Transcriptome analysis identified a novel 3-LncRNA regulatory network of transthyretin attenuating glucose induced hRECs dysfunction in diabetic retinopathy
Diabetic retinopathy (DR) is the leading cause of blindness in the working age population. Transthyretin (TTR) showed a significantly decreased concentration in DR patients and exerted a visual protective effect by repressing neovascularization. This work intended to identify long non coding RNAs (lncRNAs) and explore their potential mechanism underlying the protective role of TTR.
Transcriptome of human retinal endothelial cells (hRECs) treated with low glucose (LG), high glucose (HG) or high glucose with 4 μM TTR (HG + TTR) was conducted. Differentially expressed lncRNAs, mRNAs and TTR related lncRNAs and mRNA were acquired. Functional annotation and Gene Set Enrichment Analysis were applied to analyse TTR affected pathways and processes. Weighted gene co-expression network analysis (WGCNA) was implemented to obtain hub modules and genes. LncRNA-mRNA regulatory networks were constructed based on cis, trans and competing endogenous RNAs acting mode. QRT-PCR was conducted to validate the expression of lncRNAs in aqueous humor and serum samples from 30 DR patients and 10 normal controls.
RNA-sequencing of hRECs treated with low glucose (LG), high glucose (HG) or high glucose with 4 μM TTR (HG + TTR) was conducted. 146,783 protein-coding transcripts, 12,403 known lncRNA transcripts and 1184 novel non-coding transcripts were characterized. A total of 11,407 differentially expressed mRNAs (DE-mRNAs), 679 differentially expressed lncRNAs (DE-lncRNAs) in HG group versus LG group, 6206 DE-mRNAs and 194 DE-lncRNAs in HG + TTR versus HG group were obtained, respectively. 853 TTR-mRNAs and 48 TTR-lncRNAs were acquired, and functionally involved in cell cycle, apoptosis, inflammation signalling pathway, response to oxidative stress, neovascularization and autophagy. The WGCNA analysis identified a hub module of 133 genes, with the core function of oxidative stress response, angiogenesis, MAPK pathway, cell proliferation and apoptosis. After qRT-PCR validation, a 3-lncRNA regulatory network was proposed. At last, lncRNAs MSTRG.15047.3 and AC008403.3 showed significantly relative higher expression levels in both aqueous humor and serum samples, compared with normal controls, and FRMD6-AS2 was significantly down-regulated.
TTR regulated mRNAs and biological processes including oxidative stress, inflammation signalling and autophagy. A 3-lncRNA regulatory network was characterized underlying TTR repressing neovascularization, and showed potential diagnostic performance in DR.
KeywordsDiabetic retinopathy Transthyretin Transcriptome 3-lncRNA regulatory network, diagnosis
ABL Proto-Oncogene 1, Non-Receptor Tyrosine Kinase
Antisense non-coding RNA in the INK4 locus
BDNF Antisense RNA
BTG Anti-Proliferation Factor 1
Competing endogenous RNA
- CNC nework
Coding Potential Assessing Tool
Coding Potential Calculator 2
The Database for Annotation, Visualization and Integrated Discovery
Differentially expressed genes
Differentially expressed long non-coding RNAs
Differentially expressed mRNAs
Enhancer of Zeste 2 Polycomb Repressive Complex 2 Subunit
False discovering rate
Fasciculation And Elongation Protein Zeta 1
Fibroblast Growth Factor 13
FRMD6 Antisense RNA 2
Glucose-regulated protein 78
Gene Set Enrichment Analysis
Human brain microvascular endothelial cell
Human coronary artery endothelial cells
- HG + TTR
High glucose with 4 μM TTR
Heme Oxygenase 1
HOXA Distal Transcript Antisense RNA
Human retinal endothelial cells
Human retinal microvascular endothelial cells
Human retinal pigment epithelial cells
Integrin Subunit Alpha 5
Janus Kinase 1
Kyoto Encyclopedia of Genes and Genomes
Long non-coding RNAs
Metastasis Associated Lung Adenocarcinoma Transcript 1
Minichromosome Maintenance Complex Component 7
Molecular COmplex Detection
Maternally Expressed 3
Myocardial Infarction Associated Transcript
miRNA response element
N (Alpha)-Acetyltransferase 60, NatF Catalytic Subunit
Predictor of long non-coding RNAs and messenger RNAs based on an improved k-mer scheme
Protein Phosphatase 2 Phosphatase Activator
Polycomb Repressive Complex 2
Solute Carrier Family 5 Member 5
SOX2 Overlapping Transcript
Short Tandem Repeat
The TATA-Box Binding Protein Associated Factor 6
Transcription factor IID
Tumor Protein P53
Tripartite Motif Containing 5
TTR related DEGs
Ubiquitin Conjugating Enzyme E2 V1
Vascular Endothelial Growth Factor
Vascular Endothelial Growth Factor Receptor 1
Weighted gene co-expression network analysis
As the leading cause of vision impairment and loss in the working age population, diabetic retinopathy (DR) is a common and specific microvascular complication characterized by early retinal microvascular dysfunction . In China, the number of DR patients is growing each year . The early stage of DR is exemplified by retina endothelial cells and pericytes death . As DR progresses, vascular leakage becomes evident and may eventually lead to diabetic macular edema, the most common cause of vision loss in DR . Till now, panretinal photocoagulation, vitrectomy and anti-neovascularization drug intraocular injection remain the few therapeutic options . Accumulating evidence implicated that inflammation, cell apoptosis, oxidative stress and autophagy were major causative factors involved in the pathogenesis of DR . Therefore, uncovering potential mechanisms underlying DR pathogenesis, especially revealing hub genes and pathways in inflammation, cell apoptosis, oxidative stress and autophagy is of paramount importance for the development of effective therapeutics for patients with DR.
Transthyretin (TTR) is a 55-kDa homotetramer protein reported to transport thyroxine and retinol through retinol-binding protein (RBP) , and it is mainly secreted by human retinal pigment epithelial cells (hRPECs) and choroid in ocular tissue. In vitreous and serum samples of high myopia patients, TTR showed significantly higher expression compared with healthy donors . Interestingly, the expression of TTR in DR patients was significantly lower, compared with healthy donors [8, 9]. In our previous research, an exogenous 4 μM TTR treatment or endogenous TTR overexpression both repressed blood vessel tube forming capability of hyperglycemic cultured human retinal endothelial cells (hRECs), by attenuating pro-angiogenic genes expression, like Angiopoietin 2 (Angpt2), Vascular Endothelial Growth Factor Receptor 1 (VEGFR1), and Vascular Endothelial Growth Factor (VEGF2) [10, 11]. Additionally, TTR was reported to repress neovascularization by promoting hRECs apoptosis through direct binding to glucose-regulated protein 78 (GRP78) . Surprisingly, the exogenous added TTR efficiently crossed cell membrane and nuclear membrane, and located in both in cytoplasm and nucleus, implicating complex transcriptional and post-transcriptional regulatory role, while the exact mechanism still remains obscure.
Long noncoding RNA (lncRNA) is characterized to be transcripts with a length of more than 200 nt and of no protein-coding capability. Accumulating evidence pointed that lncRNA exerted function in cis by regulating nearby genes, or in trans by modulating distantly located genes . In addition, it’s a putative mechanism that lncRNAs may regulate target gene expression via miRNA response element (MRE), known as competing endogenous RNA (ceRNA) . The aberrant expression of lncRNAs in early diabetic retinopathy was first characterized in 2004 . Subsequent studies of individual lncRNAs playing a transcriptional or post-transcriptional regulatory role in DR pathogenesis and progression had been reported, including Maternally Expressed 3 (MEG3) , Myocardial Infarction Associated Transcript (MIAT) [17, 18], HOXA Distal Transcript Antisense RNA (HOTTIP) , Metastasis Associated Lung Adenocarcinoma Transcript 1(MALAT1) [20, 21], SOX2 Overlapping Transcript (Sox2-OT) , antisense non-coding RNA in the INK4 locus (ANRIL)  and BDNF Antisense RNA (BDNF-AS) . However, there have been limited studies of lncRNAs in TTR repressing DR neovascularization.
In this study, we proposed that there may exist certain lncRNAs, through which TTR exerts its function of repressing neovascularization in DR. RNA transcriptome of hRECs treated with low glucose concentration of 5.5 mM (LG), high glucose concentration of 25 mM (HG), and a 25 mM high glucose plus a concentration of 4 μM TTR (HG + TTR) was determined for the first time. The impact of high glucose and TTR on hRECs neovascularization was evaluated with tube formation assay. Then differentially expressed lncRNAs and mRNAs that induced by high glucose and TTR were identified, respectively. Functional annotation and network analysis of TTR regulated mRNAs significantly enriched pathways and biological processes in DR. Furthermore, the TTR regulated lncRNAs and mRNAs were predicted, and the lncRNA-mRNA regulatory network in cis, trans and ceRNA was constructed. Finally, after qRT-PCR validation in hRECs, vitreous humor and serum samples, a 3-lncRNA centered TTR response coding-non-coding network with potential diagnosis performance was proposed to explain the protective role of TTR, which would provide basis for the future potential application of TTR and lncRNAs in DR diagnosis and treatment.
Cell culture and TTR treatment
HRECs from donated human retinal tissue were extracted, sub-cultured, and authenticated by Genetic Testing Biotechnology (Shanghai, China) using Short Tandem Repeat (STR) analysis as described in 2012 in ANSI Standard (ASN-0002) by the ATCC Standards Development Organization (SDO). Nineteen short tandem repeat (STR) loci plus the gender determining locus, Amelogenin, were amplified using the commercially available EX20 Kit from AGCU. The cell line sample was processed using the ABI Prism 3130 XL Genetic Analyzer. Data were analyzed using GeneMapper ID v3.2 software (Applied Biosystems). Appropriate positive and negative controls were run and confirmed for each sample submitted. And as detected, eight core STR loci plus Amelogenin revealed that HRECs should be quite similar with HBMEC-2 (human brain microvascular endothelial cell) (Additional files 1, 2 and 3).
HRECs from passage 4 were diluted to 8 × 104/mL and incubated in 6-well plates with low glucose (LG) media until confluent. After washing with PBS, the cells were further cultured for 48 h at 37 °C in 5.5 mM glucose (low group, LG), 25 mM glucose (high glucose, HG), or 25 mM glucose with 4 μM TTR (HG + TTR) media. The concentration of low glucose of 5.5 mM and simulated high glucose concentration of 25 mM were referred to previous studies [25, 26]. The concentration of TTR of 4 μM were referred to Raquel J. Nunes et al’s study; the TTR concentration of 1, 2, 4, 8 and 10 μM were examined and 4 μM was determined as the final concentration . Cells were collected and total RNAs were extracted from hRECs, using TRIzol reagent (Takara, Dalian, China), according to the manufacturer’s instructions.
Tube formation assay
The basement membrane matrix (BD Biosciences) was placed into the well of 24-well plate, and hardened at 37 °C for 30 min; 2 × 104 hRECs cells were seeded on each well, and incubated with glucose and/ or TTR at 37 °C for 24 h. Tube formation was observed using an Olympus IX-73 microscope.
Aqueous humor and serum samples of 30 DR patients and 10 normal controls (patients without diabetes) were obtained from ophthalmology inpatients of Nanjing Medical University Affiliated Wuxi People’s Hospital. Aqueous humor of normal controls (patients without diabetes) and DR patients were collected during cataract surgery. Clinic patients’ samples processing were the same as we previously reported . Informed consents were signed for enrolled patients and healthy individuals. This study was compiled with principles of the Declaration of Helsinki and was approved by the Ethics Committee of Nanjing Medical University (2014–62). All clinical samples were obtained from ophthalmology inpatients of Nanjing Medical University Affiliated Wuxi people’s Hospital. Aqueous humor sample of DR patients and normal control were obtained from cataract surgery. All samples were stored at − 80 °C.
Transcriptome sequencing and differentially expressed genes (DEGs) identification
After total RNAs were extracted and collected, cDNA was produced and purified with AMPure XP (Beckman Coulter, Brea, CA, USA) beads. Then PCR was conducted to generate cDNA libraries, and adaptor sequences were removed, followed by low-quality reads filtering. Trinity (http://trinityrnaseq.sourceforge.net/) software and De Bruijn graph algorithm  were applied for de novo assembly. At last, unigenes were annotated with BLASTX and BLAST algorithm, by searching GenBank database (http://www.ncbi.nlm.nih.gov/) and NCBI non-redundant (NR) (ftp://ftp.ncbi.nih.gov/blast/db/), respectively. The RNA-seq data had been submitted to GEO repository, with accession number GSE117238.
In this study, we used FPKM (Fragments Per Kilobase of transcript per Million) for transcripts, instead of gene expression, and gene names corresponding to certain transcripts were also shown. For a certain gene, if more than one transcript showed significant differential expression, the gene were considered to be significantly deregulated. Specifically, Student’s t-test and fold change (FC) filtering were conducted, and to screen differentially expressed genes (DEGs) between two groups by using R software limma package. In addition to P value, a Benjamini–Hochberg procedure was conducted for multiple comparisons to control the false discovering rate (FDR) . Significantly deregulated transcripts were defined using the criteria of Fold Change≥2, P ≤ 0.05 and FDR ≤0.05. Volcano plot were construed by R software ggplot2 package, to show significantly deregulated transcripts and genes. The differential expression analysis was conducted in the transcript level, and transcripts with the converse expression fold-change trend in HG + TTR group versus HG group and HG group versus LG group were named as TTR related DEGs (TTR-DEGs).
Pathway and GO annotation and GSEA analysis
For differentially expressed genes functional annotation, KEGG pathway and Gene Ontology (GO) annotation and enrichment were conducted, based on Kyoto Encyclopedia of Genes and Genomes (KEGG) (http://www.genome.jp/kegg/) and Gene Ontology database (http://www.geneontology.org/), with DAVID (The Database for Annotation, Visualization and Integrated Discovery v6.8, https://david.ncifcrf.gov/). For GO analysis, biological processes (BP), cellular components (CC) and molecular function (MF) were applied and enriched items were shown, respectively. A Fisher exact test was used to select only significant categories. A Benjamini–Hochberg procedure was conducted for multiple comparisons to control the false discovering rate (FDR) . GO terms with corrected Q value < 0.05 were considered significant. Significantly enriched pathways and biological processes were displayed with R software ggplot2 package.
Compared with functional annotation of DEGs with KEGG pathway and GO, function of all detected genes were analyzed with Gene Set Enrichment Analysis (GSEA) . Specifically, annotation file of c2: curated gene sets-all canonical pathways, gene symbols were used to analyze significantly enriched gene sets. The GSEA analysis was applied to extract significant biological processes and pathways, for HG and TTR + HG groups. Gene expression data matrix of detected genes in LG, HG and TTR + HG groups were imported into GSEA software (v2.2.4) and the permutation index were set as 1000. Then the GS (gene set) score and P value were used, and biological processes or pathways with P < 0.05 were considered to be significant.
Protein-protein interaction (PPI) network analysis and sub-network construction
To explore the interaction of TTR-mRNAs, TTR-mRNAs lists were first imported into STRING database (https://string-db.org), and generated the protein-protein interaction (PPI) network. When constructing the PPI network, protein-protein interaction pairs with medium confidence of interaction score > 0.4 were exported and then visualized by using Cytoscape (version 3.4.0). To analyze the topological structure and relationship in PPI network, CentiScaPe was further applied for calculating centrality parameters for each node and finding the most important nodes in a network . At last, Molecular COmplex Detection (MCODE) (http://apps.cytoscape.org/apps/mcode) was applied to find densely connected regions and sub-networks within the main network.
Weighted gene co-expression network analysis (WGCNA) and module detection
The Weighted gene co-expression network analysis (WGCNA) is a well-known technique used to identify biologically meaningful and functional co-expression modules related to diseases or conditions . In this study, to explore and screen gene expression modules correlated with high glucose and TTR, the gene co-expression network was constructed by the R WGCNA package. The gene expression similarity and soft-threshold were calculated using the integrated function (pick Soft Threshold) in the package. When the soft threshold was defined as 10, a module with ME green color of 133 genes showed highest correlation with both glucose and TTR and was functionally annotated in subsequent steps.
Module genes biological pathway and function network in ClueGo
The function of 133 ME green module genes were annotated and significantly enriched (P < 0.05) pathways were labeled in Cytoscape ClueGo and CluePedia. The ClueGo and CluePedia application were developed to create and visualize a functionally grouped network of terms/pathways .
LncRNA cis and trans regulatory network construction
For lncRNA targets identification, cis and trans regulatory network was constructed. To identify cis regulatory relationship, lncRNAs that act on neighboring target genes were investigated. We searched for coding genes 10 kb/100 kb upstream and downstream of each lncRNA. To identify trans regulatory relationship, lncRNAs and target genes were identified based on their expression levels. We calculated the expression level correlation between lncRNAs and coding genes using custom scripts., and lncRNA-mRNA pairs with correlation coefficient > 0.95 or < − 0.95 were extracted for subsequent analysis. At last, the cis or trans regulatory network was shown with Cytoscape.
Competing endogenous RNAs (ceRNAs) network analysis
To reveal the roles and interactions of lncRNAs with mRNAs in glucose and TTR treated hRECs, we constructed lncRNA-miRNA-mRNA regulatory ceRNA networks. Potential miRNA response elements were searched for the sequences of lncRNAs and mRNAs, and we identified overlaps in predicted miRNA seed sequence binding sites as well as lncRNAs binding sites in the target mRNA as part of the lncRNA-miRNA-mRNA interaction. The miRNA binding sites were predicted by miRanda (http://www.microrna.org/), while the miRNA-mRNA interactions were predicted by Targetscan (http://www.targetscan.org/). The interaction network was built and displayed using Cytoscape software based on the screening of lncRNA-miRNA-mRNA pairs.
Primers of genes
SPSS 13.0 for windows (Chicago, IL, USA) was used for data analysis, and statistical significance was determined using a T test. The P and T values were calculated. P < 0.05 was considered statistically significant.
Exogenous TTR treatment repressed high glucose induced hRECs tube formation
Transcriptome sequencing characteristics of genes in glucose and TTR treated hRECs
DEGs and TTR-DEGs identification
KEGG pathway, gene ontology annotation and GSEA analysis
The TTR protein-protein interaction network and sub-networks function
WGCNA analysis identified hub modules both regulated by glucose and TTR
LncRNA-mRNA regulatory network based on cis, trans and ceRNA interaction relationship
The 5 functional lncRNA-mRNA trans networks
A 3-lncRNA centered TTR regulatory network with diagnostic performance
Recently, it was reported that dysfunction of TTR was bound up with system diseases and the intervention of TTR improved multiple clinical manifestations of hereditary transthyretin amyloidosis [49, 50]. TTR was also involved in various biological processes such as proteolysis, nerve regeneration, autophagy, glucose homeostasis and angiogenesis regulation . In ophthalmic diseases, abnormal TTR expression was reported in high myopia and diabetic retinopathy patients. Exogenous TTR treatment inhibited hRECs proliferation, migration and tube forming capability and induced apoptosis in vitro [12, 13]. However, the underlying mechanism still remains unknown. In this study, through RNA-sequencing, the basic transcriptome characteristics of hRECs underlying hyperglycemia and TTR with high glucose conditions were first determined. Then, through functional annotation, protein-protein interaction and co-expression analysis, the protective role of TTR in hRECs dysfunction including signaling pathways, biological processes and trait modules were further explored. Furthermore, potential targets of TTR related lncRNAs were predicted and 3 coding-non-coding networks were constructed based on putative lncRNA acting mode. Finally, a 3-lncRNA centered regulatory hub network with clinical significance was proposed to explain the function of TTR attenuating neovascularization.
The functional annotation of TTR related mRNAs corroborated previous reports of relative decreased expression of TTR in DR patients and neovascularization function. In our study of TTR function in hRECs, biological processes involved in cell proliferation, migration and neovascularization were significantly enriched, in lining with studies that TTR modulated human umbilical vein endothelial cells (HUVECs) apoptosis and inhibit migration, regulation of hRECs apoptosis, proliferation and migration. In addition, biological processes and pathways like response to oxidative stress, inflammatory signaling like IL2-STAT5, and regulation of autophagy were also enriched and reported for the first time. The intricate cross-talking between oxidative stress, inflammatory signaling pathway, autophagy and apoptosis were reported in melatonin ameliorating DR progression . Thus, we speculated that there may exist a similar regulatory network, composed of oxidative stress, inflammation signaling, autophagy and apoptosis, linking the protective function of TTR and DR. The 5 protein-protein interaction sub-networks in Fig. 4 implied that TTR might orchestrate pathways in transcriptional and post-transcriptional levels. Genes in sub-network 1 implied that TTR might regulate cell cycle via ubiquitin mediated proteolysis, and the function of sub-network 3 suggested that TTR modulates MAPK and NF-κB pathway through RNA transportation, splicing and translation, post-transcriptionally. In transcriptional level, the function of genes in the sub-network 5 was focused on DNA repair, an important link of oxidative stress, inflammatory signaling, autophagy and apoptosis [53, 54, 55]. Hence, the functional annotation provides a comprehensive direction for further research of how TTR rescuing high glucose induced neovascularization abnormality.
Another important part of our work was the investigation of TTR related lncRNAs and their potential regulatory role in DR. As non-coding RNAs, the functional mode of lncRNAs via regulating mRNA expression in cis, trans or as miRNA sponge has been well recognized in angiogenesis and vascular diseases . LncRNA MEG3 knockdown aggravated retinal vessel dysfunction by the activation of PI3k/Akt signaling . Through regulating p38-MAPK pathway, HOTTIP improved diabetic retinal microangiopathy . As an important inflammation regulator, lncRNA MALAT1 was reported to affect retinal endothelial cell proliferation, migration, and tube formation through cross-talking with p38 MAPK signaling pathway [20, 21]. As miR-29b and miR-150-5p sponge or VEGF ceRNA, lncRNA MIAT (Myocardial infarction associated transcript) was reported to regulate endothelial cell apoptosis . Another role that well studied in cardiovascular diseases is lncRNA ANRIL (antisense non-coding RNA in the INK4 locus), which regulated VEGF expression and function in diabetic retinopathy via miR-200b, and EZH2 (Enhancer of Zeste 2 Polycomb Repressive Complex 2 Subunit) of the PRC2 (Polycomb Repressive Complex 2) complex . However, the role of lncRNAs in TTR ameliorating hRECs dysfunction has not yet been reported. In this paper, a 3 lncRNA-mRNAs regulatory network and 5 lncRNAs-mRNAs sub-networks in cis, trans and ceRNA were constructed for the first time. The networks would contribute to understanding the underlying mechanism of TTR functions in DR from the perspective of different lncRNAs acting mode.
In addition, expression of 10 lncRNAs in lncRNA-mRNAs networks were validated by qRT-PCR and 3 lncRNAs including 2 up-regulated novel lncRNAs, i.e., MSTRG.15047.3 and AC008403.3, 1 down-regulated lncRNA, i.e., FRMD6-AS2 (FRMD6 Antisense RNA 2). The 3 lncRNAs formed a complicated potential interacting network, including 11 mRNAs and 2 miRNAs. Based on the network, the novel predicted lncRNA MSTRG.15047.3 might interact with nearby up-located CBFTA2T2 in a cis-regulatory way, trans-regulates cell cycle and neovascularization via UBE2N (Ubiquitin Conjugating Enzyme E2 N), PPP2R5C (Protein Phosphatase 2 Regulatory Subunit B′ Gamma), MCM7(Minichromosome Maintenance Complex Component 7), TAF6 (The TATA-Box Binding Protein Associated Factor 6) and PBRM1 (Polybromo 1), or just sponge miR-4651 and promote FGF13 (Fibroblast Growth Factor 13) expression. Gene UBE2N encoded a member of the E2 ubiquitin-conjugating enzyme family and was reported to be involved in cell cycle and error-free DNA repair pathway . The UBE2V1 (Ubiquitin Conjugating Enzyme E2 V1)-UBE2N heterodimer acts in concert with TRIM5 (Tripartite Motif Containing 5) to activate the MAP 3 K7/TAK1 complex and induce expression of NFκB and MAPK-responsive inflammatory genes [58, 59]. PPP2R5C gene belongs to the phosphatase 2A regulatory subunit B family. The PP2A (Protein Phosphatase 2 Phosphatase Activator) -PPP2R5C holoenzyme may specifically dephosphorylate and activate TP53 (Tumor Protein P53) and play a role in DNA damage-induced inhibition of cell proliferation, as well as regulate the ERK signaling pathway through ERK dephosphorization . MCM7 acts as component of the MCM2–7 complex (MCM complex) and may be involved in the formation of replication forks and in the recruitment of other DNA replication related proteins . The TATA-Box Binding Protein Associated Factor 6 (TAF6) are components of the transcription factor IID (TFIID) complex and modulates cell apoptosis in a TP53 dependent  and independent  way. PBRM1 encodes a subunit of ATP-dependent chromatin-remodeling complexes. In clear cell renal cancer, concomitant loss of PBRM1 rescues VHL-induced replication stress, maintaining cellular fitness and allowing proliferation . Fibroblast Growth Factor 13 possesses broad mitogenic and cell survival activities, including embryonic development, cell growth, morphogenesis and invasion . For AC008403.3, in addition to sharing 4 trans targets of TAF6, PPP2R5C, PBRM1 and FGF13 with MSTRG.15047.3 may act as a ceRNA of NAA60 (N (Alpha)-Acetyltransferase 60, NatF Catalytic Subunit) and SLC5A5 (Solute Carrier Family 5 Member 5) via sponging miR-6884-5p. NAA60 encodes an enzyme that localizes to the Golgi apparatus and transfers an acetyl group to the N-terminus of free proteins . SLC5A5 encodes a member of the sodium glucose cotransporter involved in thyroxine metabolism, a shared molecular function of TTR . As the antisense RNA and potential cis-regulator of FRMD6 (FERM Domain Containing 6), a known hippo pathway kinase regulator , lncRNA FRMD6-AS2 might interact with PBRM1, PPP2R5C and ASB (Arylsulfatase B) to regulate cell proliferation and neovascularization. Further studies will be conducted to explore the exact role of 3-lncRNA regulatory network linking TTR and DR neovascularization.
Finally, lncRNAs MSTRG.15047.3, AC008403.3 and FRMD6-AS2 showed significant differential expression levels in aqueous humor and serum samples of DR patients and normal controls (patients without diabetes), made possible 3 lncRNAs as diagnostic biomarkers in DR. Several studies investigated the clinical significance of lncRNAs in DR, like MIAT  and MALAT1 . In this study, we proposed 3 of 10 lncRNAs potentially involved in DR progression, and we would explore their function and molecular mechanism in the future. Additionally, the correlation of lncRNAs expression with clinical parameters, and especially performance as DR biomarkers would be studied, with more samples, including healthy donors and other ocular diseases.
In summary, it was concluded that TTR repressed hRECs neovascularization, via a 3-lncRNA regulatory network. This investigation was mainly based on glucose and exogenous TTR treatment of hRECs, RNA sequencing, bioinformatics analysis and qRT-PCR validation in clinical samples. There are several limitations in this study. First, hRECs cell line was used as the DR cell model, and the tube forming assay demonstrated that TTR repressed neovascularization. Whether the neovascularization repressing effect of TTR and 3-lncRNA regulatory network were restricted to hRECs, or represents a general mechanism in microvascular endothelial cells is an important question. In fact, Kun Shan .et al. employed other microvascular endothelial cell lines, like human umbilical vein endothelial cells (HUVECs), EA.hy.926 and human coronary artery endothelial cells (HCAECs) to investigate retinal vascular dysfunction . In addition, using primary human retinal microvascular endothelial cell (HRMECs) would help to illustrate the point. Secondly, this study was conducted in vitro, and investigation of TTR effect in DR animal models, like streptozotocin (STZ) administration, pancreatectomy and genetic models  would provide more solid evidence to support the conclusion. Thirdly, an exogenous TTR was employed to examine the function and a vector or virus mediated TTR overexpression approach would provide more details about the endogenous function of TTR. Lastly, we speculated that the 3-lncRNA regulatory network play a pivotal role in TTR function, and more experiments, like rescue assays of 3 lncRNAs knockdown or overexpression in TTR treatment cells, would help to support the point.
In summary, we investigated (1) the basic characteristics and landscape of RNA transcriptome in TTR and glucose treated hRECs; (2) the function of TTR-mRNAs from signaling pathway, biological process, subnetworks and hub modules; (3) lncRNA-targets networks and functional sub-networks in cis, trans and ceRNA regulatory way; (4) a 3-lncRNA centered hub network that might mediate the comprehensive function of TTR and serve as potential diagnostic biomarkers in DR. The research would pave the way for understanding the complex biology of TTR in DR progression.
We thank Genetic Testing Biotechnology (Shanghai, China) for the help of Short Tandem Repeat (STR) analysis of hRECs.
JS carried out the design of the study and drafted the manuscript. YZ carried out RNA-sequencing and bioinformatics analysis. GF carried participated in cell culture and qRT-PCR examination. YX carried performed the statistical analysis and participated in the design of the study. YY conceived of the study, and participated in its design and coordination. All authors read and approved the final manuscript.
This work was supported by National Natural Science Foundation of China (81970819), Jiangsu Young Medical Talents (QNRC2016182) and Natural Science Foundation (BK20160053). The funding body did not play any role in the study design and collection, analysis and interpretation of the data and the write-up of the manuscript.
Ethics approval and consent to participate
Informed consents were signed for all enrolled patients and healthy individuals. This study followed the tenets of the Declaration of Helsinki and the ARVO statement for research involving human subjects and was approved by the Ethics Committee of Nanjing Medical University (2014–62).
Consent for publication
The authors declare that they have no competing interests.
- 17.Zhang J, Chen M, Chen J, Lin S, Cai D, Chen C, et al. Long non-coding RNA MIAT acts as a biomarker in diabetic retinopathy by absorbing miR-29b and regulating cell apoptosis. Biosci Rep. 2017;37(2):1–10.Google Scholar
- 29.Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. JR Statist Soc B. 1995;57:289–300.Google Scholar
- 41.Singh N, Kansal P, Ahmad Z, Baid N, Kushwaha H, Khatri N, et al. Antimycobacterial effect of IFNG (interferon gamma)-induced autophagy depends on HMOX1 (heme oxygenase 1)-mediated increase in intracellular calcium levels and modulation of PPP3/calcineurin-TFEB (transcription factor EB) axis. Autophagy. 2018;14(6):972–91.PubMedPubMedCentralGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.