AAV‐mediated gene therapy improving mitochondrial function provides benefit in age‐related macular degeneration models

With an estimated 196 million people suffering from age-related macular degeneration (AMD) in 2020 and predicted to increase to 288 million by 2040,1 dry AMD, representing 70%–90% of AMD cases, represents an enormous clinical need with no current therapies. We have demonstrated that NDI1 and an optimised version of NDI1 (ophNdi1), a mitochondrial complex 1 equivalent from Saccharomyces cerevisiae, provide functional and histological benefit in two murine models of dry AMD as well as benefit in two cellular models of dry AMD. There are no drugs on the market for dry AMD. However, there are currently a small number of candidate gene therapies in clinical trial (clinicaltrials.gov). To our knowledge, this is the first demonstration that a gene therapy directly targeting mitochondrial dysfunction provides functional benefit in in vivo models of dry AMD, making this a novel approach to treating this devastating condition. Dry AMD is characterised by the formation of drusen between Bruch's membrane (BM) and the basal lamina of the retinal pigment epithelium (RPE) and atrophic changes in the choriocapillaris followed by the death of photoreceptors in the macula and geographic atrophy, with a related loss of central vision. AMD is multifactorial with genetic and environmental factors known to contribute to the disease.2 Although underlying mechanisms involved in AMD are not fully understood, mitochondrial dysfunction leading to increased oxidative stress in the RPE, DNA damage and impaired mitophagy are known to contribute to RPE and photoreceptor cell death.3 Both the RPE and photoreceptors have been shown to display mitochondrial complex 1 (of the electron transport chain) deficiency.4 The Cfh−/− mouse5 has been widely used as a dry AMD model and aged Cfh−/− mice have been reported to display impaired visual function, thinning of the retinal outer nuclear layer, changes in BM and basal laminar deposits (BlamDs).5, 6In this study, we also observed electroretinography (ERG) deficits in aged Cfh−/− mice (Figure 1A–D, Table S1), but no changes in BM or BlamDs were apparent. However, cone photoreceptors exhibited disorganised outer segments, and substantial mitochondrial alterations compared to cones of control mice. Cone mitochondria appeared shrunken and fragmented and the cytoplasm of inner segments swollen and electron-lucent. These changes in cone histology have not previously been reported and indicate mitochondrial dysfunction (Figure 1E–J). We have investigated the utility of the nuclear-encoded NDI17 gene as a candidate therapy for dry AMD. NDI1 provided benefit in models of Parkinson's disease, Leber hereditary optic neuropathy and multiple sclerosis.8 NDI1 has also been shown to reduce reactive oxygen species (ROS) and oxidative stress in disease models.7, 8We utilised a codon-optimised version of NDI1, ophNdi1, which we observed to express ∼3-fold higher than wild-type NDI1 in murine retina from recombinant adeno-associated viral (AAV) vectors following subretinal delivery (Figure S1). A range of AAV2/8 and AAV2/5 viral doses (1.0 × 107–7.5 × 109 vg) were used to deliver ophNdi1 and NDI1 subretinally to Cfh−/− mice. Significant and robust functional benefit was observed in 60 aged mice using ERG readouts, as well as reduced ROS, increased nicotinamide adenine dinucleotide (NADH) oxidation and increased cone photoreceptor numbers in treated versus control eyes (Figures 2A–O, S2, S3). Notably, with none of the doses used were negative effects observed even up to 7–9 months post-injection. In acknowledgement that no model recapitulates all aspects of dry AMD, a second murine model, the well-established sodium iodate-induced (NaIO3) model,9was also treated subretinally with AAV2/8-ophNdi1 and AAV2/5-ophNdi1. NaIO3, a strong oxidising agent, causes catastrophic damage to the RPE leading to subsequent photoreceptor loss and reduced photoreceptor cell function, including reduced ERG amplitudes when delivered systemically.10 Similar to our findings in the Cfh−/− mouse, subretinally delivered AAV-ophNdi1 provided robust ERG benefit, as well as improved optokinetic responses and increased cone photoreceptor cell numbers in treated versus control eyes (Figure 2K–O). To interrogate the mechanism behind the observed functional and histological benefit in the treated murine NaIO3 model, cellular models utilising NaIO3 were investigated; primary porcine RPE (pRPE) cells and ARPE19 cells, a well-established cell line with some characteristics of RPE. pRPE cells were transduced with AAV2/8-ophNdi1 and insulted with NaIO3. Immunocytochemistry for 8-OHdG (oxidative stress marker), CPN60 (mitochondrial marker) and phalloidin (selective for F-actin) showed high levels of oxidative and mitochondrial stress and the absence of actin filaments in NaIO3-insulted versus control cells, indicating severe stress and reduced viability. In contrast, insulted cells transduced with AAV2/8-ophNdi1 appeared similar to control cells (Figures 3A–O, S4). Similar rescue from NaIO3 insult was also observed in ARPE19 cells transduced with AAV2/8-ophNdi1 (Figures 4A–O, S5). These data suggest that AAV2/8-ophNdi1 treatment provides significant protection against mitochondrial stress, oxidative damage to DNA and cell death in the cellular NaIO3 models. Additionally, mitochondrial stress tests were performed on pRPE cells transduced with AAV2/2-ophNdi1 and insulted with NaIO3. NaIO3 insult significantly reduced basal oxygen consumption rates (OCRs), maximal OCRs and ATP production in cells. However, treatment with AAV2/2-ophNdi1 significantly increased each of these parameters indicating a rescue of mitochondrial function (OXPHOS, Figure 3P–R). Spare respiratory capacity, the difference between maximal OCR and basal OCR, was reduced with AAV2/2-ophNdi1 treatment as basal OCR was increased by more than the maximal OCR (Figure 3Q). When pRPE cells were exposed to the complex 1 inhibitor rotenone, OCRs were reduced to background levels in control and NaIO3-insulted cells. However, the addition of rotenone to AAV2/2-ophNdi1-treated cells – NDI1 is insensitive to rotenone – had minimal effect on OCR levels, which were substantially maintained (Figure 3S). Notably, similar benefits in bioenergetic profiles were also observed in NaIO3-insulted ARPE19 cells transduced with AAV2/2-ophNdi1 (Figure 4P,Q, Table S2). We tested NDI1 and ophNdi1, which target mitochondrial dysfunction, known to be a key factor in dry AMD. Robust benefit was demonstrated with multiple AAV-delivered NDI1/ophNdi1 vectors and doses in the Cfh−/− and NaIO3-induced mouse models as well as two cell models. The study represents the first demonstration globally of functional benefit in vivo in dry AMD models provided by a gene therapy directly targeting mitochondrial function. We thank Charles Murray for technical assistance. Cfh−/− mice were kindly donated by Professor Marina Botto, Imperial College London. We also thank the following funding agencies: (16/IA/4452, GJF, PH, SMW, NC), Health Research Board Ireland (HRAPOR-2015-1140, GJF), Enterprise Ireland and European Regional Development Fund (ERDF) under Ireland's European Structural and Investment Funds programme 2014–2020 (EI CF-2019-1106-Y), EU Marie Curie Innovative Training Network (StarT 813490, GJF, IJMP), Fighting Blindness Ireland – Health Research Charities Ireland (MRCG-2016-14 GJF), Irish Research Council (Ulysses 2018 TL, GJF), Fight for Sight UK (1744/45 AS). SMW, NC, MC, PFK and GJF are inventors on patent no. 10220102. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.