The expression profile of developmental stage-dependent circular RNA in the immature rat retina

Animals and tissue dissection

Sprague-Dawley rat pups (male and female) aged P3, P7, and P12 days were provided by the Experimental Animal Center at the Shanghai General Hospital in Shanghai, China. Eight rat pups were used for each group of P3, P7, and P12. Rat pups were housed with their dam under a 12 h:12 h light-dark cycle at room temperature 35–37 °C, with food and water available ad libitum until the experiments were conducted. All experimental procedures were reviewed and approved by the Animal Care Committee at the Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, and were conducted under the guidelines of the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health (National Institutes of Health Publication No. 85–23, revised in 1996) and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Every effort was made to minimize the number of animals used, and their discomfort, during all experimental procedures.

Because of the limited size of the rat retina, we combined the retinas of each group of P3, P7, and P12 rats for high-throughput sequencing of circRNA and immunohistochemistry or terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay. The basic experimental protocol was slightly modified from that in previous studies [32]. Briefly, the pups were euthanized instantaneously by decapitation, and then their eyes were rapidly dissected with fine scissors and transferred to an ice-cold (0‒4 °C) bath of artificial cerebrospinal fluid (ACSF: composition in mM: NaCl 119, KCl 2.5, K₂HPO₄ 1.0, CaCl₂ 2.5, MgCl₂ 1.3, NaHCO₃ 26.2, and D-Glucose 11). The bath was continuously bubbled with a 95% O₂/5% CO₂ gas mixture. About one fifth of the eyeball circumference at the edge of the cornea and sclera was resected using ophthalmological scissors to facilitate the perfusion of ACSF through the retina. After 1 h of recovery in ACSF (37 °C) bubbled with a 95% O₂/5% CO₂ gas mixture, the retinas were dissected and then either frozen quickly with liquid nitrogen for circRNA high-throughput sequencing or fixed in 4% paraformaldehyde at 4 °C for 24 h for immunohistochemistry and the TUNEL assay.

Cleaved caspase-3 immunohistochemistry and TUNEL

After the retinas were embedded in paraffin, retinal sections of 4 µm thickness were prepared. Three percent hydrogen peroxide was used to inactivate endogenous peroxidases, and EDTA, adjusted to pH 9.0, was used for heat-induced antigen retrieval for 8–10 min. After rinsing with PBS (1X; 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄, pH 7.4), retinal sections were incubated with rabbit anti-cleaved caspase-3 primary antibody (#9661s, dilution: 1/300, Cell Signaling Technology, Danvers, MA) was performed at 4 °C overnight, followed by incubation with goat anti-rabbit immunoglobulin G (IgG; PV-9001, ZSGB-BIO, Beijing, China) secondary antibody at 37 °C for 1 h. After washing, 3,3′-diaminobenzidine (DAB, ZLI-9017, ZSGB-BIO) was used to visualize immunoreactivity. All sections were then counterstained with hematoxylin to stain the nuclei. Finally, sections were dehydrated and mounted for microscopic examination.

For the TUNEL assay, retinal sections were deparaffinized and rehydrated. After rinsing with PBS, the sections were treated with proteinase K (Roche Applied Science, Indianapolis, IN) and quenched with 3% hydrogen peroxide; then they were incubated in a terminal deoxynucleotidyl transferase (TdT) reaction mix (Roche Applied Science) for 1 h at 37 °C. Sections were then incubated for 5 min with 4',6-diamidino-2-phenylindole (DAPI) at room temperature.

CircRNA enrichment and high-throughput sequencing

RNA was isolated using the RNeasy system (Qiagen, Dusseldorf, Germany), including on-column DNase I digestion (Qiagen) to eliminate DNA. Then the total RNA was depleted of ribosomal and linear RNA using the RiboMinus kit (Thermo Fisher Scientific, Waltham, MA) and RNase R (Epicenter, Madison, WI), respectively. CircRNAs were validated with quantitative reverse transcription (RT)–PCR using the VAHTSTM Library Quantification Kit (Vazyme Biotech Co., Ltd, Piscataway, NJ). Specifically, samples were heated to 70 °C for denaturing and then cooled to 40 °C on a thermocycler with RNase R buffer to digest linear RNA. After RNase R digestion and circRNA reverse transcription, the cDNA library was produced by PCR (conditions: 98 °C for 30 s, followed by 15 cycles of 98 °C for 10 s, 60 °C for 30 s, and 72 °C for 30 s, then 72 °C for 5 min) following depletion of the second cDNA chain with the USER enzyme. The sequencing libraries were constructed using the VAHTSTM Library Quantification Kit (Vazyme Biotech Co) and were sequenced with an Illumina HiSeq™ 2000 flowcell sequencer (Illumina, San Diego, CA).

Data analysis

The RNA-sequenced (RNA-seq) data were analyzed using FastQC software to verify the quality of the data [33]. The BWA-MEM alignment algorithm of the software BWA (the Burrows-Wheeler Alignment tool) was used to match the RNA-seq data to the genome [34]. To annotate the validated circRNAs, we used CIRI (CircRNA Identifier) software to predict the circRNAs [35] present in the sample and sequenced them using circBase [36]. The false discovery rate (FDR) was used to correct for multiple hypothesis testing, and a twofold change and FDR≤0.001 was considered statistically significant [37]. Caspase-3-positive or TUNEL-positive cells were counted in a double-blinded manner from five discontinuous views randomly selected from each sample (n = 5 retinas). Data in the figures were analyzed with GraphPad Prism 5 software (GraphPad Software Inc., San Diego, CA) and presented as mean ± standard error of the mean (SEM). A one-way ANOVA (ANOVA) followed by the Bonferroni correction, or the appropriate equivalent non-parametric test, was used for comparisons among groups, and a p-value of < 0.05 was denoted as statistically significant.

CircRNAs change during early development in the rat retina

We analyzed the raw reads or raw data of the expression of circRNA at P3, P7, and P12, using high-throughput sequencing followed by filtration and quality control (QC; Figure 1A, Table 1) to obtain clean reads as in a previous study [35]. Because exon–exon junction reads (Figure 1B-C) are important features of circRNAs that can map to the junction points of exons, the BWA software was used to match the clean reads to the genome and identify the junction reads. The results showed that the clean reads were expressed abundantly (7–9 × 10⁷/group) and largely mapped to the genome (>75% of all reads) in the early developmental retina (Figure 2A, Table 2). Most of the clean reads mapped on the exon regions could be visualized at P7 (56.41% of all reads) but were present in only small amounts at P3 (15.21% of all reads) and P12 (17.78% of all reads; Figure 2A, Table 2). After the CIRI predictions and false-positive read filtration, validated circRNAs were obtained and annotated according to circBase (Figure 1E, Table 3, Appendix 1, Appendix 2 and Appendix 3).

To identify circRNA changes in the developing rat retina, the total circular junction reads of the different stages were further separated with BWA. CircRNAs with the same sequence were regarded as one circRNA species. The number of circular junction reads at P12 (143,815) was greater than that at P3 (78,758) and P7 (78,385), while the number of circRNA species at P7 (7,021) was greater than that at P3 (2,654) and P12 (5,628; Figure 2B, Table 4).

After the data were integrated, a total of 10,890 circRNAs were obtained; additionally, the overall expression of circRNAs was different among the P3, P7 and P12 time points. For example, there were 4,123 circRNAs expressed solely at P7 (Figure 2C). By comparing the expression of circRNA between the time points, we detected differential expression (greater than a twofold change, FDR≤0.001; Figure 3A,B, Appendix 4 and Appendix 5) of 1,371 circRNAs (out of 8,223 circRNAs) between P3 and P7 and 1,112 circRNAs (out of 10,016 circRNAs) between P7 and P12. We found that 558 circRNAs and 813 circRNAs (out of 1,371 circRNAs) were upregulated and downregulated, respectively, from P3 to P7. The circRNAs upregulated from P3 to P7 included those originating from the genes Akt3, Ccdc6, Gnaz, and Mpped2; the circRNAs downregulated from P3 to P7 originated from the genes Ly6g6f, Melk, Casc5, Smarcc1, and Pank2 (Figure 3C, Appendix 4). We found that 434 or 678 circRNAs (out of 1,112 circRNAs) were upregulated or downregulated, respectively, from P7 to P12. Upregulated circRNAs included those derived from the genes Pde6a, Unc13b, Pde4b, Pi4ka, Mtr, and Dnmbp, while downregulated circRNAs included those derived from the genes Pank2, Smarcc1, Melk, Ly6g6f, Casc5, Dst, and Pak3 (Figure 3D, Appendix 5). After the circRNAs expressed at only P3 or P12 were filtered out, a total of 438 circRNAs were retained for measuring significant changes in the circRNAs across the time points (Appendix 6, Figure 2D). Of these 438 circRNAs, ten were gradually downregulated, and 17 were gradually upregulated from day P3 to P12 (Appendix 7). In addition, 145 circRNAs were present in only small amounts or were undetectable (48 circRNAs) at P7 relative to P3 and P12. In addition, 266 circRNAs of the 438 circRNAs were present at P7 but were absent or diminished at P3 and P12 (172 circRNAs appeared only at P7, Figure 2D).

Physiologic neuronal apoptosis during early development in the rat retinal ganglion cell layer

Most neurons undergoing PCD exhibit the morphological and biochemical hallmarks of apoptosis, such as cell shrinkage, activation of caspases, and DNA fragmentation [5]. Cleaved caspase-3 is produced early in the apoptotic process and is responsible for executing the apoptotic process, making it a strong biomarker of apoptosis [38]. The TUNEL assay detects apoptosis by labeling the fragmented DNA generated during PCD. For enhanced reliability of the results, we used anti-cleaved caspase-3 immunohistochemistry and the TUNEL assay to detect apoptosis in retinas from rats at ages P3, P7, and P12. The cleaved caspase-3-positive-to-negative cell ratio increased from 1.69 ± 0.25% to 3.19 ± 0.42% between P3 and P7 (p = 0.005); it then decreased to 1.69 ± 0.25% between P7 and P12 (p = 0.008; Figure 4A,C). The TUNEL-positive cell ratio increased from 6.75 ± 0.88% to 12.77 ± 1.17% between P3 and P7 (p<0.001) and then decreased to 7.16 ± 0.62% between P7 and P12 (p<0.001; Figure 4B,D). The results confirmed that normal age-dependent physiologic apoptosis of neurons was occurring.

Candidate circRNAs from host genes associated with apoptosis

We identified changes in the expression of circRNA that could be linked to normal developmental apoptosis during in the neonatal rat retina. A total of 15 circRNAs from genes associated with apoptosis were obtained after further filtration and screening (Table 5). We found that five circRNAs (from the genes Optn, Thoc1, Rbm5, Ddx19a, and Bnip2) could not be detected, and two circRNAs (from the genes Pias1 and Naa35) were not expressed at P7 but were abundantly expressed at P3 and P12. Three circRNAs (from the genes Rere, Arhgap10, and Birc6) were expressed specifically at P7, and three circRNAs (from the genes Ranbp9, Epha7, and Braf) could be visualized at P7 but were absent or diminished at P3 and P12 (Figure 5).