Nanoparticle-Mediated Expression of an Angiogenic Inhibitor Ameliorates Ischemia-Induced Retinal Neovascularization and Diabetes-Induced Retinal Vascular Leakage

The human K5 cDNA (362 bp) was amplified by PCR using a pair of primers containing a 6× histidine tag (His-tag) at the COOH-terminus of K5. For secretion of K5, a 52-bp linker encoding the signal peptide was cloned into a pcDNA3.1(+). The amplified K5 cDNA with the His-tag sequence was subcloned into pcDNA3.1(+) vector at the BamHI and XbaI sites in frame with the signal peptide sequence. The resulting pcDNA3.1(+)-SP-K5–6×His-tag expression construct was confirmed by restriction digestion and DNA sequencing.

PLGA:Chitosan nanoparticles containing the K5 expression plasmid (PLGA:CHN pK5-NPs) were prepared using a previously reported emulsion-diffusion evaporation technique (24) with some modifications. Briefly, 15.5 mg PLGA (50:50, 0.17 i.v.; Birmingham Polymers) was dissolved in 5 ml ethyl acetate. The polyvinyl acetate solution (1%, wt/vol) was prepared and then chitosan chloride (Nova Matrix PCL 113, 2.5 mg) was stirred until dissolved. Plasmid solution (1 ml containing 350 μg plasmid) was modified with 10% (wt/vol) sodium sulfate and added to the chitosan solution for complexation and DNA condensation and stirred. Both solutions were then combined and emulsified with a probe sonicator for 4 min. To the emulsion, ∼30 ml Milli-Q H₂O was added and stirred on a magnetic plate stirrer for 3 h to evaporate the solvent. The particle suspension was ultracentrifuged and resuspended in Milli-Q H₂O, and the procedure was repeated twice. Upon final resuspension in Milli-Q H₂O, the nanoparticle suspension was lyophilized to obtain dry particles.

The particle size and size distribution were determined using dynamic light scattering (Brookhaven Instruments, Holtsville, NY). The same equipment was used to determine the ζ-potential of the particles.

For plasmid loading estimation, the lyophilized product (0.2 mg) was dispersed in 1 ml methylene chloride to dissolve the polymer, followed by extraction of the plasmid into 2 ml Tris-EDTA buffer. An aliquot of the Tris-EDTA buffer fraction was analyzed for the absorbance at 260 nm to determine the plasmid content per milligrams of nanoparticles.

For in vitro assay, protein extracts were prepared from a human Müller cell line (MIO-M1) (a generous gift from Drs. Khaw and Limb [(25)]) and treated with Control-NP or K5-NP under hypoxia for 48 h. For in vivo assay, the free-floating retinas were homogenized in 150 μl ice-cold tissue lysis buffer (50 mmol/l Tris-HCl, pH 7.4, 2 mmol/l MgCl₂, 1% NP-40, 10% glycerol, 100 mmol/l NaCl) and a protease inhibitor cocktail (1:100 dilution, Sigma Aldrich, St. Louis, MO) added just before use. The lysates were cleared by centrifugation at 12,000 g for 20 min at 4°C. The protein concentration of the lysate was determined using the Bradford assay. The same amount of proteins (50 μg) were resolved by SDS-PAGE in 8% denaturing gels and transferred onto nitrocellulose membranes. Ponceau S staining (Sigma Aldrich) was carried out to verify equal protein loading. Immunoblotting and signal detection by enhanced chemiluminescences were performed as described previously by incubating the membranes with the primary antibodies against rabbit anti-VEGF (1:1,000 in 5% nonfat milk powder/0.1% TBS Tween-20, Santa Cruz), HIF-1α (1:1,000; BD Transduction Laboratories), goat anti-ICAM-1 (1:1,000; Santa Cruz), mouse anti–His-tag (1:2,000; Upstate), mouse anti-TATA box-binding protein (TATA-binding protein [TBP], 1:1,000; Abcam), or mouse anti–β-actin (1:2,000; Abcam) antibodies overnight at 4°C. The signal intensity was quantified by densitometry using software (SynGene, Frederick, MD).

Immunostaining was performed as described previously (26). Briefly, the eyes were cross-sectioned vertically through the center of the cornea and optic nerve, and both halves of the eyeball were embedded with the center facing down. Serial cryosection (5 μm thickness) was blocked with 10% goat serum in 0.1% Triton X-100 and 3% BSA in PBS. After PBS washes, the antibody specific for the His-tag was added and incubated with the sections at 4°C overnight. The sections were rinsed several times with PBS and incubated with a fluorescein isothiocyanate conjugated anti-mouse IgG antibody for 1 h. The slides were then rinsed in cold PBS and viewed under a fluorescence microscope.

The reverse transcription and real-time PCRs (RT-PCR) were performed as described previously (26). The primers used for human VEGF (5′ CAGAGCGGAGAAAGCATTTG and 3′ TGGTTCCCGAAACCCTGAGG) amplified a 180-bp fragment of VEGF. The 18S rRNA was amplified using primers (5′ TGCTGCAGTTAAAAAGCTCGT and 3′ GGCCTGCTTTGAACA CTCTAA) for normalization of the VEGF mRNA levels.

All the animal experiments were performed in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Brown Norway rats (Charles River Laboratories) were used for the OIR model following an established protocol (27). OIR rats received an intravitreal injection of K5-NP into the right eye and the same dose of Control-NP into the left eye at age of P12. Diabetes was induced and monitored in adult Brown Norway rats by an intraperitoneal injection of STZ (50 mg/kg in 10 mmol/l of citrate buffer, pH 4.5) as described previously (14,28). Diabetic rats at 2 weeks after the onset of diabetes received an intravitreal injection of Control-NP and K5-NP in the treatment group and the same dose of Control-NP in the control group.

Retinal vascular permeability was measured using the Evens blue-albumin leakage assay following an established protocol (29). Concentrations of Evans blue in the retina were normalized by total retinal protein concentrations and by Evans blue concentrations in the plasma.

Retinal vasculature was visualized by fluorescein angiography as described (13,27). The neovascular area was measured in the retina using SPOT software (Diagnostic Instruments, Sterling Heights, MI). For quantification of preretinal vascular cells, eyes were fixed, sectioned, and stained as described previously (27). The preretinal nuclei were counted, averaged, and compared using Student's t test.

Rats were dark adapted for at least 12 h. The rats were killed and pupils dilated with topical application of 2.5% phenylephrine and 1% tropicamide. Electroretinogram (ERG) responses were recorded with a silver chloride needle electrode placed on the surface of the cornea after 1% tetracaine topical anesthesia. A reference electrode was positioned at the nasal fornix and a ground electrode on the foot. The duration of light stimulation was 10 ms. The band pass was set at 0.3–500 Hz. Fourteen responses were recorded and averaged with flash intervals of 20 s. For quantitative analysis, the B-wave amplitude was measured between A- and B-wave peaks. The ERG waveforms of both eyes in the same animal were simultaneously recorded and compared using the right-to-left eye ratio of B-wave amplitude.

Nuclear and cytosolic fractions were isolated using the Fractionation System Kit (Biovision, Mountain View, CA). Nuclear and cytosolic protein concentrations were measured using the Bradford assay. The proteins were blotted with an antibody specific for hypoxia-inducible factor-1 (HIF-1α; Santa Cruz Biotechnology, Santa Cruz, CA) at 1:500 dilution.

Nanoparticle size, polydispersity index, ζ-potential, and plasmid loading were measured with four different nanoparticle batches, and various parameters are shown below (mean ± SD). The nanoparticle formulation exhibited a mean hydrodynamic diameter of 260 ± 30 nm, a polydispersity index of 0.28 ± 0.005, ζ-potential of 8.4 ± 2.4 mV, and a plasmid loading of 11 ± 1.4 μg/mg particles.

ARPE19 cells were transfected with 10 μg of K5-NP or Control-NP for 3 days. The concentrated conditioned serum-free medium from the transfected cells was blotted separately with the anti-K5 and anti–His-tag antibodies. Both of the antibodies detected significant amounts of K5 with the expected molecular weight secreted from the cells transfected with K5-NP, but not from the cells transfected with Control-NP (Fig. 1 A).

The K5 expression was also determined by immunocytochemistry using the anti–His-tag antibody. The antibody detected K5 signal in the cells transfected with K5-NP, but not in the cells transfected with Control-NP (Fig. 1 B and C).

To determine the biological effect of K5-NP, primary bovine retinal capillary endothelial cells (BRCECs) and ARPE19 cells were treated with K5-NP for 3 days, and viable cells were quantified using the MTT assay. As shown in Fig. 2,A and B, K5-NP induced a dose-dependent decrease in viable BRCEC numbers, when compared with the untreated cells and the cells treated with Control-NP (Fig. 2,A). In contrast, the same K5-NP treatment did not result in significant reduction of viable cells in ARPE19 (Fig. 2 B), suggesting that the inhibitory effect of K5-NP on cell growth is endothelial cell–specific.

Our previous studies showed that K5 inhibits angiogenesis via downregulation of VEGF expression (30). Here we evaluated the effect of K5-NP on VEGF expression by measuring VEGF secretion and VEGF mRNA levels. As shown by ELISA, ARPE19 cells overexpressed VEGF after exposure to hypoxia (1% O₂). K5-NP treatment significantly reduced the VEGF secretion induced by hypoxia, to a level comparable to that from the cells under normoxia (Fig. 2,C). Similarly, RT-PCR showed that K5-NP significantly downregulated the VEGF mRNA expression under hypoxia, whereas Control-NP did not affect VEGF expression under the same conditions (Fig. 2 D).

The effects of K5-NP were also evaluated in cultured rMc-1, a cell line derived from rat retinal Müller cells (31). As shown by Western blot analysis, K5-NP attenuated the overexpression of VEGF and decreased HIF-1α levels in Müller cells cultured under hypoxia (1% O₂) (Fig. 2 E).

To examine the expression of K5 from K5-NP in the retina, K5-NP was injected into the vitreous of the right eye (8.8 μg/eye) and the same amount of Control-NP into the left vitreous of four OIR rats at age of postnatal day 12 (P12) after exposure to 75% oxygen. At age P18, the immunohistochemistry was performed on retinal sections using the anti–His-tag antibody. The immunostaining detected high levels of K5 expression in the inner retina in the eyes injected with K5-NP, but not in the contralateral control eyes (Fig. 3 A–D).

To determine the duration of the K5 expression after a single injection of K5-NP, the retinas were dissected and homogenized at 1, 2, 3, and 4 weeks after an intravitreal injection of K5-NP in adult rats (three rats per time point) for Western blot analysis using the anti–His-tag antibody. K5 was detected with an expected molecular weight in the retinas injected with K5-NP at all the time points analyzed (Fig. 3 E), but not in the retinas injected with the Control-NP.

To evaluate the effect of K5-NP on retinal vascular leakage, K5-NP was injected intravitreally into the right eye (2.2 or 8.8 μg/eye) at age P12 and the same amount of Control-NP into the contralateral eyes. Retinal vascular leakage was measured at P16 using the Evans blue-albumin leakage method and compared with the contralateral eyes. The results showed that the eyes injected with 8.8 μg of K5-NP had significantly lower vascular permeability than that in the contralateral retinas injected with Control-NP (P = 0.011, n = 7; Fig. 4,B). The injection of 2.2 μg of K5-NP did not result in a statistically significant decrease of vascular leakage (P = 0.0531, n = 7; Fig. 4 A).

To evaluate the effect of K5-NP on retinal neovascularization, K5-NP was injected intravitreally into the right eyes (8.8 μg/eye) of the OIR rats at P12 and Control-NP into the left eyes. The retinal vasculature was visualized and examined by fluorescein retinal angiography at P18. Retinal angiographs on retinal flat mounts showed that the eyes injected with Control-NP developed severe retinal neovascularization in the OIR rats (Fig. 5,A and C). In contrast, a single K5-NP injection ameliorated the retinal neovascularization in the same rats by decreasing neovascular areas and nonperfusion areas in the retina (Fig. 5,B and D). The neovascularization was semi-quantified by measuring the ratio of the neovascular area to the total retina area, which showed that the eyes injected with K5-NP have significantly decreased retinal neovascular areas in the OIR rats, compared with those injected with Control-NP (Fig. 5 E).

The effect of K5-NP on preretinal neovascularization was also quantified by counting vascular cells growing into the vitreous space (preretinal vascular cells) on eight noncontinuous cross-sections from each eye following an established method (27). The result showed that the eyes injected with K5-NP have significantly fewer preretinal neovascular cells, compared with the contralateral eyes injected with Control-NP (Fig. 6).

To determine the effect of K5-NP on retinal inflammation, K5-NP was injected intravitreally into STZ-induced diabetic rats, with the same dose of Control-NP as vehicle control. As VEGF and ICAM-1 are both known to play important roles in the retinal inflammation and blood-retinal barrier breakdown in diabetes, we measured retinal levels of ICAM-1 and VEGF from nondiabetic rats, STZ-induced diabetic rats, and diabetic rats treated with Control-NP and K5-NP using Western blot analysis. Four weeks after the injection, the retinas from rats injected with K5-NP showed decreased retinal VEGF and ICAM-1 levels by more than 50%, compared with that in the retina injected with Control-NP and to that in untreated diabetic rats (Fig. 7,A). Consistently, leukostasis assay showed that K5-NP also decreased adherent leukocytes in the retinal vasculature of diabetic rats (supplementary Fig. 1, available in an online appendix at http://diabetes.diabetesjournals.org/cgi/content/full/db08-1327/DC1). Retinal vascular permeability in the retinas of nondiabetic and STZ-induced diabetic rats, and diabetic rats treated with Control-NP and K5-NP, was measured using the Evans blue-albumin leakage method. As shown by retinal vascular permeability assay, K5-NP reduced retinal vascular leakage by 30% in STZ-induced diabetic rats, 4 weeks after the injection of K5-NP (P < 0.001, Fig. 7 B), suggesting that the effect of K5-NP on retinal vascular leakage lasted for at least 4 weeks after a single injection. In contrast, Control-NP injection did not result in any significant difference in retinal levels of VEGF and ICAM-1 or retinal vascular leakage in the diabetic rats.

Possible toxicities of K5-NP were evaluated structurally by histological analysis. The eyes from rats with K5-NP injection and those with Control-NP injection were sectioned, stained, and examined under light microscope at 1, 2, and 4 weeks after the injection. No difference was observed in number of retinal nuclear layers or thickness of the retina among the rats injected with K5-NP and the control (Fig. 8 A–F). We also measured the retinal thickness at the same six retinal locations in each eye as described in supplementary methods. There is no significant difference in retinal thickness in the rats treated with K5-NP and controls (supplementary Fig. 2, available in the online appendix). Similarly, K5-NP did not increase apoptotic cells in the retina (supplementary Fig. 3, available in the online appendix).

Possible adverse effects of K5-NP on visual functions were evaluated by ERG recording in normal and diabetic rats. As shown in Fig. 8, the amplitudes of A- or B-waves were not significantly different among the Control-NP or K5-NP treated rats and untreated rats. STZ-induced diabetic rats showed ERG responses similar to the nondiabetic rats (Fig. 8 G and H).

To investigate the mechanism responsible for the anti-angiogenic activity of K5-NP, we have measured the effect of K5-NP on HIF-1 activation, since HIF-1 is the key transcription factor activating VEGF and is known to play a key role in retinal neovascularization (7). HIF-1α nuclear translocation is a critical step in the activation of HIF-1, and thus, we measured cytosolic and nuclear levels of HIF-1α in the human Müller cell line treated with K5-NP under hypoxia. As shown by Western blot analysis of isolated nuclear proteins, exposure of the cells to hypoxia (1% oxygen) increased HIF-1α levels in the nuclei (Fig. 9,A). K5-NP, but not Control-NP, completely blocked the increase of nuclear HIF-1α levels under hypoxia (Fig. 9 A).

Similarly, nuclear translocation of HIF-1α was also determined in the retina of the OIR rats by immunostaining of HIF-1α. OIR rats showed high levels of HIF-1α in the nuclei in the inner retina injected with Control-NP (Fig. 9,B–D). In the contralateral eyes injected with K5-NP, the HIF-1α signal in the nuclei of retinal cells was apparently decreased (Fig. 9 E–G), suggesting that K5-NP inhibited HIF-1α nuclear translocation.

Retinal vascular leakage, the direct cause of diabetic macular edema, and retinal neovascularization are the major causes of vision loss in diabetic patients (1). Long-term suppression of retinal vascular leakage and pathologic retinal neovascularization are the goals for the treatment of diabetic retinopathy. Although previous studies showed that peptide angiogenic factors such as K5, angiostatin, and so forth effectively reduce vascular leakage and inhibit retinal neovascularization in rat models (14,32), the difficulty in achieving sustained ocular delivery of the peptides represents a major hurdle for their therapeutic applications. The present study reports the first approach to combine an endogenous angiogenic inhibitor with nanotechnology in the treatment of diabetic retinopathy. Our results showed that an intravitreal injection of K5-NP mediated high-level and sustained K5 expression in the retina and resulted in sustained reductions of retinal vascular leakage in STZ-induced diabetic and OIR models. Further, K5-NP effectively blocked ischemia-induced retinal neovascularization. This study also identified a novel anti-inflammatory effect of K5 in the retina of diabetic rats. Moreover, the K5-NP injection did not result in any detectable toxicity to retinal structure and function. This study revealed therapeutic potential of nanoparticle-mediated gene delivery of angiogenic inhibitors in the treatment of diabetic retinopathy.

Nanoparticles formulated using PLGA polymers are recently being investigated as a new gene delivery system because of their sustained release characteristics, biocompatibility, biodegradability, and ability to protect DNA from degradation in lysosomes (33,34). Our immunoblotting and double immunostaining results showed that K5-NP mediated high-level and sustained K5 expression in cultured cells and in the inner retina, suggesting that K5-NP is efficiently internalized and released into the cytoplasmic compartment rather than being retained in the degradative lysosomal compartment. Moreover, the expressed K5 is secreted into the extracellular space and has biological activity of specifically inhibiting endothelial cell growth and VEGF overexpression under hypoxia and in the retina of STZ-induced diabetes.

Cytotoxicity is a potential concern in some nanoparticle-mediated gene deliveries. We have determined potential cytotoxicities of K5-NP in cultured ARP19 cells and in the retina. In ARPE19 cells treated with K5-NP, no significant reduction in cell viability was observed. In contrast, the same concentrations of K5-NP significantly inhibited endothelial cell growth, suggesting an endothelial cell-specific inhibitory effect of K5-NP. After an intravitreal injection of K5-NP, no significant alterations were observed in the retinal histology at any of the time points analyzed. Moreover, analyses of the ERG response, retinal thickness, and cell death in the retina suggest that K5-NP does not have any detectable side effects on retinal structure and functions. These data suggest that K5-NP has no severe toxicities at the dose used.

Our previous studies demonstrated that the anti-angiogenic activity of K5 is through downregulating the expression of VEGF (30). As VEGF is a major permeability and angiogenic factor, downregulation of VEGF expression is believed to be responsible for the effects of K5 on vascular leakage and retinal neovascularization (14,15). To confirm whether the expressed K5 from nanoparticles plays a role in blocking the VEGF overexpression under hypoxia, we measured VEGF expression in retinal pigment epithelium cells and Müller cells. Exposure of the cells to hypoxia increased the VEGF expression. The overexpression of VEGF under hypoxia can be effectively blocked by K5-NP, suggesting that the mechanism underlying the vascular activities of K5-NP is identical to that of the K5 peptide.

The OIR model is commonly used for studies of proliferative diabetic retinopathy (35,–37). Although it is not a diabetic model, the pathology in the retina in this model is similar to that in diabetic retinopathy in humans (27). Moreover, VEGF overexpression has been shown to be the key pathogenic factor for retinal vascular leakage and neovascularization in this model, a pathogenic mechanism similar to that in diabetic retinopathy (8,38). Our results showed that K5-NP reduced retinal vascular permeability in a dose-dependent manner, demonstrating a beneficial effect on vascular leakage induced by hypoxia.

To evaluate the effect of K5-NP on retinal neovascularization in the OIR model, we used fluorescein angiography on flat-mounted retinas. The results showed that a single injection of K5-NP significantly reduced retinal neovascular area and nonperfusion area that have been shown to correlate with the severity of diabetic retinopathy (27,35). Because preretinal neovascularization (i.e., neovasculature growing into the vitreous space) is a characteristic of proliferate diabetic retinopathy, we have also evaluated the effect of K5-NP on preretinal neovascular events. The results showed that K5-NP significantly reduced preretinal neovascular cells, compared with the contralateral eyes injected with Control-NP. These results demonstrate that K5-NP attenuates ischemia-induced retinal neovascularization. A limitation of the OIR model, however, is that both the vascular leakage and retinal neovascularization are transient, with peaks at P16 and P18, respectively (27,28).

Sustained expression is a goal of gene delivery. In the entire experimental period (4 weeks), K5-NP mediated efficient expression of K5 in the retina. To evaluate the sustained effects of K5, we have used the STZ-diabetic rat model, a commonly used type 1 diabetic model. As STZ-induced diabetic rats are known to develop vascular leakage and retinal inflammation but not neovascularization, we measured the effect of K5-NP on retinal vascular leakage and inflammation in the STZ-induced diabetic model. Our results showed that K5-NP reduced retinal vascular leakage for at least 4 weeks after a single injection in diabetic rats. Further, K5-NP attenuated the over-expression of VEGF and ICAM-1, two major inflammatory factors (3,39). Further, K5-NP also decreased leukostasis in the retina of the diabetic rats. These results indicate that K5-NP has a novel anti-inflammatory activity.

In conclusion, the present study demonstrates that PLGA nanoparticles can be used for gene delivery of angiogenic inhibitors into the retina without toxicity. The nanoparticle-mediated nonviral gene delivery of angiogenic inhibitors has therapeutic potential for the treatment of diabetic macular edema and retinal neovascularization associated with several ocular disorders such as diabetic retinopathy.

This study was supported by National Institutes of Health Grants EY015650, EY12231, and EY019309 to J.M. and EY017045 to U.B.K. (via Emory University); grant P20RR024215 from the National Center for Research Resources; and a grant from the Oklahoma Center for the Advancement of Science and Technology as well as research awards from the American Diabetes Association and the Juvenile Diabetes Research Foundation.