Identification of long non-coding RNAs co-expressed with Inherited Retinal Disease Genes

Abstract

Purpose : Long non-coding RNAs (lncRNAs) are a heterogeneous class of non-coding transcripts whose role in biological and pathological processes is still poorly defined. We aimed at identifying lncRNAs that can potentially contribute to the pathogenesis of Inherited Retinal Diseases (IRD) by meta-analysis of human retinal transcriptome data.

Methods : A total of 161 bulk RNA-Seq human retina datasets from non-visually impaired post-mortem donors were retrieved from publicly available expression databases. We re-analyzed these data using an ad-hoc designed pipeline and constructed a weighted network by calculating the Pearson correlation score to identify lncRNAs co-expressed with IRD genes. We selected the top 1% positively correlated genes for each of the analyzed IRD genes. Finally, we retained the lncRNAs co-expressed with at least ten IRD genes and displaying median expression levels greater than 100 Transcript Per Million (TPM).

We carried out an initial in silico characterization of the IRD co-expressed lncRNAs, including a) Gene Ontology (GO) and KEGG analysis of their gene co-expression networks to infer their possible function and b) evaluation of publicly available single-cell data (Spectacle, v2.0.0) to assess their putative cellular distribution within the retina. We then proceeded with experimental determination of their expression profiles by quantitative (q)RT-PCR on cDNA samples from both ocular and non-ocular tissues and by single-molecule RNA in situ hybridization assays (RNAscope or Basescope).

Results : By gene co-expression analysis, we identified 11 already annotated, but poorly characterized, lncRNAs significantly co-expressed with IRD genes. Bioinformatic analysis highlighted that the gene co-expression networks of these lncRNAs are significantly enriched for GO terms highly related to photoreceptor functions and ciliary activities.
Moreover, we found that more than 60% of them are expressed predominantly or exclusively in the retina, as assessed by qRT-PCR assays, whereas RNA in situ hybridization in adult human retina revealed, in most cases, their restricted localization to the photoreceptor cell layer, with instances of either rod- or cone-specific expression.

Conclusions : This study indicates that co-expression analysis applied to RNA-seq datasets represents a valuable tool to highlight lncRNAs with a potential role in retinal function and possibly contributing to IRD pathogenesis.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.