Meta-analysis of retinal transcriptome profiling studies in animal models of myopia

Abstract

Objective: Myopia prevalence is increasing at alarming rates, yet the underlying mechanistic causes are not understood. Several studies have employed experimental animal models of myopia and transcriptome profiling to identify genes and pathways contributing to myopia. In this study, we determined the retinal transcriptome changes in response to form deprivation in mouse retinas. We then conducted a transcriptome meta-analysis incorporating all publicly available datasets and analyzed how the results related to the genes associated with refractive errors in human genome-wide association studies (GWAS).

Methods: Form deprivation was induced in three male C57BL6/J mice from postnatal day 28 (P28) to P42. Retinal gene expression was analyzed with RNA sequencing, followed by differential gene expression analysis with DESeq2 and identification of associated pathways with the Kyoto Encyclopedia of Genes and Genomes (KEGG). A systematic search identified four similar retinal transcriptomics datasets in response to experimental myopia using chicks or mice. The five studies underwent transcriptome meta-analyses to determine retinal gene expression changes and associated pathways. The results were compared with genes associated with human myopia.

Results: Differential gene expression analysis of form-deprived mouse retinas revealed 235 significantly altered transcripts, implicating the BMP2 signaling pathway and circadian rhythms, among others. Transcriptome-wide meta-analyses of experimental myopia datasets found 427 differentially expressed genes in the mouse model and 1,110 in the chick model, with limited gene overlap between species. Pathway analysis of these two gene sets implicated TGF-beta signaling and circadian rhythm pathways in both mouse and chick retinas. Some pathways associated only with mouse retinal changes included dopamine signaling and HIF-1 signaling pathway, whereas glucagon signaling was only associated with gene changes in chick retinas. The follistatin gene changed in both mouse and chick retinas and has also been implicated in human myopia. TGF-beta signaling pathway and circadian entrainment processes were associated with myopia in mice, chicks, and humans.

Conclusion: This study highlights the power of combining datasets to enhance statistical power and identify robust gene expression changes across different experimental animal models and conditions. The data supports other experimental evidence that TGF-beta signaling pathway and circadian rhythms are involved in myopic eye growth.

Keywords: RNA-Seq; experimental models; meta-analysis; myopia; transcriptomics.

Figure 1

Ocular changes in response to form deprivation in mice used in the RNA sequencing experiment. (A) The myopic shift, expressed as the interocular difference in refractive error, showed relative myopia in the form-deprived eye of wild-type C57BL/6J mice after two weeks of form deprivation. Axial length (B) and corneal curvature (C) dynamics in response to form deprivation revealed no interaction effect between age and form deprivation. Data are mean ± SEM, n = 3. In (A), statistical analysis was performed using one-way ANOVA with Šidák's correction for multiple comparisons. In (B, C), two-way repeated measures ANOVA was used. ** p < 0.01. FDM, form-deprivation myopia.

Figure 2

Differentially expressed genes and enriched pathways in FDM mouse retinas. (A) A volcano plot of genes differentially regulated in FDM retinas from wild-type mice, illustrating the log₂ effect size and unadjusted log-transformed p values. The gray dashed horizontal line indicates an unadjusted p value of 0.05. Genes with significant changes are highlighted in red (upregulated) and blue (downregulated), with a selection of gene names indicated. A selection of KEGG pathways (B) and GO terms (C) enriched for the differentially regulated genes are highlighted.

Figure 3

Principal component analysis of the samples included in the RNA-sequencing meta-analysis. The raw RNA-Seq datasets of the studies using the mouse model of myopia (A) and chick model of myopia (C) before any processing. The RNA-Seq samples of the mouse (B) and chick studies (D) after batch correction and normalization.

Figure 4

Genes and pathways differentially regulated in mouse retinas in response to experimental myopia. (A) A volcano plot of genes differentially regulated in response to experimental myopia in mouse retinas, indicating the log₂ effect size and unadjusted log-transformed p values. The gray dashed horizontal line indicates an unadjusted p value of 0.05. A selection of genes significantly changing are highlighted in red (upregulated) and blue (downregulated), with the gene names indicated. (B) A selection of KEGG pathways enriched for the differentially regulated genes are highlighted. (C) The genes contributing to the pathways illustrated in (B) are shown on the horizontal axis, colors represent the directionality of gene expression change in experimental myopia.

Figure 5

Genes and pathways differentially regulated in chick retinas in response to experimental myopia. (A) A volcano plot of genes differentially regulated in response to experimental myopia in chick retinas, indicating the log₂ effect size and unadjusted log-transformed p values. The gray dashed horizontal line indicates an unadjusted p value of 0.05. A selection of genes significantly changing are highlighted in red (upregulated) and blue (downregulated), with the gene names indicated. (B) A selection of KEGG pathways enriched for the differentially regulated genes are highlighted. (C) The genes contributing to the pathways illustrated in (B) are shown on the horizontal axis, colors represent the directionality of gene expression change in experimental myopia.

Figure 6

Overlap between genes and enriched pathways differentially regulated in experimental myopia in mouse and chick retinas, and those implicated in refractive error in human GWA studies. (A) Venn diagram of genes differentially regulated in experimental myopia in the mouse and chick retina and genes implicated in refractive error in humans show one gene associated with all three gene groups. (B) Venn diagram of pathways enriched for genes in (A). A subset of genes and pathways most relevant in the context of the retina and refractive development are illustrated on the graphs. GWAS, genome-wide association study.