Induction of Angiogenesis by Expression of Soluble Type II Transforming Growth Factor-β Receptor in Mouse Hepatoma*

The biological effect of transforming growth factor-β (TGF-β) is cell type-specific and complex. The precise role of TGF-β is not clear in vivo. To elucidate the regulation mechanism of endogenous TGF-β on hepatoma progression, we modified the MH129F mouse hepatoma cell with a retroviral vector encoding the extracellular region of type II TGF-β receptor (TRII). Soluble TRII (TRIIs) blocked TGF-β binding to TRII on the membrane of hepatoma cells. Growth of MH129F cells was inhibited by TGF-β1 treatment; however, soluble TRII-overexpressing cells (MH129F/TRIIs) did not show any change in proliferation after TGF-β1 treatment. MH129F/TRIIs cells also increased vascular endothelial growth factor (VEGF) expression, endothelial cell migration, and tube formation. Implantation of MH129F/TRIIs cells into C3H/He mice showed the significantly enhanced tumor formation. According to Western blot and protein kinase C assay, the expression of VEGF, KDR/flk-1 receptor, and endothelial nitric-oxide synthase was enhanced, and the phosphorylation activity of protein kinase C was increased up to 3.7-fold in MH129F/TRIIs tumors. Finally, a PECAM-1-stained intratumoral vessel was shown to be 4.2-fold higher in the MH129F/TRIIs tumor. These results indicate that VEGF expression is up-regulated by a blockade of endogenous TGF-β signaling in TGF-β-sensitive hepatoma cells and then stimulates angiogenesis and tumorigenicity. Therefore, we suggest that endogenous TGF-β is a major regulator of the VEGF/flk-1-mediated angiogenesis pathway in hepatoma progression.

Transforming growth factor-␤ (TGF-␤) 1 is the potent mediator of cellular proliferation, differentiation, and production of the extracellular matrix as well as immune response (1). Many types of tumor cells secrete TGF-␤, which regulates their growth by acting as major autocrine and paracrine modulators (2). However, TGF-␤ exerts diverse biological activities in a wide range of experiments depending on the nature of the cell type and physiological conditions. In vitro, TGF-␤ acts as a potent growth inhibitor of some cancer cells; at the same time it can act as a selective growth promoter when tumor cells somehow acquire resistance to TGF-␤-mediated growth inhibition (3)(4)(5)(6)(7)(8). It is assumed that TGF-␤ may play a paradoxical role in vivo by these different processes. On one hand, TGF-␤ can stimulate malignant progression, possibly by its action on cell-matrix interaction or by its immunosuppressive activities in vivo (9,10). On the other hand, it has been shown that TGF-␤ can have tumor suppressor activities, most likely because of its growth-inhibitory activities in vivo (11)(12)(13). Additionally, Chang et al. (14) reported that increased TGF-␤ expression inhibits cell proliferation in vitro yet increases tumorigenicity and tumor growth of Meth A sarcoma cells. Thus, these findings persuade us to investigate the precise and crucial role of TGF-␤ in tumorigenesis.
In evaluating the regulatory function of TGF-␤ in the angiogenesis mechanism, many studies have reported that TGF-␤ induces angiogenesis by the production of extracellular matrix and proteolytic enzymes as well as by the expression of adhesion molecules (15)(16)(17). Inversely, contradictory reports demonstrate that the local administration of neutralizing anti-TGF-␤ antibodies has been observed to induce capillary density in tumors (18). A previous study also indicated that TGF-␤ down-regulates the expression of flk-1, the tyrosine kinase receptor for vascular endothelial growth factor (VEGF) (19,20). This same study also indicated that the down-regulation of flk-1 expression may be responsible for the inhibitory effect of this cytokine on VEGF-induced in vitro angiogenesis (21). These results imply that TGF-␤ is a regulator of the angiogenesis mechanism. However, although there are strong evidences to support the role of TGF-␤ in several aspects of vascular development (9 -13), its role and regulation mechanism in angiogenesis is not exactly known. Moreover, because there are no data available on the significance of endogenous TGF-␤ secreted in tumor cells, we demonstrated how endogenous TGF-␤ regulates the complex mechanism for determining tumor progression and subsequent tumor angiogenesis.
TGF-␤ conveys signals via two major serine/threonine kinase receptors, type I and type II. The type II TGF-␤ receptor (TRII) is the primary receptor target for TGF-␤ (22)(23)(24). It is implied that the alteration of TRII expression may potentially contribute to the disruption of the TGF-␤ signaling pathway. Therefore, to find out what effects would be caused on the tumorigenicity by the loss of endogenous TGF-␤ actions in TGF-␤-sensitive hepatoma cells, we hypothesized that overexpression of the extracellular region of TRII could interrupt the TGF-␤-mediated signal pathway by the blockade of TGF-␤ in tumor progression. In this approach, we used MH129F mouse hepatoma cells secreting a large amount of TGF-␤ and exerting antiproliferation activity to TGF-␤, and then we investigated the new angiogenesis mechanism regulated by TGF-␤.
In the present study, we demonstrate the new finding that the loss of endogenous TGF-␤ effects caused by the overexpression of the extracellular region of TRII in hepatoma cells stimulates VEGF expression and then induces angiogenesis.

EXPERIMENTAL PROCEDURES
Vector Structure and Transduction-The MFG/TRIIs retroviral vector (25) is composed of the extracellular domain (amino acids 1-159) of human TRII (22) and was inserted into the NcoI-BamHI site under the control of the LTR. An internal ribosome entry site and neomycinresistance (neo) genes were used as a backbone (Fig. 1A). The genetic modification of MH129F mouse hepatoma cells was performed by a recombinant retrovirus encoding MFG/TRIIs and control MFG/LTR vector. Briefly, each retroviral vector was transfected into the BOSC23 packaging cell line using LipofectAMINE Plus as indicated by the manufacturer (Life Technologies, Inc.). MH129F cells were incubated for 4 h with recombinant viral supernatant containing 8 g/ml of polybrene (Sigma) and selected after treatment of 800 g/ml geneticin.
Western Blot-To determine the expression pattern of soluble TRII in MH129F, MH129F/LTR, and MH129F/TRIIs, the medium was changed to serum-free medium at confluence. Conditioned medium (CM) was collected at 48 h and precipitated with 5 volumes of acetone at Ϫ70°C for 1 h. The pellet was analyzed by Western blot with goat anti-human TRII antibody (R&D Systems Inc., Minneapolis, MN) and donkey antigoat horseradish peroxidase-conjugated antibody as primary and secondary antibodies, respectively. The blot was processed with the chemiluminescence luminol reagents (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) to visualize the immunoreactive bands. Multiple exposures of films were obtained to determine the optimal exposure time. The method for cell and tissue lysis and Western blot has been described previously (26). The mouse VEGF (Santa Cruz Biotechnology, Inc.), basic fibroblast growth factor (bFGF, Santa Cruz Biotechnology, Inc.), eNOS (Transduction Laboratories Corp., Lexington, KY), flk-1 (Santa Cruz Biotechnology, Inc.), and actin (Santa Cruz Biotechnology, Inc.) antibodies were used as primary antibodies.
Enzyme-linked Immunosorbent Analysis (ELISA)-CM was prepared in each cell (MH129F, MH129F/LTR, and MH129F/TRIIs) cultured in 1% serum-containing medium for 24 h at confluent conditions. The ELISA plate was coated with 5 g/ml TGF-␤1 and incubated with CM at 37°C for 1 h. The human biotinylated TRII antibody was used for the detection of soluble TRII, and the absorbency was read at 405 nm on an ELISA reader. To determine the expression of angiogenic factors VEGF and bFGF, mouse VEGF and bFGF antibodies were used for protein detection. The mean value was determined from three independent experiments, and the measurement was performed in quadruplicate for each experiment.
Cell Growth Assay-MH129F, MH129F/LTR, and MH129F/TRIIs cells were seeded on 24-well plates (10 4 cells/well) and allowed to adhere overnight. After treatment of 5 ng/ml TGF-␤1 (R&D Systems Inc.), the cells were harvested at different time points (24,48, and 72 h) and counted by trypan blue exclusion tests and Coulter Counter (Coulter Corp., Miami, FL). Each experiment was repeated at least four times.
Tube Formation Assay-The spontaneous formation of capillary-like structures by the mouse endothelial cell MS1 (American Type Culture Collection, CRL-2279) on a basement membrane matrix (Matrigel) was used for the assessment of angiogenic potential. 24-well plates were coated with Matrigel (10 mg/ml), and MS1 cells were seeded on coated plate at 1.5 ϫ 10 5 cells/well. The CM was supplemented with the agents as indicated, and the cultures were incubated at 37°C for 24 h. CM was collected in MH129F or MH129F/TRIIs cells cultured in serum-free medium at 48 h after confluent culture, and standard culture was exposed to 2% serum-containing medium alone. Tube formation was observed and photographed over the 24-h period using a camera at-tached to an inverted phase contrast microscope. The degree of tube formation was assessed by counting the number of tubes contained in four random fields from each well.
Endothelial Cell Migration Assay-The migration assay used a monolayer denudation assay as described by Tang et al. (27). Confluent endothelial cells were wounded by scraping with a 2-200-l pipette tip and denuding a strip of the monolayer 300 m in diameter. The variation in wound diameter within experiments was ϳ5%. Cultures were washed twice with PBS and incubated with the CM supplemented with agents as indicated. The rate of wound closure was observed and photographed over a 24-h period. To confirm endothelial cell migration, polycarbonate filters (8-m pore size, Corning Costar Corp., Cambridge, MA) coated with type I collagen were used. MH129F or MH129F/TRIIs cells were grown in the outer chamber and at confluence. The inner chamber was transferred to the outer chamber, and MS1 cells were suspended in serum-free RPMI 1640 medium and seeded in the inner chamber. The cell movement was evaluated toward a gradient of CM of mock or TRIIs-overexpressing MH129F cells. The cells were incubated at 37°C for 4 h. After the incubation period, nonmigrated cells on the upper surface of the filter were removed, and the cells that had migrated onto the filter were counted manually by examination under microscope.
Tumorigenicity-C3H/He mice 4 -6 weeks of age were acclimated and caged in groups of six or fewer. MH129F, MH129F/LTR, or MH129F/TRIIs cells (5 ϫ 10 6 ) in 100 l of PBS were injected subcutaneously into the right flanks of the mice, and the animals were examined daily until the tumor became palpable. The mean tumor volume was measured by dial caliper weekly and calculated by the following formula: volume (mm 3 ) ϭ (square root of width ϫ length) 3 . The experiment was performed four times on six mice in each group, and tumor volumes were measured for 28 days.
Protein Kinase C (PKC) Activity Assay-Tumor tissue was homogenized in a lysis buffer according to a previous study (28) and sonicated three times in ice. The total PKC activity in equal amounts of homogenates was determined by measuring the phosphorylation of C1 peptide (PKC target peptide, PLSRTLSVAAC) provided in the assay kit (Promega, Madison, WI). The homogenate prepared from each tissue (MH129F, MH129F/LTR, and MH129F/TRIIs) was incubated with C1 peptide, 5ϫ reaction buffer (100 mM HEPES (pH 7.4), 6.5 mM CaCl 2 , 5 mM dithiothreitol, 50 mM MgCl 2 , and 5 mM ATP) and PKC activator 5ϫ solution (1 mg/ml phophatidylserine) at 30°C for 30 min. The C1 peptide conjugated to a fluorescent molecule was phosphorylated by PKC activity and run toward the anode direction during electrophoresis on 0.8% agarose gel (50 mM Tris-HCl (pH 8.0)). The phosphorylated bands were scanned by a densitometer after electrophoresis, and the relative intensities were quantified using ImageQuant software (Molecular Dynamics).
Immunohistochemistry and Microvessel Density-At 4 weeks after implantation, the subcutaneous tumors were excised surgically from the mice. Formalin-fixed paraffin-embedded 4-m sections were deparaffinized and rehydrated to PBS. Pretreatments included microwave antigen retrieval in a 10 mM citrate buffer for 20 min. Endothelial cells were detected using the endothelial cell adhesion molecule-1 PECAM-1 antibody (Santa Cruz Biotechnology, Inc.). The immunoperoxidase procedure was conducted as described in the immunocruz staining system, and sections were counterstained with Harris' hematoxylin. Sections labeled with the PECAM-1 antibody were scanned for the localization of vascular hot spots. The 10 most vascular areas were determined and counted for vessel number at high magnification. Stained vessels, which were clearly separated from each other, were counted. For each tumor, at least four independent sections were stained with hematoxylin and eosin and immunostained for blood vessels. Negative control slides consisted of excluding the primary antibody but retaining all the other steps.

RESULTS
First, we examined the TGF-␤ expression in MH129F mouse hepatoma cells by using reverse transcription-polymerase chain reaction and ELISA techniques. These results showed that MH129F cells expressed at both mRNA and protein levels (data not shown). Most studies involving the inhibition of TGF-␤ activity have depended on using neutralization antibodies to TGF-␤ and overexpression of the extracellular domain of TRII (18,25). Therefore, we used the approach of inducing expression of soluble TRII in hepatoma cells for binding to TGF-␤ and blocking its action. The retroviral vectors (29) encoding the extracellular domain (amino acids 1-159) of human TRII (MFG/TRIIs, Fig. 1A) and encoding only the neomycin resistance gene (MFG/LTR) were used for genetic modification of MH129F cells (MH129F/TRIIs and MH129F/LTR). To confirm the integration of the retroviral vector into chromosomal DNA, we amplified the fragment of the human soluble TRII region by genomic polymerase chain reaction. MH129F/TRIIs cells had a 0.48-kilobase pair DNA fragment encoding the extracellular domain of TRII in genomic DNA (Fig. 1B). Western blot showed the secretion of a 25-35-kDa heterogeneous glycosylated TRIIs into the CM (Fig. 1C). Because the human TRIIs antibody has less than 5% cross-reactivity with the mouse TRII protein, unclear bands were detected in MH129F and MH129F/LTR cells. According to ELISA analysis, the expression level of soluble TRII in MH129F/TRIIs cell was increased up to 22-30-fold compared with MH129F and MH129F/LTR cells (Fig. 1D). Absorbance in MH129F and MH129F/LTR cells did not show any difference with negative control (1% serum-containing medium). In addition, this fact indicates that truncated, soluble TRII retains its binding ability to TGF-␤ ligands. Therefore, soluble TRII has the capability of competing against a membrane-bound receptor and then to block the various actions of TGF-␤.
To test the sensitivity of MH129F and MH129F/TRIIs cells to TGF-␤1, we determined the effect of TGF-␤1 on cell proliferation. TGF-␤1 markedly inhibited MH129F cell growth in a time-dependent manner ( Fig. 2A). However, soluble TRII-overexpressing MH129F cells did not induce the growth inhibition after TGF-␤1 treatment (Fig. 2B). To compare with the characteristics of MH129F and MH129F/TRIIs cells at the molecular level, the expression of angiogenic factors such as VEGF and bFGF was measured by immunoblotting and ELISA analysis. Using Western blot analysis, the elevation of VEGF protein level was observed in MH129F/TRIIs cells compared with MH129F cells (Fig. 3A). Also, VEGF secretion in MH129F/ TRIIs cells increased by 6-fold compared with control cells (Fig.  3B). In contrast, we were unable to detect the change of bFGF expression in the MH129F/TRIIs cells (Fig. 3).
VEGF has been reported to contain various biological activities, e.g. it increases vascular permeability (20,30), stimulates the proliferation and migration of endothelial cells (31), and also induces tumor angiogenesis (30,32,33). Therefore, we performed the three-dimensional in vitro angiogenesis system, tube formation, on Matrigel. MS1 mouse endothelial cells were suspended on a Matrigel plate and incubated with various media for 24 h (Fig. 4). MS1 cells exposed the standard medium with 2% serum forming a tubular network as positive control (Fig. 4A), but the CM of MH129F inhibited the tube formation of MS1 cells (Fig. 4B). Interestingly, the addition of CM of MH129F/TRIIs cells formed a long cytoplasmic process and an extensive tubular network (Fig. 4C). To determine whether this potent tube formation was induced by evaluation of VEGF expression in MH129F/TRIIs cells (Fig. 3), we added the specific antibody, which prevented effects of VEGF in CM of MH129F/TRIIs. VEGF blockade in CM of MH129F/TRIIs by specific antibody inhibited tube formation (Fig. 4D), but the addition of bFGF antibody did not influence tube formation (Fig. 4E). Thus, these results indicate that induction of VEGF expression in MH129F/TRIIs cells enhances tube formation.
Tube formation in three-dimensional cultures depends on the migration of cells after plating (34). To assess the impact of MH129F/TRIIs CM on endothelial cell migration, a denudation injury model was studied, and polycarbonated filters coated with type I collagen were used to confirm the endothelial cell migration. Confluent scrape-wounded endothelial cell monolayers were incubated with various media, and the rate of closure was observed over the following 24 h (Fig. 5). MS1 cells exposed the standard medium with 2% serum migrated into the denuded area (Fig. 5A), but the CM of MH129F inhibited the migration of MS1 cells into the wounded area (Fig. 5B). Also the CM of MH129F/TRIIs cells significantly enhanced migration of endothelial cells (Fig. 5C). VEGF blockade by the specific antibody inhibited endothelial cell migration into the wounded area (Fig. 5D), but addition of the bFGF antibody did not affect the migration ability of MH129F/TRIIs CM (Fig. 5E). Thus, we found that VEGF induction in MH129F/TRIIs cells also might enhance the endothelial cell migration.
To investigate the role of endogenous TGF-␤ in tumor progression, we inhibited the action of endogenous TGF-␤ by the expression of the extracellular region of TRII in tumor cells and evaluated the consequences on tumor progression in vivo. 5 ϫ 10 6 cells (MH129F, MH129F/LTR, or MH129F/TRIIs) were injected subcutaneously into the right flank of the syngeneic C3H/He mouse. When the tumor burden was measured at day 28, MH129F and MH129F/LTR cells produced tumors of 4,404 Ϯ 712 and 4,652 Ϯ 1,551 mm 3 , respectively, whereas MH129F/TRIIs cells had produced markedly large tumors of 22,976 Ϯ 3,381 mm 3 in volume (Fig. 6A). We also showed the enhancement of vessel formation at 28 days after mouse implantation (Fig. 6B). The result of this animal experiment reveals that the expression of soluble TRII augments in vivo tumor growth of TGF-␤-sensitive hepatoma cells. In Fig. 6, we found that the tumor progression was proceeded rapidly in mice bearing MH129F/TRIIs by the loss of TGF-␤ effects. Tumor angiogenesis is permissive for tumor growth by the newly formed vessels and is regulated by a balance between positive and negative regulators. One of the elements likely to play a key role in this equilibrium is the regulation of expression of endothelial cell tyrosine kinase receptors (35,36). Therefore, we hypothesized that the induction of tumorigenicity by the loss of endogenous TGF-␤ action might be mediated by VEGF/ flk-1-induced signaling pathway. To explain this phenomenon, we examined the expression pattern of flk-1 and eNOS protein in hepatoma tissue by Western blot. The MH129F/TRIIs tumor remarkably induced the expression of flk-1 and eNOS protein (Fig. 7A). To confirm the induction of eNOS expression, we carried out another study regarding the downstream PKC activity of the flk-1 receptor. Total PKC activity in equal amounts of tissue homogenates was determined by measuring the phosphorylation of PKC target peptide. After electrophoresis, phosphorylated peptide was separated by the charge difference. The total PKC in MH129F/TRIIs tumor increased the phosphorylation of peptide by blockade of TGF-␤ action (Fig. 7B), and also the relative phosphorylation density was increased almost 3.7fold compared with control tumor (Fig. 7C). Therefore, these results imply that TGF-␤ may play the role of a regulator in VEGF/flk-1-mediated in vivo angiogenesis by the regulation of VEGF, flk-1 receptor, eNOS expression, and PKC activity.
Finally, we measured the endothelial cell population in each tumor generated by MH129F and MH129F/TRIIs cells. Toward this aim, tumor sections were stained with an anti-mouse PECAM-1 polyclonal antibody that detects vascular endothelial cells in the tumor. MH129F/TRIIs tumors showed a signif-icantly higher number of vessels compared with parental tumors (Fig. 8). The intratumoral blood vessel was determined by counting the PECAM-stained vessels in 10 different fields for each tumor. The vascular density was 4.2 times greater in MH129F/TRIIs tumors than in control tumors (Table I; p Ͻ 0.005). Taken together, these findings indicate that the blockade of endogenous TGF-␤ induces VEGF expression by the breaking of angiogenesis balance and then stimulates hepatoma progression by induction of angiogenesis. DISCUSSION In light of the well known fact that TGF-␤ regulates hepatocyte growth negatively, there is a discrepancy between high serum TGF-␤ levels and the high proliferating rate in hepatocellular carcinoma cells (37)(38)(39). This discrepancy implies that the change of growth response to TGF-␤ at the cellular level is the more important step in malignant progression. Although in recent years much attention has been focused on the biology of TGF-␤, in vivo studies of the complex tumor network have not yet yielded a clear understanding about the regulation mechanisms of TGF-␤ in the angiogenesis pathway. Therefore, we assumed that TGF-␤ may play a crucial role in the tumor progression process of TGF-␤-sensitive tumor cells and thus investigated the effects of endogenous TGF-␤ by the loss of TGF-␤ action in the phase of angiogenesis of mouse hepatoma progression.
Numerous tumor types and cultured tumor cells express TGF-␤ and its receptor but are resistant to its antiproliferative effects because of deletion (3)(4)(5), reduced expression (6), and mutation (7) of the TRII in most progressive tumor cells (2)(3)(4)(5)(6)(7)(8). Therefore, many studies have suggested the alteration of the TRII for the understanding of the mechanisms induced by the loss of TGF-␤ sensitivity in highly progressive tumors (23)(24)(25)40). In this study, we hypothesized that the overexpression of soluble TRII would interrupt TGF-␤ action through competition against wild-type TRII (Fig. 1). Soluble TRII-overexpressing MH129F/TRIIs cells could escape the growth inhibitory function (41,42) of TGF-␤ in hepatoma cells (Fig. 2), but cell morphology and doubling time of MH129F/TRIIs cells were similar to that of MH129F control cells. Especially at the molecular level, the elevation of VEGF expression was detected by the loss of endogenous TGF-␤ action (Fig. 3) and enhanced the tube formation (Fig. 4) and migration of endothelial cells (Fig.  5). These novel findings imply that the enhancement of tumor progression in TGF-␤-resistant tumor cells (5-8) may be mediated by the change in VEGF expression.
According to the tumorigenicity study, MH129F/TRIIs cells displayed a strong increase in tumor formation compared with the parent cells (Fig. 6). These results are consistent with the reports that the loss of TGF-␤ expression in benign skin pap- illomas is correlated with a high risk for malignant conversion (11)(12)(13). Based on the induction of MH129F/TRIIs tumorigenicity, one would expect that the blockade of endogenous TGF-␤ may induce the angiogenesis in the MH129F/TRIIs tumorigenesis. First, evidence of angiogenic profile in our data indicated that a loss of TGF-␤ effects induced tubule formation and migration of endothelial cells by stimulation of VEGF expression (Figs. [3][4][5]. Second, the induction of in vivo angiogenesis on MH129F/TRIIs tumor progression was established by various approaches (Figs. 6 and 7): 1) The induction of flk-1 expression by the loss of endogenous TGF-␤ effects activated downstream signaling cascade and prompted the expression of eNOS. 2) PKC plays an important role in angiogenic effects (43,44). Increased PKC activity by the blocking of TGF-␤ effects followed eNOS up-regulation. These findings are consistent with the observations that TGF-␤ down-regulates the expression of flk-1, the tyrosine kinase receptor for VEGF, and reduces the baseline expression of eNOS, the downstream PKCdependent pathway (19,20,28). Also, the expression of soluble TRII induced the formation of intratumoral vessel according to the proliferation of endothelial cells ( Fig. 8 and Table I). Therefore, these results imply that endogenous TGF-␤ is a potent regulator in proliferation of endothelial cells in vivo.
In this presentation, we present the detailed and novel function of antiangiogenic mechanism of endogenous TGF-␤ in hepatoma progression. In conclusion, these results suggest that endogenous TGF-␤ acts as a major regulator of the VEGF/flk-1-mediated angiogenesis pathway in TGF-␤-sensitive hepatoma development.

TABLE I
Induction of blood vessel density in the MH129F/TRIIs tumors 5 ϫ 10 6 MH129F, MH129F/LTR, or MH129F/TRIIs cells were implanted into C3H/He mice. 4 weeks after the injections the mice were killed, and tumors were removed as described in the legend to Fig. 6. Tumor sections were stained using a PECAM-1 antibody as described under "Experimental Procedures." The number of intratumoral blood vessels was counted in 10 different fields for each tumor. Mean Ϯ S.D. values are shown.