TGF-Beta Suppresses VEGFA-Mediated Angiogenesis in Colon Cancer Metastasis

The FET cell line, derived from an early stage colon carcinoma, is non-tumorigenic in athymic nude mice. Engineered FET cells that express TGF-α (FETα) display constitutively active EGFR/ErbB signaling. These cells readily formed xenograft tumors in athymic nude mice. Importantly, FETα cells retained their response to TGF-beta-mediated growth inhibition, and, like the parental FET cells, expression of a dominant negative TGF-beta type II receptor (DNRII) in FETα cells (FETα/DNRII) abrogated responsiveness to TGF-beta-induced growth inhibition and apoptosis under stress conditions in vitro and increased metastatic potential in an orthotopic model in vivo, which indicates metastasis suppressor activity of TGF-beta signaling in this model. Cancer angiogenesis is widely regarded as a key attribute for tumor formation and progression. Here we show that TGF-beta signaling inhibits expression of vascular endothelial growth factor A (VEGFA) and that loss of autocrine TGF-beta in FETα/DNRII cells resulted in increased expression of VEGFA. Regulation of VEGFA expression by TGF-beta is not at the transcriptional level but at the post-transcriptional level. Our results indicate that TGF-beta decreases VEGFA protein stability through ubiquitination and degradation in a PKA- and Smad3-dependent and Smad2-independent pathway. Immunohistochemical (IHC) analyses of orthotopic tumors showed significantly reduced TGF-beta signaling, increased CD31 and VEGFA staining in tumors of FETα/DNRII cells as compared to those of vector control cells. These results indicate that inhibition of TGF-beta signaling increases VEGFA expression and angiogenesis, which could potentially contribute to enhanced metastasis of those cells in vivo. IHC studies performed on human colon adenocarcinoma specimens showed that TGF-beta signaling is inversely correlated with VEGFA expression, indicating that TGF-beta-mediated suppression of VEGFA expression exists in colon cancer patients.


Introduction
Transforming growth factor beta (TGF-beta) comprises a group of multifunctional polypeptides that regulate different cellular processes, including cell proliferation, apoptosis, differentiation, migration, tumorigenecity and metastasis, through binding to TGF-beta receptors. Many studies indicate that TGF-beta signaling acts as either a tumor promoter or a tumor suppressor. The tumor promoter function of TGF-beta has been associated with its ability to induce an epithelial to mesenchymal transition (EMT), which confers resistance to the apoptotic effects of TGFbeta [1][2][3]. However, we and others have demonstrated experimentally that TGF-beta mediates tumor suppressor activity in a variety of cancers including colon cancer, and that loss of TGFbeta signaling leads to acquisition and progression of malignancy [4][5][6][7][8][9][10][11]. Our recent studies demonstrate that TGF-beta signaling suppresses metastasis in a subset of colon cancer cells in an orthotopic model in vivo [12]. Therefore identifying the mechanism(s) by which TGF-beta elicits its metastasis suppressor function is crucial for our understanding of tumor progression and for developing effective therapeutic strategies.
Angiogenesis is an important determinant of tumor progression. Solid tumors cannot grow beyond a certain size without sufficient vascular supply to provide enough oxygen and nutrients. The VEGF/VEGFR pathway is a key mediator of angiogenesis [13] and VEGFA acts as a potent tumor angiogenic factor [14]. VEGFA stimulates the growth of new blood vessels, which provide tumors with needed oxygen and nutrients [14,15]. Expression of VEGFA has been shown to be regulated at the transcriptional and translational levels [16][17][18][19]. Regulation of VEGFA expression by TGF-beta signaling has not been very well studied. It has been reported that TGF-beta induces VEGF secretion in human cytotrophoblast cells [20]. On the contrary, another group showed that TGF-beta antisense oligonucleotides increases VEGFA expression in human keratinocytes and skin fibroblasts in vitro [21]. Therefore, TGF-beta may differentially regulate VEGFA expression dependent upon cellular context.
The FET colon carcinoma cell line, which was isolated from a well differentiated early stage cancer, retains autocrine TGF-beta activity. These cells form a small tumor nodule at the site of inoculation in athymic nude mice that does not progressively grow and then regresses over 3-5 weeks [5,22]. The inability of these colon carcinoma derived cells to generate progressive tumor growth in vivo suggests that, relative to the more progressed model cell lines, FET cells retain many normal growth controls, including the inhibitory response to TGF-beta. Therefore, FET cells were engineered to express TGF-a (FETa), which constitutively activates EGFR/ErbB signaling. These cells (FETa) readily formed xenograft tumors in nude mice [22]. However, FETa cells retained their response to TGF-beta-mediated growth inhibition and rarely metastasized in vivo. Expression of a dominant negative TGF-beta type II receptor (DNRII) in FETa cells, designated FETa/DNRII, abrogated responsiveness to TGF-betainduced growth inhibition and apoptosis under stress conditions in vitro and significantly increased metastatic potential in an orthotopic model in vivo [12]. Thus, it appears that the loss of TGF-beta tumor suppressor response is permissive for the development of metastasis by FETa cells, which provides evidence in support of the metastasis suppressor function of TGF-beta signaling in this cell model. The current study utilizes the FETa/ vector and FETa/DNRII cell model system to test the hypothesis that TGF-beta signaling mediates angiogenesis through regulating VEGFA expression, which could contribute to the metastasis suppressor function of TGF-beta. In this study, we report that, in colon cancer cells, TGF-beta inhibits VEGFA expression at the post-transcriptional level, probably by decreasing VEGFA protein stability through ubiquitination and degradation. This regulation is Smad3-dependent and Smad2-independent. Our data also indicate that TGF-beta-mediated down-regulation of VEGFA expression is through a PKA-mediated pathway. In addition, we demonstrate that the loss of autocrine TGF-beta in FETa/DNRII cells results in increased expression of VEGFA as compared to the control cells both in vitro and in vivo, suggesting that one aspect of autocrine TGF-beta activity in its metastasis suppressor context represses angiogenesis. Finally, IHC studies of sections of human colonic adenocarcinomas reveal that TGF-beta signaling is inversely correlated with VEGFA expression, indicating that the link between loss of TGF-beta activity and increased VEGFA expression also exists in colon cancer patients.

Materials and Methods
All experiments involving animals were approved by the University of Nebraska Medical Center Institutional Animal care and Use Committee (IACUC #: 07-043-08-FC).

Cell Lines and Reagents
The FETa/DNRII cells were obtained by transfection of a dominant negative TGF-beta type II receptor into FETa cells, whereas FETa/vector control cells were generated by transfection of a control plasmid. FETa/vector, FETa/DNRII, HCT116, RKO, C, CBS, GEO and parental FET colon carcinoma cells [9] were cultured at 37uC in a humidified incubator with 5% CO 2 in McCoy's 5A serum-free medium (Sigma) supplemented with 5 ng/ml epidermal growth factor, 20 mg/ml insulin, and 4 mg/ml transferrin as described previously [23]. Recombinant human TGF-beta1 was purchased from R&D Systems. Cycloheximide, MG132 and PYR41 were purchased from AMRESCO, Sigma and CALBIOCHEM respectively.

Isolation of RNA and RT-PCR
Total cellular RNA from FET cells were isolated using Trizol total RNA isolation reagent (Invitrogen), according to the manufacturer's instructions. For RT-PCR, 2 mg of RNA was reverse-transcribed with M-MLV reverse transcriptase (Promega,) using a random primer. Two microliters of the cDNA product was used to amplify human VEGFA. Primer sequences for VEGFA were 59-CGGATCAAACCTCACCAAGGCC-39 (forward) and 59-CTTTCTCCGCTCTGAGCAAGGC-39 (reverse). Conditions for amplification were: one cycle at 94uC for 2 minutes followed by 30 cycles at 94uC for 30 seconds, 60uC for 30 seconds and 72uC for 30 seconds. The actin gene was used as an internal control.

Immunofluorescence Staining
The FETa/vector and FETa/DNRII cells were grown to 70-80% confluent on glass cover slips. They were washed with PBS and fixed in ice cold methanol for 5 minutes. Fixed cells were permeabilized with 0.1% Tween 20 in PBS and blocked for nonspecific binding with 10% normal goat serum in PBS for 30 minutes at 4uC. Cells were then incubated with primary anti-VEGFA antibody (1:50 dilution, Santa Cruz Biotechnology) for overnight at 4uC, followed by incubation in the dark for1 hour at room temperature with the secondary antibody Dylight488conjugated with anti-rabbit IgG (1:400 dilution, Jackson Immu-noResearch Laboratories). Stained cells were mounted in vectashield hardset mounting medium with dapi (Vector Labs) and imaged using a fluorescent microscope.

Tissue Samples
Formalin fixed paraffin embedded (FFPE) tissue blocks from the primary tumors of mice bearing FETa/vector and FETa/DNRII cells were described previously [2]. FFPE blocks of normal human colon removed for reasons other than malignancy and those containing invasive colonic adenocarcinomas were obtained from files of the Department of Pathology and Microbiology at the University of Nebraska Medical Center (UNMC). The ages of all patients were between 55 and 85 years, with material coming from both men and women. The cancer patients did not receive anticancer therapy prior to surgical removal of the tumor.
The study was performed with the approval of the ethics committee (Institutional Review Board) of UNMC. Tissue used for the study was obtained from excess material remaining in the paraffin block after issuance of the pathologic diagnosis. The surgery consent form used at UNMC includes a provision allowing the use of excess material for research purposes. Each form is on file with the Department of Pathology and Microbiology and was reviewed prior to experimentation. Therefore, the ethics committee waived the need for additional written consent from the participants.

Immunohistochemistry (IHC) Staining
Four micron thick tissue sections were cut and dewaxed in histoclear for 15 minutes and then rehydrated using 100%, 95% and 70% of graded alcohol for 5 minutes respectively. Antigen retrieval for pSmad2 was performed using Novocastra, Epitope Retrieval Solutions at pH 6.0 (Leica). Incubation with primary anti-pSmad2 antibody (1:200 dilution, Millipore) followed by NovoLink Polymer (Leica) was used to detect phosphorylation of Smad2. Antigen retrieval for VEGFA and CD31 was performed using Tris-EDTA buffer at pH 9.0. To detect VEGFA and CD31 expression, anti-VEGFA antibody (1:50 dilution, Santa Cruz) and anti-CD31 antibody (1:100 dilution, Abcam) were incubated for overnight at 4uC, followed by incubation with a HRP conjugated goat anti-rabbit secondary antibody (1:300 dilution, Jackson Laboratories) for VEGFA and with a HRP labeled polymer anti-rabbit (DAKO) for CD31. To neutralize endogenous peroxidase, Dako Dual Endogenous Enzyme Block (DAKO) was used for VEGFA and CD31 while Novocastra Peroxidase Block (Leica) was used for pSmad2. Slides were then developed using DAB Chromogen kit (DAKO) followed by counter-staining with hematoxylin. Slides were mounted with Permount (Fisher Scientific) and subjected to microscopic examinations. Staining density was measured and quantified with ImagePro plus 7.0 Software using 206 or 406 images.

TGF-beta signaling inhibits VEGFA expression in colon cancer cells
Examination of VEGFA expression in a panel of colon cancer cells indicated that cell lines with the wild type TGF-beta RII (CBS, GEO and FET [7]) expressed lower levels of VEGFA than those with mutant TGF-beta RII (RKO, C, HCT116 [7]) (Fig. 1A, left panel), suggesting an inverse relationship between TGF-beta signaling and VEGFA expression. Re-expression of the wild type TGF-beta RII in HCT116, one of the cell lines with mutant TGFbeta RII, led to reduced VEGFA expression (Fig. 1A, right panel). To directly determine whether TGF-beta signaling negatively regulates VEGFA expression, FET cells were treated with 4 ng/ml of TGF-beta. As shown in Figure 1B, exogenous TGF-beta reduced VEGFA expression in a time-dependent manner, with significant reduction after 24 hours of TGF-beta treatment. SB525334, a potent inhibitor of TGF-beta type I receptor (RI) kinase, was used to confirm the effect of TGF-beta. Treatment of FET cells with SB525334 reversed the inhibitory effect of TGFbeta on expression of VEGFA (Fig. 1C). To further confirm the suppressive effect of TGF-beta on VEGFA expression, a dominant negative TGF-beta RII (DNRII) was introduced into FETa cells to abrogate endogenous TGF-beta signaling, which was shown by reduced Smad2 and Smad3 phosphorylation (pSmad2 & pSmad3; Fig. 1D, left panel). Consequently, VEGFA expression was increased significantly in DNRII cells as compared to the vector control cells (Fig. 1D, left panel). Immunofluorescence staining of VEGFA confirmed higer expression of VEGFA in FETa/DNRII cells than in FETa/vector cells (Fig. 1D, right panel). To determine how TGF-beta suppresses VEGFA expression, mRNA expression of VEGFA was determined in FET cells treated with TGF-beta for different periods of time. RT-PCR results showed that VEGFA mRNA levels remained unchanged after TGF-beta treatment for up to 48 hours ( Fig. 2A, upper panel). Quantification of repeated experiments confirmed the observation ( Fig. 2A, lower panel). These results indicate that TGF-beta regulates VEGFA expression at the post-transcriptional level. One of the possible mechanisms of post-transcriptional regulation is that of protein stability. To determine whether TGF-beta signaling regulates the protein stability of VEGFA, cycloheximide (CHX) was added to FET cells to inhibit protein synthesis while they were treated with TGF-beta. Under these conditions, levels of VEGFA decreased faster in TGF-beta-treated samples than in control samples (Fig. 2B, upper panel). Quantification and normalization of the intensity of the VEGFA bands with that of the corresponding actin bands clearly showed that TGF-beta treatment reduced the half-life of VEGFA protein (Fig. 2B, lower  panel), indicating that TGF-beta regulates VEGFA protein stability. To further dissect the underlying mechanism, a proteasome inhibitor, MG132, or ubiquitin E1 inhibitor, PYR41, was used to treat FET cells together with TGF-beta. As shown in Figure 2C, either MG132 or PYR41 prevented TGFbeta-mediated suppression of VEGFA expression. These results suggest that TGF-beta signaling decreases VEGFA protein stability through ubiquitination and degradation.
TGF-beta signaling is mediated predominantly through Smad activation. However, Smad-independent TGF-beta signaling has also been reported in different cell types [23,24]. To determine whether TGF-beta-mediated inhibition of VEGFA expression is Smad-dependent or -independent, Smad2 and Smad3 were knocked down individually in FET cells by shRNAs specific for Smad2 or Smad3. Expression of Smad2 or Smad3 was reduced efficiently and specifically in Smad2 or Smad3 knockdown cells (Fig. 2D, left panel). As a result, increases in pSmad2 or pSmad3 upon TGF-beta treatment were abolished in Smad2 or Smad3 knockdown cells respectively (Fig. 2D, right panel). Western blot analysis indicated that inhibition of VEGFA protein expression by TGF-beta was partially reversed in Smad3 knockdown cells whereas knockdown of Smad2 had little effect (Fig. 2D, right  panel). These results indicate that TGF-beta suppresses expression of VEGFA in a Smad3-dependent and Smad2-independent manner. The partial reversal of TGF-beta-mediated suppression of VEGFA expression in Smad3 knockdown cells may be attributed to incomplete silencing of Smad3 expression in those cells or the existence of a Smad3-independent mechanism.
As shown previously, TGF-beta regulates XIAP downregulation through a PKA-mediated pathway [25]. We next determined whether PKA is involved in the regulation of VEGFA expression by TGF-beta. The expression of PKA catalytic a subunit (PKACata) was knocked down by a specific shRNA in FET cells (Fig. 2E, left panel). As a result, TGF-beta-mediated inhibition of VEGFA expression was almost completely abrogated in those cells (Fig. 2E, right panel). Taken together with the observation that PKA activation is Smad3-dependent [25], these results indicate that TGF-beta/Smad3/PKA signaling regulates downregulation of VEGFA expression in colon cancer cells.

TGF-beta signaling inhibits VEGFA expression and suppresses tumor angiogenesis in vivo
VEGFA is a key regulator of blood vessel formation and plays an important role in angiogenesis, which contributes to tumor development and progression. We have shown above that TGFbeta signaling inhibits VEGFA expression (Fig. 1B). Since TGFbeta acts as a metastasis suppressor in colon cancer cells [12], we examined whether suppression of VEGFA expression by TGFbeta contributes to metastasis suppression function of TGF-beta.
We used FETa/vector and FETa/DNRII cell model described previously [12] to address this question. VEGFA expression was much higher in FETa/DNRII cells than that in FETa/vector cells in vitro due to abrogation of TGF-beta signaling (Fig. 1D). As shown previously, FETa/DNRII cells showed significantly increased metastatic potential as compared to FETa/vector cells in an orthotopic model in vivo [12]. Primary tumor sections prepared from each mouse orthotopically transplanted with FETa/vector or FETa/DNRII cells were examined for TGFbeta signaling. Since FETa cells express both Smad2 and Smad3 (data not shown), Smad2 phosphorylation was used as an indicator for active TGF-beta/Smad2/3 signaling. IHC analyses using an anti-phospho-Smad2 (anti-pSmad2) antibody showed that nuclear Smad2 phosphorylation was significantly reduced in the primary tumors of FETa/DNRII cells as compared to those of FETa/ vector cells (Fig. 3A, left panel). Quantification of staining density of nuclear pSmad2 in multiple samples indicated that TGF-beta signaling was inhibited in the tumors of FETa/DNRII cells by about 40% (Fig. 3A, right panel, * P,0.001). To determine whether TGF-beta signaling suppresses VEGFA expression in vivo, we examined VEGFA expression in the primary tumors using IHC analyses. Results showed that VEGFA expression was much higher in the tumors of FETa/DNRII cells than those of the control cells (Fig. 3B, left panel). Quantification of staining density of VEGFA in multiple samples indicated that VEGFA expression was increased by more than two-fold in these tumors (Fig. 3B, right panel, * P,0.001), confirming the suppressive role of TGFbeta in VEGFA expression in vivo. To determine whether VEGFA expression has any biological consequences in vivo, we next analyzed microvessel formation within the tumors using an anti-CD31 antibody. CD31 is a marker of blood vessel endothelial cells. In the tumors of FETa/DNRII cells, microvessel density was more than 3-fold higher than that observed in the control samples (Fig. 3C, * P,0.001). These results indicate that inhibition of TGF-beta signaling leads to elevated VEGFA expression, which contributes to increased angiogenesis. Taken together with the results of increased metastatic potential of FETa/DNRII cells in the orthotopic model, our studies suggest that TGF-beta signaling suppresses VEGFA-mediated angiogenesis and tumor progression.

Expression of VEGFA inversely correlates with TGF-beta signaling in human colon cancer adenocarcinoma
The above in vitro and in vivo results demonstrate that TGF-beta negatively regulates VEGFA expression in human colon cancer cell lines. We next examined whether there was any correlation between TGF-beta signaling and VEGFA expression in human colonic adenocarcinomas. Sections prepared from 10 normal colon and 24 individual colon cancer patients were examined for active TGF-beta signaling and VEGFA expression. Since Smad2 and Smad3 are rarely mutated in colon cancer [26], Smad2 phosphorylation was used in this study to indicate active TGFbeta/Smad2/3 signaling. Normal human colon showed strong and distinct nuclear staining for Smad2 phosphorylation and weak staining for VEGFA in the crypt epithelial cells whereas the malignant epithelial cells in the adenocarcinoma cases displayed at most weak and diffused staining for Smad2 phosphorylation and strong staining for VEGFA by IHC analyses (Fig. 4A). Figure 4B shows the quantification of staining density of nuclear Smad2 phosphorylation and VEGFA expression of individual patient samples. Overall, normal colonic epithelial cells display higher nuclear Smad2 phosphorylation and lower VEGFA expression than adenocarcinomas, supporting the view that expression of VEGFA inversely correlates with TGF-beta signaling in human colonic adenocarcinoma.

Discussion
TGF-beta signaling has been shown to suppress tumor formation and metastasis in a subset of colon cancer cells through many different mechanisms including inhibition of cell proliferation and induction of apoptosis [4,12,25,27]. The ability of malignant cells to withstand environmental stresses, particularly those associated with metastasis, is considered a key factor in tumor development and progression [28]. Since environmental restriction on growth is very common in solid tumors such as colon carcinoma, a mechanism that enables increased angiogenesis to provide sufficient nutrients and oxygen would be highly advantageous to malignant cells and enhance their aberrant survival capacity. VEGFA is a potent angiogenic factor that plays essential roles in cancer. Tumor cells secret VEGFA to induce vasculature to develop, allowing a sufficient supply of oxygen and nutrients [14]. Therefore, VEGFA could be a potential target for cancer therapy. We have shown previously that escape from TGF-betamediated apoptosis in colon cancer cells contributes to their increased survival capacity and enhanced metastatic potential in vivo [12]. We show here that, in addition to inducing apoptosis, TGF-beta signaling can also inhibit expression of VEGFA and as a consequence, angiogenesis in colon cancer cells in vivo and that abrogation of TGF-beta enables increased VEGFA expression and angiogenesis, which could facilitate tumor growth and metastasis development. Furthermore, an inverse relationship between TGFbeta signaling and VEGFA expression is also observed in human colon cancer specimens, indicating the relevance of our studies to human cancer. Therefore, the inhibitory effect of TGF-beta on VEGFA expression and angiogenesis provides another important and novel mechanistic basis for TGF-beta tumor and metastasis suppressor function.
Although expression of VEGFA can be regulated at the transcriptional level by HIF1a [19], its regulation by TGF-beta is at the post-transcriptional level ( Fig. 2A). Our results indicate that TGF-beta treatment reduced the stability of the VEGFA protein through ubiquitination and degradation (Fig. 2B & 2C). Moreover, TGF-beta-mediated decrease in VEGFA expression occurs via a PKA-and Smad3-dependent and Smad2-independent pathway (Fig. 2D & 2E). It has been shown that activation of PKA by TGF-beta involves Smad3 but not Smad2 and that PKA activation promotes protein degradation through ubiquitination [25]. Therefore, it is possible that TGF-beta mediates VEGFA degradation through the Smad3/PKA pathway. This is a novel mechanism by which expression of VEGFA is regulated by TGFbeta signaling. Further studies will be necessary to determine the underlying mechanism.
High frequency of Smad4 mutation and inactivation is closely associated with increased metastases and poor prognosis in colon cancer [29,30]. Although TGF-beta signaling acts as a tumor suppressor in a subset of colon cancer cells (i.e. FET, GEO and CBS, etc) that express wild type Smad4 [4,10,11], Smad4independent TGF-beta signaling has been shown to promote colon cancer metastasis [31]. It has been reported that TGF-beta induces VEGF expression in Smad4-null colon cancer cells [32], suggesting that TGF-beta-mediated activation of Smad4-independent pathways (i.e. MEK-Erk and p38-MAPK, etc) is involved in the up-regulation of VEGF expression, which may cooperate with other pro-oncogenic pathways to promote metastasis. Therefore, the opposite effect of TGF-beta on VEGFA expression could contribute to its differential role as tumor suppressor or tumor promoter in different cell context.
In summary, we have identified a novel mechanism by which TGF-beta suppresses VEGFA expression, angiogenesis and metastasis in a subset of colon cancer cells. This study is significant since it has been shown that expression of VEGFA is increased in colon cancer and is associated with disease localization, stage and long-term disease-specific survival [33]. Avastin, a monoclonal antibody directed against VEGFA, has been widely used with chemotherapy to treat metastatic colon cancer with improved efficacy [34,35]. However, some patients eventually develop resistance to the treatment [36]. Therefore, understanding how VEGFA expression is regulated will facilitate development of novel strategies to effectively target it for cancer treatment. Our studies suggest that activation or restoration of TGF-beta signaling in a sub-population of colon cancer patients may improve their response to anti-VEGFA therapies.