Structural Variants Create New Topological-Associated Domains and Ectopic Retinal Enhancer-Gene Contact in Dominant Retinitis Pigmentosa

Abstract

The cause of autosomal-dominant retinitis pigmentosa (adRP), which leads to loss of vision and blindness, was investigated in families lacking a molecular diagnosis. A refined locus for adRP on Chr17q22 (RP17) was delineated through genotyping and genome sequencing, leading to the identification of structural variants (SVs) that segregate with disease. Eight different complex SVs were characterized in 22 adRP-affected families with >300 affected individuals. All RP17 SVs had breakpoints within a genomic region spanning YPEL2 to LINC01476. To investigate the mechanism of disease, we reprogrammed fibroblasts from affected individuals and controls into induced pluripotent stem cells (iPSCs) and differentiated them into photoreceptor precursor cells (PPCs) or retinal organoids (ROs). Hi-C was performed on ROs, and differential expression of regional genes and a retinal enhancer RNA at this locus was assessed by qPCR. The epigenetic landscape of the region, and Hi-C RO data, showed that YPEL2 sits within its own topologically associating domain (TAD), rich in enhancers with binding sites for retinal transcription factors. The Hi-C map of RP17 ROs revealed creation of a neo-TAD with ectopic contacts between GDPD1 and retinal enhancers, and modeling of all RP17 SVs was consistent with neo-TADs leading to ectopic retinal-specific enhancer-GDPD1 accessibility. qPCR confirmed increased expression of GDPD1 and increased expression of the retinal enhancer that enters the neo-TAD. Altered TAD structure resulting in increased retinal expression of GDPD1 is the likely convergent mechanism of disease, consistent with a dominant gain of function. Our study highlights the importance of SVs as a genomic mechanism in unsolved Mendelian diseases.

Keywords: GDPD; Hi-C; RP17; dominant retinitis pigmentosa; ectopic expression; photoreceptor precursors cells; retinal organoids; stem cells; structural variants; topologically associated domains; whole-genome sequencing.

Figure 1

Mapping of the RP17 Locus in Two Unrelated Families (A) Pedigree of Dutch NL1 family. (B) Pedigree of UK1 family. (C) Pedigrees of additional UK families with the founder haplotype on Chr17q. WGS or WES was performed in individuals highlighted in blue or red, respectively. (D) SNP haplotyping results for NL1. The refined RP17 locus (rs8078110–rs9910672) is shared by all affected individuals (n = 35) and not present in unaffected individuals (n = 28, only individuals with recombination close to or refining the critical region are depicted) with a maximum LOD score of 15.0. The horizontal numbers represent the number of individuals with this haplotype. (E) UK founder haplotype refining the RP17 locus in UK families. Representative haplotypes from several unrelated families are shown with affected (aff) individuals compared to an unaffected (unaff) individual. Black lines and arrows indicate recombination events. Shared haplotype in individuals is shaded red. (F) Overlap of refined RP17 loci in UK, NL, and previously described SA families.

Figure 2

Overview of Structural Variants within the RP17 Locus in adRP-Affected Families Breakpoints are indicated with dashed lines. Blue segments represent duplicated or triplicated regions, whereas inversions are highlighted in purple. (A) Wild-type (WT) chromosomal organization. (B) Structural variants identified in NL1 (NL-SV1) and UK founder haplotype families (UK-SV2). (C) Structural variants identified in adRP-affected families that were previously linked to the RP17 locus; SA-SV3 and CA-SV4 (unpublished data). (D) Structural variants found in a cohort of unsolved adRP-affected families; NL-SV5, UK-SV6, UK-SV7, and UK-SV8. Letters A–AI depict the genomic intervals for each SV used to analyze and annotate SV breakpoints. (E) Overview of all SV breakpoints identified in the RP17 locus. An overlapping genomic region that is duplicated or triplicated in all SVs was identified (chr17: 57,499,214–57,510,765) and is highlighted by a light blue vertical bar. The size of DHX40 is reduced, and CLTC is partially shown for the purpose of this figure.

Figure 3

YPEL2 is Located within a Structured Active Compartment that Contains Retinal-Specific Enhancers (A) The TAD landscape of the genomic region disrupted by the RP17 SVs. Hi-C map of control retinal organoids revealed a structured domain containing YPEL2. (B) YPEL2 TAD boundaries correspond with CTCF sites identified in human retina. Analysis of RNA-seq and assay for transposase-accessible chromatin using seqencing (ATAC-seq) data across the YPEL2 region shows YPEL2 retinal expression and an accessible chromatin configuration. Analysis of H3K27Ac ChIP-seq data in the same region revealed several active enhancers located within the YPEL2 TAD, which are enriched for retinal transcription factor binding sites, including NRL, CRX, and OTX2. These enhancers were located 5′ of the CTCF boundary site within LINC01476. (C) Schematic representation of the YPEL2 TAD structure.

Figure 4

RP17 SVs Create Novel Domains (neo-TADs) and Hyper-activation of Retinal Enhancers (A) Schematic modeling of the genome architecture spanning the RP17 region using Hi-C maps. The wild-type Hi-C map derived from neuronal tissue shows a TAD with CTCF boundaries containing YPEL2 and retinal enhancers, flanked by unstructured domains. TAD models of NL-SV1 and UK-SV2 (dotted vertical lines represent SV breakpoints) predict the formation of neo-TADs and ectopic interactions of the retinal enhancer with GDPD1. (B) Hi-C performed on retinal organoids (ROs) derived from control (top) and RP17 UK-SV2 individuals (bottom) (10 kb resolution; raw count map). The chromatin organization in control ROs shows the YPEL2 TAD (indicated by dashed lines). Two novel domains (neo-TAD 1 and 2) are visible in the UK-SV2 ROs, and neo-TAD 2 allows ectopic retinal enhancer contacts to GDPD1 and SMG8. The dashed circle indicates the strong chromatin contact between retinal enhancers and the GDPD1 promoter. (C) qPCR revealed significantly upregulated retinal enhancer RNA expression in UK-SV2 ROs compared to controls (n = 3, mean ± standard error of the mean, **p ≤ 0.01).

Figure 5

Convergent Mechanism of Ectopic Retinal Enhancer-GDPD1 Interaction Caused by RP17 SVs (A) In wild-type genomic context, YPEL2 expression in retina is driven by retinal enhancers in a TAD with CTCF boundaries. Neighboring genes are insulated from retinal enhancer activation. (B) The NL-SV1 duplication creates a neo-TAD with a full-length copy of YPEL2, GDPD1, and the retinal enhancers. This enables retinal-specific enhancers to ectopically interact with GDPD1, which drives its misexpression. (C) qPCR analysis of photoreceptor precursor cells (PPCs) revealed a significant upregulation of GDPD1 in NL-SV1 PPCs compared to controls. (D) The UK-SV2 duplication and inversion creates a neo-TAD with a full-length copy of GDPD1 and SMG8 and the retinal enhancers bounded by CTFC sites. (E) qPCR analysis ROs revealed a significant upregulation of GDPD1 expression in UK-SV2 ROs compared to controls (n = 3 independent ROs, mean ± standard error of the mean, **p ≤ 0.01, ****p ≤ 0.0001).