Bright field (A) and fluorescence (B) confocal images 2 days after intravitreal injection of a fluorescein isothiocyanate–labeled antisense oligodeoxynucleotide (AS-ODN) against vascular endothelial growth factor. Confocal microscopy demonstrates selective fluorescent labeling of cells at several retinal sites; AS-ODNs are localized predominantly in the internal limiting membrane (ilm), ganglion cell layer (g), inner nuclear layer (in), outer plexiform layer (op), outer nuclear layer (on), photoreceptor outer segments (os), and retinal pigment epithelium (rpe) (original magnification ×100).
Complete inhibition of iris neovascularization is seen in this animal with injection of an antisense oligodeoxynucleotide (AS-ODN) against vascular endothelial growth factor. Eleven days after laser retinal vein occlusion, brownish discoloration of the iris caused by rubeotic vessels and hemorrhage is shown in the control eye (A) compared with the AS-ODN–treated fellow eye (B). Iris angiograms display florid new vessels producing hyperfluorescence and leakage on the iris surface in the control eye (C), whereas no abnormal iris vessels are seen in the AS-ODN–treated fellow eye (D).
Fluorescein angiograms show partial inhibition in this animal after the injection of an antisense oligodeoxynucleotide (AS-ODN) against vascular endothelial growth factor, with delayed development but not suppression of iris neovascularization. Differential inhibition of iris neovascularization seen 6 days after laser retinal vein occlusion in the control eye (A) and the AS-ODN–treated eye (B) is not sustained at 11 days in the control eye (C) and the AS-ODN–treated eye (D).
Iris neovascularization grades (grade 0 to grade IV in increasing severity) for all measurements in antisense oligodeoxynucleotide (AS-ODN)–treated eyes (3μM) and control eyes during the 14-day study.
Within-animal comparisons of antisense oligodeoxynucleotide–treated eyes (3μM) and control fellow eyes. The degree of inhibition is shown as the grade differential of iris neovascularization (NVI) between test and control eyes in each animal at each measurement.
Bhisitkul RB, Robinson GS, Moulton RS, Claffey KP, Gragoudas ES, Miller JW. An Antisense Oligodeoxynucleotide Against Vascular Endothelial Growth Factor in a Nonhuman Primate Model of Iris Neovascularization. Arch Ophthalmol. 2005;123(2):214-219. doi:10.1001/archopht.123.2.214
To evaluate an antisense oligodeoxynucleotide (AS-ODN) targeted against vascular endothelial growth factor for its effects on ocular angiogenesis and its intraocular localization in a nonhuman primate model of iris neovascularization.
Bilateral laser retinal vein occlusion was performed in monkeys, followed by intravitreal injections of a vascular endothelial growth factor–specific AS-ODN or control. Serial fluorescein angiograms were graded in a masked manner to measure iris neovascularization. Localization was determined using a fluorescent-labeled AS-ODN and confocal microscopy on fixed tissue.
Intravitreally injected vascular endothelial growth factor–specific AS-ODN localized to the retina, in the ganglion cell layer, inner nuclear layer, outer plexiform layer, photoreceptor outer segments, and retinal pigment epithelium. In 8 animals tested with 3μM ODN, AS-ODN–treated eyes had a significant reduction in iris neovascularization compared with control fellow eyes (P = .006, MIXOR analysis). Overall, in 17 animals tested across a range of ODN concentrations (0.1-50.0μM), AS-ODN–treated eyes were more likely to have lower iris neovascularization grades (P = .006, McNemar test) and the absence of iris neovascularization (P< .001, mixed-effects logistic regression model).
Antisense ODNs that target vascular endothelial growth factor delivered to the retina via intravitreal injection reduced iris neovascularization in this model.
Antisense ODNs against vascular endothelial growth factor may have therapeutic potential for neovascular eye diseases.
Antisense oligodeoxynucleotides (AS-ODNs) have been investigated as therapeutic agents in neoplastic, infectious, and inflammatory diseases,1- 5 and the first antisense drug, fomivirsen, has been approved for the treatment of cytomegalovirus retinitis.6,7 Antisense technology has the potential to block target gene expression in a highly specific manner, thereby reducing the effects of nonspecific local and systemic toxicity. Inhibition of target gene expression with AS-ODN occurs primarily via sequence-specific complementary binding to target messenger RNA, leading to the enzymatic degradation of the DNA-RNA hybrid by RNase H or arrested translation by steric mechanisms.8- 11
Vascular endothelial growth factor (VEGF) represents a target gene for antisense therapy in diseases of ocular neovascularization. Vascular endothelial growth factor is a secreted endothelial cell–specific mitogen shown in vitro to be up-regulated by retinal cells in response to hypoxia.12,13 In several animal models, ocular neovascularization is associated with increased levels of VEGF,14 is correlated spatially and temporally with the increased production of VEGF,15 and is stimulated by the exogenous application or overexpression of VEGF.16,17 Furthermore, blockade of VEGF expression in animal models inhibits the formation of ocular neovascularization.18,19
In a previous study20 of a murine model of retinopathy of prematurity, intravitreal injection of VEGF-specific phosphorothioate ODNs suppressed VEGF expression and reduced retinal neovascularization observed histopathologically. Other studies in the rat21 and the rhesus monkey22 have shown inhibition of VEGF expression using AS-ODNs in models of laser-induced choroidal neovascularization. The present study used an AS-ODN that targets the human VEGF sequence in a laser retinal vein occlusion model of iris neovascularization in nonhuman primates. The ocular localization of an intravitreal injection of a fluorescein isothiocyanate (FITC)–labeled VEGF-specific AS-ODN was assessed by in situ localization studies using confocal fluorescence microscopy. Iris neovascularization was evaluated in vivo by serial fluorescein angiography after AS-ODN treatments.
Antisense phosphorothioate ODNs were synthesized on an automated synthesizer (Biosearch 8700; Milligen/Biosearch, Novato, Calif) using either the phosphoamidite or H-phosphonate procedures.23 The AS-ODN was complementary to bases 62 to 80 of the human VEGF sequence (antisense: 5′-CACCCAAGACAGCAGAAAG-3′). For controls, a sense ODN to the same sequence was constructed (sense: 5′-CTTTCTGCTGTCTTGGGTG-3′). The purity of the sequences was analyzed by capillary gel electrophoresis and ion exchange–high-pressure liquid chromatography. Endotoxin levels in the final phosphorothioate ODN preparation were examined using a limulus amebocyte lysate assay and were found to be less than 2 endotoxin units/mL.
U373 glioblastoma cells were treated with ODNs in OptiMEM (Life Technologies, Grand Island, NY) containing lipofectin, 5 μg/mL, for 6 to 8 hours. The cells were then treated with growth media containing 250μM cobalt chloride for 20 to 24 hours. The medium was assayed for protein expression using a human VEGF capture enzyme-linked immunosorbent assay as previously described elsewhere.23
All animal experiments were conducted in accordance with the Association for Research in Vision and Ophthalmology Resolution on the Use of Animals in Research and the Massachusetts Eye and Ear Infirmary guidelines for animal care. Adult cynomolgus monkeys were given a single 100μM intravitreal injection of FITC-labeled VEGF-specific ODN. Days 2 and 3 after injection, enucleations were performed on animals deeply anesthetized with intramuscular injections of ketamine hydrochloride (20 mg/kg, with 5-mg/kg supplementation as needed), acepromazine maleate (0.125 mg/kg), atropine sulfate (0.125 mg/kg), and pentobarbital sodium (5 mg/kg), followed immediately by humane killing by means of an intravenous injection of sodium pentobarbital veterinary euthanasia solution (Webster Veterinary Supply Inc, Sterling, Mass). Eyes were rinsed in cold phosphate-buffered saline solution, and eye cups were fixed in 10% formalin and placed overnight in a 0.3M sucrose solution at 4°C. Retina whole mounts (OCT, stored at −80°C) were cryostat sectioned at 7 to 10 μm and postfixed in methanol-acetone (1:1). Fluorescence confocal microscopy was performed using a BioRad MRC (BioRad Laboratories, Hercules, Calif).
Laser retinal vein occlusion was performed on adult monkeys as previously described elsewhere.15 Briefly, anesthetized cynomolgus monkeys (intramuscular ketamine hydrochloride, 20 mg/kg; acepromazine maleate, 0.125 mg/kg; atropine sulfate, 0.125 mg/kg; and topical 0.5% proparacaine hydrochloride) were treated with a 577-nm dye laser (spot size, 100-200 μm; power, 200-500 mW; duration, 0.1-0.3 second) along the major venous arcades to produce visible vessel occlusion. Iris neovascularization was monitored with serial anterior segment color photographs and iris fluorescein angiograms at baseline and every 3 to 4 days. Iris photographs and angiograms gathered from multiple experiments were graded independently on a 6-grade scale of iris neovascularization (grades 0 to V) in a masked manner by 2 readers (E.S.G. and J.W.M.) as previously described elsewhere.19
In each animal, 1 eye was randomized to receive AS-ODN injections, and the fellow eye received control (sense) ODN injections in the same concentration. Intravitreal ODN injections (0.1-50.0μM ODN in 0.1 mL of balanced salt solution) were given in a masked manner 2 days before laser treatment and again 7 days after the first injection, using a sterile technique and posttreatment gentamicin sulfate drops.
Because the data from neovascularization grades were not numeric but rather were ordered categorical outcomes (grade 0 to grade V in increasing severity), somewhat complex statistical analyses were necessary to make full use of 2 repeated factors in the study design: first, 2 eyes (the AS-ODN–injected eye and the control-injected fellow eye) were compared in each animal, and second, several measurements were compared in each eye across different times. Analysis, therefore, used the MIXOR program24 for ordinal logistic regression with random effects. An additional test was performed that simplified the analysis to a more usual random effects logistic regression by dichotomizing each measurement of each eye to either the absence of neovascularization (grade 0) or the presence of any neovascularization (grade >0). This analysis produces odds ratios as a measure of the effect of treatment, with an odds ratio greater than 1 indicating a greater chance of absence of neovascularization with AS-ODN treatment than with control treatment. Finally, the data were reduced further by classifying each animal as better on the AS-ODN–treated eye (if this eye had lower neovascularization grades than the control fellow eye on at least 1 day, and never higher grades), better on the control eye (if vice versa), or neither (if 1 eye had lower neovascularization grades than the fellow eye at some times but also higher neovascularization grades at other times). The McNemar test was then performed to calculate the probability that the observed excess of eyes with lower neovascularization with AS-ODN treatment could have arisen by chance alone.
In situ localization studies were performed in 2 monkeys injected with intravitreal FITC-ODN (100μM/0.1 mL). Eyes were enucleated 2 and 3 days after injection, and retinal tissue was fixed and studied using fluorescence confocal microscopy. The FITC-ODNs seemed to be selectively localized to cellular elements in several retinal layers, namely, the ganglion cell layer, inner nuclear layer, outer plexiform layer, photoreceptor outer segments, and retinal pigment epithelium (Figure 1). Signal seen in the internal limiting membrane, choroid, and sclera likely represents autofluorescence.
Eight monkeys receiving an intravitreal injection of 3μM AS-ODN were analyzed, and the grade of iris neovascularization in AS-ODN–injected eyes vs control ODN–injected fellow eyes was determined. Iris neovascularization grades were lower in the AS-ODN–treated eye vs the control eye at 1 or more comparisons in 6 of the 8 animals, with no comparisons greater in the AS-ODN–treated eyes (P = .03, McNemar test). Two animals showed complete inhibition (iris neovascularization in the control eye but none in the AS-ODN–injected eye at every time tested) (Figure 2). Four animals showed partial inhibition (iris neovascularization grades in the AS-ODN–injected eye lower than those in the control eye at some or all times tested) (Figure 3). No inhibition (equal grades of iris neovascularization in control and AS-ODN–injected eyes at all times tested) was seen in 2 of the 8 animals. None of the 8 animals had greater iris neovascularization in the AS-ODN–injected eye than in the control eye. Figure 4 shows the iris neovascularization grades from the 8 animals treated with 3μM ODN, comparing all measurements in AS-ODN–injected eyes with those in control ODN–injected eyes.
A total of 33 within-animal comparisons were made between treated and fellow control eyes in the 8 animals treated with 3μM ODN at points throughout the 14 days after treatment (Figure 5). For each individual animal, control-injected fellow eyes served as internal controls for AS-ODN–treated eyes. Iris neovascularization grades were lower in the AS-ODN–injected eye than in the control ODN–injected fellow eye in 19 (58%) of 33 comparisons, of which 16 (48%) were lower by 2 or more grades. In 14 comparisons (42%), the grades were equal in the AS-ODN– treated eye and the control ODN–injected eye. In none of the 33 comparisons was the grade of iris neovascularization higher in the AS-ODN–treated eye than in the control ODN–treated eye. Mixed-effects ordinal logistic regression showed that neovascularization grades were significantly lower in AS-ODN–injected eyes (2-tailed P = .006, MIXOR analysis24).
A total of 17 animals were tested across a range of ODN concentrations (0.1, 1.0, 3.0, 10.0, and 50.0μM). Statistical significance was found for the effect of ODN concentration (2-tailed P = .04, MIXOR analysis). The AS-ODN–treated eyes had lower iris neovascularization grades than the control eyes (P = .006, McNemar test); 12 of 17 animals had lower iris neovascularization grades in the AS-ODN–treated eye than in the control eye at 1 or more times, with no higher grades at any time. In comparison, only 1 of 17 animals had greater iris neovascularization grades in the AS-ODN–treated eye than in the control eye, with no lower grades at any time. For the binary outcome of iris neovascularization vs no iris neovascularization (grade ≥I vs grade 0), iris neovascularization was more likely to be absent in the AS-ODN–treated eye than in the control eye (odds ratio, 54; P < .001, mixed-effects logistic regression model).
An AS-ODN targeted against human VEGF was investigated in a simian model of laser retinal vein occlusion to determine intraocular stability and localization after intravitreal injection and to assess its effect on ocular angiogenesis. The results indicate that the AS-ODN was delivered to its target in the retina by intravitreal injection and that it reduced iris neovascularization in this primate model.
Localization of FITC-ODN was observed in the retina 2 to 3 days after intravitreal injection. The retina seemed to be accessible to ODN accumulation, including the ganglion cell layer, inner nuclear layer, outer plexiform layer, photoreceptor outer segments, and retinal pigment epithelium. Although autofluorescence has been observed in the retinal pigment epithelium layer, these results are in keeping with the findings reported in the rhesus monkey,22 rat,25 and mouse26 after intravitreal injection of AS-ODNs against VEGF; ODNs have been detected as long as 6 weeks after injection throughout the neurosensory retina, often preferentially localized to the retinal pigment epithelium. In vitro studies27 have demonstrated that VEGF is synthesized by human retinal pigment epithelium cells.
A therapeutic effect of the AS-ODN against VEGF was seen in this model of ocular angiogenesis; eyes injected with an AS-ODN exhibited decreased iris neovascularization compared with fellow eyes injected with control ODNs. Although some animals showed complete suppression of iris neovascularization with AS-ODN treatment, most showed a comparative reduction but not elimination of iris neovascularization in the treated eye. Previous in vivo studies of AS-ODN against VEGF in ocular angiogenesis have also found only a partial inhibitory effect. In a murine model of retinopathy of prematurity, AS-ODN against VEGF20 significantly reduced but did not eliminate VEGF and its effects; VEGF protein levels were reduced by 40% to 66%, and preretinal neovascular cell counts were decreased by 25% to 31%. Similarly, intravitreal injections of ODN against VEGF in models of laser-induced choroidal neovascularization in the rat21 and monkey22 only partially inhibited the formation of choroidal neovascularization. A variety of mechanisms can be postulated to explain the lack of complete inhibition of angiogenesis using antisense approaches. Phosphorothioate ODNs, as applied herein, may not provide total suppression of VEGF synthesis due to factors such as limited access or affinity for the target sequence site on VEGF messenger RNA complexes, nonspecific binding and nonspecific physiologic effects of either antisense or sense ODNs, and the possible instability of ODNs in the vitreous and retina. This final point seems unlikely because ODNs are stable for weeks in human vitreous (G.S.R., unpublished data, 1997). In addition, it is possible that alternative angiogenic pathways are unaffected or even up-regulated by blockade of VEGF messenger RNA translation. In these experiments, the effect of AS-ODNs against VEGF was assessed primarily by comparison of in vivo iris neovascularization as measured by serial fluorescein angiography. To definitively demonstrate that this effect is specifically VEGF dependent, further studies comparing measurements of VEGF message or protein are necessary.
After decades of research, the therapeutic potential of AS-ODNs remains largely unrealized. To overcome current barriers to the clinical efficacy of AS-ODNs, new approaches continue to be developed. The phosphorothioate ODNs used in these experiments represent a “first generation” of ODNs; newer generations of ODNs that incorporate novel chemical modifications are aimed at enhancing bioavailability, cellular uptake, endonuclease activation, and pharmacokinetics.28- 30 To date, the only approved application of antisense therapy is in ophthalmology, with fomivirsen for cytomegalovirus retinitis; the eye may offer an advantage in delivery to target tissues via intravitreal injection while minimizing nonspecific systemic actions.
Correspondence: Robert B. Bhisitkul, MD, PhD, Department of Ophthalmology, Beckman Vision Center, University of California San Francisco, 10 Kirkham St, K301, San Francisco, CA 94143 (firstname.lastname@example.org).
Submitted for Publication: October 30, 2003; final revision received July 7, 2004; accepted September 13, 2004.
Financial Disclosure: None.
Acknowledgment: Drs Bhisitkul and Robinson contributed equally to this study. Statistical consultation was provided by Peter Bacchetti, PhD, and Mark Segal, PhD, Department of Biostatistics, University of California San Francisco.