Rees, A.L. Harris, S.B. Fox

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.L. Nikitenko, R. Leek, S. Henderson, N. Pillay, H. Turley

Study supervision: L.L. Nikitenko, A.L. Harris, S.B. Fox

The authors thank Dr. Rajeev Gupta (UCL Cancer Institute, London, United Kingdom) for helpful discussions and comments.

This study was supported in part by the Cancer Research UK (grants C575/A6125 and C575/A13100; to L.L. Nikitenko, D. Generali, S. Henderson, A.L. Harris, and S.B. Fox), The Wellcome Trust (grant 063353/Z/00Z; to L.L. Nikitenko and M.C.P. Rees), The Cancer Society, Canterbury West Coast Division, NZ (to H.R. Morrin), Medical Research Fund, University of Oxford, United Kingdom (to L.L. Nikitenko), The Royal Society UK, Cancer Research UK and The Maurice & Phyllis Paykel Trust travel awards (to L.L. Nikitenko).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.