A genome-wide scan of non-coding RNAs and enhancers for refractive error and myopia

Abstract

Refractive error (RE) and myopia are complex polygenic conditions with the majority of genome-wide associated genetic variants in non-exonic regions. Given this, and the onset during childhood, gene-regulation is expected to play an important role in its pathogenesis. This prompted us to explore beyond traditional gene finding approaches. We performed a genetic association study between variants in non-coding RNAs and enhancers, and RE and myopia. We obtained single-nucleotide polymorphisms (SNPs) in microRNA (miRNA) genes, miRNA-binding sites, long non-coding RNAs genes (lncRNAs) and enhancers from publicly available databases: miRNASNPv2, PolymiRTS, VISTA Enhancer Browser, FANTOM5 and lncRNASNP2. We investigated whether SNPs overlapping these elements were associated with RE and myopia leveraged from a large GWAS meta-analysis (N = 160,420). With genetic risk scores (GRSs) per element, we investigated the joint effect of associated variants on RE, axial length (AL)/corneal radius (CR), and AL progression in an independent child cohort, the Generation R Study (N = 3638 children). We constructed a score for biological plausibility per SNP in highly confident miRNA-binding sites and enhancers in chromatin accessible regions. We found that SNPs in two miRNA genes, 14 enhancers and 81 lncRNA genes in chromatin accessible regions and 54 highly confident miRNA-binding sites, were in RE and myopia-associated loci. GRSs from SNPs in enhancers were significantly associated with RE, AL/CR and AL progression. GRSs from lncRNAs were significantly associated with all AL/CR and AL progression. GRSs from miRNAs were not associated with any ocular biometric measurement. GRSs from miRNA-binding sites showed suggestive but inconsistent significance. We prioritized candidate miRNA binding sites and candidate enhancers for future functional validation. Pathways of target and host genes of highly ranked variants included eye development (BMP4, MPPED2), neurogenesis (DDIT4, NTM), extracellular matrix (ANTXR2, BMP3), photoreceptor metabolism (DNAJB12), photoreceptor morphogenesis (CHDR1), neural signaling (VIPR2) and TGF-beta signaling (ANAPC16). This is the first large-scale study of non-coding RNAs and enhancers for RE and myopia. Enhancers and lncRNAs could be of large importance as they are associated with childhood myopia. We provide a confident blueprint for future functional validation by prioritizing candidate miRNA binding sites and candidate enhancers.

Fig. 1

Schematic workflow of analyses to identify candidate genetic variants in non-coding RNAs and enhancers per database. SNPs single nucleotide polymorphisms, 3' UTR 3' untranslated region, GRS genetic risk score

Fig. 2

Candidate gene expression in peripheral retinal cell types. A Normalized expression of candidate miRNA binding regions of host gene (rows) within retinal cell types (columns). Color heat map on the right of the graph depicts the level of min–max normalized expression per gene across each retinal cell type. The asterisks depict candidate host genes showing cell-type specific expression. Please note that ALKAL2 is an alias for FAM150B. B and C Normalized expression of candidate genes flanking enhancer regions (rows; B VISTA database, C FANTOM5 database) within retinal cell types (columns). The colors depict the level of min–max normalized expression per gene across each retinal cell type, similarly to the color heat map shown in A. The asterisks depict candidate flanking genes showing cell-type specific expression

Fig. 3

Venn diagram and corresponding table depicting the overlap between candidate non-coding regulatory elements. A rs6480615 and rs2394861 lie in a lncRNA region, are also in an enhancer region of DDIT4, and are also in the vicinity of the miRNA binding site in DDIT4. B overlap between enhancer regions and miRNA binding sites. Notably, rs1053639 (marked bold, DDIT4) is in high LD with enhancers (rs11000235, rs10762503). C Enhancer regions coinciding in lncRNA regions (the same SNPs are found in both regulatory databases). D MiRNA binding site regions coinciding in LncRNA regions. Note that the hs-enhancers are from VISTA Enhancer Browser and the chr:pos enhancer regions are from FANTOM5. Grey SNPs are not in chromatin accessible regions

Fig. 4

Most notable candidates and their potential corresponding ocular tissue of action. This figure depicts an overview of candidates with the highest ranking (miRNA binding sites and enhancers), notable biological pathways, and cell-type specific expression in human ocular tissue. The location of potential activity of these candidates in the eye is depicted on the right. Of some of the high ranking candidates, the location of potential activity it is still unknown. Note that the hs-enhancers are from VISTA Enhancer Browser and the chr:pos enhancer regions are from FANTOM5. n/a not applicable (no cell-type specific expression found)