November 2022
Volume 63, Issue 12
Open Access
Genetics  |   November 2022
Differential Circular RNA Expression Profiling of Orbital Connective Tissue From Patients With Type I and Type II Thyroid-Associated Ophthalmopathy
Author Affiliations & Notes
  • Huijing Ye
    State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou, China
  • Anqi Sun
    State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou, China
  • Wei Xiao
    State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou, China
  • Te Zhang
    State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou, China
  • Zhihui Xu
    State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou, China
  • Lu Shi
    State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou, China
  • Xiaotong Sha
    State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou, China
  • Huasheng Yang
    State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou, China
  • Correspondence: Huasheng Yang and Huijing Ye, State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, 1 Zhongshan Ophthalmic Centre, Sun Yat-sen University, Guangdong Provincial Key Laboratory of Ophthalmology and Visual Science, Guangzhou 510060, China; yanghs64@126.com and yehuijing@qq.com
  • Footnotes
    *  HY and AS contributed equally as co-first authors.
Investigative Ophthalmology & Visual Science November 2022, Vol.63, 27. doi:https://doi.org/10.1167/iovs.63.12.27
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Huijing Ye, Anqi Sun, Wei Xiao, Te Zhang, Zhihui Xu, Lu Shi, Xiaotong Sha, Huasheng Yang; Differential Circular RNA Expression Profiling of Orbital Connective Tissue From Patients With Type I and Type II Thyroid-Associated Ophthalmopathy. Invest. Ophthalmol. Vis. Sci. 2022;63(12):27. https://doi.org/10.1167/iovs.63.12.27.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: The purpose of this study was to investigate the molecular mechanism underlying thyroid-associated ophthalmopathy (TAO) clinical subtypes, to do so, we performed transcriptomic analysis to reveal the expression profile of circular RNAs (circRNAs) in TAO subtypes.

Methods: High-throughput RNA-sequencing was performed in six pairs of type I and type II orbital connective tissue samples from patients with TAO. The expression levels of circRNAs and mRNAs in type I and type II samples were measured by quantitative real-time polymerase chain reaction (qRT-PCR) in another three pairs of type I and type II TAO connective tissue samples. We used bioinformatics predictions to construct a circRNA-microRNA (miRNA)-mRNA network. A protein-protein interaction (PPI) network was constructed based on differential mRNA expression, and the hub genes were determined by the Cytoscape software plugin. Functional and pathway enrichment analyses were performed to elucidate circRNA function. Lentiviral-mediated overexpression of hsa_circ_0007006 and the relationship between hsa_circ_0007006 with COL1A1 and MMP2 were evaluated by Western blotting (WB). Moreover, the differential pathways were assessed by WB.

Results: RNA sequencing results predicted a total of 7489 circRNAs and 15,803 mRNAs, with 94 upregulated and 76 downregulated circRNAs and 488 upregulated and 138 downregulated mRNAs. The qRT-PCR analysis of seven dysregulated circRNAs and two major mRNAs validated the RNA-sequencing data. The competing endogenous RNA (ceRNA) network included 7 circRNAs, 23 miRNAs, and 262 mRNAs. Functional analysis revealed several important pathways. Overexpression of hsa_circ_0007006 led to decreased expression levels of COL1A1 and MMP2. Activation of the relaxin signaling pathway differed between the two subtypes.

Conclusion: We showed that circRNAs are differentially expressed between type I and type II TAO. We speculate that the hsa_circ_0007006-COL1A1 and MMP2-relaxin signaling pathways are important regulatory axes in the pathogenesis of this disease type and could be considered promising diagnostic and therapeutic targets in the future.

Thyroid-associated ophthalmopathy (TAO) is the most common orbital disease in adults.1 Autoantigens and the immune system are considered critical for the pathogenesis of TAO.2 Approximately 20 to 25% of patients with Graves’ disease have ocular symptoms, including proptosis, strabismus, and sight-threatening corneal ulceration or compressive optic neuropathy.3,4 TAO can be divided into two clinical subtypes, type I (mainly fat) and type II (mainly muscle), representing two seemingly independent but overlapping pathophysiological processes associated with very different prognoses. Compared to type I TAO, type II TAO is associated with more severe orbital fibrosis, restricted strabismus, vision loss, and even blindness.57 The mechanism underlying the clinical type of TAO is not clear. Orbital fibroblasts (OFs), acting as target and effector cells, can respond to extracellular stimulation and differentiate into adipose tissue (type I) or fibrosis (type II).8,9 The mechanism of differentiation is critical for the clinical subtype and outcomes of TAO. 
Circular RNA (circRNA) is a recently identified distinct class of noncoding RNAs (ncRNAs) that are abundant in the cytoplasm of eukaryotic cells, where the upstream exon 5′ and downstream 3′ ends form a continuous loop during the process of mRNA splicing.10 The circRNAs have many regulatory functions; for example, they act as competing endogenous RNAs (ceRNAs) by binding to cellular microRNAs (miRNAs) to block the inhibitory effect of miRNAs on their target genes.11 Increasing evidence shows that abnormal ncRNA expression profiles are closely associated with immune diseases and inflammation.12 The circRNAs have been confirmed to have regulatory functions in immune diseases, such as rheumatoid arthritis,13 retinal vascular dysfunction in diabetes mellitus,14 and systemic lupus erythematosus.15 However, the mechanism of TAO typing still lacks relevant research. 
In this study, we focused on differences in the circRNA expression profiles between type I TAO and type II TAO using high-throughput RNA sequencing (RNA-seq). Furthermore, circRNAs and mRNAs were validated by quantitative real-time polymerase chain reaction (qRT-PCR). A circRNA-miRNA-mRNA network was constructed, and comprehensive bioinformatic analysis was used for the validation of circRNAs to explore the molecular mechanism of TAO typing. This work may identify some meaningful biomarkers and provide new breakthroughs in the molecular targeted pathogenesis and therapy of TAO and other immunological diseases. 
Methods
Patient Selection and Sample Collection
We recruited 12 patients for high-throughput RNA sequencing, including 6 patients with type I TAO and 6 patients with type II TAO from Zhongshan Ophthalmic Center at Sun Yat-sen University. Orbital connective tissues were collected from patients with TAO who underwent orbital decompression surgery. High-throughput transcription sequencing and bioinformatics were performed. Another three type I TAO and type II TAO fat tissue samples were collected for the validation of the selected circRNAs and mRNAs. The enrolled patients met the criteria for TAO according to Bartley,16 and the clinical typing was performed according to Nunery.7 Patients with TAO were divided into the type I group or the type II group based on whether diplopia was present within 20 degrees of the primary position and restrictive motility impairment in any position of gaze. Subtype I was predominantly lipogenic, whereas subtype II was myogenic dominantly. Tomography showed an increase in the extraocular musculature of patients with type II TAO. In addition, the patients were in an inactive phase and moderate to severe stage that needed operation consistent with the natural history of TAO suggested by Rundle's Curve.17 Basic clinical information of the research subjects is shown in Table 1Table 2 lists the clinical features of patients with TAO for high throughput RNA-sequencing. These patients agreed to surgical treatment, understood the purpose of the investigation, and agreed to cooperate with the study. This study was approved by the Institutional Review Board of Zhongshan Ophthalmic Center (2016KYPJ028). 
Table 1.
 
Basic Clinical Information of the Research Subjects
Table 1.
 
Basic Clinical Information of the Research Subjects
Table 2.
 
Clinical Features of Patients With TAO for High Throughput RNA-Sequencing
Table 2.
 
Clinical Features of Patients With TAO for High Throughput RNA-Sequencing
Isolation of RNA and Quality Control
Total RNA of orbital connective tissue was extracted using a HiPure Total RNA Mini Kit (Magen, Guangzhou, China) according to the manufacturer's protocol. The K2800 Nucleic Acid Analyzer measures the optical density of total RNA at 260 nm and 280 nm to quantify RNA. The RNA purity was analyzed by agarose gel electrophoresis. Qubit 3.0 was used to detect the RNA sample concentration. The Agilent 2100 Bioanalyzer was used to detect RNA integrity. 
RNA Library Construction and High-Throughput Sequencing
A total RNA-seq (H/M/R) Library Prep Kit for the Illumina construct RNA Library hybridized the designed DNA probe with the RNA sample and removed the rRNA from the total RNA. Qubit 3.0 preliminary quantification and an Agilent 2100 Bioanalyzer evaluated the size range of the library and confirmed the size of the inserted target fragment, and quantitative PCR accurately quantified the effective concentration of the library (>3 nM). According to the effective concentration and target off-machine data pooling, HiSeq X10 PE150 platform sequencing was performed. The amount of sequencing data was 10 G. DCC software was used to comprehensively identify RNAs and obtain high-confidence RNAs. Furthermore, sequence prediction, expression value calculation, and expression difference analysis of the identified RNA were carried out. 
Quantitative Real-Time Polymerase Chain Reaction
The expression levels of chosen circRNA and mRNA for validation were assessed by qRT-PCR. The cDNA was composited from total RNA by PrimeScript RT Master Mix (TaKaRa, Dalian, China). The qRT-PCR was performed on a Roche LightCycler 480 (Roche, Basel, Switzerland) using TB Green Premix Ex Taq II (TaKaRa, Dalian, China). The primer pair sequences for qRT-PCR are listed in Table S1. GAPDH was selected as a reference. 
ceRNA Network Construction and Functional Enrichment Analysis
We used the miRanda, Circbank, and Circinteractome databases to screen target miRNAs of the validated circRNAs. Only predictions by the three databases can be obtained. Then, the target genes of these miRNAs were predicted with the miRwalk, TargetScan, and miRDB databases, and the intersecting genes were further filtered by the differentially expressed mRNAs retained from the RNA-seq results. The ceRNA network of circRNA-miRNA–mRNA was constructed by Cytoscape software (version 3.8.2). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional enrichment analyses further identified the function and signaling pathways of these target mRNAs using the ClueGO plug-in of the Cytoscape software. 
Construction of the PPI Network and Identification of Hub Genes
The Search Tool for the Retrieval of Interacting Genes (STRING) database was used to construct the PPI network for differentially expressed genes between type I and type II groups, and then the hub genes were identified by CytoHubba, a plugin of Cytoscape software (version 3.8.2). 
Overexpression of hsa_circ_0007006 in OFs
A lentiviral vector was constructed to transfect OFs and to overexpress hsa_circ_0007006. (Hanbio Biotechnology Co. Ltd., Shanghai, China). Puromycin was used to select for transfection efficiency. 
Western Blotting
Protein was extracted from orbital tissue and OFs with a protein extraction kit (KeyGEN, Nanjing, China). The concentration was measured by using the BCA method (Cwbiotech, Beijing, China). Protein lysates was electrophoresed, transferred to a membrane, blocked, and incubated with primary and secondary antibodies. Primary antibodies included COL1A1, MMP2, GAPDH, β-actin (Cell Signaling Technology, Beverly, MA, USA), and RLN2 (Abcam, Waltham, MA, USA). Secondary antibodies were purchased from CST. Finally, the blots were imaged with a chemiluminescence imager (Tanon Science &Technology Co. Ltd., Shanghai, China). 
Statistical Analysis
To compare the expression levels of circRNAs and mRNAs between type I and type II TAO, the Student's t-test was used. The descriptive data, such as rates and ratios, were analyzed with the χ2 test. Descriptive data were tested by paired sign. A P value < 0.05 was considered statistically significant. 
Results
circRNA Expression Profile
A total of 7489 circRNAs were predicted from RNA-seq of type I TAO and type II TAO samples using DCC software. A total of 170 circRNAs significantly differentially expressed between the 2 subtypes were selected (fold change [FC] ≥ 2.0 and adjusted P value (P-adj) < 0.05), with 94 circRNAs upregulated and 76 circRNAs downregulated in type II TAO compared with type I TAO (Fig. 1A). Hierarchical clustering analysis indicated that type I TAO and type II TAO could be discriminated according to the expression of these circRNAs (Fig. 1B). The majority of dysregulated circRNAs originated from exons, and some were from introns and intergenic regions (Fig. 1C). The majority of circRNAs were less than 1600 bp in length (Fig. 1D). 
Figure 1.
 
CircRNA expression patterns in type II TAO relative to type I TAO. (A) Volcano plot of differentially expressed circRNAs between type II and type I TAO. (B) Hierarchical cluster analysis of differentially expressed circRNAs. (C) Origination of circRNAs. (D) Length of exonic circRNAs.
Figure 1.
 
CircRNA expression patterns in type II TAO relative to type I TAO. (A) Volcano plot of differentially expressed circRNAs between type II and type I TAO. (B) Hierarchical cluster analysis of differentially expressed circRNAs. (C) Origination of circRNAs. (D) Length of exonic circRNAs.
mRNA Expression Profile
A total of 15,803 mRNAs were predicted from RNA-seq of type I TAO and type II TAO samples. A total of 626 mRNAs were significantly differentially expressed between the two subtype tissues (FC ≥ 2.0 and P-adj < 0.05), with 488 mRNAs upregulated and 138 mRNAs downregulated in type II TAO compared with type I TAO. Volcano plots and heat map patterns are shown in Figure 2
Figure 2.
 
The mRNA expression patterns in type II TAO relative to type I TAO. (A) Volcano plot of differentially expressed mRNAs between type II and type I TAO. (B) Heatmap of differentially expressed mRNAs.
Figure 2.
 
The mRNA expression patterns in type II TAO relative to type I TAO. (A) Volcano plot of differentially expressed mRNAs between type II and type I TAO. (B) Heatmap of differentially expressed mRNAs.
Validation of the Selected circRNAs and mRNAs
To validate the RNA-sequencing outcomes, qRT–PCR was used to test seven dysregulated circRNAs and two mRNAs. The validated RNAs that showed the most obvious differential expression and had clinical significance were selected. Seven circRNAs, including hsa_circ_0004447, hsa_circ_0004222, hsa_circ_0007006, hsa_circ_0097201, hsa_circ_0008775, hsa_circ_0104799, and hsa_circ_0128243, were significantly differentially expressed between the two TAO subtypes, consistent with the RNA-seq results (P < 0.05). hsa_circ_0004222 was upregulated, and other circRNAs were downregulated (Fig. 3). Table 3 lists the information of the validated circRNAs. MMP2 and COL1A1 mRNA levels were also validated. 
Figure 3.
 
Validation results for the selected circRNAs and mRNAs. The relative expression of circRNAs and mRNAs were analyzed in another 3 type II TAO, and 3 type I TAO orbital fat tissues.
Figure 3.
 
Validation results for the selected circRNAs and mRNAs. The relative expression of circRNAs and mRNAs were analyzed in another 3 type II TAO, and 3 type I TAO orbital fat tissues.
Table 3.
 
The Validated CircRNAs Ranked by Fold Change
Table 3.
 
The Validated CircRNAs Ranked by Fold Change
Construction of the ceRNA Network
The ceRNA network was constructed with the seven differentially expressed circRNAs (hsa_circ_0004447, hsa_circ_0004222, hsa_circ_0007006, hsa_circ_0097201, hsa_circ_0008775, hsa_circ_0104799, and hsa_circ_0128243) based on the ceRNA theory that circRNA may share a common binding site of miRNA response elements (MREs) with mRNA to regulate each other. Twenty-three miRNAs and 262 mRNAs were contained in this network (Fig. 4). The hsa_circ_0007006 was correlated with upregulated mRNA MMP2 and COL1A1. 
Figure 4.
 
circRNAs-miRNAs-mRNAs network. The ceRNA network map consists of 7 differentially expressed circRNAs, 23 miRNAs, and 262 mRNAs. Red inverted triangles represent the circRNAs, the blue diamonds represent the miRNAs, and the pink circles represent the mRNAs.
Figure 4.
 
circRNAs-miRNAs-mRNAs network. The ceRNA network map consists of 7 differentially expressed circRNAs, 23 miRNAs, and 262 mRNAs. Red inverted triangles represent the circRNAs, the blue diamonds represent the miRNAs, and the pink circles represent the mRNAs.
PPI Network Construction and Analysis
We established a PPI network with 262 intersecting and differentially expressed mRNAs based on data from the STRING database (Fig. 5A). The top 10 hub genes and their interactors were constructed by Cytoscape software (Fig. 5B). MMP 2 and COL1A1 were among these genes. 
Figure 5.
 
Top 10 hub genes identified in the PPI network. (A) PPI network construction for differentially expressed genes. (B) The top 10 hub genes.
Figure 5.
 
Top 10 hub genes identified in the PPI network. (A) PPI network construction for differentially expressed genes. (B) The top 10 hub genes.
Functional Enrichment Analysis of circRNA-Targeted Genes
To clarify the potential biological function and signaling pathways of these differentially expressed circRNAs, GO and KEGG analyses were applied to annotate the functions of the circRNA target genes. RNA-seq analysis was combined with the bioinformatic prediction, and 262 mRNAs were identified. According to GO functional enrichment analysis, 78 biological process (BP) terms, 37 cellular component (CC) terms, and 18 molecular function (MF) terms were included. Target genes mainly related to the BP terms were extracellular matrix organization, collagen fibril organization, T-cell activation, and T-cell differentiation; the CC terms were extracellular matrix and collagen network; and the MF terms were growth factor binding and cytokine binding. Additionally, 16 signaling pathways were revealed by KEGG analysis, and T helper 17 (Th17)-cell differentiation, the PI3K-Akt signaling pathway, extracellular matrix (ECM)-receptor interactions, and the relaxin signaling pathway were involved (Fig. 6Tables 45). 
Figure 6.
 
Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis for circRNA-target mRNAs. (A) The top 10 GO terms in biological process (BP), cellular component (CC), and molecular function (MF), respectively, for target mRNAs. (B, C) The KEGG pathways for target mRNAs.
Figure 6.
 
Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis for circRNA-target mRNAs. (A) The top 10 GO terms in biological process (BP), cellular component (CC), and molecular function (MF), respectively, for target mRNAs. (B, C) The KEGG pathways for target mRNAs.
Table 4.
 
The Top 30 Gene Ontology (GO) Terms in Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) for Target mRNAs
Table 4.
 
The Top 30 Gene Ontology (GO) Terms in Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) for Target mRNAs
Table 5.
 
Kyoto Encyclopedia of Genes and Genomes (KEGG) Analysis for circRNA-Target mRNAs
Table 5.
 
Kyoto Encyclopedia of Genes and Genomes (KEGG) Analysis for circRNA-Target mRNAs
The Correlation Between hsa_circ_0007006 and the Hub Genes
Lentivirus was used to transfect OFs and overexpress hsa_circ_0007006. The multiplicity of infection (MOI) was 100. Puromycin improved the transfection efficiency to 90%. We found that overexpression of hsa_circ_0007006 led to a decrease in the expression level of the hub genes COL1A1 and MMP2 as compared to the control group (Fig. 7). 
Figure 7.
 
The hsa_circ_0007006 was correlated with COL1A1 and MMP2. (A) Lentiviral transfected OFs to overexpress hsa_circ_0007006. (B) The expression relationship between hsa_circ_0007006 and COL1A1, MMP2.
Figure 7.
 
The hsa_circ_0007006 was correlated with COL1A1 and MMP2. (A) Lentiviral transfected OFs to overexpress hsa_circ_0007006. (B) The expression relationship between hsa_circ_0007006 and COL1A1, MMP2.
The Relaxin Signaling Pathway was Involved in TAO Subtyping
Type I and type II TAO orbital connective tissue protein were used to perform Western blotting (WB). Compared to type I, the relaxin signaling pathway was significantly more activated in type II (Fig. 8). 
Figure 8.
 
Western blotting of relaxin signaling pathway expression in type I and type II TAO.
Figure 8.
 
Western blotting of relaxin signaling pathway expression in type I and type II TAO.
Discussion
TAO is the most common orbital disease in adults.1 Patients with TAO have various clinical features. According to clinical presentations, TAO can be divided into two clinical subtypes, type I (mainly fat) and type II (mainly muscle),7 representing seemingly two independent but overlapping pathophysiological processes. Type I TAO is defined as the presence of mainly orbital fat expansion with mild to moderate extraocular muscle enlargement but without diplopia. In contrast, type II TAO involves more muscle swelling and diplopia within 20 degrees of fixation. Compared to patients with type I TAO, patients with type II TAO always have a specific inflammation period and can report the exact time when inflammation initiated and led to muscle restriction. However, the progression of type I TAO is usually “quiet” and leads to orbital adipose hypertrophy and mild to moderate extraocular muscle enlargement without diplopia. The difference in myopathy between the two subtypes indicates that immune events may target extraocular muscles.7 Fibrosis is considered to be the terminal stage of TAO, and it often accompanies chronic diseases and immune diseases.18 The different clinical types have very different outcomes; patients with type II TAO tend to have more severe orbital fibrosis, restricted strabismus, vision loss, and even blindness. OFs, acting as target and effector cells, can respond to extracellular stimulation and differentiate into adipose cells (type I) or myofibroblasts (type II).8,9 Thus, the mechanism of differentiation is critical for the clinical subtype and outcomes of TAO. Inflammation, adipogenesis, and fibrosis may be key points. 
In recent years, studies have found that epigenetic factors play an important role in the progression and development of TAO.19 A large number of studies have shown that, as a component of epigenetic factors, ncRNAs can regulate gene expression through multiple genetic mechanisms and are not inferior to proteins.2022 The circRNAs can act as ceRNAs, binding to cellular miRNAs to block the inhibitory effect of miRNAs on their target genes.11 Increasing evidence shows that abnormal ncRNA expression profiles are closely associated with immune diseases and inflammation.12 However, little research has been performed on TAO, not to mention the clinical TAO subtypes. To our knowledge, our study is the first study to evaluate the differences in the regulatory function of circRNAs between type I TAO and type II TAO. 
In this study, we investigated the circRNA expression profiles in the type I TAO and the type II TAO groups using high-throughput RNA-seq. The age of type II was significantly older than type I TAO, consistent with previous studies reported.23 TAO duration time and Clinical Activity scores (CAS) did not differ between the two groups. Overall, 170 significantly differentially expressed circRNAs, 94 upregulated circRNAs, and 76 downregulated circRNAs, were identified in type II TAO compared to type I TAO. Most of these dysregulated circRNAs originated from exons. Previous research has revealed that different circRNAs participate in the pathogenesis of TAO, such as the circRNA_14940-CCND1-wnt signaling pathway, which might regulate TAO pathogenesis.24 In our study, seven circRNAs (hsa_circ_0004447, hsa_circ_0004222, hsa_circ_0007006, hsa_circ_0097201, hsa_circ_0008775, hsa_circ_0104799, and hsa_circ_0128243) were validated through qRT–PCR, which was in accordance with the RNA-seq results. The hsa_circ_0004447 is derived from PRKCE, which is considered to be associated with the esterification of fatty acids, lipid partitioning, and the production of reactive oxygen species (ROS).25 ROS produced by mitochondria and NADPH oxidase (Nox) regulate fibrosis via the TGF-β-dependent Smad, PI3K, and MAPK signaling pathways.26,27 VPS13D is the parental gene of hsa_circ_0004222, and VPS13D is known to promote mitochondrial clearance by mitochondrial autophagy (mitophagy).28 Previous research has found mitochondrial abnormalities in the extraocular muscles of patients with TAO.29 VEZT is the host gene of hsa_circ_0008775, and it has been reported that this gene may also be inhibited by the binding of a specific miRNA to a target sequence in the 3' untranslated region (UTR) of the transcripts in several diseases, such as gastric cancer.30 The hsa_circ_0104799 originates from the acetyl-CoA carboxylase beta (ACACB) gene. Acetyl-CoA carboxylase (ACC) has 2 isoforms, ACC1 and ACC2, which coordinate the synthesis and oxidation of fatty acids in the liver.31 The hsa_circ_0104799 comes from the gene for AKAP13, which is a scaffold protein with GEF activity that could serve as an activator of NF-κB downstream of the TLR2 signaling pathway.32 In an in vitro study, the activation of PI3K in a TLR2-dependent manner increased hyaluronic acid synthase in OFs from patients with TAO.33 Taken together, these findings show that the dysregulated circRNAs identified in the orbital adipose tissues of patients with different TAO subtypes may have effects on the differentiation of OFs and thus may play a key role in TAO typing. 
The circRNAs have many regulatory functions; for example, circRNAs act as ceRNAs by binding to cellular miRNAs to block the inhibitory effect of miRNAs on their target genes.11 We constructed a ceRNA crosstalk network to further explore the role of dysregulated circRNAs in the two TAO subtypes. Seven validated dysregulated circRNAs (hsa_circ_0004447, hsa_circ_0004222, hsa_circ_0007006, hsa_circ_0097201, hsa_circ_0008775, hsa_circ_0104799, and hsa_circ_0128243) were found to target 23 miRNAs. The miRNA-557 and miR-587 were correlated with several circRNAs. Past studies have shown that miRNAs play a crucial role in the progression of immune diseases. For example, miRNA-557 inhibits the differentiation and maturation of megakaryocytes in immune thrombocytopenia.34 The miR-587 was found to regulate the TGFβ/SMAD signaling pathway. HEK293T and HCT116 cells were transfected to overexpress miR-587, and the TGFBR2 and SMAD4 genes were downregulated.35 The TGFβ/SMAD signaling pathway has multiple regulatory functions, such as cell differentiation, ECM production, angiogenesis, and cellular immune responses.36,37 By binding to the 3′UTR of the target gene, miRNAs can repress the translation of mRNAs or degrade them at the posttranscriptional level. These two miRNAs have potential functions in TAO typing. 
A total of 626 mRNAs were significantly differentially expressed in the 2 subtype tissues, with 488 mRNAs upregulated and 138 mRNAs downregulated according to RNA-seq analysis. Combined with bioinformatic prediction, 262 mRNAs were identified. To further determine the role of those target genes contained in the ceRNA network, we also performed GO and KEGG analyses, and we obtained 78 BP terms, 37 CC terms, 18 MF terms, and 16 important signaling pathways, many of which were consistent with the current knowledge on TAO. For example, extracellular matrix organization (GO: 0030198) was included as a BP term. Autoantibodies can bind to ECM proteins in vivo, playing an important role in orbital tissue inflammation.38 Through KEGG analysis, we found that the PI3K-Akt signaling pathway, Th17-cell differentiation, and the relaxin signaling pathway were involved. OFs express TSHR, a G protein-coupled cell surface protein.39 The activation of TSHR signaling can lead to increased expression of genes involved in inflammation.40 TSH could induce IL-1RA in OFs by activating the PI3K/AKT pathway.41 Th17 cells, a subset of helper T (Th) cells, are characterized by the production of cytokines, such as IL-17A and IL-17F. IL-17A has been confirmed to induce OFs to secrete proinflammatory cytokines and promote excessive ECM synthesis and deposition in TAO, dependent on MAPK activation.42 Relaxin has a molecular weight of 6 kD and is a kind of endogenous hormone synthesized and secreted by several tissues in the human body, including reproductive organs and nonreproductive organs. Relaxin 2 binds to its receptor, RXFP1, via specific pathways, such as the Notch signaling pathway and Wnt signaling pathway, to inhibit fibrosis pathological processes.4345 Therefore, these signaling pathways may be critical in the pathogenesis and classification of TAO. KEGG analysis showed that COL1A1 and MMP2, the regulatory genes of hsa_circ_0007006, participated in the relaxin signaling pathway. Furthermore, after PPI network construction and identification of the hub genes with Cytoscape, COL1A1 and MMP2 were found to be among the top 10 hub genes. The mRNA expression levels of COL1A1 and MMP2 were verified in our study. Fibrosis is considered to be the terminal stage of TAO, followed by the inflammation of Muller’s muscle, extraocular muscles, and orbital connective and adipose tissues.1,46 Excessive deposition of ECM was considered to be associated with this progression. According to the manifestations and characteristics of type I and type II TAO, the apparent distinction between them may involve fibrosis. COL1A1 is a marker of fibrosis.47 Matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs) participate in maintaining the balance of the ECM. To better investigate the relationship between hsa_circ_0007006 with COL1A1 and MMP2, a lentivirus was used to overexpress hsa_circ_0007006. We discovered that COL1A1 and MMP2 expression level was reduced compared to the control group. In addition, by performing WB experiments, we found that compared to type I, the relaxin signaling pathway was more activated in type II. In summary, hsa_circ_0007006 may regulate major genes, such as MMP2 and COL1A1, through the relaxin signaling pathway to affect TAO typing. Hence, hsa_circ_0007006 could represent a meaningful diagnostic and prognostic biomarker for TAO subtyping. Additionally, hsa_circ_0007006 has a negative correlation with COL1A1 and MMP2. These findings may present a new breakthrough to treat fibrosis. 
In conclusion, our study revealed circRNA expression profiles in TAO and identified 170 differentially expressed circRNAs between type I TAO and type II TAO. The ceRNA network and functional analysis of potential target genes highlighted the relationship between dysregulated circRNAs and the clinical subtype of TAO, which may provide a new perspective in research on TAO. We speculated that the hsa_circ_0007006-COL1A1 and MMP2-relaxin signaling pathways might be important regulatory axes that might participate in the pathogenesis of this disease type. However, there were several limitations in this study. First, the sample size was small, and only some of the patients who needed to undergo surgery were included. Second, the tissue samples were derived only from orbital adipose/connective tissues, but extraocular muscles are also an affected tissue type; thus, further validation is warranted. Third, the molecular mechanism and function of the present circRNAs should be further tested in vivo and in vitro. Our group has started to study the overall effect and in-depth mechanism of hsa_circ_0007006 and relaxin signaling in TAO, which may be promising biomarkers and therapeutic targets of TAO. 
Acknowledgments
Supported by the National Natural Science Foundation of China (81870689), and the Scientific Research Project of Guangdong Provincial Bureau of Traditional Chinese Medicine (20211077). 
Data Availability Statement: Data openly available in a public repository. 
Disclosure: H. Ye, None; A. Sun, None; W. Xiao, None; T. Zhang, None; Z. Xu, None; L. Shi, None; X. Sha, None; H. Yang, None 
References
Bahn RS . Graves' ophthalmopathy. N Engl J Med. 2010; 362(8): 726–738. [CrossRef] [PubMed]
Sahli E, Gunduz K. Thyroid-associated Ophthalmopathy. Turk J Ophthalmol. 2017; 47(2): 94–105. [CrossRef] [PubMed]
Lazarus JH . Epidemiology of Graves' orbitopathy (GO) and relationship with thyroid disease. Best Pract Res Clin Endocrinol Metab. 2012; 26(3): 273–279. [CrossRef] [PubMed]
Bartley GB, Fatourechi V, Kadrmas EF, et al. Long-term Follow-up of Graves Ophthalmopathy in an Incidence Cohort. Ophthalmology. 1996; 103(6): 958–962. [CrossRef] [PubMed]
Nunery WR, Martin RT, Heinz GW, Gavin TJ. The association of cigarette smoking with clinical subtypes of ophthalmic Graves' disease. Ophthalmic Plast Reconstr Surg. 1993; 9(2): 77–82. [CrossRef] [PubMed]
Sokol JA, Foulks GN, Haider A, Nunery WR. Ocular Surface Effects of Thyroid Disease. The Ocul Surf. 2010; 8(1): 29–39. [CrossRef] [PubMed]
Linquist RA, Symond RCA, O'Bryhim B, Whittaker TJ, Sokol JA. Cytokine profiles in clinical subtypes of ophthalmic Graves' disease. Orbit. 2014; 33(5): 363–368. [CrossRef] [PubMed]
Smith TJ . Insights into the role of fibroblasts in human autoimmune diseases. Clin and Exp Immunol. 2005; 141(3): 388–397. [CrossRef]
Dik WA, Virakul S, van Steensel L. Current perspectives on the role of orbital fibroblasts in the pathogenesis of Graves' ophthalmopathy. Exp Eye Res. 2016; 142: 83–91. [CrossRef] [PubMed]
Jeck WR, Sharpless NE. Detecting and characterizing circular RNAs. Nat Biotechnol. 2014; 32(5): 453–461. [CrossRef] [PubMed]
Ebert MS, Neilson JR, Sharp PA. MicroRNA sponges: competitive inhibitors of small RNAs in mammalian cells. Nat Methods. 2007; 4(9): 721–726. [CrossRef] [PubMed]
Marques-Rocha JL, Samblas M, Milagro FI, Bressan J, Martínez JA, Marti A. Noncoding RNAs, cytokines, and inflammation-related diseases. FASEB J. 2015; 29(9): 3595–3611. [CrossRef] [PubMed]
Ouyang Q, Wu J, Jiang Z, et al. Microarray Expression Profile of Circular RNAs in Peripheral Blood Mononuclear Cells from Rheumatoid Arthritis Patients. Cell Physiol Biochem. 2017; 42(2): 651–659. [CrossRef] [PubMed]
Shan K, Liu C, Liu B-H, et al. Circular Noncoding RNA HIPK3 Mediates Retinal Vascular Dysfunction in Diabetes Mellitus. Circulation. 2017; 136(17): 1629–1642. [CrossRef] [PubMed]
Li LJ, Huang Q, Pan HF, Ye DQ. Circular RNAs and systemic lupus erythematosus. Exp Cell Res. 2016; 346(2): 248–254. [CrossRef] [PubMed]
Bartley GB, Gorman CA. Diagnostic Criteria for Graves' Ophthalmopathy. Am J of Ophthalmol. 1995; 119(6): 792–795. [CrossRef]
Menconi F, Profilo MA, Leo M, et al. Spontaneous improvement of untreated mild Graves' ophthalmopathy: Rundle's curve revisited. Thyroid. 2014; 24(1): 60–66. [CrossRef] [PubMed]
Wynn TA, Ramalingam TR. Mechanisms of fibrosis: therapeutic translation for fibrotic disease. Nat Med. 2012; 18(7): 1028–1040. [CrossRef] [PubMed]
Khong JJ, Wang LY, Smyth GK, et al. Differential Gene Expression Profiling of Orbital Adipose Tissue in Thyroid Orbitopathy. Invest Ophthalmol Vis Sci. 2015; 56(11): 6438–6447. [CrossRef] [PubMed]
Mattick JS, Makunin IV. Non-coding RNA. Hum Mol Genet. 2006; 15(Spec No. 1): R17–R29. [PubMed]
Amaral PP, Mattick JS. Noncoding RNA in development. Mamm Genome. 2008; 19(7-8): 454–492. [CrossRef] [PubMed]
Ghildiyal M, Zamore PD. Small silencing RNAs: an expanding universe. Nat Rev Genet. 2009; 10(2): 94–108. [CrossRef] [PubMed]
Hiromatsu Y, Yang D, Bednarczuk T, Miyake I, Nonaka K, Inoue Y. Cytokine profiles in eye muscle tissue and orbital fat tissue from patients with thyroid-associated ophthalmopathy. J Clin Endocrinol Metab. 2000; 85(3): 1194–1199. [PubMed]
Wu L, Zhou R, Ciao J, et al. Differentially expressed circular RNAs in orbital adipose/connective tissue from patients with thyroid-associated ophthalmopathy. Exp Eye Res. 2020; 196: 108036. [CrossRef] [PubMed]
Raddatz K, Turner N, Grangioudakis G, et al. Time-dependent effects of Prkce deletion on glucose homeostasis and hepatic lipid metabolism on dietary lipid oversupply in mice. Diabetologia. 2011; 54(6): 1447–1456. [CrossRef] [PubMed]
Jain M, Rivera S, Monclaus EA, et al. Mitochondrial reactive oxygen species regulate transforming growth factor-beta signaling. J Biol Chem. 2013; 288(2): 770–777. [CrossRef] [PubMed]
Bracey NA, Gershkovich B, Chun J, et al. Mitochondrial NLRP3 protein induces reactive oxygen species to promote Smad protein signaling and fibrosis independent from the inflammasome. J Biol Chem. 2014; 289(28): 19571–19584. [CrossRef] [PubMed]
Anding AL, Wang C, Chang T-K, et al. Vps13D Encodes a Ubiquitin-Binding Protein that Is Required for the Regulation of Mitochondrial Size and Clearance. Curr Biol. 2018; 28(2): 287–295.e6. [CrossRef] [PubMed]
Wall JR, Stachura I, Kennerdell JH. Mitochondrial abnormalities in eye muscle fiber from three cases of thyroid-associated ophthalmopathy. Thyroid. 2006; 16(11): 1181–1183. [CrossRef] [PubMed]
Aparisi MJ, Aller E, Fuster-García C, et al. Targeted next generation sequencing for molecular diagnosis of Usher syndrome. Orphanet J Rare Dis. 2014; 9: 168. [CrossRef] [PubMed]
McGarry JD, Mannaerts GP, Foster DW. A possible role for malonyl-CoA in the regulation of hepatic fatty acid oxidation and ketogenesis. J Clin Invest. 1977; 60(1): 265–270. [CrossRef] [PubMed]
Shibolet O, Giallourakis C, Rosenberg I, Mueller T, Xavier RJ, Podolsky DK. AKAP13, a RhoA GTPase-specific guanine exchange factor, is a novel regulator of TLR2 signaling. J Biol Chem. 2007; 282(48): 35308–35317. [CrossRef] [PubMed]
Yoo HK, Park H, Hwang HS, Kim HJ, Choi Y-H, Kook KH. Ganglioside GT1b increases hyaluronic acid synthase 2 via PI3K activation with TLR2 dependence in orbital fibroblasts from thyroid eye disease patients. BMB Reports. 2021; 54(2): 136–141. [CrossRef] [PubMed]
Wang Y, Guo Y, Zhang X, et al. The role and mechanism of miR-557 in inhibiting the differentiation and maturation of megakaryocytes in immune thrombocytopenia. RNA Biol. 2021; 18(11): 1953–1968. [CrossRef] [PubMed]
Jahangirimoez M, Medlej A, Tavallaie M, Soltani BM. Hsa-miR-587 Regulates TGFbeta/SMAD Signaling and Promotes Cell Cycle Progression. Cell J. 2020; 22(2): 158–164. [PubMed]
Derynck R, Akhurst RJ, Balmain A. TGF-beta signaling in tumor suppression and cancer progression. Nat Genet. 2001; 29(2): 117–129. [CrossRef] [PubMed]
Dumont N, Arteaga CL. Targeting the TGFβ signaling network in human neoplasia. Cancer Cell. 2003; 3(6): 531–536. [CrossRef] [PubMed]
Bednarczuk T, Stolarski C, Pawlik E, et al. Autoantibodies reactive with extracellular matrix proteins in patients with thyroid-associated ophthalmopathy. Thyroid. 1999; 9(3): 289–295. [CrossRef] [PubMed]
Feliciello A, Porcellini A, Ciullo I, Bonavolontà G, Avvedimento EV, Fenzi G. Expression of thyrotropin-receptor mRNA in healthy and Graves' disease retro-orbital tissue. The Lancet. 1993; 342(8867): 337–338. [CrossRef]
Gillespie EF, Papageorgiou KI, Fernando R, et al. Increased expression of TSH receptor by fibrocytes in thyroid-associated ophthalmopathy leads to chemokine production. J Clin Endocrinol Metab. 2012; 97(5): E740–E746. [CrossRef] [PubMed]
Li B, Smith TJ. PI3K/AKT pathway mediates induction of IL-1RA by TSH in fibrocytes: modulation by PTEN. J Clin Endocrinol Metab. 2014; 99(9): 3363–3372. [CrossRef] [PubMed]
Fang S, Huang Y, Wang S, et al. IL-17A Exacerbates Fibrosis by Promoting the Proinflammatory and Profibrotic Function of Orbital Fibroblasts in TAO. J Clin Endocrinol Metab. 2016; 101(8): 2955–2965. [CrossRef] [PubMed]
Tan J, Tedrow JR, Dutta JA, et al. Expression of RXFP1 Is Decreased in Idiopathic Pulmonary Fibrosis. Implications for Relaxin-based Therapies. Am J Respir Crit Care Med. 2016; 194(11): 1392–1402. [CrossRef] [PubMed]
Wilhelmi T, Xu X, Tan X, et al. Serelaxin alleviates cardiac fibrosis through inhibiting endothelial-to-mesenchymal transition via RXFP1. Theranostics. 2020; 10(9): 3905–3924. [CrossRef] [PubMed]
Martin B, Gabris B, Barakat AF, et al. Relaxin reverses maladaptive remodeling of the aged heart through Wnt-signaling. Sci Rep. 2019; 9(1): 18545. [CrossRef] [PubMed]
Naik V, Khadavi N, Naik MN, et al. Biologic therapeutics in thyroid-associated ophthalmopathy: translating disease mechanism into therapy. Thyroid. 2008; 18(9): 967–971. [CrossRef] [PubMed]
Prabhakar BS, Bahn RS, Smith TJ. Current perspective on the pathogenesis of Graves' disease and ophthalmopathy. Endocr Rev. 2003; 24(6): 802–835. [CrossRef] [PubMed]
Figure 1.
 
CircRNA expression patterns in type II TAO relative to type I TAO. (A) Volcano plot of differentially expressed circRNAs between type II and type I TAO. (B) Hierarchical cluster analysis of differentially expressed circRNAs. (C) Origination of circRNAs. (D) Length of exonic circRNAs.
Figure 1.
 
CircRNA expression patterns in type II TAO relative to type I TAO. (A) Volcano plot of differentially expressed circRNAs between type II and type I TAO. (B) Hierarchical cluster analysis of differentially expressed circRNAs. (C) Origination of circRNAs. (D) Length of exonic circRNAs.
Figure 2.
 
The mRNA expression patterns in type II TAO relative to type I TAO. (A) Volcano plot of differentially expressed mRNAs between type II and type I TAO. (B) Heatmap of differentially expressed mRNAs.
Figure 2.
 
The mRNA expression patterns in type II TAO relative to type I TAO. (A) Volcano plot of differentially expressed mRNAs between type II and type I TAO. (B) Heatmap of differentially expressed mRNAs.
Figure 3.
 
Validation results for the selected circRNAs and mRNAs. The relative expression of circRNAs and mRNAs were analyzed in another 3 type II TAO, and 3 type I TAO orbital fat tissues.
Figure 3.
 
Validation results for the selected circRNAs and mRNAs. The relative expression of circRNAs and mRNAs were analyzed in another 3 type II TAO, and 3 type I TAO orbital fat tissues.
Figure 4.
 
circRNAs-miRNAs-mRNAs network. The ceRNA network map consists of 7 differentially expressed circRNAs, 23 miRNAs, and 262 mRNAs. Red inverted triangles represent the circRNAs, the blue diamonds represent the miRNAs, and the pink circles represent the mRNAs.
Figure 4.
 
circRNAs-miRNAs-mRNAs network. The ceRNA network map consists of 7 differentially expressed circRNAs, 23 miRNAs, and 262 mRNAs. Red inverted triangles represent the circRNAs, the blue diamonds represent the miRNAs, and the pink circles represent the mRNAs.
Figure 5.
 
Top 10 hub genes identified in the PPI network. (A) PPI network construction for differentially expressed genes. (B) The top 10 hub genes.
Figure 5.
 
Top 10 hub genes identified in the PPI network. (A) PPI network construction for differentially expressed genes. (B) The top 10 hub genes.
Figure 6.
 
Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis for circRNA-target mRNAs. (A) The top 10 GO terms in biological process (BP), cellular component (CC), and molecular function (MF), respectively, for target mRNAs. (B, C) The KEGG pathways for target mRNAs.
Figure 6.
 
Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis for circRNA-target mRNAs. (A) The top 10 GO terms in biological process (BP), cellular component (CC), and molecular function (MF), respectively, for target mRNAs. (B, C) The KEGG pathways for target mRNAs.
Figure 7.
 
The hsa_circ_0007006 was correlated with COL1A1 and MMP2. (A) Lentiviral transfected OFs to overexpress hsa_circ_0007006. (B) The expression relationship between hsa_circ_0007006 and COL1A1, MMP2.
Figure 7.
 
The hsa_circ_0007006 was correlated with COL1A1 and MMP2. (A) Lentiviral transfected OFs to overexpress hsa_circ_0007006. (B) The expression relationship between hsa_circ_0007006 and COL1A1, MMP2.
Figure 8.
 
Western blotting of relaxin signaling pathway expression in type I and type II TAO.
Figure 8.
 
Western blotting of relaxin signaling pathway expression in type I and type II TAO.
Table 1.
 
Basic Clinical Information of the Research Subjects
Table 1.
 
Basic Clinical Information of the Research Subjects
Table 2.
 
Clinical Features of Patients With TAO for High Throughput RNA-Sequencing
Table 2.
 
Clinical Features of Patients With TAO for High Throughput RNA-Sequencing
Table 3.
 
The Validated CircRNAs Ranked by Fold Change
Table 3.
 
The Validated CircRNAs Ranked by Fold Change
Table 4.
 
The Top 30 Gene Ontology (GO) Terms in Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) for Target mRNAs
Table 4.
 
The Top 30 Gene Ontology (GO) Terms in Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) for Target mRNAs
Table 5.
 
Kyoto Encyclopedia of Genes and Genomes (KEGG) Analysis for circRNA-Target mRNAs
Table 5.
 
Kyoto Encyclopedia of Genes and Genomes (KEGG) Analysis for circRNA-Target mRNAs
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×