Original Article
Colorectal Carcinoma Cell Production of Transforming Growth Factor Beta Decreases Expression of Endothelial Cell Vascular Endothelial Growth Factor Receptor 2 Elizabeth A. Kuczynski, MSc; Alicia M. Viloria-Petit, MSc, PhD; and Brenda L. Coomber, MSc, PhD
BACKGROUND: Vascular endothelial growth factor (VEGF) signaling is a target for antiangiogenic cancer therapy. The authors have previously observed that up to 40% of vessels in colorectal carcinoma (CRC) tumors are negative for VEGF receptor 2 (VEGFR2) expression. Differential activity of transforming growth factor beta (TGF-b) is a potential contributor to this receptor heterogeneity because TGF-b contributes to both angiogenesis and CRC tumor progression. METHODS: The authors analyzed VEGFR2 expression by Western blotting, and TGF-b expression in endothelial and CRC cell lines, respectively. In addition, they immunostained endothelial cells in CRC xenografts to find an association between VEGFR2 and TGF-b levels or activity. RESULTS: In bovine aortic endothelial cells (BAECs), TGF-b1 significantly repressed VEGFR2 protein in a time-dependent and dose-dependent fashion (P < .05). Serum-free conditioned media from various malignant human CRC cell lines (HCT116, 379.2, Dks8, and DLD1) induced down-regulation of VEGFR2 in BAECs. This effect was proportional to the total levels of TGF-b1 and TGF-b2 and was blocked by SB-431542 and SD-208, TGF-b receptor I inhibitors. Immunofluorescence staining of subcutaneous mouse xenografts of HCT116, 379.2, Dks8, and SW480 cells revealed vessels with an inverse relationship between TGF-b activity and VEGFR2 expression. Oxygen and bone morphogenetic protein 9 levels were shown to modulate TGF-b– induced VEGFR2 down-regulation. CONCLUSIONS: In combination with other factors, TGF-b may contribute to the C 2011 American Cancer Society. vascular heterogeneity in human colorectal tumors. Cancer 2011;117:5601–11. V KEYWORDS: colorectal carcinoma, transforming growth factor beta, vascular endothelial growth factor receptor 2, endothelial cells, angiogenesis.
Colorectal carcinoma (CRC) is the third most prevalent type of cancer in the United States; in 2010, 142,000 new cases and 51,000 deaths were estimated to occur (www.cancer.org). Antiangiogenic therapy with bevacizumab (Avastin), a monoclonal antibody against vascular endothelial growth factor (VEGF)-A, has been used clinically in combination with standard chemotherapy since its US Food and Drug Administration approval in 2004. The increase in progression-free and overall survival observed in the phase 3 clinical trial that prompted bevacizumab’s approval1 has not been reproduced in recent trials. Moreover, these trials have yielded worse survival outcomes in the anti-VEGF treatment arms.2,3 Typically, anti-VEGF therapies delay time to progression after a brief antitumor or antivascular effect, but they have poor efficacy in terms of stabilizing disease or increasing overall long-term survival.4 VEGF-A and its receptor VEGFR2 are considered to be the most important positive mediators of physiological and pathological angiogenesis through promoting vascular permeability and endothelial cell (EC) survival, migration, and proliferation.5 Our laboratory has observed that molecular heterogeneity is an inherent feature of tumor blood vessels and is characteristic of the angiopoietin receptor Tie-26 and VEGFR2.7 Among a range of tumor types, CRC has the most striking expression patterns for VEGFR2, with only 60% of vessels in clinical specimens or subcutaneous xenografts Corresponding author: Brenda L. Coomber, MSc, PhD, Department of Biomedical Sciences, Ontario Veterinary College, University of Guelph, Guelph, Ontario, Canada, N1G 2W1; Fax: (519) 767-1450;
[email protected] Department of Biomedical Sciences, University of Guelph, Guelph, Ontario, Canada Elizabeth Kuczynski’s current address: Department of Medical Biophysics, University of Toronto, Toronto, Ontario, Canada We thank Mrs. Kanwal Minhas for her technical assistance and other members of the Coomber laboratory for advice and support. Dks8 cells and 379.2 cells were generously provided by Dr. Senji Shirasawa and Dr. Bert Vogelstein, respectively. DOI: 10.1002/cncr.26247, Received: January 19, 2011; Revised: April 1, 2011; Accepted: April 13, 2011, Published online June 20, 2011 in Wiley Online Library (wileyonlinelibrary.com)
Cancer
December 15, 2011
5601
Original Article
having detectable VEGFR2 expression. Because adaptive and evasive mechanisms of resistance are typical of antiVEGF therapies, this VEGFR2 heterogeneity may have clinical significance for patient response.8 The cytokine transforming growth factor beta (TGF-b) is involved in CRC progression.9-11 Active TGF-b binds dimerized type II receptors (TGFbR-II), which recruit and phosphorylate TGbR-I (ALK5) and subsequently phosphorylate intracellular R-Smad2 and R-Smad3. R-Smads complex with Smad4, enter the nucleus, and associate with transcriptional coactivators or corepressors to regulate the expression of target genes.10 During tumorigenesis, cancer cells lose sensitivity to the growth inhibitory effects of TGF-b or usurp its tumorpromoting functions, leading to cancer cell growth, invasion, epithelial to mesenchymal transition, evasion of immune surveillance, and metastasis.9 In CRC, TGF-b is particularly important; 28% and 13% of human CRC tumors have TGbR-II and Smad4 mutations, respectively,10 and up to 85% of CRC cell lines are resistant to its growth inhibitory effects.12 TGF-b is often overexpressed in CRC, leading to high serum or plasma levels in patients, which are associated with poor prognosis.10,11 TGF-b has a biphasic effect on ECs, with low doses stimulating migration and proliferation and high doses being inhibitory.13 TGF-b, as well as the TGF-b superfamily member bone morphogenetic protein 9 (BMP9), can also activate the ALK1 type I receptor on ECs, which phosphorylates Smad1 and Smad5.14 Interestingly, ALK5 and ALK1 can modify each others’ activities, thereby either enhancing or opposing angiogenesis.15 Considerable crosstalk also occurs between VEGF and TGF-b. TGF-b up-regulates VEGF mRNA,16 and TGF-b down-regulates endothelial VEGFR2 protein and mRNA in a dosedependent manner.17 TGF-b1 has been proposed to stimulate the production of Hex suppressor in ECs, which inhibits the binding of EC-specific transcription factor GATA-2 to the 50 untranslated region of the VEGFR2 promoter.18,19 We therefore hypothesize that TGF-b present in the microenvironment of CRC tumors is potentially responsible for the heterogeneous expression patterns of VEGFR2 on blood vessels. Here, we report in vitro and in vivo evidence demonstrating an inverse relationship between TGF-b, both exogenous and from CRC cell linederived conditioned media (CM), and VEGFR2 expression by ECs. By understanding how TGF-b regulates VEGFR2 expression in colorectal tumor vessels, we can
5602
potentially optimize the efficient use of antiangiogenic therapies in a variety of cancers.
MATERIALS AND METHODS Cell Culture Primary bovine aortic endothelial cells (BAECs) were previously isolated from the aorta of adult cattle and were used before the 10th passage. Human CRC cell lines Caco2, HCT116, DLD1, and SW480 were obtained from the American Type Culture Collection. Cell line 379.2 is a p53 null cell line derived from HCT116 (p53 wt20). Dks8 was derived from DLD1 and mutant K-Ras deleted21; cells were obtained from their originators. All cultures were maintained in Dulbecco modified Eagle medium (DMEM) (Sigma-Aldrich, St Louis, Mo) supplemented with 10% fetal bovine serum (Gibco, Grand Island, NY), 1 mM sodium pyruvate (Sigma-Aldrich) and 0.25 mg/mL gentamicin (Gibco) at 37 C in 5% CO2 and 95% atmospheric air. For experiments in hypoxic conditions, cells were incubated in a Modular Hypoxic Chamber (Billups-Rothenberg, Del Mar, Calif) in .05). The active proportion of TGF-b1 or TGF-b2 was highly variable and did not differ across CRC cell lines (P > .05). Regression analysis with the line forced through (x,y) ¼ (0,1) found that the slope of the line (0.1744 0.05,376) differed significantly from zero (P ¼ .0176), hence BAEC expression of VEGFR2 and CRC cell expression of TGF-b are inversely related (Fig. 4B).
Figure 3. (A) Relative vascular endothelial growth factor receptor 2 (VEGFR2) protein expression by bovine aortic endothelial cells (BAECs) incubated with colorectal carcinoma cell line-derived serum-free conditioned media (CM) or serumfree Dulbecco modified Eagle medium (DMEM) over 24 hours is shown. Analysis of variance, P ¼ .0006; means were compared with DMEM control (n ¼ 2 or 3). (B) VEGFR2 protein expression in BAECs grown in serum-free CM derived from Caco2, HCT116, 379.2, Dks8, DLD1, and SW480 cell lines is shown. BAECs were treated for 1, 24, or 48 hours with dimethylsulfoxide (DMSO) vehicle or 5 lM SB-431542 (SB). (C) VEGFR2 levels in BAECs grown in CM from 379.2 and Dks8 cell lines for 1, 24, or 48 hours in the presence of DMSO or 2.5 lM SD-208 (SD) are shown. Graphs are representative of 2 to 3 independent experiments. *P < .05, **P < .01.
Cancer
December 15, 2011
Immunofluorescence of CRC Xenografts Five regions within each of 4 paraffin-embedded xenografted tumors per CRC cell line were analyzed (n ¼ 65120 vessels/tumor type). Vessels were analyzed from viable regions, mostly at the tumor rim. Immunofluorescence staining of sequential sections of tumors revealed heterogeneous expression of VEGFR2 in vessels (Fig. 5), as previously observed.7 Across tumor types, the proportion of VEGFR2-positive vessels was greatest in SW480 tumors. Distinct regions of phospho-Smad2 staining were observed in the nuclei of cancer cells. Most ECs were positive for phospho-Smad2, although at a lower staining intensity than cancer cells, and were not always adjacent to positive-staining cancer cells. By ANOVA, the mean proportion of each vessel type varied significantly within each tumor type for 379.2, HCT116, and SW480 (P < .05)
5605
Original Article
Figure 4. (A) Transforming growth factor beta1 (TGF-b1) and TGF-b2 concentrations normalized to cell count in colorectal carcinoma (CRC) cell line-derived conditioned media (n ¼ 2 to 6) are shown. Bar height represents total TGF-b1 and TGF-b2 levels. Analysis of variance (ANOVA) (TGF-b1); P < .0001; ANOVA (TGF-b2), P ¼ .0030. (B) Relation is shown between the mean total levels of TGF-b1 and TGF-b2 expressed by CRC cells and the relative expression of vascular endothelial growth factor receptor 2 (VEGFR2) protein by bovine aortic endothelial cells at 24 hours. Linear regression with the line forced through (x,y) ¼ (0,1) obtained a slope significantly different from zero (P ¼ .0176). DMEM, Dulbecco modified Eagle medium.
Table 1. Total (Active Plus Latent) and Active TGF-b1 and TGF-b2 in Colorectal Carcinoma Cell Line-Derived Serum-Free Conditioned Media Collected After 48 Hours
Cell Line
TGF-b1, ng/mL Mean
Caco2 HCT116 379.2 Dks8 DLD1 SW480
0.165 4.061 3.475 0.580 0.775 4.459
Total
SE
Mean
0.074 0.272 0.454 0.178 0.187 0.894
0.019 0.040 0.142 0.291 0.021 0.147
TGF-b2, ng/mL Active
SE
Mean
0.021 0.035 0.191 0.313 0.025 0.183
0.047 0.043 0.000 0.983 0.663 0.690
Total
SE
Mean
0.016 0.035 0.008 0.449 0.047 0.542
0.002 0.023 0.000 0.212 0.089 0.197
Active
SE 0.001 0.021 0.000 0.053 0.056 0.083
Abbreviations: SE, standard error; TGF-b, transforming growth factor beta.
but not Dks8 (P ¼ .2857; Fig. 5B). The number of phospho-Smad2þ/VEGFR2 and phospho-Smad2/ þ VEGFR2 vessels were analyzed by t test against doublepositive or double-negative vessels, but no significant differences were found for each tumor type (P > .05).
very low or undetectable at increasing time points. This coincided with a down-regulation of Smad2 protein (Fig. 6A). Hypoxia itself did not induce expression of TGF-b1 in either HCT116 and 379.2 cell lines after either 24 or 48 hours (Table 2).
Hypoxia and VEGFR2 Expression BAECs were incubated in
