Recombinant Human VEGF₁₆₅b Inhibits Experimental Choroidal Neovascularization

Abstract

Purpose.: Vascular endothelial growth factor (VEGF-A) is the principal stimulator of angiogenesis in wet age-related macular degeneration (AMD). However, VEGF-A is generated by alternate splicing into two families, the proangiogenic VEGF-A_xxx family and the antiangiogenic VEGF-A_xxxb family. It is the proangiogenic family that is responsible for the blood vessel growth seen in AMD.

Methods.: To determine the role of antiangiogenic isoforms of VEGF-A as inhibitors of choroidal neovascularization, the authors used a model of laser-induced choroidal neovascularization in the mouse eye and investigated VEGF-A₁₆₅b effects on endothelial cells and VEGFRs in vitro.

Results.: VEGF-A₁₆₅b inhibited VEGF-A₁₆₅–mediated endothelial cell migration with a dose effect similar to that of ranibizumab and bevacizumab and 200-fold more potent than that of pegaptanib. VEGF-A₁₆₅b bound both VEGFR1 and VEGFR2 with affinity similar to that of VEGF-A₁₆₅. After laser injury, mice were injected either intraocularly or subcutaneously with recombinant human VEGF-A₁₆₅b. Intraocular injection of rhVEGF-A₁₆₅b gave a pronounced dose-dependent inhibition of fluorescein leakage, with an IC₅₀ of 16 pg/eye, neovascularization (IC₅₀, 0.8 pg/eye), and lesion as assessed by histologic staining (IC₅₀, 8 pg/eye). Subcutaneous administration of 100 μg twice a week also inhibited fluorescein leakage and neovascularization and reduced lesion size.

Conclusions.: These results show that VEGF-A₁₆₅b is a potent antiangiogenic agent in a mouse model of age-related macular degeneration and suggest that increasing the ratio of antiangiogenic-to-proangiogenic isoforms may be therapeutically effective in this condition.

Ocular neovascularization (ONV) is the leading cause of blindness in the western world. In age-related macular degeneration (AMD), choroidal neovascularization (CNV) affects 10% of all patients and is the most severe and rapidly progressing form of the disease. In “wet” or neovascular AMD, there is rapid, devastating loss of central vision with rarely any symptoms until the later stages. Examination of pathologic specimens has demonstrated that there is an abnormal proliferation of choroidal vessels beneath the retina, which subsequently bleed, resulting in fibrosis and macular scarring. VEGF-A has been shown to be substantially upregulated in AMD, ¹ particularly in retinal pigment epithelial cells and fibroblasts. ² ONV seen in AMD is associated with increased levels of VEGF. Angiogenesis is a complex process mediated by factors from the VEGF, angiopoietin, and ephrin families. ^3,4 It is essential in normal physiology, such as in wound healing, endometrial maturation, embryogenesis, and fat deposition. However, it also underlies many disease states in addition to AMD, including retinal and other complications of diabetes, cancer, atherosclerosis, ⁵ rheumatoid arthritis, and psoriasis. ^6,7 VEGF-A is the dominant proangiogenic factor in AMD, ^8,9 stimulating endothelial cell proliferation and migration and increased microvascular permeability by activation of its cognate receptors VEGFR1 (flt-1) and VEGFR2 (KDR/flk1). ¹⁰ Anti-VEGF therapies that influence new vessel formation and that were shown to be effective in animal models ^11,12 have successfully completed clinical trials in AMD, and three agents have been widely used: pegaptanib (Macugen; Pfizer, New York, NY), ^13,14 an RNA aptamer that targets the heparin-binding domain of VEGF-A, ranibizumab (Lucentis; Genentech, South San Francisco, CA), an antibody fragment to the VEGFR binding domain of VEGF-A, and bevacizumab (Avastin; Genentech), the full-length antibody equivalent to ranibizumab. Other members of the VEGF family of proteins, (e.g., VEGF-C, VEGF-D) are formed from different gene products and are structurally distinct from VEGF-A (see Ref. 15 for review). Human VEGF-A is differentially spliced from 8 exons to form a variety of different mRNAs encoding at least 14 different proteins in two families, the proangiogenic VEGF-A_xxx family and the antiangiogenic VEGF-A_xxxb family, where xxx denotes the number of amino acids of the secreted isoform, VEGF-A₁₂₁, VEGF-A₁₆₅, (the dominant proangiogenic isoform) VEGF-A₁₆₅b, and others. ^15,16 VEGF-A_xxxb isoforms are formed by alternate splice acceptor site selection in exon 8, forming an mRNA containing 18 bases coded by exon 8b in place of the 18 bases of exon 8a. ¹⁷ This alternative splicing produces proteins of the same length as in the VEGF-A_xxx family but with a different C-terminal amino acid sequence. ¹⁸ Exons 8a and 8b both code for six amino acids, exon 8a for CDKPRR and exon 8b for SLTRKD. Therefore, exon 8b lacks the Cys residue that forms the final disulfide bond ¹⁹ and the terminal two charged Arg residues postulated to be involved with receptor signaling. ^20,21 Instead, exon 8b codes for Ser instead of Cys and a less basic C-terminal than exon 8a. The receptor binding domains are still present in VEGF-A₁₆₅b, which acts as a competitive inhibitor of VEGF-A₁₆₅ (i.e., it binds the receptors but does not stimulate angiogenesis signaling). VEGF-A₁₆₅b inhibits the proliferative, migratory, and vasodilator effects of VEGF-A₁₆₅. ¹⁷ VEGF-A₁₆₅b is antiangiogenic in the rabbit cornea, chick chorioallantoic membrane, mouse skin, ²² lactating mammary gland, ²³ and rat mesentery, and it inhibits tumor growth in xenotransplanted tumors in mice. ^24–26 Unlike angiogenic VEGF isoforms, the antiangiogenic isoform VEGF-A₁₆₅b is downregulated in renal and colorectal carcinoma and malignant prostate tissue, ^17,25,27 metastatic melanoma, ²⁸ diabetic retinopathy, ¹⁸ and Denys-Drash syndrome. ²⁹ We have also identified VEGF-A₁₆₅b protein expression in many other human tissues, including the eye. ¹⁸ To determine whether the antiangiogenic activity of VEGF-A₁₆₅b was sufficient to inhibit CNV, we used a laser-induced photocoagulation model of CNV in the mouse. We show here that VEGF-A₁₆₅b dose dependently inhibited CNV as assessed by fluorescein angiography (FA), lectin staining, and lesion size. 

Materials and Methods

Protein Extraction

Protein was extracted from human eyes obtained from the Bristol Eye Bank (Bristol, UK) with local ethics approval. The tissues were dissected manually and were treated with lysis buffer (50 mM Tris-HCl pH 8, 150 mM NaCl, 1 mM EDTA pH 8, 100 mM NaF, 1.5 mM MgCl₂, 10% vol/vol glycerol, 1% vol/vol Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 1 mM Na₃VaO₄, and 1 μg/mL of each of the protease inhibitors aprotinin, leupeptin, and pepstatin), and the solution was mechanically homogenized. Lysates were then centrifuged again for 9 minutes at 4°C, 13000 rpm, and supernatants were used for further analysis. Protein was quantified with a Bio-Rad (Hercules, CA) protein assay. Recombinant VEGF-A₁₆₅ (100 ng/well; R&D Systems, Minneapolis, MN) and VEGF-A₁₆₅b (50 ng/well, as described in Rennel et al. ³⁰ ) reconstituted in phosphate-buffered saline (PBS) were used as controls on each gel. Samples were adjusted to 150 μg total protein in PBS and were loaded in loading buffer (100 mM Tris HCl pH 6.8, 4% SDS, 20% glycerol [BDH, Poole, UK], 0.2% [wt/vol] bromophenol blue, 5% final concentration β-mercaptoethanol). Samples and recombinant human VEGF (denatured lanes) were heated for 5 minutes at 100°C. For native recombinant human VEGF, the β-mercaptoethanol and heating steps were omitted. Proteins were loaded onto a 10% polyacrylamide gel and run in 25 mM Tris, 250 mM glycine (BDH), 0.1% SDS. The gel was then trimmed, equilibrated, and transferred in 50 mM Tris, 38 mM glycine (BDH), 20% methanol, 0.1% SDS onto a polyvinylidene (PVDF) membrane (Fisher Scientific, Leicestershire, UK) at 60 V for 90 minutes. The PVDF membrane was incubated in 10% low-fat powdered milk/0.05% TBS-Tween blocking solution for 1 hour at room temperature, washed with 0.05% TBS-Tween, incubated with primary antibody, and diluted in 5% low-fat powdered milk/0.05% TBS-Tween overnight at 4°C. For detection of all VEGF-A isoforms, 1.25 μg/mL purified mouse monoclonal IgG2B (MAB293; R&D Systems) was used. For detection of VEGF-A_xxxb isoforms, mouse monoclonal antibody, raised to the nine carboxyl-terminal amino acids of VEGF-A₁₆₅b, was used (MAB3045; R&D Systems). After further washing the membrane was incubated for 1 hour in 1.4 ng/mL horseradish peroxidase (HRP)-goat anti–mouse IgG (Pierce, Cheshire, UK) diluted in 5% low-fat powdered milk/TBS-Tween. The membrane was washed again and developed using an enhanced chemiluminescence (ECL) kit (SuperSignal West Femto Maximum Sensitivity Substrate; Pierce, Cheshire, UK). 

Cell Migration Assay

Endothelial cells for migration were used in passages 3 to 6 at 70% to 80% confluence. Human microvascular endothelial cells (HMVECs) were serum starved in endothelial basal media without FBS and supplements (EBM) for 8 to 10 hours. Cells were trypsinized and resuspended in 0.1% vol/vol FBS in EBM, and 150,000 cells in 500 μL medium were seeded on attachment factor (Cascade Biologics, Portland, OR)–coated filter inserts (8 μm, 12 mm; Millipore, Billerica, MA) with the treatment in the bottom well. Each treatment was performed in triplicate. The cells were incubated at 37°C and allowed to migrate overnight. The inserts were washed with PBS, and cells were fixed with 4% PFA/PBS, pH 7.4, for 10 minutes. Nonmigrated cells were removed from the membranes, and the nuclei of migrated cells were stained with Hoechst 33258 (5 μg/mL in 0.5% Triton/PBS). Membranes were cut out of the wells and mounted on microscope slides with mounting medium (Vectashield; Vector Laboratories, Burlingame, CA). Migrated cells were counted in 10 fields per membrane under a fluorescence microscope (DM; Leica, Wetzlar, Germany; 40× objective). Change in migration was expressed relative to the basal migration rate toward zero chemoattractant and was plotted as average ± SEM. The inhibitory effect on migration of VEGF inhibitors over VEGF-A₁₆₅ was determined by increasing concentrations of inhibitor with 1 nM (40 ng/mL) VEGF-A₁₆₅. IC₅₀ was calculated from the normalized data using a variable slope sigmoidal fit (Prism 4; GraphPad, San Diego, CA). 

Surface Plasmon Resonance

To compare the binding affinities of VEGF-A₁₆₅ and VEGF-A₁₆₅b with those of VEGF receptors, Fc-VEGFR1 or Fc-VEGFR2 was amine coupled to a sensor chip (CM5; Biacore AB, Uppsala, Sweden) to an immobilization level of 630 response units (RU) VEGFR1 and 580 RU (VEGFR2). Coupling was performed using EDC/NHS and 1 mol/L ethanolamine (Biacore AB) in accordance with the manufacturer's instructions, with the VEGFR dissolved in 10 mmol/L sodium acetate (pH 4.5). A blank reference cell was formed by the same activation and deactivation process involved in amine coupling without adding antibody. Samples containing VEGF-A₁₆₅ or VEGF-A₁₆₅b diluted in HBS-EP sample buffer (HEPES-buffered saline with EDTA and P20 surfactant; Biacore AB) were then run at twofold serial dilutions in random order in duplicate. Injection was performed at 30 μL/min for 3 minutes, followed by 6 minutes of buffer only, for monitoring of dissociation. Regeneration between each interaction was performed by injection of 4 mol/L MgCl₂ at 20 μL/min for 40 seconds, followed by a 2-minute period of stabilization before the next injection. 

Induction of Choroidal Neovascularization

Thirty 6- to 8-week-old C56Bl/6J male mice (National Institutes of Health, Bethesda, MD) were used in the study. All procedures were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and approved by the University of Southern California Animal Use Committee. The mice were anesthetized for all procedures with intraperitoneal injection of 0.1 mL of a 50:50 mixture of ketamine hydrochloride (20 mg/mL) and xylazine hydrochloride (100 mg/mL; Phoenix Pharmaceutical Inc., St. Joseph, MO). The pupils were dilated with 5% phenylephrine hydrochloride and 0.8% tropicamide. Four photocoagulation lesions were delivered with a diode green laser (150 mW, 0.05 s, 75 μm; 532 nm Oculight GLx Diode laser; Iridex, Mountain View, CA) between the retinal vessels in a peripapillary distribution at a distance of 1 to 2 disc diameters in each eye. Only laser lesions with a subretinal bubble at the time of treatment were included in the study. Two microliters of saline or rhVEGF-A₁₆₅b at various concentrations (0.01, 0.1, 1, and 10 ng in each eye) was injected into the vitreous immediately after photocoagulation using a 35-gauge needle (NanoFil; World Precession Instruments, Sarasota, FL) or 100 μg rhVEGF-A₁₆₅b injected subcutaneously twice weekly for 2 weeks. Saline was used because it has previously been shown that proteins such as albumin have no effect on CNV. ³¹ The injections were repeated 7 days later. Fourteen days after laser treatment, FA of the fundi was carried out, and the animals were killed after imaging. Eyes were enucleated and fixed in 4% PFA for further histology processes. 

Fluorescein Angiography

Fluorescein angiography was carried out 14 days after photocoagulation. Pupils were dilated using 2.5% phenylephrine hydrochloride and 1% tropicamide eye drops, and 10% fluorescein sodium chloride was injected intraperitoneally into the anesthetized animals. After 180 to 200 seconds, fundus photographs were taken using an angiography camera (VK2e, KD-2UC; Kowa, Nagoya, Japan). Two experienced observers who were blinded to the study processed and scored the images. Scores were given as previously described (0, no staining; 1, slight leakage; 2, moderate leakage; and 3, heavy leakage). Images were standardized for the process. 

Whole Mount Staining

Eyes were fixed overnight in 4% paraformaldehyde and washed in PBS overnight at 4°C. After the sclera-choroid-complex was dissected and permeabilized in 1% Triton X-100 for 2 hours, biotinylated isolectin B4 (specific endothelial cell marker, 1:100; Vector Laboratories) was added and incubated at 4°C overnight. Samples were washed and incubated with streptavidin-conjugated Alexa Fluor 594 overnight. The samples were mounted (Vectashield; Vector Laboratories) after a final wash. 

Measurement of Lesion Volume and Surface Area

A laser scanning confocal microscope (LSM; model 510; Carl Zeiss Meditec, Dublin, CA) was used to image the CNV lesions. Fluorescence volume measurements were accomplished by creating image stacks of optical slices within lesions. The image stacks were generated in the z-plane by 2 μm optical sectioning, with the confocal microscope set to excite at 594 nm. Images were further processed using the LSM software, by closely circumscribing and digitally extracting the fluorescent lesion areas throughout the entire image stack. The extracted lesion was processed through the LSM topography software to generate a digital topographic image representation of the lesion and an implicit image volume. Units for implicit volume were designated as volume (in cubic micrometers). 

Cross-Section and HE-Staining

The cornea and lens were removed immediately after the animals were killed and the eye enucleated. Dissected eyecups were embedded in OCT compound (Tissue-Tek, OCT 4583; Sakura Finetek Inc., Torrance, CA) and frozen in liquid nitrogen. Air-dried and acetone-fixed 8-μm serial cross-sections of the whole specimen were stained with hematoxylin and eosin (H&E staining). The maximum cross-section of each lesion was observed and photographed using a light microscope with a digital camera (DM3000; Leica). The area of each cross-section was measured in ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html). 

Statistical Analysis

Binding curves and dose-response curves were fitted by normalization to PBS followed by nonlinear regression analysis modeled with a sigmoid curve with variable slope (Prism 4.0; GraphPad). Results are given as means and SEM. All parameters except H&E staining for the PBS treated mice were normally distributed according to the KS Normality test. Thus parametric statistics (ANOVA followed by Student's Newman-Keuls or unpaired t-test for subcutaneous VEGF-A₁₆₅b vs. PBS) were used for most parameters, but for the H&E-stained subcutaneous injections, nonparametric Mann-Whitney U test was used. 

Results

VEGF-A₁₆₅b—the Predominant VEGF-A_xxxb Isoform in the Eye

We have previously shown that VEGF-A_xxxb isoforms are expressed as a protein in the retina, iris, and vitreous of the eye. To determine whether VEGF-A₁₆₅b expression was localized to the RPE-choroid or neuronal layers of the retina, the neuronal retina was isolated from the underlying tissue by dissection of the Bruch's membrane or the RPE-choroid was separated from the overlying tissue. Figure 1 (left-hand side) shows an immunoblot of a reducing SDS-PAGE gel probed for all VEGF-A isoforms. Recombinant VEGF-A₁₆₅b and VEGF-A₁₆₅, both of which have forms at 19 kDa (unglycosylated) and 21 kDa (partially glycosylated), are detected by this antibody. A strong band at 23 and 46 kDa is consistent with expression of the 165-amino acid isoforms of VEGF-A, VEGF-A₁₆₅, and VEGF-A₁₆₅b. Weaker bands are present at 34 kDa, consistent with the 121-amino acid isoforms (VEGF-A₁₂₁ and VEGF-A₁₂₁b) as dimers. When a parallel blot was probed with an antibody specific for the VEGF-A_xxxb isoforms (Fig. 1, right-hand side), a 46-kDa and a 23-kDa protein was detected, showing that VEGF-A₁₆₅b is the predominant VEGF-A_xxxb isoform and is expressed in the neuronal layer of the retina and in the whole retina, iris, and vitreous. Interestingly, when protein from the RPE-choroid layer was included, there was significant expression in that layer as well, indicating that VEGF₁₆₅b is expressed in both the neuronal and the RPE-choroid layers of the retina. 

Figure 1.

 

VEGF-A₁₆₅b is the predominant VEGF-A_xxxb isoform in the eye. Protein was extracted from dissected tissues of donor eyes from the Bristol Eye Bank and subjected to SDS-PAGE and immunoblotting using anti-VEGF (left) or anti-VEGF_xxxb (right). Protein was extracted from eyes either by dissection of the RPE layer from the rest of the retina or by dissection of the neuronal layer from the rest of the retina or the whole retina. Recombinant human (rh) VEGF₁₆₅ and VEGF₁₆₅b were run both as denatured and native to demonstrate monomer and dimers. The expected sizes of VEGF₁₂₁, VEGF₁₈₉, and VEGF₁₄₅ are demonstrated.

VEGF-A₁₆₅b Inhibition of Endothelial Cell Migration—Similar Efficacy to Bevacizumab

To determine the relative potency of VEGF-A₁₆₅b as an inhibitor of endothelial cell migration, critical for angiogenesis, we compared the inhibitory effect of VEGF-A₁₆₅b with ranibizumab, bevacizumab, and pegaptanib. Figure 2A shows that VEGF-A₁₆₅b inhibits endothelial cell migration induced by 1 nM VEGF-A₁₆₅ (IC₅₀, 0.28 nM) at the same efficacy on a mole-for-mole basis as ranibizumab (IC₅₀, 0.24 nM) and bevacizumab (IC₅₀, 0.21 nM; P > 0.1 compared with ranibizumab and VEGF-A₁₆₅b), all of which were 200 times more potent than pegaptanib (IC₅₀, 4.08 nM; P < 0.001 compared with ranibizumab, VEGF-A₁₆₅b, and bevacizumab). To determine the binding affinity of VEGF-A₁₆₅b to its receptors, we used biotinylated VEGF-A₁₆₅b to carry out a binding-affinity experiment. Increasing doses of biotinylated-VEGF-A₁₆₅b were added to an ELISA plate coated with Fc-VEGFR1. Figure 2B shows that this resulted in an affinity of 3 × 10⁻¹¹ M. To confirm this we carried out surface plasmon resonance (SPR) on a chip coated with Fc-VEGFR1. Figure 2C shows that this resulted in a binding curve with a similar affinity of 1.9 × 10⁻¹¹ M. This compares well with previously published figures of 2.2 × 10⁻¹¹ M for SPR using VEGF-A₁₆₅. ³² We therefore used the same technique to measure binding of VEGF-A₁₆₅b to VEGFR2. Figure 2D shows that the affinity of VEGF-A₁₆₅b to VEGFR2, as measured by SPR, was 1.35 × 10⁻¹⁰ M, similar to the published values for VEGF-A₁₆₅ (2.1 × 10⁻¹⁰ M). ³³  

Figure 2.

 

VEGF-A₁₆₅b inhibits VEGF-A₁₆₅–mediated endothelial cell migration but binds to VEGFR. (A) HMVECs were seeded onto polycarbonate 8-μm pore filters, and migration across the transwell was measured to 1 nM VEGF-A₁₆₅ with increasing doses of inhibitors. (B) Fc-VEGFR1 was coated onto an ELISA plate, and increasing concentrations of biotinylated VEGF-A₁₆₅b were added. After development with HRP-SA, the percentage of bound VEGF-A₁₆₅b was calculated relative to saturated values, and IC₅₀ was calculated using a nonlinear sigmoid variable slope fit. (C) Fc-VEGFR1 was coupled to an SPR sensor chip, and increasing concentrations of VEGF-A₁₆₅b were allowed to flow over the chip. Response units relative to the reference chip were measured by surface plasmon resonance. (D) Fc-VEGFR2 was coupled to a similar chip, and the experiment was repeated. Affinity and dissociation constants were calculated by 1:1 Langmuir binding models.

Intravitreal VEGF-A₁₆₅b Inhibition of Choroidal Neovascularization

Mice were subjected to diode laser–induced photocoagulation and were injected with VEGF-A₁₆₅b on the day of laser procedure and 7 days later. We have previously shown that the intraocular terminal half-life of VEGF-A₁₆₅b is 62.5 hours. Fourteen days after laser injury, mice were subjected to FA. Figure 3A shows that the laser-induced photocoagulation produced lesions in the retina visible under FA when mice were injected with saline. However, these lesions were progressively smaller with increasing doses of VEGF-A₁₆₅b. Semiquantitative scoring of lesion size by an observer masked to treatment showed that VEGF-A₁₆₅b induced a dose-dependent decrease in lesion size, with an IC₅₀ of 18 pg/eye (log IC₅₀, −10.2 ± 0.21; Fig. 3B). To determine whether the increase in FA lesion size was the result of endothelial growth into the lesion, retinas were prepared as whole mounts and stained for endothelial cells by lectin staining. Two-dimensional epifluorescence microscopy was used to determine lesion area. Figure 3C shows a significant dose-dependent reduction in lesion area, quantified by tracing around the lesion and measurment of the area. The IC₅₀ for this was 0.6 pg/eye (log IC₅₀, −12.2 ± 2.3; Fig. 3D). To determine whether VEGF-A₁₆₅b reduced the effect of laser coagulation on the structure of the retina, the eyes of mice that underwent photocoagulation followed by VEGF-A₁₆₅b or saline injection were enucleated, fixed, embedded in paraffin, and prepared for histology of 5-μm-thick sections stained with H&E. Figure 3E shows that the lesion size was dose dependently reduced by VEGF-A₁₆₅b injection. To quantify this, the lesion was traced around and the area was measured. VEGF-A₁₆₅b induced a significant dose-dependent reduction in lesion cross-sectional area with an IC₅₀ of 8.4 pg (log IC₅₀, −11.08 ± 0.21; Fig. 3F). There was no significant difference in IC₅₀ or Hill slope between the different measurements (F test). 

Figure 3.

 

Intravitreal VEGF-A₁₆₅b inhibits CNV. Mice underwent retinal laser coagulation in both eyes and intravitreous injection with VEGF-A₁₆₅b or HBSS directly after laser procedure (day 0) and on day 7. (A) FA on day 14 showed dose-dependent inhibition of leakage (arrows). (B) Quantification using 0 to 3 scoring scheme by three independent masked observers. (C) Staining with isolectin B4 of the RPE-choroid-sclera complex showed a dose-dependent decrease in lesion size. (D) The IC₅₀ for lesion size was 0.78 pg. (E) H&E staining of the middle section of each photocoagulation area showed disrupted RPE layer and lesions. These were dose dependently reduced by VEGF-A₁₆₅b injection. (F) Measurement of the lesion size.

Subcutaneous VEGF-A₁₆₅b Reduces Lesion Size

We have previously shown that systemic administration of VEGF-A₁₆₅b reduces tumor growth in nude mice by inhibiting VEGF-A₁₆₅–mediated angiogenesis. ²⁶ To determine whether VEGF-A₁₆₅b could inhibit CNV when given systemically, mice were subjected to laser photocoagulation and treated twice weekly with 100 μg VEGF-A₁₆₅b subcutaneously. Figure 4A shows that this resulted in a reduction of fluorescein leakage as assessed by FA. Retinas were stained for endothelial cells, and, again, systemic VEGF-A₁₆₅b administration resulted in a significant reduction in lesion size (from 0.43 ± 0.03 mm² to 0.31 ± 0.27 mm²; P < 0.02, unpaired t-test; Fig. 4B). To determine whether VEGF-A₁₆₅b also reduced the volume of the endothelial lesion, a stack of images was generated through the lesion using 2-μm optical slicing, and the area of each slice of the stack was calculated. This allowed calculation of the volume of the lesion. Figure 4C shows the mean volumes of up to four lesions for each of the 10 mice in each group. It can be seen that subcutaneous treatment with VEGF-A₁₆₅b significantly reduced lesion volume. To determine the histologic effect of VEGF-A₁₆₅b, retinas were cut and sectioned for H&E staining. Figure 4D shows that subcutaneous injection also resulted in reduced lesion size by cross-sectional area. 

Figure 4.

 

Systemic VEGF-A₁₆₅b inhibits CNV. Mice underwent retinal laser coagulation in both eyes and subcutaneous injection with 100 μg VEGF-A₁₆₅b or PBS twice weekly. (A) Quantification of FA on day 14 using 0 to 3 scoring scheme by three independent masked observers showed inhibition of leakage. Representative images for each treatment are shown above the bars. (B) Staining with isolectin B4 of the RPE-choroid-sclera complex. (C) Confocal three-dimensional reconstruction was used to calculate volume size. (D) H&E staining of the middle section of each photocoagulation area showed disrupted RPE layer and lesions. These were reduced by VEGF-A₁₆₅b injection. **P < 0.02 and ***P < 0.001 compared with PBS. Unpaired t-test (A–C). Mann-Whitney U test (D).

Discussion

We show here that of the antiangiogenic isoforms, VEGF₁₆₅b appears to be the most abundant endogenous isoform in the eye. Interestingly, it is most abundant in the RPE layer and the retinal layer of normal eyes, but its expression in patients with AMD is not yet known, and this is a key area that must be addressed. Our assumption of the predominance of VEGF₁₆₅b among the antiangiogenic isoforms does depend on the affinity of the antibodies to VEGF_xxxb being the same for all the isoforms. Although this has been demonstrated for VEGF₁₂₁b, it is not yet clear for VEGF₁₈₉b, VEGF₁₄₅b, or VEGF₁₈₃b, and the conclusion that VEGF₁₆₅b is the most highly expressed must bear that assumption in mind. However, the lack of strong expression of VEGF proteins consistent with those sizes in the pan-VEGF Western blot supports that conclusion. The lack of tools for mouse studies on VEGF₁₆₅b has so far precluded investigation of the expression of VEGF₁₆₅b in mouse models of CNV. Antiangiogenic therapy has become a cornerstone of therapy for CNV resulting from AMD. Results of the MARINA ³⁴ and ANCHOR ³⁵ trials have shown clearly that VEGF antagonism reduces or reverses vision loss in wet AMD. Analysis of the ANCHOR trial shows that 30% of patients experienced improved vision (reversal of vision loss) in response to ranibizumab, ³⁵ approximately 30% experienced less vision loss than if they were untreated, and the remainder lost vision similar to control. Thus, though 95% of patients met the trial end point in the treated group compared with 65% in the control group, anti-VEGF therapy by antibodies that target all isoforms of VEGF-A is still not a universally effective treatment for AMD. One possible reason for this may be that VEGF comes in many different isoforms, including the antiangiogenic VEGF-A₁₆₅b isoforms. Anti-VEGF antibodies have been shown also to affect VEGF-A₁₆₅b. For instance, tumors expressing VEGF-A₁₆₅b, while growing more slowly than tumors not expressing VEGF-A₁₆₅b, are resistant to bevacizumab, ²⁵ suggesting that antiangiogenic therapy that does not target VEGF-A₁₆₅b may be more effective. Here we show that an alternative approach to anti-VEGF antibodies is also effective in animal models of AMD. Treatment of mice with the antiangiogenic isoforms of VEGF by injection of recombinant protein is an effective treatment for CNV in animal models. The dose range of VEGF-A₁₆₅b used (0.1–10 ng/eye) translates to a concentration in the eye of 0.01 to 1 μg/mL, or an effective dose in humans of 3 μg/eye, 1000- to 10-fold lower than ranibizumab or bevacizumab. ³⁵ Ranibizumab and bevacizumab are both human specific, so it is not possible to compare directly the relative efficacies in mice. However, VEGF-A₁₆₅b did inhibit CNV at a dose 20,000 times lower than that required by pegaptanib in rodents. ³⁶ The terminal half-life in the eye of VEGF-A₁₆₅b is similar to that of ranibizumab, ³⁷ so it can be postulated that effective dosing in humans may be monthly or longer. 

Intravitreal anti-VEGF treatment in the eye comes with relatively few side effects, ³⁸ at least in the short to intermediate term. This low side effect profile is in contrast with the systemic dosing of anti-VEGF agents in which side effects include fatal gastric perforation, ³⁹ increased risk for stroke, hypertension, and renal dysfunction and failure. ⁴⁰ Moreover, anti-VEGF agents and other antiangiogenic agents affect a relatively small proportion of patients when given systemically with chemotherapy for cancer. They do not appear to work by themselves, ³⁹ and recent studies have shown that bevacizumab does not work in earlier stages of colon cancer even with chemotherapy. ⁴¹ Moreover, resistance to such therapy is increasingly common ^42,43 and may depend on the level of VEGF₁₆₅b present in those tissues. The levels of VEGF₁₆₅b appear to vary widely from cell type to cell type ⁴⁴ and can be highly regulated. ⁴⁵ It is not widely accepted that systemic administration of anti-VEGF agents can be given for AMD. In contrast, however, VEGF-A₁₆₅b treatment of mice does not result in hypertension, proteinuria, or any other known side effects to date. ³⁰ Although this does not necessarily translate to a lack of side effects in humans, it is possible that the lack of cardiovascular effects of VEGF₁₆₅b, combined with its demonstrated cytoprotective properties in renal, ⁴⁶ gastrointestinal, ²⁵ and ocular epithelial cells, ⁴⁷ will result in VEGF-A₁₆₅b treatment having a lower side effect profile than systemic treatment with anti-VEGF agents. The finding here that VEGF-A₁₆₅b was effective when given systemically bodes well for its possible use in humans. It is perhaps surprising that a molecule with a relatively short half-life (tracer plasma clearance, ∼12 minutes) should be effective when given biweekly, but previous studies have shown that though the tracer plasma half-life is short, VEGF-A₁₆₅b is rapidly taken up into the areas of growing blood vessels, particularly tumors. ³⁰ VEGFR2 is upregulated during choroidal angiogenesis; therefore, it is likely that the effect after systemic administration is due to the uptake of VEGF-A₁₆₅b by VEGFR2 in the choroidal vasculature, although this has not been shown directly. 

In summary, we show here that VEGF-A₁₆₅b, which is equally as potent in blocking endothelial cell migration in vitro as ranibizumab and bevacizumab, potently inhibits CNV in vivo when given intraocularly. This inhibition can also be brought about by systemic administration of VEGF-A₁₆₅b. 

Footnotes

 Supported by Wellcome Trust Grants 79736 (JH) and 69029 (YQ), Fight for Sight (ER), British Heart Foundation Grant BS/06/005 (DOB), Cancer Research UK Grant C18064/A5730 (AHV), the National Eye Research Centre, the Richard Bright VEGF Research Trust and the North Bristol NHS Trust Adrian Wright Bequest into Disability in the Elderly (AM), PhiloGene Inc. (SD), National Institutes of Health Core Grant EY03040, and the Arnold and Mabel Beckman Foundation.

Footnotes

 Disclosure: J. Hua, PhiloGene Inc. (R); C. Spee, None; S. Kase, None; E.S. Rennel, None; A.L. Magnussen, None; Y. Qiu, None; A. Varey, None; S. Dhayade, None; A.J. Churchill, None; S.J. Harper, PhiloGene Inc. (F, C), P; D.O. Bates, PhiloGene Inc. (F, C, R), P; D.R. Hinton, None