Background
Age-related cataract is the predominant cause of blindness and visual impairment worldwide, and there is no consistent conclusion regarding its aetiology [
1]. It was documented that oxidative stress, DNA damage repair, response to apoptosis, autophagy, and ferroptosis are factors involved in ARC pathogenesis [
2‐
7]. Ferroptosis, a novel nonapoptotic form of oxidative cell death, is mainly induced by iron-dependent and lipid peroxidation [
8]. The lipid hydro peroxidase glutathione peroxidase 4 (GPX4) translates lipid hydroperoxides into lipid alcohols and suppresses the iron ion (Fe
2+)-dependent formation of toxic lipid reactive oxygen species (ROS) [
9]. Moreover, the vital role of GPX4 in cataract formation has been confirmed, due to the rapid development of lens opacity in GPX4 knockout mice [
10,
11]. Therefore, GPX4, as the central regulator of ferroptosis, might be crucial to the maintenance of lens transparency.
Previous studies have clarified that the expression of oxidative damage repair genes could be regulated by long noncoding RNAs (lncRNAs) [
12‐
14]. LncRNAs can regulate targeted gene expression at the epigenetic level by interacting with mRNAs [
13], microRNAs (miRNAs) [
14], or proteins [
15]. Moreover, our previous studies have demonstrated that several lncRNAs potentially exert an influence on ARC formation by different mechanisms [
12‐
14]. However, many previous studies primarily focused on the regulation of target genes by lncRNAs, with less focus on the mechanisms regulating lncRNAs expression itself. Recently, N
6-methyladenosine (m
6A) has been identified as a structural alteration affecting lncRNA expression [
16].
M
6A is the most abundant epigenetic modification within lncRNAs in eukaryotes and plays an important role in RNA biology [
17]. The modification can be regulated by multiple enzymes including m
6A methyltransferases (METTL3, METTL14, and WTAP), demethylases (FTO and ALKBH5), and m
6A-binding proteins (YTHDF1-3 and YTHDC1-2) [
18]. Our study indicated that the altered expression of ALKBH5 could contribute to changes in circRNA m
6A modifications in LECs [
6]. Numerous studies have revealed that m
6A modifications can regulate the expression of m
6A-labelled lncRNAs [
19‐
21]. Even so, research on the m
6A modification of lncRNAs is still lacking and the roles of RNA modification in ARC formation remain unknown.
Herein, we performed genome-wide altered m6A-tagged lncRNA screening in LECs from age-related cortical cataracts (ARCCs). Possible functions of m6A-modified lncRNAs were predicted by Gene Ontology (GO) annotation and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. The data suggested that the m6A-modified lncRNA ENST00000586817 might be involved in the development of ARC by cis-targeted genes to regulate ferroptosis-related GPX4 in LECs.
Methods
Study participants
ARCC patients were recruited in this study and the Lens Opacities Classification III (LOCS III) system was used for the classification of disease severity [
22]. In addition, age-matched individuals without cataracts were removed from vitreoretinal diseases and selected as controls. The inclusion and exclusion criteria for the selection of the participants in the current study were identical to our previous research [
6]. After the abovementioned screening, 19 patients with ARCC, and 19 control patients were finally included. The basic demographic data of all participants are listed in Table
1.
Table 1
The grade of lens opacity and identification codes of controls and ARCCs
No.1 | Male | 61 | C0N1P0 | | No.1 | Male | 59 | C3N1P1 |
No.2 | Male | 56 | C0N1P0 | | No.2 | Male | 64 | C3N1P1 |
No.3 | Male | 58 | C0N1P0 | | No.3 | Male | 62 | C3N1P1 |
No.4 | Male | 54 | C0N1P1 | | No.4 | Male | 67 | C3N1P1 |
No.5 | Male | 55 | C0N1P1 | | No.5 | Male | 59 | C3N1P1 |
No.6 | Male | 57 | C0N1P1 | | No.6 | Male | 64 | C3N1P1 |
No.7 | Male | 58 | C0N1P1 | | No.7 | Male | 63 | C3N1P1 |
No.8 | Male | 53 | C0N0P1 | | No.8 | Male | 65 | C4N1P2 |
No.9 | Male | 50 | C1N1P0 | | No.9 | Female | 59 | C3N1P1 |
No.10 | Male | 62 | C1N0P1 | | No.10 | Female | 58 | C3N0P1 |
No.11 | Male | 58 | C1N1P0 | | No.11 | Female | 59 | C3N1P1 |
No.12 | Female | 62 | C0N1P1 | | No.12 | Female | 55 | C4N1P2 |
No.13 | Female | 65 | C0N1P1 | | No.13 | Female | 65 | C3N1P1 |
No.14 | Female | 68 | C1N1P1 | | No.14 | Female | 68 | C4N1P1 |
No.15 | Female | 59 | C0N0P1 | | No.15 | Female | 69 | C3N2P1 |
No.16 | Female | 57 | C1N1P1 | | No.16 | Female | 61 | C4N1P1 |
No.17 | Female | 69 | C1N0P1 | | No.17 | Female | 56 | C3N1P1 |
No.18 | Female | 65 | C0N1P1 | | No.18 | Female | 54 | C4N2P1 |
No.19 | Female | 54 | C0N0P1 | | No.19 | Female | 58 | C3N1P1 |
RNA extraction and MeRIP-seq
Using TRIzol reagent (Invitrogen), we isolated total RNA from LECs. The method was performed as reported in our previous study [
6]. The MeRIP-seq method was used to measure m
6A-methylated RNA, supported by Cloudseq Biotech Inc. (Shanghai, China). In brief, m
6A immunoprecipitation was conducted with the GenSeqTM m
6A-MeRIP Kit (GenSeq Inc., China) according to the manufacturer’s instructions. The protocol for MeRIP-seq assays has been described in our previous article [
6].
High-throughput sequencing data analysis
After sequencing on the Illumina HiSeq 4000 sequencer, paired-end reads were harvested and quality controlled by Q30(Figure
S1-
S2). In addition, Table
S1 shows the alignment rates, sequencing depth, and peak metrics and Figure
S3 illustrates the PCA plots and correlation analysis of different samples. The clean data were obtained through 3’ adaptor-trimming and removal of low-quality reads using Cutadapt software (v1.9.3). After that, clean data were aligned to the reference genome (UCSC HG19) HISAT2 via software (v2.0.4). Methylated sites on lncRNAs (peaks) were identified by MACS software and diffReps, which were detected by both software overlapping with exons of lncRNA were determined and prepared with custom-made scripts. Based on the source genes of differentially methylated lncRNAs, we performed GO and KEGG analysis [
23].
Different m
6A sites were identified by diffReps, with a cut-off of
p ≤ 0.0001 and fold-change ≥ 2. Using homemade scripts, peaks were identified as overlapping with exons of lncRNA. The differential expression of mRNAs and lncRNAs between the two groups was assessed via the edgeR R package. LncRNAs with a cut-off of fold-change ≥ 2 were considered differentially expressed, while mRNAs with a cut-off of
p value < 0.05 and fold-change ≥ 2 were considered differentially expressed. The algorithm was applied to search for lncRNA
cis-regulated target genes was based on chromosome coordinates. Putative
cis-acting
regulatory DNA
elements (
cis-
elements) regulate the transcription of neighbouring genes. This study defined genes located within 10 kbp upstream or downstream of lncRNAs to be
cis regulated based on previous studies [
24,
25].
GO and pathway analysis
GO analysis and KEGG analysis were performed for functional annotation of differentially expressed mRNA and lncRNA cis targets. GO enrichment analysis was performed by the R ‘cluster Profiler’ package and KEGG enrichment analysis was tested on hypergeometric distribution by the R ‘hyper’ function. GO categories or pathways with a p value < 0.05 were considered significantly enriched.
Quantitative RT-PCR (qRT-PCR)
The lncRNA ENST00000586817 and
GPX4 levels were analysed by RT
-qPCR assay in LECs of ten ARCC patients and ten controls. In this research, the selected genes were validated by RT
-qPCR [
26]. The relative differences in gene expression between the two groups were expressed by using GAPDH as an internal control which was then compared to the target mRNA. The primer sequences and reverse primers used were as follows: ENST00000586817 forward primer CCACCAGCCACTGCTTCCT, reverse primer CACCCAACCTCCTACAACAACC; GPX4 forward primer GAGGCAAGACCGAAGTAAACTAC and reverse primer CCGAACTGGTTACACGGGAA; GAPDH forward primer TGAAGGTCGGAGTCAACGGATTTGGT, reverse primer CATGTGGGCCATGAGGTCC ACCAC. Relative fold expression changes were determined by the comparative CT (2
−△△CT) method. Online software was used to design the gene-specific primer pairs (
http://www.ncbi.nlm.nih.gov/tools/primer-blast/). All experiments were carried out in triplicate.
Transmission electron microscopy
LECs were fixed with 4% glutaraldehyde for 3 h. After the glutaraldehyde was removed, the LECs were embedded in 1% agarose and mixed with 4% glutaraldehyde. Lens anterior capsule tissues were directly fixed with 4% glutaraldehyde for 4 h and then treated with 1% osmic acid for 1.5 h. Before the samples were embedded in epoxy resin, they were sequentially dehydrated in 50%, 70%, 90%, and 100% acetone three times for 15 min each. Section (70 nm thick) were cut and stained with uranyl acetate for 15 min. Finally, samples were observed with a transmission electron microscope (Hitachi, Tokyo, Japan).
Western blot assays
Total protein was extracted from human LECs. The protocol for western blotting was been described in our previous study [
6]. Samples were incubated overnight with polyclonal rabbit anti-human-GPX4 (1:1000, Abcam) and rabbit anti-GAPDH (1:6,000, Abcam) monoclonal antibodies. After being washed, binding of anti GPX4 IgG was detected with alkaline phosphatase-conjugated goat anti-rabbit IgG antibody (1:10,000; Santa Cruz, Dallas, TX, USA).
Cell viability assay
SRA01/04 cells were seeded onto 96-well plates and treated with DMSO and RSL3 (0.2 µM) (MedChemExpress, New Jersey, USA). After 24 h, the Cell Counting Kit-8 assay (Dojindo Laboratory, Kumamoto, Japan) was used to detect cell viability. The process was as follows, after specific treatment, 10 µl of CCK-8 solution was added to each well and then incubated for 2 h at 37 °C. Then the cell viability was calculated by detecting the absorbance at 405–450 nm using a microplate reader (BioTek, Vermont).
Measurement of Fe2+ and MDA levels
The Fe2+ level was measured by the FerroOrange method, and the corresponding reagents were purchased from Dojindo Molecular Technologies Company. SRA01/04 cells were seeded in 24-well plates at a density of 5 × 105 cells/wells and cultured for 24 h, and then DMSO and RSL3 (0.2 µM) were added to the wells. Moreover, wells without treatments were prepared as a control group. After incubation for 24 h, the culture medium was removed, and the cells were washed with Hanks three times. When FerroOrange (1 µM, an intracellular Fe2+ ion probe, Ex: 543 nm, Em: 580 nm) dispersed in serum-free medium was added to each well, the cells were incubated for 30 min in a 37 °C incubator equilibrated with 95% air and 5% CO2. After the incubation was completed, fluorescence images of the cells were captured using a microscope (Leica, Germany).
Intracellular MDA concentrations were assessed using the Lipid Peroxidation MDA Assay Kit (Beyotime, Shanghai, China). After the indicated treatments, SRA01/04 cells were harvested and lysed in RIPA and CM lysis solution. After lysis at 4 degrees for half an hour, the tubes were centrifuged at 12 000 × g for 5 min and the supernatant was collected for subsequent experiments. This experiment strictly followed the manufacturer’s instructions.
Statistical analysis
Paired-end readings were obtained from the Illumina HiSeq 4000 sequencer and were then subjected to Q30 quality control. After three adaptor-trimming and poor-quality read removals steps, cutadapt software (v1.9.3) was used. First, using STAR software, clean reads from input libraries were aligned to the reference genome (UCSC HG19).
Next, HISAT2 software aligned clean reads from all libraries to the reference genome (v2.0.4). By using MACS software, methylated sites on lncRNAs (peaks) were found. DiffReps was used to find the differentially methylated sites, and the source genes of the differentially methylated lncRNAs were used to conduct GO and pathway enrichment analyses. These peaks identified by both software programs overlapping with exon lncRNAs were identified and selected by homemade scripts. All of the results was expressed as the means ± SDs of experiments that were repeated at least three times. All statistical analyses were performed using SPSS software, version 25.0 (IBM SPSS, Armonk, NY, USA) and GraphPad Prism software 7.0 (GraphPad Software). One-way ANOVA and Student’s t test were used for statistical analyses, with p < 0.05 was considered to indicate statistical significanc.
Discussion
Emerging evidence has demonstrated the involvement of deregulated lncRNAs in the pathogenesis of multiple ocular diseases, including ARC [
12‐
14]. LncRNA degradation is not significantly different from mRNA degradation. Recently, increasing attention has been focused on m
6A RNA modification which is regarded as a new epigenetic event and has been demonstrated to affect the degradation of targeted lncRNAs and participate in the progression of various age-related diseases [
26‐
28]. However, the function of this new RNA modification of lncRNAs in ARC formation has not been characterized.
In this study, we used genome-wide profiling of m
6A-labelled lncRNAs between ARCCs and controls. Our results showed that the expression of differentially regulated lncRNAs was comparable between the two groups. There were 1617 upregulated m
6A-lncRNAs and 976 downregulated m
6A-modified lncRNAs in ARCCs. Our results suggest that the modification of m
6A-modified lncRNAs is significantly upregulated in ARC LECs. M
6A modification is regulated by RNA methyltransferases, which have been named “writer”, and can be removed by RNA demethylases, including FTO and ALKBH5 [
29]. In our previous study, the expression of ALKBH5 was significantly increased in LECs of ARCCs among five major methyltransferases [
6]. We speculated that the altered methyltransferase expression induced the m
6A RNA modification change and contributed to the downregulation of lncRNAs. Then, we compared the expression of m
6A modified lncRNAs, and non-m
6A-modified lncRNAs between ARCCs and controls. Overall, the m
6A-levels of lncRNAs were negatively correlated with the expression levels of lncRNAs. In ARCC, m
6A-levels of lncRNAs were negatively correlated with the expression levels of lncRNAs. The analysed data showed that the expression levels of m
6A-labelled lncRNAs and non-m
6A-labelled lncRNAs in the ARCC patients were all higher than those in controls. In addition, these data underscored the dynamic process regulating the m
6A methylation levels on lncRNAs and the expression level of m
6A-labelled lncRNAs in ARCCs.
M
6A modifications of lncRNAs is reversible, and which may act as a switch to control lncRNA functionality and further impact cellular function by various underlying mechanisms associated with human diseases. Many previous studies have found that lncRNAs can regulate gene expression to mediate ARC formation through multiple mechanisms [
12]. One mechanism is functioning as a competing endogenous (ce)RNA to bind to miRNA. As shown in our study, lnc-PLCD3 regulated PLCD3 expression by sponging miR-224-5p in ARC as a ceRNA [
14]. lncRNA H19 could be a useful prognostic marker of early ARC and a promising therapeutic target for ARC treatment [
12]. Antisense lncRNAs can downregulate gene expression by binding to target genes. For example, RNA GPX3-as decreased the apoptosis of LECs by directly promoting GPX3 expression [
13]. In addition to the above mechanisms, lncRNAs function in a wide range of biological processes and can regulate gene expression in
cis mechanisms [
30]. Although that most efforts have concentrated on individual lncRNAs, researchers have attempted to classify intergenic lncRNAs based on their genomic positions relative to protein coding loci and have found a remarkable pattern of
cis regulation by divergent lncRNAs at adjacent sites [
31]. One classic
cis-acting intergenic lncRNA Haunt (HOXA upstream noncoding transcript) is located on 40 kb upstream of the HOXA cluster [
32,
33]. Interestingly, the lncRNA and its genomic locus exert opposing influences in regulating the same target genes. Through the Haunt DNA locus, which includes potential HOXA enhancers, Haunt RNA acts in
cis to prevent abnormal high-level transcription, thus promoting fine-tuned expression of HOXA genes during embryonic stem cell differentiation [
32]. The intergenic lincRNA Peril provides another example of how a genomic lncRNA sequence can serve as a
cis enhancer for nearby transcription [
34]. The expression of Peril and nearby SOX2 in mESCs was abolished via genomic deletion of a superenhancer upstream of the first intron of Peril, contributing to a global expression change and reduced proliferation of mESCs [
34]. Thus, intergenic lncRNA can regulate nearby gene expression and involve in a number of biological processes [
31]. Overall,
cis regulation of nearby transcription by intergenic lncRNAs plays an important role in regulation of gene expression. Here we aimed to identify lncRNA regulation of gene expression through
cis mechanisms. In this study, we aimed to identify lncRNA regulation of gene expression through
cis mechanisms. Combined with analysis of DE-m
6A-lncRNAs, DE-lncRNAs, DE-mRNAs and the intersection of genes associated with ARC pathogenesis including DNA damage repairs and cell death pathways, we enriched the m
6A-lncRNA-mRNA network related to the mechanisms of ARC. Among this, GPX4 appeared in the downstream target molecule in the lncRNA-mRNA network diagrams. Based on the location of the intergenic lncRNA ENST000000586817 in the genome, we found that the transcript of GPX4 is adjacent. We therefore surmise that downregulated m
6A-modified lncRNA (ENST00000586817) might regulate GPX4 expression in ferroptosis networks.
Ferroptosis is a recently discovered iron-dependent cell death process characterized by the abnormal accumulation of iron-dependent oxidative damage products in cells. In contrast to apoptosis, necroptosis, and other forms of non-apoptotic cell death, it is unique in the central involvement of iron-dependent lipid ROS accumulation and can be triggered by small molecules that block GSH synthesis or GPX4 activity. Ferroptosis is manifested by an increase in Fe
2+ and lipid peroxidation [
35]. Many studies have explored the relationship between iron concentration and cataract formation. For instance, the levels of iron in cataractous human lenses were increased compared to those in clear lenses [
36,
37]. These increased iron contents could be identified as “redox available” in the cataractous lens and might be potentially cataractogenic [
38]. In addition, the experimental findings of several groups support that lipid peroxidation could be one of the initial mechanisms of cataractogenesis [
39‐
42]. In this study, we found that downregulation of the expression of GPX4 in SRA01/04 cells using the GPX4 inhibitor RSL3 led to higher more lipid ROS and Fe
2+ levels, which induced ferroptosis. Cell viability was also significantly inhibited. In terms of epigenetics, research on the occurrence and regulation of ferroptosis is still in its infancy. Another study screened for transcript levels of some key ferroptosis-related genes in ARCs by transcripyome sequencing analysis, but the upstream regulatory mechanism has not yet been studied [
7]. Therefore, our study aimed to enrich the knowledge on ferroptosis epigenetics, and provides a new direction for understanding the pathogenesis of ARCs.
Although, this study focused on m6A-lmodified lncRNA networks, some limitations still exist. Based on bioinformatics analysis, there is a lack of experiments to validate these results in this study. Another limitation is the small sample size and future larger sample study was required to validate our results. Although, our study offers a new perspective for the lncRNA regulation through the m6A-modification for further studies with larger sample sizaes are exploring the mechanisms and functions of m6A-lncRNAs for ARC. Moreever, FISH and nuclear- and cytoplasmic-fractionated RT-qPCR both will be needed to indicate the effects of lncRNA and GPX4 in the future experiments. Furthermore, we also need to systematically illustrate the morphological characteristics and metabolic regulation of mitochondria in the regulation of ferroptosis. Finally, experiments in related in vivo and in vitro models could be the future direction to certify these speculations.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.