Purpose: The role for the hypoxia-inducible angiogenic factor adrenomedullin (AM) in tumor growth and progression has been suggested. Calcitonin receptor–like receptor (CL) is a G protein–coupled receptor (GPCR) that mediates effects of AM, but little information is available on its expression and functional state in human tumors. The present study attempted to determine CL potential for antiangiogenic therapy of uterine leiomyoma.
Experimental Design and Results: GPCR CL is transported to the cell surface and recognized by AM only when terminally/mature glycosylated. The presence and localization of this form of the receptor in tumor and surrounding myometrial tissues obtained from leiomyoma-bearing uteri were examined using deglycosylation, immunoblotting, and immunofluorescence analysis. The mature CL glycoprotein was expressed in both tissues and localized exclusively in normal and tumor endothelium within leiomyoma-bearing uteri. The functionality of the receptor expressed in myometrial microvascular endothelial cells (MMVEC) was examined in vitro using receptor internalization and angiogenic assays. The mature CL glycoprotein expressed by primary MMVECs was functional because AM interacted with this GPCR and induced its internalization as well as angiogenic effects (proliferation and migration) in MMVECs in vitro. Finally, the levels of tissue-expressed mature CL glycoprotein as a functional form of this GPCR were analyzed by immunoblotting. The expression of this functional form of the receptor in vivo was significantly decreased (P = 0.01) in leiomyoma tissue, and this was concurrent with the decrease in microvascular density (measured by Chalkley counting) in tumor compared with surrounding myometrium (P = 0.031).
Conclusions: Our findings suggest that GPCR CL mediates angiogenic effects of AM in myometrium and that further evaluation of the properties of the CL expressed in both normal and tumor endothelium in vivo may be essential before targeting this endothelial GPCR for antiangiogenic therapies.
Angiogenesis, or new blood vessel growth, is an essential process during tumor growth and development (1, 2). Despite some initial setback and negative clinical trial results, major progress has been made over the past few years in targeting tumor angiogenesis. Several clinical studies using inhibitors of key angiogenic factors, such as vascular endothelial growth factor (VEGF)-A, not only validated the notion that angiogenesis is an important target for cancer but also revealed the phenomenon of the resistance of some tumors to antiangiogenic therapy (3, 4). There is now evidence that the effects of key angiogenic factors may be replaced by other pathways as the disease progresses or that other known as well as novel angiogenic molecules might be responsible for breakthrough angiogenesis in hypoxic tumor microenvironments during antiangiogenic strategies (5). In addition, potent tissue-specific angiogenic factors may be responsible for such resistance to antiangiogenic therapy (6). Therefore, the current challenge in translating angiogenesis research into the clinic is to identify differences in molecular mechanisms of angiogenesis between individual tumor types/stages by defining the role for the known and novel key as well as tissue-specific angiogenic factors. This would enable the selection of a specific antiangiogenic therapy from a range of those that are currently used in clinical trials or under development as well as to determine the likelihood of its effectiveness for an individual patient (5).
Uterine leiomyomas, also known as fibroids, are common benign solid tumors arising from the myometrium. They affect one third of adult women and are a significant cause of menorrhagia, pelvic pain, infertility, and pregnancy loss (7). Currently, the most effective treatments for fibroids are surgical and they are the commonest indication for hysterectomy. It is currently believed that the local aberrant angiogenesis is essential for development and growth of leiomyomas (7). The expression of various angiogenic factors, including VEGF, basic fibroblast growth factor, platelet-derived growth factor, and adrenomedullin (AM), has been recently documented in leiomyomas, and their role in tumor growth has been suggested (8, 9). However, AM is the only angiogenic factor to directly correlate with the increase in both vascular density and endothelial cell proliferation index in fibroids and therefore might play a key angiogenic role in the pathogenesis of this tumor (9).
AM was originally identified as a hypotensive peptide and then shown to be an endothelial antiapoptotic, growth, and angiogenic factor (10, 11). It belongs to the calcitonin family of peptide hormones, which comprises six known members [calcitonin, amylin, two calcitonin gene–related peptides (CGRP-α and CGRP-β), AM, and intermedin; ref. 12]. AM expression and secretion is regulated by inflammatory cytokines (e.g., interleukin-1 and tumor necrosis factor α and β modulated by estrogen and progesterone in the uterus) and hypoxia (11). Hypoxia is a frequent feature of the microenvironment in solid tumors and constitutes one of the driving forces of cancer growth and progression (13), and a role for AM as a promoter of these processes via induction of angiogenesis has been suggested for various tumors, including leiomyomas (9, 14–16). However, previous studies focused on analyzing AM expression in normal versus leiomyoma-bearing uteri, and no comparisons have been made between fibroid and myometrial tissues within leiomyoma-bearing uteri themselves. In addition, no studies have been done to evaluate the expression of AM receptors in vivo within leiomyoma or any other tumors, where AM expression is up-regulated (reviewed in ref. 17).
Effects of AM are mediated via heterodimeric receptors composed of calcitonin receptor–like receptor (CRLR, now known as CL) and one of the three receptor activity modifying proteins (RAMP; ref. 18). CL belongs to the family B of seven-transmembrane G protein–coupled receptors (GPCR). The RAMP family comprises three members (RAMP1, RAMP2, and RAMP3) that share <30% sequence identity but a common topological organization (reviewed in ref. 19). RAMPs are essential for terminal glycosylation, cell surface targeting, and ligand-binding selectivity of CL (18). The glycosylation state of the CL receptor is crucial to its properties. Only mature, fully glycosylated CL species are expressed at the cell surface and recognized by ligands (20). Therefore, mature CL glycoprotein is the functional (biologically active) form of this GPCR. Ligand-binding selectivity of the mature CL glycoprotein is defined by the formation of CL/RAMP heterodimers at the cell surface (18). RAMP1 promotes the expression of CGRP receptor, whereas coexpression of RAMP2 or RAMP3 with CL leads to the formation of AM receptors, termed AM1 and AM2, respectively (12, 18).
Based on CL/RAMP cotransfectant models as reported above, it has been proposed that the known effects of AM on vascular cells are mediated via endogenous CL receptor in a similar manner (21). However, only recently, experimental evidence from in vitro models confirmed that endogenous GPCR CL expressed in human endothelial cells is essential in AM-induced angiogenesis (22, 23). Endothelial CL is fully glycosylated (biologically active) and interacts with AM, playing a role as a functional AM-sensitive receptor in vitro (23). Anti-human CL (hCL) antibodies have been successfully used to block AM-induced endothelial migration and tube formation in vitro (22). In vivo, CL is predominantly expressed on endothelium (23–25) and therefore belongs to a unique group of few other recently discovered endothelial cell–specific GPCRs (6, 26). Diverse members of the GPCR superfamily participate in a variety of physiologic functions and are major targets of pharmaceutical drugs due to their often tissue-specific or cell type–specific expression (27). It follows that those GPCRs that show an endothelial phenotype and mediate the effects of angiogenic factors might represent attractive targets for antiangiogenic therapies (6, 28). However, no studies have been done to evaluate the expression, functional state, and localization of the CL in human tumors or to determine its role in myometrial endothelial cell biology as well as the potential for antiangiogenic therapy of fibroids (reviewed in ref. 17).
The present study was undertaken to clarify the role for AM and its receptor in leiomyoma angiogenesis as well as to test the hypothesis that CL could be a novel target for noninvasive antiangiogenic therapy of these benign but clinically problematic uterine tumors.
Materials and Methods
Source of tissue
Uteri were obtained by hysterectomy from nonpregnant women with fibroids ages 30 to 49 years. All had regular menstrual cycles (26-30 days) and had used neither oral nor intrauterine contraception nor had received any hormonal treatment for at least 6 months before. No other pelvic pathology was seen at surgery, which was confirmed by histologic examination done by an independent pathologist.
Uteri were bisected: one half was used for cell isolation and preparation of frozen samples and the other half was processed for histologic examination. The matched pairs of leiomyoma and adjacent myometrial tissue samples (N = 12) from leiomyoma-bearing uteri from individual patients (i.e., “patient-matched” samples) were snap frozen in liquid nitrogen. The Central Oxfordshire Research Ethics Committee (C00.147) approved the study.
Construction of hCL cDNA vector and transient expression of the receptor
The human endothelial CL cDNA was cloned into the pcDNA3.1 expression vector and termed hCLpcDNA (23). Myc-RAMP2 cloned into pcDNA3.1 was a gift from D. Poyner (Aston University, Birmingham, United Kingdom) and was originally prepared by Dr. S. Foord and coworkers (GlaxoSmithKline, Stevenage, United Kingdom).
DNA (1 μg) per plasmid was transfected for each T-75 cm2 flask containing 50% to 60% confluent HEK293T cells. Cells were transfected with hCLpcDNA alone or cotransfected with RAMP2 cDNA and then left for 24 hours before collecting protein lysates (as described below) for immunoblotting analysis solely for the purpose of validation of specificity of our own polyclonal antibody in the present study (see below).
Rabbit polyclonal antibody LN-1436 was raised against synthetic peptide corresponding to residues 427 to 461 (HDIENVLLKPENLYN) at the extreme COOH terminus of hCL protein (accession nos. AAC41994 and AAA62158; ref. 23).
Synthetic human AM was from Bachem (St. Helens, Merseyside, United Kingdom).
Deglycosylation experiments were done according to the supplier's protocols (Roche, Lewes, United Kingdom) before SDS-PAGE and immunoblotting.
SDS-PAGE and immunoblotting
Protein lysates from cell lines and tissues were obtained and subjected to SDS-PAGE and immunoblotting as described previously (23, 24). The density of bands was analyzed on an AlphaImager 1220 documentation and analysis system version 5.5 using linked background subtraction.
RNA isolation, reverse transcription-PCR, and Northern blotting
RNA isolation, reverse transcription-PCR (RT-PCR), and Northern blotting were done as described previously (24, 28).
Reverse transcription-PCR. Amplifications were routinely done using 25 to 30 cycles in the Perkin-Elmer (Beaconsfield, United Kingdom) GeneAmp PCR System 2400 for β-actin control, AM, CL, and RAMPs using primers designed for their specificity and spanning neighboring exons to enable the detection of a possible genomic DNA contaminations (Table 1).
|Gene .||Accession no. .||Primers (5′-3′) forward .||Primers (5′-3′) reverse .||Sequence .||PCR fragment size .|
|exon 3||exon 4|
|exon 12||exon 13|
|exon 2||exon 3|
|exon 3||exon 4|
|exon 2||exon 3|
|exon 4||exon 5|
|Gene .||Accession no. .||Primers (5′-3′) forward .||Primers (5′-3′) reverse .||Sequence .||PCR fragment size .|
|exon 3||exon 4|
|exon 12||exon 13|
|exon 2||exon 3|
|exon 3||exon 4|
|exon 2||exon 3|
|exon 4||exon 5|
Northern blotting. To generate the probes, full-length hCL, RAMP1, and RAMP2 cDNAs were RT-PCR amplified and cloned into pcDNA3.1 vector and full-length RAMP3 and 410 bp PCR-amplified fragment of AM cDNAs into TOPO vector. All resulting vectors were sequenced using an Applied Biosystems (Warrington, United Kingdom) 377 Genetic analyzer, and sequences were checked against the Genbank database. Ubiquitin probe was from BD Biosciences (Oxford, United Kingdom). Inserts were excised with restriction enzymes and labeled with [32P]dCTP using MegaPrime labeling kit (Amersham, Little Chalfont, United Kingdom). After hybridization and stringent washes as described previously (24), the blots were exposed to Hyperfilm (Amersham) and then to Phosphoscreen. Exposure was monitored to avoid saturation of the signal. The hybridization signals were further analyzed using ImageQuant software and linked background subtraction.
Endothelial cell isolation and culture
The method used in the present study was based on that developed for endometrial endothelium (Supplementary Fig. S1; ref. 29). In brief, positive selection of myometrial microvascular endothelial cells (MMVEC) was done using lectin Ulex europaeus agglutinin-1 (UEA-1) covalently bound to tosylactivated ‘Dynabeads’ M-450 (Dynabeads, Dynal, Norway). Purified endothelial cells were grown in EGM-2MV Bullet kit medium (BioWhittaker, Wokingham, United Kingdom). Second round of positive selection using UEA-1-coated beads was done after first passage to remove remaining contaminating myometrial myocytes and vascular smooth muscle cells.
In vitro angiogenesis assays
In vitro angiogenesis assays included endothelial cell proliferation, network formation, and migration.
Endothelial cell Matrigel network formation assay. For the Matrigel network formation assay, each well of a 24-well Falcon (Fisher Scientific, Loughborough, United Kingdom) tissue culture plate was evenly coated with 150 μL Matrigel (BD Biosciences). MMVECs (1 × 104 to 5 × 104) were seeded in triplicate per well in full EGM-2MV medium. Network formation was accessed after 24 hours by photographing the matrices using Zeiss (Welwyn Garden City, United Kingdom) light microscope and Nikon (Kingston Upon Thames, United Kingdom) CoolPix 990 digital camera.
Cultured cells and cryostat sections. Cultured MMVECs and 7 to 8 μm cryostat sections of myometrial and fibroid tissues were prepared and processed for immunofluorescence as described previously (23, 29).
Antibodies. Cell type phenotyping was done with monoclonal anti-CD34 (endothelial cell marker; Qbend 10, Novocastra, Newcastle upon Tyne, United Kingdom), anti-CD31 (endothelial cell marker), anti–smooth muscle actin (smooth muscle cell marker), anti-CD45 (peripheral blood mononuclear cell marker), and anti-CD68 (macrophage marker) antibodies and polyclonal antibody against von Willebrand factor (endothelial cell marker; all from DAKO, Ely, United Kingdom). Intracellular structures were identified using monoclonal anti-GM130 (Golgi matrix protein of 130 kDa), anti–early endosome–associated antigen 1, anti-calnexin (endoplasmic reticulum marker), and anti-CD107a (lysosome-associated membrane protein 1; all from BD Biosciences PharMingen, Oxford, United Kingdom) antibodies. Secondary antibodies were Texas red–conjugated and FITC-conjugated horse anti-mouse and horse anti-rabbit IgG (all from Vector Laboratories, Peterborough, United Kingdom).
Double immunofluorescence. Double immunofluorescence was done as described previously using anti-CL antibody LN-1436 and markers of individual cell types (as detailed in the “Antibodies”; refs. 23, 29). Visualization was carried out with the use of a Leitz Diaplan or Leitz DMRBE microscopes (Leica, Milton Keynes, United Kingdom), a Hamamatsu Orca C4742-95 digital camera, and OpenLab software (both from Improvision, Coventry, United Kingdom).
CD34/Mib1 double immunohistochemistry. Immunohistochemical staining for detection of proliferating endothelial cells was done using monoclonal anti-CD34 (endothelial cell–specific marker) and anti-Ki-67 (marker of proliferating nuclei) antibodies [Qbend 10 and Mib1 (DAKO)] essentially as described previously (9).
Determination of vascular density and endothelial cell proliferation index
Vascular density was determined by Chalkley counting and endothelial cell proliferative index as described previously (9).
Human MMVECs were supplemented with 0.5% FCS-EGM-2 medium for 16 hours. Cells were then treated with or without 100 nmol/L AM in 0.5% FCS-EGM-2 medium for 30 minutes. Internalization was accessed by immunofluorescence as described above using anti-hCL antibody and cell surface, early endosome, and lysosome markers.
Data on microvascular density, immunoblotting ratio signals, and mRNA expression levels in patient-matched samples of myometrial and leiomyoma tissues were analyzed by Wilcoxon signed rank test. Ps < 0.05 were considered significant.
Functional CL receptor is expressed in endothelium in leiomyoma
Mature glycosylated receptor is the predominant CL form in fibroids. We have currently raised and characterized an anti-hCL polyclonal antibody LN-1436 (23). The antibody specifically recognizes unglycosylated, core-glycosylated, and mature glycosylated forms of CL receptor as shown using hCL-transfected or empty vector–transfected cell lines and deglycosylation assays (23). In the present study, we used this antibody to examine which forms, and if the mature CL glycoprotein (functional form of the receptor) in particular, are present in lysates from fibroid and adjacent myometrial tissues from leiomyoma-bearing uteri. We also did deglycosylation experiments to confirm the glycosylation status of the receptor.
Several distinct bands were observed in untreated myometrial and fibroid samples (Fig. 1B, lanes 1 and 4). Endoglycosidases F and H were used to differentiate between mature and core N-linked CL glycoproteins. The ∼55 kDa (strong band) form was reduced to a 37 kDa form by treatment with endoglycosidase F, showing that the additional mass units represent carbohydrate residues (Fig. 1B, lanes 2 and 5). However, this form is resistant to endoglycosidase H (Fig. 1B, lanes 5 and 6), indicating that CL has been terminally glycosylated, an event normally associated with transit through the Golgi complex and the production of mature glycoproteins. The majority of CL glycoprotein is a mature glycosylated and not core-glycosylated form (Fig. 1B; see Fig. 3B for comparison, representing the immunoblot of deglycosylation experiments done on lysates from MMVECs, where all three forms of CL could be detected). Thus, unglycosylated and mature CL glycoproteins are the predominant forms of the receptor in myometrial and fibroid tissues. It follows that the presence of the mature CL glycoprotein in these tissues suggests the functional state of the receptor expressed in vivo (20, 23).
CL is predominantly expressed by microvascular endothelium in vivo in myometrium and fibroid. We then examined CL localization in fibroid and adjacent myometrial samples by double immunofluorescence. In myometrium, CL was localized in microvascular endothelium but was absent in mural cells (vascular smooth muscle cells/pericytes) and leukocytes (Fig. 1C). A similar pattern was observed in fibroids (Fig. 1C). Thus, CL is expressed exclusively by endothelium in both fibroids and surrounding myometrium. It follows that functional CL exhibits endothelial-specific pattern of expression in vivo.
Myometrial endothelial cells express functional CL receptor that interacts with AM in vitro
We then did in vitro studies to analyze the functionality of the CL expressed in MMVECs. We isolated primary MMVECs and examined the glycosylation status of the CL and also investigated whether AM interacts with this endogenously expressed receptor and has angiogenic effects on these cells in vitro.
Isolation and characterization of primary human MMVECs. Initially, we have isolated and characterized primary MMVECs. We used UEA-1-coated magnetic beads because, similarly to CL, this lectin is expressed specifically in myometrial endothelium in vivo (Fig. 2A). We obtained pure populations of MMVECs (beads positive) and myometrial smooth muscle cells (beads negative) as shown by double immunofluorescence using antibodies against MMVEC-specific and myometrial smooth muscle cell–specific markers von Willebrand factor and α-smooth muscle actin, respectively (Supplementary Fig. S1). Primary myometrial endothelial cells formed typical “cobblestone” morphology and “capillary-like” structures when seeded on Matrigel (Fig. 2B) and retained their characteristics in culture as shown by analysis of the expression of endothelial cell–specific markers von Willebrand factor (by immunofluorescence; Supplementary Fig. S1), CD31 (by immunoblotting; Fig. 2C), VEGF receptor 2, and Delta4 (by RT-PCR; Fig. 2D; refs. 30, 31).
Mature and core-glycosylated CL forms are expressed in MMVECs in vitro. We then investigated whether primary MMVECs maintained the expression of the mature CL glycoprotein (the functional form of the receptor) in culture. We found that both core and mature CL glycoproteins are present in isolated MMVECs (Fig. 3A; compare with the expression of only mature CL glycoprotein in tissues, Fig. 1B). The amount of both forms was comparable as determined by quantification of CL/β-actin ratio.
Primary MMVECs continue to express CL and other components of the AM receptor system in vitro. We then investigated whether primary MMVECs maintained the expression of components of AM receptor system (i.e., CL and RAMPs) as well as AM itself compared with the in vivo expression of these molecules (Fig. 3B). Both primary MMVECs and myocytes express AM mRNA in vitro (Fig. 3B). CL and RAMP2, but not RAMP1 or RAMP3 mRNA, are expressed in MMVECs (Fig. 3B). CL mRNA was virtually absent in myocytes in vitro (Fig. 3B), which is consistent with the immunofluorescence findings on endothelium-specific distribution of CL protein in myometrium (Fig. 1C). Myocytes express RAMP1 and RAMP2 but not RAMP3 mRNA. All receptor components (CL and three RAMPs) were expressed in tissues (Fig. 3B).
Presence of the terminally glycosylated CL form in MMVECs suggests its cell surface expression and the presence of functional endogenous AM receptors. We have examined this by doing subcellular localization, proliferation, and migration studies.
Correlation with cell surface expression. Presence of both terminally and core-glycosylated forms of CL suggests the cell surface and intracellular localization of this GPCR in MMVECs. This is because, when RAMP expression is insufficient, core-glycosylated CL does not undergo terminal glycosylation and is retained in the endoplasmic reticulum and not transported to the cell surface (18). We analyzed the subcellular localization of CL in MMVECs by immunofluorescence.
The majority of MMVECs displayed visible staining concentrated in a perinuclear region, where CL protein is localized in the endoplasmic reticulum but not in Golgi apparatus (Supplementary Fig. S2). Although absence or little surface staining was shown by lack of colocalization with the plasma membrane marker CD31 (Supplementary Fig. S2), further internalization studies confirmed that significant part of the receptor is presented at the cell surface (as described below). Thus, our findings show both cell surface and intracellular localization of the CL in MMVECs.
Correlation with the presence of functional endogenous AM-sensitive receptors. We then investigated whether the expression of mature CL glycoprotein correlates with the presence of functional endogenous AM receptors in MMVECs. We therefore did ligand-induced proliferation and migration studies in vitro. Fully active endogenous AM receptors were found in human MMVECs in vitro as shown by comparable magnitude of an agonist-mediated proliferation (Fig. 4A) and migration (Fig. 4B).
AM interacts with endogenous MMVECs surface-expressed CL and induces its internalization. The presence of fully active endogenous receptors for AM as well as RAMP2 mRNA expression in MMVECs suggests that endogenous CL might be involved in generating AM-sensitive receptors via formation of CL/RAMP2 heterodimers, which are known to interact with AM (20, 32). We tested whether MMVEC-expressed CL interacts with AM by analyzing the dynamics of receptor internalization in response to AM (10 nmol/L to μmol). Agonist-mediated internalization of surface-expressed endogenous CL was observed in response to AM (Fig. 4C and Supplementary Fig. S3). A significant proportion of internalized CL was targeted to early sorting endosomes and not lysosomes on exposure to the ligand. Localization of endoplasmic reticulum–associated portion of the receptor in perinuclear region remained unaltered (Fig. 4C and Supplementary Fig. S3). These findings show that mature CL glycoprotein expressed in myometrial endothelium interacts with its ligand angiogenic factor AM.
Functional CL receptor is down-regulated and its expression is concurrent with alterations in RAMP2 mRNA levels and microvascular density in fibroids
Expression of mature CL glycoprotein is down-regulated in fibroids. Endothelial-specific expression of the receptors for angiogenic factors is often a prerequisite for successful antiangiogenic therapies (26, 33, 34). Ideally, these receptors should be expressed at higher levels in tumor vasculature and lower, or even undetectable, levels in normal vasculature. The presence of differentially expressed receptors would enable selective targeting of fibroid vasculature. We therefore analyzed the expression levels of the functional form of the CL in leiomyoma and surrounding myometrium by immunoblotting. We found that the mature glycoprotein (functional receptor) levels were significantly down-regulated in fibroids (P < 0.01; Fig. 5A).
Concurrence with alterations in RAMP expression. Because expression of mature CL glycoprotein was down-regulated in fibroids and because RAMPs are essential for terminal glycosylation, cell surface targeting, and ligand-binding selectivity of this GPCR (18), we studied the expression of RAMPs. Unfortunately no anti-human RAMP antibodies that reliably distinguish their isoforms (1, 2, and 3) or show expression of endogenous monomers (which should be observed in case of a heterodimeric CL/RAMP receptor) of expected molecular weight are available. Therefore, we studied RAMP mRNA expression by Northern blotting. Only RAMP2 mRNA expression was significantly altered (P < 0.05) in fibroids compared with myometrium (Fig. 5B). AM and RAMP1 were unchanged in fibroids (Fig. 5B). RAMP3 mRNA levels were below the Northern blotting detection limit but could be found in normal myometrium and fibroids by RT-PCR (see Fig. 3B for details). Thus, down-regulation of the functional CL is concurrent with the reduction in RAMP2 mRNA expression in fibroids.
Concurrence with alterations in microvascular density. Because GPCR CL was localized in microvascular endothelium in both fibroids and surrounding myometrium (Fig. 1C), we examined microvessel density in both tissues (Fig. 5C). The vascular density of leiomyoma tissue in leiomyoma-bearing uteri was significantly lower than in neighboring myometrial tissue (P < 0.05; Fig. 5C). The endothelial proliferation index was unaltered in leiomyoma (data not shown). Thus, down-regulation of the functional form of CL receptor is concurrent with the reduction of the microvascular density in leiomyoma.
AM is a potent angiogenic factor. Several studies used animal models to show that AM overexpression in xenografted tumors enhances tumor growth via angiogenesis, whereas blocking AM antibodies have the reverse effect (35). Although these observations point to a potential role for AM and its receptors in tumor angiogenesis, their role in human cancer is as yet poorly defined (reviewed in ref. 17). The lack of information about the distribution, regulation, and function of endogenous AM receptors in tumors makes it difficult to assess the potential for anti-AM strategies for noninvasive antiangiogenic therapy and its possible effects on surrounding tissue (reviewed in ref. 17). Here, we present the first report on expression of the functional form of GPCR CL in human tumors.
We investigated the expression and localization of GPCR CL in uterine leiomyomas, a common solid tumor arising from the myometrium, where a role for AM in angiogenesis has been suggested based on correlation of its expression with tumor vascularization alone (9). Here, we used both in vitro and in vivo approaches to clarify the role for both AM and its receptor CL in fibroid angiogenesis. First, we studied the distribution and functional state of the CL in leiomyomas. Second, we examined if AM has an angiogenic effect on microvascular endothelial cells isolated from a specific microvascular bed, where we found endothelial-specific expression of the CL and where a role for AM in aberrant angiogenesis has been recently proposed (i.e., from myometrium). This is because the endothelium from different vascular beds shows a differential response to angiogenic factors due to the variations in expression of their receptors (36). Third, we examined whether CL expressed by MMVECs is functional (i.e., interacts with AM) by doing ligand-mediated receptor internalization studies and analyzed which RAMPs might form heterodimers with CL. Finally, we analyzed expression levels of mature CL glycoprotein (functional form of the receptor) in vivo in leiomyomas and compared this with the expression levels of RAMPs and microvascular density in this tumor.
To our knowledge, there are no reports on expression of the functional form of the CL receptor in human tumors. McLatchie et al. (18) showed that terminal (mature) glycosylation of CL receptor requires expression of RAMPs by using overexpression cell models. This form of CL is considered to be biologically active (i.e., interacting with ligands; ref. 20). We found that mature CL glycoprotein is expressed in leiomyomas. Thus, we were able to access the functional status of CL directly in human tumor even without evaluating the expression of translated RAMPs. In contrast, the presence of the mature CL glycoprotein in tissues could be used to show indirectly the expression of these accessory proteins. To our knowledge, our study is the first to introduce the methodologic approach that enables to access directly both the presence of endogenous RAMP proteins and to detect the levels of the functional CL receptor in human tumors.
In vivo, the mature CL glycoprotein was localized exclusively in vascular endothelium in both myometrium and fibroids. Although our report is the first to show the expression and distribution of this functional form of the CL receptor in human tumor tissue, more studies are required to investigate the distribution and the role for this GPCR in other tumor types and cancers. It could not be excluded that, in other tumors, the expression pattern of functional form of the CL is different to the endothelial-specific expression observed in leiomyoma, and it should be expected that CL might be also expressed on tumor cells.6
L.L. Nikitenko et al., unpublished observations.
Our findings suggest the essential role for GPCR CL in myometrial endothelial cell biology. The endothelial cell–specific expression of the functional form of the CL receptor in vivo in leiomyoma tissue, its internalization in response to agonist stimulation, and effects of AM on growth and migration of MMVECs in vitro concur with previous observations that elevated AM levels are associated with increased vascular density and endothelial proliferation index in leiomyoma, thus supporting the view that AM might play a role in aberrant angiogenesis in fibroids (9) by inducing an angiogenic response in the normal myometrial endothelium. However, our data also show that mature glycosylated CL levels are lower in fibroid tissue within leiomyoma-bearing uteri. We hypothesized that the observed decrease could be due to (a) reduced gene transcription (i.e., reduced CL mRNA expression), (b) decrease in expression of other components of the AM receptor system (e.g., RAMPs), (c) alterations in vascular density and/or number of endothelial cells, (d) chronic activation and degradation of the AM receptors in fibroids, and (e) possible alterations in the myometrial vasculature in the presence of leiomyomas (9). We therefore analyzed these possibilities in more detail.
Our findings suggest that reduced hCL gene transcription in leiomyomas is unlikely to account for the reduction in mature glycoprotein production. In contrast, observed down-regulation of RAMP2 mRNA expression in fibroids could possibly account for decreased levels of mature CL glycoprotein because RAMPs are known to regulate CL glycosylation (18). Furthermore, both reduced terminally glycosylated CL and RAMP2 mRNA levels could be due to the observed decrease in vascular density. This is because CL is expressed exclusively in endothelium in both myometrium and fibroids and also because MMVECs express only RAMP2 mRNA in vitro. The reduction of microvascular density in fibroids might be due to the immature state of blood vessels commonly observed in dividing tumors or due to the higher content of fibrous connective tissue, which is avascular (37). Hague et al. (9) reported that the vascular density of leiomyomas is comparable with that of normal myometrium and that vascular density of myometrium was higher in leiomyoma-bearing uteri compared with normal controls. Thus, although no direct comparison was made between myometrium and fibroid within the same uterus, the data presented by Hague et al. (9) concur with our findings and several other reports showing decrease in microvascular density in fibroid tissue within leiomyoma-bearing uteri (38, 39). Finally, it could not be excluded that mature CL glycoprotein levels might be reduced as a result of increased ligand-induced activation and subsequent internalization and degradation of the receptor. This is because agonist-promoted internalization is common to a large number of GPCRs and is often related to the loss of receptor activity and its targeting to the degradation pathway (40). If this was the case, then an increase in either ligand expression or receptor degradation should be observed. Our Northern blotting data show no change in the levels of AM mRNA in fibroids compared with surrounding myometrium, and internalization studies showed that activated CL is not immediately targeted to lysosomes for degradation. Similarly, other research groups showed that the expression of VEGF, another angiogenic factor, is also unchanged in myometrial and fibroid tissues within the same uterus (8). Thus, our data suggest that the decreased levels of the mature CL glycoprotein in leiomyoma are due to the alterations in vascular density and/or number of endothelial cells and not due to the alterations in the receptor expression in tumor endothelium.
Because reduced levels of endothelial mature CL glycoprotein in fibroids correlate only with RAMP2 mRNA expression, it is likely that CL/RAMP2 heterodimer (AM receptor), but not CL/RAMP1 (CGRP receptor), is present in vivo in endothelium in leiomyoma. This is supported by the presence of RAMP2, but not RAMP1 or RAMP3 mRNA, in isolated MMVECs. However, also our study shows that, in cultured MMVECs, there was an approximately equal proportion of the mature and core-glycosylated receptor forms and that CL was localized not only on the cell surface but also in the endoplasmic reticulum in MMVECs. When RAMP expression is insufficient, core-glycosylated CL is retained in endoplasmic reticulum and not transported to the cell surface (20). Therefore, it is possible that RAMP expression is altered in MMVECs in vitro. It follows that in vivo RAMP3 may also play a role in the production of a functional heterodimeric AM receptor (AM2; ref. 12) in endothelium by facilitating the glycosylation of the remaining proportion of the CL. This view is supported by our finding on expression of the mRNA for this accessory protein in myometrial tissue. The role for RAMP1 in this process is unlikely because RAMP1 mRNA is predominantly expressed by myometrial myocytes in vivo as shown by our previous in situ hybridization studies (24). Thus, we conclude that it could not be excluded that both CL/RAMP2 (AM1 receptor) and CL/RAMP3 (AM2 receptor), but not CL/RAMP1 (CGRP receptor), heterodimers form a mixed in vivo–expressed pool of CL-associated receptors in microvascular endothelium in both myometrium and fibroids.
The presence of differentially expressed receptors would enable selective targeting of fibroids, and several studies attempted to identify these molecules (41, 42). However, no changes have been observed in gene expression profiles in endothelium from fibroid and myometrium within leiomyoma-bearing uteri as yet. The functional CL receptor (AM1 or AM2 receptor) expressed in fibroid endothelium in vivo retains its potential to be useful for targeting tumor vasculature even despite the relatively unaltered expression in leiomyoma vessels (when taking into account reduction in microvessel density in leiomyoma and endothelial-specific expression of the receptor) as well as unchanged expression of its ligand AM within the leiomyoma-bearing uterus. This is because cellular components other than just CL and RAMPs could be also important for the expressed phenotype and the properties of CL-associated receptor subtypes, such as magnitude of agonist-mediated responses and mechanisms of internalization and desensitization (reviewed in refs. 19, 43, 44). For example, the role for one of these molecules (i.e., receptor component protein) in AM signaling and in endothelial cells in female reproductive tract in particular has been recently suggested (45, 46). In addition, it has been shown that the GPCR kinases play a pivotal role in desensitization of GPCRs without loss of their cell surface expression (47). Recently, the role for GPCR kinases in GPCR-mediated signaling in endothelial cells (48) as well as CL-mediated signaling has been suggested (49). Finally, the internalization and trafficking of the AM receptors can be regulated by the expression of Na+/H+ exchanger regulatory factor-1 (44). Findings from these studies suggest that whether mature CL glycoprotein (CL/RAMP2 or CL/RAMP3 complex) expressed in fibroid vasculature is functionally “more efficient” or has higher turnover/recycling rate compared with surrounding myometrial vasculature remains to be investigated. The detailed preclinical expression studies for these components of AM signaling system and the modulators of CL expression and function in particular would provide such information in the future. The outcome of such studies would be essential to enable further evaluation of the potential for targeting endothelial GPCR CL for antiangiogenic therapies before the use of respective inhibitors in clinical trials to avoid adverse effect on normal tissues within the tumor-bearing organ.
In summary, we present the first report showing the expression of the functional form of GPCR CL in human tumors. Our findings suggest the role for CL in mediating angiogenic effects of AM and that the endothelial-specific phenotype determines the alterations in the expression of the functional form of the CL in uterine leiomyoma. We conclude that further evaluation of the properties of the CL in normal and tumor endothelium is essential before targeting this endothelial GPCR for antiangiogenic therapies in uterine leiomyoma and possibly other tumors.
Grant support: The Wellcome Trust (L.L. Nikitenko and M.C.P. Rees), Medical Research Fund, University of Oxford, United Kingdom (L.L. Nikitenko), Cancer Research UK (L. Campo, H. Turley, R. Leek, and R. Bicknell), and Royal Society UK Travel Award (L.L. Nikitenko).
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
T. Cross is a Final Honour School student from St. Anne's College, Oxford.
We thank Robin Roberts-Gant for assistance with formatting figures and Drs. Veronica Carroll and Cecilia Lai for the critical reading of the article.