• Open Access

Anti-angiogenic and anti-tumor effects of TAK-593, a potent and selective inhibitor of vascular endothelial growth factor and platelet-derived growth factor receptor tyrosine kinase


To whom correspondence should be addressed

E-mail: hori_akira@takeda.co.jp


We recently reported that TAK-593, a novel imidazo[1,2-b]pyridazine derivative, is a highly potent and selective inhibitor of the vascular endothelial growth factor (VEGF) and platelet derived growth factor (PDGF) receptor tyrosine kinase families. Moreover, TAK-593 exhibits a uniquely long-acting inhibitory profile towards VEGF receptor 2 (VEGFR2) and PDGF receptor β (PDGFRβ). In this study, we demonstrated that TAK-593 potently inhibits VEGF- and PDGF-stimulated cellular phosphorylation and proliferation of human umbilical vein endothelial cells and human coronary artery smooth muscle cells. TAK-593 also potently inhibits VEGF-induced tube formation of endothelial cells co-cultured with fibroblasts. Oral administration of TAK-593 exhibited strong anti-tumor effects against various human cancer xenografts along with good tolerability despite a low level of plasma exposure. Even after the blood and tissue concentrations of TAK-593 decreased below the detectable limit, a pharmacodynamic marker (phospho VEGFR2) was almost completely suppressed, indicating that its long duration of enzyme inhibition might contribute to the potent activity of TAK-593. Immunohistochemical staining indicated that TAK-593 showed anti-proliferative and pro-apoptotic effects on tumors along with a decrease of vessel density and inhibition of pericyte recruitment to microvessels in vivo. Furthermore, dynamic contrast-enhanced magnetic resonance imaging revealed that TAK-593 reduced tumor vessel permeability prior to the onset of anti-tumor activity. In conclusion, TAK-593 is an extremely potent VEGFR/PDGFR kinase inhibitor whose potent anti-angiogenic activity suggests therapeutic potential for the treatment of solid tumors.

Angiogenesis, the development of new blood vessels, is required for solid tumors to grow beyond a certain size, and provides oxygen, nutrients, and a potential route for subsequent metastasis.[1] Therefore, inhibition of tumor angiogenesis is an attractive and well-validated anticancer therapy. The process of tumor angiogenesis involves multiple and complex steps.[2] First, vascular endothelial growth factor (VEGF) is induced by hypoxia and activates vascular endothelial cells in the vessels around a tumor.[3, 4] Vascular endothelial growth factor stimulates endothelial cell proliferation through the VEGF receptor family, especially VEGF receptor 2 (VEGFR2), which is almost exclusively located in endothelial cells.[5, 6] Following the endothelial cell basement membrane degradation by proteases, endothelial cells migrate from the existing vessels towards the source of VEGF, and begin to proliferate[7, 8] and assemble into tubes with a parent lumen. After the formation of small blood vessels, recruitment of smooth muscle cell-like pericytes to the immature vascular structures is required to stabilize these new vessels. To effect this process, endothelial cells release platelet-derived growth factors (PDGFs), especially PDGF-BB, that are potent chemoattractants and mitogens for pericytes expressing PDGF β-receptor. Loss of PDGF-BB or PDGF β-receptor in mouse models causes microvascular leakage, hemorrhage, and almost complete failure of pericyte development.[9-11] Therefore, PDGF signaling has essential roles in establishing stable vascular structures by pericytes recruitment.

The clinical effectiveness of blocking VEGF signaling has been demonstrated with the use of bevacizumab, a neutralizing monoclonal antibody targeting VEGF, in combination with chemotherapy.[12] However, the results of some clinical trials have suggested that anti-VEGF therapy alone may not be enough for efficient blocking of tumor angiogenesis.[13, 14] Not all tumors are sensitive to VEGF blockade, and some tumors that are initially sensitive can become resistant.[15-18] Mature tumor vessels with more abundant pericytes appear to be less sensitive to the anti-VEGF therapy.[19-21] Increased expression of PDGF-BB has been detected around the vessels of tumors that developed resistance to anti-VEGF therapy.[22] Based on these results, an anti-angiogenic strategy targeting both endothelial cells and pericytes by inhibiting both VEGFR and PDGFR kinases could possibly improve the response over that obtained with anti-VEGF therapy alone.[19, 23-27]

TAK-593 is a highly selective small molecular weight inhibitor of the VEGFR2 and PDGFRβ tyrosine kinases that has been shown to exhibit a markedly long residence time on its targets.[28] In this study, we describe the cellular and anti-tumor activities of TAK-593 as well as its effects on the morphology of intratumoral blood vessels and vascular permeability.

Materials and Methods

Test compound

TAK-593,N-[5-({2-[(cyclopropylcarbonyl)amino]imidazo[1,2-b]pyridazin-6-yl}oxy}- 2-methylphenyl)-1,3-dimethyl-1H-pyrazole-5-carboxamide was synthesized at Takeda Pharmaceutical Company.[28] TAK-593 was initially prepared as 10 mM stock solution in dimethyl sulfoxide for in vitro use and as a 0.5 w/v% methylcellulose (Shin-Etsu Chemical, Tokyo, Japan) solution for in vivo use.

Cell proliferation assay

Human umbilical vein endothelial cells (HUVECs) and coronary artery smooth muscle cells (CASMCs) were obtained from Cambrex (Walkersville, MD, USA). Cancer cell lines were obtained from the American Type Culture Collection (ATCC, Menassas, VA, USA) or European Collection of Animal Cell Cultures (ECACC, Salisbury, UK). Human umbilical vein endothelial cells were treated with TAK-593 and recombinant human VEGF (R&D Systems, Minneapolis, MN, USA) for 5 days. Coronary artery smooth muscle cells were starved overnight in serum-free medium and treated with TAK-593 and recombinant human PDGF-BB (PeproTech EC, London, UK) for 6 days. Human cancer cell lines and MRC-5 human fibroblasts were treated with TAK-593 for 3 days. After incubation, cell proliferation was determined using the Cell Counting Kit-8 (10 μL/well; DOJINDO Laboratories, Kumamoto, Japan) as directed by the manufacturer. The concentration causing 50% inhibition (IC50) was calculated from the dose-response curve generated by least-squares linear regression analysis of the response using sas software and the NLIN procedure (version 5.0; SAS Institute Japan, Tokyo, Japan).

Receptor phosphorylation assay

Human umbilical vein endothelial cells and CASMCs were treated with TAK-593 for 2 h prior to stimulation with recombinant human VEGF or PDGF-BB for 5 min. Cell lysates were analyzed by Western blotting with antibodies for phospho-Tyr951-VEGFR2 (Cell Signaling Technology, Beverly, MA, USA), VEGFR2 (Cell Signaling Technology), and PDGFRβ antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Phospho PDGFRβ was detected with an anti-phosphotyrosine antibody (4G10, Upstate Systems, Charlottesville, VA, USA) after immunoprecipitation with the anti-PDGFRβ antibody. Quantitation was done using LAS lumino-image analyzer and Multi Gauge ver. 3.3 software (FUJIFILM, Kanagawa, Japan).

Tube formation assay

Human umbilical vein endothelial cells and normal human dermal fibroblasts (NHDF; Cambrex) were co-cultured with VEGF and TAK-593 for 7 days. The cells were stained with mouse anti-human CD31 monoclonal antibody (R&D systems) and Alexa Fluor 488-conjugated chicken anti-mouse IgG (Invitrogen, Carlsbad, CA, USA) for visualization by fluoroscopic imaging (Discovery-1 system, Molecular Devices, Sunnyvale, CA, USA) according to the manufacturer's protocol, “Angiogenesis Tube Formation”. The tube area was quantified by using Metamorph software (Molecular Devices).

Tumor xenograft models

All animal experiments were conducted according to the guidelines of the Takeda Experimental Animal Care and Use Committee. Athymic nude mice (BALB/cA Jcl-nu/nu) and severe combined immunodeficient (SCID) mice (C.B17/Icr-scid/scid Jcl) were purchased from Japan CLEA (Tokyo, Japan). Animals were housed in a barrier facility with 12 h light/dark cycles and were provided with sterilized food and tap water ad libitum. Tumor cells were implanted subcutaneously into the hind flank of mice. The human gastric cancer cell line MKN45 and human primary renal cell carcinoma xenograft tissue RCC-02-JCK were obtained from RIKEN BRC Cell Bank (Ibaraki, Japan) and the Central Institutes for Experimental Animals (Kawasaki, Japan), respectively. RCC-02-JCK xenografts were implanted subcutaneously as tissue fragments approximately 2 mm in diameter. For the intracranial xenograft model, U87 MG human glioblastoma cells (5 × 105 cells/5 μL) were injected near the bregma using a microsyringe. After the tumor xenografts were established, TAK-593 and the vehicle were given orally to the animals twice daily. Tumor volumes were assessed by measurement of two dimensions with vernier caliper at least twice weekly and were calculated as length × width2 × 1/2. As an index of anti-tumor activity, the mean change in tumor volume over the treatment period was compared between the control and treated groups, and the treated/control ratio (T/C) was calculated as a percentage. In the intracranial model, the mice were monitored for survival. Body weight was also measured on the day of tumor volume assessment. Statistical comparisons were performed with the one-tailed Williams' test or Shirley-Williams test using sas system.

Pharmacokinetic and pharmacodynamic analyses

Nude mice bearing human lung carcinoma cell line A549 xenografts were orally administered TAK-593. Then the TAK-593 concentration in plasma and lung tissue was determined by HPLC/MS/MS or HPLC with a fluorescence detector at the indicated times. Plasma concentration-versus-time data were analyzed by a non-compartmental pharmacokinetic method to determine the area under the plasma concentration-time curve for 24 h (AUC0–24 h) and the maximum drug concentration (Cmax). For pharmacodynamic analysis, VEGF was given intravenously to mice 5 min before death, and the lung tissues were collected and subjected to Western blotting for detection of phospho-VEGFR2 and VEGFR2 (Invitrogen).


Mice bearing A549 tumors were treated with TAK-593 at the indicated doses and schedules. Cryosections of optimal cutting temperature compound embedded tumor tissues dissected from the xenografts of mice were prepared. For visualization of the blood vessels, immunohistochemical staining for CD31 was performed by incubation with a rat anti-CD31 monoclonal antibody (BD Biosciences, San Jose, CA, USA) followed by the immunoperoxidase technique with diaminobenzidine tetrahydrochloride as the chromagen. Proliferating cells and pro-apoptotic cells were immunostained with an anti-Ki67 antibody (Dako Cytomation; Dako Glostrup, Denmark) and were assayed by the TUNEL method using “ApopTag Peroxidase In Situ Apoptosis Detection Kit” (Millipore, Billerica, MA, USA). All sections were examined under an Axiovert 200M microscope (Carl Zeiss Microimaging, Oberkochen, Germany). Three areas of 8.25 mm2 each were randomly selected and captured with a digital camera. Image processing was performed with AxioVision Release 4.5 digital imaging analysis software (Carl Zeiss). To assess pericyte recruitment in the A549 xenograft nude mouse model, tumor tissues were prepared by perfusion fixation with paraformaldehyde and embedding in paraffin. Endothelial cells and pericytes were stained with anti-CD31 (Spring Bioscience, Fremont, CA, USA) and anti-α smooth muscle actin (αSMA) antibodies (Dako), respectively. The fluorescent immunohistochemical staining for CD31 and αSMA was performed with AlexaFluor594-labeled donkey anti-rabbit IgG and AlexaFluor488-labeled donkey anti-mouse IgG (Invitrogen), respectively. Subsequently, the slides were scanned with a NanoZoomer Digital Pathology instrument (Hamamatsu Photonics, Shizuoka, Japan) and the pericyte area was quantified by analysis of five randomly selected fields from each tumor using WinROOF software (Mitani, Tokyo, Japan).

Dynamic constant enhanced magnetic resonance imaging

Under isoflurane anesthesia (Merck KGaA, Darmstadt, Germany), human colon carcinoma cell line HT-29 tumor-bearing mice were secured on an acrylic board to immobilize the tumor region and put into the MRI unit (Varian Unity INOVA 4.7T, Varian, Palo Alto, CA, USA). After acquiring data for a T1 map of the tumor tissue using a gradient-echo sequence (TE/TR=6/20 msec, flip angle = 3, 5, 10, 15, 20, and 30 degrees, FOV = 4 × 4 cm, slice thickness = 1 mm, NEX = 2), dynamic constant enhanced magnetic resonance imaging (DCE-MRI) was carried out with a 2-dimensional multislice, RF-spoiled gradient-echo technique (TE/TR = 6/20 ms, flip angle = 60 degrees, FOV = 4 × 4 cm, slice thickness = 1 mm, NEX = 2). Gadolinium tetraazocyclododecane-tetraacetate (Gd-DTPA, 0.2 mL of a 10 mM Gd-DTPA solution, Magnevist, Bayer Healthcare Pharmaceuticals, Leverkusen, Germany) was administered into the tail vein within 10 s of 10 preliminary scans. After the bolus of contrast, 61 scans were obtained continuously over about 10 min. Magnetic resonance imaging was initially performed before oral administration of TAK-593. TAK-593 was administered twice daily for 3 days, and MRI was repeated at 1 h after the initial dose on Day 4. On the basis of a two-compartment model (bidirectional kinetic mode),[29] the Gd-DTPA concentration in tumor tissue (Ct) and the Gd-DTPA concentration in plasma (Cp) were computed from Formula 1, with R1 (=1/T1) acquired in this experiment being substituted. Then the values thus obtained were substituted for Ct and Cp in Formula 2 to obtain the Ktrans (Kps).

Formula 1:

display math

C: concentration (mM), t1d: T1 of tissue (s), α: flip angle, tr: repetition time (s), r1: relaxivity (mM/s), SI: signal intensity after treatment, SIpre: signal intensity before treatment.

Formula 2:

display math

The unit of Kps is mL/min per mL tissue. Conversion of the Varian FID data into the TIFF images and compensation of the signal intensity of the chronological data (71 scans) were carried out before the Ktrans maps were computed by the MATLAB programs (version 7.5, Cybernet systems, Tokyo, Japan). To measure the tumor tissue volume, the dimensions of the tumor on each slice (thickness = 1 mm) were measured as pixels, which were converted into millimeters.


TAK-593 potently and selectively inhibits VEGF and PDGF signaling in cellular assays

TAK-593 strongly inhibited VEGF-induced phosphorylation in HUVECs and PDGF-BB-induced phosphorylation in CASMCs (a cell line similar to pericytes) with IC50 values of 0.34 and 2.1 nM, respectively (Fig. 1a). TAK-593 also inhibited the signaling molecules downstream of VRGFR2 including Akt and ERK, which may correlate with the suppression of the VEGFR2 phosphorylation Suppl. Fig. S1). In addition, TAK-593 caused concentration-dependent inhibition of the VEGF- or PDGF-BB-stimulated proliferation of HUVECs and CASMCs, with IC50 values of 0.30 and 3.5 nM, respectively (Table 1, Suppl. Fig. S2). In contrast, TAK-593 had a much weaker effect on the proliferation of fibroblasts (MRC5) and various cancer cell lines, with IC50 values of 8.6–30 μM (Table 1, Suppl. Fig. S3). The effect of TAK-593 was also assessed in tube formation assay to mimic neovascularization. Endothelial cells co-cultured with fibroblasts produced tube-like structures in the presence of VEGF. TAK-593 strongly inhibited this tube formation in agreement with its anti- proliferative effect on HUVECs, with a IC50 value of 0.32 nM (Fig. 1b). These data revealed high selectivity of TAK-593 for VEGFR and PDGFR family kinases in cell-based assays.

Table 1. Inhibitory activity of TAK-593 against cellular proliferation
TypeCell lineIC50 (nM)95% confidence interval (nM)
  1. a

    Proliferation of Human umbilical vein endothelial cells (HUVECs) was induced by vascular endothelial growth factor (VEGF) (60 ng/mL).

  2. b

    Proliferation of coronary artery smooth muscle cells (CASMCs) was induced by platelet-derived growth factor-BB (PDGF-BB) (200 ng/mL). The experiments were performed in quadruplicate.

Endothelial cellHUVECa0.300.28–0.32
Smooth muscle cellCASMCb3.503.00–4.00
LungA54930 00025 000–37 000
PancreasCFPAC-113 00010 000–17 000
ProstrateDU-14518 00017 000–21 000
ColonHT-2914 00013 000–15 000
BreastMDA-MB-238 6006 800–11 000
Lung fibrobastMRC-524 00019 000–30 000
Figure 1.

TAK-593 inhibits cellular vascular endothelial growth factor/platelet-derived growth factor (VEGF/PDGF) signaling and VEGF-induced tube formation. (a) Human umbilical vein endothelial cells (HUVECs) or coronary artery smooth muscle cells (CASMCs) were treated with TAK-593 and then stimulated with VEGF (100 ng/mL) or PDGF-BB (20 ng/mL), respectively. (b) Human umbilical vein endothelial cells co-cultured with normal human dermal fibroblasts (NHDF) were treated with TAK-593 in the presence of VEGF (10 ng/mL) for 7 days. Endothelial cells were visualized and quantified with anti-CD31 staining and fluorescent imaging; representative images are shown (×20). The experiments were performed in duplicate.

TAK-593 demonstrates broad-spectrum anti-tumor activity in xenograft models

Twice daily oral administration of TAK-593 suppressed tumor growth in mouse xenograft models of lung, colon, breast, prostate, pancreas, renal, thyroid, ovary, and gastric cancer (Table 2). Statistically significant inhibition of tumor growth (< 0.025) was observed in six out of eight tumor models (A549, HT-29, MDA-MB-231, DU145, CFPAC-1, and SK-OV-3) at a dose of 0.25 mg/kg. Almost complete inhibition of tumor growth (T/C < 10%) was observed in six out of 10 models (A549, HT-29, MDA-MB-231, DU145, CFPAC-1, and MKN45) at 4 mg/kg without significant adverse effects on body weight (data not shown). To determine the effect of TAK-593 in larger, more established tumors, treatment was initiated when A549 lung cancer xenografts had reached an average volume of 430 mm3 (about fourfold larger than the initial volume of the standard tumor model in Table 2). In this model, twice-daily oral administration of TAK-593 for 2 weeks at 0.125 and 0.25 mg/kg strongly inhibited tumor growth, with T/C values of 33% and 16%, respectively, while actual tumor regression was seen at doses of 1.5 or 3 mg/kg (Fig. 2a). After treatment with TAK-593 was ceased on day 42, tumor regrowth was observed. However, re-initiation of treatment at Day 50 exhibited stasis at 0.25 mg/kg or regression at 1.5 or 3 mg/kg doses when administered for an additional 5 weeks. TAK-593 treatment did not lead to a difference of mean body weight between the control and treatment groups through the treatment period indicating general tolerability (Suppl. Fig. S4). The anti-tumor activity of TAK-593 was also investigated in the U87 MG orthotopic xenograft mouse model. Oral administration of TAK-593 at 1 and 4 mg/kg twice daily significantly prolonged the median survival time of mice with intracranial implantation of U87 MG human glioblastoma (Fig. 2b, ≤ 0.025).

Table 2. Anti-tumor efficacy of TAK-593 in subcutaneous xenograft mouse model
TypeCell lineNumber of miceInjected cell number (cells)Time to establish tumor prior to treatment (days)Tumor size at start of treatment (mm3)Treatment duration (days)T/C (%) Dose (mg/kg, bid)
  1. TAK-593 was orally administered at the indicated doses twice daily for 10–21 days. Data are shown as T/C ratios (= 5–6). * 0.025 versus the vehicle group by one-tailed Williams' or Shirley-Williams' test. N.D., not done.

LungA54955 × 106131201434*8*−8*
ColonHT-2955 × 106171201451*26*2*
BreastMDA-MB-23153 × 106321201451*5*−18*
ProstateDU14555 × 106202001443*13*−1*
PancreasCFPAC-151.25 × 106151701432*−21*−38*
RenalRCC-02-JCK5Tissue fragment1417014N.D.53*23*
ThyroidTT65 × 10641120217848*20*
GlioblastomaU87MG55 × 10627190148854*24*
OvarySK-OV-351 × 106561702153*37*24*
GastricMKN4553 × 1061112010N.D.14*0*
Figure 2.

TAK-593 causes tumor regression and prolongs survival in a mouse xenograft model. (a) Data represent the mean and standard deviation (SD) (n = 5). The T/C (%) was calculated relative to the vehicle control group on Day 42. *≤ 0.025 versus the vehicle control group by one-tailed Williams' test. (b) U87 MG human glioblastoma cells were orthotopically implanted into the brains of nude mice. Twice daily oral administration of TAK-593 (1 and 4 mg/kg) or vehicle was initiated at 4 days after inoculation, and the mice were monitored for survival (n = 10). #≤ 0.025 versus vehicle control by Tarone's test.

Pharmacodynamic and pharmacokinetic profile of TAK-593

Oral administration of TAK-593 at 0.125 mg/kg in mice led to an AUC0–24 h of 0.078 μg・h/mL and a Cmax of 0.069 μg/mL in plasma at 15 min post-dosing (Fig. 3a). Administration of TAK-593 at 1 mg/kg resulted in comparatively proportional AUC0–24 h and Cmax values in plasma (0.883 μg・h/mL and 0.451 μg/mL at 1 h, respectively) and lung tissue (0.556 μg・h/mL and 0.242 μg/mL at 1 h, respectively). The pharmacokinetic profile of TAK-593 relatively showed rapid absorption and clearance. TAK-593 was not detectable in plasma or lung tissue of two out of three mice at 8 h after dosing (Fig. 3b). Despite low plasma and tissue levels, phospho-VEGFR2 in lung tissue was almost completely suppressed until 8 h after oral administration of TAK-593 at 1 mg/kg and was restored to baseline at 16 h after administration (Fig. 3c). The durability of this effect is likely a reflection of the long residence time of TAK-593 on VEGFR and PDGFR.

Figure 3.

Pharmacodynamic and pharmacokinetic correlation of TAK-593. (a,b) TAK-593 was orally administered at (a) 0.125 mg/kg and (b) 1 mg/kg. Plasma and lung tissue concentration of TAK-593 were determined by High performance liquid chromatography (HPLC)/MS/MS or HPLC with a fluorescence detector at the indicated times (n = 3). (c) Athymic nude mice (n = 3) were treated with a single oral dose of TAK-593 (1 mg/kg) and the lung tissue was collected at the indicated times after vascular endothelial growth factor (VEGF)injection (20 μg/mouse). Tissues were lysed and subjected to western blot analysis for phospho-VEGFR2 and VEGFR2.

TAK-593 inhibits tumor angiogenesis

Mice bearing A549 xenografts were orally treated with TAK-593 (0.25 and 1 mg/kg, twice daily). Xenografts were resected on Days 3 and 7 to analyze the tumor microvessel density, tumor cell proliferative activity, and tumor cell apoptosis. There was a strong anti-tumor effect with T/C values of 49 and 17% after 7 days of treatment at 0.25 and 1 mg/kg, respectively (Suppl. Fig. S5). Consistent with this, immunostaining of CD31 showed that TAK-593 treatment significantly decreased tumor vessel density in a dose- and time-dependent manner (Fig. 4a). Notably, the vessel density of tumors from TAK-593-treated mice was lower than before treatment, indicating that TAK-593 caused tumor vessel regression. After 3 days of treatment with TAK-593 at 1 mg/kg, we observed a reduction in Ki-67 positive (proliferating) cells and an increase in TUNEL-positive (pro-apoptotic) cells indicating that TAK-593 treatment had both anti-proliferative and pro-apoptotic effects on A549 lung cancer cells in vivo (Fig. 4b,c and Suppl. Fig. S6).

Figure 4.

TAK-593 shows anti-angiogenic, anti-proliferative, and pro-apoptotic effect in vivo. TAK-593 was orally administered to A549 xenograft mice for up to 7 days and tumors were harvested on Days 3 and 7. (a) Endothelial cells were visualized and quantified by CD31 immunostaining. (b) Tumor cells were immunostained with anti-Ki67 antibody and proliferative activity was quantified. (c) Apoptosis of tumor cells was quantified by the terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) method. Data represent the mean and standard deviation (SD) (n = 4). *≤ 0.025 versus the vehicle control group by the one-tailed Williams' test.

TAK-593 inhibits pericyte recruitment to tumor blood vessels

To assess the effect of TAK-593 on pericyte coverage towards tumor vasculature, endothelial cells and pericytes were analyzed by immunohistochemistry. TAK-593 was orally administered once daily at doses of 0.5 or 1.5 mg/kg for 3 and 7 days to mice with A549 xenograft tumors. As a result, tumor growth was similarly suppressed with the T/C values of 42% and 18% after 7 days of treatment at 0.5 and 1.5 mg/kg, respectively. For visualization of endothelial cells and pericytes, tumor tissue sections were immunostained for CD31 (a marker of endothelial cells) and αSMA (a marker of pericytes),[30, 31] followed by fluorochrome-labeled secondary antibodies. TAK-593 significantly inhibited pericyte recruitment to tumor blood vessels in a dose-dependent manner (Fig. 5a). Representative photomicrographs are shown in Figure 5b. In tumors from TAK-593-treated mice, the microvessels were surrounded by a pericyte layer that was substantially thinner than that seen in vehicle-treated tumor vessels. This observation suggests that TAK-593 inhibits the stabilization of tumor vessels by blocking pericyte recruitment.

Figure 5.

TAK-593 decreases pericyte coverage of vessels. TAK-593 or the vehicle was administered to A549 xenograft mice. (a) Tumors were resected and processed for immunohistochemical analysis at 4 h after the last dose. Sections were simultaneously stained for CD31 and α-smooth muscle actin (α-SMA) to detect endothelial cells and pericytes, respectively. The region positive for α-SMA surrounding the CD31-positive vessels was quantified in five randomly selected fields. Data represent the mean and standard deviation (SD) (n = 3). *≤ 0.025 versus the vehicle control group by the one-tailed Williams' test. (b) Representative tumor sections from mice treated with the vehicle or TAK-593 (1.5 mg/kg). White arrowhead in tumors from TAK-593-treated mice indicates the pericyte layer that was substantially thinner than that seen in vehicle-treated tumors. Bar = 100 μm. Red = CD31-positive endothelial cells; Green = α-SMA-positive pericytes

TAK-593 decreases tumor vascular permeability

Dynamic constant enhanced-MRI can be used to noninvasively assess vascular permeability as an indirect marker of angiogenesis.[32] Because VEGF inhibition is known to decrease tumor vessel permeability, DCE-MRI has been used in clinical and preclinical studies of various VEGF signaling inhibitors.[33, 34] In this study, the effects of TAK-593 on tumor vascular permeability in HT-29 human colon cancer xenografts were assessed by DCE-MRI. TAK-593 was orally administered at 0.25 and 1.5 mg/kg twice daily for 3 days and was confirmed to show a significant antitumor effect in this model (Suppl. Fig. S7). On Day 4 (one hour post-dose), mice were analyzed by DCE-MRI to directly visualize vascular permeability. Tumor vessel permeability (Ktrans value) was significantly decreased after 4 days of treatment with TAK-593 at 0.25 and 1.5 mg/kg when compared with baseline (pretreatment status) (Fig. 6a). Figure 6b shows representative images of tumors before and after 4 days of treatment with TAK-593 (1.5 mg/kg). The area of high permeability (colored region) is clearly reduced in the tumor from TAK-593-treated mice.

Figure 6.

TAK-593 affects tumor vascular permeability on dynamic constant enhanced magnetic resonance imaging (DCE-MRI). TAK-593 or the vehicle was administered for 3 days and once on Day 4 to HT-29 xenograft mice. Before and after treatment, gadolinium tetraazocyclododecane-tetraacetate (Gd-DTPA was injected intravenously and MRI was conducted. Ktrans values were obtained by calculating the Gd-DTPA concentrations in tumor tissue and plasma on the basis of a two-compartment model (n = 6–8 per group). (a) Ktrans values of tumors from vehicle or TAK-593-treated mice. Data represent the mean and SD. #≤ 0.025 versus pretreatment by the one-tailed Shirley-Williams' test. (b) Representative Ktrans maps of tumors before and after treatment with TAK-593 or the vehicle. The colored portion of the spectrum map represents higher vascular permeability. Bar = 2 mm.


There is considerable evidence that VEGF and PDGF signaling are essential to the process of tumor angiogenesis. Therefore, we sought to generate potent and selective compounds that inhibit both VEGFR2 and PDGFRβ protein kinases as potential anti-cancer drugs. In the course of optimizing such molecules, we noticed that some compounds bound to the targets with exceptionally slow off-rates (known as “pseudo-irreversible inhibition”).[28] Accordingly, such compounds were optimized using the pseudo-irreversible inhibition because the effect seemed to contribute to potent cellular activity. We ultimately selected the ATP-competitive inhibitor, TAK-593, for development based on its potency and selectivity towards VEGFR2 and PDGFRβ tyrosine kinases (IC50 = 0.95 and 13 nM, respectively). Notably, TAK-593 weakly inhibited cKIT, FGFR1, and BRAF with IC50 values of 100, 350, and 8400 nM, respectively, while the IC50 values were over 10 μM for EGFR, IGF-1R, Tie2, cMET, and Aurora-A.[28] This selective profile of TAK-593 distinguishes it from several previously reported small molecular weight VEGFR kinase inhibitors such as Sorafenib (BAY-43-9006), Cediranib (AZD-2171), Vandetanib (AZD-6474), XL880, KRN951, and ABT-869 that were reported to show more potent inhibition against BRAF, FGFR, EGFR, cMET, Tie2, and cKIT, respectively.[35-41] In cellular assays, TAK-593 potently inhibited the phosphorylation of VEGFR2 and PDGFRβ (IC50 = 0.34 and 2.1 nM, respectively), and also inhibited VEGF-induced proliferation of HUVECs and PDGF-induced proliferation of CASMCs (IC50 = 0.3 and 3.5 nM, respectively). In addition, consistent with its selective enzyme inhibition profile, TAK-593 showed much weaker inhibition of the proliferation of five cancer cell lines and fibroblasts, with an over 28 000-fold difference versus cellular VEGF inhibition. These features of potency and selectivity may be due to the “pseudoirreversible” effect of TAK-593 on VEGFR2 and PDGFRβ.

TAK-593 showed considerable anti-tumor effect against a broad spectrum of xenograft tumors, but its potency varied among the tumor types. Although these differences of sensitivity were likely due to variations in the angiogenesis or VEGF signaling dependency of each tumor, the precise mechanism remains unclear. TAK-593 also showed significant anti-tumor activity in a xenograft nude mouse model of well-established A549 human lung cancer (T/C = 33% on Day 42) at a dose of 0.125 mg/kg twice daily. Pharmacokinetic analysis showed that AUC0–24 h was 0.078 μg•h/mL at 0.125 mg/kg and total AUC by twice-daily treatment was calculated to be 0.156 μg•h/mL per day. In addition, almost complete tumor regression was achieved at 1 mg/kg twice daily. Pharmacokinetic analysis also showed that AUC0–24 h was 0.883 μg•h/mL at 1 mg/kg and total AUC was calculated to be 1.77 μg•h/mL. Moreover, although the plasma concentration was nearly undetectable at 8 h after oral dosing with TAK-593 at 1 mg/kg, almost complete inhibition of VEGFR2 phosphorylation was still observed along with its potent anti-tumor activities. These results are distinct from previous reports for more reversible kinase inhibitors such as Sunitinib for which continuous or adequate plasma concentration is necessary to show the inhibition of VEGFR2 tyrosine phosphorylation in the tumor, as well as in a surrogate tissue (lung), which strongly correlated with tumor growth inhibition.[42, 43] Thus, it may be possible that the pseudo-irreversible effect of TAK-593 (long duration of action on VEGFR2 and PDGFRβ) greatly contributes to its strong anti-tumor activity in vivo.

Several lines of evidence indicate that the in vivo anti-tumor effect of TAK-593 is mediated through an anti-angiogenic mechanism. First, we have demonstrated suppression of endothelial tube formation in vitro. Consistent with this finding, we have demonstrated that administration of TAK-593 to xenograft tumor-bearing mice leads to a decrease in microvessel density. Although TAK-593 had much weaker inhibition of A549 cell proliferation in vitro, we observed strong inhibition of tumor growth as well as increased survival in tumor-bearing mice with anti-proliferative and pro-apoptotic effects on A549 cells in vivo. Taken together, these data show that the anti-tumor effect of TAK-593 is mediated through an anti-angiogenic mechanism rather than a direct effect on tumor cell growth. Pericytes are contractile cells that exist in close contact with endothelial cells in the capillaries, and activated tumor pericytes are characterized by expression of α-SMA.[44] It has been shown that PDGF-BB secreted by endothelium is necessary for the recruitment of pericytes to newly formed blood vessel. Treatment with TAK-593 also decreased pericyte coverage into endothelial cells, indicating that inhibition of PDGF signaling by TAK-593 had additional inhibitory activity on tumor angiogenesis. Tumor vessels are heavily dependent on VEGF signaling for their survival,[45] whereas inhibition of VEGF alone leads to increased pericyte coverage of tumor blood vessels for the protection of the endothelial cells from loss of VEGF signaling.[46] Therefore, dual inhibition of VEGF and PDGF signaling by TAK-593 would be more effective against tumor vessel development through disruption of pericyte-endothelial cell association.

In addition, we have demonstrated that non-invasive DCE-MRI techniques can be used to visualize the effects of TAK-593 on the tumor vasculature. For the DCE-MRI technique, early Gd-DTPA enhancement in the tumor has been accepted as indicating well-vascularized tumors, while delayed or no Gd-DTPA enhancement reveals a poorly-vascularized tumor or a necrotic region.[32-34] In this study, Gd-DTPA enhancement was calculated as the Ktrans value to assess vascular permeability, revealing that TAK-593 reduced Gd-DTPA enhancement and tumor vessel permeability from an early stage of treatment. The change of Ktrans was greater than those of tumor volume. Thus, Gd-DTPA enhancement could be used as a noninvasive pharmacodynamic biomarker for TAK-593.

In conclusion, TAK-593 had dual potent inhibition of both VEGF and PDGF signaling. The strong activities of TAK-593 on tumor growth and angiogenesis results in marked tumor regression combined with good tolerability. Accordingly, TAK-593 is a novel therapeutic agent for use against solid tumors.


We thank Drs Takayasu Ito, Takeshi Watanabe, and Ms Michiko Tawada for their contribution to this project and Ms Midori Yamasaki for pharmacological assistance. We also thank Dr Takashi Yokawa (BioView, Yokohama, Japan) for the technical support of DCE-MRI experiments. We finally thank Drs Masami Kusaka, Osamu Nakanishi, Shuichi Furuya, and Hideaki Nagaya for their supports during this project.

Disclosure Statement

The authors have no conflict of interest.


alpha smooth muscle actin


area under the plasma concentration-time curve for 24 hours


twice daily


coronary artery smooth muscle cells


maximum drug concentration


dynamic constant enhanced magnetic resonance imaging


Gadolinium tetraazocyclododecane-tetraacetate


Hanks' balanced salt solution


human umbilical vein endothelial cells


concentration causing 50% inhibition


normal human dermal fibroblasts


once daily


severe combined immunodeficiency


platelet derived growth factor


PDGF receptor β


treated/control ratio


terminal deoxynucleotidyl transferase dUTP nick-end labeling


vascular endothelial growth factor


VEGF receptor 2