Article Text


Gene therapy for the eye
  1. MRC Human Genetics Unit, Western General Hospital
  2. Crewe Road, Edinburgh EH4 2XU

    Statistics from

    Gene therapy implies the delivery of genes to somatic tissues for therapeutic purposes.12 The eye is an attractive target for gene therapy because of its accessibility and its immune privilege. However, over the past few years gene therapy has taken some hard knocks, both from a disillusioned stock market and, more tellingly, from a distinguished peer review committee.34The latter concluded that more time should be spent on understanding and improving potential strategies at a basic science level and less on publishing hasty, poorly controlled clinical experiments. The premature scramble towards clinical studies prompted the taunt ‘less hype, more biology’ in a Science article.5 The question arises as to whether there are lessons here for the expanding group of researchers working towards ocular gene therapy.

    One of the most stringent tests of a gene therapy would be to replace the function of a defective ocular gene, since this requires the introduction and long term expression of a functional copy into many, if not a majority of, target cells. In the case of the retina, for example, specific genetic defects have been found in 20 of the 87 mapped genes causing genetic eye disorder.6 Most of these disorders result from single gene defects which are both highly disabling and potentially correctable. As an alternative, transferred genes expressing growth factors may be therapeutic in some of the degenerative disorders.7 These aspects of gene therapy have therefore attracted a growing body of researchers bent on revolutionising their treatment.8-14

    As long term gene expression is the goal, success depends on (i) efficient uptake into the target cells; (ii) avoidance of endocytosis and lysosomal degradation; (iii) import into the nucleus; (iv) stable retention in the nucleus, either as a circular episome (for example, adenovirus) or by integration into the host genome; (v) target cell specific expression of the therapeutic gene, driven by the natural promoter and enhancer elements; (vi) appropriate translation and subcellular localisation of the gene product (Fig 1). The most difficult steps are probably (iii), (iv), and (v)—import of the gene into the nucleus and achieving a stable and correct level of expression. These are major hurdles since they are often poorly understood and difficult to evaluate. What are the available options for retinal gene therapy?

    Figure 1

    Schematic diagram of the steps involved in cellular uptake and expression of a therapeutic gene. The filled arrows indicate the most difficult steps in achieving effective expression. See text for description.

    There are several possible sites for introducing genes into the retina. Intravitreal injection is a relatively safe and easy method of approach but potential complications such as vitreous haemorrhage, retinal detachment, and endophthalmitis cannot be ignored. However, studies using animal models have found that while this is an effective means of transducing ganglion cells, it is much less so for the outer retinal layers.8913 Our present understanding suggests that most severe human retinal degenerations are primarily disorders of photoreceptors, even though retinal pigment epithelium (RPE) cells may be the initial focus of disease. Subretinal injection, which produces a local and often transient detachment of the retina, has so far been the most favoured means of accessing the photoreceptor layer.8912-14 This reinforces the idea that very long term expression is necessary to avoid repeated entry. Other possible strategies15 include scleral or choroidal gene implants or episcleral injections, with or without iontophoresis to drive the vector into the retina; propulsion of DNA coated gold particles into the retina with a pressurised gene gun; intravenous or intracarotid injection of the transducing agent with the aim of achieving uptake by RPE—all offering radically different approaches. None of these can be completely discounted, bearing in mind the scepticism that greeted vaccination using naked DNA.16 New ideas are always threatening.

    In principle, there are several different approaches to obtaining therapeutic expression of introduced genes in the eye, only some of which have been used. These can be broken down into ex vivo and in vivo approaches:


    in vivo injection of viral derived vector expressing a therapeutic gene


    in vivo injection of genes carried by non-viral agents, such as liposomes or vectors designed for receptor mediated endocytosis


    ocular transplantation of ex vivo modified cells.

    In vivo injection of viral derived vectors

    The goal is to achieve efficient entry, nuclear uptake, and stable expression in photoreceptors. The native viral DNA is stripped of as many genes as possible to reduce possible toxicity and above all to ‘cripple’ it by removal of those genes required for replication. Signals required for efficient packaging into the viral coat and for integration or nuclear circularisation (for example, terminal repeats) are retained. Proteins for packaging the DNA into a viral coat are generally supplied by growing the vector in a packaging cell line which supplies them ‘in trans’ (that is, coded by another chromosome or genome). The choice of currently fashionable viral derived vectors for gene delivery is determined more by availability than suitability. Those currently used include adenovirus, adeno associated virus, and herpes simplex type 1 (HSV1) based vectors. Retroviral vectors can be used to transduce dividing cells such as ocular tumours but are not at present useful for non-dividing, differentiated cells such as photoreceptors (although see below). The advantages and disadvantages of the available vectors have been discussed elsewhere.217 What then has been achieved by applying viral derived vectors to the retina?

    Transfer of a readily detectable ‘reporter gene’ (lacZ, β galactosidase) to retinal cells and expression lasting 6–13 weeks has been described following subretinal injection of adenoviral (AV) vectors in mice.89 The RPE was efficiently transduced at viral titres of more than 107 plaque forming units (pfu). Almost all RPE cells were transduced within the area of subretinal injection but only a small percentage of photoreceptors, except in immature (5–7 day old) or degenerating retinas.89Extraocular tissues were not transduced and retinal toxicity appeared to be low. Transgene expression declined with time because of factors such as low grade immune reaction, switch off, or loss of the episomal transgene. Intravitreal injection was not effective in staining the outer retina or RPE even at titres as high as 109 pfu. Similar results have been obtained using adeno associated virus (AAV) based vectors but with more efficient transduction of photoreceptors11-13; however, this may result from the presence of contaminating wild type AV.1819 In each case no more than 1% of photoreceptors were successfully transduced and expression was not stable. However, Bennett and coworkers14 using AV vectors expressing the rod cGMP phosphodiesterase gene did succeed in delaying the photoreceptor degeneration in homozygous rd mutant mice for several weeks.

    In vivo injection of non-viral agents

    A number of non-viral methods for introducing genes into the retina have yet to be tried, including receptor mediated endocytosis. For example, Hart and coworkers20 have described an integrin binding peptide coupled to a polylysine chain which binds vector DNA, facilitating gene uptake into cultured cells. The use of cationic liposomes to target photoreceptors gave unpromising results initially13 although this method is notoriously sensitive to the particular liposome used and no systematic study using different liposome preparations has been reported. These approaches circumvent some of the safety concerns raised by viral vectors and the risk of an immune response to the introduced vector proteins is greatly diminished. In addition, liposomes have been reported to facilitate transduction of retroviruses into cells and so may at the least provide a useful adjunct to therapy.21

    The most promising application of non-viral agents to the retina used a mixture of liposome and expressing plasmid complexed with HMG1 chromosomal proteins, which was then coated with the envelope of Sendai virus (haemagglutinating virus of Japan, HVJ) to promote fusion with the cell membrane.22 The HMG1 proteins contain a nuclear localisation signal to facilitate nuclear uptake and may also protect the DNA from degradation. Intravitreal and subretinal injection of alacZ ‘reporter gene’ in mice and rats resulted initially in strong blue staining of the photoreceptors, indicating expression, with expression persisting for up to 30 days after injection. Staining of other neurons and glial cells was also seen but this was much less evident in RPE. No evidence of an immune response was seen. As a result of HVJ induced membrane fusion and cytoplasmic entry, the encapsulated plasmid is thought to bypass endocytosis and lysosomal degradation. Nuclear entry was enhanced by the presence of HMG1 proteins, increasing expression up to 10-fold. However, expression appeared more transient than with AV vectors since by 30 days after injection there was only weak or no blue staining.22

    Ocular transplantation of ex vivo modified cells

    The problems of both efficiently and stably transducing photoreceptors suggest an alternative ex vivo approach which currently shows promise in the treatment of degenerative disorders of the CNS.223 The aim is to introduce modified cells, such as fibroblasts taken from the patient, into the retina to act as a source of the missing protein. A retroviral vector is used to stably introduce the functional gene into the recipient cell genome under the control of a strong promoter. The expressed gene product can be modified so that it is secreted from the cell with the hope that it is then phagocytosed by target cells such as photoreceptors. Primary fibroblasts survive for extended periods of time within rodent (up to 18 months) and primate brains, although a decline in gene expression has been observed with time, suggesting the need for the concomitant expression of appropriate growth factors to maintain cell viability.23 This approach deserves to be explored because of the inherent difficulties of obtaining long term expression in cells such as photoreceptors using an in vivo approach. The most suitable site for transplanting the modified cells is unclear but the subretinal space, vitreous, and choroid are possibilities.

    Future developments

    It is difficult to predict the key ingredients required for success in retinal gene therapy. Less immunogenic vectors will certainly be helpful but one worrying possibility is that unless chromosomal integration of the introduced gene occurs, expression will be too short lived (for example, not more than 1 year) to be useful. None of the currently used vectors (AV, AAV, HSV1) integrates at an appreciable frequency, and although this can be seen as an advantage for dividing cells (less chance of oncogenic damage) it is probably a disadvantage in post-mitotic cells. This may necessitate the use of modified retroviral vectors which have low immunogenicity and now have the potential for integration into non-dividing neural cells.24

    Good animal models are essential to evaluate the efficacy, safety, and potential for therapy in humans. Targeted disruption of specific genes and the creation of transgenic mutants are confined to rodents at present, but recent work suggests that this may change so that larger animal models become available.25 This will facilitate the evaluation process and shorten the step from animals to humans.

    Adenoviral vectors are not efficient at transducing photoreceptors and it is possible that the success in delaying the progress of therd mouse retinal degeneration results from highly efficient RPE uptake and expression spilling over into adjacent photoreceptors. Bennett and coworkers noted that diffusion of lacZ protein was occurring from RPE to photoreceptors.14 The use of a photoreceptor specific promoter to drive expression should resolve this. Even if this is the case, fewer immunogenic AV vectors should lead to a longer duration of expression and improved photoreceptor rescue, if by an unexpected route. Similarly, if the apparently superior efficacy of AAV for transducing photoreceptors11-13 results from a combination of its small size and the presence of contaminating wild type AV18-19 or even AAV, it is a rather oblique success. These and related studies provide valuable proofs of principle. However, the confusion relating to actual mechanisms brings to mind the plea for more basic research and the dangers of rushing into clinical trials before the efficacy of gene therapy techniques is adequately understood. This is nowhere more important than in the treatment of genetic eye disease, where the watchwords must remain animal models, systematic investigation and, above all, safety. The field clearly has promise but there will be no quick fixes. Research money is justified but it must go to those carrying out the hard groundwork.


    I would like to thank Dr B Fleck, Dr B Dhillon, Professor D Porteous, and Dr K Dry for helpful discussions.


    View Abstract

    Request permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.