Widespread Expression of an Exogenous Gene in the Eye after Intravenous Administration

Abstract

purpose. Gene-targeting technology and tissue-specific gene promoters were used to produce widespread expression of an exogenous gene throughout the eye, including the retinal pigmented epithelium, after noninvasive intravenous administration of a nonviral plasmid formulation.

methods. An expression plasmid encoding bacterial β-galactosidase, under the influence of either the simian virus (SV)40 promoter or the glial fibrillary acidic protein (GFAP) gene promoter, was packaged in the interior of 85-nm pegylated immunoliposomes (PIL) targeted to transferrin receptor (TfR)-rich structures with the rat 8D3 monoclonal antibody (mAb) to the mouse TfR. Plasmid DNA was packaged in the 8D3-PIL and injected intravenously into adult female BALB/c mice at a dose of 5 μg DNA per mouse. The eyes were removed 48 hours later, and frozen sections were prepared for β-galactosidase histochemistry and rhodopsin, TfR, or GFAP immunocytochemistry.

results. There was diffuse expression of the SV40/β-galactosidase gene in the retinal pigmented epithelium. The cellular specificity of gene expression was influenced by the promoter used in the gene construct, evidenced by the fact that gene expression in the inner retina was induced with the GFAP promoter. The β-galactosidase gene was also widely expressed in the TfR-rich epithelial structures of the eye, including the ciliary body, the iris, the sebaceous glands of the tarsal plate, and the epithelium of the cornea. When the anti-TfR mAb on the PIL was replaced with the rat IgG isotype control, there was no gene expression in the eye.

conclusions. Widespread expression of an exogenous gene throughout the retina pigmented epithelium or other structures of the eye is possible with a noninvasive intravenous administration of a nonviral plasmid that is reformulated with gene-targeting technology. The cellular specificity of gene expression in the eye can be regulated with the use of cell-specific gene promoters.

Many disorders of the retina and other structures of the eye are potentially amenable to treatment with gene therapy. ¹ In the case of retinitis pigmentosa (RP), many of the gene defects that cause the disease are known. ^{2 3} The limiting factor in gene therapy for the eye is the gene delivery system. Both viral and nonviral cationic liposome systems have been used to target exogenous genes to the retina and other structures of the eye. Because neither the viral nor the cationic liposome gene delivery system crosses the blood–retinal barrier (BRB), it is necessary to inject the gene directly into either the vitreous cavity or the subretinal space. The exogenous gene is expressed at the local site of injection, with no gene expression in the retina at distances as little as 100 μm away from the injection site. ⁴ It has been estimated that only 6% of the human retina would be transduced after a single subretinal injection of an exogenous gene. ¹  

It would be advantageous to have a retinal gene-targeting system that transduces essentially 100% of the retina with the therapeutic gene. This would require targeting the exogenous gene through the retinal microvasculature comprising the BRB, which would allow for gene distribution to all parts of the retina. In this approach, the exogenous gene is administered noninvasively, by an intravenous or subcutaneous route, for example. This would be possible if a retinal gene-targeting technology were developed that is capable of delivering an exogenous gene across the two principle barriers that segregate the blood from the intracellular space of the retina: the BRB and the plasma membrane of cells within the retina. 

The widespread expression of an exogenous gene throughout the target organ after an intravenous (IV) injection is possible with gene-targeting technology that uses nonviral expression plasmids encapsulated in pegylated immunoliposomes (PIL). ⁵ The double-stranded, supercoiled plasmid DNA is encapsulated in the interior of a 75- to 100-nm liposome (Fig. 1A) . This encapsulation in the interior of the liposome renders the exogenous gene resistant to the ubiquitous endonucleases present in vivo. The liposome is stabilized for in vivo circulation and prolonged plasma residence time by conjugating to the surface of the liposome several thousand strands of 2000-dalton polyethylene glycol (PEG), designated PEG²⁰⁰⁰ (Fig. 1A) . A receptor-specific monoclonal antibody (mAb) is tethered to approximately 1% to 2% of the tips of the PEG²⁰⁰⁰ strands. ⁵ The mAb targets a receptor present on both the BRB and the plasma membrane of cells within the retina or other ocular structures. In the formulation used in the present studies for gene targeting to the mouse retina, the rat 8D3 mAb to the mouse transferrin receptor (TfR) was used. ⁶ The TfR is widely expressed at both the BRB and the plasma membrane of cells in the retina and other structures of the eye. ^{7 8 9 10 11}  

Once the exogenous gene is delivered to cells, the expression of the gene is regulated by the promoter that is inserted in the expression plasmid (Fig. 1A) . Prior work has shown that gene expression is observed in both brain and peripheral TfR-rich organs, when the widely active simian virus (SV)40 promoter is used to drive the expression of the exogenous gene. ^{5 12} Conversely, brain-specific expression of an exogenous gene after IV administration is possible with a combined use of the PIL gene-targeting technology and a brain-specific promoter, such as the glial fibrillary acidic protein (GFAP) gene promoter. ⁶ The GFAP gene is expressed in the normal retina in the ganglion cell layer (GCL). ¹³ Therefore, with the rat 8D3 mAb to the mouse TfR, we examined the expression of an exogenous gene, bacterial β-galactosidase, that was controlled by either the SV40 promoter or the human GFAP promoter and was packaged in the interior of a PIL targeted to the mouse retina. 

Materials and Methods

Materials

Adult female BALB/c albino mice (25–30 g) were purchased from Harlan (Indianapolis, IN). The pSV–β-galactosidase expression plasmid under the influence of the SV40 promoter was obtained from Promega (Madison, WI). The β-galactosidase staining kit was purchased from Invitrogen (San Diego, CA). The pGfa-lacZ β-galactosidase expression plasmid was provided by Jose Segovia of the Center of Investigation and Advanced Studies (San Pedro Zacatenco, Mexico). In this plasmid, the lacZ gene is driven by the human GFAP promoter, as described previously. ^{14 15} The 8D3 hybridoma line, secreting a rat IgG to the mouse TfR, ¹⁶ was obtained from Britta Engelhardt of the Max Planck Institute (Bad Nauheim, Germany), and the 8D3 mAb was purified as described previously. ⁶ The 1D4 mouse mAb against bovine rhodopsin ¹⁷ was obtained from Dean Bok of the UCLA School of Medicine (Los Angeles, CA). The immunodetection kit (Vector MOM), 3-amino-9-ethylcarbazole (AEC) substrate kit for peroxidase, and hematoxylin QS counterstain were purchased from Vector Laboratories (Burlingame, CA). Optimal cutting temperature compound (OCT; Tissue-Tek) was purchased from Sakura FineTek (Torrance, CA). The mouse mAb against porcine GFAP, pooled rat immunoglobulin G (IgG), and all other reagents were purchased from Sigma (St. Louis, MO). 

In Vivo Administration of Pegylated Immunoliposomes

The preparation of the 8D3-PIL carrying either the pSV–β-galactosidase (designated SV40/β-galactosidase) or the pGfa-lacZ (designated GFAP/β-galactosidase) has been described previously. ^{5 6} All animal procedures were approved by the UCLA Animal Research Committee and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Female BALB/c mice of 25 to 30 g body weight were anesthetized with intraperitoneal ketamine (50 mg/kg) and xylazine (4 mg/kg). Experimental animals (n = 8) were injected intravenously through the femoral vein with 8D3-PIL carrying either pSV-β-galactosidase plasmid DNA (n = 3) or the pGfa-lacZ plasmid DNA (n = 5) at a dose of 5 to 6 μg per mouse. For control studies, the plasmid DNA was encapsulated within PILs targeted with the rat IgG isotype control antibody, and mice (n = 5) were injected intravenously with 5 to 6 μg per mouse of plasmid DNA carried by the rat IgG-PIL. Prior work has shown that if the targeting mAb is replaced with an isotype control antibody that does not recognize the targeted receptor, then there is no gene expression in brain, liver, spleen, lung, or heart in either rats or mice. ^{5 6 12}  

Mice injected with either the 8D3-PIL or the rat IgG-PIL were killed at 48 hours after the single IV injection of the gene. Eyes were removed and frozen in OCT embedding medium on dry ice and stored at −70°C. Alternatively, some mice underwent perfusion fixation with 4% paraformaldehyde in situ before removal of the eyes. These eyes were immersion fixed in 4% paraformaldehyde, cryoprotected overnight in 30% sucrose, and frozen in OCT medium before preparation of frozen sections. Mouse kidney was also removed as a positive control for the β-galactosidase histochemistry. ^{6 12}  

β-Galactosidase Histochemistry

β-Galactosidase histochemistry was performed on frozen sections of eyes similar to prior work reported in brain tissue. ^{5 6 12} Horizontal frozen sections of 18-μm thickness were cut on a microtome cryostat (model HM505; Micron Instruments, San Diego, CA) and fixed with 0.5% glutaraldehyde in 0.01 M PBS (pH 7.4) for 5 minutes. After a wash in PBS, sections were incubated in 5-bromo-4-chloro-3-indoyl-β-d-galactoside (X-gal) staining solution (4 mM potassium ferricyanide, 4 mM potassium ferrocyanide, 2 mM MgCl₂, and 1 mg/mL X-gal[pH 7.4]) at 37°C overnight. After the staining, sections were briefly washed in distilled water and lightly counterstained with hematoxylin. Frozen sections of mouse kidney were also processed in the histochemistry assay. Unlike the eye, kidneys from control, uninjected rodents express endogenous β-galactosidase–like enzyme activity that gives a positive histochemical reaction at a neutral pH. ^{6 12} Therefore, including kidney sections in the β-galactosidase histochemistry assay serves as a positive internal control, particularly for eye specimens taken from rat IgG-PIL–injected animals that show no evidence of β-galactosidase gene expression. 

Immunohistochemistry

Immunohistochemistry for rhodopsin, GFAP, or mouse TfR was performed with the avidin biotin complex (ABC) immunoperoxidase method (Vector Laboratories). Frozen sections of the eyes from the control and experimental animals were fixed in 2% paraformaldehyde for 20 minutes at 4°C. Endogenous peroxidase was blocked with 0.3% H₂O₂ in 0.3% horse serum for 5 minutes. Nonspecific binding of proteins was blocked with mouse immunoglobulin blocking solution (MOM; Vector) for 1 hour, when a mouse primary antibody was used. Sections were then incubated in the primary antibody for 30 minutes at room temperature, and three primary antibodies were used: the 1D4 mouse anti-bovine rhodopsin mAb (5 μg/mL), the mouse anti-porcine GFAP mAb (8 μg/mL), and the rat 8D3 mAb to the mouse TfR (10 μg/mL). Control primary antibodies were either mouse or rat IgG at the same concentration. After a wash in PBS, sections were incubated in biotinylated secondary antibody for 10 minutes and then in ABC horseradish peroxidase (Vectastain Elite; Vector) for 5 minutes. After development with AEC, sections were lightly counterstained with hematoxylin and mounted with glycerol-gelatin. The thickness of the layers of the retina in different sections varied, owing to differences in sectioning angle of frozen, unfixed eyes. 

Results

β-Galactosidase histochemistry was performed on eye sections obtained from mice that were injected with the β-galactosidase plasmid encapsulated within the control rat IgG-PIL, and the histochemistry shows the absence of any endogenous β-galactosidase enzyme product (Fig. 1B) . The β-galactosidase histochemistry of the eye removed 48 hours after a single IV injection of the 8D3-PIL carrying the β-galactosidase expression plasmid driven by the SV40 promoter is shown in Figure 1C . This demonstrates diffuse expression of the exogenous gene in the outer retina corresponding to the retinal pigmented epithelium (RPE), as well as gene expression in the ciliary body and iris (Fig. 1C) . 

Ocular expression of the β-galactosidase gene was examined at 48 hours after a single IV injection of the GFAP/β-galactosidase plasmid encapsulated within the 8D3-PIL. The gene was expressed in both the inner retina and the RPE, as shown in Figure 1D . The β-galactosidase gene driven by the GFAP promoter was also expressed in the conjunctival epithelium and in the epithelium of the ciliary body and the iris (Fig. 1D) . 

The reactivity of the rat 8D3 mAb to the mouse TfR with ocular structures was examined by immunocytochemistry using either the 8D3 mAb (Fig. 1E) or the control rat IgG (Fig. 1F) . Immunoreactive TfR was present in the ganglion cell layer (GCL), the outer plexiform layer (OPL), the inner segments (IS) of the photoreceptor cells, and the RPE (Fig. 1E) . 

High-magnification microscopy of the retina removed from control mice shows the absence of the endogenous β-galactosidase histochemical product (Fig. 2A) . After the IV administration of the SV40/galactosidase plasmid encapsulated within the 8D3-PIL, the predominant site of gene expression was the RPE (Fig. 2B) . There was minimal gene expression in the inner retina after the administration of the SV40/galactosidase plasmid (Fig. 2B) . 

The administration of the GFAP/β-galactosidase plasmid resulted in increased gene expression in the inner retina corresponding to the GCL and the inner nuclear layer (INL; Fig. 2C ). GFAP immunocytochemistry was performed on parallel sections (data not shown), and there was overlap of the GFAP-immunopositive cells in the inner retina and the β-galactosidase histochemical product (Fig. 2C) . 

Higher-magnification views of the outer retina are shown in Figures 2D and 2H for β-galactosidase histochemistry and for rhodopsin immunocytochemistry, respectively. The outer segments (OS) and the IS of the photoreceptor cells, which were immunopositive and immunonegative, respectively, for rhodopsin are visible in Figure 2H . Based on comparison of the β-galactosidase histochemistry (Fig. 2D) and the rhodopsin immunocytochemistry (Fig. 2H) , it appears that exogenous gene expression in the outer retina was restricted to the RPE when either the SV40 or GFAP promoter was used. 

The exogenous gene was also expressed in the epithelium of the ciliary body and iris (Fig. 2E) . The tarsal plate sebaceous gland epithelium, with the characteristic foamy appearance, expressed the exogenous gene (Fig. 2F) . The epithelium of the cornea demonstrated expression of the transgene (Fig. 2G) . 

Discussion

The results of these studies lead to the following conclusions: First, it is possible to target an exogenous gene throughout the entire RPE and other structures of the eye with a simple IV injection, if gene-targeting technology is used. Second, the expression of the exogenous gene in different cell layers of the retina can be modulated with cell-specific promoters, as shown by the selective expression of the β-galactosidase gene in the GCL after administration of the expression plasmid driven by the GFAP promoter (Fig. 2C) . The use of tissue-specific promoters can eliminate gene expression in peripheral tissues, as shown in the mouse. ⁶ Gene expression in the eye was demonstrated in the present studies at a single time point, 48 hours, after IV injection (Figs. 1 2) , although prior work with either Southern blot or histochemical analysis for either brain or liver has demonstrated that expression of the exogenous gene is detectable for at least 6 days after a single IV administration of the gene. ¹²  

The nonviral PIL formulation (Fig. 1A) used in the present investigations can be contrasted with conventional nonviral cationic liposome-DNA mixtures. The latter form layered structures of DNA and lipid in aqueous solutions of low ionic strength. ¹⁸ However, cationic liposomes rapidly aggregate into structures of more than 1 μm in physiologic saline, ¹⁹ and cationic liposomes are inhibited by serum proteins. ²⁰ Cationic liposomes aggregate in blood, and more than 99% of organ gene expression is found in the pulmonary vascular bed after IV administration of cationic liposomes. ^{21 22} There is no gene expression in the brain after IV injection of cationic liposomes. ²³ In contrast to the cationic liposomes, with which the DNA is exposed to blood constituents, the plasmid DNA is packaged in the interior of the PIL (Fig. 1A) . The PIL does not aggregate in saline solution and has prolonged plasma residence times, owing to the use of pegylation technology. ⁵ Pegylation of the liposome is accomplished by conjugation of several thousand strands of PEG²⁰⁰⁰ to the surface of the liposome (Fig. 1A) . A pegylated liposome, per se, would be inert in vivo and would not be specifically targeted to cells in vivo after IV administration. ²⁴ When a nonspecific IgG molecule is conjugated to the tips of the PEG strands, in lieu of the targeting mAb (Fig. 1A) , and injected intravenously in rats or mice, there is no expression of the exogenous gene in brain or peripheral tissues that express the TfR. ^{5 6 12} In the present studies, there was no gene expression in the mouse eye when the PIL was targeted with a nonspecific isotype control IgG (Fig. 1B) . However, the PIL was delivered to tissues based on the tissue-specific expression of the targeted receptor (Fig. 1A 1R) , if a targeting mAb was conjugated to the PEG strands. An mAb to the TfR was used to target the PIL carrying the DNA to cells in vivo. Gene expression in brain cells is possible with the PIL gene-targeting technology, because the TfR is expressed on both the brain capillary endothelium, which forms the blood–brain barrier (BBB), ²⁵ and on the plasma membrane of brain cells. ²⁶  

The anti-TfR mAb can also be used to target exogenous genes to the eye, owing to the abundant expression of the TfR in ocular structures. ⁸ The TfR is expressed on the plasma membrane of multiple cells of the eye, including cells of the GCL, the INL, the RPE, and the IS of the photoreceptor cells. ^{7 8 9} This pattern of retinal TfR expression was confirmed in the present studies with immunocytochemistry with the 8D3 mAb to the mouse TfR (Fig. 1E) , which is the same mAb used to target exogenous genes to the eye (Figs. 1C 1D) . The TfR is also expressed in the epithelium of the conjunctiva, the iris, the ciliary body, and the cornea. ^{10 11} The pattern of gene expression in the eye (Figs. 1C 1D) parallels the diffuse expression of the TfR in the eye. Owing to the dual expression of the TfR on both the BRB and the plasma membrane of ocular cells, the PIL can deliver the exogenous gene across the different cellular barriers separating the blood from the intracellular compartment of cells within the eye. Once inside the retinal cells, the expression of the exogenous gene is influenced by the tissue specificity of the promoter. ⁶ The GFAP/β-galactosidase gene was expressed in epithelial structures such as the iris and the ciliary body, indicating the GFAP promoter is expressed in these structures (Figs. 1C 1D) . This observation correlates with results in prior studies showing GFAP expression in both the iris ²⁷ and the ciliary body ²⁸ of the rodent eye. The GFAP/β-galactosidase gene was expressed in cells within the inner retina (Fig. 2C) , which appeared to be Müller cell processes located primarily within the GCL and, to a lesser extent, in the INL (Fig. 2C) . This pattern of GFAP/β-galactosidase gene expression correlates with the cellular distribution of GFAP expression in the normal retina, which includes Müller cell processes within the GCL, and to a lesser extent in the IPL and INL. ²⁹ There was minimal expression of the β-galactosidase gene in the GCL when the plasmid was under the control of the SV40 promoter (Fig. 2B) . The correlation of the β-galactosidase histochemistry (Fig. 2D) and rhodopsin immunocytochemistry (Fig. 2H) suggest gene expression is located primarily in the RPE, with minimal expression in the photoreceptor cells. However, it may be possible to induce expression of an exogenous gene in the photoreceptor segments, if an opsin gene promoter is inserted in the expression plasmid, as has been shown previously with viral delivery systems. ³⁰  

In summary, the present studies demonstrate that it is possible to obtain diffuse expression of an exogenous gene throughout the RPE after an IV injection of a nonviral formulation. With the expression plasmid used in the present studies, the exogenous gene is expressed within cells as an extrachromosomal episome, and gene expression is necessarily transient. The half-life of expression of an exogenous gene driven by the SV40 promoter and delivered with the PIL system to organs in the rat is approximately 6 days. ¹² Therefore, an exogenous gene therapy would have to be administered repeatedly at periodic intervals, based on the persistence of the transgene. Repeated administration of the PIL should be possible, because the only immunogenic component of the formulation is the mAb, and the immunogenicity of the mAb in humans can be reduced or eliminated with genetic engineering and the production of humanized antibodies. 

Alternatively, it may be desirable to have long-term expression of the transgene in the retina, and this can be achieved with viral vectors, such as adeno-associated virus (AAV) or lentivirus. ^{30 31} However, both of these viral vectors result in stable integration into the host genome, and the long-term effects of such viral integration in humans is unknown. The risk of such viral integration may not be acceptable until it is demonstrated that the therapeutic gene has the intended effect on visual function in humans. The transient and reversible expression of a therapeutic gene in the retina is possible with the PIL gene-targeting technology. The therapeutic gene can be targeted across the BRB in humans with a genetically engineered humanized mAb. If the transient effect on visual function in humans of the therapeutic gene is favorable, then the decision could be made to transduce the retina long-term with viral vectors that integrate into the host genome. 

The PIL gene-targeting technology described in the present studies could be used in humans by replacement of the anti-TfR mAb with an mAb to the human insulin receptor (HIR). ³² A genetically engineered HIR mAb readily crosses the primate BBB in vivo and binds to the human BBB in vitro, owing to the high level of expression of the insulin receptor at the BBB. ³³ Because the insulin receptor is also expressed at the BRB, ³⁴ the genetically engineered HIR mAb may also target PIL-encapsulated genes into the retina in humans. The HIR mAb is particularly suited for targeting therapeutic genes, because this antibody targets the nucleus and yields very high levels of expression of the therapeutic gene in human cells. ³⁵  

 Submitted for publication November 27, 2001; revised April 16, 2002; accepted May 20, 2002.

 Commercial relationships policy: P (WMP); N (all others).

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

(A) A nonviral, supercoiled, double-stranded plasmid DNA driven by a cell-specific promoter was packaged in the interior of a 75- to 100-nm liposome. The surface of the liposome was labeled with several thousand strands of PEG²⁰⁰⁰, and the tips of 1% to 2% of the PEG strands were conjugated with a receptor (R)-specific targeting mAb. The mAb triggered the receptor-mediated transcytosis of the pegylated immunoliposome across the blood-retinal barrier, and triggered the receptor-mediated endocytosis of the PIL into cells of the eye expressing the targeted receptor. The PIL was prepared with either the rat 8D3 mAb to the mouse TfR or the rat IgG isotype control antibody. The control rat IgG antibody did not target the TfR. (B) β-Galactosidase histochemistry in an eye of a mouse injected with the GFAP/β-galactosidase plasmid encapsulated in PILs targeted with nonspecific rat IgG showed no expression of the β-galactosidase gene. (C) β-Galactosidase histochemistry in an eye obtained 48 hours after IV injection of the SV40/β-galactosidase plasmid encapsulated within the PIL targeted with the 8D3 mAb. There was diffuse expression of the transgene in the RPE and in the iris and ciliary body. (D) β-Galactosidase histochemistry in the mouse eye obtained 48 hours after IV injection of the GFAP/β-galactosidase plasmid encapsulated within the PIL targeted with the 8D3 mAb. There is diffuse expression of the transgene in the inner retina and at the RPE and in the iris and conjunctival epithelium. (E) Immunocytochemistry of mouse retina with the rat 8D3 mAb to the mouse TfR. (F) Immunocytochemistry of mouse retina with the rat IgG control antibody. All specimens were counterstained with hematoxylin. (E, F) Immunopositive TfR is reddish brown and the counterstain is purple. Scale bars: (C, D) 106 μm; (E, F) 27 μm.

Figure 2.

 

β-Galactosidase histochemistry: (A) The eye of a control, uninjected mouse. (B) The retina of a mouse injected with the SV40/β-galactosidase plasmid encapsulated within the 8D3-PIL showing diffuse expression in the RPE with minimal gene expression in the inner retina or GCL. (C) An eye of a mouse injected with the GFAP/β-galactosidase expression plasmid encapsulated within the 8D3-PIL, showing gene expression in the RPE, the GCL of the inner retina, and at the border of the IPL and the INL. (D) The outer retina in a high-magnification view. A comparison of the distribution of rhodopsin, shown in (H) indicates the transgene is expressed in the RPE. (E) Expression of the transgene in the epithelium of the iris and ciliary body. (F) Expression of the transgene in the foamy epithelia of the sebaceous gland of the conjunctival tarsal plate. (G) Expression of the transgene in the epithelium of the cornea. (H) Rhodopsin immunocytochemistry. All specimens were counterstained with hematoxylin. The specimens used for histochemistry (A–G) were fixed in 0.5% glutaraldehyde for 5 minutes, and that used for immunocytochemistry (H) was fixed in 2% paraformaldehyde for 20 minutes. Scale bars: (A, B, E) 37 μm; (C) 60 μm; (D, F, G, H) 15 μm.