Gene Therapy (2006) 13, 752–760 & 2006 Nature Publishing Group All rights reserved 0969-7128/06 $30.00 www.nature.com/gt ORIGINAL ARTICLE Proliferating endothelial cell-specific expression of IGF-I receptor ribozyme inhibits retinal neovascularization

LC Shaw1, H Pan1, A Afzal1, SL Calzi1, PE Spoerri1, SM Sullivan2 and MB Grant1 1Department of Pharmacology and Therapeutics, University of Florida, Gainesville, FL, USA and 2Department of Pharmaceutics, College of Medicine, University of Florida, Gainesville, FL, USA

Insulin-like growth factor-I (IGF-I) and its receptor (IGF-IR) are drives expression of the IGF-IR ribozyme. This promoter essential for normal ocular development and are expressed limited retinal expression of the reporter gene to proliferating in numerous ocular cell types including lens epithelial cells, endothelial cells in two mouse models of proliferative retinal pigment epithelial cells, Mu¨ller cells and endothelial retinopathy. In addition, expression of the IGF-IR ribozyme cells. Endothelial cell proliferation is a common feature of by this promoter inhibited aberrant retinal angiogenesis in proliferative retinopathies and involves abnormal growth of both models while preserving normal vessels. These results blood vessels within and on the surface of the retina. In an demonstrate the feasibility of IGF-IR ribozyme expression in a effort to inhibit the formation of these aberrant blood vessels, selective manner for safer treatment of abnormal angiogen- we cloned an IGF-IR ribozyme into an expression vector that esis associated with retinopathy. limits expression of the ribozyme to proliferating endothelial Gene Therapy (2006) 13, 752–760. doi:10.1038/sj.gt.3302718; cells. An endothelin enhancer and Cdc6 promoter chimera published online 26 January 2006

Keywords: retinopathy; IGF-I; ribozyme; proliferating; endothelial; promoter

Introduction and the maturing avascular retina becomes hypoxic resulting in VEGF accumulation in the vitreous. These Insulin-like growth factor-I (IGF-I) and its receptor (IGF- authors went on to hypothesize that as IGF-I increases to IR) are key proteins involved in proliferation, differentia- a critical level, retinal neovascularization is triggered. tion and malignant transformation and are expressed Their data indicate that serum IGF-I levels in premature in numerous ocular cell types, including retinal endo- infants can predict which infants will develop ROP. thelial cells (RECs), lens epithelial cells,1 retinal pigment The role of IGF-I has not been well established in epithelial cells,2 cone photoreceptor cells3 and Mu¨ ller normal retinal vessel development in humans. Retinal cells.4 IGF-I and its receptor are required for normal vessel morphology was evaluated in patients with genetic ocular development including the vasculature. IGF-I defects of the IGF-I axis. Patients with low serum levels of has been shown to play a key role in retinopathy of IGF-I during and after normal retinal vessel growth had prematurity (ROP).5 significantly less retinal vascularization, as evidenced by Hellstrom et al.6 showed that lack of IGF-I in knockout lower numbers of vascular branching points, compared mice prevents normal retinal vascular growth, despite with the reference group of normal controls. This is the the presence of vascular endothelial growth factor first study to provide genetic evidence for a role of IGF-I (VEGF). Inadequate levels of IGF-I in vitro result in loss system in retinal vascularization in humans.6 of VEGF-induced activation of protein kinase B (Akt), a The IGF-I/IGF-IR axis has also been shown to kinase critical for endothelial cell survival.5 These studies influence the progression of proliferative diabetic retino- emphasize the salient role of IGF-IR in orchestrating the pathy (PDR). Pivotal studies by Merimee et al.7 showed function of other growth factors in the eye. the first association of serum IGF-I levels with disease Hellstrom et al.6 postulated that if sufficient serum severity. They identified a group of diabetic patients with levels of IGF-I are maintained after birth in premature markedly accelerated neovascularization. These patients infants, normal vessel development occurs and ROP does had very high serum IGF-I levels, compared to diabetic not develop. However, when serum IGF-I, which reflects patients with slower advancing disease.7 This correlation retinal IGF-I, is persistently low, vessels cease to grow, was corroborated by others.8–10 Chantelau et al.11 found that elevated IGF-I serum levels preceded PDR in patients undergoing intense insulin treatment. Correspondence: Dr MB Grant, Department of Pharmacology and However, vitreous levels of IGF-I may be more Therapeutics, University of Florida, Box 100267, 1600 SW Archer Rd, relevant to the initiation of vascular pathology. We Gainesville, FL 32610, USA. E-mail: [email protected]fl.edu examined IGF-I levels in vitreous and demonstrated Received 26 May 2005; revised 13 October 2005; accepted 30 a threefold increase in levels of IGF-I in diabetics with November 2005; published online 26 January 2006 PDR compared to non-diabetic individuals.12 This was Cell-specific ribozyme expression in mouse retina LC Shaw et al 753 later confirmed by Meyer-Schwickerath.13 In vitro studies HRECs and fibroblasts were transfected with lucifer- also established the presence of IGF-IR on human retinal ase expression plasmids driven by the proliferating endothelial cells (HRECs) and correlated receptor num- endothelium promoter (ETe/Cdc6p). The luciferase ber with growth rates of cells, with rapidly proliferating assays are shown in Figure 2 and Table 1 and they demo- 14 cells showing the highest receptor number. nstrate that the 4 Â ETe and the 7 Â ETe promoters yielded These studies support a critical role for IGF-I in 200-fold higher expression of luciferase in the endothelial normal development, suggesting that serum and tissue cells compared with fibroblasts. A 12-fold higher expres- levels of IGF-I and IGF-IR surface expression in cells sion of luciferase was observed in endothelial cells must be carefully regulated for normal development. compared to fibroblasts transfected with the construct In contrast, ocular pathologies are associated with dys- containing the ribozyme. Expression of the constructs regulated levels of both IGF-I and IGF-IR.7 Thus, we containing the IGF-IR hammerhead/hairpin ribozymes sought to target the IGF-IR exclusively in only prolifer- was lower than constructs containing the promoter and ating endothelial cells within the retinal vasculature. This luciferase alone. This reduction was attributed to loss of approach provides a unique antiangiogenic strategy free the polyA signal as a result of the hairpin ribozyme of toxicity to resident endothelial cells and other non- cleaving its target at the 30 end of the IGF-IR ribozyme. endothelial cell types in the retina but deleterious only to proliferating endothelial cells. The ETe/Cdc6p enhancer/promoter expresses luciferase specifically in RECs in vivo Results The intravitreal injection of pLUC1298 into the OIR mouse model (Figure 3) or the adult mouse model of

The ETe/Cdc6p enhancer/promoter expresses laser-induced retinopathy (Figure 4) resulted in expres- luciferase specifically in cultured endothelial cells sion of luciferase (green) only in the endothelial cells Figure 1 shows the expression cassette maps of the of blood vessels. Staining of retina with rhodamine plasmids. The proliferating endothelial promoters are conjugated to agglutinin identified the retinal vascula- composed of either a 4 Â (1297) or a 7 Â (1298) 46-mer ture (red). When luciferase is superimposed with the rhodamine, it appears yellow. Immunohistochemical multimerized endothelin enhancer (ETe) upstream of a analysis demonstrated that luciferase expression was human Cdc6 promoter (Cdc6p). Consensus intron/exon intervening sequence (IVS) 12 and IVS 8 exist for mRNA restricted to the vessels as there was only luciferase processing. The IGF-IR ribozyme and hairpin processing surrounding the lumens perfused with rhodamine- ribozyme were cloned into the 30 end of the luciferase labeled dextran. gene directly after the luciferase stop codon. The IGF-IR ribozyme and hairpin processing ribozyme have been previously described.15 The IGF-IR ribozyme is a 34 nt ribozyme consisting of two 6 nt targeting arms and an internal 4 bp stem with a tetra-loop. The hairpin processing ribozyme is cloned 6 bp downstream of the IGF-IR ribozyme. A hairpin cleavage site recognized by the hairpin processing ribozyme is inserted between the IGF-IR ribozyme and the hairpin processing ribozyme. Self-cleavage by the hairpin processing ribozyme at this cleavage site generates a specific 30 end for the IGF-IR ribozyme and results in a transcript with five additional nucleotides at the 30 end of the IGF-IR ribozyme. This configuration limits the number of nucleotides at the 30 end of the IGF-IR ribozyme that could interfere with proper folding of this ribozyme with its target. Figure 2 The ETe/Cdc6p limits expression of luciferase to HRECs. Expression of luciferase by pLUC1297 and pLUC1298 in HRECs is approximately 200-fold greater than in human fibroblasts. Expres- sion of luciferase by pLUC1297 and pLUC1298 in HRECs is approximately 12-fold greater than expression of luciferase by pLUC1297HHHP and pLUC1298HHHP in HRECs owing to transcript instability resulting from the lack of a polyA signal.

Table 1 Expression of luciferase in human fibroblasts and HRECs in relative fluorescent units

Fibroblasts HRECs, donor 1 HRECs, donor 2 Figure 1 The ET /Cdc6 expression cassettes. In the pLUC1297 and e p 7 4 7 6 7 5 pLUC1298 plasmids, the ETe/Cdc6p drives the expression of pLUC1297 1.4 0.29 Â 10 2.98 0.16 Â 10 5.9 2.5 Â 10 7 4 7 6 7 5 luciferase and contains a polyA signal. In the pLUC1297HHHP and pLUC1297 1.9 0.58 Â 10 3.59 0.16 Â 10 8.6 0.62 Â 10 7 3 7 5 7 4 pLUC1298HHHP plasmids, the ET /Cdc6 drives the expression of pLUC1297HHHP 1.7 1.1 Â10 3.3 0.01 Â10 6.9 0.03 Â 10 e p 7 2 7 5 7 4 luciferase and the IGF-IR ribozyme and a hairpin ribozyme that self- pLUC1298HHHP 8.3 1.7 Â 10 3.0 0.12 Â 10 1.2 0.006 Â 10 7 2 7 2 7 2 cleaves the transcript at the 30 end of the IGF-IR ribozyme. These two NT 3.4 0.15 Â 10 3.6 0.17 Â 10 3.2 0.45 Â 10 plasmids contain no polyA signal. The 1297 plasmids contain four ET enhancers and the 1298 plasmids contain seven ET enhancers. HRECs, human retinal endothelial cells; NT, non-transfected cells.

Gene Therapy Cell-specific ribozyme expression in mouse retina LC Shaw et al 754

Figure 3 Expression of luciferase by pLUC1298 is limited to the rapidly dividing endothelial cells of the retinal vasculature in the OIR mouse model. The vasculature has been stained with rhodamine-agglutinin (red). Luciferase is visualized by fluorescently labeled secondary antibody (green). Colocalization of rhodamine and luciferase results in a yellow color. Magnification:  200 (a),  200 (b) and  400 (c). A z-axis projection of panel c was produced from the stack of confocal images using Image J and can be found in the Supplementary Movie.

The ETe/Cdc6p-driven expression of the IGF-IR ribozyme reduces abnormal pre-retinal neovascularization in the OIR mouse model Figure 5 shows that expression of the IGF-IR ribozyme from pLUC1298HHHP reduces abnormal angiogenesis in the injected eye (Figure 5a) compared to the uninjected eye (Figure 5b). The density of blood vessels, visualized with rhodamine-agglutinin staining, is reduced in the injected eye (Figure 5a). Figure 5c–g demonstrates neovascular tufts, where expression of luciferase, shown as green in the images (Figure 5d–h), is limited to the rapidly proliferating endothelial cells (blue) (Figure 5e, g and i), within pre-retinal blood vessels. Figure 5c shows the rhodamine-stained vessels. Figure 5d demonstrates luciferase expression in endothelial cells of this vessel and Figure 5e is the combined image of Figure 5c and d. Figure 5f shows proliferating cells within this same vessel. Figure 5g is a composite of the three fluorescent markers and demonstrates proliferating endothelial cells as white cells. Figure 5h and i represents rhodamine- stained retinal vessels from the contralateral uninjected eye. Figure 5h shows the rhodamine and FITC channels, whereas Figure 5i shows a tuft of proliferating endothe- lial cells with superimposed images from the rhodamine and the blue channel. Figure 6 represents a summary of the quantitative analysis of pre-retinal nuclei per cross-section and demon- strates that the pLUC1297HHHP and pLUC1298HHHP plasmids (both expressing the IGF-IR ribozyme) reduced pre-retinal neovascularization by 48712% (P ¼ 1.4 Â 10À15) and 54711% (P ¼ 4.3 Â 10À8), respectively, when compared to the uninjected eyes. The pGEHHHP1298 plasmid (expressing the IGF-IR ribozyme but not luciferase) reduced pre-retinal neovascularization by 5978% (P ¼ 5.7 Â 10À28). As expected, the parent plasmids, Figure 4 Expression of luciferase by pLUC1298 is limited to areas pLUC1297 and pLUC1298 not containing the ribozyme of rapidly dividing endothelial cells in the retinal vasculature of constructs, showed no significant difference in the number mice undergoing the adult model of laser-induced neovasculariza- of pre-retinal nuclei between the injected and uninjected tion. Confocal image taken of flatmounted retina where blood eyes (P ¼ 0.5 and 0.4, respectively). vessels have been labeled with endothelial cell-specific agglutinin conjugated to rhodamine (red) is shown. Luciferase is visualized by The ET /Cdc6 -driven expression of the IGF-IR fluorescently (FITC) labeled secondary antibody (green). Colocali- e p zation of agglutinin and luciferase results in a yellow color. ribozyme reduces abnormal pre-retinal Magnification: Â 400. A z-axis projection of this image was neovascularization in the adult model of produced from the stack of confocal images using Image J and laser-induced retinopathy can be found in the Supplementary Movie. In addition, Image J was used to produce a three dimensional image from the same Figure 7 shows retinas from adult mice undergoing stack of confocal images and can be found in the Supplementary laser-induced branch vessel occlusion to stimulate Movie. neovascularization. The retinas were perfused with

Gene Therapy Cell-specific ribozyme expression in mouse retina LC Shaw et al 755

Figure 5 The ETe/Cdc6p-driven expression of the IGF-IR ribozyme reduces abnormal pre-retinal neovascularization in the OIR mouse model. Images were taken of retinal flatmounts from mice undergoing the OIR model. The retinal vasculature has been delineated with rhodamine-agglutinin. Luciferase is visualized by an FITC-labeled secondary antibody (green). Panels f, g and i visualize rapidly proliferating vasculature with antibodies to PCNA (blue). Colocalization of rhodamine and luciferase is yellow (seen best in panel e), rhodamine and PCNA is violet (seen best in panel i) and rhodamine, luciferase and PCNA is white (seen in panel g only). (a) Retina from an eye injected with the pLCHHHP1298 plasmid shows that luciferase expression is limited to the vasculature ( Â 200). (b) Retina from the uninjected eye of the same mouse as in panel a ( Â 100). (c–g) High magnification ( Â 400) of a retinal vascular tuft from an eye injected with pLCHHHP1298. (c) Red channel showing vascular endothelial cells stained with rhodamine-agglutinin. (d) Green channel showing luciferase-expressing cells. (e) Red and green channels showing overlap of endothelial cells expressing luciferase (yellow). Luciferase- expressing cells are associated with endothelial cells only. (f) Blue channel showing PCNA-stained nucleus of proliferating endothelial cells. (g) Red, green and blue channels showing that luciferase is being expressed in the proliferating endothelial cells of the vasculature (white). (h, i) Uninjected control eyes visualized with images from both red and green channels (h) and red and blue channels (i).

rhodamine-labeled dextran in 4% paraformaldehyde to identify vessels and demonstrate leakage. Non-lasered eyes from age-matched mice (Figure 7a) show no leaky vessels. Figure 7b, in contrast, shows a representative image from a lasered eye injected with the control plasmid, pLUC1298, with numerous areas of vascular leakage (arrows). Figure 7c shows a representative image from a lasered eye injected with pLUC1298HHHP (expressing the IGF-IR ribozyme) showing significantly fewer areas with vasculature leakage when compared to Figure 7b.

Discussion

Figure 6 The ETe/Cdc6p-driven expression of the IGF-IR ribozyme We have previously shown reduction of IGF-IR mRNA reduces abnormal pre-retinal neovascularization in the OIR mouse and protein expression in HRECs using an IGF-IR- model. Summary of the quantitative analysis of the number of pre- retinal nuclei per cross-section is shown. The pLUC1297HHHP, specific ribozyme and that this same ribozyme inhibited pLUC1298HHHP and pGEHHHP1298 plasmids equally reduce pre-retinal neovascularization in the OIR mouse pre-retinal neovascularization. model.15 A CMV enhancer/chicken b-actin promoter

Gene Therapy Cell-specific ribozyme expression in mouse retina LC Shaw et al 756

Figure 7 The ETe/Cdc6p-driven expression of the IGF-IR ribozyme reduces abnormal pre-retinal neovascularization in the adult model of laser-induced retinopathy. Retinal flatmounts from mice were perfused with rhodamine-labeled dextran to visualize leaky retinal vessels. (a) Non-lasered eye showing the normal retinal vasculature of the eye. (b) Laser-treated eye intravitreally injected with pLUC1298 showing many areas containing leaky vessels (arrows). (c) Laser-treated eye intravitreally injected with pLUC1298HHHP showing a significant reduction in leaky vessels when compared to panel c. Magnification: Â 50.

drove the expression of the IGF-IR ribozyme in these Our in vivo studies of luciferase expression from studies. This promiscuous enhancer/promoter resulted pLUC1298 demonstrate that expression is limited to in ribozyme expression in many cell types and in rapidly proliferating endothelial cells of the retinal both quiescent and proliferating cells of the mouse vasculature, as typified by the neovascularization retina. This enhancer/promoter did not restrict expres- observed in the OIR mouse model (Figures 3 and 5) and sion of the IGF-IR ribozyme to areas of neovascu- in the adult mouse model of laser-induced retinopathy larization only. (Figure 4). Coexpression of luciferase and the IGF-IR Because IGF-I and its receptor play a major role in ribozyme from pLUC1298HHHP showed exclusive vascular development of both mouse and human eyes,16 expression of luciferase in the rapidly proliferating the indiscriminant loss of IGF-IR could result in altered endothelial cells of the retinal vasculature in the OIR vascular development and propagate the ischemia mouse model (Figure 5f, g and i). observed in ROP infants or OIR mice. Therefore, we In addition, expression of the IGF-IR ribozyme from developed a promoter that would restrict expression to this particular plasmid also demonstrated that the IGF-IR rapidly proliferating endothelium for the purpose of ribozyme was able to inhibit the pre-retinal neo- minimizing any adverse impact on normal ocular vascularization produced in the OIR mouse model development. (Figure 6) without deleteriously influencing the deeper

The ETe/Cdc6p enhancer/promoter is a composite retinal vasculature. Pre-retinal neovascularization was promoter, which, as described previously,17 shows reduced by as much as 54711% after injection of selective gene expression in proliferating mouse en- pLUC1298HHHP. This is comparable to levels of inhibi- dothelial cells in vitro. It is composed of a multimerized tion of neovascularization when expression of IGF-IR endothelin enhancer combined with a cell cycle-regu- ribozyme is driven by the CVM enhancer/b-actin lated Cdc6 promoter. Interestingly, ligation of the IGF-IR promoter (6576%).15 The IGF-IR ribozyme can inhibit hammerhead ribozyme and the hairpin ribozyme caused pre-retinal neovascularization produced in the adult the endothelin enhancers to be deleted in several strains model of laser-induced neovascularization. This is of bacteria, DH5a, Stable 2s and Top Tens. While initially apparent from the dramatic reduction in leaky vessels the placement of luciferase was used solely to validate in the ribozyme-treated lasered eye compared to the cells expressing the construct, the placement of the untreated lasered eye (Figure 7). Therefore, our in vivo hammerhead and hairpin ribozymes at the 30 end of results demonstrate that we can limit expression of the luciferase prevented the deletion of the endothelin potentially therapeutic ribozyme to pathological prolif- enhancer. erating endothelial cells of the retina while not affecting Figure 2 demonstrates strong expression of luciferase healthy quiescent endothelium. from the pLUC1297 and pLUC1298 clones in HRECs Another remarkable observation is the duration of and essentially no expression of luciferase in human expression of the luciferase reporter gene. Systemic fibroblasts. Luciferase expression from plasmids expres- administration of plasmid DNA alone by hydrodynamic sing the IGF-IR ribozyme (pLUC1297HHHP and administration results in initial high levels of expression pLUC1298HHHP) was lower than with parent plasmids. 24 h after injection, which decrease to 7% of the peak This would be expected, as these transcripts have no value by day 10.18,19 Delivery of plasmid alone in the OIR polyA tails. Therefore, it is expected that these transcripts model and formulated plasmid in the adult laser model would have a higher turnover rate than transcripts with demonstrates a duration of expression that exceeds the a polyA tail. duration of expression observed in other organs follow-

Previously, we showed that the multimerized ETe/ ing systemic administration. Luciferase expression

Cdc6p promoter restricted expression to endothelial and ribozyme activity are observed up to 17 days cells in vitro and in vivo.17 Figure 2 confirms these following administration in the OIR model and up to in vitro results. However, we wanted to examine expression 21 days in the adult model. These time points represent properties specifically in RECs. Our results support the required termination of the experiment and thus specific expression of the IGF-IR ribozyme in vitro to expression may be even longer. This expression also HRECs and in vivo to retinal endothelium of both mouse elicits the biological response of preventing aberrant models. ocular neovascularization.

Gene Therapy Cell-specific ribozyme expression in mouse retina LC Shaw et al 757 Although we specifically targeted the IGF-1 system in strated that, in a variety of experimental settings, this study, other key factors critical to development of interference with IGF-IR function results in inhibition retinal neovascularization include VEGF. VEGF binds in of cancer cell proliferation,47 survival,32 anchorage- a distinct pattern to three structurally related tyrosine independent growth in vitro,48 inhibition of tumor kinase receptors. Of these, VEGFR1 and VEGFR2 are growth and formation of metastasis in vivo36,41,43,44,49 considered to be the most important for regulation of and sensitization of cancer cells to various chemother- neovascularization. VEGFR2 has long been considered apeutic and radiation regimens.46 Experimental evidence the major inducer of angiogenesis. VEGF binding to clearly supports pursuit of IGF-IR as a target in oncology. VEGFR2 stimulates the phosphorylation of a variety of Antagonistic antibodies and small molecular mass proteins in endothelial cells including phospholipase kinase inhibitors hold therapeutic promise. Drug C-g, PI-3 kinase, Ras GTPase activating protein and the discovery approaches to target IGF-IR also include src family of proteins, as well as the generation of nitric antisense oligonucleotides and recombinant IGF-binding oxide. VEGF induces endothelial cell growth by activat- proteins.50,51 ing the RAF–MEK–ERK pathway. VEGF, similar to IGF-I, In summary, we have focused on inhibiting ocular mediates endothelial cell survival by activating the PI-3 angiogenesis using an IGF-IR ribozyme and proliferating kinase–Akt pathways. The published clinical trials and endothelial cell-specific promoters. IGF-I plays a promi- FDA approval of the anti-VEGF monoclonal antibody nent role in ROP and PDR and orchestrates the function bevacizumab (Avastin, Genentech) for the treatment of of VEGF in the eye.5,52 The clinical significance of IGF-I in colorectal cancer marked a milestone for antiangiogen- ocular pathology is appreciated, as serum IGF-I levels in esis therapy. In addition to colorectal cancer, VEGF premature infants can predict which infants will develop inhibitors are being investigated in the treatment of renal ROP. Targeted reduction of IGF-IR in rapidly proliferat- cell carcinoma, head and neck carcinoma, lung cancer, ing vasculature rather than in quiescent endothelium breast cancer, prostate cancer and a variety of hemato- will be critical to treatment of ROP infants as well as logical malignancies.20 New therapies directed at block- other disorders where the preservation of quiescent ing VEGF receptors are currently being developed.21 vasculature is essential. Our results demonstrate a proof Angiogenic factors such as IGF-1 and VEGF are of concept for using plasmid DNA encoding a ribozyme balanced by antiangiogenic growth factors in the normal against IGF-IR in a selective manner for safe treatment of eye. An imbalance of these factors induces growth of neovascularization associated with diseases such as PDR aberrant new blood vessels. The presence of angiogenic and ROP. growth factors such as IGF-I and VEGF in the retinas and vitreous of patients with vascular retinopathies is well documented.5,22–24 In addition to proangiogenic factors, it Materials and methods has recently become apparent that a variety of endogen- ous antiangiogenic factors like pigment epithelium- Cell culture derived factor (PEDF), endostatin and thrombospondin Human eyes were obtained from National Disease may contribute to vascular quiescence. PEDF, one of the Research Interchange within 36 h of death (n ¼ 6 donors). endogenous antiangiogenic factors found in the eye, was HRECs were prepared and maintained as previously first purified from the conditioned medium of human described.53 Cells in passages 3–4 and ascertained RPE cells25 and is a member of the serine protease positive for acetylated LDL were used for this study. inhibitor (serpin) family with neuroprotective, neuro- Human skin fibroblasts (FG 152 cell line) were a kind gift trophic25 and antiangiogenic activities.26 PEDF has of Dr Dietmeir Siemen at the University of Florida. potential as a therapeutic agent for retinal and choroidal diseases and has been extensively tested in animal Expression plasmid construction models27–29 and is currently being evaluated in a clinical pLUC1297 and pLUC1298 were isolated from DH5a trial.30 The counterbalance of VEGF and PEDF is Escherichia coli (Qiagen Mega Kit). Plasmids were supported by the previous demonstrations that either digested with XbaI (New England Biolabs) at base pair inhibition of the VEGF system or overexpression of PEDF 2036 and XhoI (New England Biolabs) at base pair 2254, inhibits neovascularization.28,31 which represented proliferating endothelial cell-specific Epidemiological and mechanistic evidence link IGF-IR promoters.17 The proliferating endothelial promoter activation and signaling to beyond simply pathological is composed of a 4 Â (1297) and 7 Â (1298) 46-mer angiogenesis and also to tumor biology. Blocking IGF-IR multimerized endothelin enhancer upstream of a human function in tumor cells may result in therapeutic benefits Cdc6 promoter. There is a consensus intron/exon for patients with cancer.32–35 Increased levels of this IVS 12 and IVS 8 for mRNA processing. The plasmid receptor and its cognate ligands have been observed in a pGEHHHP1298 was derived from pLUC1298HHHP by variety of solid human tumors.36–39 IGF-IR function is removal of the luciferase reporter gene. Synthetic implicated in the hallmarks of cancer – self-sufficiency in oligomers for the anti-IGF receptor hammerhead/pro- growth signals, evasion from apoptosis, tissue invasion cessing hairpin ribozyme15 (Midland Scientific, Midland, and metastasis, as well as angiogenesis. While we used TX, USA) containing Xba1 and Xho1 complementary ribozymes in this study, many other approaches have overhang sequences were ligated into the digested been used to study IGF-IR function including dominant- backbone. StblII cells were transformed with the ligation negative mutants, kinase-defective mutants, antisense reaction and selected with kanamycin. DNA from oligonucleotides, antisense expression plasmids, IGF- bacterial clones was isolated and analyzed for a unique binding proteins, soluble forms of the receptor, antag- MnII restriction site. For construction of pLUC1297 onistic and/or neutralizing antibodies40–44 or small HHHP and pLUC1298HHHP, the IGF-IR ribozyme/ molecule kinase inhibitors.45,46 These studies demon- hammerhead processing ribozyme was cloned into the

Gene Therapy Cell-specific ribozyme expression in mouse retina LC Shaw et al 758 30 end of the luciferase gene directly after the luciferase IACUC of the University of Florida. C57BL6/J timed stop codon for both 4 Â and 7 Â multimerized endothe- pregnant mice and adult mice were obtained from lin enhancers with the Cdc6 promoter. A glycerol master Jackson Laboratories (Bar Harbor, ME, USA). The mice cell bank was created, plasmid DNA was isolated (Giga were housed in the University of Florida Health Science Prep Kit, Qiagen) and sequenced for both the ribozyme Center Animal Resources Facilities.

insert and the ETe/Cdc6 promoter. Intraocular injection into the mouse model Plasmid formulation for adult mouse eye gene transfer of oxygen-induced retinopathy Plasmid DNA was purified from StblII bacteria (Invitro- gen) and condensed with a cationic lipid (Lipid 89 In the neonatal mouse model of oxygen-induced retino- pathy, 7-day-old mice were placed with their nursing Genzyme Corporation) in 40% ethanol/5% dextrose. The 55 ratio of cationic lipid to plasmid DNA nucleotide was 3:1 dams in a 75% oxygen atmosphere for 5 days. Upon (mol/mol). The cationic lipid was mixed with a helper return to normal air, these mice develop retinal neovas- lipid mixture composed of lysophosphatidylcholine:mono- cularization, with peak development occurring 5 days acylglycerol:free fatty acid (1:4:2; mol/mol/mol). The after their return to normoxia. In this study, the mouse m lengths of the acyl chain helper lipids were 18:2 and the pups received a 0.5 l intravitreal injection of plasmid ratio of cationic lipid to helper lipid mixture was 10:90. (2 mg/ml) OD on postnatal day one (P1). After the fifth Ethanol was removed by dialysis against PBS and the day following return to normoxia (P17), the animals were final DNA concentration was 0.5 mg/ml. killed and the eyes removed and fixed in 4% parafor- maldehyde and embedded in paraffin. Three hundred Transfection of HRECs with lipofectamine serial sections (6 mm) were cut sagittally through the HRECs grown to confluence on 100 mm plates were cornea parallel to the optic disc. Every thirtieth section transfected with the IGF-IR ribozyme plasmid constructs was placed on slides and stained with hematoxylin– using Lipofectamine 2000 as described.54 A 728 ml eosin. Ten sections from each eye were scored in a portion of Opti-MEM I was combined with 52 mlof masked fashion using light microscopy by counting Lipofectamine 2000 (Invitrogen, Rockville, MD, USA) endothelial cell nuclei extending beyond the inner limiting membrane into the vitreous, as described and allowed to stand at room temperature for 5 min. 55 Then, 780 ml of Opti-MEM I was combined with 13 mg previously. The efficacy of treatment with each plasmid of DNA and allowed to stand at room temperature for was then calculated as the percent average nuclei per 5 min. These two solutions were then combined and section in the injected eye versus the uninjected complexed for 20 min at room temperature. While the contralateral eye. Selected animals were perfused with solutions were complexing, the cultures were removed 4% paraformaldehyde. For these animals, their retinas from the incubator and the medium from each Petri dish were removed for immunohistochemistry. was aspirated and replaced with fresh medium. These plates were then returned to the incubator until the time Adult model of laser-induced retinopathy of addition of the complexes. After 20 min of complex- In this model, laser photocoagulation of retinal vessels ing, the Opti-MEM I/DNA/Lipofectamine 2000 was was used to stimulate the angiogenesis within and on the added to the cultures at 1560 ml per Petri dish. The retina as described by Grant et al.56 Six-to eight-week-old culture medium was replaced after 24 h. The cells were C57BL/6 mice (Jackson Laboratories) were intravitreally harvested after 72 h of incubation for further analysis. injected with 2 ml of AAV-VEGF. Expression of VEGF from this construct stimulates the formation of retinal Luciferase assay blood vessels when followed by laser injury. The HRECs were isolated as previously described and plated contralateral eye served as a control. One month onto 100 mm dishes. Human skin fibroblasts (FG 152 cell following intraocular AAV-VEGF, the mice underwent line) were also plated onto 100 mm dishes. Both cell laser occlusion of large venous vessels in the retina. types were transfected with the plasmids (pLUC1297 Briefly, mice were anesthetized with an intraperitoneal and pLUC1298) using Lipofectamine 2000 reagent injection of ketamine (70 mg/kg body weight) and (Invitrogen) according to the manufacturer’s protocol. xylazine (14 mg/kg body weight). An argon green laser At 72 h post-transfection, the cells were washed twice system (HGM Corporation, Salt Lake City, UT, USA) was with PBS (Cellgro) and harvested with 100 mlof used for retinal vessel photocoagulation with the aid of a luciferase cell culture lysis reagent (Promega) containing 78-diopter lens. The blue-green argon laser (wavelength protease inhibitor cocktail (Sigma). BCA protein assay 488–514 nm) was applied to selected venous sites next to was performed (Pierce) on the harvested cells. Luciferase the optic nerve. Venous occlusion was accomplished assay system (Promega) was used to measure the using laser parameters of 1 s duration, 50 mm spot size luminescence in the samples according to the manufac- and an average of 600 mW intensity with an average of turer’s protocol. The relative luminescence units were 44 burns per retina. Following laser injury, the lasered read on a luminometer with a 2 s delay followed by a eye received a 2 ml injection of plasmid containing the 10 s reading time. The samples were normalized to IGF-IR ribozymes. Plasmids were formulated with a protein levels. cationic lipid formulation to facilitate transfection of the adult mouse endothelium, as the adult mouse eye has Animals low transfection efficiencies with plasmid DNA alone. All animals were treated in accordance with the ARVO The pLUC1298HHHP plasmid was compared to the statement for the use of animals in Ophthalmic and pLUC1298 plasmid with same promoter driving lucifer- Vision Research and with the Guide for the Care and Use of ase expression without the ribozyme. The formulated Laboratory Animals. All protocols were approved by the plasmid had a final DNA concentration of 0.3 mg/ml.

Gene Therapy Cell-specific ribozyme expression in mouse retina LC Shaw et al 759 Three weeks following laser injury and plasmid injec- retinal neovascularization by insulin-like growth factor-1 recep- tion, each mouse was anesthetized and perfused through tor. Nat Med 1999; 5: 1390–1395. the left ventricle with 3 ml of 4% paraformaldehyde. The 6 Hellstrom A, Carlsson B, Niklasson A, Segnestam K, Bogus- eyes were enucleated and the retinas removed and zewski M, de Lacerda L et al. IGF-I is critical for normal flatmounted for histopathological studies. vascularization of the human retina. J Clin Endocrinol Metab 2002; 87: 3413–3416. Histopathological studies 7 Merimee TJ, Zapf J, Froesch ER. Insulin-like growth factors: 1 studies in diabetics with and without retinopathy. N Engl J Med The retinas were incubated overnight at 4 C in 1:1000 1983; 309: 527–530. rhodamine-conjugated agglutinin in 10 mM HEPES, 8 Grant MB, Russell B, Fitzgerald C, Merimee TJ. Insulin-like 150 mM NaCl and 0.1% Tween 20 (Buffer F) (pH 7.5). growth factors in vitreous: studies in control and diabetic After 24 h, the retinas were washed in 10 mM HEPES subjects with neovascularization. Diabetes 1986; 35: 416–420. and 150 mM NaCl (Buffer B) at 41C, the samples were 9 Dills DG, Moss SE, Klein R, Klein BE. Association of elevated permeabilized in Buffer F for 24 h at 41C and then IGF-I levels with increased retinopathy in late-onset diabetes. washed twice in Buffer B. Incubation in ice-cold Diabetes 1991; 40: 1725–1730. methanol followed for 2 h on ice. Thereafter, the retinas 10 Hyer SL, Sharp PS, Brooks RA, Burrin JM, Kohner EM. A two- were incubated in 0.7 M NaOH for 30 min and washed year follow-up study of serum insulinlike growth factor-I in twice in PBS. Then, the retinas were incubated overnight diabetics with retinopathy. Metabolism 1989; 38: 586–589. at 41C in a 50/50 solution of Mouse on Mouse (MOM) 11 Chantelau E, Eggert H, Seppel T, Schonau E, Althaus C. Mouse Ig blocking reagent (Vector Laboratories Inc., Elevation of serum IGF-1 precedes proliferative diabetic Burlingame, CA, USA) and 10% normal goat serum retinopathy in Mauriac’s syndrome. Br J Ophthalmol 1997; 81: (Sigma, Saint Louis, MO, USA) in PBS (blocking buffer) 169–170. to block nonspecific binding. The following day, the 12 Grant MB, Ellis EA, Caballero S. The role of insulin-like growth samples were incubated in MOM diluent for 10 min and factor-I in proliferative diabetic retinopathy: a modulator of free then in 20 mg/ml rabbit anti-PCNA IgG (Novus Biologi- radical derived oxidant generation. In: Shigeta Y, King GL (eds). cals, Littleton, CO, USA) and 20 mg/ml mouse anti- Recent Advances in Endothelial Cell Dysfunction in Diabetes. Churchill Livingstone: Osaka, Japan, 1994, pp 265–280. luciferase IgG1 (Zymed Laboratories, San Francisco, CA, USA) in blocking buffer overnight at 41C. After two 13 Meyer-Schwickerath R, Pfeiffer A, Blum WF, Freyberger H, Klein M, Losche C et al. Vitreous levels of the insulin-like washes in PBS, samples were incubated in MOM growth factors I and II, and the insulin-like growth factor biotinylated anti-mouse IgG reagent for 2 h at room binding proteins 2 and 3, increase in neovascular eye disease. temperature, washed again and incubated with Fluor- Studies in nondiabetic and diabetic subjects. J Clin Invest 1993; escein Avidin DCS plus 1:400 dilution goat anti-rabbit 92: 2620–2625. 1 CY5 (Chemicon International Inc.) overnight at 4 C. They 14 Grant MB, Mames RN, Fitzgerald C, Ellis EA, Caballero S, were then washed and mounted for confocal analysis. Chegini N et al. Insulin-like growth factor I as an angiogenic The retinas were visualized using an MRC-1024 Confocal agent. In vivo and in vitro studies. Ann NY Acad Sci 1993; 692: Laser Scanning Microscope at the Optical Microscopy 230–242. Facility at the University of Florida (Gainesville). ImageJ 15 Shaw LC, Afzal A, Lewin AS, Timmers AM, Spoerri PE, Grant (ImageJ 1.32j, Wayne Rasband, National Institutes of MB. Decreased expression of the insulin-like growth factor 1 Health, USA, http://rsb.info.nih.gove/ij/) was used for receptor by ribozyme cleavage. Invest Ophthalmol Vis Sci 2003; 44: analysis of the confocal images. 4105–4113. 16 Hellstrom A, Engstrom E, Hard AL, Albertsson-Wikland K, Statistical analysis Carlsson B, Niklasson A et al. Postnatal serum insulin-like Data were analyzed by one-way analysis of variance growth factor I deficiency is associated with retinopathy of prematurity and other complications of premature birth. followed by Bonferroni t-test, with either hyperoxia- or Pediatrics 2003; 112: 1016–1020. vehicle-treated used as a determinant, as appropriate. 7 17 Szymanski P, Anwer K, Sullivan S. Prolliferating endothelium Results are reported as mean s.e. Po0.05 was consid- specific gene expression. J Gene Med 2006 (in press). ered to be significant. 18 Anwer K, Kao G, Proctor B, Rolland A, Sullivan S. Optimization of cationic lipid/DNA complexes for systemic gene transfer to tumor lesions. J Drug Target 2000; 8: 125–135. References 19 Anwer K, Meaney C, Kao G, Hussain N, Shelvin R, Earls RM et al. Cationic lipid-based delivery system for systemic cancer gene 1 Walker JL, Zhang L, Zhou J, Woolkalis MJ, Menko AS. Role for therapy. Cancer Gene Ther 2000; 7: 1156–1164. alpha 6 integrin during lens development: evidence for signaling 20 Rhee J, Hoff PM. Angiogenesis inhibitors in the treatment of through IGF-1R and ERK. Dev Dyn 2002; 223: 273–284. cancer. Expert Opin Pharmacother 2005; 6: 1701–1711. 2 Rosenthal R, Wohlleben H, Malek G, Schlichting L, Thieme H, 21 Wakelee HA, Schiller JH. Targeting angiogenesis with vascular Bowes Rickman C et al. Insulin-like growth factor-1 contributes endothelial growth factor receptor small-molecule inhibitors: to neovascularization in age-related macular degeneration. novel agents with potential in lung cancer. Clin Lung Cancer Biochem Biophys Res Commun 2004; 323: 1203–1208. 2005; 7 (Suppl 1): S31–S38. 3 Zygar CA, Colbert S, Yang D, Fernald RD. IGF-1 produced by 22 Adamis AP, Miller JW, Bernal MT, DJ DA, Folkman J, Yeo TK et cone photoreceptors regulates rod progenitor proliferation in the al. Increased vascular endothelial growth factor levels in the teleost retina. Brain Res Dev Brain Res 2005; 154: 91–100. vitreous of eyes with proliferative diabetic retinopathy. Am J 4 Li F, Cao W, Steinberg RH, LaVail MM. Basic FGF-induced Ophthalmol 1994; 118: 445–450. down-regulation of IGF-I mRNA in cultured rat Muller cells. Exp 23 Simo R, Lecube A, Segura RM, Garcia Arumi J, Hernandez C. Eye Res 1999; 68: 19–27. Free insulin growth factor-I and vascular endothelial growth 5 Smith LE, Shen W, Perruzzi C, Soker S, Kinose F, Xu X et al. factor in the vitreous fluid of patients with proliferative diabetic Regulation of vascular endothelial growth factor-dependent retinopathy. Am J Ophthalmol 2002; 134: 376–382.

Gene Therapy Cell-specific ribozyme expression in mouse retina LC Shaw et al 760 24 Freyberger H, Brocker M, Yakut H, Hammer J, Effert R, receptor antibody A12 on androgen-dependent and androgen- Schifferdecker E et al. Increased levels of platelet-derived growth independent xenograft human prostate tumors. Clin Cancer Res factor in vitreous fluid of patients with proliferative diabetic 2005; 11: 3065–3074. retinopathy. Exp Clin Endocrinol Diabetes 2000; 108: 106–109. 42 Cohen BD, Baker DA, Soderstrom C, Tkalcevic G, Rossi AM, 25 Tombran-Tink J, Chader GG, Johnson LV. PEDF: a pigment Miller PE et al. Combination therapy enhances the inhibition of epithelium-derived factor with potent neuronal differentiative tumor growth with the fully human anti-type 1 insulin-like activity. Exp Eye Res 1991; 53: 411–414. growth factor receptor monoclonal antibody CP-751,871. Clin 26 Kommonen B, Dawson WW, Parmer R. Pigment epithelial Cancer Res 2005; 11: 2063–2073. function in canine retina. Am J Vet Res 1991; 52: 1341–1344. 43 Wang Y, Malkowski M, Hailey J, Turek-Etienne T, Tripodi T, 27 Mori M, Ghyselinck NB, Chambon P, Mark M. Systematic Pachter JA. Screening for small molecule inhibitors of insulin- immunolocalization of retinoid receptors in developing and like growth factor receptor (IGF-1R) kinase: comparison of adult mouse eyes. Invest Ophthalmol Vis Sci 2001; 42: 1312–1318. homogeneous time-resolved fluorescence and 33P-ATP plate 28 Mori K, Gehlbach P, Ando A, Dyer G, Lipinsky E, Chaudhry AG assay formats. J Exp Ther Oncol 2004; 4: 111–119. et al. Retina-specific expression of PDGF-B versus PDGF-A: 44 Ludwig DL, Witte L, Hicklin DJ, Prewett M, Bassi R, Burtrum D vascular versus nonvascular proliferative retinopathy. Invest et al. Conservation of receptor antagonist anti-tumor activity by Ophthalmol Vis Sci 2002; 43: 2001–2006. epidermal growth factor receptor antibody expressed in trans- 29 Gehlbach P, Demetriades AM, Yamamoto S, Deering T, Duh EJ, genic corn seed. Hum Antibodies 2004; 13: 81–90. Yang HS et al. Periocular injection of an adenoviral vector 45 Li W, Favelyukis S, Yang J, Zeng Y, Yu J, Gangjee A et al. encoding pigment epithelium-derived factor inhibits choroidal Inhibition of insulin-like growth factor I receptor autophos- neovascularization. Gene Therapy 2003; 10: 637–646. phorylation by novel 6-5 ring-fused compounds. Biochem 30 Rasmussen H, Chu KW, Campochiaro P, Gehlbach PL, Haller JA, Pharmacol 2004; 68: 145–154. Handa JT et al. Clinical protocol. An open-label, phase I, single 46 Warshamana-Greene GS, Litz J, Buchdunger E, Hofmann F, administration, dose-escalation study of ADGVPEDF.11D (AD- Garcia-Echeverria C, Krystal GW. The insulin-like growth factor- PEDF) in neovascular age-related macular degeneration (AMD). I (IGF-I) receptor kinase inhibitor NVP-ADW742, in combination Hum Gene Ther 2001; 12: 2029–2032. with STI571, delineates a spectrum of dependence of small cell 31 Krzystolik MG, Afshari MA, Adamis AP, Gaudreault J, lung cancer on IGF-I and stem cell factor signaling. Mol Cancer Gragoudas ES, Michaud NA et al. Prevention of experimental Ther 2004; 3: 527–535. choroidal neovascularization with intravitreal anti-vascular 47 Byron SA, Yee D. Potential therapeutic strategies to interrupt endothelial growth factor antibody fragment. Arch Ophthalmol insulin-like growth factor signaling in breast cancer. Semin Oncol 2002; 120: 338–346. 2003; 30: 125–132. 32 Hofmann F, Garcia-Echeverria C. Blocking the insulin-like 48 Vasilcanu D, Girnita A, Girnita L, Vasilcanu R, Axelson M, growth factor-I receptor as a strategy for targeting cancer. Drug Larsson O. The cyclolignan PPP induces activation loop-specific Discov Today 2005; 10: 1041–1047. inhibition of tyrosine phosphorylation of the insulin-like growth 33 Wang Y, Sun Y. Insulin-like growth factor receptor-1 as an anti- factor-1 receptor. Link to the phosphatidyl inositol-3 kinase/Akt cancer target: blocking transformation and inducing apoptosis. apoptotic pathway. Oncogene 2004; 23: 7854–7862. Curr Cancer Drug Targets 2002; 2: 191–207. 49 Ye JJ, Liang SJ, Guo N, Li SL, Wu AM, Giannini S et al. Combined 34 Baserga R. The contradictions of the insulin-like growth factor 1 effects of tamoxifen and a chimeric humanized single chain receptor. Oncogene 2000; 19: 5574–5581. antibody against the type I IGF receptor on breast tumor growth 35 Baserga R. The IGF-I receptor in cancer research. Exp Cell Res in vivo. Horm Metab Res 2003; 35: 836–842. 1999; 253: 1–6. 50 Pollak MN, Schernhammer ES, Hankinson SE. Insulin-like 36 Arteaga CL, Osborne CK. Growth inhibition of human breast growth factors and neoplasia. Nat Rev Cancer 2004; 4: 505–518. cancer cells in vitro with an antibody against the type I 51 Zhang H, Yee D. The therapeutic potential of agents targeting somatomedin receptor. Cancer Res 1989; 49: 6237–6241. the type I insulin-like growth factor receptor. Expert Opin Investig 37 Khalili K, Del Valle L, Wang JY, Darbinian N, Lassak A, Safak M Drugs 2004; 13: 1569–1577. et al. T-antigen of human polyomavirus JC cooperates with 52 Smith LE, Kopchick JJ, Chen W, Knapp J, Kinose F, Daley D et al. IGF-IR signaling system in cerebellar tumors of the childhood- Essential role of growth hormone in ischemia-induced retinal medulloblastomas. Anticancer Res 2003; 23: 2035–2041. neovascularization. Science 1997; 276: 1706–1709. 38 Min Y, Adachi Y, Yamamoto H, Ito H, Itoh F, Lee CT et al. Genetic 53 Grant MG, Guay C. Plasminogen activator production by blockade of the insulin-like growth factor-I receptor: a promising human retinal endothelial cells of nondiabetic and diabetic strategy for human pancreatic cancer. Cancer Res 2003; 63: origin. Invest Ophthalmol Vis Sci 1991; 32: 53–64. 6432–6441. 54 Ohki EC, Tilkins ML, Ciccarone VC, Price PJ. Improving the 39 Furstenberger G, Morant R, Bolliger B, Senn HJ. Serum levels of transfection efficiency of post-mitotic neurons. J Neurosci IGF-1 and IGFBP-3 during adjuvant chemotherapy for primary Methods 2001; 112: 95–99. breast cancer. Breast 2003; 12 (Suppl 1): S31. 55 Smith LE, Wesolowski E, McLellan A, Kostyk SK, D’Amato R, 40 Maloney EK, McLaughlin JL, Dagdigian NE, Garrett LM, Sullivan R et al. Oxygen-induced retinopathy in the mouse. Connors KM, Zhou XM et al. An anti-insulin-like growth factor Invest Ophthalmol Vis Sci 1994; 35: 101–111. I receptor antibody that is a potent inhibitor of cancer cell 56 Grant MB, May WS, Caballero S, Brown GA, Guthrie SM, proliferation. Cancer Res 2003; 63: 5073–5083. Mames RN et al. Adult hematopoietic stem cells provide 41 Wu JD, Odman A, Higgins LM, Haugk K, Vessella R, Ludwig DL functional hemangioblast activity during retinal neovasculariza- et al. In vivo effects of the human type I insulin-like growth factor tion. Nat Med 2002; 8: 607–612.

Supplementary Information accompanies the paper on the Gene Therapy website (http://www.nature.com/gt).

Gene Therapy