Background
Diabetic retinopathy (DR) is the most common complication of diabetes mellitus and the leading cause of blindness in middle-aged and elderly individuals [
1,
2]. It is classified as non-proliferative diabetic retinopathy (PDR) and PDR [
3]. Retinal neovascularization is a key feature of PDR [
4]. The main therapeutic strategy is the use of anti-vascular endothelial growth factor (VEGF) [
5]. However, resistance to neovascularization remains challenging owing to its ineffectiveness.
Long noncoding RNAs (lncRNAs) are non-protein-coding transcripts larger than 200 nucleotides. They can participate in gene regulation at transcriptional, post-transcriptional, and translational, such as regulatory transcription factors and endogenous competitive RNA [
6]. LncRNAs have been implicated in a wide range of physiological processes and in the pathophysiology of several diseases, and DR is no exception. They are increasingly being recognized as important players in the development of DR.
Aberrant Expression of lncRNAs in PDR may be relevant to the molecular etiology of DR. Some researchers performed microarray-based gene expression analysis designed to serve as a resource for elucidating lncRNA-mediated DR pathogenesis [
7,
8]. The identification of dysregulated lncRNAs is a key step in understanding the significance of lncRNAs in DR. The vitreous humor of patients with PDR can represent a reservoir of pathological signaling molecules because of tissue accessibility [
9]. The lncRNAs that are differentially expressed in the vitreous humor of patients with PDR remain inadequately explored [
9]. Therefore, this is a novel research field for PDR.
In addition, it is necessary to identify lncRNA expression profiles in the vitreous fluid of patients with PDR after anti-VEGF therapy. Adjuvant intravitreal injection of anti-VEGF drugs before vitrectomy, which reduces the difficulty of surgery and recurrence of vitreous hemorrhage, is beneficial [
10]. Nevertheless, in some cases, the patient’s eyes fail to respond adequately. Thus, a complete understanding of the relationship between lncRNAs and anti-VEGF drugs may enable add another dimension to its therapeutic directions [
11].
In this study, we aimed to detect differences in the expression of lncRNAs and messenger RNAs (mRNAs) between patients with PDR and those with idiopathic macular hole (IMH) and between patients with PDR treated with anti-VEGF therapy and untreated patients with PDR. To reveal the functional significance of lncRNAs in DR, we performed bioinformatic analysis.
Methods
Patient recruitment
This clinical study adhered to the provisions of the Declaration of Helsinki for research involving human subjects. This study was approved by the Ethical Review Committee of the First People’s Hospital of Zunyi (Zunyi, China; project number 2019–019). All patients who underwent pars plana vitrectomy surgery for IMH or PDR at the First People’s Hospital of Zunyi between October 2019 and September 2020 were enrolled consecutively.
The patients were divided into three groups: group A consisted of patients with IMH without diabetes, group B consisted of patients with PDR pretreated with conbercept 3–7 days before surgery, and group C consisted of patients with PDR who underwent surgery alone. All subjects underwent a complete ophthalmologic examination, including medical history, best-corrected visual acuity measurement, intraocular pressure measurement, slit-lamp examination, fundus examination, ocular ultrasonography, optical coherence tomography, and fundus fluorescein angiography (as necessary). According to the international classification, PDR is defined as neovascularization and/or vitreous/preretinal hemorrhage [
3].
Subjects with other systemic diseases such as renal failure or malignant tumors were excluded. Subjects with other eye diseases, including glaucoma, uveitis, age-related macular degeneration, retinal artery occlusion, retinal vein occlusion, rhegmatogenous retinal detachment, endophthalmitis, or ocular trauma, were also excluded. Patients who had undergone one or more of the following: eye surgery, retinal laser photocoagulation, intravitreal steroids, or anti-VEGF therapy (only 3–7 days before surgery were allowed) were excluded from the analysis. Also, subjects were excluded if they showed any evidence of systemic or local inflammation within 6 months.
Finally, three patient samples from each group were analyzed using microarray technology. The remaining 39 patients (6 in group A, 8 in group B, and 25 in group C) were included in the confirmation cohort. There were no statistically significant differences in the age or sex ratios (Table
1).
Table 1
Clinical characteristics of the enrolled patients
Screening Cohort |
Group A (n = 3) | 55.67 ± 12.34 | 2 (66.67) | 4.90 ± 0.40 |
Group B (n = 3) | 57.67 ± 3.06 | 2 (66.67) | 10.43 ± 2.49 |
Group C (n = 3) | 58.67 ± 8.02 | 2 (66.67) | 9.17 ± 2.20 |
P | 0.91 | 1.00 | 0.02 |
PDR Confirmation Cohort |
Group C (n = 11) | 56.55 ± 9.86 | 7 (63.64) | 8.26 ± 1.30 |
Croup A (n = 6) | 57.33 ± 10.50 | 5 (83.33) | 4.90 ± 0.45 |
P | 0.88 | 0.60 | < 0.01 |
Anti-VEGF Confirmation Cohort |
Group B (n = 8) | 49.00 ± 19.54 | 5 (62.50) | 10.26 ± 4.04 |
Croup C (n = 14) | 58.07 ± 8.72 | 7 (50.00) | 8.69 ± 2.34 |
P | 0.15 | 0.68 | 0.34 |
Sample processing
A blood sample (approximately 6 mL) was drawn from the forearm vein into a tube containing an anticoagulant following overnight fasting on the day of surgery. The mixture was immediately centrifuged at 4000×g for 10 min at 4 °C. A vitreous sample (approximately 1 mL) was carefully collected into a 2 mL sterile syringe using a 25-gauge vitreous cutter and manual suction before opening the intraocular irrigation system. If vitreous hemorrhage was present, the surgeon avoided collecting blood components as much blood as possible. All samples were stored in cryopreservation tubes and immediately cooled at − 80 °C until analysis.
After sample collection, total RNA was extracted using TRIzol LS reagent (Invitrogen, Carlsbad, CA, USA) combined with miRNeasy Micro Kit (Qiagen, Hilden, Germany). RNA quality and integrity were measured using Nanodrop (Thermo Fisher Scientific, Waltham, MA, USA) and Agilent 4200 TapeStation.
Microarray analysis
The noncoding RNA and coding RNA transcriptome analysis of the vitreous humor were detected using Clariom D Pico Assay (Affymetrix, Bedford, MA, USA), which has 13,574 transcripts. After assessing RNA quality and quantity, chip analysis was performed by Gminix Biotechnology Company (Shanghai, China). Briefly, sample labeling, hybridization, and washing were performed according to the manufacturer’s protocol. The microarrays were scanned using a GeneChip Scanner 3000 7G (Affymetrix). Raw intensity CEL files generated by GeneChip™ Command Console™ were imported into Transcriptome Analysis Console 4.0.2 (TAC 4.0.2). The data were analyzed with the Robust Multi-chip Analysis algorithm using Affymetrix default analysis setting and global scaling as the normalization method. Quality control graphs were used to assess the quality of sample files. The Limma Bioconductor package (implemented in TAC 4.0.2) was used to analyze expression data based on linear models.
The final difference result was obtained according to the filter condition | fold change | ≥ 1.5 and
P values < 0.05. Hierarchical clustering was performed to show distinguishable noncoding RNA and coding RNA transcript expression patterns among the samples. All the original data were uploaded to the Gene Expression Omnibus public database (
https://www.ncbi.nlm.nih.gov/geo; accession number GSE191210).
Gene ontology (GO) enrichment analysis and Kyoto encyclopedia of genes and genomes (KEGG) pathway enrichment analysis
The GO database describes our knowledge of the biological domain regarding molecular functions, cellular components, and biological processes. According to the relationship of pathways in the KEGG database, the interaction network of a significant pathway was constructed to find the core pathway that plays a key role [
12‐
14]. Fisher’s exact test was used to select significant GO categories and KEGG pathways, and the significance threshold was defined as
P values < 0.05.
Co-expression network
Co-expression networks were constructed to determine the potential roles of noncoding RNA transcripts. To find the co-expression relationship between the differences, Pearson correlation was calculated to identify significantly correlated pairs. It was constructed using Cytoscape, according to the expression value distribution of differential RNA transcripts in different groups. The Pearson correlation value cut-off was = 0.95, with P values < 0.05.
Analysis of neighbor genes of the noncoding RNA transcripts
We searched for noncoding RNA transcripts and their associated coding gene pairs, including the same strand gene with overlap, upstream gene with 10,000 bp, downstream gene with 10,000 bp, and complementary strand gene with overlap.
Quantitative real-time polymerase chain reaction (qRT-PCR)
To further verify the gene chip results, the threshold for the differential expression of lncRNAs was set to a
P value < 0.01. Differentially expressed lncRNAs were sorted according to the absolute value of the fold change. Several candidate lncRNAs with multi-variable shear or difficult primer designs were excluded. Five target genes were selected for subsequent qRT-PCR analysis. Total RNA was reverse-transcribed using a PrimeScript RT reagent kit (TaKaRa, Dalian, China), and qRT-PCR was performed using the CFX96 real-time PCR detection system (Bio-Rad, Hercules, CA, USA). Transcript levels were determined using the PCR Master Mix (Solarbio, Beijing, China). The primer pairs used are listed in Table S
1. The specificity of the qRT-PCR products was estimated using a dissociation curve. qRT-PCR was performed in duplicate for each sample. The relative gene expression was calculated using the 2
-ΔΔCt method. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as the internal control.
Statistical analysis
All statistical analyses were performed using the SPSS software (version 18.0; SPSS Inc., Chicago, IL, USA). Using the Shapiro–Wilk normality test, the numeric variables were first tested for the normality of distributions. Comparisons between two groups were made using Student’s t-test or Mann–Whitney U test. Also, comparisons among three groups were analyzed using a one-way analysis of variance or the Kruskal–Wallis H rank sum test. Categorical variables were determined using Fisher’s exact test. Statistical significance was defined as a P value less than 0.05.
Discussion
The multifactorial pathogenesis of DR is not completely understood. As a new class of modulatory molecules, the essential roles of lncRNAs in the etiology of a broad spectrum of diseases have attracted considerable attention [
6,
15,
16]. PDR is usually considered to be a neovascular disorder, and several lncRNAs have been implicated in neovascularization. ANRIL regulates VEGF expression and function in DR mediated by the PRC2 complex, thereby promoting new vessel formation [
17]. MALAT1 promotes high glucose-induced human retinal endothelial cells (HRECs) by upregulating endoplasmic reticulum stress [
18] or via suppressing the VE-cadherin/β-catenin complex by targeting miR-125b [
19]. MIR497HG is downregulated after high-glucose treatment, and it suppresses the neovascularization of HRECs by targeting the miRNA-128-3p/SIRT axis [
20]. SNHG7 inhibits high-glucose-induced angiogenesis by regulating the miR-543-mediated SIRT1/VEGF pathway [
21]. TDRG1 promotes neovascularization by upregulating VEGF during DR [
22]. Thus, dysregulated lncRNA expression is relevant to the molecular etiology of PDR.
Some researchers have focused on the expression of lncRNAs in DR using high-throughput screening technologies. Yan et al. first reported that approximately 303 lncRNAs are differentially expressed in the retinas of diabetic rats [
7]. Likewise, to demonstrate the relationship between lncRNAs and anti-VEGF drugs, Wang et al. reported that 427 lncRNAs were differentially expressed after anti-VEGF treatment [
23]. However, existing data are insufficient for the vitreous in PDR [
9]. The transcriptional landscape of the vitreous, a reservoir of pathological signaling molecules, is a novel research field for PDR.
We analyzed the noncoding RNA transcript expression profiles of three patient groups: IMH, PDR treated with conbercept, and PDR alone. As a control for cytokine analyses in PDR, vitreous samples with idiopathic epiretinal membrane and/or IMH have been used in many studies [
24‐
26]. Nonetheless, it has been reported that the activation conditions of inflammation and fibrosis in eyes with idiopathic epiretinal membranes should be carefully considered as a control group [
27]. This study only included patients with IMH in the control group.
We identified that 1067 noncoding RNA transcripts were aberrantly expressed in patients with PDR. We also evaluated the effects of anti-VEGF therapy on the expression of lncRNA in patients with PDR. Ultimately, 835 dysregulated noncoding RNA transcripts were identified. We followed up on the differential expression analysis with qRT-PCR tests in a separate validation cohort, including six patients in group A, eight patients in group B, and twenty-five patients in group C. Most transcriptomes by qRT-PCR (6/10) were consistent with the results of gene microarray analysis, verifying the reliability of the microarray data.
Plasma samples from the same patient were subjected to qRT-PCR, which offers an alternative noninvasive strategy and helps to analyze whether a differential expression is more attributable to the local effect of the ocular vitreous or to the mutual influence of systemic and vitreoretinopathy modifications. The differential expression of all qRT-PCR-verified transcripts in plasma was not statistically significant, suggesting that these transcripts are likely to result from local differential expression of the ocular vitreous.
In this study, LINC01210 expression levels were significantly lower in the vitreous humor of patients with IMH patients in those study. LINC01210 is associated with the proliferative, migratory, and invasive abilities of cells [
28,
29]. We need further investigation to verify the potential role of LINC01210 in PDR. To date, the effects of most lncRNAs tested by qRT-PCR in our study are not explicitly understood. In the future, comprehensive studies on the function of lncRNAs in the pathogenesis of PDR will help determine new and effective diagnostic and therapeutic targets.
Most lncRNAs were poorly annotated. Bioinformatics analysis was used to investigate further differentially expressed lncRNAs. Using bioinformatics analysis, we found that the mRNA expression levels of frizzled class receptor 6 (FZD6) and proteasomal subunit α4s (PSMA8), which associated with the Wnt signaling pathway and Alzheimer disease, were downregulated in the vitreous of patients with PDR. Of interest, increased Wnt signaling is one of the causes of pathological ocular neovascularization of DR [
30,
31]. Further, multiple factors of DR have been shown to play a vital role in the development of neurodegeneration in Alzheimer’s disease [
32]. These potential correlations will be explored in future studies; in vivo and in vitro studies should be performed to elucidate the molecular mechanisms of lncRNA-mediated PDR pathogenesis.
This study had certain limitations. First, the number of patients included in this study was relatively small. However, these results were statistically significant. We provided one of the few lncRNA expression profiles in individual vitreous samples of patients with PDR. Second, vitrectomy surgery for patients with DR was performed out of necessity, such as in cases of vitreous hemorrhage and retinal detachment [
33]. Therefore, this study did not exclude patients with vitreous hemorrhage. If vitreous hemorrhage was incorporated, the surgeon avoided collecting as much blood as possible. Blood-borne molecules in vitreous samples cannot be completely eliminated [
9]. Finally, a selection bias may exist. The effects of anti-VEGF drugs were evaluated in different cohorts because it is difficult to collect dissimilar vitreous samples from same patient with PDR.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.