¹Institute of Ophthalmology, University College London, UK

²Institute of Child Health, University College London, UK

Correspondence to: RR Ali, Institute of Ophthalmology, University College London, UK

We have evaluated the transduction profiles of an HIV-based lentiviral vector delivered regionally to ocular tissues in vivo. Following subretinal injection, a green fluorescent protein (GFP) reporter gene was efficiently and stably expressed in retinal pigment epithelial (RPE) cells. Limited transduction of adjacent photoreceptors occurred in newborn mice, but was inefficient in adult animals. Injection of the vector into the anterior chamber resulted in efficient and stable transduction of corneal endothelial cells. Efficient in vivo gene transfer into cells of the corneal endothelium and retinal pigment epithelium by lentiviral vectors may therefore offer a valuable approach to the treatment of disorders of the cornea and outer retina.

Gene Therapy 2001 8, 1665-1668.

gene transfer; lentivirus; cornea; retina

The eye has several advantages as a target organ for gene transfer. It is readily accessible for the delivery of vectors and for examination in vivo by slit-lamp biomicroscopy and ophthalmoscopy. The eye comprises a range of tissues including epithelia (cornea, iris, ciliary body, lens), muscle (ciliary body), neural crest (corneal endothelium), neurons (retina) and endothelium (vasculature). These tissues are well defined and anatomically compartmentalised enabling the precise delivery of vector suspensions into target areas by microsurgical techniques. The blood-retina and blood-aqueous barriers may help to maintain the concentration of vectors in target areas such that transduction efficiency is optimised. Adenoviral (Ad) and adeno-associated virus (AAV)-based vector systems have been evaluated in some detail, and show distinct spatial and temporal patterns of transduction. For example, AAV vectors (serotype 2) mediate efficient and stable transduction of both retinal pigment epithelium (RPE) cells and photoreceptors without evidence of direct toxicity,^1,2 but transduce tissues in the anterior of the eye with low efficiency. In contrast, Ad-mediated transduction of corneal endothelial cells and RPE is transient and associated with significant immunological response.³

Lentiviruses are attractive vectors for gene transfer because they have the potential to stably transduce non-dividing cells, and are deleted for all potentially pro-inflammatory viral components. In ocular tissues, lentiviral vectors have been shown to mediate efficient and sustained transgene expression in human corneal endothelial cells in vitro and in situ.⁴ Similarly, transduction of both RPE and photoreceptors has been reported after subretinal injection of an HIV-based lentiviral vector.^5,6 We have performed a systematic evaluation of a VSV-G-pseudotyped HIV-1-based lentiviral vector delivered to specific ocular sites in order to further define the potential roles of lentiviral vectors in gene transfer to the eye.

VSV-G pseudotyped vectors were produced by transient transfection of three plasmids into 293T cells, as has been described previously.⁷ The transfer vector used in this study was generated by modification of a basic self-inactivating vector pHR'SIN-CE (which itself incorporates a CMV promoter and enhanced green fluorescent protein (eGFP) reporter gene¹¹) with the woodchuck post-transcriptional regulatory element (WPRE), and the central polypurine tract cis-acting sequence (cPPT) and central termination sequence (CTS) of HIV-1. The WPRE was first rescued from pBlueScript-WPRE-B11 (gift from Thomas J Hope, Salk Institute, La Jolla, CA, USA) and inserted downstream of the eGFP stop codon in pBlueScript containing the eGFP open reading frame. Subsequently, pHR'SIN-CEW was constructed by substitution of the BamHI-XhoI eGFP cassette of pHR'SIN-CE with eGFP-WPRE. As described previously,⁸ the 178 bp fragment, encompassing the central polypurine tract, was amplified by PCR from pCMVR8.91 (gift from Pierre Charneau, Insitut Pasteur, Paris, France) and inserted upstream of the expression cassette to create the final vector pHR'SIN-cPPT-CEW (Demaison et al, submitted). For production of recombinant virus, a total of 10⁷ 293T cells were seeded in one 150-cm² flask overnight before transfection. Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% foetal calf serum (FCS), penicillin (100 IU/ml), and streptomycin (100 g/ml) in a 5% CO₂ incubator, and the medium was changed 2 h before transfection. A total of 100 g of plasmid DNA was used for the transfection of one flask: 17.5 g of the envelope plasmid, 32.5 g of packaging plasmid, and 50 g of transfer vector plasmid were pre-complexed with 0.25 mm PEI (22 kDa) in 10 ml Optimem at room temperature for 15 min. The DNA plus PEI complexes were then added to the cells. After 4 h incubation at 37°C in a 5% CO₂ incubator, the medium was replaced by fresh DMEM supplemented with 10% FCS. Virus particles were concentrated 20- to 100-fold by ultracentrifugation at 50 000 g for 90 min at 4°C. The pellet was resuspended in serum-free X-VIVO10 medium (BioWhittaker, Walkersville, MD, USA) and kept at -80°C until use. Vector titres were determined by transduction of HeLa cells, and for the studies described here, were approximately 2 ´ 10⁸ transducing units per ml.

C57Bl-6J mice used for this study were managed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Mice were anaesthetised with intraperitoneal injection of 0.2 ml Hypnorm (Janssen Pharmaceutical, Oxford, USA) and Hypnovel (Roche, Welwyn Garden City, UK) mixed 1:1:6 with PBS. The palpebral fissure was opened if necessary with a No.11 scalpel blade and the pupil dilated with topical 1% Tropicamide (1% Mydriacyl, Alcon Labs, Watford, UK). The injection procedure was performed under direct vision by means of an operating microscope using a 5 mm 34-gauge needle mounted on a 10 l Hamilton syringe. For anterior chamber injections the needle tip was advanced through the peripheral cornea and directed into the anterior chamber. For intravitreal and subretinal injections the fundus was visualised by means of a contact lens system consisting of a drop of 1% hypromellose solution on the cornea, covered with a glass coverslip. For intravitreal injections the needle tip was advanced through the sclera 1 mm posterior to the corneoscleral limbus into the vitreous cavity. For subretinal injections the needle tip was advanced through the sclera at the equator into the subretinal space, injection of viral suspension causing a localised retinal detachment as previously described.¹ Approximately 2 l of the same viral suspension (2 ´ 10⁸ transducing units/ml) was injected into the anterior chamber (12 eyes), vitreous cavity (six eyes) and subretinal space (six eyes) of adult mice. Approximately 1 l of virus supension was also injected into the subretinal space of 5-day-old mouse pups (six eyes).

Injection of vector suspension into the anterior chamber resulted in expression of GFP in corneal endothelial cells. GFP expression was established by 7 days after injection (four eyes) and typically extended across approximately two-thirds of the corneal diameter Figure 1Expression of GFP was observed predominantly in corneal endothelial cells and was also present in keratocytes at sites of needle trauma. Although GFP expression was most prominent in corneal endothelial cells, at high magnification GFP fluorescence was also observed in cells of the trabecular meshwork in some sections. GFP fluorescence was not observed in corneal epithelial cells, iris pigment epithelium or ciliary epithelium. The efficiency and pattern of GFP expression in corneal endothelium was unchanged at 6 weeks (four eyes) and 12 weeks (four eyes) after injection. No clinical signs of toxicity were seen on biomicroscopy. There was no evidence of an inflammatory cell infiltrate on histological sections of transfected corneas stained with haematoxylin and eosin and no stromal or epithelial oedema suggestive of endothelial failure (data not shown). Six weeks following anterior chamber injection of vector suspension, fluorescence of expressed corneal GFP was imaged in vivo by means of confocal laser scanning ophthalmoscopy (cSLO, four eyes) Figure 2This technique offers a valuable non-invasive means of monitoring the time-course of transgene expression across the corneal endothelium in an individual animal in vivo and is a sensitive tool which can provide detailed anatomical information for the diagnosis and management of corneal disorders in human patients.^9,10 GFP fluorescence was detected in an irregular distribution across the central cornea and consistently in a circular distribution towards the corneal periphery.

Injection of vector suspension into the subretinal space of adult mice resulted in efficient GFP expression by cells of the retinal pigment epithelium across approximately 50% of its area. Expression was established at 1 week after injection (three eyes) and persisted for at least 6 weeks (three eyes) Figure 3GFP expression in photoreceptor cells was limited to the immediate area of needle trauma at the injection site in adult mice. GFP expression in photoreceptors of 5-day-old pups, 1 week (two eyes) and 2 weeks (two eyes) after subretinal injection, extended beyond the immediate area of the injection site trauma, but was of low efficiency (data not shown).

Injection of vector suspension into the vitreous cavity failed to produce detectable expression of GFP in any intraocular cell type (not shown). The integrity of the anterior hyaloid face may prevent virus delivered into the vitreous body from reaching the anterior chamber in sufficient concentrations required to transduce corneal endothelial cells. Consistent with results following subretinal injection, this type of lentiviral vector clearly transduces cells of the neuroretina poorly.

The main findings of this study are that VSV-G pseudotyped HIV-based lentiviral vectors are able to transduce mitotically inactive corneal endothelial cells and RPE in vivo and that expression of the transgene is sustained in the absence of an inflammatory response. Transgene expression was consistently observed at the corneal periphery. This distribution may relate to aqueous flow patterns in the anterior chamber or reflect differences in populations of endothelial cells across the cornea. Gene transfer to corneal endothelial cells offers a potentially valuable approach to the management of disorders of the cornea and anterior segment. The corneal endothelium is a nonvascular monolayer of highly metabolic, mitotically inactive¹¹ cells which lines the posterior corneal surface and is bathed in the aqueous of the anterior chamber. This tissue is responsible for corneal transparency by maintaining dehydration of the corneal stroma.¹² A reduction in endothelial cell density resulting from surgical trauma, inherited degeneration, ageing or transplantation, leads to failure of endothelial function, loss of corneal transparency and visual loss. The corneal endothelium also plays a central role in the allogeneic response to corneal transplants¹³ and contributes to aqueous homeostasis.¹⁴ Although an in situ approach to gene transfer to the endothelium of donor cornea in culture has potential clinical application in the control of corneal allograft rejection, the management of many corneal disorders will require an in vivo strategy, to which lentiviral vectors may therefore be ideally suited. The trabecular meshwork is a complex structure situated in the iridocorneal angle which plays a critical role in the regulation of intraocular pressure and is an attractive target for the delivery of therapeutic genes to treat glaucoma. Transfection of trabecular meshwork by lentiviral vectors has previously been reported in human eyes in ex vivo organ culture,¹⁵ but has not previously been described in vivo. Our findings lend support to a potential role for lentiviral vectors in gene therapy of glaucoma.

Subretinal injection of an HIV-based lentiviral vector in adult mice led to efficient, stable transduction of the RPE. Transduction of the adjacent photoreceptor cell layer was restricted, however, to the immediate area of needle trauma at the injection site. These results are consistent with previous reports,^5,6 although in our study we found that the efficiency of photoreceptor transduction was low even in newborn pups. This finding however is consistent with recent unpublished reports by other groups which describe inefficient lentivirus-mediated transduction of photoreceptor cells in rats¹⁶ and in non-human primates.¹⁷ Although we can offer no clear explanation for the differences in tranduction efficiency observed by different investigators, it remains possible that changes in the configuration of the vector, particularly with regard to the expression system, may be responsible. The ability of lentiviral vectors to transduce RPE with relative tissue specificity may be valuable for the therapeutic gene transfer of anti-angiogenic and neurotrophic factors in the management of acquired retinal diseases, and for correction of inherited genetic diseases specifically affecting this cell layer.

Acknowledgements

JWBB is a Wellcome Trust Research Training Fellow. AJT is a Wellcome Trust Senior Clinical Fellow. KF and CD are supported by grants from the Primary Immunodeficiency Association and the Chronic Granulomatous Disease Research Trust.

1 Ali RR. Gene transfer into the mouse retina mediated by an adeno-associated viral vector. Hum Mol Genet 1996; 5: 591-594, Article MEDLINE

2 Flannery JG. Efficient photoreceptor-targeted gene expression in vivo by recombinant adeno-associated virus. Proc Natl Acad Sci USA 1997; 94: 6916-6921, Article MEDLINE

3 Reichel MB. Immune responses limit adenovirally mediated gene expression in the adult mouse eye. Gene Therapy 1998; 5: 1038-1046, MEDLINE

4 Wang X. Efficient and sustained transgene expression in human corneal cells mediated by a lentiviral vector. Gene Therapy 2000; 7: 196-200, MEDLINE

5 Miyoshi H, Takahashi M, Gage FH, Verma IM. Stable and efficient gene transfer into the retina using an HIV-based lentiviral vector. Proc Natl Acad Sci USA 1997; 94: 10319-10323, Article MEDLINE

6 Galileo DS, Hunter K, Smith SB. Stable and efficient gene transfer into the mutant retinal pigment epithelial cells of the Mitf(vit) mouse using a lentiviral vector. Curr Eye Res 1999; 18: 135-142, MEDLINE

7 Zufferey R. Self-inactivating lentivirus vector for safe and efficient in vivo gene delivery. J Virol 1998; 72: 9873-9880, MEDLINE

8 Zennou V. HIV-1 genome nuclear import is mediated by a central DNA flap. Cell 2000; 101: 173-185, MEDLINE

9 Cho BJ, Gross SJ, Pfister DR, Holland EJ. In vivo confocal microscopic analysis of corneal allograft rejection in rabbits. Cornea 1998; 17: 417-422, MEDLINE

10 Stave J. Keratinocyte density of the cornea in vivo. Automated measurement with a modified confocal microscopy. Klin Monatsbl Augenheilkd 1998; 213: 33-44,

11 Joyce NC, Meklir B, Joyce SJ, Zieske JD. Cell cycle protein expression and proliferative status in human corneal cells. Invest Ophthalmol Vis Sci 1996; 37: 645-655, MEDLINE

12 Green K. Corneal endothelial structure and function under normal and toxic conditions (published erratum appears in Cell Biol Rev 1991; 25: 343). Cell Biol Rev 1991; 25: 169-207,

13 Donnelly JJ. Class II alloantigen induced on corneal endothelium role in corneal allograft rejection. Ophthalmol Vis Sci 1990; 31: 1315-1320,

14 Francis BA, Alvarado J. The cellular basis of aqueous outflow regulation. Curr Opin Ophthalmol 1997; 8: 19-27, MEDLINE

15 Francis BA, Alvarado J. The cellular basis of aqueous outflow regulation.

16 Francis BA, Alvarado J. The cellular basis of aqueous outflow regulation.

17 Francis BA, Alvarado J. The cellular basis of aqueous outflow regulation.

Figure 1 GFP expression 6 weeks after anterior chamber injection of lentivirus. At intervals after injection, mice were killed by cervical dislocation. The eyes were removed, fixed in 4% paraformaldehyde for 4 h, cryoprotected in 20% sucrose and frozen in optimal cutting temperature medium. Serial 20 m cryosections were counterstained with propridium iodide, mounted and viewed by fluorescent microscopy. (a) At low magnification, a representative whole eye shows bright confluent GFP fluorescence across the greater part of the corneal diameter (c). There is no GFP expression in the lens (l) or retina (r). (b) At higher magnification GFP expression in the cornea is restricted to the endothelium (end) and is not seen in the epithelium (ep) or stromal keratocytes (s) in this section. (c) At high magnification GFP fluorescence is apparent in cells of the trabecular meshwork (tm) in the iridocorneal angle.

Figure 2 Laser scanning ophthalmoscope images of corneal GFP fluorescence in vivo. The confocal laser scanning ophthalmoscope (cLSO) used for GFP fluorescence measurements was a prototype Zeiss SM 30-4042 (Zeiss, Oberkochen, Germany). The cLSO has an Argon (Ar+) laser line at 488 nm that generates an illumination power of 140 W at the cornea. All images were acquired in 40 degree field of view mode using direct confocal aperture 3 (FWHM <600 m). The confocal aperture prevents light from out of the plane of focus reaching the detector. Consequently, only fluorescence from the cornea is detected. The cLSO detector is a Hamamatsu photomultiplier (Hamamatsu Photonics KK, Hamamatsu City, Japan) that has an extended red, high sensitivity multialkali photocathode. Fluorescence images were generated by illuminating the fundus with the argon laser line while a high pass filter (cut off at 521 nm) was in place in front of the detector. All cLSO images were recorded at 25 frames per second on to a SVHS videotape recorder (Panasonic AG7330, Matsushita, Japan) and digitised at 768 ´ 576 ´ 8 bit resolution using a Matrox Millennium frame grabber (Matrox Imaging Products Group, Quebec, Canada) installed on a Gateway P5/133 PC. 64 consecutive frames (or 2.5 s of data) were acquired. Using custom written software these frames were then aligned on a common feature, and averaged to reduce variations in corneal illumination and to increase the signal to noise of the image. (a) A reflectance image of the eye shows the upper (ul) and lower eyelid (ll) margins and a corneal light reflex (c). (b) A fluorescence image of the same eye shows GFP fluorescence across the cornea.

Figure 3 GFP expression 6 weeks after subretinal injection of lentiviral vector. Eyes were processed as previously described. Bright GFP fluorescence is evident across a large area of retinal pigment epithelium (rpe) in a section of an eye of an adult mouse. There is no evidence of transfected photoreceptors in the neurosensory retina (r) in this section. GFP fluorescence observed in the region of the neuroretina originates from detached RPE microvilli.