RNA modifications are post-transcriptional chemical alterations of RNA nucleobases that play crucial roles in gene regulation in mammals. Major RNA modifications such as N ⁶-methyladenosine (m6A), 5-methylcytidine (5-mC), pseudouridine (Ψ), Adenosine-to-Inosine (A-to-I) editing, dynamically regulate RNA metabolism, including splicing, nuclear export, stability, translation, and degradation (Cui L. et al., 2022). Among these, A-to-I editing is a unidirectional RNA modification catalyzed by Adenosine Deaminase Acting on RNA (ADAR), which specifically convert adenosine-to-inosine within double-stranded RNA (dsRNA) regions (Wang et al., 2017). In humans, the ADAR family comprises three members: ADAR1, ADAR2, and the catalytically inactive ADAR3. While ADAR2 is predominantly expressed in the central nervous system, ADAR1 is broadly expressed across various tissues. ADAR1 exists in two major isoforms generated through alternative splicing: the constitutively expressed nuclear isoform ADAR1 p110, and the interferon-inducible cytoplasmic isoform ADAR1 p150, which contains a Z-DNA-binding domain (George and Samuel, 1999; Sun et al., 2021). Notably, ADAR1 p150 prevents inappropriate activation of innate immune signaling pathways by editing cytoplasmic dsRNAs (Liddicoat et al., 2015; Chung et al., 2018; Nakahama et al., 2021). In addition to its enzymatic activity, ADAR1 also functions as an RNA-binding scaffold protein independent of A-to-I-editing. For instance, it directly interacts with Dicer to facilitate pre-miRNA processing and enhance miRNA loading into the RNA-induced silencing complex, highlighting its role in RNA interference (Ota et al., 2013). Furthermore, ADAR1 p110 has been shown to stabilize antiapoptotic mRNAs following UV irradiation by inhibiting Staufen1-mediated mRNA decay, thereby supporting cell survival (Sakurai et al., 2017).

Advancements in next-generation sequencing technologies have enabled transcriptome-wide identification of A-to-I editing sites (Mansi et al., 2021; Picardi et al., 2017). The functional impact of A-to-I editing varies depending on the location of the modification within the mRNA (Yuan et al., 2023). A significant proportion of editing occurs in the 3′ untranslated regions (3′UTRs) and introns, where dsRNA structures formed by Alu repeat elements serve as editing substrates (Ouyang et al., 2018; Qiu et al., 2016; Athanasiadis et al., 2004). Intronic editing can influence alternative splicing by creating or abolishing splice sites, as observed in genes such as NARF and CDDC15 (Lev-Maor et al., 2007; Tang et al., 2020). Editing within 3′UTRs can disrupt miRNA binding, thereby modulating mRNA stability and translational efficiency. For example, ADAR1-mediated editing of DHFR and METTL3 mRNAs has been linked to increased expression of these genes and tumor progression in breast cancer cells (Nakano et al., 2017; Li et al., 2022). A-to-I editing can also occur within coding regions, leading to amino acid substitutions that may alter protein function. ADAR2-mediated editing of the AMPA receptor subunit GluA2 exemplifies this, as it controls calcium permeability and suppresses excitotoxicity in neurons (Sommer et al., 1991). Additionally, A-to-I editing of lncRNAs can affect RNA secondary structure and miRNA binding, thereby regulating gene expression. In the case of NEAT1, such modifications enhance RNA folding stability and prevent competition with miR-9-5p, ultimately modulating JAK-STAT signaling and interferon-responsive genes expression (Chattopadhyay et al., 2024).

Recent studies have proposed that A-to-I editing plays a regulatory role in the DNA damage response and DNA repair pathways. For example, ADAR1-mediated editing within the exon of NEIL1 mRNA, which encodes a DNA glycosylase, results in a K242R amino acid substitution. The edited NEIL1 protein exhibits diminished capacity to recognize and excise damaged bases (Yeo et al., 2021; Lotsof et al., 2022). A-to-I editing has also been reported in the 3′UTRs of BRCA2, ATM, and POLH mRNAs, where increased editing correlates with transcript expression levels (Sagredo et al., 2018). In the case of BRCA2, editing prevents miRNA binding, thereby stabilizing the mRNA and enhancing protein levels, contributing to cisplatin resistance (Liu et al., 2024). These findings highlight the need for systematic investigation into the interplay between A-to-I editing and genome maintenance, particularly through factors involved in DNA repair and the DNA damage responses. To address this, we aimed to comprehensively identify inosine-modified transcripts associated with DNA repair pathways.

To explore the potential crosstalk between A-to-I RNA editing and DNA repair factor transcripts, we focused on the human lymphoblastoid cell line TK6, which is extensively used in genome stability and genotoxicity research. TK6 cells retain functional p53 and exhibit a stable karyotype, making them a reliable model for studying the relationship between RNA editing and genome maintenance. We employed the EpiPlex RNA assay (Sendinc et al., 2024), a technique that enables epitranscriptome-wide mapping of both m⁶A and inosine modifications in cellular RNAs by enriching for modifications in fragmented RNA, to perform a comprehensive analysis of RNA modification dynamics. The resulting epitranscriptomic mapping revealed significant enrichment of A-to-I editing within DNA repair-related pathways. Leveraging this dataset, we further investigated the isoform-specific roles of ADAR1 by comparing inosine modifications in DNA repair transcripts under conditions deficient in either ADAR1 p110 or p150. Notably, the nucleotide excision repair factor XPA was found to undergo splicing regulation via intronic inosines. This study provides suggests a possible link between RNA inosine modification and DNA repair processes.

Recent studies have revealed that A-to-I RNA editing plays a regulatory role in DNA repair mechanisms. For example, in the human breast carcinoma cell line ZR-75-1, A-to-I editing of ATM and POLH transcripts is enhanced, accompanied by elevated expression levels of these genes, as shown by RNA-seq analysis (Sagredo et al., 2018). In human U2OS cells, depletion of ADAR2 has been reported to increase γH2AX levels and the frequency of micronucleus formation, indicating the accumulation of DNA damage (Jimeno et al., 2021). Moreover, ADAR1-mediated A-to-I editing influences cellular responses to DNA-damaging agents: knockdown of ADAR1 significantly reduces cell viability following cisplatin treatment in cholangiocarcinoma HuCCT1 and RBE cell lines (Liu et al., 2024). This phenotype can be rescued by reintroducing wild-type ADAR1, but not by a catalytically inactive mutant (ADAR1 E912A), suggesting that A-to-I RNA editing activity is essential for cisplatin resistance. Thus, A-to-I modification likely plays a critical role in genome stability, and elucidating its broader functions may contribute to novel therapeutic strategies in cancer. Therefore, this study’s analysis of the relationship between A-to-I RNA editing and DNA repair pathways provides insight that may help to understand the physiological relevance of inosine modifications in genome maintenance.

In this study, the EpiPlex assay, a recently developed for epitranscriptome analysis, was applied to the human lymphoblastoid TK6 cell line, identifying 870 A-to-I-modified transcripts across two independent biological replicates (Supplementary Table S2). WT cells contained on average 68,801 A-to-I editing sites. EpiPlex peaks called in WT cells were also validated using the REDIportal v3 database (http://srv00.recas.ba.infn.it/atlas/) (Picardi et al., 2017), containing on average 156,556 sites reported by the database (Supplementary Figure S1A). A-to-I density within EpiPlex peaks showed a wide distribution in WT and p150 KO cells, with a right-skewed tail representing peaks of particularly high A-to-I density, indicative of hyper-editing regions (Supplementary Figures S1B,C). The majority of editing sites were located in intronic regions, followed by 3′UTRs (Figure 1B). This distribution pattern aligns with the global annotation of RNA editing sites annotated in the RADAR database of human transcripts (Weirick et al., 2019; Ramaswami and Li, 2014). According to GO enrichment analysis, these edited transcripts were significantly enriched in pathways such as “chromatin remodeling” and “DNA repair” (Figure 1C; Supplementary Table S3). Notably, associations between specific biological pathways and RNA modifications appear to be cell type-dependent. For example, in brain tissues, A-to-I editing targets are enriched in transcriptional regulation and neurological disorder–related pathways (Sakurai et al., 2014), while in granulosa cells from patients with polycystic ovary syndrome, editing events are associated with the mitotic cell cycle and membrane transport (Kong et al., 2023). In 14 human cell lines from the ENCODE project, modified RNAs are enriched in cell division, antiviral response, and translational control pathways (Park et al., 2012). Collectively, these findings indicate that the functional significance of A-to-I RNA editing varies by tissues and cell type. Our analysis using TK6 cells revealed that the presence of Alu elements within DNA repair–related loci serves as a major target of A-to-I editing, consistent with previous observations that repetitive elements frequently harbor editing sites.

Notably, we identified A-to-I–modifications in mRNAs of 27 DNA repair-related genes, including previously reported targets ATM and POLH (Figure 1E) (Sagredo et al., 2018). Of modified sites identified in these targets, >95% were annotated as Alu elements by REDIportal (Supplementary Figure S1D). Furthermore, the A-to-I peak sequences of these transcripts shared conserved sequence motifs (Figure 1F), some of which showed homology to Alu subfamily elements such as AluYh7, AluYc, and AluSq4. This is consistent with earlier reports that A-to-I editing preferentially occurs within Alu elements (Sawyer et al., 2004; Godoy et al., 2012; Nishikura, 2016), suggesting that A-to-I editing in DNA repair transcripts relies on double-stranded RNA structures formed by Alu-derived sequences. In addition to the DNA repair pathway, GO analysis indicated enrichment in “chromatin remodeling” (Figure 1C), including genes encoding topoisomerase I (TOP1) and histone-modifying enzymes, (KDM2A, KDM4B, NSD1, and NSD2). These enzymes are known to contribute to the regulation of DNA damage responses (Iwasaki et al., 2024; Pommier et al., 2006; Khoury-Haddad et al., 2015; Wagner et al., 2025), implying that A-to-I editing may support genome integrity through broader chromatin-based repair mechanisms.

Using the EpiPlex assay to compare wild-type, p150 KO, and p150/p110 KO cell lines, we assessed isoform-specific contributions of ADAR1 to A-to-I editing. Inosine peak levels were reduced by approximately 73.4% in p150 KO cells and by 99.9% in p150/p110 KO cells relative to wild-type cells. We note that p150 uniquely harbors a Zα binding domain (Figure 2A), which enables recognition of Z-DNA and Z-RNA and contributes to its broader editing activity compared with p110. Previous reports have demonstrated that this domain of p150 plays an important role in A-to-I editing activity and accounts for the majority of editing events in cells (Nakahama et al., 2021), which may explain the predominant contribution of p150 to RNA editing observed in our study. In contrast, the number of m6A peaks remained unchanged, suggesting that A-to-I and m6A modifications are independent processes (Figure 3A). Cluster analysis of A-to-I editing revealed 839 genes as strongly p150-dependent, 303 as moderately p150-dependent, and 176 as p150-independent. Additionally, a distinct subset of genes was edited specifically by the p110 isoform (Figure 3B). This contrasts with a previous report in HEK293 cells where no p110-specific editing was observed (Sun et al., 2021), suggesting that ADAR1 editing targets vary by cell type and reflect isoform-specific localization and activity. Consistently, A-to-I editing has been shown to occur in a cell type–specific manner in human brain cells (Cuddleston et al., 2022). Moreover, under stress conditions, p110 undergoes phosphorylation, promoting transient cytoplasmic translocation (Sakurai et al., 2017). Among the DNA repair factors identified in this study, isoform dependency appeared to vary across individual genes, implying potential gene-specific contributions of p150 and p110 to A-to-I editing.

Regarding the effects of ADAR1 deficiency on cell viability, our study demonstrated that TK6 cells lacking ADAR1 remain viable, in contrast to previous reports where ADAR1 loss induced cell death in HeLa cells. In ALT (alternative lengthening of telomeres)-negative cells like HeLa, ADAR1p110 is required for telomerase-dependent telomere maintenance by editing RNA:DNA hybrids and resolving telomeric R-loops (Shiromoto et al., 2021): Loss of ADAR1 in these cells destabilizes telomeres, leading to lethal proliferative arrest. In contrast, ALT-positive cells, characterized by heterogeneous telomere lengths, maintain telomere through telomerase-independent mechanisms (Loe et al., 2020). The viable phenotype of TK6 cells lacking ADAR1 may suggest the presence of alternative telomere maintenance pathways compensating for the loss of inosine-mediated stability. Supporting this idea, a previous study reported that telomerase does not contribute to telomere elongation following ionizing radiation in TK6 cells (Berardinelli et al., 2012).

Gene expression profiling of ADAR1-deficient cells revealed upregulation of interferon-stimulated genes in both p150 KO and p150/p110 KO cells compared to WT cells (Supplementary Figure S3), consistent with studies showing that the loss of A-to-I editing leads to accumulation of endogenous double-stranded RNAs and activation of type I interferon responses (Chung et al., 2018; Kuster et al., 2015; Bardon et al., 2016). RT-qPCR analysis revealed that ATM mRNA levels were downregulated in p150/p110 KO cells (Figure 4A), consistent with previous findings that ADAR1 knockdown breast cancer cells reduces ATM expression without affecting mRNA stability in ZR-75-1 (Sagredo et al., 2018). In contrast, expression levels of ATR and FANCA remained unchanged (Figure 4A), indicating gene and locus-specific effects of A-to-I editing on transcript expression. It should be noted that no DNA repair-related genes shown in Figure 1D were detected as DEGs in RNA-seq, indicating that the limited sensitivity of the RNA-seq analysis in quantifying subtle changes in the transcript abundance. It is also plausible that A-to-I editing affects mRNA stability. More specialized approaches, such as RNA metabolic labeling or transcriptome-wide decay profiling, for example, RATE-seq (Neymotin et al., 2014), will be required in future studies to determine whether and how A-to-I editing affects transcript stability. Notably, our RNA-seq analysis revealed that PARP1, a DNA repair factor, was upregulated in ADAR1-deficient cells (Supplementary Figure S3A). Since PARP1 was not identified as an inosine-modified transcript in our dataset, this upregulation is unlikely to be a direct consequence of A-to-I editing, but rather represents an indirect effect mediated by DDR-associated stress responses (Pathikonda et al., 2020).

Interestingly, increased expression of novel splicing variants was observed in both p150 KO and p150/p110 KO cells relative to WT (Figures 4B,C). These alternative splicing events correspond to the pathogenic non-coding transcript variants NR_149093.2 and NR_149094.2 (Tanioka et al., 2005). It is plausible that A-to-I editing within central intronic region of XPA alters splicing factors binding specificity, thereby reshaping splicing patterns and promoting the production of alternative variants. Supporting this possibility, intronic A-to-I editing has been implicated in splicing regulation at other gene loci. For example, SRSF1 binding to an A-to-I site promotes exonization in the HNRPLL gene (Chen et al., 2018), while SRSF7 binding enhances exon inclusion in CDDC15 (Tang et al., 2020). In the RELL2 gene, ADAR2 binds to a double-stranded RNA structure near the exon, interfering with U2AF65 binding and inducing exon skipping (Tang et al., 2020). These findings raise the possibility that ADAR1 directly regulates splicing patterns in XPA through similar mechanisms. At the protein level, notably, the intact XPA protein was expressed physiologically relevant levels in WT, p150, and p150/p110 cells (Figure 4D). This may implicate that the alternative splicing variant is not efficiently translated into detectable protein products and could be subject to post-transcriptional degradation, such as through nonsense-mediated mRNA decay (Lykke-Andersen and Jensen, 2015). On the other hand, markedly elevated levels of γH2AX were detected in p150 KO and p150/p110 KO cells compared to WT cells (Figure 4D), indicating altered DDR signaling linked to the absence of ADAR1 function.

To further dissect the DNA damage-specific phenotypes caused by ADAR1 deficiency, we assessed cell survival following exposure to different DNA-damaging agents. Strikingly, both p150 KO and p150/p110 KO cells exhibited a clear reduction in cell survival compared with WT cells after cisplatin treatment (Figure 5A). In contrast, sensitivity to CPT was limited: p150/p110 KO cells showed a significant decrease only at a single dose (1 nM), while p150 KO cells exhibited a modest but statistically not significant decrease compared to WT cells (Figure 5B). Interestingly, ADAR1-deficient cells exhibited no reduction in survival relative to wild-type cells after UVC irradiation (Figure 5C). These distinct patterns of sensitivity likely reflect the types of DNA lesions induced by each agent. Cisplatin generates mono-adducts and intra-strand crosslinks, which are repaired mainly by nucleotide excision repair and related pathways, and to a lesser extent interstrand crosslinks that are processed by the interstrand crosslink repair machinery (Ahmed et al., 2024). The accumulation of such lesions is expected to eventually give rise to DSBs, requiring DSB repair pathways. CPT, on the other hand, induces DNA breaks specifically during DNA replication via formation of DNA-protein crosslinks. UVC irradiation generates UV-induced dimers, which are repaired by nucleotide excision repair. Notably, a key factor in the UV tolerance is translesion DNA synthesis by Pol η, which can bypass UV-induced dimers with remarkable efficiency and high fidelity (Masutani et al., 1999). The preserved resistance to UVC observed in ADAR1-deficient cells is therefore likely explained by the activity of such factors, which may compensate for subtle post-transcriptional effects on DNA repair–related RNAs. A-to-I RNA editing may broadly influence a wide range of DNA repair pathways, and the combined effects of these alterations are likely reflected in the survival outcomes following exposure to DNA-damaging agents.

In line with our results, previous studies have shown that ADAR1 is crucial for cellular responses to various genotoxic stresses. ADAR1 promotes cell survival following cisplatin treatment through A-to-I editing of the BRCA2 transcript (Liu et al., 2024). Another recent study found that elevated ADAR1 expression confers chemoresistance to 5-fluorouracil and cisplatin by editing the 3′UTR of the stearoyl-CoA desaturase gene transcript (Wong et al., 2023). ADAR1 also exhibits anti-apoptotic effects in response to UV-induced stress (Sakurai et al., 2017). Under CPT treatment, ADAR1 p110 functions non-catalytically by accumulating at R-loops and recruiting DHX9 and DDX21, thereby enhancing ATR activation (Cui H. et al., 2022; Zhang et al., 2023). Our analysis extends this understanding by demonstrating that ADAR1 broadly edits transcripts encoding genome maintenance factors. These observations suggest that the diverse responses to genotoxic agents under ADAR1 deficiency may result from a combination of impaired A-to-I editing across multiple DNA repair-related transcripts and disruption of ADAR1’s non-catalytic roles, such as R-loop resolution. These findings suggest a potential multifaceted role of ADAR1 in genome stability. However, the precise contribution of each isoform still remains unclear. Further investigations, including isoform-specific functional analyses, will be necessary to clarify how ADAR1 regulates genome stability, particularly in relation to transcriptomic remodeling and RNA–protein interactions in DNA repair pathways.