Antiangiogenic therapies targeting VEGFA have been commonly used in clinics to treat cancers over the past decade. However, their clinical efficacy has been limited, with drawbacks including acquisition of resistance and activation of compensatory pathways resulting from elevated circulating VEGFB and placental growth factor (PlGF). To bypass these disadvantages, we developed a novel glycosylated soluble decoy receptor fusion protein, VEGF-Grab, that can neutralize VEGFA, VEGFB, and PlGF. VEGF-Grab has the second and third immunoglobulin (Ig)-like domains of VEGF receptor 1 (VEGFR1) fused to IgG1 Fc, with three potential glycosylation sites introduced into the third Ig-like domain of VEGF-Grab by mutagenesis. Compared with VEGF-Trap, VEGF-Grab showed more potent decoy activity against VEGF and PlGF, mainly attributed to the VEGFR1 backbone. Most importantly, the negatively charged O-glycans attached to the third Ig-like domain of VEGFR1 counterbalanced the originally positively charged VEGFR1 backbone, minimizing nonspecific binding of VEGF-Grab to the extracellular matrix, and resulting in greatly improved pharmacokinetic profile. These advancements led to stronger and more durable antiangiogenic, antitumor, and antimetastatic efficacy in both implanted and spontaneous tumor models as compared with VEGF-Trap, while toxicity profiles were comparable with VEGF-Trap. Collectively, our results highlight VEGF-Grab as a promising therapeutic candidate for further clinical drug development. Mol Cancer Ther; 14(2); 470–9. ©2014 AACR.

This article is featured in Highlights of This Issue, p. 315

VEGFA is a critical regulator of tumor angiogenesis, mainly through the activation of its primary receptor, VEGF receptor 2 (VEGFR2; refs. 1–3). VEGFA is expressed in most tumor cells and corresponding stromal cells throughout tumor progression, whereas VEGFR2 is highly expressed in growing tumor vessels, leading to the formation of structurally and functionally malformed tumor blood vessels (1, 4). VEGFA specifically binds to the second immunoglobulin (Ig) homology domain (D2) of the extracellular region of VEGFR2, resulting in activation of proangiogenic signaling (5). For the past decade, much effort has been devoted to targeting this VEGFA/VEGFR2 signaling pathway using monoclonal antibodies, soluble decoy receptor fusion proteins, and small-molecular inhibitors in patients with cancer (6–9). Although the current therapeutic blockade of VEGFA/VEGFR2 signaling provides clinical benefits, the anticancer effect is transient, eventually giving rise to acquired resistance through the activation of alternative proangiogenic pathways and further recruitment of proangiogenic cells such as tumor-associated macrophages (TAM; refs. 10–12). These limitations highlight current unmet needs in antiangiogenic cancer treatment strategies, which must be addressed for successful therapy development.

VEGFA also binds to VEGFR1 with higher affinity (<10–20 pmol/L) than VEGFR2 (<100–125 pmol/L; ref. 13). In addition, VEGFR1 is a receptor for other proangiogenic ligands, VEGFB and placental growth factor (PlGF), which have recently been highlighted as alternative targets for antiangiogenic therapy (14–17). Because of its ability to bind multiple proangiogenic ligands, VEGFR1 has been considered as a potential backbone for the development of a novel decoy receptor fusion protein for therapeutic purposes. However, the efficiency of a decoy receptor fusion protein, which consisted of the first three Ig domains of VEGFR1 fused with the Fc region of IgG1 (VEGFR1-Fc), proved unsatisfactory due to nonspecific binding to the extracellular matrix (ECM) attributed to the abundant positively charged residues in the third Ig domain (VEGFR1 D3) and its high isoelectric point (pI; ref. 7). Nonetheless, this finding inspired the invention of VEGF-Trap (Aflibercept, Regeneron), consisting of VEGFR1 D2 and VEGFR2 D3 fused to IgG1 Fc. By switching VEGFR1 D3 to VEGFR2 D3, net pI of VEGF-Trap was decreased, resulting in less ECM bindings and an improved pharmacokinetic (PK) profile compared with VEGFR1-Fc (7). However, because VEGFR2 D3 was used instead of VEGFR1 D3, the high-affinity binding of VEGFA and PlGF was disturbed (18). Hence, the important issue to be addressed now is how to incorporate VEGFR1 D3 into a decoy receptor while minimizing nonspecific ECM binding.

Glycosylation is a posttranslational modification resulting in the addition of carbohydrate chains to specific asparagine (N-linked glycosylation) or serine/threonine (O-linked glycosylation) residues. Glycosylation of secreted and membrane proteins affects their biochemical and biologic properties. It usually provides a negative charge and increases solubility, thus diminishing nonspecific binding to ECM. Moreover, glycosylation grants resistance to proteolysis and extended serum half-life, enhancing a protein's PK profile (19). Glyco-engineered therapeutic proteins such as Aranesp (erythropoietin) from Amgen and Gazyva (obinutuzumab) from Genentech are good examples that exploited these advantages (20, 21).

Here, we developed a novel VEGF decoy receptor fusion protein, VEGF-Grab. Parental VEGFR1-Fc (VEGFR1 D2–D3 fused to Fc) was used as a backbone, and new potential glycosylation sites were introduced into the positively charged patch of VEGFR1 D3 by site-directed mutagenesis. This engineered VEGF-Grab showed significantly improved decoy efficiency and a dramatic decrease in net pI, thus attenuating nonspecific ECM binding and enhancing PK profiles. Furthermore, VEGF-Grab strongly suppressed tumor angiogenesis, progression, and metastasis via effective capturing of three VEGFR1 ligands, VEGFA, VEGFB, and PlGF.

Cell lines

Human umbilical vein endothelial cell (HUVEC; CC-2519, Lonza), dhfr-deficient CHO cells (CRL-9096, ATCC), and Lewis lung carcinoma (LLC) cells (CRL-1642, ATCC), which were tested and authenticated according to the ATCC guidelines, were obtained from indicated manufacturer. They have been cultured as reported earlier (22) for fewer than 6 months after receipt.

Recombinant proteins expression and purification

Details for the construction of recombinant DNAs are described in Supplementary Methods and Supplementary Table S1. Parental VEGFR1-Fc, VEGF-Grabs, or VEGF-Trap constructs were transfected into dhfr-deficient CHO cells using Lipofectamine 2000 (Life technologies) and cultured as previously reported (22). Secreted VEGFR1-Fc, VEGF-Grabs, and VEGF-Trap into cultured media were purified by protein A-sepharose affinity chromatography (3, 23).

Antibodies

Antibodies used in this study are listed in Supplementary Table S3.

Solid phase binding assay

Solid phase binding assay was performed as previously described (22). Briefly, varying amounts (0.1 nmol/L to 10 μmol/L) of VEGF-Grabs or VEGF-Trap were added to MaxiSorp 96-well plates (Nunc) where either hVEGFA165 (150 ng/mL), PlGF (62.5 ng/mL), or hVEGFB (125 ng/mL) was coated. VEGF-Grabs or VEGF-Trap bound to coated proteins were detected by ELISA with horseradish peroxidase (HRP)–conjugated anti-human Fc antibody.

Isoelectric focusing analysis

Of note, 5 μg of proteins (VEGF-Grabs and VEGF-Trap) and IEF marker (Novex) were loaded on IEF gels ranging pH 3 to 10 (Novex) and run as previously described (24). After electrophoresis, gels were stained with silver staining kit (Elpis).

Glycosylation analyses

N-glycosylation mapping was performed by LC/MS as described previously (25). O-glycosylation was analyzed by glyco-analytical multispecific proteolysis (glyco-AMP; ref. 26). In addition, N-glycan compositions were analyzed using nano-LC/MS and MALDI-MS and identified by GP Finder software package (27).

In vitro ECM-binding assay

In vitro ECM-binding assay was performed as previously described (22). Briefly, varying amounts of VEGF-Grabs or VEGF-Trap (5–80 nmol/L) were added to the ECM-coated plate (Matrigel Thin-Layer Plates, BD Biosciences). ECM-bound proteins were detected by ELISA with HRP-conjugated anti-human Fc antibody (22).

Inhibition of VEGFR2 signaling by VEGF-Grabs

HUVECs were treated with VEGF-Grabs or VEGF-Trap (14 nmol/L) 15 minutes followed by VEGFA (R&D systems) treatment (1 nmol/L) for 10 minutes. Then, cells were washed and lysed in lysis buffer (3). Of note, 50 μg of total proteins were loaded on 10% SDS-PAGE. VR2, ERK, and its phosphorylated form were detected by the Western blot analysis with corresponding antibody (22).

Cell survival assay

When the seeded HUVECs became confluent, cells were starved in OPTI-MEM (Invitrogen) concurrently treated with VEGF-Grabs or VEGF-Trap (0.35, 0.7, 3.5, 7, 35, and 70 nmol/L) in the presence or absence of VEGFA (0.2 nmol/L). After 36 hours, WST-1 (water-soluble tetrazolium salt, DOGEN) was added and absorbance was measured at 450 nm.

Migration assay

When seeded HUVECs on the “culture-insert of μ-Dish” (Cat#.81176, Ibidi) became confluent, the culture-inserts were removed to generate wound gap (22). Then, migrated cells within the wound were monitored for 12 hours in the presence of VEGFA (1 nmol/L) and indicated proteins (14 nmol/L) in EBM-2 (Lonza).

Tube formation assay

HUVECs were seeded (5 × 104 cells/well) on Matrigel (4 mg, BD Biosciences)-coated 24-well plates and treated with VEGF-Grabs or VEGF-Trap (14 nmol/L). After 15 minutes, VEGFA (1 nmol/L) was added and tube formations were monitored for 12 hours (28).

Mice

Specific pathogen-free male C57BL/6J and female MMTV-PyMT transgenic mice (FVB/N) were purchased from Jackson Laboratory. All mice were anesthetized with 80 mg/kg of ketamine and 12 mg/kg of xylazine, before sacrifice. All animal care and experimental procedures were performed under the approval (KA2013-42) from the Animal Care Committee of Korea Advanced Institute of Science and Technology (KAIST; Daejeon, Korea).

PK analysis

Ten-week-old C57BL/6J mice were given subcutaneous injections of 4, 10, or 25 mg/kg VEGF-Grabs and VEGF-Trap (n = 3, respectively). Blood samples were collected from tails at 1, 2, 4, 8, 12, 24, 48, 96, and 144 hours. The levels of indicated proteins in serum or organs were measured by ELISA with HRP-conjugated anti-human Fc antibody (22).

Tumor models and treatment regimes

LLC cells (1 × 106 cells/100 μl) were subcutaneously injected into the dorsal flank of C57BL/6J (8–10 week old; refs. 22, 28). Tumor volume was calculated according to the formula: 0.5 × length × width2. VEGF-Grab3 or VEGF-trap (indicated dose) was intraperitoneally injected at given time points. For the combined therapy of VEGF-Grab3 with chemotherapeutics, cisplatin (10 mg/kg, Sigma-Aldrich) was intraperitoneally injected at day 9 into LLC tumor-bearing mice that were receiving injections of either VEGF-Grab3 or VEGF-Trap. MMTV-PyMT mice (12-weeks-old) received IP-injections of either VEGF-Grab3 or VEGF-Trap (25 mg/kg) twice per week for 3 weeks. Mice were anesthetized and their primary tumors, lymph nodes (LN), and organs were harvested for histologic analyses.

Histologic analyses

Tumors were processed and stained as previously described (28). Samples were sectioned and then stained with hematoxylin and eosin (H&E) or corresponding antibodies. To detect hypoxia, Hypoxyprobe-1 (60 mg/kg, pimonidazole hydrochloride, Hypoxyprobe, Inc.) was intravenously injected 90 minutes before sacrifice as previously reported (22). The CD11b+, cytokeratin+, and caspase-3+ areas were calculated as a percentage per total sectional area.

Statistical analyses

Statistical differences between means were determined by independent sample t test. Statistical significance was set at P < 0.05.

Design of VEGF-Grabs and their enhanced binding affinities for VEGFA and PlGF

VEGFR1 has seven Ig-like domains in the extracellular domain (Fig. 1A). Among them, VEGFR1 D2 is the primary contributor to VEGFA and PlGF binding. In addition, residues in VEGFR1 D3 also participate in the high-affinity binding of VEGFA and PlGF (18). Therefore, the VEGFR1 D2–D3 is the minimal required domain to bind both VEGFA and PlGF with high affinity. To design a novel VEGF decoy receptor fusion protein, we first analyzed the model structures of the VEGFR1 D2–D3/VEGFA complex generated by MODELLER (29) using template structures (PDB ID: 1FLT and 2X1X; Fig. 1B and C). The electrostatic potential analysis revealed that VEGFR1 D3 has abundant positively charged amino acids (shown in blue), responsible for the high pI of this domain compared with VEGFR2 D3 (Fig. 1B). In an attempt to reduce the net pI of VEGFR1 D3 and improve in vivo half-life of VEGF decoy receptor fusion protein, we designed three VEGFR1 variants; VEGF-Grab1, VEGF-Grab2, and VEGF-Grab3 (Fig. 1A). Three positive-charged residues, R135, K138 and the R172, within the VEGFR1 D3 loop region were mutated to negative-charged residues (R135S, K138T, and R172N) where glycans could be attached (Fig. 1A and Supplementary Fig. S1). All VEGF-Grab constructs consist of VEGFR1 D2–D3 variants fused to Fc (Fig. 1A). The model structure of the VEGFA/VEGFR1 D2–D3 complex shows that these selected residues are not involved in ligand binding, but are located within the flexible loops. Thus, it was predicted that these mutants would maintain their high affinities to VEGFA/PlGF and avoid structural disruption (Fig. 1C). Unfortunately, the expression levels of parental VEGFR1-Fc and VEGF-Grab2 in CHO cells were too low, so further analyses were performed only with VEGF-Grab1 and VEGF-Grab3. All of the proteins, VEGF-Grab1, VEGF-Grab3, and VEGF-Trap (hereafter, abbreviated as G1, G3, and VT, respectively), were produced from CHO cell and purified by ProteinA affinity chromatography. Purified G1 and G3 showed diffuse band patterns in reduced SDS-PAGE condition (Fig. 1D), which is a typical characteristic of glycosylated proteins (20). Under nonreduced conditions, VEGF-Grabs were dimer due to the disulfide bond in the Fc (Fig. 1D). The in vitro binding affinities of VEGF-Grabs and VT to proangiogenic ligands, VEGFA, PlGF, and VEGFB, showed that G1 (KD = 7.9 × 10−10 mol/L) and G3 (KD = 5.6 × 10−10 mol/L) had 1.1- and 1.5-fold higher affinity to VEGFA than VT (KD = 8.4 × 10−10 mol/L; Fig. 1E). However, the binding affinities of G1 and G3 to VEGFB, which only requires VEGFR1-D2 for binding, were similar to VT (Fig. 1G). Intriguingly, the binding affinities of G1 (KD = 2.8 × 10−9 mol/L) and G3 (KD = 6.9 × 10−9 mol/L) to PlGF were 18.5- and 6.7-fold more potent than VT (KD = 4.6 × 10−8 mol/L), respectively (Fig. 1F), suggesting that we successfully generated new glycosylated-VEGF decoy receptor fusion proteins, G1 and G3, which have significantly higher affinities to both VEGFA and PlGF.

Figure 1.

Generation and characterization of VEGF-Grab. A, schematic diagram of VEGF-Grabs (VEGF-Grab1, VEGF-Grab2, and VEGF-Grab3; hereafter referred to as G1, G2, and G3, respectively) and VEGF-Trap (VT). Mutated residues in VEGF-Grabs are indicated by brown at the bottom. B, electrostatic potential of the D2–D3 domain in VEGFR1 (left) and VEGFR2 (right). Positively and negatively charged residues are colored in blue and red, respectively. C, model structure of the VEGFA/VEGFR1 D2–D3 complex. VEGFA dimer is colored in yellow and green. VEGFR1 D2 and D3 are colored in tan and mustard. Residues indicated by blue stick are target sites for mutagenesis. D, SDS-PAGE analysis of G1 and G3 in reduced (R) and nonreduced (NR) conditions. E–G, binding affinities of G1, G3, and VT for VEGFA (E), PlGF (F), or VEGFB (G) *, P < 0.05 G1 versus VT; #, P < 0.05 G3 versus VT. For each group, n = 3. Values are mean ± SD.

Figure 1.

Generation and characterization of VEGF-Grab. A, schematic diagram of VEGF-Grabs (VEGF-Grab1, VEGF-Grab2, and VEGF-Grab3; hereafter referred to as G1, G2, and G3, respectively) and VEGF-Trap (VT). Mutated residues in VEGF-Grabs are indicated by brown at the bottom. B, electrostatic potential of the D2–D3 domain in VEGFR1 (left) and VEGFR2 (right). Positively and negatively charged residues are colored in blue and red, respectively. C, model structure of the VEGFA/VEGFR1 D2–D3 complex. VEGFA dimer is colored in yellow and green. VEGFR1 D2 and D3 are colored in tan and mustard. Residues indicated by blue stick are target sites for mutagenesis. D, SDS-PAGE analysis of G1 and G3 in reduced (R) and nonreduced (NR) conditions. E–G, binding affinities of G1, G3, and VT for VEGFA (E), PlGF (F), or VEGFB (G) *, P < 0.05 G1 versus VT; #, P < 0.05 G3 versus VT. For each group, n = 3. Values are mean ± SD.

Close modal

Identification of newly added O-glycan on VEGF-Grabs

Because glycosylation can alter the pI of proteins (30), we measured the pI of the VEGF-Grabs. Because of the diverse composition of attached glycans, VT, G1, and G3 exhibited micro-heterogeneity with diverse isoforms on the isoelectric focusing gel (Fig. 2A). Intriguingly, the pIs of G1 and G3 were dramatically decreased to 8.0 and 7.4, respectively (Fig. 2B), compared with that of parental VEGFR1-Fc (pI: 9.4; ref. 7). These pI values were comparable with that of VT (pI: 7.8). After digestion with PNGaseF, the molecular weights of the VEGF-Grabs and VT were decreased (Fig. 2C), indicating the presence of N-linked glycosylation in VT, G1, and G3. In contrast with VT, even after PNGaseF digestion, G1 and G3 still displayed diffuse band patterns suggesting their O-linked glycosylation.

Figure 2.

VEGF-Grab3 exhibits low ECM binding and prolonged PK profiles. A, isoelectric point analysis of G1, G3, and VT. Red lines on each band were used to analyze net pI. B, comparison of pI for each protein. Net pI of each protein was the mean pI of each isoform denoted as lines. C, PNGase F digestion to remove N-linked glycan. D, analysis of O-linked glycosylation at Serine135 of G3. E, schematic diagram for glycosylated sites analyzed by mass spectrometry. Occupied N-glycan (red); occupied O-glycan (blue); unoccupied N- or O-glycosylation sites (gray). F and G, PK profiles analysis of VT (F) and G3 (G) at varying doses. H, comparison of AUC of VT and G3. I and J, tissue distribution of G3 and VT 48 hours after subcutaneous injection (4 mg/kg). I, accumulated VT and G3 in tumor. J, relative accumulated levels of VT and G3 in liver and kidney compared to tumor. K, analysis of in vitro ECM-binding affinities with Matrigel-coated plates. *, P < 0.05 G3 versus VT. Each group, n = 3. Values are mean ± SD.

Figure 2.

VEGF-Grab3 exhibits low ECM binding and prolonged PK profiles. A, isoelectric point analysis of G1, G3, and VT. Red lines on each band were used to analyze net pI. B, comparison of pI for each protein. Net pI of each protein was the mean pI of each isoform denoted as lines. C, PNGase F digestion to remove N-linked glycan. D, analysis of O-linked glycosylation at Serine135 of G3. E, schematic diagram for glycosylated sites analyzed by mass spectrometry. Occupied N-glycan (red); occupied O-glycan (blue); unoccupied N- or O-glycosylation sites (gray). F and G, PK profiles analysis of VT (F) and G3 (G) at varying doses. H, comparison of AUC of VT and G3. I and J, tissue distribution of G3 and VT 48 hours after subcutaneous injection (4 mg/kg). I, accumulated VT and G3 in tumor. J, relative accumulated levels of VT and G3 in liver and kidney compared to tumor. K, analysis of in vitro ECM-binding affinities with Matrigel-coated plates. *, P < 0.05 G3 versus VT. Each group, n = 3. Values are mean ± SD.

Close modal

To further confirm the presence of N- or O-glycans at the mutated sites of VEGF-Grabs, we performed a mass spectrometry for glycosylation mapping. Calculated masses of both deglycosylated and unglycosylated tryptic peptides that encompass a potential site for N-glycosylation are listed in Supplementary Fig. S2A. LC/MS data indicate that the mutated Asn172 on G3 was not glycosylated, whereas three original N-glycosylation sites, Asn61, Asn93, and Asn308 (Asn328 for VT), were all fully occupied (Fig. 2E and Supplementary Fig. S1). Furthermore, from the MS/MS spectrum using glyco-AMP (26), O-Glycopeptides (STPSPV+HexNAc1Hex1NeuAc1) were observed in both G1 and G3 (Fig. 2D and Supplementary Fig. S3). Typical glycan fragments for HexNAc, NeuAc-H2O, NeuAc, Hex1NeuAc1, and HexNAc1Hex1NeuAc1 confirmed that VEGF-Grabs are O-glycosylated at Ser135. However, we found no evidence of O-glycosylation at Thr138.

We then analyzed the overall N-glycan compositions of G1, G3, and VT. N-glycan profile was quantified by both nano-LC/MS and MALDI-MS (27). These data indicated that G1 and G3 exhibited an increase in high mannose glycosylation, a decrease in fucosylated and complex-undecorated glycans (C/H-Fuc) as compared with VT (Supplementary Fig. S2B). Particularly, G3 displayed increased sialylation compared with G1 leading to G3′s lower pI as compared with G1 (Fig. 2A). This mass spectrometry analysis showed that both G1 and G3 have only one additional O-glycosylation site (Ser135) among the mutated residues (Fig. 2E and Supplementary Figs. S1–S3), as well as three N-glycosylation sites at 61N, and 93N in VEGFR1 D2 and 308N (328N for VT) in the Fc. In addition, G3 was revealed to be more sialylated than VT or G1.

VEGF-Grab3 exhibits decreased ECM bindings and enhanced PK profiles

Proteins with high pI values bind nonspecifically to the ECM, resulting in poor PK and bioavailability (24). We confirmed that the reduced pI values of VEGF-Grabs indeed led to the decreased in vitro ECM binding, comparable with that of VT (Fig. 2K). In vivo ECM-binding analysis also revealed that, although VEGFR1-Fc bound to the tumor section nonspecifically, no VT or G3 was detected in the tumor section (Supplementary Fig. S4), suggesting that the charge conversion at the three mutation sites and additionally added glycans to Ser135 of G3 allowed it to effectively overcome the intrinsic problems of VEGFR1-Fc, nonspecific binding to ECM. To test whether G1 and G3 displayed an improved in vivo half-life, we analyzed their PK profiles at 4 mg/kg dose for 4 days. Interestingly, G3 displayed improved PK profiles (AUC: 59.44 μg × days/mL) compared with G1 (AUC: 29.08 μg × days/mL; Supplementary Fig. S5). Therefore, we chose G3 for further in vivo anticancer studies. Then, we evaluated the PK profiles of VT and G3 at varying doses (4, 10, 25 mg/kg) for 6 days (Fig. 2F and G). VT showed an AUC of 37.57, 34.07, and 65.02 μg × days/mL, whereas G3 showed an AUC of 64.36, 65.68, and 117.5 μg × days/mL after 4, 10, and 25 mg/kg injections, respectively (Fig. 2H). This reflected a 1.7-, 1.9-, 1.8-fold increase over VT, respectively. This enhanced PK profile of G3 also supports low ECM binding of G3 in vivo. Moreover, PK profiles demonstrated that VT was mostly eliminated by 6 days postinjection, whereas G3 levels at day 6 remained 2- to 5-fold higher than VT (Fig. 2F and G), indicating that G3 has a prolonged half-life in serum. We also examined the accumulated levels of VT and G3 in liver, kidney, tumor, and urine of LLC tumor-bearing mice 48 hours after subcutaneous injections (4 mg/kg). Higher G3 levels were detected in liver, kidney, and particularly in tumor (18.9-fold increase) than for VT (Fig. 2I and Supplementary Fig. S6). However, the relative amounts of accumulated G3 in liver and kidney versus tumor were much lower than that of VT (Fig. 2J), suggesting that most G3 accumulate at the tumor site, where VEGFA and PlGF are predominantly produced. No G3 or VT was detected in urine under our experimental conditions. Taken together, these findings suggest that VEGF-Grab3 has lower ECM-binding properties and prolonged half-life.

VEGF-Grabs inhibit EC survival, migration, and tube formation via suppression of the VEGF signaling pathway

VEGFA promotes proliferation, migration, and survival of endothelial cells through VEGFR2 activation (5). Accordingly, we examined VEGFR2 signaling in HUVECs. Both VEGF-Grabs and VT attenuated VEGFA-induced phosphorylation of VEGFR2 and its downstream ERK (Fig. 3A–C and Supplementary Figs. S7A–S7C and S8). In addition, VEGF-Grabs inhibited VEGFA-induced proliferation of HUVECs with an IC50 at 1.7 and 2.4 nmol/L, which are 2.8- and 2-fold more efficient than VT (IC50 = 4.8 nmol/L), respectively (Fig. 3D). Also, VEGF-Grabs and VT strongly suppressed VEGFA-induced migration (Fig. 3E and F and Supplementary Fig. S7D and S7E) and tube formation (Fig. 3G and H and Supplementary Fig. S7F and S7G) of HUVECs. These results indicate that VEGF-Grabs and VT inhibit VEGFA-induced endothelial cell activation at comparable levels, through VEGFA sequestration.

Figure 3.

VEGF-Grabs inhibit EC survival, migration, and tube formation via suppression of the VEGF signaling pathway. A–C, inhibition of VEGFA-induced phosphorylation of VEGFR2 and ERK in HUVECs by the treatment of G1, G3, and VT. Immunoblotting (A) and quantification (B and C). D, cell survival assay with HUVECs after G1, G3, and VT treatment (0.35, 0.7, 3.5, 7, 35, 70 nmol/L) in the presence of VEGFA (0.2 nmol/L). E and F, cell migration assay with HUVECs in the presence of VEGFA and indicated proteins. Images (E) and quantification (F) of migration area. Wound-healing areas are indicated in red. G and H, images (G) and quantification (H) for tube formation assay with HUVECs in the presence of VEGFA and indicated proteins. *, P < 0.05 versus control; #, P < 0.05 G3 versus VT. Each group, n = 3. Values are mean ± SD.

Figure 3.

VEGF-Grabs inhibit EC survival, migration, and tube formation via suppression of the VEGF signaling pathway. A–C, inhibition of VEGFA-induced phosphorylation of VEGFR2 and ERK in HUVECs by the treatment of G1, G3, and VT. Immunoblotting (A) and quantification (B and C). D, cell survival assay with HUVECs after G1, G3, and VT treatment (0.35, 0.7, 3.5, 7, 35, 70 nmol/L) in the presence of VEGFA (0.2 nmol/L). E and F, cell migration assay with HUVECs in the presence of VEGFA and indicated proteins. Images (E) and quantification (F) of migration area. Wound-healing areas are indicated in red. G and H, images (G) and quantification (H) for tube formation assay with HUVECs in the presence of VEGFA and indicated proteins. *, P < 0.05 versus control; #, P < 0.05 G3 versus VT. Each group, n = 3. Values are mean ± SD.

Close modal

VEGF-Grab3 displays enhanced antitumor activity

To evaluate the antitumor effects of G3, we used the LLC tumor model (31) and treated them with either VT or G3 at 25 mg/kg. G3 treatment resulted in 61% and 71% reduction in tumor volume and weight, whereas VT treatment showed 37% and 29% decreases, respectively (Fig. 4A and C). In addition, intratumoral necrosis was more dramatic in G3-treated tumors (35%) than VT-treated tumors (21%; Fig. 4B and D). Furthermore, G3-treated tumors exhibited superior antiangiogenic effects in both peri- and intratumoral regions and antimetastatic effects compared with VT-treated tumors, even though there was no significant difference in the lymphatic vascular density (Fig. 4E–I).

Figure 4.

VEGF-Grab3 effectively suppresses tumor growth, angiogenesis, and metastasis in LLC tumors. A–N, mice were treated with proteins on the indicated days (arrows). A and C, comparison of tumor growth (A) and tumor weights (C). B and D, images (B) and quantification (D) of intratumoral necrotic area stained with H&E. Dotted line demarcates intratumoral necrosis. E and F, images (E) and quantification (F) of CD31+ blood vessels in the peri- and intratumoral area. G, images showing cytokeratin+ tumor cell metastasis (red) in inguinal LNs. Each indicated region (squares) is magnified in the bottom panel. H and I, quantifications of lymphatic vascular densities (LVD; H) and cytokeratin+ tumor cell metastasis (I). J and K, images (J) and quantifications (K) of hypoxyprobe+ hypoxic areas (green) in tumors. L and M, images (L) and quantifications (M) of CD11b+ myeloid cells (red) in tumor. N, comparisons of mRNA expression levels of various genes in intratumoral tissue after treatment with G3 and VT. Values indicate fold changes over control tumors. O–R, comparative dose responses of VT and G3 on tumor growth and metastasis. LLC tumor-bearing mice were treated with either VT or G3 at the indicated days (arrows), respectively. O and P, comparison of tumor growth after VT (O) or G3 (P) treatment. Q and R, images (Q) and quantification (R) of cytokeratin+ tumor cell metastasis in inguinal LNs. Scale bars, 100 μm. *, P < 0.05 versus control; #, P < 0.05 G3 versus VT. S–U, combination therapy of cisplatin (green arrows, at day 9 after implantation) with either VT or G3 at the indicated days (black arrows). S, comparison of LLC tumor growth. Images (T) and comparison (U) of caspase-3+ apoptotic cells (red) in intratumoral area. For each group n = 5. Values are mean ± SD. Scale bars, 100 μm. *, P < 0.05 versus cisplatin; #, P < 0.05 cis+G3 versus cis+VT.

Figure 4.

VEGF-Grab3 effectively suppresses tumor growth, angiogenesis, and metastasis in LLC tumors. A–N, mice were treated with proteins on the indicated days (arrows). A and C, comparison of tumor growth (A) and tumor weights (C). B and D, images (B) and quantification (D) of intratumoral necrotic area stained with H&E. Dotted line demarcates intratumoral necrosis. E and F, images (E) and quantification (F) of CD31+ blood vessels in the peri- and intratumoral area. G, images showing cytokeratin+ tumor cell metastasis (red) in inguinal LNs. Each indicated region (squares) is magnified in the bottom panel. H and I, quantifications of lymphatic vascular densities (LVD; H) and cytokeratin+ tumor cell metastasis (I). J and K, images (J) and quantifications (K) of hypoxyprobe+ hypoxic areas (green) in tumors. L and M, images (L) and quantifications (M) of CD11b+ myeloid cells (red) in tumor. N, comparisons of mRNA expression levels of various genes in intratumoral tissue after treatment with G3 and VT. Values indicate fold changes over control tumors. O–R, comparative dose responses of VT and G3 on tumor growth and metastasis. LLC tumor-bearing mice were treated with either VT or G3 at the indicated days (arrows), respectively. O and P, comparison of tumor growth after VT (O) or G3 (P) treatment. Q and R, images (Q) and quantification (R) of cytokeratin+ tumor cell metastasis in inguinal LNs. Scale bars, 100 μm. *, P < 0.05 versus control; #, P < 0.05 G3 versus VT. S–U, combination therapy of cisplatin (green arrows, at day 9 after implantation) with either VT or G3 at the indicated days (black arrows). S, comparison of LLC tumor growth. Images (T) and comparison (U) of caspase-3+ apoptotic cells (red) in intratumoral area. For each group n = 5. Values are mean ± SD. Scale bars, 100 μm. *, P < 0.05 versus cisplatin; #, P < 0.05 cis+G3 versus cis+VT.

Close modal

Anti-VEGF therapy induces hypoxia, which in turn stimulates the recruitment of TAM into intratumoral hypoxic region. TAM usually express profound proangiogenic factors to revascularize the tumor (32). Interestingly, despite such a significant increase in hypoxia in both VT- and G3-treated tumors (Fig. 4J and K), the level of macrophage infiltration in the G3-treated group remained nearly the same as that in nontreated control tumor (Fig. 4L and M), which, we believe, is attributed to the increased affinity of G3 to PlGF compared with VT (14). We next compared the various gene expression profiles in tumors (Supplementary Table S2). G3 treatment downregulated proangiogenic genes, including Vegfa, Plgf, Vegfc, Vegfr1, and Vegfr2. G3 also reduced Bv8 and Ccl2 expression compared with VT. These two genes are critical in myeloid cell recruitment to the tumor which can subsequently cause refractoriness to antiangiogenic therapy (Fig. 4N; refs. 33, 34). Furthermore, no obvious differences were observed in H&E-stained sections of vital organs, including heart, lung, liver, and kidney, of VT- or G3-treated mice (Supplementary Fig. S9). We also assessed the antitumor effect of G3 on established macroscopic tumors (>500 mm3). These results showed 27% and 46% delays in tumor growth after VT and G3 treatment, respectively (Supplementary Fig. S10), implying that G3 is a promising agent even against bulky macroscopic tumors.

To confirm dose responsiveness, tumor-bearing mice received injections of varying doses (5, 10, 25, and 50 mg/kg) of VT or G3 every 2 days. The tumor growth of the G3-treated group was gradually reduced in a dose-dependent manner with the maximal effect at 50 mg/kg. However, no distinct differences on tumor growth were observed between 25 mg/kg and 50 mg/kg in the VT-treated group, whose efficiency, strikingly, was comparable with that of 10 mg/kg treatment of G3 (Fig. 4O and P). In terms of tumor metastasis, VT and G3 both showed maximal antimetastatic effects at 50 mg/kg (Fig. 4Q and R). No significant changes were found in kidney or liver stained with H&E for any dosage of VT or G3 (Supplementary Fig. S11). Taken together, these results suggest that the higher affinity to VEGFA and PlGF and prolonged half-life of G3 contribute to its higher efficacy in cancer treatment over VT.

Combining anti-VEGF therapy with chemotherapeutics is a valuable therapeutic strategy, as anti-VEGF therapy alone is not sufficient to induce complete regression of bulky tumors (35). The combined therapy of G3 and cisplatin displayed the most potent antitumor effect (78% reduction in tumor volume) in comparison with cisplatin monotherapy (33%) or combined therapy of VT and cisplatin (57%; Fig. 4S). In addition, intratumoral apoptosis of tumor cells increased by >2-fold in the cisplatin + G3 combination group compared with cisplatin monotherapy (Fig. 4T and U and Supplementary Fig. S12). These results highlight G3 as a potent therapeutic option for combination chemotherapies.

VEGF-Grab3 also suppresses tumor growth, angiogenesis, and metastasis in a spontaneous breast cancer model

To determine whether G3 consistently inhibits tumor progression in other tumor models, we confirmed our findings using a spontaneous breast cancer model, MMTV-PyMT mice (36). VT and G3 treatment reduced the average size of tumor nodules by 34% and 61% compared with the control, respectively (Fig. 5B). Tumor sections stained with H&E showed that control MMTV-PyMT tumors displayed solid sheets of invasive tumor cells (Inv) with no remaining mammary gland structure. In contrast, in G3-treated tumor nodules, more early carcinoma lesions (Ea) were observed, in which the boundaries (dotted lines) between early carcinoma and surrounding adipose tissue (Adi) were well preserved (Fig. 5A). G3 reduced tumor vascular densities by 45% and 53% in peri- and intratumoral regions versus control, respectively, as compared with the 27% and 28% decreases observed in VT-treated tumors (Fig. 5C and D). Tumor cell apoptosis was also significantly increased in G3-treated tumors compared with controls or VT-treated tumors (Fig. 5E and F). The metastatic tumor cells were 64% and 52% less abundant in the axillary LNs of G3 and VT-treated mice, respectively, compared with the control, whereas we could not identify any significant differences in the lymphatic vascular density (Fig. 5G–I). These findings demonstrate that the increased antiangiogenic activity of G3 effectively suppresses tumor growth and metastasis in a breast cancer model as well.

Figure 5.

VEGF-Grab3 delays tumor growth and suppresses neovessel formation and metastasis in a spontaneous breast cancer model. A–I, female MMTV-PyMT mice (12 weeks old) received intraperitoneal injections of VT or G3 (25 mg/kg) twice per week for 3 weeks. A, tumor sections stained with H&E. Invasive tumor cells (Inv), early carcinoma lesions (Ea), and surrounding adipose tissue (Adi) are denoted by dotted lines. B, comparison of volumes of tumor nodules. Lines denote mean values. C and D, images (C) and comparison (D) of CD31+ blood vessels in the peri- and intratumoral areas. E and F, images (E) and comparison (F) of caspase-3+ apoptotic cells (red) in tumor. G, images showing cytokeratin+ tumor cell metastasis (red) in axillary LNs. H and I, quantifications of LVD (H) and cytokeratin+ tumor cell metastasis (I) in axillary LNs. Unless otherwise noted, for each group, n = 4. Values are mean ± SD. Scale bars, 100 μm. *, P < 0.05 versus control. #, P < 0.05 G3 versus VT.

Figure 5.

VEGF-Grab3 delays tumor growth and suppresses neovessel formation and metastasis in a spontaneous breast cancer model. A–I, female MMTV-PyMT mice (12 weeks old) received intraperitoneal injections of VT or G3 (25 mg/kg) twice per week for 3 weeks. A, tumor sections stained with H&E. Invasive tumor cells (Inv), early carcinoma lesions (Ea), and surrounding adipose tissue (Adi) are denoted by dotted lines. B, comparison of volumes of tumor nodules. Lines denote mean values. C and D, images (C) and comparison (D) of CD31+ blood vessels in the peri- and intratumoral areas. E and F, images (E) and comparison (F) of caspase-3+ apoptotic cells (red) in tumor. G, images showing cytokeratin+ tumor cell metastasis (red) in axillary LNs. H and I, quantifications of LVD (H) and cytokeratin+ tumor cell metastasis (I) in axillary LNs. Unless otherwise noted, for each group, n = 4. Values are mean ± SD. Scale bars, 100 μm. *, P < 0.05 versus control. #, P < 0.05 G3 versus VT.

Close modal

VEGF-Grab3 provides durable suppression of tumor angiogenesis

It has been reported that new vascular sprouts begin to regrow from remaining tumor vasculatures shortly after the cessation of antiangiogenic therapy (12). To examine the durability of G3, we treated tumor-bearing mice with either VT or G3 repeatedly at given time points (Fig. 6A, green arrows). We then withdrew treatment and analyzed the tumor vasculatures at D17 and D19 (Fig. 6A, blue arrows). Although the control tumor vessels showed a 17% increase in tumor vascular density and a 24% increase in vascular sprouts between D17 (2 days off-treatment) and D19 (4 days off-treatment), VT-withdrawn tumor vessels at D19 showed 51% increase in vascular density and a 91% increase in vascular sprouts compared with D17, indicating a vigorous regrowth of tumor vessels upon cessation of conventional anti-VEGF treatment. In contrast with VT, G3-withdrawn tumor vessels showed only 20% and 22% increase in vascular density and vascular sprouts at D19 compared with D17, which are comparable with those of the control (Fig. 6B–D). These findings demonstrate a lasting suppressive effect of G3 on tumor angiogenesis compared with VT.

Figure 6.

VEGF-Grab3 exerts more durable suppression of tumor angiogenesis. A–D, tumor vessel regrowth after VT or G3 treatment. A, experimental scheme. Mice were treated with proteins on the indicated days (green arrows). Treatment was withdrawn, and analyzed at D17 and D19 (blue arrows). B–D, changes in the vascularity after treatment with either VT or G3. White arrowheads indicate representative new vascular sprouts. Images (B) and quantifications of CD31+ blood vessels (C) and sprouts numbers (D). *, P < 0.05 VT D17 versus VT D19.

Figure 6.

VEGF-Grab3 exerts more durable suppression of tumor angiogenesis. A–D, tumor vessel regrowth after VT or G3 treatment. A, experimental scheme. Mice were treated with proteins on the indicated days (green arrows). Treatment was withdrawn, and analyzed at D17 and D19 (blue arrows). B–D, changes in the vascularity after treatment with either VT or G3. White arrowheads indicate representative new vascular sprouts. Images (B) and quantifications of CD31+ blood vessels (C) and sprouts numbers (D). *, P < 0.05 VT D17 versus VT D19.

Close modal

In this study, we developed a novel glycosylated soluble decoy receptor fusion protein containing VEGFR1 D2–D3 and Fc, called VEGF-Grab, which demonstrates a prolonged PK profile and sequesters both VEGF and PlGF. Although VEGFR1 binds to VEGFA and PlGF with higher affinity than VEGFR2, the development of therapeutic decoy proteins with the VEGFR1 backbone has proven to be difficult thus far. The major reason behind this was the high pI value VEGFR1 due to the positively charged residues at theVEGFR1 D3 region. This causes nonspecific ECM binding and poor PK profiles, leading to a shortened half-life, a subsequent decrease in efficacy, and even toxic side effects (7, 37). To overcome this intrinsic property in VEGFR1 D3, we mutated three positive residues within VEGFR1 D3 loop that were predicted to be irrelevant to ligand binding. These positive residues were altered to become potential glycosylation sites (Ser, Thr, or Asn).

Creating new decoy receptor fusion proteins using this glycosylation strategy results in several advantages. First, ECM binding of VEGF-Grab was dramatically decreased by introducing these glycosylation sites into the VEGFR1 D3 region. These sites counterbalance the positively charged residues with negatively charged residues or newly attached negative glycans. In addition, G3 contained the increased sialylation compared with G1 or VT. Terminal sialic acid is critical for in vivo half-life of proteins because the asialo-glycoprotein receptors in the liver bind to nonsialylated glycoproteins and remove them from the serum by endocytosis (38). This lower ECM binding and increased sialylation of G3 appears to facilitate an enhanced PK profile. Specifically, compared with VT, the AUCs of G3 were increased by 1.7- to 1.9-fold, suggesting that bioavailability of G3 is superior to VT. Second, G3 containing the VEGFR1 D2–D3 showed more potent decoy activity against VEGFA and PlGF, compared with VT; G3 bound 1.5-fold and 6.7-fold higher to VEGFA and PlGF as compared with VT. This was evidenced by our in vitro experiments demonstrating strong suppression of EC proliferation, migration, and tube formation after G3 treatment. Consistent results were observed in vivo, where G3 showed much stronger antiangiogenic, antitumor, and antimetastatic effects in both implanted and spontaneous tumor models compared with VT. The binding affinity of G3 to PlGF, which is critical for TAM recruitments, was comparable with that of anti-PlGF antibody (14), resulting in the decreased macrophage infiltration in G3-treated tumor compared with those treated with VT. Considering that the LLC tumor model is known to be relatively resistant to anti-VEGF therapy (28, 31), these findings suggest the possibility of overcoming resistance to anti-VEGF therapy by concurrent blockade of PlGF with VEGF-Grab. Third, G3 demonstrated improvements in toxicity profile as compared with the parental VEGFR1-Fc. Ascites formation and mortality were reported with the use of VEGFR1-Fc due to its nonspecific interaction with the ECM (7, 37). During our animal experiments, G3 was treated for long periods lasting 2 to 3 weeks, with no signs of ascites formation or mortality (data not shown). Furthermore, histologic analyses of vital organs did not reveal any significant differences in comparison with VT (Supplementary Fig. S9), implying that the toxicity of parental VEGFR1-Fc can be overcome by introducing additional glycosylation. However, adverse effects of systemic anti-VEGF therapy have been reported, including delayed wound healing and hemorrhage (39). Recently, Sticky-trap, which locally inhibits angiogenesis, has been developed and confirmed to have no systemic side effects (40). Although the potential side effects of G3 in clinics should be carefully studied at varying dosages, time points, and after long-term treatment, it could be further improved by adopting the Sticky-trap concept.

Currently, anti-VEGF agents are approved for clinical use in combination with chemotherapy for the treatment of various tumors (6). Here, we also demonstrated the potential application of VEGF-Grab as a candidate for combinational chemotherapy with its additive and synergistic effects. As VEGF-Trap (aflibercept) was also FDA approved for the treatment of age-related macular degeneration (AMD) exhibiting higher efficiency than a single approach inhibiting VEGFA (41), VEGF-Grab can be also applicable to angiogenic ocular diseases, including AMD.

In conclusion, our evidence suggests that VEGF-Grab3 is a potent and effective recombinant decoy for both VEGF and PlGF. Through the enhanced PK profile, VEGF-Grab3 showed durable suppression of tumor angiogenesis, growth, and metastasis. Clinical applicability of this novel fusion protein should be explored through further preclinical and clinical studies.

No potential conflicts of interest were disclosed.

Conception and design: J.-E. Lee, C. Kim, S.C. Kim, G.Y. Koh, H.M. Kim

Development of methodology: J.-E. Lee, C. Kim, N. Oh, S. Hua, S.C. Kim, G.M. Lee, G.Y. Koh, H.M. Kim

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.-E. Lee, C. Kim, I. Park, N. Oh, S. Hua, H.J. An, S.C. Kim, H.M. Kim

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.-E. Lee, C. Kim, N. Oh, S. Hua, H. Jeong, H.J. An, S.C. Kim, G.Y. Koh, H.M. Kim

Writing, review, and/or revision of the manuscript: J.-E. Lee, C. Kim, I. Park, S. Hua, H. Jeong, H.J. An, S.C. Kim, G.Y. Koh, H.M. Kim

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.-E. Lee, C. Kim, H. Yang, N. Oh, S. Hua, H. Jeong, S.C. Kim

Study supervision: G.M. Lee, G.Y. Koh, H.M. Kim

The authors thank Tae-Chang Yang for his technical assistance.

This work was supported by the National Research Foundation of Korea (NRF; 2012R1A1A1010456 and 2013M3A9B6075938 to H.M. Kim and NRF-2013M3A9B6046565 to G.Y. Koh), the Intelligent Synthetic Biology Center of Global Frontier Project (2011-0031955 to H.M. Kim), KAIST Future Systems Healthcare Project (to H.M. Kim) from the Ministry of Science, ICT and Future Planning, and the venture research program for graduate and PhD students, KAIST (N01130003 to J.-E. Lee).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Ferrara
N
,
Alitalo
K
. 
Clinical applications of angiogenic growth factors and their inhibitors
.
Nat Med
1999
;
5
:
1359
64
.
2.
Sung
HK
,
Michael
IP
,
Nagy
A
. 
Multifaceted role of vascular endothelial growth factor signaling in adult tissue physiology: an emerging concept with clinical implications
.
Curr Opin Hematol
2010
;
17
:
206
12
.
3.
Jung
K
,
Lee
D
,
Lim
HS
,
Lee
SI
,
Kim
YJ
,
Lee
GM
, et al
Double anti-angiogenic and anti-inflammatory protein Valpha targeting VEGF-A and TNF-alpha in retinopathy and psoriasis
.
J Biol Chem
2011
;
286
:
14410
8
.
4.
Egeblad
M
,
Nakasone
ES
,
Werb
Z
. 
Tumors as organs: complex tissues that interface with the entire organism
.
Dev Cell
2010
;
18
:
884
901
.
5.
Olsson
AK
,
Dimberg
A
,
Kreuger
J
,
Claesson-Welsh
L
. 
VEGF receptor signalling - in control of vascular function
.
Nat Rev Mol Cell Biol
2006
;
7
:
359
71
.
6.
Ferrara
N
,
Hillan
KJ
,
Gerber
HP
,
Novotny
W
. 
Discovery and development of bevacizumab, an anti-VEGF antibody for treating cancer
.
Nat Rev Drug Discov
2004
;
3
:
391
400
.
7.
Holash
J
,
Davis
S
,
Papadopoulos
N
,
Croll
SD
,
Ho
L
,
Russell
M
, et al
VEGF-Trap: a VEGF blocker with potent antitumor effects
.
Proc Natl Acad Sci U S A
2002
;
99
:
11393
8
.
8.
Shibuya
M
. 
Vascular endothelial growth factor and its receptor system: physiological functions in angiogenesis and pathological roles in various diseases
.
J Biochem
2013
;
153
:
13
9
.
9.
Sullivan
LA
,
Carbon
JG
,
Roland
CL
,
Toombs
JE
,
Nyquist-Andersen
M
,
Kavlie
A
, et al
r84, a novel therapeutic antibody against mouse and human VEGF with potent anti-tumor activity and limited toxicity induction
.
PLoS ONE
2010
;
5
:
e12031
.
10.
Bergers
G
,
Hanahan
D
. 
Modes of resistance to anti-angiogenic therapy
.
Nat Rev Cancer
2008
;
8
:
592
603
.
11.
Potente
M
,
Gerhardt
H
,
Carmeliet
P
. 
Basic and therapeutic aspects of angiogenesis
.
Cell
2011
;
146
:
873
87
.
12.
Mancuso
MR
,
Davis
R
,
Norberg
SM
,
O'Brien
S
,
Sennino
B
,
Nakahara
T
, et al
Rapid vascular regrowth in tumors after reversal of VEGF inhibition
.
J Clin Invest
2006
;
116
:
2610
21
.
13.
Davis-Smyth
T
,
Chen
H
,
Park
J
,
Presta
LG
,
Ferrara
N
. 
The second immunoglobulin-like domain of the VEGF tyrosine kinase receptor Flt-1 determines ligand binding and may initiate a signal transduction cascade
.
EMBO J
1996
;
15
:
4919
27
.
14.
Fischer
C
,
Jonckx
B
,
Mazzone
M
,
Zacchigna
S
,
Loges
S
,
Pattarini
L
, et al
Anti-PlGF inhibits growth of VEGF(R)-inhibitor-resistant tumors without affecting healthy vessels
.
Cell
2007
;
131
:
463
75
.
15.
Fischer
C
,
Mazzone
M
,
Jonckx
B
,
Carmeliet
P
. 
FLT1 and its ligands VEGFB and PlGF: drug targets for anti-angiogenic therapy?
Nat Rev Cancer
2008
;
8
:
942
56
.
16.
Adini
A
,
Kornaga
T
,
Firoozbakht
F
,
Benjamin
LE
. 
Placental growth factor is a survival factor for tumor endothelial cells and macrophages
.
Cancer Res
2002
;
62
:
2749
52
.
17.
Laurent
J
,
Hull
EF
,
Touvrey
C
,
Kuonen
F
,
Lan
Q
,
Lorusso
G
, et al
Proangiogenic factor PlGF programs CD11b(+) myelomonocytes in breast cancer during differentiation of their hematopoietic progenitors
.
Cancer Res
2011
;
71
:
3781
91
.
18.
Anisimov
A
,
Leppanen
VM
,
Tvorogov
D
,
Zarkada
G
,
Jeltsch
M
,
Holopainen
T
, et al
The basis for the distinct biological activities of vascular endothelial growth factor receptor-1 ligands
.
Sci Signal
2013
;
6
:
ra52
.
19.
Elliott
S
,
Lorenzini
T
,
Asher
S
,
Aoki
K
,
Brankow
D
,
Buck
L
, et al
Enhancement of therapeutic protein in vivo activities through glycoengineering
.
Nat Biotechnol
2003
;
21
:
414
21
.
20.
Egrie
JC
,
Browne
JK
. 
Development and characterization of novel erythropoiesis stimulating protein (NESP)
.
Nephrol Dial Transplant
2001
;
16
Suppl 3
:
3
13
.
21.
Ratner
M
. 
Genentech's glyco-engineered antibody to succeed Rituxan
.
Nat Biotechnol
2014
;
32
:
6
7
.
22.
Koh
YJ
,
Kim
HZ
,
Hwang
SI
,
Lee
JE
,
Oh
N
,
Jung
K
, et al
Double antiangiogenic protein, DAAP, targeting VEGF-A and angiopoietins in tumor angiogenesis, metastasis, and vascular leakage
.
Cancer Cell
2010
;
18
:
171
84
.
23.
Kim
JY
,
Kim
YG
,
Lee
GM
. 
CHO cells in biotechnology for production of recombinant proteins: current state and further potential
.
Appl Microbiol Biotechnol
2012
;
93
:
917
30
.
24.
Jung
K
,
Lee
JE
,
Kim
HZ
,
Kim
HM
,
Park
BS
,
Hwang
SI
, et al
Toll-like receptor 4 decoy, TOY, attenuates gram-negative bacterial sepsis
.
PLoS ONE
2009
;
4
:
e7403
.
25.
Hua
S
,
Williams
CC
,
Dimapasoc
LM
,
Ro
GS
,
Ozcan
S
,
Miyamoto
S
, et al
Isomer-specific chromatographic profiling yields highly sensitive and specific potential N-glycan biomarkers for epithelial ovarian cancer
.
J Chromatogr A
2013
;
1279
:
58
67
.
26.
Hua
S
,
Hu
CY
,
Kim
BJ
,
Totten
SM
,
Oh
MJ
,
Yun
N
, et al
Glyco-analytical multispecific proteolysis (Glyco-AMP): a simple method for detailed and quantitative Glycoproteomic characterization
.
J Proteome Res
2013
;
12
:
4414
23
.
27.
Strum
JS
,
Nwosu
CC
,
Hua
S
,
Kronewitter
SR
,
Seipert
RR
,
Bachelor
RJ
, et al
Automated assignments of N- and O-site specific glycosylation with extensive glycan heterogeneity of glycoprotein mixtures
.
Anal Chem
2013
;
85
:
5666
75
.
28.
Kim
C
,
Yang
H
,
Fukushima
Y
,
Saw
PE
,
Lee
J
,
Park
JS
, et al
Vascular RhoJ is an effective and selective target for tumor angiogenesis and vascular disruption
.
Cancer Cell
2014
;
25
:
102
17
.
29.
Yang
Z
,
Lasker
K
,
Schneidman-Duhovny
D
,
Webb
B
,
Huang
CC
,
Pettersen
EF
, et al
UCSF Chimera, MODELLER, and IMP: an integrated modeling system
.
J Struct Biol
2012
;
179
:
269
78
.
30.
Sola
RJ
,
Griebenow
K
. 
Effects of glycosylation on the stability of protein pharmaceuticals
.
J Pharm Sci
2009
;
98
:
1223
45
.
31.
Shojaei
F
,
Wu
X
,
Malik
AK
,
Zhong
C
,
Baldwin
ME
,
Schanz
S
, et al
Tumor refractoriness to anti-VEGF treatment is mediated by CD11b+Gr1+ myeloid cells
.
Nat Biotechnol
2007
;
25
:
911
20
.
32.
Murdoch
C
,
Muthana
M
,
Coffelt
SB
,
Lewis
CE
. 
The role of myeloid cells in the promotion of tumour angiogenesis
.
Nat Rev Cancer
2008
;
8
:
618
31
.
33.
Shojaei
F
,
Wu
X
,
Zhong
C
,
Yu
L
,
Liang
XH
,
Yao
J
, et al
Bv8 regulates myeloid-cell-dependent tumour angiogenesis
.
Nature
2007
;
450
:
825
31
.
34.
Qian
BZ
,
Li
J
,
Zhang
H
,
Kitamura
T
,
Zhang
J
,
Campion
LR
, et al
CCL2 recruits inflammatory monocytes to facilitate breast-tumour metastasis
.
Nature
2011
;
475
:
222
5
.
35.
Kabbinavar
F
,
Hurwitz
HI
,
Fehrenbacher
L
,
Meropol
NJ
,
Novotny
WF
,
Lieberman
G
, et al
Phase II, randomized trial comparing bevacizumab plus fluorouracil (FU)/leucovorin (LV) with FU/LV alone in patients with metastatic colorectal cancer
.
J Clin Oncol
2003
;
21
:
60
5
.
36.
Lin
EY
,
Jones
JG
,
Li
P
,
Zhu
L
,
Whitney
KD
,
Muller
WJ
, et al
Progression to malignancy in the polyoma middle T oncoprotein mouse breast cancer model provides a reliable model for human diseases
.
Am J Pathol
2003
;
163
:
2113
26
.
37.
Kuo
CJ
,
Farnebo
F
,
Yu
EY
,
Christofferson
R
,
Swearingen
RA
,
Carter
R
, et al
Comparative evaluation of the antitumor activity of antiangiogenic proteins delivered by gene transfer
.
Proc Natl Acad Sci U S A
2001
;
98
:
4605
10
.
38.
Bork
K
,
Horstkorte
R
,
Weidemann
W
. 
Increasing the sialylation of therapeutic glycoproteins: the potential of the sialic acid biosynthetic pathway
.
J Pharm Sci
2009
;
98
:
3499
508
.
39.
des Guetz
G
,
Uzzan
B
,
Chouahnia
K
,
Morere
JF
. 
Cardiovascular toxicity of anti-angiogenic drugs
.
Target Oncol
2011
;
6
:
197
202
.
40.
Michael
IP
,
Westenskow
PD
,
Hacibekiroglu
S
,
Greenwald
AC
,
Ballios
BG
,
Kurihara
T
, et al
Local acting Sticky-trap inhibits vascular endothelial growth factor dependent pathological angiogenesis in the eye
.
EMBO Mol Med
2014
;
6
:
604
23
.
41.
Stewart
MW
. 
Aflibercept (VEGF Trap-eye): the newest anti-VEGF drug
.
Br J Ophthalmol
2012
;
96
:
1157
8
.