Ocular A-to-I RNA editing signatures associated with SARS-CoV-2 infection

Ophthalmic manifestations have recently been observed in acute and post-acute complications of COVID-19 caused by SARS-CoV-2 infection. Our precious study has shown that host RNA editing is linked to RNA viral infection, yet ocular adenosine to inosine (A-to-I) RNA editing during SARS-CoV-2 infection remains uninvestigated in COVID-19. Herein we used an epitranscriptomic pipeline to analyze 37 samples and investigate A-to-I editing associated with SARS-CoV-2 infection, in five ocular tissue types including the conjunctiva, limbus, cornea, sclera, and retinal organoids. Our results revealed dramatically altered A-to-I RNA editing across the five ocular tissues. Notably, the transcriptome-wide average level of RNA editing was increased in the cornea but generally decreased in the other four ocular tissues. Functional enrichment analysis showed that differential RNA editing (DRE) was mainly in genes related to ubiquitin-dependent protein catabolic process, transcriptional regulation, and RNA splicing. In addition to tissue-specific RNA editing found in each tissue, common RNA editing was observed across different tissues, especially in the innate antiviral immune gene MAVS and the E3 ubiquitin-protein ligase MDM2. Analysis in retinal organoids further revealed highly dynamic RNA editing alterations over time during SARS-CoV-2 infection. Our study thus suggested the potential role played by RNA editing in ophthalmic manifestations of COVID-19, and highlighted its potential transcriptome impact, especially on innate immunity. Supplementary Information The online version contains supplementary material available at 10.1186/s12864-024-10324-z.


Introduction
Since 2019, the world has been facing a pandemic of coronavirus disease 2019 (COVID-19) caused by the SARS-CoV-2 virus, resulting in significant impacts on global healthcare and economies [1].Symptoms of COVID-19 vary from being asymptomatic to severe [2,3].SARS-CoV-2 infection could cause viral pneumonia [4], but can also affect the heart, liver, kidney, brain [5][6][7][8][9], and eyes [10].Understanding such infections in other tissues and organs than the lung is essential for the control and treatment of acute and post-acute COVID-19 sequelae.
Adenosine-to-inosine (A-to-I) RNA editing mediated by adenosine deaminases that act on RNA (ADARs) was the most common canonical RNA editing in mammals [36].Three members of ADARs are encoded in the human genome, including ADAR, ADARB1, and ADARB2.ADAR (also known as ADAR1) and ADARB1 (also known as ADAR2) are expressed in many tissues and demonstrate catalytic activity for adenosine deamination, whereas ADARB2 (also known as ADAR3) is mainly present in the brain and no adenosine deaminase activity has been reported [37][38][39].As an important component of epigenetics, RNA editing plays an important role in various physiological and pathological processes [40].It is associated with the pathogenesis of neurodegenerative diseases such as amyotrophic lateral sclerosis (ALS), Parkinson's disease, and Alzheimer's disease [41].In addition, A-to-I RNA editing is an important component of innate and adaptive immunity and plays an important role in the host's antiviral responses [38], such as those to the Ebola virus, hepatitis virus, SARS, and SARS-CoV-2 [42,43].Meanwhile, ADARs influenced SARS-CoV-2 infection in vivo [44].Moreover, our recent study revealed a possible link between host RNA editing and infection with singlestrand RNA viruses, including SARS-CoV-2 in mouse models [45].However, ocular A-to-I RNA editing during SARS-CoV-2 infection remains uninvestigated in COVID-19 patients.
Herein, we performed a transcriptome-wide analysis to examine the RNA editing profiles of SARS-CoV-2 infections in five ocular tissues, to identify SARS-CoV-2 infection-associated signatures of host RNA editing across tissues.Our findings highlight both the similarities and differences in host RNA editing during SARS-CoV-2 infections and provide valuable insights into the epigenetic mechanisms of RNA editing underlying the ophthalmic manifestations of SARS-CoV-2 infection.

RNA-Seq dataset downloads
We downloaded three RNA-Seq datasets containing 37 samples of five ocular tissues from the European Nucleotide Archive (ENA) (https:// www.ebi.ac.uk/ ena).These datasets include PRJNA790648 [46], PRJNA688734 [47], and PRJNA731890 [48].Dataset PRJNA688734 contains samples from the cornea, limbus, and sclera isolated from human donor tissues and passaged in tissue culture.Cells of the three tissue were infected with SARS-CoV-2 in triplicate and their RNA was collected at 24 h postinfection (hpi), and compared to mock samples (N = 3 each).The PRJNA790648 contains mock (N = 4) and SARS-CoV-2 infected (N = 3) ex vivo cultures of an airliquid interface organotypic conjunctival epithelial model epithelia infected.Dataset PRJNA731890 [48] contained samlpes of human stem cell-derived retinal organoids that were mock or infected with SARS-CoV-2 collected at 24 and 96 hpi (N = 3 each).Details of these datasets are shown in Table 1.

Read alignment
The raw sequencing reads were processed using a workflow as previously described [49].In summary, raw sequencing data quality was assessed using FASTQC [50].Read alignment was performed using RNA STAR (version 2.7.0e) and the human genome reference sequence (UCSC hg38) [50].Base quality scores were then recalibrated using GATK (version 4.1.3)[51] after filtering duplicated reads using SAMtools (version 1.9) [52].

RNA editing identification
VarScan (version 2.4.4) and the Ensembl Variant Effect Predictor (VEP) were used for the identification and annotation of A > G single nucleotide variations (SNVs) [53,54].SNVs meeting specific criteria were selected, including a base quality of ≥ twenty-five, total sequencing depth of ≥ ten, alternative allele depth of ≥ two, and alternative allele frequency (AAF) of ≥ 1%.SNVs found in the REDIportal V2.0 database were considered highconfidence A-to-I RNA editing sites [55].Additional filtering criteria were applied to the remaining SNVs, such as excluding those located in homopolymer runs (≥ five nucleotides) or simple repeats, mitochondrial genes, within six nucleotides from splice junctions, within one nucleotide from RNA insertion-deletion (INDEL), within 4% of the ends of reads, annotated as known variants in the dbSNP database Build 142, and those with AAF values of 100% or between 40 and 60% in more than 90% of NC and infected samples.High-confidence A-to-I RNA editing sites were retained for subsequent data analysis, defined as those with editing levels ≥ 1% and observed in two or more samples.

RNA binding protein (RBP) binding site prediction
To gain deeper insights into the potential functional consequences of RNA editing, RBPmap (http:// rbpmap.techn ion.ac.il) was used to predict RNA binding protein sites that coincided with RNA editing sites [57].

Statistical analysis
The general linear model (GLM) and likelihood ratio test were used to compare RNA editing levels between NC and SARS-CoV-2 infected tissues and identify differential RNA editing (DRE), which could be associated with SARS-CoV-2 infection.Empirical P-values (P GLM ) were calculated using the likelihood ratio test.For sites with P GLM less than 0.05, Fisher's exact test was further used to compare the sequencing depth of the reference and alternative alleles in both groups.We applied the Benjamini-Hochberg method to calculate the false discovery rate (FDR).RNA editing sites with FDR less than 0.05 were considered differentially edited.Additionally, we analyzed the correlation between RNA editing and gene expression levels using Spearman's correlation.

A-to-I RNA editing profiles in ocular tissues
To assess whether and how ocular A-to-I RNA editing was involved during SARS-CoV-2 infection, we searched and downloaded the publicly available RNA-seq datasets containing ocular tissues or organoids infected with SARS-CoV-2.The results showed the average RNA editing level was decreased in the conjunctiva, limbus, and sclera, but increased in the cornea, with no significant changes between mock and infected retinal organoids (Fig. 1A).Meanwhile, upon infection, ADAR expression decreased in the conjunctiva, whereas ADARB1 expression decreased in the limbus and sclera (Fig. 1B, C).Furthermore, a total of 178558 editing sites were identified in the conjunctiva, 1084 in the limbus, 1140 in the cornea, 941 in the sclera, and 44716 in retinal organoids (Fig. 1D).

Functional relevance of SARS-CoV-2 infection-associated A-to-I RNA editing in ocular tissues
Functional enrichment analysis using these DRE genes was then used to understand the impact of A-to-I RNA editing changes on biological functions during SARS-CoV-2 infection.Notably, the results in Fig. 3 showed more evident functional enrichment of DRE in the conjunctiva and retinal organoids than in the other three tissues included in the current study.The common enrichment in DRE genes consisted of biological processes mainly related to ubiquitin-dependent protein catabolic process, regulation of transcription, RNA splicing, and positive regulation of gene silencing by miRNA (Fig. 3A); common molecular functions mainly related to RNA binding, protein binding, cadherin binding, and metal ion binding (Fig. 3B); common cellular components included the cytosol, nucleoplasm, nucleus, and cytoplasm (Fig. 3C), and common KEGG pathways were mainly related to ubiquitin-mediated proteolysis and herpes simplex virus 1 infection (Fig. 3D).Tissue type also had its unique enriched features, especially those related to viral infection, such as biological processes related to defense response to virus, negative regulation of viral genome replication, and positive regulation of type I interferon-mediated signaling pathway were uniquely found in the cornea, (Fig. 3A), and KEGG pathways of   coronavirus disease-COVID-19 were unique to the retinal organoids (Fig. 3D).

SARS-CoV-2 infection-associated A-to-I RNA editing in ocular surface tissues
We then focused on the role of A-to-I RNA editing during this process.1454 DRE sites in 991 genes, 41 DRE sites in 33 genes, 51 DRE sites in 38 genes, and 51 DRE sites in 35 genes were identified in the conjunctiva, limbus, cornea, and sclera infected (Fig. S1).Notably, the top 50 DRE sites (ranked by empirical P-values) in the different tissue types (Fig. S2) were strongly correlated with ADAR and ADARB1 expression (Table 3), pointing to an active role of ADARs during ocular infection of SARS-CoV-2.In conjunctiva, the RNA editing levels of DNA polymerase gamma 2 (POLG2:chr17:64,495,902), inositol-trisphosphate 3-kinase C (ITPKC:chr19:40,729,566), and inflammation and lipid regulator with UBA-like and NBR1-like domains (ILRUN:chr6:34,652,970) decreased significantly after SARS-CoV-2 infection (Fig. 4A).In the limbus, the editing level of lysosomal associated membrane protein 2 (LAMP2:chrX:120,437,993), SON DNA and RNA binding protein (SON:chr21:33,550,969) deceased, and that of collagen type XII alpha 1 chain (COL12A1:chr6:75,130,950) increased (Fig. 4B).In the cornea, the editing level of methyltransferase like 7A (METTL7A:chr12:5,093,040) deceased after SARS-CoV-2 infection and tripartite motif containing 56 (TRIM56:chr7:101,091,531 and chr1:101,091,593) increased after SARS-CoV-2 infection (Fig. 4C).Additionally, both METTL7A and TRIM56 expressions significantly increased after SARS-CoV-2 infection.The editing level and mRNA expression level of METTL7A showed a negative correlation, whereas TRIM56 showed a positive correlation (Fig. S3A-S3D).SON (chr21:33,550,969) also showed a decreased editing level in the limbus and sclera upon infection (Fig. 4B,  D).Among these sites in the sclera, the most important is that apolipoprotein B mRNA editing enzyme catalytic subunit 3C (APOBEC3C), a member of the cytidine deaminase gene family related to C-to-U RNA editing, showed a decreased editing level (Fig. 4D), which was negatively correlated its up-regulated gene expression (Fig. S3E).

Temporal dynamics of A-to-I RNA editing in the retinal organoids during SARS-CoV-2 infection
A total of 1696 DRE sites in 799 DRE genes were identified in the retinal organoids transcriptome upon SARS-CoV-2 infection (Fig. S1).The retinal organoids infected were divided into 24 hpi and 96 hpi groups to investigate the possible role of A-to-I RNA editing during infection.More DRE sites were found at 24 hpi than 96 hpi, with 465 and 272 DRE sites found in 346 and 231 DRE genes at 24 hpi and 96 hpi, respectively (Fig. 5A).These DRE sites varied a lot from 24 to 96 hpi, with only 4 sites and 85 genes shared by 24 hpi and 96 hpi (Fig. 5A).The genes with four common DRE sites included intracisternal A particle-promoted polypeptide (IPP:chr1:45,699,736), myocardial infarction associated transcript (MIAT:chr22:26,672,672), pleckstrin homology Table 3 Sites correlated with ADAR and ADARB1 domain containing A5 (PLEKHA5:chr12:19,271,334), and caprin family member 2 (CAPRIN2:chr12:30,720,208).The editing level of IPP and PLEKHA5 decreased both at 24 hpi and 96 hpi (Fig. 5B).In contrast, the editing levels of MIAT and CAPRIN2 increased both at 24 hpi and 96 hpi compared to mock retinal organoids (Fig. 5B).
Our study analyzed RNA sequencing data from publicly available ocular tissue datasets, including the conjunctiva, limbus, cornea, sclera, and retinal organoids, to determine the potential role of A-to-I RNA editing in SARS-CoV-2 infection-related ocular diseases.The results showed significant changes in RNA editing profiles in various ocular tissues after COVID-19 infection.Specifically, the limbus, cornea, and sclera showed fewer RNA editing sites and genes when compared to the conjunctiva and retinal organoids (Fig. 1D, E), possibly due to the lower sequencing depth of the data in the tissues.
Although emerging studies have reported ocular symptoms in COVID-19 patients, such as conjunctival congestion, blurred vision, and foreign body sensation [24,59], it remains unclear how these symptoms are caused by SARS-CoV-2.Our results found that infection with SARS-CoV-2 in ocular tissues substantially altered RNA editing, suggesting a potential role of RNA editing in the ocular infection of SARS-CoV-2 and its ocular manifestations.
Most of these ocular RNA editing sites were within Alu repetitive elements (Fig. 1H).Alu elements are abundant short interspersed nuclear elements in the human genome and have been shown to play a critical role in regulating gene expression and alternative splicing [60].In addition, Alu elements are also involved in RNA editing as they were hot spots recognized by ADAR enzymes.Our findings suggested that these repetitive elements may play a key role in RNA editing, which contributes to the relationship between RNA editing and ocular SARS-CoV-2 infection.
Common DRE sites were observed in two or more tissues of the conjunctiva, limbus, cornea, sclera, and retinal organoids (Fig. 2C, E).For example, DRE in LAMP2 was shared among the conjunctiva, limbus, and sclera might suggest the involvement of lysosome-related functions in these tissues during SARS-CoV-2 infection.Importantly, our study also highlighted MAVS and MDM2 genes with common DRE sites across different  [61], and the inhibition of MDM2 might protect the eye from SARS-CoV-2 infections [62].
Our study also found RNA editing sites significantly correlated with gene expression, suggesting possible cis-regulating of gene expression in ocular tissues after SARS-CoV-2 infection.In addition, DRE varied among ocular tissues, particularly the retinal organoids, and conjunctiva, in which SARS-CoV-2 infections might affect the visual and immune functions [63,64].In addition, each tissue type had its unique DRE genes, especially those related to viral infections, suggesting a critical yet divergent role of A-to-I RNA editing in the immune response to ocular viral infections due to functional differences among conjunctiva, cornea, and retinal organoids (Fig. 3D).
In addition, we also found that APOBEC3C (Fig. 4D), a catalytic subunit of lipoprotein B mRNA editor associated with C-to-U RNA editing, had I-to-I RNA editing level decreased with its gene expression significantly increased.Such a finding might indicate that apart from A-to-I RNA editing, C-to-U RNA editing was also involved in ocular SARS-CoV-2 infections.Studies have shown that APOBEC3C is associated with infections of RNA viruses such as hepatitis and HIV [65,66].This finding thus implies the involvement of Fig. 5 Temporal dynamics of SARS-CoV-2 infection-associated A-to-I RNA editing in the retinal organoids.A 4 sites and 85 genes were shared by 24 hpi and 96 hpi groups, respectively.B The editing level difference of four shared sites between 24 and 96 hpi groups multiple RNA editing types during SARS-CoV-2 infection, which needs to be explored in further studies.
The study has observed dramatic changes in A-to-I RNA editing in ocular tissues upon SARS-CoV-2 infection.The biological impacts of such epigenetic changes on the eye remain largely undermined and are complex and context-dependent.The effects of such A-to-I RNA editing changes might be beneficial or adverse, depending on specific gene changes, which needs further investigation.Nevertheless, as ADARs and A-to-I RNA editing are considered important antiviral components in mammalian cells, such RNA editing response to SARS-CoV-2 infection is likely to contribute to the cellular antiviral process.For example, the MAVS gene showed the most evident changes in RNA editing across different tissues upon the SARS-CoV-2 infection.Interestingly, MAVS also showed significant expression changes.MAVS encodes a crucial intermediary protein in the virus-induced beta interferon signaling pathways and is essential for activating transcription factors that control the expression of beta interferon and play a role in the innate immune response against viruses [67][68][69].Upon SARS-CoV-2 infection, MAVS could probably be activated and contribute to the production of interferons, which are key antiviral chemokines that limit viral replication and spread [70,71].However, abnormal A-to-I RNA editing has also been recently implicated in diseases, especially immune-related pathogenesis.Persistent SARS-CoV-2 infection might lead to overactivated, A-to-I RNA editing, which could be adverse to the eye.Further research is needed to understand further the consequences of A-I RNA editing in the eye and to determine whether it is ultimately beneficial or adverse.
While providing valuable insights into the potential role of A-to-I RNA editing in ocular manifestations during SARS-CoV-2 infection, our current study has some limitations.Firstly, our study was based on bioinformatics analysis of publicly available datasets, and further experimental validation is needed to confirm the findings.Further functional studies could be important for future perspectives to validate and investigate the functional consequences of the RNA editing changes identified in the current study, which could provide a deeper understanding of the underlying mechanisms.Secondly, the current study focused on A-to-I RNA editing, whereas other types of RNA editing, such as C-to-U editing, which may also play a role in ocular SARS-CoV-2 infections, as well as the interplay between these different types of RNA editing, need to be investigated in further study.
In conclusion, our findings revealed substantial changes in A-to-I RNA editing in ocular tissues upon SARS-CoV-2 infection and provided new insights into understanding ocular manifestations of COVID-19.

Fig. 1 AFig. 2
Fig.1A-to-I RNA editing was identified from transcriptomes of the five ocular tissues in the current study.A The average A-to-I RNA editing level of the conjunctiva, limbus, cornea, sclera, and retinal organoids between mock and SARS-CoV-2 infection.B-C The fold change in the expression level of ADAR and ADARB1 genes, (D-E) the number of A-to-I RNA editing sites and genes, and (F) the functional types of variants resulting from A-to-I RNA editing in mock and SARS-CoV-2 infection.G SIFT prediction of the missense RNA editing variants.H RNA editing sites distribution in repetitive elements.A two-tailed unpaired Student's t-test was used to analyze the significance; *P < 0.05; **P < 0.01; ***P < 0.001; ns, no significance

Fig. 3
Fig. 3 Functional enrichment analysis of SARS-CoV-2 infection-associated A-to-I RNA editing in the five ocular tissues.A-D The enrichment analysis results of biological processes (A), molecular functions (B), cellular components (C), and KEGG pathway (D) enriched by genes with DRE are shown

Fig. 4
Fig. 4 SARS-CoV-2 infection-associated A-to-I RNA editing in ocular surface tissues.The main DRE sites (no more than 50, ranked by P-values) in the conjunctiva (A), limbus (B), cornea (C), and sclera (D) are shown

Table 1
Details of the datasets analyzed in the current

Table 2
DRE sites in MDM2 and MAVS