Vascular endothelial growth factor-B gene transfer exacerbates retinal and choroidal neovascularization and vasopermeability without promoting inflammation

Production, purification, and characterization of recombinant adeno-associated virus vectors

Three recombinant AAV vectors expressing green fluorescent protein (GFP) and the 167 and 186 amino acid isoforms of mouse VEGF-B (AAV-GFP, AAV-VEGF-B167, and AAV-VEGF-B186, respectively) were prepared by the AAV Vector Unit at ICGEB Trieste. All the vectors were pseudotyped following a cross-packaging approach whereby the AAV type 2 vector genome was packaged into AAV capsid serotype 9 [59,60]. Methods for production and purification were previously described [61,62]. Briefly, viral vector stocks were obtained in HEK293 cells by high-scale calcium phosphate co-precipitation of the AAV backbone plasmid along with helper plasmids, expressing the AAV genes coding for the replicative and capsid proteins and the adenovirus helper functions supporting AAV replication. Viral vector particles were then purified from crude cell lysates by cesium chloride gradient ultracentrifugation followed by dialysis of pooled fractions. Among the different serotypes that were used for retinal gene delivery, AAV serotype 9 (AAV9) showed robust and widespread transduction in the outer retina, inner retina, retinal ganglion cell (RGC) layers [63,64], and synaptic layers [65,66]. Based on these results, we decided to develop an AAV9-based vector for VEGF-B expression in the retina in our experimental settings. The vectors used in this study express the exogenous gene under the control of the strong, constitutive, immediate early promoter from the human cytomegalovirus (CMV).

Mice for gene transduction studies

Pathogen-free C57BL/6 mice from Charles River Laboratories (Wilmington, MA) were treated in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research, the guidelines of the Johns Hopkins University Animal Care and Use Committee, and the guidelines of the Institute for Laboratory Animal Research (Guide for the Care and Use of Laboratory Animals).

Intraocular injections of adeno-associated virus vectors in mice

Subretinal or intravitreous injections were administered with a Harvard pump microinjection apparatus and pulled glass micropipets, as previously described [67]. Briefly, under a dissecting microscope, the sharpened tip of a micropipette was passed through the sclera, just behind the limbus into the vitreous cavity. Each micropipet was calibrated to deliver 1 μl of vehicle containing approximately 1×10⁹ viral genome copies (vgc) of AAV-VEGF-B167, AAV-VEGF-B186, or AAV-GFP, upon depression of a foot switch.

Real-time PCR

Total RNA was isolated from retinas using the RNeasy Mini Kit (Qiagen, Alameda, CA), and RNA quality was measured using a NanoDrop1000 spectrophotometer (Thermo Scientific, Waltham, MA). The cDNA was prepared using the SuperScript First-Strand Synthesis System (Invitrogen, Carlsbad, CA), then used as a template for real-time PCR amplification to detect the expression levels of murine VEGF-B167 and 186. Real-time PCR was performed using iQ SYBR Green Supermix and MyiQ Single-Color Real-Time PCR Detection System (BioRad, Hercules, CA). The forward and reverse primers that were used are shown in Table 1. All primers were synthesized by Integrated DNA Technologies (Coralville, IA). Five serial eightfold dilutions of one of the samples were used to establish a standard curve in each experiment with the primer Actb (β-actin). The expression level of each gene was normalized to that of Actb. Samples were prepared in at least triplicate, and real-time PCR measurement of each sample was done in triplicate. The formation of a single product was also confirmed by observing the melting-curve graph that was generated by the thermal cycler machine, MyiQ, for each reaction tube. Agarose gel electrophoresis was used to confirm that reaction products had the expected size.

Fluorescence microscopy after subretinal injection of AAV-GFP

To verify the expression of AAV vectors in retinas, 1 to 6 weeks after subretinal injection of the AAV-GFP vector or phosphate buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 8 mM Na₂HPO₄, and 2 mM KH₂PO₄), mice were humanely sacrificed and eyes were rapidly removed and immersed in 10% formalin for at least 3 h at room temperature. Some eyes were used to make choroidal or retinal flatmounts, the others for preparing cryosections. To prepare retinal flatmounts, the anterior segments of eyes were removed and the entire retina was carefully dissected from the remaining eyecups. Radial cuts were made from the edge to the equator of the retina, and the retina was flatmounted in aquamount (Thermo Fisher Scientific, Waltham, MA) with the photoreceptors facing down. For choroidal flatmounts, radial cuts were also made in the remaining eyecups, which were flatmounted with the sclera facing down. To prepare cryosections, the cornea and lens were removed and the eyecups were cryoprotected in 20% sucrose in PBS overnight at 4 °C. Eyes were then embedded in optimal cutting temperature embedding compound (OCT; Miles Diagnostics, Elkhart, IN) and frozen in a slurry of dry ice and isopentane. Ten-micron frozen sections were cut, then dried in air for 20 min, rinsed in PBS, and mounted with aquamount containing 0.2 μg/ml 4´, 6-diamidino-2´-phenylindole, dihydrochloride. Flatmounts and sections were examined using fluorescence microscopy (Axioskop microscope; Zeiss, Thornwood, NY), and images were digitized using a three-color charge-coupled device video camera (IK-TU40A; Toshiba, Tokyo, Japan) and a frame grabber.

Immunofluorescence staining for vascular endothelial growth factor-B

Mice treated with a subretinal injection were humanely killed after 4 weeks, and the eyes were removed and fixed with 4% paraformaldehyde in PBS for 15 min. Then, the anterior segments were removed, and the remaining posterior retina was immersed again in a 4% paraformaldehyde solution at 4 °C overnight. Eyes were sequentially washed with PBS and 12.5% and 25% sucrose, and then embedded in OCT. Ten-micron frozen sections were cut, air dried, and washed with PBS. Specimens were blocked with 10% normal goat serum in PBS for 30 min at room temperature to prevent nonspecific binding. The slides were incubated with a 1:50 dilution of mouse monoclonal antihuman/antimouse VEGF-B antibody (MAB751; R&D), which identifies both VEGF-B isoforms, in 2% normal goat serum /PBS overnight at 4 °C in a humidified chamber, then washed with PBS three times, 5 min each time, and incubated with Alexa Fluor 594 goat antimouse IgG (H⁺L) secondary antibody (Molecular Probes, Carlsbad, CA) for 1.5 h at 37 °C. After washing, the slides were mounted with aquamount containing 0.2 μg/ml 4´, 6-diamidino-2´-phenylindole, dihydrochloride. For a parallel negative control, the primary antibody was omitted and PBS was substituted. Sections were examined using fluorescence microscopy (Axioskop microscope; Zeiss) and confocal microscopy (Zeiss LSM 710), and images were taken under the same conditions for optimal comparison.

Histological examination

Four weeks after vector injections into 4-week-old normal C57BL/6J mice, mice from each group were euthanized. Cryosections were prepared as described above. Eyes were serially sectioned at 10 μm thickness, mounted onto Superfrost/Plus microscope slides (Fisher Scientific, Pittsburgh, PA), stained with hematoxylin and eosin, and examined by light microscopy.

Quantitative measurement of leukostasis

Leukocytes were labeled with fluorescein isothiocyanate (FITC)-conjugated Concanavalin A (Vector Labs, Burlingame, CA) as previously described [68]. Briefly, mice were perfused with PBS to remove erythrocytes and non-adherent leukocytes, followed by a perfusion with fluorescein-conjugated concanavalin A to label leukocytes. Another PBS perfusion was used to flush out unbound fluorescein. Retinal flatmounts were prepared and examined with the Axioskop microscope. The total numbers of leukocytes adhering to the retinal vessels were counted at 200× by the same investigator, with the investigator being masked as to the nature of the specimen.

Quantitative assessment of the blood–retinal barrier

Quantitative measurement of the breakdown of the BRB was performed as previously described [68,69]. Briefly, mice subretinally injected with AAV-VEGF-B167, AAV-VEGF-B186, or AAV-GFP control vector were sedated and given an intraperitoneal injection of 1 μCi/g bodyweight of [³H]-mannitol after 4 weeks. One hour after injection, the mice were sedated, retinas were rapidly removed and dissected to free them from the lens, vitreous, and any retinal pigment epithelium (RPE) that has been extruded. The retinas were then placed into preweighed scintillation vials. The thoracic cavity was opened, and the left superior lobe of the lung was removed, blotted to remove excess blood, and placed in another preweighed scintillation vial. A left dorsal incision was made, and the retroperitoneal space was entered without entering the peritoneal cavity. The renal vessels were clamped with a forceps, and the left kidney was removed, cleaned of all fat, blotted, and placed into a preweighed scintillation vial. The remaining droplets on the tissues were allowed to evaporate for 20 min. The vials were weighed, and the tissue weights were calculated and recorded. One ml of NCSII solubilizing solution was added, and the vials were incubated overnight in a 50 °C water bath. The solubilized tissue was brought to room temperature and decolorized with 20% benzoyl peroxide in toluene in a 50 °C water bath. After re-equilibrating it to room temperature, 5 ml of Cytoscint ES (Fisher Scientific) and 30 μl of glacial acetic acid (Sigma, St. Louis, MO) were added, and the vials were stored for several hours in darkness at 4 °C to eliminate chemoluminescence. Radioactivity was counted with a Wallac 1409 Liquid Scintillation Counter. The CPM/mg tissue measured for the lungs, kidneys, and retinas from the experimental and control groups was used to calculate the retina-to-lung (RLLR) and retina-to-renal leakage ratios (RRLR). The ratios obtained for retinas treated with AAV-VEGF-B167 or AAV-VEGF-B186 were compared with those treated with AAV-GFP control vectors by Student’s unpaired t test for populations with unequal variances.

Animal model of oxygen-induced ischemic retinopathy

Postnatal day 7 (P7), C57BL/6J mice were given an intravitreal injection of 1 μl of vehicle containing approximately 1×10⁹ vgc of AAV-GFP, AAV-VEGF-B167, or AAV-VEGF-B186 and placed in a 75% oxygen box for 5 days. At P12, the mice were returned to room air. At P17, the mice were humanely sacrificed. Eyes were enucleated and fixed in 4% paraformaldehyde for 30 min at room temperature. Retinas were dissected and stained overnight at room temperature with fluoresceinated Griffonia Simplicifolia Isolectin B4 (Alexa Fluor 594–conjugated I21413; 1:100 dilution; Molecular Probes) in PBS. Following three washes, the retinas were whole-mounted onto Superfrost/ Plus microscope slides with the photoreceptor side down.

Quantification of retinal NV was performed as described previously [70,71]. Briefly, images of each of the four quadrants of whole mounted retinas were taken at 5× magnification on a Zeiss Axioplan 2 microscope and imported into Adobe Photoshop (Adobe Systems, Inc., San Jose, CA). Retinal segments were merged to produce an image of the entire retina. Neovascular tuft formation was quantified by comparing the number of pixels in the affected areas with the total number of pixels in the retina. Percentages of NV from mouse retinas treated with AAV-VEGF-B167 or 186 were compared with those from the retinas treated with AAV-GFP. Measurement of retinal NV was done blind to the identity of the sample.

Animal model of laser-induced choroidal neovascularization

Two weeks after vector injection, Bruch’s membrane was ruptured by laser photocoagulation at three locations in each eye as previously described [72]. Briefly, C57BL/6 mice were anesthetized with ketamine hydrochloride (100 mg/kg bodyweight), and their pupils were dilated. Laser photocoagulation (75 μm spot size, 0.1 s duration, 120 mW) was performed in the 9, 12, and 3 o’clock positions of the posterior pole of each eye with the slit-lamp delivery system of an OcuLight GL diode laser (Iridex, Mountain View, CA), and a handheld coverslip was used as a contact lens to view the retina. Production of a tissue bubble by the laser, which indicates ruptures of Bruch’s membrane, is an important factor in obtaining CNV; therefore, only burns in which a bubble was produced were included in the study.

After 2 weeks, the mice were perfused with 1 ml of PBS containing 50 mg/ml of fluorescein-labeled dextran (2×10⁶ Da average molecular mass; Sigma-Aldrich, St. Louis, MO), and choroidal flatmounts were examined by fluorescence microscopy. Images were captured at 20× magnification on a Zeiss Axioplan 2 microscope. Image analysis software (Image-Pro Plus; Media Cybernetics, Silver Spring, MD) was used to measure the total area of CNV at each rupture site, with the investigator masked with respect to treatment group. This is a widely used and accepted method for assessing experimental CNV [72,73].

Statistical analysis to determine significance

Data were expressed as mean±SEM. Statistical analysis was performed using Student’s t test; p<0.05 was considered significant.

Construction and characterization of recombinant adeno-associated virus vectors

Three vectors based on the AAV serotype 9 (AAV9) were constructed, respectively, expressing the 167 and 186 amino acid isoforms of mouse VEGF-B (AAV-VEGF-B167 and AAV-VEGF-B186) or the green fluorescence protein gene (AAV-GFP) under the control of the human CMV promoter (Figure 1A). For the three vector preparations, titers were approximately 1.0×10¹² vgc/ml.

Transgene expression of recombinant adeno-associated virus vectors in the mouse retina

To determine the location and the time course of rAAV expression in mouse retinas, 4-week-old C57BL/6J mice received subretinal injections of AAV-GFP or PBS. Animals were killed 1–6 weeks after injection. Under fluorescence microscopy, GFP expression was clearly demonstrated at 4 weeks postinjection, primarily in the RPE and the outer retina (Figure 1B-D) in mice treated with 10⁸ or 10⁹ vgc, but not 10⁷ or 10⁶ vgc, of the AAV-GFP vector, whereas animals injected with PBS showed no green fluorescence during the 1 to 6 weeks after subretinal injection. GFP expression was maximal at 4 weeks.

Analogous results were obtained by subretinal injection with the AAV vectors expressing VEGF-B167 and 186. Immunofluorescence performed with the mouse monoclonal antibody reacting with both isoforms of mouse VEGF-B at 4 weeks after injection showed that expression of VEGF-B167 (Figure 1E) or VEGF-B186 protein (Figure 1F) was greater in retinas of AAV-VEGF-B injected mice, mainly in the RPE, outer segments (OS), outer plexiform layer (OPL), inner nuclear layer (INL), inner plexiform layer (IPL), ganglion cell layer (GCL), and endothelial cells than in mice from the GFP control group (Figure 1G). Data was not shown for the other time points. Real-time PCR assay further confirmed that the mRNA expression levels of VEGF-B167 (Figure 1H) and 186 (Figure 1I) in treated groups were significantly higher (5.8 fold and 12-fold respectively) than in the AAV-GFP control group. Administration of 10⁸ vgc or less of AAV-VEGF-B167 or 186 did not result in increased expression, probably because this was a subthreshold level or because the increase in expression was too small for our assay to detect. Our data for GFP expression was similar to those of previous reports [74,75]; therefore, the same dosage of 10⁹ vgc/μl of AAV vectors was used for all ensuing experiments.

Neither VEGF-B167 nor 186 induced an obvious inflammatory response

Based on published references and our own studies, VEGF-A notably induced not only neovascularization, but also an inflammatory response and vasopermeability [68,76-78]. Therefore, we questioned whether intraocular overexpression of both isoforms of VEGF-B would induce similar inflammatory effects. Hematoxylin and eosin staining did not show evidence of inflammatory cell infiltration or edema on retinal sections from the control and experimental groups (Figure 2A-C) and the retinas and their vasculature had relatively normal configuration.

To determine more definitively the inflammatory response to VEGF-B167 and 186, a leukostasis assay was performed and the numbers of leukocytes adherent to retinal vessels were counted. Four weeks after subretinal injection of AAV-vectors, neither AAV-VEGF-B167 nor AAV-VEGF-B186 caused significant leukostasis compared to AAV-GFP controls (AAV-VEGF-B167: 8.8±0.9 adherent leukocytes/retina, n=10; AAV-VEGF-B186: 7.75±1.2 adherent leukocytes/retina, n=8; AAV-GFP: 7.3±0.9 adherent leukocytes/retina, n=10; P_{VEGF-B167/GFP}=0.14; P_{VEGF-B186/GFP}=0.38) (Figure 2D-G). A comparative image showing VEGF-A-induced retinal leukostasis is shown in Figure 2H.

Overexpression of VEGF-B186, but not VEGF-B167, significantly increased vasopermeability of the retina

Our previous experiments have demonstrated that determining the amount of [3H]-mannitol that enters the retina from the blood, relative to the amount that enters from the lung or kidney, is a sensitive and reproducible technique for quantifying the breakdown of the BRB [55]. Mice treated with a subretinal injection of AAV-VEGF-B167 showed that the RLLR (Figure 3A) and RRLR (Figure 3B) were 1.44±0.5 (n=8) and 1.48±0.5 (n=8), respectively, which was almost a threefold increase. This was not statistically different compared to those ratios from the AAV-GFP control mice (RLLR: 0.55±0.13, n=5; RRLR: 0.52±0.09, n=7; P_RLLR=0.10; P_RRLR=0.05). However, mice treated with AAV-VEGF-B186 showed an RLLR or RRLR of 2.58±0.86 (n=7) or 1.92±0.61 (n=9), respectively, about a four- to fivefold increase that was statistically different compared to those ratios from the AAV-GFP control mice (P_RLLR=0.04; P_RRLR=0.03).

Both VEGF-B167 and VEGF-B186 increased choroidal neovascularization

To determine whether the overexpression of VEGF-B would provide any increase in NV, we compared the AAV-VEGF-B167, AAV-VEGF-B186, and AAV-GFP-treated eyes in an experimental model of CNV. Four weeks after vector injection and two weeks after rupture of Bruch’s membrane with laser photocoagulation, the area of CNV at Bruch’s membrane rupture sites appeared larger in eyes that had been given subretinal injections of AAV-VEGF-B167 or AAV-VEGF-B186 (Figure 4A,B) when compared with those given a subretinal injection of AAV-GFP (Figure 4C). Measurement of the area of CNV at rupture sites by image analysis, with the investigator masked with respect to treatment group, confirmed that the mean area of CNV was significantly greater in eyes given subretinal injections of either AAV-VEGF-B167 (0.019±0.003 mm², n=22, p=0.02) or AAV-VEGF-B186 (0.021±0.017 mm², n=19, p=0.02) (Figure 4D) when compared with eyes given AAV-GFP (0.012±0.002 mm², n=20) by the corresponding route of administration.

Both VEGF-B167 and VEGF-B186 increased retinal neovascularization in oxygen induced ischemic retinopathy

Given that overexpression of VEGF-B167 and 186 significantly increased the area of CNV, we next explored whether they would increase the area of pathological retinal NV. Developing retinal NV in the mouse model of OIR has proven reliable and quantifiable over 17 days, and is widely used to clarify the molecular changes in neovascular eye diseases and to screen anti-angiogenic compounds [79-81]. At P7, pups received an intravitreal injection of AAV-VEGF-B167, AAV-VEGF-B186, or AAV-GFP vectors and were then placed in a 75% oxygen box for 5 days. Thereafter, pups were returned to room air for another 5 days. At day 3, 6, 8, and 10 after vector delivery, three AAV GFP-injected pups were killed to observe the GFP protein expression, which was initially and focally seen in the retina at day 3, and became stronger and more diffuse at day 6 and 10 (Figure 5). Therefore, protein expression of the GFP reporter gene at the above time points indicated that AAV VEGF-B167 or 186 could also effectively be expressed at these time points. At P17 (10 days after receiving the AAV vectors), increased expression of VEGF-B was also confirmed in AAV-VEGF-B-injected pups by quantitative PCR. At P17, mice treated with AAV-VEGF-B167 (18.6%±2.1%, n=10, p<0.001) or AAV-VEGF-B186 (16.2%±0.89%, n=8, p=0.001) developed more extensive retinal NV than control mice treated with AAV-GFP (9.97% ± 1.3%, n=13; Figure 6A-D).