|Home | About | Journals | Submit | Contact Us | Français|
Endothelin-1 (ET-1) and its receptors are overexpressed in human cancers, but much less is known about the roles of ET-2 and ET-3 in cancer etiology. We sought to examine human and rat colon tumors for dysregulation of ET-2 and ET3 expression, and determine the underlying mechanisms. Human primary colon cancers and carcinogen-induced rat colon tumors were subjected to real-time RT-PCR, immunoblotting, and immunohistochemistry; EDN2 and EDN3 genes were examined by methylation-specific PCR, bisulfite sequencing, and pyrosequencing; and forced expression of ET-2 and ET-3 was conducted in human colon cancer cells followed by real-time cell migration and invasion assays. Rat and human colon tumors had markedly reduced expression of ET-2 and ET-3 mRNA and protein compared with matched controls. Mechanistic studies revealed hypermethylation of EDN2 and EDN3 genes in human primary colon cancers, and in a panel of human colon cancer cell lines. Forced expression of ET-2 and ET-3 attenuated significantly the migration and invasion of human colon cancer cells. We conclude that epigenetic inactivation of ET-2 and ET-3 occurs frequently in both rat and human colon cancers. Current therapeutic strategies target overexpressed members of the ET axis via small molecule inhibitors and receptor antagonists, but this work supports a complementary approach based on the re-expression of ET-2 and ET-3 as natural antagonists of ET-1 in colon cancer.
There is growing interest in the endothelin (ET) axis and its dysregulation in cancer development1-3. This “axis” comprises the 21-amino acid peptides ET-1, ET-2, and ET-3, the G-protein-coupled receptors ETAR and ETBR, and upstream processing enzymes such as endothelin converting enzyme 11-3. ET-1 is by far the best characterized isoform, being a potent vasoconstrictor with key roles in normal physiology and in various pathophysiological conditions4,5. ET-2 and ET-3 differ from ET-1 by two and six amino acids, respectively, and undergo similar processing of the precursor peptides (preproETs). Overexpression of ET-1, ETAR, and ETBR has been documented in various human cancers, but much less is known about the roles of ET-2 and ET-3.
Whereas ET-1 is expressed in many tissue types, ET-2 is restricted mainly to the gastrointestinal tract, sex organs, and pituitary gland. ET-3 is more ubiquitous, and has been detected in brain, kidney, lung, spleen, stomach, and intestine. In human breast cancer cells, ET-2 acts as a hypoxia-induced autocrine survival factor with a potential role in tumor invasion6,7. ET-3 was reported to be decreased in cervical cancer compared with normal cervical epithelial cells8, and loss of ET-3 gene (EDN3) expression was observed in human breast cancer9.
Given this background information, we sought to examine the roles of ET-2 and ET-3 in colon cancer development. In addition to studies of primary human colon cancers and colon cancer cell lines, a preclinical rat model was used to examine both early and late stages of colon carcinogenesis. Mechanistic work focused on epigenetic inactivation of EDN2 and EDN3, coupled with experiments involving forced expression of ET-2 and ET-3 in human colon cancer cells.
Thirty primary human colon cancers and patient-matched controls were provided under an IRB-approved protocol by Steven F. Moss and Lelia Simao (Rhode Island Hospital, Providence, RI). The tumors were characterized as late-stage adenocarcinomas and were the primary source for the molecular studies. In a separate but parallel arm of the investigation, immunohistochemical analyses of ET-2 and ET-3 were performed on tissue slides of human normal colon (ab4327, Abcam) and on human colon cancer tissue microarrays (TMAs).10 Rat tumors and adjacent normal looking tissues were from a 1-year study10 in which male F344 rats were treated with 2-amino-1-methyl-6-phenylimidazo[4,5- b]pyridine, a heterocyclic amine from cooked meat. An interim sacrifice was included in order to assess early molecular changes before tumor onset. Specifically, 24 h after the last dose of carcinogen, rats (5-6/group) were euthanized, the colon was removed and opened longitudinally, and the mucosa was scraped and frozen in liquid nitrogen. The study received prior approval from the Institutional Animal Care and Use Committee.
Frozen samples of tumor and normal tissue were thawed and the mRNA was extracted using the RNeasy kit (Qiagen, Valencia, CA). RNA (2 μg) was reverse-transcribed in 20 μl using the SuperScriptR III first-strand synthesis system for RT-PCR (Invitrogen). EDN2 and EDN3 mRNA levels were measured by qPCR and normalized to glyceraldehyde-3- phosphate dehydrogenase (GAPDH). Forty cycles of PCR (95°C/10s, 58°C/10s, 72°C/10s) were run on a LightCycler 480 II system (Roche, Indianapolis), in a 20-μl total reaction volume containing cDNAs, SYBR Green I dye (Roche), and primers. The amount of specific mRNA was quantified by determining the point at which the fluorescence accumulation entered the exponential phase (Ct), and the Ct ratio of the target gene to GAPDH was calculated for each sample. At least three separate experiments were performed for each sample.
Proteins (20 μg) were subjected to SDS-PAGE and immunoblotted according to reported methods10. Primary monoclonal mouse antibodies were to ET-2 (1:500 dilution, Novus Biologicals, H00001907-MO1) and ET-3 (1:500 dilution, Novus, H00001908-MO1). Proteins were visualized by Western Lightning chemiluminescence Reagent Plus (PerkinElmer Life Sciences, Boston MA, USA) and quantified using an Alpha Innotech Image System (Alpha Innotech, San Leandro, CA). For immunohistochemistry studies, tissues were sectioned at 4-5 μm and processed in a Dako Autostainer. Polyclonal rabbit anti-ET-2 (Atlas, Sweden, Lot# 27432) was used at 1:50 dilution (human tissues) or 1:25 dilution (rat tissues). Polyclonal rabbit anti-ET3 (Phoenix Pharmaceuticals, Cat# H-023-17) was used at 1:200 dilution. Primary antibodies were applied for 0.5 h at room temperature. Dako Universal negative rabbit (N1699) served as a negative control. MaxPoly-One Polymer HRP Rabbit Detection solution (MaxVision Biosciences) was applied at room temperature, followed by Nova Red (Vector Laboratory SK-4800) as the chromagen and Dako hematoxylin (S3302) as the counterstain.
Genomic DNA (0.5 μg) was bisulphite-modified using the EZ DNA methylation kit (Zymo Research Corp, Orange County, CA) and eluted in 20 μl Tris-Buffer (10 mM). Quantitative methylation-specific PCR (MSP) was performed with primers that specifically recognized unmethylated or methylated EDN2 and EDN3 promoter sequence after bisulphite conversion. Primer sequences were designed by MethPrimer software. Forty cycles of PCR were run on a LC480 LightCycler (Roche) in a 10-μl reaction containing bisulphite-treated DNA, SYBR Green I dye (Roche), and primer set. PCR conditions were 95 °C /10s, 58 °C /10s, and 72 °C /10s. DNA was quantified by determining the point at which the fluorescence accumulation entered the exponential phase (Ct), and the Ct ratio of target to β-actin was calculated for each sample. At least two independent experiments were performed per sample. For pyrosequencing, primers were designed with PyroMark Software SW2.0 (Qiagen), and targeted EDN2 (−198 to −60) or EDN3 (−250 to −77). Forty-five cycles of PCR were run on a GeneAmp system 9700 (Applied Biosystems) in a 20-μl reaction containing PyroMark PCR Master Mix (Qiagen), bisulphite-treated DNA, and EDN2 or EDN3 primer set. After confirming the correct amplicon size, PCR products and sequencing primers were sent to the Protein and Nucleic Acid Facility (Stanford University) for pyrosequencing on a PyroMark Q24 system.
Human colorectal cancer lines Caco2, HCT116, HT29, SW48 and SW480 were obtained from American Type Culture Collection (ATCC, Manassas, VA). Caco2 cells were maintained in MEM medium (ATCC) with 20% heat-inactivated fetal bovine serum (FBS, Hyclone Laboratories). Other cell lines were maintained in McCoy’s 5A medium (Invitrogen) supplemented with 10% FBS, 100 units/ ml penicillin, and 100 μg/ml streptomycin at 37°C in 5% CO2. Cells typically were seeded at a density of 1×105 in a six-well plate format. 5′-Aza-2′-deoxycytidine (Sigma-Aldrich) was added to a final concentration of 10 μM, whereas controls were treated with the equivalent volume of PBS. Cells were harvested on day 4 for DNA methylation analyses, and on day 5 for quantification of mRNA expression. In ET-2 and ET-3 overexpression experiments, human EDN2 and EDN3 cDNA clones were obtained from OriGene Technologies (Rockville, MD). Cells (3 ×105) were seeded in 6-well plates overnight. The media was aspirated and replaced with fresh antibiotic free-transfection media containing 6 μl lipofectmine2000 and 0.5 μg EDN2 or EDN3 plasmid DNA, or 0.5 μg XL5 empty vector (OriGene). At 72 h after transfection, cells were harvested for protein analyses and cell migration/invasion assays. The latter were conducted in real time using the xCELLigence System (Roche). Colon cancer cells were transfected with EDN2, EDN3, or empty vector XL5, and harvested at 72 h. Cells (0.2 × 106) in 100 μl serum-free media were loaded onto the upper chamber of CIM plates coated with or without 2.5% matrigel (BD Bioscience, San Diego, CA), to assess invasion and migration, respectively. The lower chamber was filled with MEM medium containing 20% FBS. Measurements were taken at intervals of 10 min for at least 40 h.
Data were plotted as mean±SE and compared using Student’s t-test, or by Anova for group comparisons, unless stated otherwise. In the figures, significant outcomes were shown as follows: *P < 0.05, **P < 0.01, ***P < 0.001.
Human primary colon cancers and patient-matched controls were immunoblotted for preproETs (Fig. 1A). Colon tumors had markedly lower levels of preproET-2 and preproET-3, whereas no such loss of preproET-1 was detected. Based on these initial findings, EDN2 and EDN3 mRNA levels were determined by qPCR and normalized to GAPDH. Human primary colon cancers had significantly reduced expression of EDN2 and EDN3 compared with patient-matched controls (Fig. 1B). Specifically, EDN2 expression had a relative value of 4.27±1.38 in controls compared with 0.37±0.11 in the colon tumors (mean±SE, n = 30, P < 0.01), i.e., an 11-fold reduction overall. In the case of EDN3, the corresponding values were 8.42±1.89 versus 1.04±0.38 (mean±SE, n = 30, P < 0.001), i.e., 8-fold lower expression in the colon tumors.
Loss of preproET-2 and preproET-3 expression also was detected in carcinogen-induced colon tumors compared with adjacent normal-looking tissue (Fig. 1C), and in normal-looking colonic mucosa obtained several weeks before the onset of frank tumors (Fig. 1D). The relative expression of Edn2 mRNA in rat colon tumors was 0.02±0.01 (Fig. 1E, left panel, black bar), ~20-fold lower than in normal-looking tissue adjacent to tumor (0.40±0.06, P < 0.001), and colonic mucosa from untreated controls (0.43±0.06, P < 0.001). No significant difference was detected for Edn2 expression when comparing normal colon from untreated rats and normal-looking tissue adjacent to tumors (Fig. 1E, left panel, white versus grey bar, P>0.05). In contrast, Edn3 expression was significantly lower not only in colon tumors compared with adjacent normal-looking tissue (Fig. 1E, right panel, black versus grey bars, P<0.01), but also in adjacent normal-looking tissue compared with colonic mucosa from untreated rats (Fig. 1E, right panel, grey versus white bars, P <0.001). The corresponding relative mRNA expression levels were as follows: 0.26±0.03 (colon tumors) < 0.45±0.05 (adjacent normal-looking tissue) < 0.88±0.09 (control colonic mucosa).
Immunostaining for ET-2 (Fig. 2A) revealed intense expression in the propria mucosae of normal human colon tissue, mainly in macrophages, whereas neurons of the autonomous nerve plexus in the tunica muscularis were moderately stained and enterocytes of the lamina epithelialis were lightly stained. In human colon cancer TMAs, normal-looking areas stained positive in macrophages, whereas adenocarcinoma regions had little or no detectable ET-2. In rat normal colon, neurons of the autonomous nerve plexus and segments of the epithelial brush border were strongly stained for ET-2, whereas epithelial cells per se were faintly stained. Colon tumors in the rat were largely ET-2 negative, or exhibited faint background staining.
In the case of ET-3 (Fig. 2B), both human and rat normal colonic epithelium stained strongly, most notably in the basal and apical enterocytes within colonic crypts. ET-3 also was detected in enteric ganglion cells from both human and rat. In human colon cancer TMAs, most cores were negative, except a few that exhibited cytoplasmic punctate ET-3 staining. In the rat, sections with tumor (T) adjacent to normal (N) tissue typically revealed positive staining for ET-3 only in the latter regions (Fig. 2B, lower right, 40x magnification).
A search using MethPrimer software confirmed previous findings9 of two regions with high CpG density in EDN3, and revealed a putative CpG island in EDN2, located at position -5 to +142 relative to the translation initiation site (Fig. 3A). Using primers that flanked these regions, methylation-specific PCR (MSP) detected a 30-fold increase in the methylation of EDN2 in human primary colon cancers compared with patient-matched controls (Fig. 3B, left panel, 19.80±4.9 versus 0.64±0.08, n = 30, P < 0.001). MSP also revealed a ~2-fold increase in EDN3 methylation in cancers versus controls (Fig. 3B, right panel, 2.49±0.31 versus 1.40±0.15, n = 30, P < 0.001).
A panel of human colon cancer cell lines was examined by qPCR for EDN2 mRNA expression, normalized to GAPDH. After treatment with the DNA demethylating agent 5- aza-2′-deoxycytidine (5-aza), EDN2 expression was increased 7-fold in Caco-2 cells, 46- fold in HCT116 cells, 24-fold in HT29 cells, 23-fold in SW48 cells, and 6.5-fold in SW480 cells (Fig. 4A). Four days after treatment with 5-aza, MSP analyses revealed a shift from high to low EDN2 methylation status (Fig. 4B,C). In HCT116 cells, for example, 5-aza treatment decreased by ~7-fold the EDN2 methylated PCR product (MSP-M) and increased by 15-fold the corresponding unmethylated signal (MSP-U). Similar trends were observed for EDN3 in the panel of five colon cancer cell lines (data not shown). In Caco-2 cells, for example, EDN3 mRNA expression was increased 2-fold after 5-aza treatment, there was a corresponding loss of the methylated product in MSP analyses (P < 0.01), and a 2.8-fold increase was detected for the unmethylated product (P < 0.001).
Bisulfite sequencing in human colon cancer cells treated with and without 5-aza provided initial confirmation of the predicted methylation sites in EDN2 and EDN3 (data not shown), and pyrosequencing subsequently was used to generate quantitative data. In the case of EDN2, six separate CpG sites were shown to be hypermethylated in human colon cancer cells (Fig. 5A, grey bars). At site CpG1 for example, 100% methylation was reduced to 60% methylation after 5-aza treatment (P < 0.001, compare grey bar versus solid black bar). Similar results were obtained for seven CpG sites in EDN3 (Fig. 5B). For example, at site CpG1, 85% methylation was reduced to <50% methylation by 5-aza (P < 0.001).
Human colon cancer cells were transiently transfected with constructs expressing ET-2 or ET-3. Empty vector XL5 served as the control. Immunoblotting confirmed that preproET-2 and preproET-3 levels were increased relative to the vector control at 72-h post-transfection (Fig. 6A). Forced expression of ET-2 or ET-3 inhibited cell migration significantly, relative to vector control (Fig. 6B, inset). Cell invasion also was inhibited significantly by ET-2 overexpression (P < 0.05), whereas ET-3 was less effective under the same assay conditions (Fig. 6C). Results shown here were from real-time monitoring assays in Caco-2 cells, but similar findings were obtained in HCT116 colon cancer cell lines (data not presented).
Despite the growing interest in the ET axis and its role in cancer,1-3 most attention to date has focused on ET-1 and its receptors, with surprisingly little information on ET-2 and ET-3. In fact, there is a significant gap in our general understanding of the lesser studied ET isoforms and their functions in normal physiology and pathophysiology, especially in the case of ET-2. As noted in the introduction, ET-2 differs by only two amino acids from ET-1, has the same affinity as ET-1 for the receptors ETAR and ETBR, and may co-exist in the same tissue compartments. However, there is converging evidence for a distinct ET-2 pathway, and this “forgotten isoform”11 may have critical roles in ovary development, immunology, the cardiovascular system, and cancer.
This investigation has provided the first evidence for the involvement of ET-2 in colon cancer development. In marked contrast to ET-1,1,2,12,13 ET-2 and ET-3 expression was low or undetectable in both rat and human colon cancers, suggesting a possible tumor suppressor function in the colon. Importantly, in the rat model, loss of ET-2 and ET-3 was detected in colonic mucosa several weeks before the onset of frank tumors. This suggests that reduced expression of ET-2 and ET-3 might provide an early biomarker of cancer development in the colon.
Interestingly, epigenetic inactivation of ET-3 recently was identified in human breast cancer.9 We have extended these observations by showing hypermethylation of both EDN2 and EDN3 in human primary colon cancers, and in a panel of widely used human colon cancer cell lines. Although there was variation in the constitutive levels of EDN2 and EDN3 among the cell lines examined, in all cases treatment with 5-aza produced a significant increase in the corresponding gene expression. One potentially important observation was that, in any given cell line, the relative mRNA expression following 5-aza treatment was consistently higher for ET-2 than for ET-3 (Fig. 4A and data not shown). Moreover, in human primary colon cancers, MSP analyses revealed a 30-fold increase in ET-2 after 5-aza treatment, compared with only a 2-fold increase in ET-3. One interpretation is that ET-2 is the more critically silenced factor in the colon, and that its re-expression effectively antagonizes dysregulated ET-1. It is noteworthy that in normal colon of both rat and human, ET-2 was restricted largely to proprial macrophages and enteric ganglion cells, with mild expression in the epithelium and brush border (Fig. 2A). Similar findings were reported in the mouse;14 ET-2 was detected at the bottom of villi in the duodenum, ileum and jejunum, but the pattern was reversed in crypts of the colon and rectum.
We did not perform a comprehensive screening of all putative DNA methylation sites for the genes of interest, but focused on six CpGs in EDN2 and seven CpGs in EDN3. Computational analyses identified these as likely sites for DNA methylation, which was confirmed by both bisulfite sequencing and pyrosequencing. The results for EDN3 corroborate and extend prior studies of epigenetic silencing in human breast cancer,9 whereas the DNA methylation data reported here for EDN2 are novel.
Based on these observations, we hypothesized that ET-2 and ET-3 might be silenced early in cancer development so as to circumvent competition with ET-1 for its receptors, thereby avoiding “mixed messages” in the ET axis. In support of this idea, forced expression of ET-2 and ET-3 inhibited migration and/or invasion of human colon cancer cells in vitro (Fig. 6), whereas ET-1 had the opposite effect (data not shown). Monitoring assays in real time suggested a greater impact of ET-2 than ET-3 on cell invasion, and of ET-3 compared with ET-2 on cell migration. However, this was not always consistent among the colon cancer cell lines tested, and was dependent to some extent on the efficiency of transient transfection. We prescreened the colon cancer cell lines for expression of ETAR and ETBR (qPCR data not shown), and on this basis selected Caco2 and HCT116 cells for invasion/migration assays. It would be interesting to repeat the studies in HT29 and SW48 cells, since these lines had low or undetectable ETAR and ETBR levels yet responded to 5-aza treatment with strong re-expression of ET-2 and ET-3. This might provide insights in other cases, such as breast cancer, where ET-2 is a reported macrophage chemoattractant with a possible role in tumor cell invasion.15,16
We did not formally investigate the competition among ET-1, ET-2, and ET-3 for ETAR and ETBR receptors under the present conditions, but it is interesting to speculate on ligand versus receptor “silencing” as a driving mechanism in the disease context. Pao et al. examined the promoter methylation status of the gene coding for ETBR (EDNRB) and noted that colon normal tissue had levels of methylation that occasionally exceeded those in cancers of the bladder and prostate.17 These authors also examined a panel of human colon cancers, observing that 4/8 tumors had regions of the EDNRB promoter with methylation status exceeding 50%.17 Might ETBR downregulation in the colon serve as a trigger for reduced expression of selected ligands, such as ET-2 or ET-3? This seems to be unlikely, at least in the rat model, where ETAR and ETBR levels were consistently overexpressed, not silenced, in the colon tumors (R. Wang et al., unpublished results). Since loss of ET-2 and ET-3 expression occurred before the onset of frank tumors, we conclude that ET-2 and ET-3 silencing was not a driver for subsequent downregulation of ETA or ETB receptors during rat colon carcinogenesis. A more likely driver of ETAR and ETBR expression was active ET-1 peptide, being consistently elevated in rat colon tumors compared with normal colonic mucosa (R. Wang et al., manuscript in preparation). Further work is in progress to clarify these relationships and their effects on cancer cell migration and invasion.
In summary, the present work has shed new light on two largely forgotten isoforms in the ET axis, namely ET-2 and ET-3, and their roles in cancer etiology. Epigenetic silencing of EDN2 and EDN3 was clearly implicated in the loss of ET-2 and ET-3 expression in both human and rat colon tumors. The rat colon carcinogenesis model provided insights into the timing of epigenetic silencing, with evidence for loss of ET-2 and ET-3 several weeks before the onset of frank tumors. This raises the possibility that diminished expression of ET-2 and ET-3 might serve as an early biomarker of colon cancer risk. Mechanistic studies supported an inhibitory role of ET-2 and ET-3 on colon cancer cell migration and invasion, but further work is needed to clarify the molecular details and applicability of these findings to other cancers characterized by dysregulation of the ET axis. Nevertheless, this investigation has opened an avenue for potential new anticancer therapies targeting the re-expression of silenced members of the ET axis. This would provide for an alternative or complementary strategy to the small molecule inhibitors and receptor antagonists that are currently undergoing clinical evaluation.1,2
The endothelin axis is dysregulated in cancer development. Most studies to date have focused on ET-1 overexpression and the use of small molecule inhibitors or receptor antagonists. We now report that rat and human colon cancers exhibit epigenetic inactivation of ET-2 and ET-3, and that their forced re-expression inhibited migration/invasion of human colon cancer cells. These findings suggest a new therapeutic strategy based on the activation of natural antagonists of ET-1 in the colon.
We thank Carmen P. Wong for technical assistance with pyrosequencing. Steven F. Moss and Lelia Simao (Rhode Island Hospital, Providence, RI) kindly provided the panel of primary human colon cancers and patient-matched controls. Zhiqiang Yang assisted with statistical analyses. Cell invasion and migration assays were conducted in the Cell Imaging and Analysis Core of the Environmental Health Sciences Center at Oregon State University.
Grant Sponsor: US National Institutes of Health grants CA090890, CA65525, CA122906, CA122959, CA80176, and ES00210
Contributors RW was responsible for the carcinogenicity bioassay and molecular studies; WMD assisted with the carcinogenicity bioassays, necropsies, and cell culture; CVL and KAF performed histopathology and immunohistochemistry; JAG assistance with invasion/migration assays; HA provided TMAs; EH, DEW, and MRD were responsible for critical revision of the manuscript; RHD provided the study concept, and was responsible for data interpretation and drafting/editing of the manuscript.