Inhibition of Growth and Metastasis of Mouse Mammary Carcinoma by Selective Inhibitor of Transforming Growth Factor-β Type I Receptor Kinase In vivo

Materials and Methods

Reagents and antibodies. Human recombinant TGF-β1 (1 μg/mL; Austral Biologicals, San Ramon, CA) was dissolved in 4 mmol/L HCl and 1 mg/mL bovine serum albumin (Sigma, St. Louis, MO). SD-093 and SD-208 are selective chemical inhibitors of the TβRI receptor kinase that inhibit cellular responses to TGF-β with an IC₅₀ of 20 and 80 nmol/L, respectively (42, 44, 45). For in vitro studies, SD-093 (Scios, Inc., Sunnyvale, CA) was dissolved in DMSO (Sigma) and stored at −70°C. For in vivo studies, SD-208 was suspended in 1% (w/v) methylcellulose in water. Rabbit polyclonal antibodies directed against Smad2 and Smad3 were obtained from Zymed Laboratories (South San Francisco, CA), and a mouse monoclonal antibody directed against Smad4 was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit polyclonal antibodies directed against phosphorylated Smad2 (pSmad2) and phosphorylated Smad3 (pSmad3) were produced in our laboratory (46, 47). A monoclonal antimouse E-cadherin antibody was obtained from BD Transduction Laboratories (Bedford, MA). Alexa Fluor 488–conjugated phalloidin was obtained from Molecular Probes (Eugene, OR).

Cell culture. NMuMG, a spontaneously immortalized, nontumorigenic cell line derived from a normal murine mammary gland, was obtained from Dr. Daniel DiMaio (Yale University, New Haven, CT). NMuMG cells were maintained in high-glucose (4.5 g/L) DMEM (Life Technologies, Grand Island, NY) supplemented with 10% (v/v) fetal bovine serum (Sigma), 10 μg/mL insulin (Sigma), and 10 μg/mL gentamicin (Life Technologies). MDA-MB-435 human breast carcinoma cells (derived from a malignant pleural effusion) were maintained in DMEM/F-12 (Invitrogen, Carlsbad, CA) supplemented with 10% (v/v) fetal bovine serum, pyruvate (10 mmol/L; Life Technologies), and nonessential amino acids (1 mmol/L; Life Technologies). R3T mammary carcinoma cells (48) were maintained in MEM-α (Invitrogen) supplemented with 8% (v/v) fetal bovine serum with the addition of 3 μg/mL puromycin (Sigma), 800 μg/mL geneticin (G418; Invitrogen), and 10 μg/mL gentamicin (Invitrogen). 4T1 cells (refs. 49, 50; kindly provided by Dr. Fred Miller, Michigan Cancer Foundation, Detroit, MI) were maintained in DMEM supplemented with 10% (v/v) fetal bovine serum. TMLC cells (kindly provided by Dr. Daniel Rifkin, New York University, New York, NY) were maintained in RPMI 1640 (Life Technologies) supplemented with 10% (v/v) fetal bovine serum with the addition of 200 μg/mL geneticin (G418).

Cell proliferation assays. Cells were plated at 2 × 10⁴ per well in 24-well cluster dishes (Corning, Inc., Corning, NY). Twenty-four hours later, cultures were treated with 1 μmol/L SD-093 or vehicle only followed 15 minutes later by addition of 100 pmol/L (2.5 ng/mL) TGF-β1. Cells were counted 72 hours later using a Beckman Coulter counter (Model 0039, Beckman, Inc., Miami, FL).

Epithelial-to-mesenchymal transdifferentiation. For assessment of subcellular F-actin fiber distribution, cells were washed with PBS and fixed with buffered formalin for 10 minutes. Following washes with PBS, cells were permeabilized using 0.1% (v/v) Triton X-100 in PBS for 5 minutes and then incubated with 0.165 μmol/L Alexa Fluor 488–conjugated phalloidin with 1% (w/v) bovine serum albumin in PBS for 20 minutes at 20°C in the dark. For E-cadherin immunostaining, cells were washed with PBS and fixed for 5 minutes using methanol precooled to −20°C. Air-dried slides were then incubated with 5% (v/v) goat serum for 20 minutes at room temperature followed by incubation with 2 μg/mL mouse monoclonal anti-E-cadherin antibody in 2.5% (v/v) goat serum for 1 hour at room temperature. The cells were then washed 3 × 5 minutes with PBS followed by incubation with 2 μg/mL rhodamine-conjugated goat anti-mouse IgG for 45 minutes in the dark. In both cases, stained dishes were then washed 3 × 5 minutes with PBS, mounted using Vectashield mounting medium (Vector Laboratories, Inc., Burlingame, CA), and viewed using a Zeiss epifluorescence microscope (model 090477, Carl Zeiss, Microimaging, Inc., Chester, VA) equipped with a MTI charge-coupled device camera (model DC 330E, DAGE-MTI, Inc., Michigan City, IN).

Western blot analysis. For detection of Smad proteins, semiconfluent cell cultures were lysed in situ using buffer composed of 150 mmol/L NaCl, 10 mmol/L Tris-HCl (pH 8.0), 1 mmol/L EDTA, 1 mmol/L EGTA, and 1% (v/v) Triton X-100 in the presence of protease inhibitors (Complete Mini Protease Inhibitor Cocktail Tablets, Roche Diagnostics Corp., Indianapolis, IN) for 30 minutes at 4°C. Cell lysates were subjected to Western blot analysis as described previously (42). Activated Smad2 (pSmad2) and activated Smad3 (pSmad3) were detected using our own rabbit anti-pSmad2 and anti-pSmad3 antibodies at a 1:1,000 dilution. Total Smad2, Smad3, and Smad4 were detected using rabbit anti-Smad2, rabbit anti-Smad3, and mouse anti-Smad4 antibodies, respectively, at a 1:500 dilution.

In vitro cell migration and invasion assays. For migration assays, uncoated polyethylene terephthalate track etched membrane (24-well insert; pore size, 8 μm; Becton Dickinson, Franklin Lakes, NJ) inserts were equilibrated by adding 0.5 mL cell culture medium to the upper and lower chambers followed by incubation at 37°C for 2 hours. For invasion assays, BD Biocoat Growth Factor Reduced Matrigel Invasion Chambers (24-well insert; pore size, 8 μm; BD Biosciences, Bedford, MA) were rehydrated by adding 0.5 mL warm (37°C) culture medium to the upper chambers followed by 2-hour incubation at 37°C. For both assays, medium used for equilibration was removed and 10⁵ cells were plated in the upper chambers. TGF-β1 (100 pmol/L, 2.5 ng/mL), SD-093 (1 μmol/L), both agents, or vehicle only were added to both upper and lower chambers. Following a 24-hour incubation at 37°C, cells in suspension were aspirated, the inserts were washed twice with PBS, and the cells adherent to the top of the inserts were removed by scraping the upper surface of the membrane with a cotton tip applicator. The cells that had migrated to the underside of the inserts were fixed and stained using the DiffQuick (Dade Behring, Newark, DE) staining kit. Cells in 10 random squares of 100 × 100 μm in each well were counted at ×200 magnification using quadruplicate wells per assay condition and the results were expressed as number of cells/mm².

Plasminogen activator inhibitor/luciferase assay. The plasminogen activator inhibitor (PAI)/luciferase assay was done as described by Abe et al. (51) with minor modifications. Briefly, 1.6 × 10⁴ TMLC cells suspended in 100 μL medium were plated in 96-well tissue culture dishes and allowed to attach for 3 hours at 37°C in a 5% CO₂ incubator. After 3 hours, the medium was replaced with 100 μL medium containing either TGF-β at concentrations ranging from 0 to 100 pmol/L (0-2.5 ng/mL) or a suspension of 3.2 × 10⁴ 4T1 or R3T cells. The cells were lysed 14 hours later at room temperature and luciferase activity in cell lysates was determined using the Promega Luciferase Reporter Assay System following the protocol recommended by the manufacturer (Promega, Madison, WI) using a TD-20/20 Luminometer (Turner Designs, Sunnyvale, CA).

Animal experiments. Viral antibody-free 8- to 9-week-old female athymic (Harlan Laboratory, Indianapolis, IN) 129S1 and BALB/cJ (The Jackson Laboratory, Bar Harbor, ME) were weighed and randomly assigned to different treatment groups. Tumor cells (1 × 10⁶) were injected orthotopically into the right second or third fat pads (52). Mice were then treated with single daily 0.2 mL doses of vehicle [1% (w/v) methylcellulose, SD-208 (20 mg/kg), or SD-208 (60 mg/kg)] by gavage beginning 1 day following tumor cell inoculation. Animal body weight and tumor sizes were measured thrice weekly. R3T cells were inoculated into 129S1 or athymic nude mice, whereas 4T1 cells were injected into syngeneic BALB/cJ mice. Primary tumors were resected from athymic nude mice 18 days after R3T cell injection or from BALB/cJ mice 18 days after 4T1 cell injection, and animals continued to be treated postoperatively for the indicated periods. Blood was collected 2 hours following the penultimate dose to determine plasma levels of SD-208. At sacrifice, liver and body weight were measured. Lungs, liver, kidneys, adrenal glands, and major lymph node groups were visually inspected for the presence of tumor metastases and then processed for routine histology. The burden of lung metastases was estimated by digitally imaging H&E-stained cross-sections of fixed and paraffin-embedded lung preparations and measuring the areas of individual metastases using IP Lab (version 3.6.5, Scanalytics, Inc., Billerica, MA) image analysis software.

Pharmacokinetic analysis. Descriptive pharmacokinetic variables were determined by standard model-independent methods. Noncompartmental analysis was done using WinNonlin version 4.0.1 (Pharsight, Mountain View, CA). Because individual mice were not used to describe a full profile, variables were calculated using mean data. Samples with SD-208 concentration below quantifiable limits (10 ng/mL) were assigned the value of 0 for the analysis. Nominal time points were used for all calculations. AUC_(0-8) is the area under the plasma concentration-time curve from time 0 to 8 hours for animals dosed with SD-208.

Cell proliferation, apoptosis, and angiogenesis. Tissue sections were deparaffinized, rehydrated, and stained with H&E, rat antimouse monoclonal CD34 IgG2a (1:100; CL8927AP; Cedarlane, Hornby, British Columbia, Canada), or rabbit polyclonal anti-Ki-67 (1:100; ab833-500; Novus Biologicals, Littleton, CO). Biotinylated secondary antibodies (1:150; Zymed Laboratories) were used for detection. Apoptotic cells were identified by terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) assay using the In situ Cell Death Detection kit (Roche Molecular Biochemicals, Palo Alto, CA). Streptavidin-alkaline phosphatase (1:100) was added, and the staining was developed with naphtol as substrate and levamisole as inhibitor of endogenous alkaline phosphatase (Fast Red Tablets; Roche Molecular Biochemicals). The negative control for CD34 was normal rat IgG2a (CBL605; Chemicon International, Temecula, CA). The negative control for Ki-67 staining was normal rabbit IgG (SC-2027; Santa Cruz Biotechnology). The total number of CD34⁺ microvessels were counted in five randomly selected high-power (×400) fields in areas of viable tumor. To assess the percentage of proliferating cells, the proportion of Ki-67-positive nuclei was determined. At least 600 nuclei were counted in five randomly selected high-power (×400) fields in areas of viable tumor. To assess the degree of apoptosis, TUNEL-positive cells were counted in the tumor in five randomly selected high-power (×400) fields in areas of viable tumor.

CTL activity. Splenocytes (7 × 10⁶) harvested from 4T1 tumor-bearing mice were restimulated with 1.4 × 10⁵ irradiated 4T1 cells in a 24-well plate in a total of 2 mL tissue culture medium in the presence of 50 μmol/L β-mercaptoethanol and 1 μmol/L SD-093. Cultures were maintained for 5 days at 37°C, 5% CO₂, after which nonadherent effector cells were harvested. 4T1 cells were labeled in 100 μCi ⁵¹Cr (Perkin-Elmer Life and Analytical Sciences Inc., Wellesley, MA) for 1 hour at 37°C and then washed thrice with warm medium. ⁵¹Cr-labeled 4T1 target cells (1 × 10⁴) and effector cells were then added at various E:T ratios in a total of 200 μL TCM to 96-well round-bottomed plates. Plates were incubated for 4 hours at 37°C, 5% CO₂, 100 μL supernatant was removed, and released ⁵¹Cr was quantified using a gamma counter (Packard Bioscience, Meriden, CT). Percentage of specific lysis was calculated using the formula: (experimental release − spontaneous release) × 100 / (maximal release in 5% Triton X-100 − spontaneous release).

Real-time Taqman PCR. Transcript levels of individual genes were assayed in frozen R3T tumor tissue specimens using quantitative real-time PCR on an ABI Prism 7700 Sequence Detector (Applied Biosystems, Foster City, CA) as described previously (44).

Microarray gene expression profiling. RNA was extracted from snap-frozen livers using Trizol reagent (Invitrogen) and purified using RNeasy mini columns (Qiagen, Valencia, CA) according to the manufacturer's instructions. The RNA concentration was adjusted to 1 μg/μL and its quality was assessed on a RNA chip using an Agilent Bioanalyzer (Agilent Technologies, Palo Alto, CA). Isolated total RNA was processed as recommended by Affymetrix, Inc. (Santa Clara, CA). Biotin-labeled cRNA was size fractionated to 35 to 200 bases long using Affymetrix protocols and hybridized to the mouse genome 430 2.0 GeneChip set at 45°C for 16 hours in an Affymetrix GeneChip Hybridization Oven 320. This chip was designed using the most recent publicly available draft of the mouse genome and contains >45,000 probe sets covering >39,000 transcripts and variants representing >34,000 well-substantiated mouse genes. Each GeneChip was then washed and stained with streptavidin-phycoerythrin using an Affymetrix Fluidics Station 400 and scanned on a Affymetrix Gene Array scanner. Scanned image files were analyzed using the Microarray Suite 5.0 software (Affymetrix). Scaling and normalization were carried out using the 500 Normalization Control probe set included on the 430 2.0 GeneChip set.

Statistical analyses. For analyses of tumor growth rates, metastasis burden, microvessel densities, Ki-67 staining, apoptosis rates, and mRNA expression levels one-way ANOVA tests were done using InStat version 3.0 (GraphPad Software, Inc., San Diego, CA) or two-way repeated-measures ANOVA tests using JMP version 5.1 (SAS Institute, Inc., Cary, NC).

Results

Effects of TβRI kinase inhibitor on mammary carcinoma cell growth. TGF-β is a known inhibitor of cell cycle progression of epithelial cells. As shown in Fig. 1A , TGF-β strongly inhibited growth of NMuMG cells in a TβRI kinase-dependent manner. In contrast, growth of R3T and 4T1 mammary carcinoma cells was not significantly inhibited by exogenous TGF-β. Moreover, treatment with the TβRI kinase inhibitor SD-093 by itself did not affect growth of NMuMG or R3T cells and caused only a small but statistically significant inhibition of 4T1 cell growth (unpaired t test with Welch correction).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

A, effects of TβRI kinase inhibitor on anchorage-dependent growth. R3T and 4T1 cells were treated with vehicle control, 100 pmol/L (2.5 ng/mL) TGF-β, 1 μmol/L SD-093, or 100 pmol/L TGF-β + 1 μmol/L SD-093 for 72 hours after attachment. TGF-β strongly inhibited the growth of NMuMG cells, and this effect was effectively blocked by pretreatment with SD-093. In contrast, TGF-β had no significant effect on the growth of R3T and 4T1 compared with vehicle control. However, treatment with SD-093 alone caused a small but statistically significant inhibition of 4T1 cell growth. Columns, mean of four separate experiments, triplicate wells per condition; bars, SE. Results for each cell line were analyzed by one-way ANOVA (NMuMG: P < 0.0001, R3T: P = 0.2469, and 4T1: P = 0.0074). B, effects of TβRI kinase inhibitor on EMT. Phase-contrast pictures (magnification, ×100) show that the R3T and 4T1 become spindeloid following TGF-β (100 pmol/L, 2.5 ng/mL) treatment, which can be completely blocked by pretreatment with SD-093 (1 μmol/L). Treatment with TGF-β resulted in acquisition of actin stress fibers (magnification, ×400) and loss of E-cadherin (magnification, ×200) from cell margins. Pictures were taken using a Zeiss epifluorescence microscope attached to a MTI charge-coupled device camera. C, effects of TβRI kinase inhibitor on cell migration in vitro. Cells were plated onto polyethylene terephthalate membranes as described in Materials and Methods and incubated for 24 hours with TGF-β (100 pmol/L, 2.5 ng/mL), SD-093 (1 μmol/L), or both. TGF-β strongly inhibited motility of NMuMG cells, an effect that was completely blocked by pretreatment of cells with SD-093. In contrast, TGF-β significantly stimulated motility of both R3T and 4T1 cells, and pretreatment with SD-093 effectively blocked this effect. In fact, in R3T cells, SD-093 treatment strongly inhibited cell motility compared with vehicle treatment, independently of the presence of exogenous TGF-β (P < 0.001). Columns, mean number of cells/mm² of four separate experiments, triplicate wells per condition; bars, SE. Results for each cell line were analyzed by one-way ANOVA (NMuMG: P < 0.0001, R3T: P < 0.0001, and 4T1: P < 0.0001). D, effects of TβRI kinase inhibitor on invasion in vitro. Cells were plated onto (Growth Factor Reduced) Matrigel Invasion cell culture inserts as described in Materials and Methods and incubated for 24 hours with TGF-β (100 pmol/L, 2.5 ng/mL), SD-093 (1 μmol/L), or both. TGF-β significantly stimulated the invasive capacity of both R3T and 4T1 cells, an effect that was blocked by preincubation of the cells with SD-093. Columns, mean number of cells/mm² of three separate experiments, triplicate wells per condition; bars, SE. Results for each cell line were analyzed by one-way ANOVA (R3T: P = 0.0034 and 4T1: P = 0.0001). E, mammary carcinoma cells activate PAI-1 promoter activity in neighboring TMLC cells. To generate a standard curve, 1.6 × 10⁴ TMLC cells were plated in 96-well tissue culture dishes and allowed to attach for 3 hours at 37°C in a 5% CO₂ incubator. After 3 hours, the medium was replaced with 100 μL medium containing TGF-β at the indicated concentrations. Luciferase activity was determined in cell lysates 14 hours later. Left, TGF-β induced activation of the PAI-1 promoter and luciferase expression in a dose-dependent manner. To determine whether mammary carcinoma cells produce and/or activate TGF-β in their microenvironment, TMLC cells were coincubated with 4T1 or R3T cells, or vehicle only, in the presence or absence of 1D11 murine pan-TGF-β-neutralizing antibody. Luciferase activity was determined in cell lysates 14 hours later. Points, mean of two independent experiments, triplicate wells per condition; bars, SD. Conditions were compared pairwise by Student's t test. Both cell lines induced activation of the PAI-1 promoter and luciferase expression in neighboring TMLC cells. In both cases, this effect was blocked in the presence of TGF-β-neutralizing antibody.

TβRI kinase inhibitor blocks TGF-β-induced EMT. As shown in Fig. 1B, TGF-β induced EMT in NMuMG cells as manifested by spindle-cell morphology, reduced cell-cell cohesion, and cellular redistribution of F-actin and E-cadherin. These effects were blocked by pretreatment with the TβRI kinase inhibitor SD-093. Both mammary cancer cell lines displayed some degree of EMT even in the absence of exogenous TGF-β. 4T1 cells were less cohesive than R3T cells presumably because they express little or no E-cadherin at their cell surface (Fig. 1B; ref. 53). As treatment with SD-093 by itself caused cells to assume a more epithelioid phenotype than untreated cells, the basal level of EMT seems to be dependent on endogenous TGF-β signaling. Conversely, EMT became more pronounced with the addition of exogenous TGF-β, an effect that was blocked by treatment with SD-093. EMT was most clearly detectable at 24 to 36 hours following the addition of TGF-β. Thus, in these two mammary carcinoma cell lines, the ability of TGF-β to induce EMT has been retained, whereas its ability to suppress growth has been lost.

TβRI kinase inhibitor affects cell migration and invasion. Several studies have suggested that tumor cell migration and invasion might be driven by TGF-β (29, 54, 55). As shown in Fig. 1C, migration of NMuMG cells was strongly inhibited by TGF-β in a TβRI kinase-dependent manner. In contrast, under the same culture conditions, TGF-β stimulated migration of both mammary carcinoma lines, and this effect was also dependent on TβRI kinase activity. Moreover, treatment with SD-093 alone inhibited motility of R3T cells, indicating that the migratory phenotype of these cells is driven, at least in part, by endogenous TGF-β signaling.

Besides cell migration, treatment with exogenous TGF-β significantly stimulated the ability of R3T and 4T1 cells to invade Growth Factor Reduced Matrigel (3.3- and 2-fold, respectively). This effect was completely inhibited by pretreating the cells with SD-093 (Fig. 1D). Moreover, treatment of R3T cells with SD-093 alone inhibited invasion by 40%, indicating that their invasive phenotype is also partly dependent on endogenous TGF-β signaling.

To investigate whether excessive production and/or extracellular activation of TGF-β was responsible for activation of TGF-β signaling, we conducted coculture experiments using TMLC cells that express a TGF-β-inducible PAI-1 promoter driving a luciferase reporter gene construct (51). As shown in Fig. 1E, exogenous TGF-β resulted in a concentration-dependent increase in luciferase activity in TMLC cells. Cocultivation of TMLC cells with mammary carcinoma cells induced a significant increase in luciferase activity (Fig. 1E). In both cases, the effect was abolished in the presence of pan-TGF-β antibody, thereby showing that both mammary cancer cell lines produce biologically active TGF-β capable of inducing specific gene responses in neighboring cells.

SD-208 inhibits mammary tumor growth and metastasis in vivo. Because of its superior oral bioavailability compared with SD-093, the SD-208 compound was used for all in vivo experiments. In a typical assay, R3T or 4T1 carcinoma cells were inoculated into the mammary fat pad, and mice were treated with single daily doses of vehicle, SD-208 (20 mg/kg), or SD-208 (60 mg/kg) beginning 1 day after tumor cell inoculation. As shown in Fig. 2A , SD-208 treatment inhibited primary R3T tumor growth in syngeneic 129S1 mice in a dose-dependent manner. In addition, both the number and the size of lung metastases were significantly reduced by SD-208 treatment (Fig. 2B). The antitumor activity of SD-208 was not limited to the R3T mammary carcinoma: SD-208 also caused a dose-dependent statistically significant growth delay of 4T1 orthotopic tumors in syngeneic BALB/c mice (Fig. 2C) as well as a dose-dependent reduction in the number of metastases to the lungs (Fig. 2D) and other organs, including liver, adrenal glands, stomach, and retroperitoneal lymph nodes (Fig. 2D, inset). When R3T cells were inoculated into athymic nude mice, the rate of growth of primary tumors was substantially higher than in 129S1 mice (Fig. 3A ). Consequently, the experiments had to be terminated sooner than in 129S1 mice, possibly accounting for the lower number of lung metastases found at autopsy (Figs. 2B and 3B). In contrast to the antitumor effects seen in syngeneic mice, SD-208 failed to retard the growth of R3T tumors in athymic nude mice (Fig. 3A). In addition, treatment with SD-208 had little effect on the size or numbers of lung metastases (Fig. 3B).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Effects of SD-208 on R3T and 4T1 mammary carcinomas in vivo. R3T (A and B) or 4T1 (C and D) cells (1 × 10⁶) were injected orthotopically into the right second or third mammary fat pads of 8- to 9-week-old 129S1 or BALB/c mice, respectively (8-10 mice per group). Mice were treated with single 0.2 mL daily doses of vehicle [1% (w/v) methylcellulose], SD-208 (20 mg/kg), or SD-208 (60 mg/kg) by gavage starting 1 day after tumor cell inoculation. Fat pad tumor sizes were measured thrice weekly. Primary tumors were resected when they reached a size of ∼1,000 mg. Median tumor weights for each treatment group. Bars, SE. Animals continued to be treated postoperatively and were autopsied at the end of the experiment. At that time, lungs, liver, kidneys, adrenal glands, and major lymph node groups were visually inspected for the presence of tumor metastases and then processed for routine histology. Lung metastases were quantified by digitally imaging H&E-stained cross-sections of fixed and paraffin-embedded lung preparations and measuring the areas of individual metastases using IP Lab version 3.6.5 image analysis software. SD-208 treatment inhibited both primary and metastatic tumor growth in 129S1 mice. A, R3T primary tumor growth. Treatment with SD-208 caused significant primary tumor growth delay (P = 0.0026, univariate test of drug effect over time, repeated-measures multivariate ANOVA). B, R3T lung metastases. The number of metastases per group of surviving mice (left) and the sizes (right, box plot; boxes, median values with upper and lower quartiles; filled square, mean; whiskers, 90th and 10th percentiles) of lung metastases in SD-208-treated mice were significantly smaller than in vehicle-treated control animals (P < 0.001 for either 20 or 60 mg/kg/d SD-208 compared with vehicle, Kruskal-Wallis nonparametric ANOVA). C, 4T1 primary tumor growth. Treatment with SD-208 caused significant primary tumor growth delay (P = 0.017, univariate test of drug effect over time, repeated-measures multivariate ANOVA). D, 4T1 metastases. The total number of metastases per group of surviving mice (top left) as well as the sizes of lung metastases (top right) in high-dose SD-208-treated mice were significantly smaller than in low-dose or vehicle-treated control animals (P = 0.0016 and P < 0.0001, respectively, for 60 mg/kg/d SD-208 compared with either vehicle or 20 mg/kg/d SD-208, Kruskal-Wallis nonparametric ANOVA). In addition, the total number of macroscopically visible metastases to other sites also varied per treatment group (bottom).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Effects of SD-208 on R3T mammary carcinoma in athymic nude mice in vivo. R3T cells (1 × 10⁶) were injected orthotopically into the right second or third mammary fat pads of 8- to 9-week-old athymic nude mice, 8 to 10 mice per group. Treatment and evaluation as described in Fig. 2. A, R3T primary tumor growth. Treatment with SD-208 failed to affect primary tumor growth (P = 0.9766, univariate test of drug effect over time, repeated-measures multivariate ANOVA). B, lung metastases. The size (right) and number (left) of lung metastases were similar across treatment groups (P = 0.5946 and 0.4602, respectively, Kruskal-Wallis nonparametric ANOVA).

Pharmacokinetic and pharmacodynamic properties of SD-208. As shown in Fig. 4A , SD-208 inhibited TGF-β-induced Smad2 phosphorylation in cultured cells in a dose-dependent manner, with an IC₅₀ of ∼80 nmol/L, indicating that it is slightly less potent than SD-093, which has an IC₅₀ of 20 nmol/L. Figure 4B depicts the kinetics of SD-208 in plasma of each of the mouse strains following a single oral dose of either 20 or 60 mg/kg. Peak plasma levels as well as areas under the curve varied as function of the dose given in each of the three mouse strains. Although levels achieved at 1 hour following a dose of 60 mg/kg were not significantly different across the three mouse strains (ANOVA, P = 0.1893), trough levels at 24 hours were significantly lower in athymic than in 129S1 mice (ANOVA, P = 0.0291). Moreover, the area under the curve achieved in 129S1 mice was approximately twice as high as in athymic animals (Fig. 4B). These differences may account, in part, for the lower efficacy against R3T tumors we observed in athymic compared with 129S1 mice (see above). Plasma levels obtained at 2 hours following the penultimate dose of SD208 in treatment experiments were very similar to those achieved following a single dose, indicating that prolonged repeated dosing did not have adverse effects on pharmacokinetics (data not shown). SD-208 treatment was tolerated without observable toxicity for up to 56 days of continuous daily administration.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Properties of SD-208. A, pretreatment of MDA-MB-435 breast cancer cells with TβRI kinase inhibitors inhibited TGF-β-induced Smad2 phosphorylation in a dose-dependent manner, with an IC₅₀ of ∼20 and 80 nmol/L for SD-093 and SD-208, respectively. B, levels of SD-208 in plasma collected following a single oral dose of either 20 or 60 mg/kg were determined by high-performance liquid chromatography. Points, average of three mice per group; bars, SD. Area under the curve was estimated as described in Materials and Methods. C, determination of pSmad2 and pSmad3 levels in protein extracts from individual flash frozen carcinomas growing in mammary fat pads by Western blot analysis. A, control tumors; B, tumors from animals treated with 20 mg/kg/d SD-208; C, tumors from animals treated with 60 mg/kg/d SD-208. Endogenous pSmad2 and pSmad3 levels in R3T and 4T1 tumors were clearly decreased by SD-208 treatment.

To determine whether treatment with SD-208 resulted in detectable inhibition of TβRI kinase activity in vivo, we measured pSmad2 and pSmad-3 levels in protein extracts from individual mammary fat pad tumors harvested and flash frozen within 2 hours of the penultimate SD-208 dose (Fig. 4C). 4T1 tumors seemed to express slightly higher steady-state levels of pSmad2 and pSmad3 than R3T tumors. However, in both cases, pSmad levels were significantly reduced in tumors obtained from SD-208-treated animals compared with those from control animals.

To further substantiate inhibition of TGF-β signaling by SD-208 treatment in vivo, we compared the levels of mRNA expression of several TGF-β-regulated target genes in snap-frozen R3T tumors by quantitative reverse transcription-PCR. As shown in Fig. 5A , the levels of mRNA of Serpine1 (PAI-1), connective tissue growth factor, Col1a2 (procollagen I, α2), and matrix metalloproteinase-2 were all significantly reduced in tumors that had been exposed to SD-208.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Effects of SD-208 on TGF-β-regulated genes in vivo. A, real-time PCR: mRNA levels for the Serpine1 (PAI-1), connective tissue growth factor, Col1a1, and matrix metalloproteinase-2 genes were decreased in R3T tumors growing in 129S1 mice treated with 20 mg/kg SD-208 compared with control animals (Student's t test). B, effects of SD-208 on mRNA transcript levels of individual genes were compared by ANOVA. Six genes that were repressed by SD-208 treatment in a dose-dependent manner are shown as example. Boxes, median values with upper and lower quartiles; filled square, mean; whiskers, 90th and 10th percentiles.

In addition to measuring the expression of TGF-β target genes in tumor tissue, we also examined the effects of SD-208 treatment on the gene expression profile of normal mouse tissue. Because liver tissue is easily accessible and relatively homogeneous in terms of its cellular composition, we compared the gene expression profiles in livers obtained within 2 hours of the last dose of SD-208 from 129S1 mice in each of the three treatment groups. RNA extracted from four individual livers was pooled, and two such pools from different groups of mice were subjected to Affymetrix GeneChip analysis. The experiment was repeated using pooled RNAs obtained from the second 129S1 in vivo treatment study. Genes (n = 12,612) were scored as “present” on all GeneChips, indicating that mRNA was expressed under all 12 conditions (3 dose levels, 2 pooled RNA samples per experiment, and 2 separate experiments). As we ended up with four replicates for each condition, expression levels for each feature could be compared across the three treatment groups by ANOVA. A total of 1,693 Affymetrix IDs representing 1,539 unique genes (79% with annotation) were significantly differentially expressed between the three treatment groups (ANOVA, P < 0.05). These included 729 genes that were down-regulated and 810 genes that were up-regulated in livers of SD-208-treated animals compared with vehicle-treated control animals. We used the DAVID 2.0 software tool (56) to assign the genes to biological processes and molecular functions (Supplementary Fig. 1A and B). A modification of the Fisher's exact test (EASE score) was used to determine whether biological categories to which genes from our list were assigned were overrepresented (P ≤ 0.05) compared the mouse transcriptome in general (56). Genes that were down-regulated in SD-208-treated animals are predominantly involved in transcription, cell cycle control, intracellular signaling, and apoptosis. Several studies have examined the spectrum of TGF-β-regulated genes in the mouse mammary epithelial cell line NMuMG (9, 11, 13) and in mouse embryo fibroblasts (12). Table 1 lists 31 genes that were repressed in the livers of SD-208-treated mice and have been reported previously to be induced by TGF-β in NMuMG mouse mammary epithelial cells in vitro (13). Figure 5B illustrates the dose-dependent reduction of transcript levels of six of these genes by SD-208 treatment. In summary, SD-208 treatment clearly inhibited TβRI kinase activity and TGF-β-regulated gene expression in vivo.

In contrast to the repressed genes, the vast majority of genes that were up-regulated in SD-208-treated animals encoded enzymes and cofactors involved in the intermediary metabolism of carbohydrates, proteins, and lipids, oxidative phosphorylation, or detoxification reactions (Supplementary Fig. 1C and D). Rather, the spectrum of induced genes was strikingly similar to that found in rat or mouse liver tissue in response to a variety of toxins (57–60). Thus, the profile of SD-208-induced genes likely reflects a generalized activation of hepatic metabolic pathways in response to chronic exposure to SD-208.

Effects of SD-208 on tumor-specific CTL activity. The notable difference in therapeutic efficacy of SD-208 against R3T tumors growing in syngeneic compared with immunodeficient mice suggested that enhancing antitumor immunity might represent an important mechanism of action of this agent in vivo (45). To test this hypothesis in a second independent model, we harvested splenocytes from 4T1 tumor-bearing mice from each of the three treatment groups and restimulated these with irradiated 4T1 cells for 5 days in the presence of SD-093. Nonadherent effector cells were then incubated with ⁵¹Cr-labeled 4T1 target cells, and ⁵¹Cr release was quantified using a gamma counter (Fig. 6A ). Specific CTL activity was significantly increased by in vivo treatment with SD-208 in a dose-dependent manner. Moreover, this effect was only observed as long as SD-093 was present during restimulation, indicating that 4T1 produce sufficient amounts of bioactive TGF-β to suppress CTL activity or expansion in vitro. Similar effects, although of lower magnitude, were observed using freshly isolated splenocytes without in vitro restimulation (Fig. 6A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Mechanisms of action of SD-208 in vivo. A, effects of treatment with SD-208 on tumor-specific CTL activity. Day 5: Splenocytes harvested from 4T1 tumor-bearing mice from each of the three treatment groups (▪, vehicle; •, 20 mg/kg; ⧫, 60 mg/kg) were restimulated with irradiated 4T1 cells for 5 days in the presence of SD-208, after which nonadherent effector cells were harvested. ⁵¹Cr-labeled 4T1 target cells and effector cells were then incubated for 4 hours at 37°C, and released ⁵¹Cr was quantified using a gamma counter. At an E:T of 100:1, 4T1-specific CTL activity was significantly increased in mice treated with 20 and 60 mg/kg/d SD-208 (P = 0.04 and 0.02, respectively, Student's t test). Points, mean from two independent experiments, triplicate wells per condition; bars, SD. Day 0: Splenocytes harvested from 4T1 tumor-bearing mice from each of the three treatment groups (▪, vehicle; •, 20 mg/kg; ⧫, 60 mg/kg) were immediately incubated with ⁵¹Cr-labeled 4T1 target cells for 4 hours at 37°C, and released ⁵¹Cr quantified using a gamma counter. 4T1-specific CTL activity was increased as a function of the SD-208 dose the animals had received. B, sections of R3T tumors from SD-208-treated 129S1 mice were stained by H&E as well as for markers of cell proliferation (Ki-67), apoptosis (TUNEL), and capillaries (CD34). Magnification, ×200. R3T tumors are undifferentiated spindle-cell carcinomas. Moreover, R3T tumors have a high proliferative cell fraction and a low rate of apoptosis. However, in some R3T tumors, SD-208 induced adenosquamous differentiation. Moreover, SD-208 treatment was associated with a striking eosinophil infiltrate. C, treatment with SD-208 did not affect tumor cell proliferation (Ki-67; P = 0.3066, one-way ANOVA) or apoptosis (TUNEL assay; P = 0.5474, one-way ANOVA). However, microvessel density, as determined by anti-CD34 immunostaining, was significantly reduced in tumors of animals treated with 60 mg/kg/d SD-208 compared with either vehicle-treated control animals or animals treated at the 20 mg/kg/d dose level (P = 0.0143, one-way ANOVA). For each of these analyses, a total of 20 randomly selected high-power (×400) fields in areas of viable tumor were examined in tumor sections from each treatment group.

To further elucidate the mechanisms of the antitumor effects of SD-208 in vivo, we examined the rates of cell proliferation and apoptosis of R3T tumors as well as their neovascularization (Fig. 6B and C). We failed to detect any significant differences in the rate of tumor cell proliferation (Ki-67 staining) or apoptosis (TUNEL) between R3T tumors of untreated and SD-208-treated animals (Fig. 6C). However, tumor angiogenesis, as reflected by CD34⁺ microvessel density, was significantly reduced in the tumors from animals treated with the highest dose of SD-208 (Fig. 6C). This finding suggests that, besides enhancing antitumor immunity, SD-208 also has modest antiangiogenic properties in vivo.

Finally, treatment with SD-208 was associated with a striking infiltration of tumors with eosinophilic leukocytes (Fig. 6B). In addition, in contrast to the undifferentiated spindle-cell appearance of control tumors, some of the R3T carcinomas in SD-208-treated animals were much more differentiated with clearly recognizable ductal structures and areas of keratinization (Fig. 6B). This morphologic appearance was reminiscent of the metaplastic squamous carcinomas that develop in the mammary glands of Smad4 conditional knockout mice and in carcinogen-treated transgenic mice that express a dominant-negative TβRII receptor (20, 61). Thus, chemical or genetic inhibition of TGF-β signaling seems to alter mammary carcinoma cell differentiation in similar ways.

Discussion

The discovery that many cancers produce or induce bioactive TGF-β, which, in turn, acts as a tumor-promoting oncogene, has generated a great deal of enthusiasm for targeting tumor-associated bioactive TGF-β as cancer therapy (see refs. 37–40 for recent reviews). Our study describes, for the first time, the inhibition of tumor growth and metastatic efficiency in vivo of mouse mammary carcinomas by a novel selective chemical inhibitor of the TβIR kinase, SD-208. This agent belongs to a class of highly selective and potent pyridopyrimidine type TβRI kinase inhibitors that block TGF-β-induced Smad phosphorylation, reporter gene activation, and cellular responses at submicromolar concentrations (42). These chemicals bind to the ATP-binding site of the TβRI kinase and maintain the enzyme in its inactive configuration (62).

The phenotype of mammary carcinoma lines differed from that of nontransformed NMuMG cells in several important ways. First, in contrast to NMuMG cells, both R3T and 4T1 cells were refractory to TGF-β-mediated growth suppression, consistent with previous reports (reviewed in refs. 22, 32, 33, 63). Both R3T and 4T1 cells seem to have retained a typical epithelioid morphology in vitro, although 4T1 cells were clearly less cohesive probably because they do not express E-cadherin (53). Exogenous TGF-β clearly induced EMT in both tumor lines in vitro in a TβRI kinase-dependent manner, whereas treatment with SD-093 alone seemed to increase the cohesion of R3T cells. Thus, the growth-inhibitory effect of TGF-β had become uncoupled from the EMT response. In contrast to NMuMG, TGF-β not only failed to inhibit migration of R3T and 4T1 cells but also acted as a stimulus of cell motility, an effect that could be blocked by SD-093, indicating that it was mediated by the TβRI kinase. In addition, TGF-β treatment stimulated the ability of the mammary carcinoma cells to invade Matrigel, an effect that was also inhibited by SD-093. Interestingly, TGF-β inhibits in vitro invasion of normal trophoblast cells (64), whereas an extensive literature attests to the fact that TGF-β signaling stimulates or even drives the invasive behavior of malignant cells (see refs. 32, 33 for reviews). Thus, TGF-β seems to undergo a switch from being an antimigratory and anti-invasive to a promigratory and proinvasive factor during malignant transformation (29, 43, 55).

In contrast to their similar in vitro phenotype, R3T tumors in vivo were composed of highly spindle-shaped cells, whereas 4T1 tumor-derived carcinomas were more differentiated and epithelioid. This apparent discrepancy between in vitro and in vivo phenotypes is likely due to a combination of cell intrinsic factors (e.g., activation of the H-ras gene in R3T but not in 4T1 cells) and a permissive microenvironment (53, 65). Thus, one might speculate that differences in local TGF-β activation in tumors in vivo might influence the degree of EMT. If this were correct, one would expect treatment with SD-208 to induce a reversal of the fibroblastoid to an epithelioid phenotype. This is, in fact, what we observed in R3T tumors in animals treated with the highest dose of SD-208. In contrast to the undifferentiated spindle-cell appearance of control tumors, these carcinomas were much more differentiated, with clearly recognizable ductal structures and areas of keratinization. This morphologic appearance was reminiscent of the metaplastic squamous carcinomas seen in the mammary glands of Smad4 conditional knockout mice and in carcinogen-treated transgenic mice that express a dominant-negative TβRII receptor (20, 61). Treatment with SD-208 was associated with a striking infiltration of tumors with eosinophilic leukocytes. The significance and the mechanism for this phenomenon are unclear. However, Takaku et al. (66) reported that gastric polyps that develop in Smad4 heterozygous mice are associated with an eosinophilic cell infiltrate. Conversely, a nasally delivered DNA encoding TGF-β1 can suppress pulmonary eosinophilic responses to infectious agents (67). Thus, attenuation of TGF-β signaling seems to promote eosinophilia, whereas increased expression of TGF-β1 suppresses it.

As reported previously, orthotopic 4T1 or R3T invasive carcinomas efficiently metastasized to the lung parenchyma (48, 68). Most importantly, treatment of the mice with SD-208 significantly retarded primary tumor growth in both models in a dose-dependent manner. The antitumor effects of SD-208 were more pronounced against R3T than against 4T1 tumors possibly as a consequence of the higher growth rate of 4T1 tumors or the stronger TGF-β activation by R3T cells (Fig. 1E). Alternatively, the activation of Ha-ras and/or expression of PyVMT in R3T may have conferred particular sensitivity to TGF-β antagonists. Besides retarding the growth of primary tumors, treatment with SD-208 significantly reduced the number of lung metastases in both models. However, 4T1 metastases were significantly larger than those derived from R3T tumors, and the effect of SD-208 on their size was notably less. The therapeutic effect of SD-208 against 4T1 was entirely consistent with those observed in the 4T1 model by other investigators using different types of TGF-β antagonists, including dominant-negative TβRII receptors (29), soluble TβRII exoreceptors (69), neutralizing anti-TGF-β antibodies (70), or other chemical TβR kinase inhibitors (71). Our observations suggest that the effects of TGF-β antagonists on tumor growth rates may be quite distinct from those on metastatic efficiency. Consistent with this idea, expression of a dominant-negative TβRII receptor in 4T1 or in human MCF10CA1a mammary carcinoma cells reduced their metastatic ability without affecting primary tumor growth (29, 72). Similarly, administration of soluble TβRII exoreceptor to syngeneic 4T1 or EMT6 mammary carcinoma-bearing animals had a modest inhibitory effect on primary tumor size, whereas lung metastases were inhibited by 60% to 70% (69). In addition, soluble TβRII exoreceptor inhibited the development of lung metastases in neu transgenic mice without affecting the incidence or growth of primary tumors (73). Overall, the antimetastatic effects of TGF-β antagonists against syngeneic mouse tumors seem to be retained in human xenograft models, whereas tumor growth rates may or may not be affected. This suggests that effects on the host may primarily mediate the inhibition of tumor growth, whereas the reduction in metastatic clonal efficiency is more likely caused by cell-autonomous effects of the antagonists.

The mechanisms of the antitumor effects of SD-208 are probably multifactorial. First, although growth of R3T tumors was strongly inhibited in syngeneic host animals, SD-208 treatment failed to affect R3T tumor growth in athymic nude mice. Several factors may account, at least in part, for the difference in efficacy of SD-208 between the two experiments. First, R3T tumors grew significantly more rapidly in athymic animals compared with 129S1 mice. In addition, athymic mice seem to clear SD-208 significantly more rapidly than 129S1 mice. Thirdly, and perhaps most importantly, the difference may be due to induction of tumor-specific cytotoxic T-cell (CTL) activity (35, 36). Independent evidence for enhancement of CTL activity was provided by our finding that splenocytes isolated from 4T1 tumor-bearing SD-208-treated mice were significantly more cytotoxic against 4T1 cells in vitro than those from vehicle-treated control animals. Interestingly, this difference was uncovered only in the presence of a TβRI kinase inhibitor, indicating that active TGF-β, presumably produced by 4T1 cells, was inhibiting CTL activity even in vitro. These results are consistent with a recent report by Suzuki et al. (74), showing that the therapeutic efficacy of a soluble TβRII exoreceptor against transplanted murine malignant mesotheliomas was largely dependent on CD8⁺ T cells.

Besides the effects on CTL activity, TGF-β antagonists may stimulate antitumor natural killer cell activity. Arteaga et al. (75) showed that i.p. injections of a pan-TGF-β-neutralizing mouse antibody, 2G7, suppressed in vivo growth and lung metastases of MDA-MB-231 human breast cancer cells in athymic nude mice with a concomitant increase in mouse spleen natural killer cell activity, although the effect that was not seen in natural killer–deficient mice. Similarly, Uhl et al. (45) recently reported that the in vivo antitumor effects of SD-208 against murine SMA-560 gliomas correlated with restoration of lytic activity of polyclonal natural killer cells against glioma cells in vitro. Finally, Friese et al. (76) reported that silencing of TGF-β1 and TGF-β2 by small interfering RNAs in human LNT-229 malignant glioma cells suppressed their tumorigenicity in nude mice, and natural killer cells isolated from these mice showed an activated phenotype. A third immune mechanism that may play a role in the antitumor efficacy of TGF-β antagonists is enhancement of dendritic cell function (77). In one study, the TGF-β-neutralizing monoclonal antibody 2G7 enhanced the ability of dendritic cell vaccines to inhibit the growth of established 4T1 murine mammary tumors in vivo (78). However, in our case, we failed to observe any antitumor effect of SD-208 in athymic nude mice.

Besides the evidence in favor of an immune-mediated effect of SD-208 summarized above, we also noted a significant decrease in microvessel density in R3T tumors obtained from SD-208-treated animals compared with control tumors. Although Uhl et al. (45) failed to detect any change in microvessel density in SD-208-treated gliomas, this may be due to disparities between the tumor models, as several other studies of human tumor xenograft models have found TGF-β antagonists to inhibit tumor angiogenesis (79–83). These observations are consistent with previous reports showing that tumor-associated TGF-β contributes to angiogenesis and that hypoxia and TGF-β synergistically up-regulated vascular endothelial growth factor mRNA expression (84).

That SD-208 treatment truly inhibited TGF-β signaling in vivo was shown both by the reduction in phosphorylated Smad levels in SD-208-treated tumors and by the dose-dependent down-regulation of mRNA levels of TGF-β target genes in tumor tissue. These results are consistent with the report by Bonniaud et al. (44), showing similar effects of SD-208 in lung tissue in a rat lung fibrosis model.

Moreover, SD-208 also affected transcription of TGF-β target genes in normal tissue. Previous studies of the effects of short-term TGF-β treatment of cultured cells on the global gene expression profile have shown that several hundred genes are up-regulated by TGF-β. Although several of these previously described TGF-β target genes were repressed in liver tissue by SD-208 treatment, the majority of repressed genes had not been previously identified as TGF-β targets. This is, perhaps, not surprising as short-term effects of TGF-β on cultured cells are likely to be quite different from long-term effects of a TGF-β antagonist on animal tissue cells in vivo. Perhaps more illuminating is the fact that the genes affected by SD-208 treatment are predominantly involved in the regulation of the cell cycle, cell death, and apoptosis. Thus, our findings are highly consistent with the known biological effects of TGF-β on normal epithelial cells in general and normal hepatocytes in particular (i.e., context-dependent inhibition of the cell cycle at the G₁-S checkpoint or the induction of apoptosis; refs. 85, 86).

Finally, it is important to note that SD-208 was remarkably free of clinically observable toxicity even after prolonged treatment of mice in vivo. This extends previous observations in which SD-208 was given by gavage to rats for short periods of time (44). Although Uhl et al. (45) also gave SD-208 to mice for up to 40 days, the drug was dissolved in the drinking water. Thus, ours is the first preclinical study that shows the relative clinical safety of this agent when given at high doses by daily gavage. Moreover, we failed to detect any evidence of organ damage postmortem, including breakdown of mucosal antimicrobial defenses or inflammation. Although we did note a dose-dependent increase in relative liver weight in all three strains of mice, the histologic appearance of the livers was unremarkable. Given our finding that SD-208 treatment was associated with a dose-dependent increase in mRNA expression of a wide range of genes that encode enzymes involved in carbohydrate, protein, and fat metabolism as well as detoxification reactions, it is likely that the resulting global increases in protein accounted for the increase in liver weight. The majority of the induced genes are not normally regulated by TGF-β signaling per se but seem to reflect a generalized activation of hepatic metabolic pathways in response to chronic exposure to SD-208. This spectrum of induced genes was strikingly similar to that found in rat or mouse liver tissue in response to exposure to a wide variety of chemicals (57–60, 87).

In summary, our studies show, for the first time, that the novel selective chemical inhibitor of the TβRI kinase, SD-208, is capable of inhibiting the metastatic efficiency as well as growth of mouse mammary carcinomas in vivo. The antitumor effects of SD-208 were observed in the absence of any obvious signs of toxicity, and pharmacodynamic studies of pSmad levels and TGF-β target gene expression confirmed inhibition of the target enzyme in vivo. The inhibitory effect of SD-208 on tumor growth rates seemed to be mediated by enhancement of tumor-specific CTL activity and, to a lesser extent, by inhibition of angiogenesis, whereas the inhibition of metastatic efficiency was more likely due to cell-autonomous effects of the drug on the tumor cells themselves. The observed antitumor activity was modest in magnitude. As the oncogenic role of TGF-β signaling seems to come into play at a relatively late stage of tumor progression, blocking this pathway by itself is unlikely to be sufficient to eradicate tumors and will have to be combined with strategies that block the primary drivers of tumor growth, such as cell cycle activators, inhibitors of apoptosis, and immortalization. Our results should help inform the clinical development of TGF-β antagonists in general and chemical TβRI kinase inhibitors in particular.