A combined RNA-seq and whole genome sequencing approach for identification of non-coding pathogenic variants in single families

Abstract

Inherited retinal degenerations (IRDs) are at the focus of current genetic therapeutic advancements. For a genetic treatment such as gene therapy to be successful, an accurate genetic diagnostic is required. Genetic diagnostics relies on the assessment of the probability that a given DNA variant is pathogenic. Non-coding variants present a unique challenge for such assessments as compared to coding variants. For one, non-coding variants are present at much higher number in the genome than coding variants. In addition, our understanding of the rules that govern the non-coding regions of the genome is less complete than our understanding of the coding regions. Methods that allow for both the identification of candidate non-coding pathogenic variants and their functional validation may help overcome these caveats allowing for a greater number of patients to benefit from advancements in genetic therapeutics. We present here an unbiased approach combining whole genome sequencing (WGS) with patient-induced pluripotent stem cell (iPSC)-derived retinal organoids (ROs) transcriptome analysis. With this approach, we identified and functionally validated a novel pathogenic non-coding variant in a small family with a previously unresolved genetic diagnosis.

Figure 1

Family OGI-081 variant segregation scheme and retinal phenotypes. (A) The OGI-081 pedigree with variant segregation scheme. (B) Fundus (upper panel) and OCT image (lower panel) for the OGI-081-197 at age 8; area of retinal degeneration is indicated by the red bar.

Figure 2

RO differentiation. (A) Schema of the differentiation process and a light microscopy image of a typical RO. Arrow head indicating photoreceptors, scale bar = 100 microns. (B–G) Immunocytochemistry on cryosections of ROs. (B and E) NR2E3 staining of rod nuclei (green). (C and F) Mature cones show staining of cone opsins in the cone photoreceptor outer segments (red). (C) S opsin. (F) ML opsin. (D and G) Overlay of rod and cone staining. All cones are stained with ARR3 (purple). Scale bars = 20 microns.

Figure 3

Comparison of IRD gene expression and splice junctions. Human normal retina (HNR; N = 3, gray), RO from the unaffected sibling (N = 5, blue), skin-sun exposed (SSE; N = 473, green) and whole blood (WB; N = 407, red). (A and B) Average TPM values of IRD genes. (A) IRD genes are sorted by their expression in HNR overlaid with RO, SSE or WB. (B) Violin plot. (C and D) Number of splice junctions detected by MAJIQ. (C) All annotated genes. (D) IRD genes.

Figure 4

Alternative splicing in the NCALD and CNGB3 genes. (A) Venn diagram of genes found to have alternative splicing events in OGI-081 comparison of affected vs. unaffected siblings and genes found to have segregating allelic pairs (green). Alternative splicing analysis was conducted by MAJIQ (blue) and CASH (red). (B) Collapsed diagram of exons (black boxes) from all isoforms of the NCALD and CNGB3 genes. DNA variants (red); MAJIQ (blue) and CASH (yellow) alternative splicing events (E). Events detected by MAJIQ are depicted as split reads arches. The event range detected by CASH is depicted by the left (L) and right (R) borders. Genomic locations of variants, junctions and event borders are given in Supplementary Material, Table S8.

Figure 5

Aberrant splicing of CNGB3 in the affected vs. unaffected siblings. (A) Sashimi plot presenting RNA-seq results showing a cryptic exon spliced into the isoform as a result of the intronic variant chr8:g.87618576G>A. The cryptic exon is only present in the affected sibling (lower panel, red) and not in the unaffected sibling (upper panel, blue). The splice junction between exon 14b and exon 15 is not represented by split reads in the Sashimi plot due to an alignment error (Supplementary Material, Fig. S1). (B) RT-PCR using primers from the canonical exon14 and exon 15. All three siblings express the normal size isoform lacking exon 14b (lower band). A larger abnormal band containing exon14b (upper band) is present in the two affected siblings Af1 and Af2 but not in the unaffected sibling (Un). Negative controls lacking RNA template in the RT reaction (NC1) and NC1 used as template for PCR amplification (NC2). Sanger sequencing of the larger band confirming the inclusion of exon 14b. (C) Schematic representation of the protein domains in W.T CNGB3 and the two mutant alleles found in the affected siblings of OGI-081.

Figure 6

Mislocalization of the CNGB3 truncated proteins. Immunocytochemical analysis of day 262 ROs from the heterozygous parent OGI-081-200 (A–C) and an affected sibling OGI-081-197 (D–F). In the heterozygote, both ML opsins (red) and CNGB3 (green) are localized to the photoreceptor outer segments whereas in the affected sibling, CNGB3 localizes to the photoreceptor inner segments. An exemplary photoreceptor outer segment is indicated by the white brackets. Nuclei are counterstained with DAPI (blue). Scale bars = 20 micron.