Integrative Analysis of Proteomics and DNA Methylation in Orbital Fibroblasts From Graves' Ophthalmopathy

Abstract

Background: Graves' ophthalmopathy (GO) is a frequent extrathyroidal complication of Graves' hyperthyroidism. Orbital fibroblasts contribute to both orbital tissue inflammation and remodeling in GO, and as such are crucial cellular elements in active GO and inactive GO. However, so far it is largely unknown whether GO disease progression is associated with functional reprogramming of the orbital fibroblast effector function. Therefore, the aim of this study was to compare both the proteome and global DNA methylation patterns between orbital fibroblasts isolated from active GO, inactive GO and healthy controls.

Methods: Orbital fibroblasts from inactive GO (n=5), active GO (n=4) and controls (n=5) were cultured and total protein and DNA was isolated. Labelled and fractionated proteins were analyzed with a liquid chromatography tandem-mass spectrometer (LC-MS/MS). Data are available via ProteomeXchange with identifier PXD022257. Furthermore, bisulphite-treated DNA was analyzed for methylation pattern with the Illumina Infinium Human Methylation 450K beadchip. In addition, RNA was isolated from the orbital fibroblasts for real-time quantitative (RQ)-PCR. Network and pathway analyses were performed.

Results: Orbital fibroblasts from active GO displayed overexpression of proteins that are typically involved in inflammation, cellular proliferation, hyaluronan synthesis and adipogenesis, while various proteins associated with extracellular matrix (ECM) biology and fibrotic disease, were typically overexpressed in orbital fibroblasts from inactive GO. Moreover, orbital fibroblasts from active GO displayed hypermethylation of genes that linked to inflammation and hypomethylated genes that linked to adipogenesis and autoimmunity. Further analysis revealed networks that contained molecules to which both hypermethylated and hypomethylated genes were linked, including NF-κB, ERK1/2, Alp, RNA polymerase II, Akt and IFNα. In addition, NF-κB, Akt and IFNα were also identified in networks that were derived from the differentially expressed proteins. Generally, poor correlation between protein expression, DNA methylation and mRNA expression was observed.

Conclusions: Both the proteomics and DNA methylation data support that orbital fibroblasts from active GO are involved in inflammation, adipogenesis, and glycosaminoglycan production, while orbital fibroblasts from inactive disease are more skewed towards an active role in extracellular matrix remodeling. This switch in orbital fibroblast effector function may have therapeutic implications and further studies into the underlying mechanism are thus warranted.

Keywords: DNA methylation; epigenetics; graves’ ophthalmopathy; orbital fibroblast; proteomics.

Figure 1

Proteomic profiles of orbital fibroblasts from inactive GO, active GO and healthy controls. (A) Cluster analysis of the differential protein expression between orbital fibroblasts from patients with inactive GO (n = 5), active GO (n = 4) and healthy controls (n = 5). Differentially expressed proteins grouped in three main clusters. Proteins within cluster 1 proteins were expressed at slightly higher level in orbital fibroblasts from active GO compared with orbital fibroblasts from inactive GO and controls. Proteins within cluster 2 displayed higher expression in orbital fibroblasts from inactive GO compared with orbital fibroblasts from active GO and controls. Proteins within cluster 3 were expressed at a higher level in orbital fibroblasts from active GO compared to inactive GO and controls. (B) Gene and protein expression level from proteins within cluster 2. Protein expression levels were compared to high abundant protein expression level. Four proteins (PSMB4, FBN2, COL6A1 and NCAM2) up-regulated in inactive GO orbital fibroblasts compared to active GO were related to extracellular matrix biology and further determined by RQ-PCR. * and *** indicate p-value < 0.05 and < 0.001, respectively. Individual symbols represent orbital fibroblast cultures from individual patients. Horizontal bar depicts the median. (C) Gene and protein expression level from proteins within cluster 3. Protein expression levels were compared to high abundant protein expression level. Seven proteins (PACSIN3, NFKB1, SMC3, GFER, GSDMD, UGDH and MT1X) down-regulated in inactive GO orbital fibroblasts compared to active GO related to inflammation, hyaluronan and adipogenesis were determined by RQ-PCR. * and ** indicate p-value < 0.05 and < 0.01, respectively. Individual symbols represent orbital fibroblast cultures from individual patients. Horizontal bar depicts the median.

Figure 2

Proteome based network analysis. (A) Network 1: linked to cellular development and connective tissue disorders. Differentially expressed proteins are highlighted in grey. The differentially expressed proteins labelled in red were the up-regulated while differentially expressed proteins labelled in blue were down-regulated in inactive GO comparing to active GO orbital fibroblasts. (B) Network 2: linked to free radical scavenging, DNA replication, recombination and repair and cellular assembly and organization. Differentially expressed proteins are highlighted in grey. The differentially expressed proteins labelled in red were up-regulated while differentially expressed proteins labelled in blue were down-regulated in inactive GO comparing to active GO orbital fibroblasts.

Figure 3

Comparison of orbital fibroblast global DNA methylation profiles with FDR <0.05. (A) DNA methylation in active GO orbital fibroblasts (n = 4), inactive GO orbital fibroblasts (n = 4) and healthy control orbital fibroblasts (n = 4) were compared. Global DNA methylation was measured using the Illumina Infinium Human Methylation 450K beadchip on bisulphite-treated DNA and analyzed by GenomeStudio methylation analysis package. Genes with differential DNA methylation were further corrected for multiple testing using FDR < 0.05. (B) Methylation level of SLC39A8. Methylation level ranges from 0 (unmethylated) to 1 (fully methylated). Individual symbols represent orbital fibroblast cultures from individual patients. Horizontal bar depicts the median. (C) SLC39A8 expression was determined by real-time quantitative (RQ)-PCR and normalized to the control gene ABL. Data from mRNA expression was analyzed using ANOVA and subsequently analyzed with the Mann Whitney U test. Individual symbols represent orbital fibroblast cultures from individual patients. Horizontal bar depicts the median.

Figure 4

Hypermethylated genes in orbital fibroblasts from active GO in comparison to orbital fibroblasts from inactive GO and healthy controls with a fold difference ≥ 2. DNA methylation in active GO orbital fibroblasts (n = 4) was compared with inactive GO orbital fibroblasts (n = 4). Global DNA methylation was measured using the Illumina Infinium Human Methylation 450K beadchip on bisulphite-treated DNA and analyzed by GenomeStudio methylation analysis package. Hypermethylated genes with differences in DNA methylation level of at least 2-fold in active GO orbital fibroblasts were clustered as shown in the heatmap.

Figure 5

Hypomethylated genes in orbital fibroblasts from active GO in comparison to orbital fibroblasts from inactive GO and healthy controls with a fold difference ≥ 2. DNA methylation in active GO orbital fibroblasts (n = 4) were compared with inactive GO orbital fibroblasts (n = 4). Global DNA methylation was measured using the Illumina Infinium Human Methylation 450K beadchip on bisulphite-treated DNA and analyzed by GenomeStudio methylation analysis package. Hypomethylated genes with differences in DNA methylation level of at least 2-fold in active GO orbital fibroblasts were clustered as shown in the heatmap.

Figure 6

Comparison of DNA methylation, mRNA expression and proteomic data. (A) Comparison of hypermethylated genes, mRNA expression and proteomic data. Gene expression was determined by real-time quantitative (RQ)-PCR and normalized to the control gene ABL. Data from mRNA expression was analyzed using ANOVA and subsequently analyzed with the Mann Whitney U test. Individual symbols represent orbital fibroblast cultures from individual patients. Horizontal bar depicts the median. (B) Comparison of hypomethylated genes, mRNA expression and proteomic data. Gene expression was determined by real-time quantitative (RQ)-PCR and normalized to the control gene ABL. Data from mRNA expression was analyzed using ANOVA and subsequently analyzed with the Mann Whitney U test. Individual symbols represent orbital fibroblast cultures from individual patients. Horizontal bar depicts the median.

Figure 7

Methylation based network analysis. (A) Network 1 derived from hypermethylated genes in orbital fibroblasts from active GO: linked to neurological disease, organismal injury and abnormalities, and nervous system development and function Network analysis was further performed on these differentially hypermethylated genes (highlighted in grey) using Ingenuity (Qiagen) filtering only for those networks and pathways that are based on experimentally obtained information. (B) Network 2 derived from hypermethylated genes in orbital fibroblasts from active GO: linked to embryogenic development, nervous system development and function, and organ development. Network analysis was further performed on these differentially hypermethylated genes (highlighted in grey) using Ingenuity (Qiagen) filtering only for those networks and pathways that are based on experimentally obtained information. (C) Network 3 derived from hypermethylated genes in orbital fibroblasts from active GO: linked to cardiovascular system development and function, organismal development, and cellular assembly and organization. Network analysis was further performed on these differentially hypermethylated genes (highlighted in grey) using Ingenuity (Qiagen) filtering only for those networks and pathways that are based on experimentally obtained information. (D) Network 4 derived from hypomethylated genes in orbital fibroblasts from active GO: linked to cellular growth and proliferation, nervous system development and function, and cell-to-cell signaling and interaction. Network analysis was further performed on these differentially hypomethylated genes (highlighted in grey) using Ingenuity (Qiagen) filtering only for those networks and pathways that are based on experimentally obtained information.

Figure 8

Model depicting orbital fibroblast effector characteristics upon GO disease progression. During active disease orbital fibroblasts are skewed towards a functional phenotype involved in in inflammation, glycosaminoglycan synthesis and adipogenesis. With disease progression the orbital fibroblasts undergo a switch from this “inflammatory/pro-adipogenic primed” effector cell towards a “remodelling/pro-fibrotic” type of effector cell.