Nanopore Deep Sequencing as a Tool to Characterize and Quantify Aberrant Splicing Caused by Variants in Inherited Retinal Dystrophy Genes

The contribution of splicing variants to molecular diagnostics of inherited diseases is reported to be less than 10%. This figure is likely an underestimation due to several factors including difficulty in predicting the effect of such variants, the need for functional assays, and the inability to detect them (depending on their locations and the sequencing technology used). The aim of this study was to assess the utility of Nanopore sequencing in characterizing and quantifying aberrant splicing events. For this purpose, we selected 19 candidate splicing variants that were identified in patients affected by inherited retinal dystrophies. Several in silico tools were deployed to predict the nature and estimate the magnitude of variant-induced aberrant splicing events. Minigene assay or whole blood-derived cDNA was used to functionally characterize the variants. PCR amplification of minigene-specific cDNA or the target gene in blood cDNA, combined with Nanopore sequencing, was used to identify the resulting transcripts. Thirteen out of nineteen variants caused aberrant splicing events, including cryptic splice site activation, exon skipping, pseudoexon inclusion, or a combination of these. Nanopore sequencing allowed for the identification of full-length transcripts and their precise quantification, which were often in accord with in silico predictions. The method detected reliably low-abundant transcripts, which would not be detected by conventional strategies, such as RT-PCR followed by Sanger sequencing.


Introduction
Inherited retinal dystrophies (IRDs) constitute a group of conditions affecting the retina, which is a thin layer of neuronal cells lining the back of the eye.These disorders are characterized by deterioration or congenital dysfunction of the photoreceptors or the retinal pigment epithelium, which ultimately results in impaired vision or blindness.The genetic basis of IRDs is highly heterogeneous, with variants in over 300 loci known to lead to these disorders (RetNet, https://sph.uth.edu/RetNet/accessed on 28 June 2024).Molecular diagnostics for this group of conditions is further aggravated by high clinical (phenotypic) heterogeneity [1].Despite this complexity, recent studies (2018-2022) reported an overall diagnostic yield for mixed IRDs of 64.2% [2].This figure improves to 73.5% when considering exclusively studies using exome sequencing (WES) [2].
Whole-gene sequencing and functional characterization of candidate splicing variants in ABCA4 have been the subject of extensive efforts for the resolution of missing heritability in IRDs, which led to the identification of many pathogenic variants that WES would not detect [6][7][8][9][10][11][12][13][14][15][16][17].These studies have identified and characterized many deepintronic and non-canonical splice site variants affecting splicing.Similarly, a recent study curated pathogenicity classification for the 2246 ABCA4 variants reported within the LOVD database [18].They classified 1248 variants to be likely pathogenic or pathogenic; among these, 254 (20.4%) variants may affect splicing, including 52 (4.2%) non-canonical or deepintronic variants [18].All of these studies demonstrated that splicing variants remain underreported in the literature due to the difficulty in predicting their effects in silico and the inability to detect them, as deep-intronic regions are not covered by WES.
A multitude of in silico tools have been developed to predict the impact of a variant on splicing [19][20][21][22][23][24][25][26].The performance of some of these tools has been benchmarked with splicing variants in the ABCA4, MYBPC3, and NF1 genes that had been functionally characterized [27,28].These studies found that deep learning tools (areas under the curve (AUCs) of 0.72-0.99)often outperform classical machine learning (AUCs of 0.69-0.80)and motif-based tools (AUCs of 0.72-0.86)[27,28].However, Riepe et al. found that the variant context played an important role in determining which was the best-performing tool [28].
Splicing prediction tool scores can help in genetic testing during variant filtering and prioritization.However, in the absence of functional assays, these scores are insufficient for categorizing variants outside of canonical splice sites as likely pathogenic or pathogenic [29].Depending on the accessibility of the tissue expressing the mutated gene, reverse-transcriptase (RT)-PCR or RNA-seq can provide insights into potential splicing defects in patient-derived cells [30].Alternatively, minigene constructs containing the genomic region surrounding the candidate variant can enable the identification and characterization of aberrant splicing [31].Many genes associated with IRD pathogenesis are not stably expressed in readily accessible tissues, such as whole blood [5], which makes minigene assays particularly interesting for variants in these genes [6,17,[32][33][34][35][36][37][38][39].
The aim of this study was to test the performance of Nanopore deep sequencing for the characterization and quantification of variant-induced aberrant splicing events.For this purpose, we selected 19 candidate variants that may affect splicing, which were detected in patients affected by IRDs.We functionally characterized these variants using minigene assay or patient-derived peripheral blood RT-PCR, combined with Nanopore sequencing, to identify and quantify alternative splicing products.The results of these functional readouts were compared to the predictions obtained with in silico tools.Thirteen of the nineteen variants led to aberrant splicing events and may have a negative impact on protein function.

Variant Selection
Nineteen rare variants that may affect splicing were selected for functional characterization by Nanopore sequencing (Table 1).These variants occur in genes expressed in the retina, previously associated with an IRD phenotype.The candidate variants were identified in IRD patients; some of these patients remained undiagnosed after WES, wholegene long-range PCR sequencing (LR-PCR), WGS, or a combination of these methods, as reported in previous studies [40,41].Demographic data of these patients is summarized in Table S8.

Splicing Predictions for Candidate Variants
All variants selected for functional characterization were assessed for possible effects on splicing by in silico predictions using several algorithms, including those present in Alamut ® Visual Plus, SpliceAI [19], and Pangolin [20] (Supplementary Tables S1-S3).Table 2 lists the findings, and the most likely variant-caused aberrant splicing events based on these predictions.The average variant-induced difference in splice site strengths computed by tools in Alamut Visual Plus will be referred to as the "Effect score".Supplementary Figure S1 provides a visual representation of the splicing predictions from Alamut ® Visual Plus for each variant and its flanking sequences.
Table 2.In silico splicing predictions summary.The EX-skip prediction column reports the chance (likelihood) of exon skipping computed by EX-skip; it compares the exonic splicing enhancer (ESE) and silencer (ESS) sequences in the reference and variant exons.The last column lists the conclusions based on integrating the predictions for the natural (canonical) splice sites and cryptic splice sites that may be affected by the variants, along with the ESE/ESS profiles.Based on the predictions, the variants in ABCA4, CACNA1F, CHM, OCA2, PDE6C, and PROM1 are expected to lead to partial exon skipping, as they only affect natural acceptor and/or donor sites.Similarly, the ATF6, IMPG2, KIF11, POC1B (NM_172240.2:c.677-2A>G), and REEP6 variants affect the natural splice sites; however, they may also influence cryptic splice sites, by either creating new or strengthening pre-existing sites, and most of them impact the exonic splicing enhancer (ESE) to exonic splicing silencer (ESS) binding sites ratios.As a result, these variants are predicted to cause partial exon skipping and/or partial usage of an alternative (cryptic) acceptor or donor splice site.In silico predictions suggest that the FZD4, POC1B (NM_172240.2:c.1033-327T>A), and RPGR variants create cryptic acceptor sites and could lead to partial usage of these alternative cryptic acceptor sites.Variant NM_172240.2:c.1033-327T>Aconcomitantly abolishes a natural acceptor splice site of the noncoding POC1B transcript NR_037659.2.Finally, the TIMP3 and USH2A variants only slightly affect nearby cryptic splice sites; however, both decrease the ESE/ESS ratio of the predicted pseudoexon, making its inclusion in the transcript more likely when compared to the reference sequence.

Minigene Assays for Candidate Variants
At least one minigene plasmid was successfully constructed for each variant (Table 3), except for the KIF11 variant.The KIF11 variant was functionally tested using whole blood cDNA from the affected family (see Section 2.4).Most minigene constructs (16/19) were based on the pcDNA3_RHO_ex3-5_plasmid (refer to Materials and Methods Section 4.5).The FZD4 minigene contains the entire FZD4 locus.For the ATF6 variants, minigenes containing exons 1, 2, and 9 or 13, with flanking introns, were constructed from three PCR products.For the IMPG2 variant, a large (including exons 15-18) and a minimal (including only exon 17) minigene were created.A circular view of the features of each plasmid is available in Supplementary Figure S2.
The expected major (WT) transcript (highlighted in green under the coverage plots in Figures A1-A20) was identified in all assays for the reference minigene.Its relative abundance as measured by Nanopore sequencing, however, varied greatly from only 1.1% to 98.7% of total reads.We also identified the transcript composed only of RHO exons in most reference minigene assays (10/16) at levels ranging from 1.6% to 80.9% (Figures A1-A20 and Tables A1-A20).
Reference minigenes that resulted in low abundance (<50%) of WT transcript included RHO_minigene_ATF6_int8-9 (32.5%, Figure A3 and Table A3), RHO_minigene_CACNA1F_ int14-18 (11.2%, Figure A6 and Table A6), RHO_minigene_IMPG2_int15-18 (1.1%, Figure A9 and Table A9), RHO_minigene_OCA2_int5-7 (17.6%, Figure A11 and Table A11), and RHO_minigene_PROM1_int23-26 (3.7%, Figure A16 and Table A16).In all these cases, another transcript resulting from at least one exon-skipping event was present in the reference minigene splicing assay.These findings can be partially explained by the splice site strengths of the skipped exons (Supplementary Table S4).In fact, ATF6 exon 9 has a relatively weak acceptor site (56%), CACNA1F exons 16 and 17 and OCA2 exons 6 and 7 are flanked by very weak splice sites, and the donor splice site defining IMPG2 exon 17 is relatively weak (44%).On the other hand, RHO exons 3 and 5 are characterized by strong donor and acceptor (average transformed Alamut scores of 75% and 67%, respectively).Exon skipping due to an imbalance in splice site strengths has been postulated previously in a similar study [17].This may explain the presence of the transcript composed only by RHO exons in most minigene assays.Additionally, shorter transcripts are preferentially amplified during PCR amplification, which can lead to a bias.
The minigene assays revealed aberrant splicing events for 12/18 variants (Table 4).Six additional variants in ABCA4 (NM_000350.Generally, the relative abundance of WT transcript in variant minigene assays was reduced when aberrant splicing events were present.However, comparing the relative abundance of WT transcripts can be misleading.The differences in WT transcript abundance between reference and variant minigenes for the CACNA1F (Figure A6 and Table A6) and one of the PROM1 (NM_006017.3:c.2490-2A>G, Figure A16 and Table A16) variants are small (−11.2% and −3.7%, respectively), but it is important to notice that the variant minigene showed complete depletion of the WT transcript.The delta value is low, only due to the low abundance of the WT transcript in the reference minigene assays in these cases.An overview of the gel electrophoresis results for RT-PCR products of each minigene can be accessed in Supplementary Figure S3.The ATF6 (NM_007348.3:c.1096-15G>A and NM_007348.3:c.1534-9A>G),PDE6C (NM_006204.3:c.864+1G>A),POC1B (NM_172240.2:c.677-2A>G), and RPGR (NM_ 001034853.1:c.1415-9A>G)variants were predicted to create or strengthen a cryptic acceptor or donor site and to weaken or disrupt the natural acceptor or donor splice site (Table 2).The minigene assays for these variants supported these predictions.
Variant NM_007348.3:c.1096-15G>A(ATF6) was functionally tested using two different minigene constructs: one based on the RHO backbone, and one containing the endogenous ATF6 exons 1 and 2 inserted upstream of ATF6 exon 9 (Table 3).The RHO_minigene_ATF6_int8-9 assay resulted in the identification of the predicted alternative acceptor splice site at position c.1096-13, being used in 2.8% of the variant minigene reads (Figure A3 and Table A3).Similarly, the ATF6_minigene_ex1-2-9 found the cryptic acceptor site in 6.1% of the reads.Additionally, this minigene assay identified another cryptic acceptor site at position c.159+275 (part of ATF6 intron 2) in 3.0% of the reads (Figure A4 and Table A4).The reduction of WT transcript for both assays is relatively low (+1.4% and −25.2%, respectively), which suggests that this variant may have a mild effect on splicing.A mild effect could be expected based on in silico predictions, which anticipated a 7.1% Effect score and 17% SpliceAI reduction in natural splice site strength (Supplementary Table S1), and the creation of the c.1096-13 acceptor splice site with a 25% Effect score and 20% SpliceAI and Pangolin scores (Supplementary Table S2).
The ATF6_minigene_ex1-2-13 assay for variant NM_007348.3:c.1534-9A>G(ATF6) confirmed the activation of an alternative acceptor splice site at position c.1534-8 that was used in 77.3% of the reads (Figure A5 and Table A5).The relative abundance of the WT transcript was reduced from 87.5% in the reference minigene to 6.9% in the variant minigene.The c.1534-8 acceptor splice site was expected, based on in silico predictions, with a strength of 64.5% Effect score, 89% SpliceAI, and 82% Pangolin scores (Supplementary Table S2).
The PDE6C variant (NM_006204.3:c.864+1G>A)abolished the weak natural donor site and was expected to cause exon 4 skipping.The assay, however, identified the usage of two cryptic donor sites instead in the variant minigene (Figure A12 and Table A12); one was used in 65.5% of the reads and it corresponds to position c.864+128 in intron 4 (extending exon 4 by 128 nucleotides), and the second one is located in exon 4 at position c.801 (shortening exon 4 by 63 nucleotides), present in 28.6% of the reads.
The POC1B variant NM_172240.2:c.677-2A>Gabolished the natural acceptor splice site of exon 7 and is predicted to create an alternative weak acceptor splice site (8.9%Effect score) at position c.684.Nanopore sequencing of the minigene cDNA confirmed the use of the predicted alternative acceptor splice site in 86.2% of the variant minigene reads, with no WT transcript detected (Figure A13 and Table A13).
The findings of the RHO_minigene_RPGR_int10-13 construct, characterized using gel electrophoresis and Sanger sequencing, were published in Koller et al., 2023 [39].Briefly, we reported that the RPGR variant (NM_001034853.1:c.1415-9A>G)led to the extension of exon 12 by 8 nucleotides at the 5 ′ end in the vast majority of the transcripts.We also reported that partial exon 12 skipping was detected in the reference and variant minigene results.Nanopore sequencing of the minigene cDNAs revealed a similar, but more complex, set of transcripts (Figure A18 and Table A18).The alternative acceptor site is part of 70.5% of the transcripts in the variant minigene (transcripts T9, T10, and T11).Exon 12 skipping was found in 4.6% and 14.6% of reference and variant minigene transcripts (transcripts T4, T6, T12, and T13), respectively.A shorter exon 12 (using a cryptic acceptor site at position c.1427) was detected in 3.0% of the reference minigene transcripts (transcripts T5 and T8).Additionally, exon 11 was found to be alternatively spliced by using two alternative acceptor splice sites at positions c.1337 and c.1390.Finally, Nanopore sequencing revealed evidence of exon 13 skipping in the variant minigene transcripts T12 and T14, representing 1.4% of the reads.
The NM_000390.4:c.1413G>C(CHM) missense variant is located at the exon-intron boundary of CHM exon 11 and is predicted to weaken the natural donor splice site (Table 2).The sequencing results highlighted complete exon 11 skipping for the variant minigene (Figure A7 and Table A7).
The synonymous variant in exon 23 of PROM1 (NM_006017.3:c.2358C>T)influences the exonic splicing enhancer (ESE) and silencer binding sequences (ESS) ratio, leading to the variant exon being more likely skipped during splicing (Table 2 and Supplementary Table S3).While the Effect score on natural splice sites is marginal, SpliceAI and Pangolin estimate exon splicing loss with a chance of about 27.7% (Supplementary Table S1).The assay revealed exon 23 was skipped in 40.9% of the variant minigene transcripts (Figure A15 and Table A15).
Based on in silico predictions, the NM_005183.4:c.2239+5C>Gvariant upstream of CACNA1F exon 16 is expected to cause partial exon skipping (Table 2) as SpliceAI computes a donor loss likelihood of 23% and Pangolin predicts splice loss with a score of 47% (Supplementary Table S1).Transcripts analysis revealed 18 unique transcripts represented by at least 0.5% of the reads (Figure A6 and Table A6).The main variant-induced aberrant splicing events were increased exon 16 skipping and alternative acceptor splice site for exon 17 (transcripts T1, T2, and T8) with 45.9% against 71.6% of reads in the reference versus variant minigene, respectively (Figure A6 and Table A6).The WT transcript could not be detected in the variant minigene.
Variant NM_016247.4:c.3423-7_3423-4del was functionally tested using two minigene constructs based on the RHO backbone: one containing IMPG2 exons 16-18 and a minimal minigene containing only exon 17 and flanking introns (Table 3).The variant is predicted to weaken the natural acceptor site of exon 17 (−19.7%Effect score) and to strengthen a cryptic acceptor splice site (+1.1% Effect score) of 80 nucleotides upstream of exon 17 (Table 2 and Supplementary Tables S1 and S2).The RHO_minigene_IMPG2_int15-18 minigene resulted in increased exons 16-17 skipping for the variant minigene (52.2% against 36.7% in the reference minigene; Figure A9 and Table A9).Sequencing of the cDNA from the reference minigene also highlighted unexpectedly low levels of WT transcript (1.1%).The results for the minimal minigene (RHO_minigene_IMPG2_int16-17) supported increased exon 17 skipping as a variant-dependent effect (67.6% against 42.2% in the reference minigene); however, they also revealed a transcript using the cryptic acceptor site located at c.3423-80 in 18.6% of the reads (Figure A10 and Table A10).The WT transcript was drastically reduced from 56.1% in the reference minigene, as opposed to 9.2% in the variant minigene (Figure A10 and Table A10).
The NM_001329556.3:c.517G>Amissense variant lies at the exon-intron 4 boundary of REEP6.It severely weakens the natural donor splice site (−52.1% Effect score) and moderately strengthens a cryptic donor splice site, located at c.517+4 (+1.2%Effect score; Supplementary Tables S1 and S2).The assay demonstrated evidence of exon 4 skipping (56.3% of the variant minigene reads), and the use of an alternative donor splice site located at c.517+43 in a transcript, representing 10.4% of the reads (Figure A17 and Table A17).The alternative donor splice site c.517+43 was predicted by SpliceAI and Pangolin with scores of 46% and 20%, respectively (Supplementary Table S2).The WT transcript was severely reduced in the variant minigene (10.6% against 66.6% in the reference minigene).

Pseudoexon Inclusion in POC1B
Variant NM_172240.2:c.1033-327T>A is predicted to completely disrupt the natural acceptor site of exon 9 of the noncoding POC1B transcript NR_037659.2 and to create a stronger cryptic acceptor site, located 5 nucleotides upstream of it (c.1033-325).The minigene assay revealed that the WT transcript was strongly reduced in the variant minigene (43.3% against 91.3% in the reference minigene; Figure A14 and Table A14) and that the predicted novel acceptor site (c.1033-325) was used in 38.3% of the reads in combination with the pre-existing natural donor splice site from exon 9 of POC1B transcript NR_037659.2(Figure A14 and Table A14).Therefore, a large portion of the transcripts include a pseudoexon (or an elongated version of exon 9 of NR_037659.2),which is not part of the "normal" splicing of the protein-coding transcripts.This variant has been functionally characterized previously with a minigene assay and patient-derived blood cDNA.Both assays revealed pseudoexon (or the elongated exon 9 of NR_037659.2) inclusion as a consequence of the variant [42].

Splicing Assays on Blood cDNA for KIF11 and CACNA1F
The KIF11 variant (NM_004523.3:c.1875+2T>A) was functionally characterized using cDNA derived from the blood of three family members; the index patient and both parents.The index patient and the mother carry the variant heterozygously; the father is homozygous for the major allele at this position.The variant is located at the natural dinucleotide donor splice site of KIF11 exon 14 and is predicted to lead to a 91% chance of donor loss and 6% chance of donor gain at position c.1785 by SpliceAI (Supplementary Tables S1 and S2).
As expected, the sequencing results for the father revealed the WT transcript in at least 90.0% of reads (Figure 1 and Table 5).An additional 6.8% of the reads were lacking either exon 13 or 16 (transcripts T2-T5), which probably represents incomplete reads, as the PCR used primers binding to these exons for the amplification from cDNA.Therefore, it is likely that these reads represent incompletely sequenced transcripts.
Table 5. Transcript identification and quantification for the KIF11 variant M_004523.3:c.1875+2T>A for reference (WT) and variant (MT) sequences.The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Length WT (%) MT (%) ∆ MT-WT (%)
Counts WT Transcript quantification of the cDNA from the mother resulted in 41.5% of WT transcript (up to 63.1% if T2-T5 are considered incomplete WT transcripts), which was expected as the index and the mother are heterozygous for the variant.Sequencing allowed for the identification of the predicted alternative acceptor splice site at position c.1785 being used in 15.5% (transcript T6), and a transcript characterized by exon 14 skipping in 5.3% of the reads (transcript T7).Additionally, sequencing identified alternative donor or acceptor splice sites used for exons 15 and 16 (transcript T8).Finally, a minority of the reads (1.9%, transcript T9) were distinguished by exon 14 skipping and the inclusion of a pseudoexon from intron 14 (c.1875+661_1875+851).Similar results were found for the index patient.As discussed in Section 2.3.3, the minigene assay for the CACNA1F variant (NM_005183.4:c.2239+5C>G)revealed a complex collection set of transcripts, with the main variant-induced aberrant transcript being characterized by exon 16 skipping and an elongated exon 17 (Figure 2 and Table 6).Similarly, PCR amplification of exons 15-17 from blood cDNA of the index patient confirmed the main variant-induced transcript (40.8% of the reads) to be characterized by exon 16 skipping and the elongated exon 17 (corresponding to T1 in the minigene results).Additionally, a transcript was detected differing from the WT only by the elongation of exon 17 (T2, 35.7% of the reads), which was not identified in the minigene assay.Finally, a transcript was found with exon 16 skipping, intron 15 retention (c.2118+1_2129), and the elongated exon 17 (19.6%),corresponding to T10 in the minigene assay (Figure A6 and Table A6).Table 6.Transcript identification and quantification for the CANCA1F variant M_005183.4:c.2239+5C>G for the variant (MT) sequence.The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in the variant (MT) sequence, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.the WT only by the elongation of exon 17 (T2, 35.7% of the reads), which was not identified in the minigene assay.Finally, a transcript was found with exon 16 skipping, intron 15 retention (c.2118+1_2129), and the elongated exon 17 (19.6%),corresponding to T10 in the minigene assay (Figure A6 and Table A6).

Discussion
We applied Nanopore deep sequencing for the identification and quantification of aberrant splicing events in minigene and blood-derived cDNA assays due to a diverse set of candidate splicing variants in IRD genes.The method allowed for the characterization of complete transcripts and the identification of rare splicing events.Thirteen out of nineteen variants were found to lead to highly variable levels of aberrant splicing.

Discussion
We applied Nanopore deep sequencing for the identification and quantification of aberrant splicing events in minigene and blood-derived cDNA assays due to a diverse set of candidate splicing variants in IRD genes.The method allowed for the characterization of complete transcripts and the identification of rare splicing events.Thirteen out of nineteen variants were found to lead to highly variable levels of aberrant splicing.Five variants led to an alternative splice site used in the main transcript: ATF6:c.1096-15G>A,ATF6:c.1534-9A>G,PDE6C:c.864+1G>A,POC1B:c.677-2A>G, and RPGR:c.1415-9A>G.Three variants caused exon skipping: CHM:c.1413G>C,PROM1:c.2358C>T, and PROM1:c.2490-2A>G.Multiple aberrant splicing events, including exon skipping and alternative splice sites, were recognized for four variants: CACNA1F:c.2239+5C>G,IMPG2:c.3423-7_3423-4del,REEP6:c.517G>A, and KIF11:c.1875+2T>A.Finally, pseudoexon inclusion was found to be the main variantinduced event for a deep-intronic variant in POC1B (NM_172240.2:c.1033-327T>A).Conversely, the ABCA4:c.573C>T,ABCA4:c.5586T>A,FZD4:c.313A>G,TIMP3:c.205-3117T>C, and USH2A:c.652-22287T>Cvariants had no measurable effect on splicing in these assays.The assay outcomes were used to re-classify variants according to ACMG guidelines (functional evidence PS3/BS3 criterion).This led to a higher class for nine variants, a lower class for four variants, and an unchanged class for six variants (Table 7).
The splicing prediction algorithms included in this study proved to be reliable tools in predicting the nature of variant-induced aberrant splicing, as well as the magnitude of the effect for most variants.In particular, SpliceAI and Pangolin identified and scored accurately the effects of variants ATF6:c.1534-9A>G(average splice loss score of 78.5% and 80.6% WT transcript loss measured), CHM:c.1413G>C(average splice loss score of 79.5% and 85.1% WT transcript loss measured), PDE6C:c.864+1G>A(average splice loss score of 89.5% and 98.0% WT transcript loss measured), POC1B:c.677-2A>G(average splice loss score of 92.5% and 92.3% WT transcript loss measured), and PROM1:c.2358C>T(average splice loss score of 30.5% and 39.1% WT transcript loss measured).A notable exception is the predictions for CACNA1F:c.2239+5C>G,which suggested a loss of WT transcript in the range of 35%, and the use of a cryptic donor site (located at c.2200) in 16% of the transcripts.Transcript quantification resulted in a complete loss of WT transcript in the variant minigene and blood cDNA assays.Additionally, no transcript using the predicted cryptic donor site was identified.Similarly, predictions for ABCA4:c.573C>T and ABCA4:c.5586T>A,FZD4:c.313A>G,TIMP3:c.205-3117T>C, and USH2A:c.652-22287T>C were mild, and no aberrant splicing could be detected.While the minigene assay results did not support aberrant splicing for these variants, it may be possible that tissue-specific splicing might result in different quantities of alternatively spliced transcript in the relevant tissue.
An important caveat regarding this study is that minigene assays are simplified models of splicing, often limited to small portions of the gene that is being functionally tested.The same applies to blood-derived cDNA assays unless blood is the diseaserelevant tissue.Aberrant splicing events detected using these models may not reflect the actual physiological variant-induced effects in the relevant cell type(s) or tissues.It has been previously shown that splicing variants can have different effects and magnitude depending on the model used [47].While the exact nature of the effect on transcripts may not be extrapolated from these assays alone, the fact that aberrant splicing events are detected is a strong indication that the variant does affect splicing processes.For this reason, minigene assays remain particularly useful for the characterization of variantinduced aberrant splicing for inaccessible tissues, such as the retina.Combining these assays with Nanopore sequencing allows for unparalleled precision in the identification of transcripts and their relative abundance.Additionally, a limitation of the method is represented by its reliance on PCR amplification, which is prone to biases that could confound the results.Nevertheless, this method will help streamline and improve the analysis of novel candidate splicing variants, particularly for complex splicing patterns with multiple transcripts.Understanding the exact nature of aberrant splicing events could be crucial in the development of personalized therapies (e.g., antisense oligonucleotidebased therapies).
Table 7. Study outcome overview with adjusted ACMG classification.* ACMG recomputed on the Franklin platform by manually curating Functional Studies evidence based on this study results (Evidence categories PS3/BS3).Functional Studies evidence set on "Strong" for most assays. 1 Functional Studies evidence set on "Moderate" based on aberrant splicing evidence from assays. 2 Functional Studies evidence set on "Very strong" based on aberrant splicing evidence from assays. 3 ACMG classification is unchanged because the variant affects the protein function by altering the amino acid sequence.

Patient Cohort
Index patients were referred to us for genetic testing from large specialized medical centers in Switzerland.Blood samples were collected from index patients and available family members.Written informed consent was obtained from all patients and family members included in this study.This study was conducted in accordance with the 2013 Declaration of Helsinki.A subset of the patients included in this study have been included in previous studies from our group [39][40][41].

Genetic Testing
Genomic DNA (gDNA) was extracted from whole blood in duplicate with the automated Chemagic MSM I system, according to the manufacturer's specifications (PerkinElmer Chemagen Technologie GmbH, Baesweiler, Germany).Genetic testing strategies included WES, whole-gene sequencing by long-range PCR, or WGS.WES was performed as previously described [48].Whole-gene sequencing was performed as previously described [40].WGS was performed as previously described [41].
Rare variants that may affect splicing identified in patients affected by IRDs were selected for functional assays.

Minigene Assays
The effect of most variants in this study was functionally tested in a cellular system using minigene constructs.Most minigene constructs are based on the previously published pcDNA3.1 backbone (Invitrogen, Carlsbad, CA, USA), containing the genomic region encompassing exons 3 to 5 of the gene RHO with an artificial start codon introduced in RHO exon 3 [39,48,51,52].To introduce the genomic region of interest, exon 4 of RHO and part of the flanking introns were excised from the construct by digestion, using the restriction enzymes PflMI and EcoNI.The genomic regions of interest were amplified by PCR from patient's gDNA with Phusion High-Fidelity DNA Polymerase (New England Biolabs, Ipswich, MA, USA) for a total volume of 50 µL, containing 1× Phusion High-Fidelity Buffer, 0.5 µM of each primer, 0.2 mM dNTPs, 0.02 U/µL Phusion High-Fidelity DNA Polymerase, and 10 ng of gDNA.PCR reactions were performed on a Veriti thermal cycler (Applied Biosystems, Foster City, CA, USA) according to the following conditions: 98 • C for 30 s, 35 cycles of 98 • C for 10 s, 58-62 • C (depending on the primers) for 30 s, 72 • C for 5 min, and 72 • C for 10 min.PCR products were verified by electrophoresis on 1% agarose gels.
The minigene constructs for the variants in ATF6, FDZ4, and NRL are not based on the RHO backbone.Instead, exons 1 and 2 of the native gene were included.Specifically, for the ATF6 minigene constructs (one variant located in intron 8 and the other located in intron 12), exons 1 and 2 of ATF6 and the exon downstream of the variant (exons 9 and 13) were cloned into the pcDNA3.1 backbone (Invitrogen, Carlsbad, CA, USA) using the Takara In-Fusion HD cloning kit (Takara, Kusatsu, Japan).Similarly, the entire coding sequence and UTRs of the FZD4 gene and NRL (NM_006177.3) were inserted into the pcDNA3.1 backbone (Invitrogen, Carlsbad, CA, USA), using the Takara In-Fusion HD cloning kit (Takara Bio, Kusatsu, Japan).The genomic region of interest was amplified by PCR, as described in the previous paragraph.
Sanger and/or long-range PCR sequencing were performed to verify the genotype of the region of interest in selected clones, as previously described [40,48].
The plasmids were transfected into HEK293T cells by Xfect Transfection Reagent (Takara, Kusatsu, Japan), according to the manufacturer's instructions.Cells were harvested after 24 h and total RNA was isolated, and reverse transcribed into cDNA with the NucleoSpin RNA Plus (Macherey-Nagel, Düren, Germany) and SuperScript III First-Strand Synthesis SuperMix (Invitrogen, Waltham, MA, USA) kits, according to the manufacturer's instructions.
Primer sequences used for the amplification of the genomic regions of interest are listed in Supplementary Table S6.

Blood RNA Assays
Whole blood was collected in PAXgene Blood RNA Tubes (PreAnalytiX, Hombrechtikon, Switzerland) from index patients and family members, when available.The PAXgene Blood RNA Kit (PreAnalytiX, Hombrechtikon, Switzerland) was used to extract total RNA, as previously described [48].Total RNA was then reverse transcribed into cDNA with the Super-Script III First-Strand Synthesis SuperMix (Invitrogen, Waltham, MA, USA) with oligo(dT)20 primers, according to the manufacturer's instructions.

Nanopore Sequencing
In the case of RHO-backbone constructs, primers binding to RHO exons 3 and 5 were used to amplify the minigene-derived transcripts.Primers binding to the T7 promoter region and the BGH terminator were used to amplify ATF6, FZD4, and NRL minigenederived transcripts.These primers also contained adapter sequences for the Nanopore PCR Barcoding Kit SQK-PBK004 (TTTCTGTTGGTGCTGATATTGC-forward primer sequence, and ACTTGCCTGTCGCTCTATCTTC-reverse primer sequence; Oxford Nanopore Technologies, Oxford, UK).Primer sequences are available in Supplementary Table S7.
Similarly, primers containing the adapter sequences for the Nanopore PCR Barcoding Kit were designed to amplify the CACNA1F exons 15-17 and KIF11 exons 13-16 regions from whole blood cDNA.Primer sequences are available in Supplementary Table S7.
The transcripts of interest were first amplified by PCR in 50 µL volume, according to the Phusion High-Fidelity DNA Polymerase protocol (New England Biolabs, Ipswich, MA, USA), using the GC-Buffer and 100 ng of cDNA with the following conditions: 98 • C for 30 s, 35 cycles of 98 • C for 10 s, 63 • C with the RHO primers or 53 • C with T7/BGH primers for 30 s, 72 • C for 9 min, and 72 • C for 10 min.PCR products were verified by electrophoresis on 1% agarose gels.PCR reactions were purified with AMPure XP beads (Beckman Coulter Life Sciences, Indianapolis, IN, USA) with a 1:1.5 (PCR: beads) ratio and eluted in 50 µL of 1× Tris-EDTA (TE) buffer (Integrated DNA Technologies, Coralville, IA, USA), according to the manufacturer's instructions.Concentrations of purified PCRs were measured with the QuBit dsDNA High Sensitivity Assay Kit (Thermofisher Scientific, Waltham, MA, USA).These data were used to dilute each purified PCR in ddH 2 O to 10 ng/µL, for a final volume of 24 µL.
Subsequently, an indexing PCR was performed by adding 25 µL of Long Amp Taq 2X Master Mix (New England Biolabs, Ipswich, MA, USA) and 1 µL of barcoded universal primers with rapid attachment chemistry from the Nanopore PCR Barcoding Kit SQK-PBK004 (Oxford Nanopore Technologies, Oxford, UK), with the following conditions: 94 • C for 1 min, 30 cycles of 94 • C for 30 s, 62 • C for 30 s, 65 • C for 2 min, and 65 • C for 5 min.Indexing PCRs were purified using AMPure XP beads with a 1:1 (PCR: beads) ratio and eluted in 22 µL of Resuspension Buffer (Illumina, San Diego, CA, USA).Concentrations of indexing PCRs were quantified with the QuBit dsDNA High Sensitivity Assay Kit (Thermofisher Scientific, Waltham, MA, USA).The size distribution of PCR products was measured with a Bioanalyzer High-Sensitivity DNA kit on a Bioanalyzer 2100 instrument (Agilent Technologies, Santa Clara, CA, USA).Concentration and size distribution data were used to pool purified PCRs to a total of 50-90 fmol in a final volume of 10 µL.Finally, the rapid 1D sequencing adapters were attached by the addition of 1 µL of RAP to the PCRs pool, which was then incubated for 5 min at room temperature.

Splice Junctions Characterization and Usage Quantification
The JWR_checker.pyscript from NanoSplicer v1.0 [55] was used to detect splice junctions from the minimap2 alignment results for the minigene assays.The resulting output file was used to identify and quantify high-quality transcripts (reads) from the sequencing data.Briefly, only transcripts characterized by at least one high-quality splice junction (JAQ = 1) were kept.In the case of RHO-backbone constructs, only transcripts including both RHO exons were considered further.For each transcript identified, the number of reads representing them, and the mean junction quality (JAQ), were calculated.Only transcripts represented by at least 0.5% of the high-quality reads were kept.A construct-specific gff3 file was used to annotate known junctions and exons included in the transcripts and to calculate their length in base pairs.The in-house scripts used to transform the JWR_checker.pyoutputs are available on GitHub (https://github.com/jordimaggi/Minigene_transcripts_quantification_Nanopore; accessed on 21 July 2024).
When unknown splice sites were detected, the resulting transcript table was manually curated; the location of unknown acceptor and donor sites was verified on Alamut Visual for the existence of cryptic splice sites.If the predictions software on Alamut showed no scores at the splice site location identified during sequencing, the splice junction was assumed to be wrongly called and manually corrected to the most likely nearby splice junction.To visualize the identified transcripts, a gff3 file was created.

Table A1.
Transcript identification and quantification for the ABCA4 variant NM_000350.2:c.573C>T for reference (WT) and variant (MT) minigenes (construct RHO_minigene_ABCA4_int4-6).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Table A2.
Transcript identification and quantification for the ABCA4 variant NM_000350.2:c.5586T>A for reference (WT) and variant (MT) minigenes (construct RHO_minigene_ABCA4_int38-41).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the Table A2.Transcript identification and quantification for the ABCA4 variant NM_000350.2:c.5586T>A for reference (WT) and variant (MT) minigenes (construct RHO_minigene_ABCA4_int38-41).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Abbreviations: WT, wildtype (or reference); MT, mutant (or variant); ex, exon; bp, base pairs; Δ, delta.

Figure A2
. Functional characterization of the ABCA4 variant NM_000350.2:c.5586T>A using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A2
. Functional characterization of the ABCA4 variant NM_000350.2:c.5586T>A using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.
Table A3.Transcript identification and quantification for the ATF6 variant NM_007348.3:c.1096-15G>A for reference (WT) and variant (MT) minigenes (construct RHO_minigene_ATF6_int8-9). The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A4.Transcript identification and quantification for the ATF6 variant NM_007348.3:c.1096-15G>A for reference (WT) and variant (MT) minigenes (construct ATF6_minigene_ex1-2-9).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Transcript
Length  Table A4.Transcript identification and quantification for the ATF6 variant NM_007348.3:c.1096-15G>A for reference (WT) and variant (MT) minigenes (construct ATF6_minigene_ex1-2-9).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A5.Transcript identification and quantification for the ATF6 variant NM_007348.3:c.1534-9A>G for reference (WT) and variant (MT) minigenes (construct ATF6_minigene_ex1-2-13).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Transcript
Length  Table A5.Transcript identification and quantification for the ATF6 variant NM_007348.3:c.1534-9A>G for reference (WT) and variant (MT) minigenes (construct ATF6_minigene_ex1-2-13).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A7.Transcript identification and quantification for the CHM variant NM_000390.4:c.1413G>C for reference (WT) and variant (MT) minigenes (construct RHO_minigene_CHM_int9-11). The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A8.Transcript identification and quantification for the FZD4 variant NM_012193.4:c.313A>G for reference (WT) and variant (MT) minigenes (construct FZD4_minigene_ex1-2).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A8.Transcript identification and quantification for the FZD4 variant NM_012193.4:c.313A>G for reference (WT) and variant (MT) minigenes (construct FZD4_minigene_ex1-2).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A9.Transcript identification and quantification for the IMPG2 variant NM_016247.4:c.3423-7_3423-4del for reference (WT) and variant (MT) minigenes (construct RHO_minigene_IMPG2_int15-18).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A10.Transcript identification and quantification for the IMPG2 variant NM_016247.4:c.3423-7_3423-4del for reference (WT) and variant (MT) minigenes (construct RHO_minigene_IMPG2_int16-17).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Transcript
Length   17).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Transcript
Length WT (%) MT (%)  The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A12.Transcript identification and quantification for the PDE6C variant NM_006204.3:c.864+1G>A for reference (WT) and variant (MT) minigenes (construct RHO_minigene_PDE6C_int3-4).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The  Table A13.Transcript identification and quantification for the POC1B variant NM_172240.2:c.677-2A>G for reference (WT) and variant (MT) minigenes (construct RHO_minigene_POC1B_int6-7).
The table lists the transcripts identified, along with their characteristics, such as length, their relative Table A13.Transcript identification and quantification for the POC1B variant NM_172240.2:c.677-2A>G for reference (WT) and variant (MT) minigenes (construct RHO_minigene_POC1B_int6-7).
The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The  The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A17.Transcript identification and quantification for the REEP6 variant NM_001329556.3:c.517G>A for reference (WT) and variant (MT) minigenes (construct RHO_minigene_REEP6_int1-5).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The    Table A18.Transcript identification and quantification for the RPGR variant NM_001034853.1:c.1415-9A>G for reference (WT) and variant (MT) minigenes (construct RHO_minigene_RPGR_int10-13).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A18.Transcript identification and quantification for the RPGR variant NM_001034853.1:c.1415-9A>G for reference (WT) and variant (MT) minigenes (construct RHO_minigene_RPGR_int10-13).
The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.Table A19.Transcript identification and quantification for the TIMP3 variant NM_000362.4:c.205-3117T>C for reference (WT) and variant (MT) minigenes (construct RHO_minigene_TIMP3_int1-3).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Figure 1 .
Figure 1.Functional characterization of the KIF11 variant NM_004523.3:c.1875+2T>A using patientderived blood cDNA.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) sequences.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure 1 .
Figure 1.Functional characterization of the KIF11 variant NM_004523.3:c.1875+2T>A using patientderived blood cDNA.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) sequences.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure 2 .
Figure 2. Functional characterization of the CACNA1F variant NM_005183.4:c.2239+5C>G using patient-derived blood cDNA.The top panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for variant (MT) sequence.An overview of each transcript (name T#) identified and its relative abundance in MT can be seen underneath the coverage plots.

Figure 2 .
Figure 2. Functional characterization of the CACNA1F variant NM_005183.4:c.2239+5C>G using patient-derived blood cDNA.The top panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for variant (MT) sequence.An overview of each transcript (name T#) identified and its relative abundance in MT can be seen underneath the coverage plots.
B.-G., C.G.-K.and W.B.; data curation, J.M.; writing-original draft preparation, J.M.; writing-review and editing, J.M., S.F., J.G., K.M., R.B.-G., C.G.-K., S.K. and W.B.; visualization, J.M.; supervision, W.B.; project administration, J.M. and W.B.; funding acquisition, J.M. and W.B. All authors have read and agreed to the published version of the manuscript.Funding: This research was funded by Velux Stiftung, grant number 1371 (to W.B.).Institutional Review Board Statement: This study was conducted in accordance with Good Clinical Practices and the guidelines of the Declaration of Helsinki, and approval for genetic testing in human patients was awarded to the Institute of Medical Molecular Genetics by the Federal Office of Public Health (FOPH) in Switzerland.Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Figure A1 .
Figure A1.Functional characterization of the ABCA4 variant NM_000350.2:c.573C>T using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A1 .
Figure A1.Functional characterization of the ABCA4 variant NM_000350.2:c.573C>T using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A3 .
Figure A3.Functional characterization of the ATF6 variant NM_007348.3:c.1096-15G>A using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A3 .
Figure A3.Functional characterization of the ATF6 variant NM_007348.3:c.1096-15G>A using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A4 .
Figure A4.Functional characterization of the ATF6 variant NM_007348.3:c.1096-15G>A using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A4 .
Figure A4.Functional characterization of the ATF6 variant NM_007348.3:c.1096-15G>A using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A5 .
Figure A5.Functional characterization of the ATF6 variant NM_007348.3:c.1534-9A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.Table A6.Transcript identification and quantification for the CACNA1F variant M_005183.4:c.2239+5C>G for reference (WT) and variant (MT) minigenes (construct RHO_minigene_CACNA1F_int14-18).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.TranscriptLength WT (%) MT (%) Δ MT-WT (%)Counts WTCountsMT Effect on Transcript

Figure A5 .
Figure A5.Functional characterization of the ATF6 variant NM_007348.3:c.1534-9A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A6 .
Figure A6.Functional characterization of the CACNA1F variant NM_005183.4:c.2239+5C>G using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.TableA7.Transcript identification and quantification for the CHM variant NM_000390.4:c.1413G>C for reference (WT) and variant (MT) minigenes (construct RHO_minigene_CHM_int9-11). The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Figure A6 .
Figure A6.Functional characterization of the CACNA1F variant NM_005183.4:c.2239+5C>G using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A7 .
Figure A7.Functional characterization of the CHM variant NM_000390.4:c.1413G>C using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A8 .
Figure A8.Functional characterization of the FZD4 variant NM_012193.4:c.313A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A8 .
Figure A8.Functional characterization of the FZD4 variant NM_012193.4:c.313A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A9 .
Figure A9.Functional characterization of the IMPG2 variant NM_016247.4:c.3423-7_3423-4del using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A9 .
Figure A9.Functional characterization of the IMPG2 variant NM_016247.4:c.3423-7_3423-4del using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A10 .
Figure A10.Functional characterization of the IMPG2 variant NM_016247.4:c.3423-7_3423-4del using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.Table A11.Transcript identification and quantification for the OCA2 variant NM_000275.3:c.574-53C>G for reference (WT) and variant (MT) minigenes (construct RHO_minigene_OCA2_int5-7).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Figure A10 .
Figure A10.Functional characterization of the IMPG2 variant NM_016247.4:c.3423-7_3423-4del using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A11 .
Figure A11.Functional characterization of the OCA2 variant NM_000275.3:c.574-53C>G using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.TableA12.Transcript identification and quantification for the PDE6C variant NM_006204.3:c.864+1G>A for reference (WT) and variant (MT) minigenes (construct RHO_minigene_PDE6C_int3-4).The table lists the transcripts identified, along with their character-

Figure A12 .
Figure A12.Functional characterization of the PDE6C variant NM_006204.3:c.864+1G>A using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A13 .
Figure A13.Functional characterization of the POC1B variant NM_172240.2:c.677-2A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A13 .
Figure A13.Functional characterization of the POC1B variant NM_172240.2:c.677-2A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A14 .
Figure A14.Functional characterization of the POC1B variant NM_172240.2:c.1033-327T>A using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.TableA15.Transcript identification and quantification for the PROM1 variant NM_006017.3:c.2358C>T for reference (WT) and variant (MT) minigenes (construct RHO_minigene_PROM1_int20-23).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by Figure A14.Functional characterization of the POC1B variant NM_172240.2:c.1033-327T>A using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A15 .
Figure A15.Functional characterization of the PROM1 variant NM_006017.3:c.2358C>T using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.Table A16.Transcript identification and quantification for the PROM1 variant NM_006017.3:c.2490-2A>G for reference (WT) and variant (MT) minigenes (construct RHO_minigene_PROM1_int23-26).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Figure A15 .
Figure A15.Functional characterization of the PROM1 variant NM_006017.3:c.2358C>T using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A16 .
Figure A16.Functional characterization of the PROM1 variant NM_006017.3:c.2490-2A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A16 .
Figure A16.Functional characterization of the PROM1 variant NM_006017.3:c.2490-2A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A17 .
Figure A17.Functional characterization of the REEP6 variant NM_001329556.3:c.517G>A using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A17 .
Figure A17.Functional characterization of the REEP6 variant NM_001329556.3:c.517G>A using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A18 .
Figure A18.Functional characterization of the RPGR variant NM_001034853.1:c.1415-9A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of Figure A18.Functional characterization of the RPGR variant NM_001034853.1:c.1415-9A>G using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A19 .
Figure A19.Functional characterization of the TIMP3 variant NM_000362.4:c.205-3117T>C using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.TableA20.Transcript identification and quantification for the USH2A variant NM_206933.2:c.652-22287T>C for reference (WT) and variant (MT) minigenes (construct RHO_minigene_USH2A_int3).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.TranscriptLength WT (%) MT (%) Δ MT-WT (%) Counts WT Counts MT Effect on Transcript T1 RHO_ex3-RHO_ex5 119 bp 98.69 98.06 −0.63 56,927 87,450 WT T2 RHO_ex3-USH2A_pe4a-RHO_ex5 158 bp 0.95 1.41 0.46 546 1253 pe4a Abbreviations: WT, wildtype (or reference); MT, mutant (or variant); ex, exon; bp, base pairs; pe, pseudoexon; Δ, delta.

Figure A19 .
Figure A19.Functional characterization of the TIMP3 variant NM_000362.4:c.205-3117T>C using minigene assays.The panel shows an IGV screenshot highlighting the construct's characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Figure A20 .
Figure A20.Functional characterization of the USH2A variant NM_206933.2:c.652-22287T>C using minigene assays.The panel shows an IGV screenshot highlighting the construct s characteristics, followed by the coverage plots for the reference (WT) and variant (MT) minigenes.An overview of each transcript (name T#) identified and its relative abundance in WT and MT can be seen underneath the coverage plots.The green transcript represents the expected reference (WT) transcript.

Table 1 .
Nineteen candidate splicing variants in 15 IRD-associated genes.Classification according to American College of Medical Genetics and Genomics (ACMG) guidelines from Varsome and Franklin.

Table 4 .
Minigene assay results summary.The aberrant splicing events column reports the main aberrant splicing events induced or favored by the variant.The last column lists the difference in relative abundance of the expected reference (WT) transcript between variant and reference minigenes.

Table 6 .
Transcript identification and quantification for the CANCA1F variant M_005183.4:c.2239+5C>G for the variant (MT) sequence.The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in the variant (MT) sequence, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.MT, mutant (or variant); ex, exon; int, intron; alt, alternative; AS, acceptor splice site; DS, donor splice site; bp, base pairs; Δ, delta.
Abbreviations: ACMG, American College of Medical Genetics and Genomics guidelines; VUS, variant of unknown significance; AS, acceptor splice site; DS, donor splice site; and PE, pseudoexon.

Table A3 .
Transcript identification and quantification for the ATF6 variant NM_007348.3:c.1096-15G>A for reference (WT) and variant (MT) minigenes (construct RHO_minigene_ATF6_int8-9). The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Table A6 .
Transcript identification and quantification for the CACNA1F variant M_005183.4:c.2239+5C>G for reference (WT) and variant (MT) minigenes (construct RHO_minigene_CACNA1F_int14-18).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.
table is sorted by relative abundance.
table is sorted by relative abundance.

Table A14 .
Transcript identification and quantification for the POC1B variant NM_172240.2:c.1033-327T>A for reference (WT) and variant (MT) minigenes (construct RHO_minigene_POC1B_int9-10).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Table A15 .
Transcript identification and quantification for the PROM1 variant NM_006017.3:c.2358C>T for reference (WT) and variant (MT) minigenes (construct RHO_minigene_PROM1_int20-23).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Table A16 .
Transcript identification and quantification for the PROM1 variant NM_006017.3:c.2490-2A>G for reference (WT) and variant (MT) minigenes (construct RHO_minigene_PROM1_int23-26).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.

Table A17 .
Transcript identification and quantification for the REEP6 variant NM_001329556.3:c.517G>A for reference (WT) and variant (MT) minigenes (construct RHO_minigene_REEP6_int1-5).The table lists the transcripts identified, along with their characteristics, such as length, their relative abundance in reference (WT) and variant (MT) minigenes, the difference (delta) in relative abundance between MT and WT sequencing results, the absolute number of reads representing each transcript, and the effect on the transcript.The table is sorted by relative abundance.