Preclinical studies of an AAV8-CYP4V2 gene therapy VGR-R01 for the treatment of Bietti crystalline dystrophy

Abstract

Bietti crystalline dystrophy (BCD) is an autosomal recessive disorder caused by loss-of-function mutations in the CYP4V2 gene, characterized by crystal-like lipid deposits in the retina, progressive photoreceptor loss, and retinal pigment epithelium (RPE) deterioration. Currently, there are no approved treatments for BCD. VGR-R01, an investigational gene therapy, uses subretinal administration of recombinant adeno-associated virus type 8 (AAV8) vector to deliver the human CYP4V2 gene. This therapy is now undergoing phase 1/2 clinical trials (NCT05694598). The pre-clinical study results for VGR-R01 are summarized, with a focus on its pharmacology, pharmacokinetics, and toxicology. The in vitro cellular studies demonstrated that VGR-R01 induces a dose-dependent expression of the CYP4V2 protein, which significantly enhances fatty acid hydroxylase activity and reduces lipid droplet accumulations in the RPE cells. In vivo, VGR-R01 showed effectiveness in improving electroretinogram (ERG) amplitudes in 8-month-old Cyp4v3^−/− mice. VGR-R01 was well tolerated in New Zealand rabbits and non-human primates (NHPs). Furthermore, after subretinal administration, VGR-R01 was primarily distributed in the ocular tissues, especially in the retina, with minimal systemic presence, notably in the gonads. Overall, these results support the potential for clinical application of VGR-R01 in treating BCD.

Graphical abstract

In this paper, Zhu and colleagues evaluated the efficacy and safety of an AAV8-CYP4V2 gene therapy VGR-R01 in in vitro cellular studies, animal models, New Zealand rabbits, and NHPs that provides critical support for its clinical translation in Bietti crystalline dystrophy.

Introduction

Bietti crystalline dystrophy (BCD) is an autosomal recessive disorder caused by loss-of-function mutations in the CYP4V2 gene, leading to the accumulations of crystal-like lipid deposits in the retina. This condition, a form of retinitis pigmentosa, manifests through a progressive loss of photoreceptors and deterioration of the retinal pigment epithelium (RPE).^1,2 Clinically, BCD presents with symptoms such as night blindness (nyctalopia), flashes of light (photopsia), diminished visual acuity, myopia, blurred vision, and visual field constriction, often progressing to legal blindness by middle age.¹ While BCD is relatively rare globally with a prevalence of approximately 1/67,000, it is more common in the East Asian populations.^3,4 In China, for instance, the carrier frequency for CYP4V2 mutations is about 0.005 per individual, translating to roughly 21,000 affected persons.⁵ Currently, treatment options are limited to supportive care, similar to therapies used for other types of retinitis pigmentosa, which include vasodilators, vitamins, and anti-vascular endothelial growth factor (VEGF) injections to manage choroidal neovascularization.^3,6

The genetic foundations of BCD were established in 2004, identifying mutations in the CYP4V2 gene as the principal cause.² The gene is critical for fatty acid metabolism in the RPE cells, and its mutations result in disrupted lipid metabolism with subsequent accumulation of cholesterol or cholesterol esters in various tissues, notably the retina and cornea.^3,7 Studies have indicated that RPE cells with CYP4V2 mutations suffer from altered fatty acid homeostasis, leading to an accumulation of polyunsaturated fatty acids (PUFAs), increased mitochondrial reactive oxygen species, impaired respiratory functions, and cell apoptosis.⁸ By restoring fatty acid metabolism in the RPE cells, it could potentially alleviate the structural and functional damages, thus preserving vision.⁹

Gene replacement therapy has shown promise as an effective treatment for loss-of-function monogenic diseases, with FDA-approved adeno-associated virus (AAV)-based therapies available for conditions like congenital amaurosis. Research conducted in 2020 by Qu and colleagues on a Cyp4v3^−/− mouse model, which mimics the BCD phenotype, demonstrated that subretinal delivery of a human CYP4V2 gene using an AAV vector could slow disease progression, highlighting the potential of AAV-based therapies in BCD treatment.¹⁰ However, further development and comprehensive preclinical evaluations are necessary to advance such therapies to clinical application.

VGR-R01, an investigational gene therapy product currently in phase 1/2 clinical trials (NCT05694598), utilizes a recombinant AAV serotype 8 (AAV8) vector to deliver the CYP4V2 gene. This therapy has been optimized for effective expression in RPE cells, incorporating tailored promoter and codon usage alongside essential non-coding regulatory elements. Comprehensive preclinical studies focusing on pharmacology, pharmacokinetics, and toxicology have been conducted to support the clinical translation of VGR-R01. This article provides a summary of these extensive pre-clinical research results for VGR-R01.

Results

Optimization and characterization of a CYP4V2 AAV vector expression cassette

The expression cassette for VGR-R01 was comprehensively optimized, focusing on key elements such as promoters, regulatory elements, and the CYP4V2 coding sequence. Given the broad expression of CYP4V2 in RPE, photoreceptor, and choroidal cells, a universal CAG promoter was selected for its ability to significantly enhance the expression of the CYP4V2 protein, as shown in studies together with the published CMV-hCYP4V2 construct (Figure S1A).¹⁰ The coding sequence was modified to increase the codon usage efficiency in the human cells, thereby optimizing the CYP4V2 protein expression levels. Additionally, modifications were made to reduce CpG islands, aiming to minimize potential TLR9-mediated immune responses to reduce the risk of gene silencing in vivo.¹¹ A total of 16 optimized human CYP4V2 coding sequences were designed and cloned under the CAG promoter, resulting in 17 constructs named BCD1–BCD17 (Table S1). Of these, BCD16 was chosen for its consistently high CYP4V2 expression in both HEK293 and ARPE-19 cells (Figure S1B), featuring zero CpG and no CpG islands, a significant improvement from the wild-type sequence that contains 50 CpGs with CpG islands. The final transgene cassette of VGR-R01 is depicted in Figure 1A.

Figure 1.

Optimization and characterization of the AAV vector expression cassette

(A) Illustration of the CYP4V2 transgene expression cassette of VGR-R01. (B) VGR-R01-mediated expression of CYP4V2 in RPE cells differentiated from iPSCs of a BCD patient (CYP4V2^−/− RPE cells). The cells were transduced with VGR-R01 at MOIs from 1 × 10⁵ to 1 × 10⁶, assessed by western blot. RPE cells treated with AAV8-EGFP (MOI = 1 × 10⁶) were used as the negative control. Histogram was constructed according to the gray values of the protein bands and presented as mean ± SD. (C) VGR-R01 restored fatty acid hydroxylase activity in the RPE cells. ARPE-19 cells were treated with VGR-R01 at different MOIs. The hydroxylase activity of the CYP4V2 protein was shown as the production rate of 12-hydroxydodecanoic acid or 14-hydroxytetradecanoic acid, detected using the LC-MS/MS method.

To assess the VGR-R01-mediated expression of the CYP4V2 protein, we conducted in vitro expression studies across several cell lines and RPE cells differentiated from induced pluripotent stem cells (iPSCs) of a BCD patient. The VGR-R01, which was a recombinant AAV8 carrying the BCD16 expression cassette, were produced. The cell lines ARPE-19, A549, HEK293, HepG2, and SH-SY5Y were transduced with VGR-R01 at various multiplicities of infection (MOI, ranging from 1 × 10⁵ to 5 × 10⁶ vg/cell). All the tested cell lines demonstrated an MOI-dependent increases in the CYP4V2 protein levels (Figure S2), confirming that expression driven by the universal CAG promoter is independent of cell types. In addition, tests on the CYP4V2^−/− RPE cells, differentiated from iPSCs using a two-dimensional cytokine induction method,¹² VGR-R01 again showed MOI-dependent expression of CYP4V2, ranging from 1 × 10⁵ to 1 × 10⁶ vg/cell as illustrated in Figure 1B. This MOI-dependent CYP4V2 expression was also confirmed in the iPSC-derived RPE cells from healthy donors (data not shown). The aforementioned findings showed that VGR-R01 can effectively transduce disease relevant cells to express the CYP4V2 protein, indicating its potential for therapeutic application.

VGR-R01 restored fatty acid hydroxylase activity in the RPE cells

To evaluate the ability of VGR-R01 to restore hydroxylase activity in the transduced cells, an in vitro enzymatic activity study was performed. CYP4V2, a member of the cytochrome P450 family, is known for its hydroxylase activity crucial for maintaining fatty acid homeostasis in RPE cells. This protein effectively catalyzes the omega-hydroxylation of medium-chain saturated fatty acids, particularly lauric and myristic acids, which serve as sensitive substrates.^13,14 ARPE-19 cells were co-cultured with VGR-R01 at various MOIs (1 × 10⁵, 5 × 10⁵, 2 × 10⁶, and 5 × 10⁶), and subsequently the cell lysates were incubated with cofactors cytochrome b5, cytochrome P450 reductase, and reduced NADPH, along with lauric or myristic acids as substrates. Hydroxylation enzyme activity was measured by the production rates of 12-hydroxylauric acid or 14-hydroxynutmeg acid, quantified using liquid chromatography-tandem mass spectrometry (LC-MS/MS) with a detection limit of 0.5 ng/mL, translating to an activity detection threshold of 4.2 pg/mL/min.

The detected activity levels, after correcting for background at the 0-min mark, are depicted in Figure 1C. Following transduction with VGR-R01, there was a significant MOI-dependent increase in hydroxylation activity. These findings suggest that VGR-R01 has the potential to restore cellular fatty acid hydroxylase activity to re-establish fatty acid metabolic homeostasis in the retinas of BCD patients with CYP4V2 hydroxylase deficiencies.

VGR-R01 reduced lipid accumulation in the CYP4V2^−/− RPE cells

It has been reported that RPE cells with CYP4V2 mutations are defective in fatty acid homeostasis, leading to excessive accumulations of the PUFAs.⁸ We have successfully generated differentiated RPE cells from both H9 human embryonic stem cells and BCD patient-derived pluripotent stem cells, named CYP4V2^+/+ and CYP4V2^−/− RPE cells, respectively. As shown in Figure 2A, these differentiated cells exhibited prominent pigment deposition, characteristic of the RPE cells.

Figure 2.

VGR-R01 reduced lipid droplet accumulation in the CYP4V2^−/−RPE cells

(A) Prominent pigment deposition in RPE cells differentiated from iPSCs of a BCD donor. (B and C) The accumulations of lipid droplets in RPE cells differentiated from a healthy donor (CYP4V2^+/+ RPE cells), a BCD donor (CYP4V2^−/− RPE cells), and CYP4V2^−/− RPE cells transduced with VGR-R01 at MOI of 5 × 10⁵ vg/cell, analyzed by staining of BODIPY. Integrated densities, calculated by mean fluorescence × area, were shown in bar graph. Data are presented as mean ± SD. Ordinary one-way ANOVA was used for significance analysis. ∗∗p < 0.01, ∗∗∗p < 0.0001.

To assess the impact of VGR-R01 on the fatty acid metabolic homeostasis in CYP4V2^−/− RPE cells, we employed BODIPY-493/503 staining to visualize the intracellular lipid accumulation. Images and quantitative analysis of lipid droplet accumulation, captured using fluorescence confocal microscopy, are presented in Figures 2B and 2C. The staining results showed that the lipid droplet accumulation in the CYP4V2^−/− RPE cells was significantly higher compared to the CYP4V2^+/+ RPE cells, confirming the metabolic disturbance in the BCD patient-derived cells. Notably, treatment with VGR-R01 (MOI = 5 × 10⁵ vg/cell) led to a significant reduction in lipid droplet accumulation in the CYP4V2^−/− RPE cells (p < 0.0001) compared to the untreated cells, demonstrating a restorative effect on fatty acid metabolism in these cells.

In vivo efficacy in the Cyp4v3^−/− mice

To investigate the in vivo effects of VGR-R01 in ameliorating the retinal function, an efficacy study was conducted using the Cyp4v3^−/− mice, a model for BCD. The mouse Cyp4v3 gene is a close homolog of the human CYP4V2 gene, sharing 82% amino acid sequence identity.¹⁵ Taking into account the relative slow progression of this disease in humans, 8-month-old Cyp4v3^−/− mice were used. As depicted in Figure 3A, the baseline electroretinograms (ERGs) showed significantly reduced b-wave amplitudes in both dark- and light-adapted ERGs of the 8-month-old Cyp4v3^−/− mice compared to the wild-type mice (Figure 3A), indicating impairments in the retinal function of the mutant mice.

Figure 3.

VGR-R01 consistently and steadily enhanced the ERG amplitudes in the Cyp4v3^−/− mice

Forty 8-month-old Cyp4v3^−/− mice (20/sex) were randomly assigned into 2 groups, 10/sex/group, and were given vehicle control or 4.0 × 10⁷ vg/eye of VGR-R01 through unilateral (the right eye) subretinal injection. C57BL/6J mice were used as the wild-type control. (A) The dark-adapted and light-adapted B wave amplitudes of ERG prior to administration. (B) The OCT representative images prior to injection (top), immediately after injection (middle), and after recovery (bottom). (C–E) Dark-adapted and light-adapted B wave amplitudes of ERG after VGR-R01 treatment in 3 (C), 6 (D), and 9 months (E). Dark adaptation with an intensity of 0.01, 0.1, or 1 cd s/m². Light adaptation with an intensity of 3 or 10 cd s/m². OD, oculus dexter, right eye; OS, oculus sinister, left eye. Data are presented as mean ± SD. Ordinary one-way ANOVA was used for significance analysis. ∗p < 0.05.

Forty 8-month-old Cyp4v3^−/− mice (20 of each sex) were divided into two groups to receive either VGR-R01 or vehicle through unilateral subretinal injections in the right eye. Wild-type C57BL/6J mice served as control. A dose of 4.0 × 10⁷ vg/eye was selected based on its tolerability in a preliminary dose-range finding study in this model. After subretinal injections, optical coherence tomography (OCT) showed retinal elevation and detachment at the injection site (Figure 3B). Follow-up OCTs every three months showed reattachment of the retinal nerve sensory layer, with intact retinal structures and clear demarcation of the outer nuclear layer and ellipsoid zone. There were no differences between wild-type C57BL/6J mice and Cyp4v3^−/− mice in general ophthalmological examination, OCT, intraocular pressure, fundus photography (FP), and fundus auto fluorescein (AF) before and after treatment. No abnormal changes related to VGR-R01 were observed.

ERG assessments at 3, 6, and 9 months post-treatment indicated that while b-wave amplitudes in the vehicle control Cyp4v3^−/− mice remained significantly lower than in the wild-type mice, those in VGR-R01 treated mice showed improvement (Figures 3C–3E). By 3 months, the b-wave amplitudes in the treated right eyes nearly returned to the wild-type levels. By 6 months, the left eyes also showed significant increases approaching the wild-type levels. At 9 months, significant improvements under dark adaptation were observed in both eyes, with b-wave amplitudes nearly matching those of the wild-type mice. These results suggest that unilateral subretinal injection of VGR-R01 not only improves the visual function in the injected eyes of Cyp4v3^−/− mice but may also have beneficial effects on the other untreated eyes. Furthermore, VGR-R01 was well tolerated in the Cyp4v3^−/− mice, with no adverse effects on body weight, general ophthalmological health, intraocular pressure, or other clinical parameters compared to the vehicle control mice.

VGR-R01 was well tolerated in the New Zealand rabbits and cynomolgus monkeys

The safety profile of VGR-R01 was evaluated in cynomolgus monkeys (Table 1) and New Zealand rabbits (Table 2) using the same route of administration and dose regimen proposed for the clinical trial. The CYP4V2 amino acid sequences of the New Zealand rabbit and the cynomolgus monkey share 84.76% and 99.05% sequence similarity, respectively, with the human sequence, which is among the highest of the commonly used lab animal species. In addition, the eyes of these two species are structurally similar to the human eyes, allowing for subretinal injection and eye examinations to be conducted similarly in human clinical studies.

Table 1.

Study design of GLP toxicology study in the NHP

Design of non-human primate toxicology study
	Animalsa		Dose (vg) (60 μL/eye)
Group	Male	Female	Right eye	Left eye
1	5	5	0 (vehicle)	0 (vehicle)
2	5	5	2.4 × 1011	2.4 × 1011
3	5	5	2.4 × 1012	2.4 × 1012

^aThree/sex/group animals were euthanized after 4 weeks, and the remaining two/sex/group animals were euthanized after 13 weeks.

Table 2.

Study design of GLP toxicology study in the New Zealand rabbits

Design of New Zealand rabbits toxicology study
	Animalsa		Dose (vg) (40 μL/eye)
Group	Male	Female	Right eye	Left eye
1	5	5	–	0 (vehicle)
2	5	5	–	1.6 × 1010
3	5	5	–	1.6 × 1011
4b	5	5	–	1.6 × 1011

^aThree/sex/group animals were euthanized after 4 weeks, and the remaining two/sex/group animals were euthanized after 13 weeks.

^bThe animals in group 4 were only used for biodistribution analysis.

After a single bilateral subretinal injection (in both eyes), VGR-R01 was well tolerated in monkeys. Compared to the vehicle control, no changes in clinical observations, body weight, food consumption, clinical pathology, or histopathological examinations (except the eyes) were observed in monkeys treated with VGR-R01. Ophthalmological findings, such as inflammatory cell exudation in the vitreous, needle insertion site injuries, alterations in the coverage area of the local and/or peripheral drug solution (representative data and figures were shown in Tables S2, S3, and Figure S3), as well as changes in both the amplitude and implicit time of ERG (data not shown), were observed in monkeys from both the vehicle control and VGR-R01-treated groups on D29. This suggests that these changes were most likely due to the injection procedure and/or ophthalmological examination procedure. The administration of VGR-R01 may exacerbate changes in vitreous inflammatory cell exudation, alterations in the local and/or peripheral drug solution coverage area, and ERG amplitude and implicit time. By D92, none of the ophthalmological findings showed obvious trend of recovery except for the changes in the amplitude and implicit time of ERG. Additionally, ocular histopathological findings (data not shown), including inflammatory cell infiltration of the choroid/retina/sclera/ciliary body, with or without vitreous inflammatory cell exudation were also noted in monkeys from both the vehicle control and VGR-R01-treated groups on D29. These findings still present on D92 but were less severe and less frequent, indicating potential reversibility. Based on the effects of the subretinal injection procedure and the results of ERG, the maximum tolerated dose (MTD) of VGR-R01 was determined to be 2.4 × 10¹¹ vg/eye for monkeys in this study.

In rabbits, findings similar to those in monkeys were observed following a single unilateral (left eye) subretinal injection of VGR-R01. The injection was well tolerated in rabbits without any systemic adverse effects. No abnormal changes related to VGR-R01 were observed in the general ocular examination, OCT, ERG, FP, or intraocular pressure in the non-injected eyes of rabbits. Ophthalmological findings of posterior vitreous detachment, inflammatory cell exudation in the vitreous, injuries at the needle insertion site, and relevant changes in the local injection area (representative data and figures were shown in Tables S2, S4, and Figure S4), as well as ocular histopathologic findings of retinal atrophy (data not shown) were observed in the injected eyes of rabbits from the vehicle control and VGR-R01-treated groups on D29, suggesting that these changes were most likely attributed to the injection procedure and/or ophthalmological examination procedure. The administration of VGR-R01 may aggravate changes of vitreous inflammatory cell exudation and alterations in the local injection area. However, a decrease in the amplitude of ERG was noted in the injected eye of males from the 1.6 × 10¹¹ vg/eye group (data not shown). On D92, these changes did not show any obvious recovering trends except for posterior vitreous detachment and inflammatory cell exudation in the vitreous. Additionally, ocular histopathological findings (data not shown), including inflammatory cell infiltration of the choroid/retina, inflammatory exudate of the vitreous cavity, and fibrosis of the choroid, were noted in VGR-R01-treated rabbits on D29; these findings were also observed on D92 but were of less incidence and severity and, therefore, seemed reversible. Based on the effects of subretinal injection and the results of ERG, the MTD and the no observed adverse effect level of VGR-R01 for rabbits in this study were determined to be 1.6 × 10¹¹ and 1.6 × 10¹⁰ vg/eye, respectively.

Biodistribution and/or shedding in New Zealand rabbits and cynomolgus monkeys

The biodistribution and shedding of VGR-R01 were investigated as part of the toxicology studies, including biodistribution of CYP4V2 DNA and mRNA in the New Zealand rabbits and the cynomolgus monkeys and shedding in the cynomolgus monkeys.

In rabbits, at a dose of 1.6 × 10¹¹ vg/eye, DNA was detected in the blood, which peaked on D2, then gradually declined, and was undetectable on D8. CYP4V2 DNA and mRNA were mainly distributed in the eyes. After unilateral eye injection, they were distributed in both eyes with a higher distribution in the retina. CYP4V2 DNA was only detected in the optic tract, optic chiasma, and spleen. The levels of CYP4V2 DNA and mRNA remained relatively stable in the retina of the administered eye over time but they were eliminated over time in other tissues (Figure 4A).

Figure 4.

Biodistribution of VGR-R01 in New Zealand rabbits and cynomolgus monkeys

The biodistribution of VGR-R01, including CYP4V2 DNA and mRNA, were evaluated in the ocular and non-ocular tissues from the high dose group in the New Zealand rabbits (A) and cynomolgus monkeys (B). The collected tissues were analyzed to detect CYP4V2 DNA using qPCR and the tissue samples with detectable CYP4V2 DNA were further analyzed to measure the levels of the CYP4V2 mRNA using RT-qPCR. Data are presented as mean ± SD.

In monkeys, at a dose of 2.4 × 10¹¹ vg/eye, CYP4V2 DNA was detected in the blood, which peaked on D2, then declined over time, and became undetectable on D92. CYP4V2 DNA and mRNA were mainly distributed in the eye tissues, with a higher distribution in the retina. CYP4V2 DNA was only detected in the spleen, thymus, lymph nodes, liver, heart, kidney, optic tract, and gonads. The biodistribution of CYP4V2 DNA and mRNA was stable in the retina but declined over time and became undetectable in other tissues (Figure 4B). Additionally, CYP4V2 DNA was detected to shed through tears, nasal and oral mucosa, urine, and feces (data not shown).

Discussion

BCD is an autosomal recessive disorder characterized by biallelic mutations in the CYP4V2 gene, which leads to dysregulated lipid metabolism and accumulations of cholesterol or cholesterol esters in various tissues, particularly the retina. Gene replacement therapy has emerged as a promising treatment for monogenic diseases by correcting the genetic defects via functional gene replacement. Qu and colleagues¹⁰ demonstrated that subretinal delivery of a human CYP4V2 gene via an AAV vector could alleviate the disease progression in a genetic mouse model of BCD, highlighting the potential of CYP4V2 gene therapy in treating BCD, a disabling disease without any approved treatment. VGR-R01, a recombinant AAV8 vector-based gene therapy encoding a functional CYP4V2 protein, is currently in phase 1/2 clinical trials (NCT05694598). Preclinical studies to support the clinical development of VGR-R01 have comprehensively assessed its in vitro and in vivo efficacy, pharmacodynamics, biodistribution, shedding, and safety.

Preliminary characterization of VGR-R01 showed that it dose-dependently mediated the CYP4V2 protein expression across multiple cell lines and in RPE cells derived from BCD patient-derived pluripotent stem cells, increased fatty acid hydroxylase activity, and significantly reduced lipid droplet accumulations. The data suggested that VGR-R01 has the potential to be an effective first-in-class gene therapy for treating BCD patients to improve visual impairments.

In the publication by Qu et al., they used high-fat diet to accelerate the disease progression of the Cyp4v3^−/− mice. However, BCD patients typically experience a very slow disease progression, often taking several decades to develop legal blindness.^4,16 In order to better mimic the slow pathophysiological progression of BCD patients, we conducted an efficacy study using older Cyp4v3^−/− mice maintained on a normal diet. At 8 months of age, which correlates approximately to 30–40 human years, the age when BCD patients typically begin to exhibit visual impairments, these mice displayed a phenotype that partially resembled human BCD; in particular, a significant reduction in the ERG amplitudes was observed. Following treatment with VGR-R01, the ERG assessments demonstrated significant improvements in amplitudes at 3, 6, and 9 months post treatment, indicating the potential of VGR-R01 to enhance visual functions in BCD patients. However, translating these findings to clinical settings presents challenges due to differences in disease manifestation between mice and humans, and the impracticality of using subjective visual function tests such as the best corrected visual acuity, visual fields, and mobility testing in mice. Additionally, the absence of predictable biomarkers for BCD complicates the efficacy prediction of VGR-R01 in clinical trials. These factors underscore the complexities of bridging preclinical results with clinical outcomes in developing effective BCD therapies.

The biodistribution of a gene therapy product is significantly influenced by the AAV serotype used, the administration mode, and the dosage. In our study, a single subretinal injection of VGR-R01 in rabbits and monkeys resulted in the primary distribution of CYP4V2 DNA and mRNA within the ocular tissues, predominantly in the retina, with minimal systemic distribution, notably in the gonads. Over time, the concentration of CYP4V2 DNA and mRNA remained stable in the retina but decreased in other tissues to be undetectable over time. These findings are consistent with the previous report of the subretinal injection of an AAV8 vector in NHPs.¹⁷ Moreover, we observed that CYP4V2 DNA and mRNA were detectable in the contralateral, non-injected eyes of the rabbits, suggesting potential binocular benefits from unilateral subretinal injections, as reported by David and colleagues that Lenadogene nolparvovec, an AAV2-MT-ND4 gene therapy, could migrate to the contralateral eye in NHPs after unilateral intravitreal administration.¹⁸ Some possible mechanisms for this phenomenon include inter-orbital or systemic circulation DNA transfer and the migration of mitochondrial components (RNA, protein) between eyes.^18,19 Indeed, in our studies, CYP4V2 DNA was detected in the optic chiasm, optic tracts, and optic nerves of both eyes in NHPs, but only transiently detected in the blood shortly after administration. These findings support the hypothesis of inter-orbital DNA transfer, although further research would be needed to fully elucidate the underlying mechanisms.

In our toxicology studies, we observed ophthalmological changes in both the vehicle control and VGR-R01-treated groups across all animal subjects. These included posterior vitreous detachment, inflammatory cell exudation in the vitreous, injuries at the needle insertion site, and alterations in the coverage area of the drug solution, as well as changes in the ERG amplitude and implicit time. Histopathological observations included retinal atrophy and infiltration of inflammatory cells into the choroid, retina, sclera, and ciliary body, occasionally accompanied by exudation of vitreous cells. These adverse events observed most likely attributed to the injection or/and the ophthalmological examination procedures. In addition, the administration of VGR-R01 exacerbated ocular inflammation responses that are consistent with those reported for other AAV-based subretinal therapies,²⁰ highlighting the immune challenges inherent to subretinal delivery of exogenous substances, like viral vectors or transplanted cells. Despite the immune protection afforded by the subretinal space, immune surveillance mechanisms remain active, and the delivery of AAV through this route can trigger responses that may amplify ophthalmological changes induced by the surgical procedures. In clinical practice, the injections will be performed by experienced ophthalmologists using smaller needles (38G compared to 32G and 33G used in the rabbits and monkeys), which would likely result in less damages. In addition, co-administration of steroids could be applied to mitigate the inflammatory responses. Therefore, while the ophthalmological safety risks from subretinal AAV administration are significant, they can be effectively managed and must be carefully assessed during clinical trials.

In conclusion, the preclinical studies presented here provide promising data to support the clinical development of VGR-R01, a gene therapy by subretinal administration of AAV8 vectors to deliver a functional human CYP4V2 gene for treating BCD. In vitro, VGR-R01 demonstrated dose-dependent expression of the CYP4V2 protein and elevated fatty acid hydroxylase activity that results in the reduced lipid droplet accumulations in the RPE cells. In vivo, VGR-R01 was effective in alleviating the retinal dysfunction of Cyp4v3^−/− mice and was shown to be safe and well tolerated in the New Zealand rabbits and NHPs. The ongoing phase 1/2 clinical trial (NCT05694598) aims to further assess the therapeutic benefits and safety profile of VGR-R01 in BCD patients, potentially offering a new treatment avenue for this challenging genetic disorder.

Materials and methods

Plasmid construction

Wild-type or codon-optimized human CYP4V2 coding sequences which contained Kozak sequence on their 5′- end were commercially purchased from Genewiz and others. An ssAAV construct with an expression backbone which had a CAG promoter, a WPRE and an SV40 polyA were used. Kozak sequence together with the coding sequences were cut and ligated to the ssAAV backbone, between the CAG promoter and the WPRE sequence.

Cell culture and transfection

HEK293 and ARPE-19 cells were cultured in the Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and passaged every 3 days using TrypLE. The day prior to transfection, HEK293 cells were seeded into a 24-well plate at a density of 1 × 10⁵ cells/cm², while ARPE-19 cells were seeded at a density of 7 × 10⁴ cells/cm². Plasmids were transfected using Lipofectamine 3000 Transfection Reagent (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s instructions. After 72 h, cells were harvested in RIPA lysis buffer (Beyotime, Shanghai, China) supplemented with a protease inhibitor for western blot analysis.

AAV production

The VGR-R01 manufacturing process involved a three-plasmid transient transfection. The upstream process started with the HEK293 cells (from working cell bank) recovery, followed by cell expansion in shaker flask, WAVE, and 50 L bioreactors. Then, the cells were then transfected by transfer plasmid, helper plasmid, and packaging plasmid. The virus was assembled in the cells and partially released into the cell culture supernatant, the crude AAV was harvested by lysis of cells, and clarification was done by depth filtration. The downstream process adopted scalable industrial chromatography techniques. First, the AAV particles were captured by affinity chromatography, and some process impurities were removed. Then, the empty capsid AAV and the full viruses were separated by anion exchange chromatography to obtain high-purity AAV products. Subsequently, the buffer solution was replaced by tangential flow filtration concentration and buffer exchange, and the AAV was transferred into the final formulation buffer. Finally, the AAV products were obtained by sterilization filtration and filling, and the products were stored at −80°C.

Cell culture and AAV transduction

HEK293, HepG2, SH-SY5Y, and ARPE-19 cells were cultured in DMEM supplemented with 10% FBS. A549 cells were cultured in F-12K Nutrient Mixture supplemented with 10% FBS. After digestion and centrifugation, cells were resuspended and cultured overnight with the complete medium containing 2 mM hydroxyurea (Sigma-Aldrich, St. Louis, MO). Cells were seeded at 1 × 10⁵/well in a 24-well plate and transduced with VGR-R01 at MOIs of 1 × 10⁵, 5 × 10⁵, 2 × 10⁶, and 5 × 10⁶ vg/cell. After an incubation of 48–72 h, cells were harvested in RIPA lysis buffer (Thermo Fisher Scientific, Waltham, MA) supplemented with a protease inhibitor for western blot analysis.

The differentiated RPE cells (CYP4V2^+/+ and CYP4V2^−/− RPE cells) were cultured in the medium which contained 5% KOSR in DMEM and then seeded into a 96-well plate at a density of 1 × 10⁵ cells/well. The following day, these RPE cells were administered with VGR-R01 (MOI: 1 × 10⁵, 3 × 10⁵, and 1 × 10⁶ vg/cell) or AAV8-EGFP (PackGene, Guangzhou, China; MOI: 1 × 10⁶ vg/cell). 27 days after AAV treatment, the supernatant was aspirated, and the cells underwent a single wash with PBS before the addition of 50 μL of Accutase (STEMCELL Technologies, Vancouver, Canada), followed by incubation at 37°C. The cells were then harvested in RIPA lysis buffer supplemented with a protease inhibitor for western blot analysis.

Hydroxylation enzyme activity of the CYP4V2 protein

ARPE-19 cells were cultured and transduced with VGR-R01 at MOIs of 1 × 10⁵, 5 × 10⁵, 2 × 10⁶, and 5 × 10⁶ vg/cell as mentioned previously. After a transduction of 144–168 h, cells were harvested in an ice bath and subjected to sonication using a cell disruptor. Subsequently, the protein concentration was determined using the BCA Protein Quantification Kit (YEASEN, Shanghai, China), and the total protein concentration was adjusted to 1,000 μg/mL with PBS. The cell lysate was subsequently incubated with a series of cofactors including cytochrome b5 (Sigma-Aldrich, St. Louis, MO), cytochrome P450 reductase (R&D, Minneapolis, MN), and the reduced form of coenzyme II, NADPH (Sangon Biotech, Shanghai, China) at a molar ratio of 1:2:1, respectively, along with either lauric acid or myristic acid for durations of 0 or 120 min. Hydroxylation products were identified using LC-MS/MS, and the hydroxylation enzyme activity was determined based on the rate of product formation. The LC-MS/MS detection method is briefly described as follows: 50% methanol and internal standard working solution (30 ng/mL diclofenac in 5% trifluoroacetic acid) were added in the samples. After vortexing, all samples were treated with methyl tert-butyl ether and finally reconstituted with 30% methanol-water solution for analysis. The mobile phase consisted of solution A, containing 0.1% formic acid in water, and solution B, containing 0.1% formic acid in acetonitrile.

Differentiation of PSCs into RPE cells

The human embryonic stem cell (ESC) line H9 was used as a wild-type control. The peripheral blood mononuclear cells from a BCD patient with the CYP4V2 genotype of IVS6-8del17bp/insGC; c.958C>T were separated and induced to iPSC utilizing the Sendai virus. All the aforementioned cells were gifts from the Institute of Zoology, Chinese Academy of Sciences. The ESC and iPSC were cultured in the Essential 8 medium (Thermo Fisher Scientific, Waltham, MA). Differentiation of ESC or iPSC into RPE cells was performed as previously described.¹² Briefly, the N2B27 medium was conducted 7 days prior. On day 7, the cells were transferred to the medium which contained 10% KOSR in DMEM with 50-ng/mL recombinant human activin A until the appearance of the pigment spots. Then the cells were switched to the medium which contained 10% KOSR in DMEM until approximately 35%–40% of the cell culture’s bottom surface area displayed pigment deposition. Following the coating of the transwell, the cells were resuspended in 1-mL Accutase (STEMCELL Technologies, Vancouver, Canada), and the cell suspension was filtered through a 40-μm cell strainer. After 3 days, the cells were transferred to the medium which contained 5% KOSR in DMEM for future investigation.

Lipid staining

The differentiated RPE cells (CYP4V2^+/+ and CYP4V2^−/− RPE cells) were seeded into 4-well chamber slides at a density of 6 × 10⁵ cells per well. The following day, these RPE cells were transduced with VGR-R01 at an MOI of 5 × 10⁵ vg per cell. 23 days later, the cells were stained with BODIPY 493/503 (Thermo Fisher Scientific, Waltham, MA) at room temperature and protected from light for 15 min. Following washing, the chamber slides were mounted onto glass slides using anti-fluorescence quenching reagent and observed under a fluorescence confocal microscope for imaging. Integrated densities, calculated by mean fluorescence × area, were shown in bar graph. Data are depicted as the mean and SD. Statistical analysis was performed and plotted using GraphPad. Ordinary statistical comparisons between groups were conducted using one-way ANOVA.

Western blot

The cell lysate was mixed with 5× SDS-PAGE loading buffer and denatured for 15 min at 95°C. Samples were separated in 4%–20% SDS-PAGE gel (Genscript, Nanjing, China) and blotted onto polyvinylidene fluoride membrane, followed by blocking for 1 h at room temperature. The protein expression levels of CYP4V2 and the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were detected by antibodies against the human CYP4V2 (Sigma-Aldrich, St. Louis, MO) and GAPDH (Abcam, Waltham, MA), respectively. After washing, the CYP4V2 and GAPDH bands were detected and quantitated by adding the enhanced chemiluminescence exposure reagent. Histogram was constructed according to the gray values of the protein bands. Data are depicted as the mean and SD.

In vivo Cyp4v3^−/− mouse study

Details of the Cyp4v3^−/− mice, established by State Key Laboratory of Stem Cell and Reproductive Biology, Institute of Zoology, Chinese Academy of Sciences (Beijing, China), are described in the previous study.¹⁰ The Cyp4v3^−/− mice were bred by Shanghai Southern Model Biological Technology Co., Ltd.

Forty 8-month-old Cyp4v3^−/− mice (20 mice per sex) were randomly divided into two groups with 10 mice per sex in each group. These mice were given 0 (vehicle control) or 4.0 × 10⁷ vg/eye of VGR-R01 through a single unilateral (the right eye) subretinal injection at 1 μL/eye. In addition, a total of eighteen C57BL/6J mice, including 8 females and 10 males, were chosen as the wild-type control group and were given the vehicle control through a single unilateral (the right eye) subretinal injection. After injection, all mice were observed for a duration of 9 months. The endpoints included clinical observations, body weight, and detailed ophthalmological examinations, including general ophthalmological examination, OCT by Heidelberg laser ophthalmic diagnostic instrument, ERG by Diagnosys visual electrophysiology instrument, intraocular pressure by TonoLab tonometer, FP and AF by a CRO fundus imaging machine. All mice were housed at a pathogen-free (SPF) facility, with unrestricted access to feed (Co⁶⁰-irradiated SPF rodent maintenance feed) and water, following a 12-h light/dark cycle. All animal experiments were conducted in strict compliance with institutional policies and guidance principles regarding animal welfare. Data was plotted using GraphPad. SPSS Statistics 21.0 was used for ERG parameters statistical analysis. One-way ANOVA was used in the analysis between groups.

New Zealand rabbit toxicity study

A 13-week good laboratory practice (GLP) toxicity study of VGR-R01 was performed in New Zealand rabbits. New Zealand rabbits were obtained from Qingdao Kangda Biotechnology Co., Ltd. and housed at the JOINN Laboratories (Suzhou) Co., Ltd. A total of 40 rabbits (20/sex, 3–4 months old) were randomly assigned to four groups, with five per sex per group, and were administered 0 (group 1, vehicle control), 1.6 × 10¹⁰ (group 2, low dose), or 1.6 × 10¹¹ vg/eye (groups 3 and 4, high dose) of VGR-R01 through a single unilateral (the left eye) subretinal injection with a dose volume of 40 μL/eye. Following injection, the rabbits were observed for 4 or 13 weeks and necropsied on day 29 or day 92. Throughout the study, the animals underwent regular examinations for clinical observations, body weight, food consumption, clinical pathology (including hematology coagulation, serum chemistry, and urinalysis), ophthalmoscopic assessments (including general ocular examinations, intraocular pressure measurements, ERG, OCT, and FP), and pathology. A psychologically and physically comfortable environment was maintained for the animals, with a room temperature ranging from 18°C to 26°C, humidity maintained between 40% and 70%, and a 12-h light/dark cycle. Animal care was compliant with the SOPs of JOINN Laboratories (Suzhou) and the related Guides. The protocol of animal use in this study was approved by the Institutional Animal Care and Use Committee (IACUC) at JOINN Laboratories (Suzhou).

Nonhuman primate toxicity study

A 13-week GLP toxicity study of VGR-R01 was performed in cynomolgus monkeys. Cynomolgus monkeys were obtained from Guangxi Grandforest Scientific Primate Co., Ltd. and housed at JOINN Laboratories (Suzhou) Co., Ltd. A total of 30 monkeys (15/sex, 2.9–4.1 years old) were randomly assigned to three groups, with five per sex per group, and were administered 0 (vehicle control), 2.4 × 10¹¹, or 2.4 × 10¹² vg/eye of VGR-R01 through a single bilateral (both eyes) subretinal injection with a dose volume of 60 μL/eye. Following injection, the monkeys were observed for 4 or 13 weeks and necropsied on day 29 or day 92. Throughout the study, the animals underwent regular examinations for clinical observations, body weight, food consumption, clinical pathology (including hematology coagulation, serum chemistry, and urinalysis), ophthalmoscopic assessments (comprising general ocular examinations, intraocular pressure measurements, ERG, OCT, and FP), and pathology. A psychologically and physically comfortable environment was maintained for the animals, with a room temperature ranging from 18°C to 26°C, humidity maintained between 40% and 70%, and a 12-h light/dark cycle. Animal care was compliant with the SOPs of JOINN Laboratories (Suzhou) and the related guides. The method of animal use in this study was approved by the IACUC at JOINN Laboratories (Suzhou).

Subretinal injection

In Cyp4v3^−/− mice, the eyes for administration (the right eye) were prepared with a 2.5% povidone iodine solution and rinsed with sterile saline. The subretinal bleb was created in the retina using a needle (NanoFil, 36G) and then followed by the injection process. For the injection, the needle was inserted through the conjunctiva and sclera, gently contacting the retina. Upon insertion of the beveled-tip into the superficial retinal layer, the force of the test article or the vehicle control article stream emerging from the fine-diameter cannula punctured the neural retina to establish initial access, subsequently expanding the subretinal space.

Similar subretinal injection procedures were performed for the New Zealand rabbits and cynomolgus monkeys. Briefly, animals were given a mydriatic agent and then anesthetized using Zoletil 50 via intramuscular injection before the subretinal injection. Subsequently, ophthalmic surgical scissors were used to cut and open the conjunctival sac at the injection site of the eye for administration (the left eye for rabbits and both eyes for monkeys). A needle was used to puncture the stoma of the perforating membrane, and then a flat-end injection needle (NanoFil, 32G for rabbits and 33G for monkeys) attached to a microinjector was inserted through the upper scleral wound into the vitreous cavity. Medical hyaluronan gel or sterile saline was instilled into the cornea, a retinoscope was positioned on the cornea, and then the subretinal cavity was injected at the tapetal retina area within 15–25 s. Following the completion of the subretinal injection, the injection site was immediately checked under the microscope for successful administration, and the closure of the scleral stoma was observed. FP and OCT were then performed immediately to confirm the successful administration.

Electroretinograms

In Cyp4v3^−/− mice, ERGs were recorded using the Espion V6 recording system (Diagnosys LLC, Lowell, MA) following the International Society for Clinical Electrophysiology of Vision (ISCEV) standard. Prior to ERG examination, all animals underwent overnight dark adaptation. Subsequently, after pupil dilation with tropicamide phenylephrine eye drops, the anesthetized mouse was positioned on the stage and electrode placement was ensured. Scotopic recordings were conducted on dark-adapted mice at incremental light intensities of 0.01, 0.1, and 1 cd s/m² as per the 3-step protocol, with three sweeps recorded for each test. Photopic recordings were performed following 10 min of light adaptation intervals under a background light intensity maintained throughout photopic recordings. Photopic recordings were conducted at light intensities of 3 and 10 cd s/m² for two steps, with ten sweeps recorded for each test. Additionally, two extra flash intensities of 3 and 10 cd s/m² were recorded for two steps, with thirty sweeps recorded for each test. Statistical analysis was performed on the amplitudes of scotopic and photopic a- and b-waves.

In New Zealand rabbits and cynomolgus monkeys, full-field ERG examinations were performed employing an RETI-SCAN 21 SLO+ERG System (Roland Consult, Brandenburg an der Havel, Germany). Animals were placed on dark background for at least 30 min for dark adaptation immediately after sedation for the examination and topical instillation of tropicamide phenylephrine eye drops (25 mg/mL). A rod-specific response was elicited by a dim white flash (0.01 cd s/m²) after dark adaptation and pupil dilation. Mixed rod and cone responses and oscillatory potentials ERG were obtained using the standard bright white flashes (3.0 cd s/m²) under scotopic condition. To evaluate the function of the cone photoreceptors, monkeys were light adapted for at least 10 min, then the photopic ERGs (3.0 cd s/m² and 10.0 cd s/m²) and photopic flicker ERG (30 Hz) were elicited on the background light.

Biodistribution and shedding

In New Zealand rabbits, the biodistribution was assessed in the animals from the high dose group. Blood was collected prior to dosing, at 1–2 h post-dosing, and on D2, D4, D8, D29, and D92. Tissues (eye tissues and other major tissues/organs) were collected on D29 and D92. Samples were analyzed to detect CYP4V2 DNA using quantitative polymerase chain reaction (qPCR) method. Samples positive for CYP4V2 DNA were then further analyzed to detect CYP4V2 mRNA using quantitative reverse transcription polymerase chain reaction (RT-qPCR) method.

In cynomolgus monkeys, the biodistribution and the shedding were assessed. Blood was collected prior to dosing, at 1–2 h post-dosing, and on D2, D4, D8, D29, and D92. Tissues (eye tissues and other major tissues/organs) were collected on D29 and D92. In addition, shedding samples were collected prior to dosing, and on D2, D4, D8, D27–28, and D90–91. Samples were then analyzed same as the New Zealand rabbits.

DNA was extracted from the blood, tissues, and shedding samples and purified using a DNeasy Blood & Tissue Kit (Qiagen, Germantown, MD) following the manufacturer’s protocol and subsequently dissolved in nuclease-free water. The concentration of DNA was quantified via spectrophotometric analysis using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA). qPCR was employed to determine the copy number of target DNA on a LightCycler 480 Ⅱ (Roche, Indianapolis, IN) using TaqMan probes and the AceQ universal U⁺ Probe Master Mix V2 (Vazyme, Beijing, China) in a 20-μL reaction mixture. Negative control reactions were included in each run. The concentrations of the PCR products were interpolated from the cycle threshold (CT) values, and the duplicate concentration values were averaged.

Samples positive for CYP4V2 DNA were then further analyzed to detect CYP4V2 mRNA. Total RNA was extracted and purified using TRIzol Reagent (Thermo Fisher Scientific, Waltham, MA) following the manufacturer’s instructions and dissolved in DEPC-treated water. RT-qPCR was utilized to quantify the copy number of CYP4V2 mRNA on a LightCycler 480 Ⅱ using the same TaqMan probes as the same in the determination the copy number of target DNA. The HisScript II U+ One-Step RT-qPCR Probe Kit (Vazyme, Beijing, China) and a 20-μL reaction mixture were employed. Negative control reactions were included in each run. Concentrations were interpolated from the CT values, and the duplicate concentration values were averaged. Statistical analysis was performed and plotted using GraphPad.

Data availability

All of the study data generated in the article and/or supplemental information during the present study are available from the corresponding author on reasonable request.

Acknowledgments

We thank the Institute of Zoology, Chinese Academy of Sciences for providing cells for our research. We gratefully acknowledge the State Key Laboratory of Stem Cell and Reproductive Biology, Institute of Zoology, Chinese Academy of Sciences for the Cyp4v3^−/− model mouse. Graphical abstract was drawn by Figdraw. We also acknowledge the Science and Technology Commission of Shanghai Municipality (23S11903400).

Author contributions

L.L. and X. Zhu designed experiments, supervised data analysis and results interpretation, and revised the manuscript. L.G. and W.L. conducted experiments, analyzed data, and wrote the manuscript. N.H. managed the project and analyzed data. Y.T. provided experimental support and suggestions during manuscript development. S.-S.T. provided experimental support and suggestions during manuscript development and revised the paper. Y.Z. and Y.C. conducted experiments and analyzed data. X. Zhao provided critical insights.

Declaration of interests

The authors declare no competing interests.