Preclinical assessment of splicing modulation therapy for ABCA4 variant c.768G>T in Stargardt disease

Abstract

Background: Stargardt disease type 1 (STGD1) is a progressive retinal disorder caused by bi-allelic variants in the ABCA4 gene. A recurrent variant at the exon-intron junction of exon 6, c.768G>T, causes a 35-nt elongation of exon 6 that leads to premature termination of protein synthesis.

Methods: To correct this aberrant splicing, twenty-five 2'-O-methoxyethyl antisense oligonucleotides (AONs) were designed, spanning the entire exon elongation.

Results: Testing of these AONs in patient-derived photoreceptor precursor cells and retinal organoids allow the selection of a lead candidate AON (A7 21-mer) that rescues on average 52% and 50% expression of wild-type ABCA4 transcript and protein, respectively. In situ hybridization and probe-based ELISA demonstrate its distribution and stability in vitro and in vivo. No major safety concerns regarding off-targets, immunostimulation and toxicity are observed in transcriptomics analysis, cytokine stimulation assays in human primary immune cells, and cytotoxicity assays.

Conclusions: Additional optimization and in vivo studies will be performed to further investigate the lead candidate. Considering the high prevalence of this variant, a substantial number of patients are likely to benefit from a successful further development and implementation of this therapy.

Fig. 1. The effect of ABCA4 c.768G>T variant on splicing.

a A schematic overview of splice donor sites (SDS) of the correct and aberrant transcript. The variant c.768G>T is indicated and the score values assessed by various prediction tools are shown above the SDS. b RT-PCR analysis showing the effect of the c.768G>T splicing in fibroblasts, PPCs and ROs. PPCs: photoreceptor precursor cells, ROs: retinal organoids, CON: control individual, PAT: patient-derived material, ISO: isogenic control of patient-derived material, CHX: cycloheximide. c A schematic overview of the region included in the oligo walk. The variant c.768G>T is indicated and AON sequences are listed in Supplementary Table 4.

Fig. 2. AON screening in PPCs and ROs.

a A schematic overview of the differentiation procedure. Photoreceptor precursor cells (PPCs) are differentiated for 30 d, whereas retinal organoids (ROs) are differentiated for at least 180 d. b Representative RT-PCR analysis of PPCs treated with 2’MOE/PS AONs targeting c.768G>T for 10 d at 5 μM final concentration (n = 2). c Fiji semi-quantification of WT transcript level based on RT-PCR analysis in (b). d Representative RT-PCR of ROs treated with 2’MOE/PS AONs targeting c.768G>T for 10 d at 10 μM final concentration (n = 2). e Fiji semi-quantification of WT transcript level based on RT-PCR analysis in (d). b, d ISO denotes isogenic control and PAT represents patient-derived PPCs/ROs. Three different transcripts were detected, aberrant transcript containing 35-nt exon 6 elongation, correct/wild-type transcript, and ex 6 del referring to a transcript containing a 96-bp deletion of exon 6 (c.673-c.768). ACTB was used as loading control. c, e The %WT transcript level illustrated is based on the ratio of aberrant, correct, and partial exon 6 deletion transcripts in each sample, with total signal quantified as 100%. The semi-quantification was performed on at least two PCR products per biological replicate. Each dot denotes the mean of technical replicates of one biological replicate. Significance was calculated by one-way ANOVA followed by post-hoc Dunnett test comparing each AON-treated sample to SON.

Fig. 3. A7 21-mer showed the best potency in rescuing ABCA4 c.768G>T at RNA and protein level.

A RT-PCR analysis of ROs treated with different lengths and concentrations of A7 for 10 d. ISO denotes isogenic control and PAT represents patient-derived ROs. Three different transcripts were detected, aberrant transcript containing 35-nt exon 6 elongation, correct/wild-type transcript, and ex 6 del referring to a transcript containing the 96-bp deletion of exon 6 (c.673-c.768). ACTB was used a loading control. B Western blot analysis of ROs treated with A7 19, 20 and 21-mer for 30 d at 10 μM final concentration. C Fiji semi-quantification of ABCA4 expression based on the band at ~250 kDa, normalized to β-TUBULIN expression, relative to Iiogenic control. Each dot denotes the mean of technical replicates of one biological replicate. Significance was calculated by one-way ANOVA followed by post-hoc Dunnett test comparing each AON-treated sample to SON. D Immunohistochemal analysis of ABCA4 localization (red, 5B4 antibody) upon 30 d treatment with 10 μM A7 21-mer. PNA (green) is used to mark the outer segments of the ROs, whereas DAPI was used as nuclear marker. Representative images from three biological replicates are shown. Scale bar equals to 50 μm.

Fig. 4. A7 is stable both in vitro and in vivo.

a Detection of A7 21-mer (red) in ROs 10 and 30 d post-delivery by means of in situ hybridization, counterstained with DAPI as nuclear marker. Scale bar equals 50 μm. b Detection of A7 21-mer in C57BL/6 J mouse retina 1 week post-injection by means of in situ hybridization. Scale bar equals 100 μm. ONL: outer nuclear layer, INL: inner nuclear layer. c Quantification of A7 21-mer level in C57BL/6 J mouse retina over time (W: week) employing probe-based ELISA. Each dot represents measurement from one retina.

Fig. 5. No potential off-targets that cause differential gene expression were identified in total transcriptomics analysis.

a Heatmap of top 500 variable genes comparing all samples included in the analysis. b MA-plots of pairwise comparison of differential gene expression. Blue dots represent genes differentially expressed in the pairwise comparison. c Volcano plots of pairwise comparison of differential splicing. Blue dots represent genes with deltapsi <−0.3 or >0.3 (data: Supplementary Data 3), red dashed-line denotes FDR threshold of 0.05. Arrows indicate the aberrant transcript associated with ABCA4 c.768 G > T and the correct transcript. ROs were treated for 10 d prior to analysis. ISO: isogenic control, PAT: patient-derived ROs, AON: A7 21-mer, SON: sense oligonucleotides of A7 21-mer. Data presented in b, c were generated from two biological replicates shown in (a).

Fig. 6. Safety evaluation of A7.

a Heatmap of fold-change cytokines and chemokines released upon 24 h PBMC stimulation with A7 21-mer or controls relative to vehicle (PBS). Geometric mean from 18 healthy donors are presented, with the exception for TNFα (9 donors). b Cytotoxicity measured by LDH-release 24 h and 10 d post-delivery into ROs. Three independent experiments including 3 ROs per experimental condition are presented. DOX: doxorubicin, cytotoxic compound used as positive control. For both assessments, significance was calculated by one-way ANOVA followed by post-hoc Dunnett test comparing each AON-treated sample to vehicle (PBS).