Systematic review of differential methylation in rare ophthalmic diseases

Rare ophthalmic diseases have a devastating impact on a patient’s vision and consequently negatively affect their independence, ability to work and overall quality of life. Methylation is an important emerging biomarker of disease and may improve understanding of rare ophthalmic disorders. This systematic review sought to identify and evaluate literature on methylation and rare ophthalmic disease. MEDLINE, EMBASE, PubMed, Cochrane Database of Systematic Reviews and grey literature resources were searched for publications prior to 20 August 2019. Articles written in English which featured key terms such as ‘methylation’ and rare ophthalmic diseases were included. Titles, abstracts, keywords and full texts of publications were screened, as well as reference lists for reverse citations and Web of Science ‘cited reference search’ for forward citation searching. Study characteristics were extracted, and methodological rigour appraised using a standardised template. Fourteen articles were selected for full inclusion. Rare ophthalmic conditions include congenital fibrosis of extraocular muscles, retinitis pigmentosa, Fuchs endothelial corneal dystrophy, granular corneal dystrophy, choroideraemia, brittle cornea syndrome, retinopathy of prematurity, keratoconus and congenital cataracts. Outcomes include identification of methylation as contributor to disease and identification of potential novel therapeutic targets. The studies included were heterogeneous with no scope for meta-analysis following review; a narrative synthesis was undertaken. Differential methylation has been identified in a small number of rare ophthalmic diseases and few studies have been performed to date. Further multiomic research will improve understanding of rare eye diseases and hopefully lead to improved provision of diagnostic/prognostic biomarkers, and help identify novel therapeutic targets.

To investigate the role of CpG methylation in preterm infants risk of retinopathy of prematurity.
Population: Pre-term newborn babys with prethreshold retinopathy of prematurity at 25.1 gestational weeks compared to newborn babies who did not at 26.1 gestational weeks.
Methylation measurement: Illumina Infinium MethylationEPIC array used to analyse bisulphite converted DNA samples.

Methodological rigour: weak
Increased methylation levels at eleven probes was associated with a decreased risk of retinopathy of prematurity (probes: (cg10266336, cg15484375, cg12907644, cg06979684, cg00473501, cg01565803, cg07694252, cg06740893, cg11151395, cg14619064, and cg22331200), while increased methylation in five different probes was associated with an increased risk of retinopathy of prematurity (cg14970975, cg05199346, cg07159484, cg15483907, and cg00989220 Methylation measurement: Methylation array data collected in a previous study, [44] was subjected to subanalysis focusing on 2,227 miRNA probes of 463 miRNA genes. MethyLight analysis was also performed on an additional cohort of FECD cases and controls. Methodological rigour: weak controls, predominantly DNA hypermethylation (71%, 154 probes, of FECD samples were hypermethylated compared to just 29%, or 62 hypomethylated probes). Differential methylation was predominantly seen within promoter regions (96%) and no significant methylation changes were seen with reference to age or sex.
Of the 154 hypermethylated probes, 74% were within intragenic sequences. Similarly, 71% of the 62 hypomethylated probes occurred in intragenic sequences, 44% of these being within intronic regions. MethyLight validation confirmed the differential methylation between cases and controls from the HumanMethylation450 array, with higher mean DNA methylation in FECD patients, as well as specific methylation of the miRNAs: miR-199A1 and miR-23B in FECD cases compared to controls.
An inverse relationship between hypermethylation and gene expression levels was see in 18 miRNAs, in particular the almost complete silencing of miR-199b-5p expression (which contained the highest level of promoter hypermethylation), suggesting a role in FECD pathogenesis.
A dual-luciferase reporter assay found that miR-199b-5p directly regulated Snail and ZEB1 mRNA transcripts, reducing expression levels significantly in human corneal endothelial cells. Farinelli, P. 2014 Reference number: 53 DNA methylation and differential gene regulation in To elucidate patterns of differential DNA methylation In vivo interventional study using murine models. Population: In all RP mice models, there was a general increase of methylation in the outer nuclear layer compared to any retinal layer of healthy Mice and rat models of RP, known as rd1 strains, used irrespective of gender and corresponding wild type strains. Ages were used corresponding to stages of retinal degeneration.
Methodological rigour: weak wild types, but not in the inner retinal layers of RP models. 5methylcystosine antibody co-labelled with dying cells in terminal deoxynucleotidyl transferase dUTP nick end labelling assay, indicating a link between DNA methylation and photoreceptor degeneration, particularly evident at the late stages of photoreceptor cell death.
DNA methylation measurement at whole tissue level found little difference between rd1 and wt, indicating that in contrast to cellular approach, whole tissue measurement of methylation may be insufficiently sensitive to detect a small number of cells showing strong DNA methylation.
Microarray analysis showed 1284 hypermethylation genes in rd1 compared to wt in intragenic regions.
Hypermethylation of rd1 mice appeared to be greater in binding motifs of transcription factors YY1, E2F3 and NRL, all of which have possible roles in RP, and their target genes compared to wt.
DNMT i inhibition using decitabine increased survival of photoreceptors in short term cultures (4 days) from rd1 animal retinae compared to control (not treated) but not in long term cultures (11 days Methylation measurement: Conducted using the digoxigenin-labeled M27 beta probe to examine the XLRP2 gene methylation.

Methodological rigour: weak
The opposite methylation pattern was observed in a phenotypically normal carrier.
There is a potential association between extreme skewed X chromosome inactivation and clinical severity.
No correlation between differential methylation status of X chromosomes and clinical severity. Observational design: case-control.
Population: Three RP and three normal control participants, as well as a fourth additional RP participant with an asymmetric allele ratio of 4:1 including an extra copy of the wild-type allele. Methylation measurement: Bisulphite conversion of DNA was processed using a methylation specific PCR assay, followed by cloning of PCR products in a vector (pCR4-TOPO) which Bisulphite converted DNA from 20 clones showed no significantly different methylation pattern between healthy controls and RP participants, as well as no significant difference in methylation levels between alleles containing the FSCN2 c.72delG mutation and alleles without, indicating no preferential monoallelic imprinting. Methodological rigour: weak Of the probes for COL8A1, TCF4, SLC4A11, and AGBL genes, previously identified to be associated with late onset FECD, the vast majority did not yield a statistically significant difference in methylation with the exception of SLC4A11 (14/16 probes). However, 10,961 other probes showed a significant difference in methylation with a false discovery rate of 0.01. Of these, 59% displayed hypermethylation and 41% displayed hypomethylation. No probes were significantly differentially methylated when comparing age or sex.
Gene ontology analysis showed gene body hypomethylation occurred disproportionately in Methodological rigour: weak Levels of H3K4me3 were high in wild type cells compared to GCD2 cells in TGFβ1 promoter region, whilst H3K27me3 p levels were low in both cells. DNA methylation levels were not significantly different between the two cell types.
TGFβ1 increased active H3K4me1 q and H3K4me3 levels in TGFβ1 and other extra cellular matrix associated gene promoters in wt fibroblasts but not in GCD2-homozygous cells, correlated with increased expression of TGFβ1 induced genes. The repressive H3K27me3 decreased in wildtype fibroblasts following TGFβ1 treatment, but again no effect was found in GCD2-homozygous cells.

Increases in H3K4me1/3 and decreases in
H3K27me3 at TGFBIp and extra cellular matrix gene promoters plays a role in GCD2 pathogenesis. In vivo interventional study using murine models.
Population: Murine models with germ-line Nrl and Pde6b knock outs.
Methylation measurement: Bisulphite converted DNA from murine retinal tissue was amplified via PCR and analysed using the bisulphite sequencing DNA methylation software to retrieve CpG methylation quantification. Methodological rigour: weak Germline deletion of Nrl on Pde6b retinal degeneration mice, as well as condition in activation of the Nrl gene in adult mice, supresses the retinal degeneration phenotype through rod gene degeneration and gain of cones. However, only partial re-programming was seen at P44 in Nrl knock out mice compared to embryonic Nrl knock out as this was insufficient to reprogramme DNA methylation patterns (unchanged at the Rho and Opn1sw loci), suggesting DNA methylation of these loci may play a role in RP development. Observational design: case control.

Population:
Samples from five patients with type 2 brittle cornea syndrome (n=4 skin and eye fibroblasts, n=1 expression construct), compared to six samples from controls eye pathology (n=4 skin fibroblasts, n=1 postmortum eye sample and n=1 postenucleation from corneal trauma sample).
Methylation measurement: ChIP microarray and mass spectrometry was used to measure PRDM5 and its interaction with histone methyltransferases.
Methodological rigour: weak PRDM5 proteins were found to interact with HP1BP3, which binds to H3K9 methylated genomic regions. This interaction was found to be lost in mutated PRDM5 proteins. H3K9me2 was enriched in unaffected controls at the genes COL13A1, NTN1 and COL15A1, and decreased in patients with type 2 brittle cornea syndrome compared to unaffected controls. Therefore, H3K9me2 mediated repression is thought to have a role in normal vascular development and maintenance through PRDM5-target genes.