Conflict of interest: The authors declare no conflict of interest.
Endothelin-1 (ET-1) is produced by and stimulates colorectal cancer cells. Fibroblasts produce tumour stroma required for cancer development. We investigated whether ET-1 stimulated processes involved in tumour stroma production by colonic fibroblasts. Primary human fibroblasts, isolated from normal tissues adjacent to colon cancers, were cultured with or without ET-1 and its antagonists. Cellular proliferation, migration and contraction were measured. Expression of enzymes involved in tumour stroma development and alterations in gene transcription were determined by Western blotting and genome microarrays. ET-1 stimulated proliferation, contraction and migration (p < 0.01 v control) and the expression of matrix degrading enzymes TIMP-1 and MMP-2, but not MMP-3. ET-1 upregulated genes for profibrotic growth factors and receptors, signalling molecules, actin modulators and extracellular matrix components. ET-1 stimulated colonic fibroblast cellular processes in vitro that are involved in developing tumour stroma. Upregulated genes were consistent with these processes. By acting as a strong stimulus for tumour stroma creation, ET-1 is proposed as a target for adjuvant cancer therapy.
Colorectal cancer accounts for almost a million cancer cases a year, and the incidence is rising worldwide. The initiation and progression of epithelial cancer types, including colorectal, is dependent on the development of a supportive tumour stroma. This is a glycoprotein-rich extracellular matrix (ECM) containing blood vessels and nonepithelial cells. It provides a scaffold and source of nutrients to allow tumours to grow and invade.1 In cancers, fibroblasts—the major stromal cell type—become activated and characteristically express α-actin. These cancer associated fibroblasts (CAFs) are mostly derived from resident stromal fibroblasts that are quiescent in the normal gut.2, 3 CAFs are found in increased numbers at the cancer invasion front and within more aggressive tumours.4, 5 Fibroblasts may be activated by various growth factors or peptides, such as endothelin-1 (ET-1).
The 21-amino acid vasoactive peptide ET-1 is the most ubiquitously expressed member of the endothelin family. It has multiple actions including the stimulation of smooth muscle contraction, angiogenesis, wound healing and fibrosis; signals are mediated via two seven-pass transmembrane G-protein linked receptors, ETA and ETB. Raised levels of ET-1 and its precursor have been found in early colonic adenomas and adenocarcinomas, with increased levels associated with a poor prognosis.6, 7In vitro, ET-1 stimulates proliferation and survival in colorectal cancer cell lines.8, 9 These actions are predominantly associated with the ETA receptor, which is upregulated in colorectal cancer.10
Fibroblasts from a number of tissues express both ETA and ETB,11 with ET-1 shown to stimulate various cellular processes, e.g., contraction and expression of ECM remodelling components.12–14 ET-1 has also been reported to stimulate migration and contraction in human colonic fibroblasts,15 although effects on cell proliferation, gene regulation and enzyme expression are unknown.
We hypothesize that ET-1 activates quiescent fibroblasts into CAFs, cells capable of promoting epithelial cancer development. In this study, we examined the response of human colonic fibroblasts to ET-1 (our in vitro CAF model) to determine whether the peptide stimulated: proliferation, contraction, migration or expression of matrix modifying proteins, all functions associated with activated fibroblasts. Additionally, specific receptor antagonists were used to delineate whether ETA or ETB receptors mediated these responses. Finally, we identified genes upregulated and downregulated by ET-1, with particular emphasis on those known to function in some of the cellular processes under investigation.
Material and Methods
ET-1 and the specific antagonists BQ-123 and BQ-788 (against ETA and ETB, respectively) were purchased from Bachem (UK) (St. Helens, Merseyside, United Kingdom). ET-1 was used at 10−7 M and antagonists at 10−6 M (optimum concentrations determined previously). All other solutions and reagents were obtained from Sigma-Aldrich Co. (Irvine, Ayrshire, United Kingdom), unless otherwise stated.
Primary human fibroblasts were isolated from areas of normal colon within resected specimens of six patients with colon cancer. The cancer specimens were all adenocarcinomas from both sexes: one moderately differentiated Dukes A; three moderately differentiated Dukes B; one poorly differentiated Dukes B; one poorly differentiated Dukes C. None had treatment prior to surgery. Cells were extracted from macerated tissue adjacent to, but not part of, the cancer tissue. (Informed consent was obtained from patients prior to surgery; study awarded ethical approval UCLH, 2003.) Antibody coated magnetic beads were used to remove epithelial and endothelial cells; the remaining cells (β-actin positive) were grown in selective medium to favour fibroblast growth. Fibroblasts were routinely cultured at 37°C, 5% CO2/air in Dulbecco's Modified Eagle Medium with 10% foetal calf serum, gentamicin 0.05 mg/ml; and passaged using a 1:3 or 1:4 split. Fibroblasts were used between passages 4–12 and thereafter discarded due to loss of fibroblastoid phenotype (e.g., cell shape started changing from the classical elongated spindle to a more rounded morphology).15 Fibroblasts grown from individual patients were designated as “strains” [cells propagated in vitro from primary cultures, with a finite lifetime (nontransformed)].
ET-1 and receptor antagonist incubations were carried out in serum free medium. For each assay described below (e.g., proliferation and migration), we used up to six fibroblast strains, separately. There were up to six independent repeats for each strain.
Cytospin slides were prepared (∼0.25 ml, 500,000 fibroblasts/ml) (Shandon Scientific, Cheshire, United Kingdom), air dried and fixed in acetone (3 min, −20°C). Slides were washed (phosphate buffered saline, PBS) prior to immunostaining for the expression of ET-1, ETA or ETB (antibodies at 1:500, 1:200, 1:200; Alomone Labs, Israel). We used the Vectastain universal ABC-AP kit, which uses a biotinylated universal secondary antibody and vector red substrate for visualisation. (Vector Labs, Peterborough, Cambridgeshire, United Kingdom). Slides were counterstained with haematoxylin/eosin.
Ninety-six-well plates were seeded with 10,000 fibroblasts/well and cultured in serum containing medium for 24 hr or until ∼60% confluence. Medium was discarded, cells were washed (PBS) and incubated with ET-1 and/or specific receptor antagonists (serum free medium) for 48 hr. Cell number was determined by the colourimetric methylene blue (MB) assay.16 Briefly, MB dye chelates with nucleic acids and therefore the larger the cell number the more dye is bound. Linearity between cell number and absorbance at 650 nm (of eluted dye) was previously validated.
Scratch wound migration
Migration was assessed using a modified scratch wound assay.15, 17 Twelve-well plates were seeded with 50,000 fibroblasts/well and grown to confluence in serum containing medium. Cells were serum starved for 24 hr in medium containing 0.1% bovine serum albumin (BSA); ‘scratches’ were made through confluent monolayers with a pipette tip (∼1.3-mm width) and washed (PBS). Cells to be treated with receptor antagonists (alone/in combination) were preincubated with antagonists alone (1 hr). ET-1/receptor antagonist solutions (with 1 μg/ml mitomycin C to block division) were added and photographs taken at 6-hourly intervals up to 24 hr. Migration was measured semiquantitatively, by the amount of encroachment into the acellular area as percentage of 0-hr gap size.18
Type 1 rat tail collagen gels were prepared and impregnated with 50,000 fibroblasts/gel in each well of a 24-well plate.19 Gels to be treated with receptor antagonists (alone/in combination) were preincubated with antagonists alone (1 hour). Gels were then “floated” by adding 1 ml/well ET-1 with/without antagonist solution, incubated for 72 hr, then fixed with formaldehyde. Gel contraction was determined by loss of water as measured by weight loss and photographs were taken.
Near confluent fibroblasts were serum starved as above and then preincubated with antagonists alone as appropriate (1 hr). After 24-hr exposure to ET-1 and/or receptor antagonists, cells were lysed using CellLytic M mammalian cell lysis/extraction reagent. Proteins underwent sodium dodecyl sulphate 12% polyacrylamide gel electrophoresis (SDS-PAGE)13; transferred onto nitrocellulose membranes (30V, 90 min); and blocked for 1 hr with 5% nonfat milk in 0.2%Tween-20/PBS. Antigens were detected using mouse or rabbit monoclonal antibodies against: the ECM modulators matrix metalloproteinases MMP-2, MMP-3 and their tissue inhibitor TIMP-1 (1:500, all from Calbiochem, Beeston, Nottinghamshire, United Kingdom); connective tissue growth factor (CTGF); collagen XI alpha 1 chain; (both at 1:1,000, SantaCruz Biotechnology, Santa Cruz, CA). Primary antibody was visualised using an antimouse or antirabbit secondary IgG biotin conjugate, as appropriate (Vector Labs). The antigen–antibody complexes were incubated with ABC reagent (Vector Labs) and detected by chemiluminescence (GE Healthcare/Amersham, Amersham, Buckinghamshire, United Kingdom). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the housekeeping protein. Immumoblots for each protein were performed at least twice, for each cell strain.
Gene expression analysis
Near confluent fibroblasts were serum starved as above and incubated with/without ET-1 (4 hr). Total RNA was isolated using Trizol (Invitrogen, Paisley, Renfrewshire, United Kingdom). Microarray analysis was carried out as previously described.14, 20 Briefly, RNA was reverse transcribed into cDNA and then in vitro transcribed into biotinylated cRNA. Target cRNA was fragmented and hybridized to the Affymetrix human U133A array (Santa Clara, CA), following a standard Affymetrix protocol. Hybridization of cRNA to U133A chips, signal amplification and data collection were performed using an Affymetrix fluidics station and chip reader.
Chip files were scaled to an average intensity of 100/gene and analysed using Affymetrix Version 5.0 (MAS5) comparison analysis software. Transcripts were defined as upregulated or downregulated when identified as present by the Affymetrix detection algorithm and as significantly increased/decreased as determined by the Affymetrix change algorithm (p < 0.01). Fold changes between treated and control samples had to be at least 1.5-fold to identify an altered transcript. Expression for 20,631 genes was analysed; variations in gene expression between treated and control samples were grouped into functional classes.
Data are presented as means with standard deviations or percentage of control values, as appropriate. Graphical presentation of growth, migration and contraction combine data from all fibroblast strains used (e.g., means of means, or taking each control data set as 100% to facilitate visualisation). Statistical testing was performed on the original values using one-way ANOVA followed by post-hoc analysis using Tukey's honestly significant difference test.
Colonic fibroblasts expressed ET-1 and ETA/ETB receptors
ET-1 and the specific endothelin receptors ETA and ETB were detected in the four fibroblast strains tested (Fig. 1a). Strong ETA and ETB and weaker ET-1 staining was seen throughout the cytoplasm. ETB appeared to be more specifically localised in cytoplasmic organelles. This is congruent with transfection experiments where ETB was detected in the endosomes and endoplasmic reticulum.21
ET-1 stimulated growth via ETA receptors
The effect of ET-1 in the absence or presence of each specific receptor antagonist on cell growth was investigated in all six fibroblast strains; each showed the same response pattern (Fig. 1b). Generally, ET-1 stimulated only a small increase in cell numbers, reaching significance in one strain only. However, the effect ET-1 is exerting on cell growth is more clearly demonstrated by ET receptor blockade. When the ETA receptor was antagonised, fibroblasts showed a reduction in cell growth of 35% (p < 0.001). When fibroblasts were incubated with ET-1 and at the same time ETA was blocked, fibroblast growth was reduced by 19% (p < 0.001). Antagonism of ETB receptors did not result in a significant reduction in proliferation.
ET-1 stimulated migration predominantly via ETB receptors
The migratory response of fibroblasts to ET-1 stimulation was variable. Four fibroblast strains migrated strongly in response to ET-1, while two strains showed only a weak response. ET-1 stimulated the four responsive fibroblast strains to migrate by 75–95% compared with control migration of 25% (p < 0.01), over 24 hr (Fig. 2). This effect was reduced to control levels by ETB blockade. ETA antagonism only partially reduced migration (p < 0.05).
ET-1 stimulated contraction via ETA and ETB receptors
All six fibroblast strains responded to ET-1 treatment: ET-1 induced contraction of gels resulted in 40% reduction in weight compared with controls (p < 0.001). The response was inhibited by both ETA and ETB blockade. ETA antagonism reduced ET-1 stimulated contraction by 16% and ETB antagonism by 18% (p < 0.01 v ET-1 group) (Fig. 3).
ET-1 increased levels of ECM modifying proteins
Levels of matrix metalloproteinase enzymes MMP-2, MMP-3 and the MMP inhibitor TIMP-1 in response to ET-1 and receptor antagonists were examined in four fibroblast strains. ET-1 caused an increase in TIMP-1 protein levels, which was blocked by both ET receptor antagonists. This inhibition was more pronounced with ETA antagonism than ETB antagonism. MMP-2 levels were also increased by exposure to ET-1 and again this was blocked to a greater extent with ETA than ETB antagonism. MMP-3 levels remained unchanged regardless of treatment (Fig. 4a).
Following gene microarray results from two fibroblast strains, we examined the expression of two more proteins: CTGF and collagen XI - products of two of the most strongly upregulated genes in response to ET-1. CTGF protein levels were increased more than ten-fold. Collagen XI expression was increased about 2.5-fold (Fig. 4b).
ET-1 altered gene expression
Gene expression in response to ET-1 was examined in two strains: 141 altered genes common to both strains were identified (see Additional Supporting Information); while the majority (85%) were mostly concerned with normal homeostasis or biosynthesis, the remaining were considered relevant and grouped according to function (Table 1).
Table 1. Genes commonly upregulated or downregulated in two fibroblast strains, after ET-1 stimulation. Fold-changes shown as averages of the two values
Upregulated genes included growth factors and receptor regulators, e.g., CTGF, a number of genes belonging to the major intracellular signalling transduction pathways such as MAPK, SMAD, PLC and cyclin B. However, some components of the SMAD and PLCβ pathways were also downregulated. Other upregulated genes produce intracellular molecules that control actin organisation and in turn determine the myofibroblast phenotype, such as WAS/WASL, WIPF1 and TRIO.
In terms of the ECM and molecules that determine cell–matrix interactions, gene alterations in response to ET-1 concerned collagen deposition and adhesion modulators, such as fibronectin (upregulated) and integrins (downregulated). The ephrin receptor A2 (EPHA2) that promotes cell motility is upregulated as well as a number of molecules controlling the bioavailability of insulin-like growth factor (IGF). A diagrammatic summary of these genes and their actions is shown in Figure 5. Notably, the majority of downregulated genes are mediators of inflammation.
Fibroblast activation with subsequent production and release of growth factors and remodelling of the tumour stroma is integral to the development and progression of solid tumours. This study shows that ET-1, a peptide produced by the majority of colorectal adenocarcinomas,7 is a stimulatory factor for human colonic fibroblasts in vitro. The fibroblasts used were resident stromal fibroblasts taken from normal colon areas adjacent to colonic tumours. These cells are the major pool for recruitment of activated CAFs in vivo. We showed that the fibroblasts express ET-1 and its receptors and that ET-1 stimulates fibroblast functions involved in tumour development: growth, migration, contraction and the production of ECM modifying proteins. None of the fibroblast strains used behaved in a consistently ‘different’ way across all investigations carried out in this study, i.e., there was not a generally unresponsive, or highly responsive cell strain.
This is the first study to demonstrate that ET-1 contributes to the growth of human colonic fibroblasts. Previously it has only been demonstrated in dermal and lung fibroblasts.22–24 This effect, as with proliferation of rat cardiac fibroblasts, is mediated via the ETA receptor. In the only other study on the effect of ET-1 on growth of human colon fibroblasts (strain isolated from a 2.5-month old), the authors found no significant response.15 However, Kernochan et al.15 had not investigated the effect of ET receptor antagonists on proliferation. When we blocked ETA receptors, we significantly reduced cell numbers. This suggests that ET-1, especially endogenously produced, contributed to cell growth/survival. Any further discrepancies may be accounted for by age-related differences (which are widely recognised for fibroblasts) or by natural variation between cells isolated from different individuals.
ET-1 has been shown to stimulate migration in hepatic stellate and dermal fibroblasts.13, 17, 25 In our experiments, the effect appeared to be mediated through the ETB rather than the ETA receptor. These findings support a previous report.15 In rat granulation tissue pouches and in human lung fibroblasts, the ETA receptor was the main signal transducer for ET-1 stimulated contraction.13, 26 In the present study, ET-1 stimulated contraction via both ETA and ETB as in the previous human colon fibroblast work.15
The MMP system is a tightly controlled group of proteases and inhibitors that are produced predominantly by fibroblasts. The molecules are involved in a number of key tumourigenic processes: matrix degradation, activation of growth factors and degrading enzymes and modulation of angiogenesis. The MMP system is complex and the role of ET-1 within it is still being delineated.27, 28 MMP-2 and 3 and TIMP-1 were investigated in this study, because they have been reported as upregulated in colorectal tumours. We found both TIMP-1 and MMP-2 expression was increased by ET-1 and blocked by both ETA and ETB receptor antagonists, while MMP-3 levels were unaffected. MMP-2 and MMP-3 have been detected by ELISA in colorectal tumours and MMP-2 plasma levels correlated with lymph node status.29 Survival analyses suggest MMP-2 and 3 are independent predictors of poor prognosis in colorectal cancer.30 In the latter, immunohistochemical images appear to show MMP-3 almost exclusively expressed by cancer epithelial glands. The poor expression of MMP-3 in fibroblasts may explain why ET-1 did not alter MMP-3 levels in our experiments. mRNA for TIMP-1 has been reported as raised in colorectal cancers, especially so in metastatic disease. In situ hybridisation localised mRNA to stromal fibroblast-like cells.31 TIMP-1 is a high affinity inhibitor of many MMPs, including MMP-2 and 3, and its expression is critical to prevent uncontrolled destruction of the ECM. Other functions of the MMP inhibitors include stimulation of proliferation in various cell types, and this may contribute to their role in the progression of cancers.32 No previous studies have demonstrated increases in MMP levels in human colonic fibroblasts in response to ET-1, although in fibroblasts from other sites, such as lung fibroblasts, ET-1 stimulates both MMP and TIMP expression.33
Microarray genome-wide screening identified a number of ET-1 induced gene alterations, such as CTGF, that activates intracellular pathways resulting in increased migration, proliferation and MMP production, e.g., MMP-2.34CTGF upregulation is a common outcome of ET-1 fibroblast stimulation, with both gene and protein expression confirmed in dermal and lung fibroblasts.35, 36 Increased CTGF expression is also seen in colonic fibroblasts from patients with Crohn's disease. The resulting fibroblast activation and stroma production may contribute to stricture formation.37 CTGF stimulated proliferation can be augmented by IGF, an autocrine mitogen for colonic fibroblasts.38 The glutaminase gene was upregulated by ET-1 in our study, and this was previously reported for lung fibroblasts from scleroderma patients39; the product is responsible for releasing and activating IGF.
ET-1 modulated expression of genes involved in migration and contraction: Upregulated TRIO and F-actin binding protein control actin cytoskeletal organisation, enabling cell shape changes and promoting motility.40, 41Fibronectin III has previously been reported to be increased in colonic fibroblasts in response to TGFβ1 stimulated migration.42 There was a concomitant downregulation in the adhesion molecules integrins. Reduced adhesion, increased motility and cytoskeletal reorganisation are the key components to increased cellular movement.18 The gene for collagen type XI alpha 1 chain was upregulated, and this was confirmed at the protein level. The protein is not found in the healthy colon, but its overexpression has been reported in stromal fibroblasts in most sporadic colorectal carcinomas.43
Downstream intracellular signalling molecules modulated by ET-1 include messengers for the major signalling pathways MAPK, SMAD and PLC.24, 44, 45 The upregulation in fibroblasts of two genes: the ephrin receptor A2 (EPHA2) and AML1 are reported here for the first time. The ephrin-ephrin receptor system is responsible for cell motility and invasion46 and associated with tumour growth and progression. Raised levels of ephrin receptor have been detected in various cancers47; EPHA2 specifically has been proposed as an oncogene in intestinal tumours.48 The AML1 (RUNX1) gene is a frequent target for translocations in various leukemias, especially childhood acute myeloid leukemia. The gene product has been shown to bind to SMAD3 and compromise its ability to stimulate downstream targets, therefore interfering with the action of TGFβ.49 Possible mechanisms linking key upregulated genes and biological processes stimulated by ET-1 in fibroblasts are shown in Figure 5.
ET-1 is a multifunctional molecule in cancer, stimulating epithelial cancer cell growth, angiogenesis and as shown here the activation of cancer adjacent fibroblasts.50 Our findings support the hypothesis that within colorectal cancers, ET-1 can activate fibroblasts from inactive pools of stromal fibroblasts. Tumour–stromal interactions are critical in the development of tumours, and their disruption is a promising method of anticancer treatment. This in vitro study identifies ET-1 as a hormone that could act to modulate tumour stroma development. The fibroblast responses to ET-1 were due to stimulation by both ETA and ETB receptors. Both receptors should be considered as targets for antitumour therapies, including prophylactic regimens for patients who are high risk for developing metastatic disease.
The authors are grateful to the Royal College of Surgeons for awarding J.K. a research fellowship.