Inhibiting ADAR1‐Mediated Excessive Epigenetic A‐to‐I RNA Editing Improves the Immune Microenvironment and Increases Sensitivity to Immunotherapy in Lung Adenocarcinoma
Sihui Wang, Sufei Zheng, Chengming Liu, Chaoqi Zhang, Xinfeng Wang, Zhanyu Wang, Yan Wang, Xiaoli Feng, Qi Xue, Nan Sun, Jie He
TL;DR
Inhibiting ADAR1 in lung adenocarcinoma improves the tumor immune environment and makes immunotherapy more effective.
Contribution
This study shows ADAR1 inhibition enhances immunotherapy by improving immune cell infiltration and activating antitumor pathways.
Findings
ADAR1 suppression increases PD-L1 and CD8+ T-cell infiltration in tumors.
ADAR1 deficiency leads to dsRNA accumulation, activating RIG-I, MAVS, and IFN-β signaling.
Reduced ADAR1 expression correlates with better immunotherapy responses in LUAD patients.
Abstract
Adenosine‐to‐inosine (A‐to‐I) RNA editing, predominantly catalyzed by the enzyme adenosine deaminase acting on RNA 1 (ADAR1), has attracted interest due to its essential functions in regulating immune response and cancer progression. This research investigates ADAR1 inhibition as a promising strategy aimed at improving immunotherapy efficacy in lung adenocarcinoma (LUAD) and explores the underlying mechanisms. Findings from murine models demonstrate that ADAR1 suppression within tumors notably improves the immune microenvironment, marked by increased PD‐L1 expression and enhanced CD8+ T‐cell infiltration, as well as elevated levels of CXCL9, CXCL10, and CXCL11. These changes promote antitumor T‐cell immune responses and amplify the effects of immunotherapy. Mechanistic investigations further reveal that deficiency in ADAR1 leads to an increase in double‐stranded RNA (dsRNA), which…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6- —National Natural Science Foundation of China
- —the Beijing Natural Science Foundation
- —the Fundamental Research Funds for the Central Universities
- —the National High Level Hospital Clinical Research Funding and Beijing Hope Run Special Fund of Cancer Foundation of China
- —the CAMS Innovation Fund for Medical Sciences
- —the Beijing Municipal Science & Technology Commission
- —the National High Level Hospital Clinical Research Funding
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsRNA regulation and disease · interferon and immune responses · Chromatin Remodeling and Cancer
Introduction
1
Lung adenocarcinoma (LUAD), the predominant histological type of lung cancer, continues to be the foremost cause of cancer‐related mortality globally, impacting both genders [1, 2, 3]. While prognosis for patients has risen significantly since the advent of molecule‐targeted therapies, obstacles involving treatment resistance and undesirable side effects persist even though targeted epidermal growth factor receptor (EGFR) drugs have been developed into a third generation [4, 5]. Recently, persistent clinical efficacy has been observed in certain patients undergoing immunotherapy with immune checkpoint inhibitors (ICIs). The medication achieves its therapeutic effects by obstructing inhibitory signals and activating cytotoxic T cells, therefore positioning targeted therapy and immunotherapy as the foundation of first‐line treatment for LUAD [6, 7, 8, 9, 10, 11, 12]. However, as with targeted therapy, primary resistance to immunotherapy continues to be a major obstacle, leading to suboptimal responses [13, 14]. Consequently, further investigation into the mechanisms of resistance to immunotherapy and the establishment of effective treatments for disease management are urgently required [14, 15, 16].
RNA editing occurs as a form of post‐transcriptional process that alters certain RNA bases without changing the corresponding DNA sequence [17]. This process involves insertion, deletion, or substitution, with the most common type being adenosine‐to‐inosine (A‐to‐I) editing. The enzyme adenosine deaminase acting on RNA (ADAR) catalyzes this activity, which mostly uses double‐stranded RNA (dsRNA) as substrates. Inosine is interpreted as guanosine (G) during translation, which might modify the amino acid sequence and, as a result, the protein's structure and function [18]. The three primary members of the ADAR family are ADAR1, ADAR2, and ADAR3, with ADAR1 being the primary driver of A‐to‐I editing and the most highly expressed in humans [19]. There is growing evidence that ADAR1‐mediated A‐to‐I RNA editing plays a part in immunological modulation and the development of cancer [20, 21]. According to recent research, tumors lacking ADAR1 exhibit noticeably higher sensitivity to ICIs treatment in mouse melanoma models, accompanied by increased CD8^+^ T‐cell infiltration [22]. Results indicate that RNA editing is crucial for establishing an inflammatory tumor microenvironment (TME) that facilitates immune cell infiltration. While a high density of CD8^+^ tumor‐infiltrating lymphocytes (TILs) in the TME is frequently linked to better immunotherapy outcomes, the suppression of the TME, linked to poor immunotherapy outcomes, is not solely characterized by a lack of immune cells [23, 24, 25].
A classic four‐group classification system, based on degree of infiltration of CD8^+^ T cells and expression level of PD‐L1, was developed to predict patient response to ICIs [26, 27]. This classification system comprises four subtypes, characterized respectively by TIL^+^/PD‐L1^+^ (type I), TIL^−^/PD‐L1^−^ (type II), TIL^−^/PD‐L1^+^ (type III), and TIL^+^/PD‐L1^−^ (type IV). Previous research has revealed that type II TME, defined by immunological ignorance, is most prevalent in LUAD patients and associated with a poor prognosis [28]. Converting cold immune‐rejecting tumors into hot immune‐responsive tumors is a crucial approach to conquering resistance to immunotherapy [29]. High‐throughput sequencing research has demonstrated an overabundance of A‐to‐I RNA editing level in LUAD compared with the corresponding normal tissue [30, 31], and ADAR1 has been found to be overexpressed in LUAD, promoting tumorigenesis and cellular lethality through PKR activation [32, 33, 34]. Given that the TME is regulated not only by genetic factors but also by epigenetic alterations, and the significance of ADAR1 in melanoma regulation of immunotherapy resistance [22, 35, 36], this study focuses on RNA editing mediated by ADAR1 to explore the mechanisms underlying immunological ignorance and provide new insights into predicting and improving immunotherapy responses in LUAD.
During this investigation, A‐to‐I RNA editing levels were evaluated in LUAD, then ADAR1 catalytic activity was analyzed using public datasets and validated in LUAD cell lines. A mouse model was developed to examine the role of ADAR1 in LUAD immunotherapy; subsequently, CD8^+^ TILs, PD‐L1 expression, and cytokine profiles within TME were examined to illuminate the underlying mechanisms. Additionally, dsRNA, the substrate of RNA editing, as well as its downstream signaling pathways, was investigated to elucidate the mechanism of ADAR1‐induced immune resistance. The clinical prognostic value of ADAR1 expression was also explored in LUAD patients undergoing immunotherapy.
Results
2
Over A‐to‐I Editing Levels in LUAD Were Positively Correlated With ADAR1 Expression
2.1
To investigate changes in A‐to‐I editing levels in LUAD compared to normal tissues, data from The Cancer Editome Atlas (TCEA) were extracted. The findings demonstrated that LUAD had more editing sites per sample than normal tissues (Figure 1A) [37]. To determine whether ADAR1 or ADAR2 contributed to the overediting in LUAD, the Alu editing index (AEI), an indicator of the RNA editing level, along with expression levels of ADAR1 and ADAR2 in 92 LUAD cell samples, was extracted from the REDIportal database [38, 39]. The analysis revealed that AEI positively correlated with levels of ADAR1 (Figure 1B) but not with ADAR2 (Figure 1C). mRNA expression data from the Gene Expression Omnibus (GEO) showed that ADAR1 expression was higher in LUAD tissues than normal tissues, whereas ADAR2 expression was lower in tumor samples (Figure 1D,E). IHC staining conducted on 101 LUAD patients from the CICAMS cohort confirmed these findings, showing higher ADAR1 expression in cancer tissues relative to paracancerous and normal tissues, whereas ADAR2 expression did not follow the same pattern (Figure 1F,G). Representative IHC images are shown in Figure 1H. Further validation was performed in our lab using BEAS‐2B cells (normal human bronchial epithelium) and six LUAD cell lines (A549, H1975, A427, H358, H322, and H1299). The levels of A‐to‐I editing in the cell line were characterized by the editing level at representative sites, which were assessed via Sanger sequencing; simultaneously, ADAR1 expression levels were determined by qRT‐PCR. LUAD cells demonstrate higher A‐to‐I editing levels than BEAS‐2B cells as shown in Figure S1A–H, and these levels correlated with ADAR1 expression (Figure 1I). To confirm ADAR1's role in RNA editing, ADAR1‐knockdown LUAD cell lines were established (Figure 1J,K). The A‐to‐I editing levels were significantly decreased in these ADAR1‐knockdown cell lines (Figure 1L). To verify whether ADAR's role in RNA editing is dependent on enzymatic catalysis, rescue experiments were performed. In ADAR1‐knockdown LUAD cells, we reintroduced either wild‐type ADAR1 or an enzymatically inactive ADAR1 mutant (Figure S1I,K). Sanger sequencing revealed that, compared to control cells, A‐to‐I RNA editing levels were significantly restored after reintroduction of WT ADAR1 in ADAR1‐knockdown cells. However, overexpression of the enzyme‐inactive ADAR1 mutant failed to rescue editing levels, which remained significantly lower (Figure S1J,L). The results suggest that the elevated A‐to‐I editing in LUAD is predominantly attributed to ADAR1.
*A‐to‐I editing levels were high in LUAD and correlated with ADAR1 expression. (A) TCEA data showing the number of edited sites per sample in LUAD and normal tissues. (B) Correlation between ADAR1 expression and AEI in 92 LUAD cell lines, analyzed using data from REDIportal. (C) Correlation between ADAR2 expression and AEI in 92 LUAD cell lines, analyzed using data from REDIportal. (D) ADAR1 mRNA expression in LUAD and normal tissues from GEO(GSE10072). (E) ADAR2 mRNA expression in LUAD and normal tissues from GEO(GSE10072). (F) IHC analysis of ADAR1 expression in LUAD, paracancerous, and normal tissues in the CICAMS cohort. (G) IHC analysis of ADAR2 expression in LUAD, paracancerous, and normal tissues in the CICAMS cohort. (H) Representative images of ADAR1 and ADAR2 expression in LUAD tissues. (I) Correlation between ADAR1 mRNA expression and A‐to‐I editing levels in BEAS‐2B cells and six LUAD cell lines. Sanger sequencing was used to assess A‐to‐I RNA editing levels at three typical sites (AZIN1: chr8:103841636, COG3: chr13:46090371, GIPC1: chr19:14593693) in cell lines. (J) Western blotting analysis of ADAR1‐p110 and ADAR1‐p150 isoforms in ADAR1‐knockdown LUAD and control cells. sh2, sh3, and con refer to ADAR1‐knockdown LUAD cells constructed using shRNA‐2, ADAR1‐knockdown LUAD cells constructed using shRNA‐3, and control LUAD cells, respectively. (K) qRT‐PCR analysis of ADAR1 mRNA expression in ADAR1‐knockdown LUAD and control cells. (L) Sanger sequencing to detect A‐to‐I RNA editing levels at three typical sites mentioned above in ADAR1‐knockdown LUAD and control cells (cor, Pearson correlation coefficient; *p < 0.05; **p < 0.01; ***p < 0.001; ***p < 0.0001; ns, not significant; t‐test; error bars represent standard deviation).
Inhibiting ADAR1 Enhanced the Sensitivity to Immunotherapy in LUAD
2.2
To investigate the role of ADAR1 in the response of LUAD to immunotherapy, an ADAR1‐knockdown mouse model was developed (Figure 2A) by generating ADAR1‐knockdown LLC cells (Figure S2A), subcutaneously transplanting them into C57BL/6J mice, and confirming the successful establishment of the immunocompetent model through IHC staining of the resulting subcutaneous tumors (Figure S2B,C). Subsequently, four groups were formed, and mice from each group were administered either anti‐PD‐L1 antibody or the control isotype. Upon treatment with the anti‐PD‐L1 monoclonal antibody, tumor growth was noticeably suppressed in the group with reduced ADAR1 compared to the control group (Figure 2B,C). CD8^+^ TILs and PD‐L1 levels in the TME across the four groups were subsequently assessed (Figure 2D–F). The results showed a marked reduction in ADAR1 expression in the knockdown group. Obviously, the group treated with both ADAR1 knockdown and anti‐PD‐L1 monoclonal antibody exhibited an increased presence of CD8^+^ TILs (Figure 2E) and elevated PD‐L1 level (Figure 2F). Since type‐I IFN signaling is activated by ADAR1‐mediated A‐to‐I RNA editing, with downstream chemokines potentially contributing to TILs recruitment [40, 41, 42, 43], the levels of IFN signaling, encompassing IFN‐β and chemokines (CXCL9, CXCL10, and CXCL11), were quantified in tumor lysates. Under the influence of anti‐PD‐L1 monoclonal antibody IFN‐β, the levels of CXCL9, CXCL10, and CXCL11 were markedly elevated in the ADAR1‐knockdown group versus the control group (Figure 2G–K). These observations were additionally confirmed by IHC staining of tumor tissues, yielding similar results (Figure 2L–O).
*Inhibition of ADAR1 enhances sensitivity to immunotherapy in the LUAD mouse model. (A) Schematic diagram of tumor construction from LLC cells, transplantation, and treatment schedule in C57BL/6J mice. (B) Images of tumors derived from sh‐LLC and control‐LLC cells treated with PD‐L1 mAb and isotype control. The groups sh‐LLC and con‐LLC refer to ADAR1‐knockdown LLC cells prepared using shRNA and control LLC cells, respectively. n = 5 mice per group. (C) Tumor growth curves for the four groups. n = 5 mice per group. (D) IHC images of CD8 and PD‐L1 expression in the corresponding groups. (E, F) IHC analysis of CD8+ T‐cell infiltration in the corresponding groups. (G–K) ELISA analysis of IFN‐β, IFN‐γ, CXCL9, CXCL10, and CXCL11 concentrations in the corresponding groups. (L–O) IHC analysis of IFN‐β, CXCL9, CXCL10, and CXCL11 expression in the corresponding groups (*p < 0.05; **p < 0.01; ***p < 0.001; ***p < 0.0001; t‐test; error bars represent standard deviation).
High‐ADAR1 Expression Induced an Immunosuppressive TME in LUAD
2.3
To validate these results, IHC analysis of the TME was performed on 227 LUAD patients from the CICAMS cohort. Patients were categorized into low‐ and high‐ADAR1‐expression groups according to the median IHC score, with a cut‐off of 6. The low‐ADAR1‐expression group demonstrated elevated amounts of CD8^+^ TILs (Figure 3A) and increased PD‐L1 expression (Figure 3B). Upon classification into the aforementioned four subgroups, the low‐ADAR1‐expression group exhibited the highest prevalence of type I (TIL^+^/PD‐L1^+^) TME, signifying adaptive immune resistance [26]. Conversely, the high‐ADAR1‐expression group contained the highest proportion of type II (TIL^−^/PD‐L1^−^) TME, suggesting immunological ignorance (Figure 3C). The negative correlation between ADAR1 and IFN‐β, CXCL9, CXCL10, and CXCL11 in the mouse model was further validated in LUAD patients via IHC staining. Consistent with the mouse model, the concentrations of IFN‐β, CXCL9, CXCL10, and CXCL11 were elevated in the low‐ADAR1‐expression subgroup (Figure 3D–G). Representative IHC images are shown in Figure S3. Thereby, ADAR1 significantly contributes to the formation of an immunosuppressive TME. To investigate the function of A‐to‐I editing in TME modification, editing levels in LUAD patients were assessed. Four typical sites with high‐editing levels located in exons were selected from the REDIportal database. Additionally, 32 frozen surgical specimens and corresponding FFPE tissues were obtained to assess A‐to‐I RNA editing levels and examine TME characteristics. The findings revealed that the low‐editing group exhibited a markedly greater proportion of type I TME (TIL^+^/PD‐L1^+^), indicating that diminished A‐to‐I RNA editing correlates with a more immune‐activated TME that exhibits better responsiveness to immunotherapy (Figure 3H–J). The clinical significance of ADAR1 was further examined, revealing that low‐ADAR1 expression was a predictor of better survival (Figure 3K).
*ADAR1 suppressed the activation of TME and led to poor prognosis in LUAD. (A) IHC analysis of CD8+ TILs in tissue samples from 227 LUAD patients. ADAR1‐low and ADAR1‐high refer to the low‐ADAR1‐expression and high‐expression groups, respectively, based on the median IHC score. (B) IHC analysis of PD‐L1 in tissue samples from 227 LUAD patients. (C) Percentage of TME types in 227 LUAD patients. (D) IHC analysis of IFN‐β in tissue samples from 227 LUAD patients. (E) IHC analysis of CXCL9 in tissue samples from 227 LUAD patients. (F) IHC analysis of CXCL10 in tissue samples from 227 LUAD patients. (G) IHC analysis of CXCL11 in tissue samples from 227 LUAD patients. (H) CD8+ TILs in the low‐ and high‐editing‐level groups of 32 frozen surgical specimens. (I) PD‐L1 expression in the low‐ and high‐editing‐level groups of 32 frozen surgical specimens. (J) Correlation between ADAR1 expression and A‐to‐I editing level in 32 LUAD patients. (K) Survival analysis of 227 LUAD patients in low‐ and high‐ADAR1‐expression groups (***p < 0.001; ***p < 0.0001; chi‐square test; t‐test; error bars represent standard deviation).
Inhibiting ADAR1 Activated Type‐I IFN Signaling Triggered by dsRNA
2.4
To investigate the mechanism through which ADAR1 knockdown enhances type‐I IFN signaling activation in LUAD, dsRNA levels, the substrate of ADAR1, were evaluated in LUAD cell lines. The results showed higher dsRNA accumulation in ADAR1‐knockdown cells compared to controls, particularly in the cytoplasm (Figure 4A,B). The conclusions remain consistent in LLC models (Figure S2D,E). Earlier studies have identified dsRNA as a source of damage‐associated molecular patterns (DAMPs) that initiate innate immunity. Once recognition by pattern recognition receptors (PRRs) such as RIG‐I and MDA5 occurs, downstream molecules like TBK, IRF3, and IRF7 are activated, resulting in elevated expression of IFN‐stimulated genes (ISGs) [44, 45]. To prevent excessive autoimmune responses, RNA editing marks dsRNA as “self,” thereby hindering receptor recognition—a mechanism exploited by tumors to evade the immune system [46, 47, 48]. To determine if ADAR1 knockdown stimulates the RIG‐I/MDA5/MAVS pathway via dsRNA in immune regulation, Western blotting was utilized to evaluate the expression of dsRNA and their receptors. The outcomes revealed increased levels of RIG‐I, MDA5, MAVS, p‐TBK1, and p‐IRF3 (Figure 4C). In line with these findings, Western blotting showed upregulation of PD‐L1 and IFN‐β in ADAR1‐knockdown cells (Figure 4D). Subsequent qRT‐PCR and Western blot investigations revealed that the reduction of RIG‐I or MDA5 resulted in reduced IFN‐β levels, hence confirming the involvement of dsRNA in the activation of innate immunity (Figure 4E–J). The findings indicate that via the RIG‐I and MDA5 pathways, knockdown of ADAR1 stimulates the IFN‐β immune response by accumulating dsRNA.
*IFN‐β signaling induced by dsRNA was activated in ADAR1‐knockdown cells. (A) Immunofluorescence (IF) analysis of dsRNA in ADAR1‐knockdown A549 and H1975 cells and control cells, stained with DAPI (blue) and anti‐dsRNA antibody J2 (red). (B) Quantitative analysis of fluorescence intensity of dsRNA in ADAR1‐knockdown A549 and H1975 cells and control cells. (C) Western blotting analysis showing the expression of RIG‐I, MDA5, MAVS, TBK, p‐TBK, and p‐IRF3 in ADAR1‐knockdown LUAD cells and control cells. (D) Western blotting analysis of PD‐L1 and IFN‐β in ADAR1‐knockdown LUAD cells and control cells. (E) MDA5 expression in MDA5‐knockdown A549 cells, measured by qRT‐PCR. (F) IFN‐β expression in MDA5‐knockdown A549 cells, measured by qRT‐PCR. (G) IFN‐β expression in MDA5‐knockdown A549 cells, measured by Western blotting. (H) RIG‐I expression in RIG‐I‐knockdown A549 cells, measured by qRT‐PCR. (I) IFN‐β expression in RIG‐I‐knockdown A549 cells, measured by qRT‐PCR. (J) IFN‐β expression in RIG‐I‐knockdown A549 cells, measured by Western blotting (*p < 0.05; *p < 0.01, t‐test; error bars represent standard deviation n = 3 independent cell culture experiments).
IFN‐β‐Dependent Effect of ADAR1 in Remodeling Tumorigenesis and the TME of LUAD
2.5
After investigating the function of ADAR1 in triggering the IFN‐β response, the reverse effects of IFN‐β on tumor behavior were further examined by stimulating cells with IFN‐β in vitro. No differences in cell proliferation were observed between ADAR1‐knockdown and control cells in both A549 and H1975 cell lines (Figure 5A,B). However, ADAR1‐knockdown cells exhibited significantly reduced cell growth when treated with IFN‐β (Figure 5C,D). A colony formation experiment demonstrated analogous results, showing a significant reduction in the colony count of ADAR1‐knockdown cells in the presence of IFN‐β (Figure 5E,F). To further explore how ADAR1 influences cell growth, apoptosis was assessed, revealing a substantial rise in apoptosis in ADAR1‐knockdown cells stimulated with IFN‐β (Figures 5G,H and S4A,B). These results indicate that ADAR1‐deficient cells are more sensitive to extracellular IFN‐β. Notably, treatment with IFN‐β elevated the expression of PD‐L1 and ADAR1‐p150 in wild‐type LUAD cells, suggesting a feedback mechanism linking ADAR1 knockdown and activation of the IFN‐β immune response pathway (Figure 5I,J). Additionally, chemotaxis experiments showed increased infiltration of CD8^+^ TILs in ADAR1‐knockdown tumor cells upon IFN‐β stimulation (Figures 5K,L and S4C,D). Thus, ADAR1 knockdown, through IFN‐β pathway activation, not only promotes tumorigenesis but also upregulates the level of PD‐L1 and enhances CD8^+^ T‐cell infiltration.
*ADAR1‐knockdown cells are more sensitive to IFN‐β treatment. (A, B) Proliferation curves of ADAR1‐sh A549, H1975 cell lines, and respective control cells (CCK‐8 assay). (C, D) Proliferation curves of ADAR1‐sh A549, H1975 cell lines, and respective control cells treated with IFN‐β (CCK‐8 assay). (E) Colony formation assay of ADAR1‐sh A549 and control cells, untreated or treated with IFN‐β. (F) Colony formation assay of ADAR1‐sh H1975 and control cells, untreated or treated with IFN‐β. (G) Flow cytometry analysis of apoptosis in ADAR1‐sh A549 and control cells, untreated or treated with IFN‐β, stained with annexin V‐APC/PI. (H) Flow cytometry analysis of apoptosis in ADAR1‐sh H1975 and control cells, untreated or treated with IFN‐β, stained with annexin V‐APC/PI. IFN‐β+ and IFN‐β− refer to the IFN‐β‐treated and control groups, respectively. (I, J) Western blotting showing PD‐L1 and ADAR1 expression in A549 and H1975 cell lines when treated with IFN‐β. (K) Flow cytometry analysis of CD8+ T‐cell infiltration in ADAR1‐sh A549 and control cells in chemotaxis experiments, with CD8 stained using annexin APC. (L) Flow cytometry analysis of CD8+ T‐cell infiltration in ADAR1‐sh H1975 and control cells in chemotaxis experiments, with CD8 stained using annexin APC (*p < 0.05; **p < 0.01; ***p < 0.001; ***p < 0.0001; ns, not significant; t‐test; error bars represent standard deviation n = 3 independent cell culture experiments).
Clinical Validation of ADAR1 in Improving Immunotherapy Resistance to LUAD
2.6
The improved immunotherapy response observed in the ADAR1‐knockdown mouse model (Figure 2) suggested the potential of ADAR1 in predicting and enhancing responses to ICIs. To assess the clinical prognostic value of ADAR1, 20 LUAD patients receiving ICI therapy were selected from CICAMS. IHC analysis was conducted to evaluate ADAR1 levels, and patients were classified into low‐ and high‐ADAR1‐expression cohorts according to the median IHC score. Kaplan–Meier survival analysis demonstrated that patients with decreased ADAR1 expression experienced improved progression‐free survival (PFS; Figure 6A). IHC staining of CD8 and PD‐L1 showed higher levels of CD8^+^ TILs and PD‐L1 expression within the group with low‐ADAR1 expression (Figure 6B,C). Representative IHC images are shown in Figure 6D. Immunotherapy response, evaluated based on the RECIST criteria, revealed higher ADAR1 expression in the PR/SD group (Figure S5A), and the PR/SD group had elevated CD8+ TILs and PD‐L1 expression compared to the PD group (Figure S5B,C). Therefore, low‐ADAR1 expression is associated with a more active TME and improved response to immunotherapy.
*ADAR1 is associated with poor immunotherapy response in patients with LUAD. (A) Kaplan–Meier survival curves for progression‐free survival (PFS) based on ADAR1 IHC scores. (B) IHC analysis of CD8+ TILs in the low‐ and high‐ADAR1‐expression groups. (C) IHC analysis of PD‐L1 in the low‐ and high‐ADAR1‐expression groups. (D) Representative IHC images for ADAR1, CD8, and PD‐L1 staining in the corresponding groups (*p < 0.05; *p < 0.01, t‐test; error bars represent standard deviation).
Discussion
3
ICI‐based immunotherapy has provided long‐term clinical benefits and revolutionized treatment strategies for LUAD patients. However, the efficacy of immunotherapy remains lower in cold tumors, and the underlying mechanisms of primary drug resistance urgently require exploration to improve immunotherapeutic responses. Given A‐to‐I RNA editing is a vital epigenetic phenomenon intricately associated with immune regulation and cancer progression [49, 50], this study primarily investigated the effect of ADAR1‐mediated A‐to‐I RNA editing on the immunotherapy response in LUAD. Our findings revealed that elevated RNA editing mediated by ADAR1 contributes to immunotherapy resistance in LUAD by inducing an immunosuppressive TME, marked by decreased CD8^+^ TILs and low PD‐L1 expression. Additionally, ADAR1 and A‐to‐I editing could serve as attractive treatment targets for forecasting and enhancing immunotherapy resistance in LUAD patients.
Consistent with the results of in silico analysis in a previous pancancer study [30], cell experiments and tissue analysis were conducted to validate that RNA editing levels are upregulated in both LUAD tissues and cells. The editing levels were observed to correspond with ADAR1 levels but not with ADAR2 (Figure 1). A mouse model was created using ADAR1‐knockdown and control LLC cells, thereafter treated with anti‐PD‐L1 antibody or a control isotype to investigate the role of ADAR1 in immunotherapy resistance (Figure 2A–C). Findings indicated that tumors in the ADAR1‐knockdown group demonstrated better responsiveness to anti‐PD‐L1 therapy, consistent with findings from melanoma studies [22]. In addition to examining lymphocyte infiltration mechanisms identified in melanoma, this study comprehensively analyzed the TME, comprising PD‐L1 expression and chemokines. The TME is a critical site for antitumor immunity, where CD8^+^ T cells are among the most significant immune cell types, PD‐L1 serves as a crucial immunological checkpoint, and cytokines act as vital signal transducers. In this study, the infiltration of CD8^+^ T cells, along with PD‐L1 expression, IFN‐β, and chemokines related to CD8^+^ TILs chemotaxis (CXCL9, CXCL10, and CXCL11), was elevated in ADAR1‐knockdown LLC cells (Figure 2D–O). These findings indicate that ADAR1‐knockdown activates the type‐I IFN pathway, upregulating IFN‐stimulated chemokines, which increases CD8^+^ TILs and in turn enhances PD‐L1 expression. IHC analysis of LUAD patients confirmed these findings. In the high‐ADAR1‐expression cohort, the percentage of CD8^+^ TILs and PD‐L1 expression was decreased. As described by Teng et al., patients were divided into four TME subtypes, and the high‐ADAR1‐expression group had the highest proportion of samples with an immunological ignorance type (TIL^−^/PD‐L1^−^), indicating that ADAR1 contributes to immune evasion [26]. Cytokine analysis showed upregulated levels of IFN‐β, CXCL9, CXCL10, and CXCL11, consistent with the results observed in the mouse model (Figure 3A–G). Moreover, a negative correlation between A‐to‐I editing and CD8^+^ TILs and PD‐L1 expression was found, suggesting that excess ADAR1‐mediated editing suppresses TME activity. Additionally, ADAR1 expression was positively correlated with shorter overall survival (OS) in patients with LUAD (Figure 3H–K). Besides its role in oncogenesis [33], ADAR1‐induced immune suppression may partly explain the poor prognosis observed in these patients.
As a typical DAMP, dsRNA in the cytoplasm of cancer cells is sensed by TLRs and RLRs, triggering interferon signaling [51]. In the present study, increased accumulation of dsRNA was observed in ADAR1‐knockdown cells. Furthermore, receptors for dsRNA, such as RIG‐I and MDA5, were upregulated in ADAR1‐knockdown cells, accompanied by the activation of the downstream type‐I IFN pathway. IFN‐β and PD‐L1 expression was also elevated in ADAR1‐knockdown cells (Figure 4A–D); however, CXCL9, CXCL10, and CXCL11 were not detected. Considering that not only tumor cells but also other cells within TME, such as macrophages and fibroblasts, can secrete CXCL9, CXCL10, and CXCL11 [52, 53], the type‐I IFN immune response triggered by accumulated dsRNA in cancer cells may promote the secretion of IFN‐β into the TME. This, in turn, activates the secretion of CXCL9, CXCL10, and CXCL11 from other TME cells, which may modulate PD‐L1 expression in neoplastic cells by IFN‐β. However, ADAR1 modulates this process by converting dsRNA into edited forms that are not recognized by their receptors [54]. In the present study, after ADAR1 knockdown, although edited dsRNA levels decreased, dsRNA accumulation increased within the nucleus and cytoplasm of neoplastic cells, indicating that ADAR1 inhibition promotes immune activation through multiple mechanisms.
Interestingly, ADAR1 not only regulates the TME but also affects the response of tumor cells to extracellular cytokines from the TME. Given the increased IFN‐β levels in the TME following ADAR1 knockdown, the effects of IFN‐β stimulation on tumor cells were further investigated. The results revealed that the ADAR1‐p150 isoform and PD‐L1 expression were considerably elevated in LUAD cells treated with IFN‐β, suggesting a negative feedback mechanism triggered by ADAR1 knockdown (Figure 5I,J). Although suppression of oncogenicity in ADAR1‐knockdown cells was not immediately observed as in previous research [32], significant differences between ADAR1‐knockdown and control cells were evident following extracellular IFN‐β stimulation (Figure 5A–H). Notably, ADAR1‐knockdown cells exhibited increased CD8^+^ T‐cell infiltration in the presence of IFN‐β (Figure 5K,L). Accumulation of dsRNA triggers an immune response, which is amplified within the TME as other immune cells become involved. ADAR1‐knockdown cells are activated by upregulating molecules in the IFN pathway, resulting in an enhanced response to extracellular IFN‐β. Simultaneously, the ADAR1‐p150 isoform and PD‐L1 are upregulated to prevent excessive immune activation.
In a clinical cohort of LUAD patients receiving ICIs, the low‐ADAR1‐expression group exhibited a better response, higher CD8^+^ TILs, and elevated PD‐L1 expression. Conversely, ADAR1 expression was higher and CD8^+^ TILs were lower in the group with PD response, with PD‐L1 expression following a similar trend (Figure 6). These findings suggest that ADAR1 knockdown disrupts immune suppression and may serve as an indicator and enhancer of immunotherapy response.
Activation of TME is critical for the success of antitumor immunotherapy. Tumor mutations during oncogenesis serve as a source of neoantigens that stimulate immune response [55]. Recent studies have emphasized the role of ADAR1‐mediated RNA editing in preventing dsRNA antigen presentation [48, 56]. ADAR1 is positioned at the intersection of immune regulation and cancer development due to its dual role in immune suppression and tumor progression, making it a focal point in cancer immunotherapy research [18, 22, 57, 58, 59]. In this study, ADAR1 induced resistance to ICIs in LUAD by restructuring the TME and reducing tumor cell reactivity to extracellular stimuli, offering new insights into strategies to enhance immunotherapy responses.
This study possesses certain drawbacks. The quantity of patients receiving immunotherapy was relatively small, so larger and multicenter studies are needed for further validation. Further research is needed to fully explore the potential of ADAR1 and RNA editing as biomarkers given the limited sample size. Such research could provide a more comprehensive analysis of tumor immunogenicity and contribute to more precise predictions when combined with tumor mutation burden (TMB). Moreover, epigenetic therapies targeting A‐to‐I RNA editing or ADAR1‐p150 isoforms should be explored to enhance immunotherapy efficacy in LUAD patients [60, 61, 62].
A‐to‐I RNA editing facilitated by ADAR1 induces a suppressive TME characterized by low CD8^+^ T‐cell infiltration and reduced PD‐L1 expression through the inhibition of IFN‐β signaling, contributing to immunotherapy resistance in LUAD. Conversely, ADAR1‐knockdown tumors were more vulnerable to the immune‐activating TME. Our findings reveal a novel mechanism of resistance and demonstrate the clinical potential of RNA editing and ADAR1 as biomarkers to predict and improve immunotherapy outcomes in LUAD patients.
In conclusion, our research uncovers a novel mechanism of immunotherapy resistance in LUAD, where excessive A‐to‐I RNA editing, catalyzed by ADAR1, plays a central role. This editing activity fosters an immunosuppressive TME, marked by decreased CD8^+^ TILs and reduced PD‐L1 expression, and inhibits IFN‐β signaling, ultimately undermining LUAD's responsiveness to immunotherapy. Moreover, our results suggest that ADAR1 depletion enhances the tumor's sensitivity to immune activation in the TME. These findings imply that inhibiting ADAR1 could be a promising strategy to augment the therapeutic benefits of immunotherapy in LUAD patients.
Materials and Methods
4
Specimen Collection and Immunohistochemical Analysis
4.1
The clinical information and corresponding formalin‐fixed, paraffin‐embedded (FFPE) tissue specimens from 227 LUAD patients were collected at the Cancer Hospital and Institute, Chinese Academy of Medical Sciences (CICAMS, Beijing, China). The CICAMS cohort included cancer, paracancerous, and normal tissues from 101 patients. Target proteins in these tissues were detected using immunohistochemistry (IHC). For antigen retrieval, tissue slides were submerged in pH 8.0 EDTA buffer. Primary antibodies for IHC staining included ADAR1 (Abcam, #88574), ADAR2 (Santa Cruz, #sc‐73409), CD8 (Zsbio Tech, #ZA‐0508), PD‐L1 (Ventana, #740‐4907), IFN‐β (Abcam, #140211), CXCL9 (Proteintech, #22355‐1‐AP), CXCL10 (R&D, #MAB2662), and CXCL11 (Proteintech, #10707‐1‐AP). TPS, the PD‐L1 tumor proportion score, and the percentage of CD8+ T cells were assessed based on established criteria from prior publications [63]. Samples with CD8^+^ T‐cell infiltration > 10% were categorized as CD8 TIL^+^ samples, and those with a TPS > 1% were considered PD‐L1^+^ samples. The IHC score for other parameters was determined utilizing a formula as follows: overall staining intensity score (Negative: 0; Light (pale yellow): 1; Medium (light brown): 2; Strong (brown): 3) × the percentage of tumor cells stained (≤ 10%: 1; 11%–50%: 2; 51%–75%: 3; > 75%: 4). Patients were categorized into high‐ and low‐expression groups according to the median IHC score [64].
Clinical Cohort and Response Evaluation
4.2
A collection of 20 patients with LUAD receiving ICI therapy were recruited from CICAMS between April 2016 and July 2019. Treatment responses to ICIs were evaluated based on the Response Evaluation Criteria in Solid Tumors, version 1.1. FFPE specimens were available from all patients before initiating ICI treatment. The Ethics Committee at CICAMS approved this study (approval number: 20/242‐2438). All LUAD specimens were acquired following the obtaining of written informed consent from the patients. The clinical and demographic characteristics of the patients are provided in Table S1.
Cell Culture and Cell Transfection
4.3
The LUAD cell lines used in this study (A549, H1975, A427, H358, H1299, and H322) were acquired from the American Type Culture Collection (ATCC). Tumor cells were cultivated in RPMI 1640 media (Corning, #10‐040‐CV) augmented with 10% fetal bovine serum (FBS, Corning, #35‐081‐CV) and 1% antibiotics (penicillin and streptomycin, Gibco, #15140‐122), whereas normal human bronchial epithelial cells (BEAS‐2B) were maintained in BEGM (Lonza, #CC‐3170). All cells were incubated at 37°C with 5% CO_2_. ADAR1‐knockdown cell lines were generated using short hairpin RNAs (shRNAs; OBiO Technology, Shanghai), and small interfering RNAs (siRNAs) were used to silence MDA5 (Syngen, Beijing). The sequences of shRNAs and siRNAs are listed in Table S2.
A‐to‐I RNA Editing Detection and Qualitative Analysis
4.4
A total of 32 LUAD tissues were quick‐frozen in liquid nitrogen immediately postresection and preserved at −80°C until future utilization. These patients had not received any treatment prior to surgery and were pathologically diagnosed with LUAD. The reagent for RNA extraction from tissues and cells was Trizol (Invitrogen, #15596026). Sanger sequencing was performed on PCR products of cDNA for typical sites (AZIN1: chr8:103841636, GIPC1: chr19:14593693, SON: chr21:34923319, COPA: chr1:160302244, COG3: chr13:46090371), and editing events were verified while editing levels were evaluated using ImageJ software. Table S3 lists the editing sites and primers used during PCR amplification.
Construction of a Mouse Model and Immunotherapy Treatment
4.5
ShRNAs targeting mouse ADAR1 were synthesized (sequences listed in Table S2) and used to establish an ADAR1‐knockdown Lewis lung carcinoma (LLC) cell line. C57BL/6J mice were obtained from Huafukang Bioscience (Beijing, China). Approximately 2.0 × 10^6^ sh‐ADAR1 or control cells, diluted in 100 µL PBS, were injected into 4‐week‐old female mice subcutaneously. Two tumor types were subsequently categorized into four groups, each including five mice, upon the tumors reaching a diameter of 6 mm. The mice were treated with PD‐L1 (10 mg/kg three times a week, Bio X Cell, BE0101) and isotype control on Days 1, 4, 7, and 10. Tumor volume was calculated utilizing the formula V = L × W ^2^/2 (V represents tumor volume, L denotes tumor length, and W signifies tumor breadth). Mice were humanely euthanized via carbon dioxide inhalation, and tumors were excised for further analysis. All mouse studies adhered to the ethical guidelines established by CICAMS, and all operations were examined and approved by the Animal Care and Use Committee of CICAMS (permit number: #NCC2020A163).
Analysis of Changes in the Tumors of Mice
4.6
IHC staining was conducted on FFPE tumor tissues from mice to detect CD8 (CST, #98941), PD‐L1 (Abcam, ab238697), IFN‐β (Abclonal, A1575), and ADAR1 (Proteintech, 14330). Cytokine levels of IFN‐β (RayBio, #ELM‐IFNb1), CXCL9 (RayBio, #ELM‐IFNg), CXCL10 (RayBio, #ELM‐MIG), and CXCL11 (RayBio, #ELM‐CRG2, #ELM‐ITAC) were evaluated by enzyme‐linked immunosorbent assay (ELISA). Tumors were ground into a powder in liquid nitrogen and collected in 2‐mL Eppendorf tubes. The Protease/PhosSTOP Phosphatase Inhibitor Cocktail (Beyotime, #P1048) and lysis buffer were added to the powdered samples; the supernatant was obtained following ultrasonication and centrifugation. A BCA assay (Thermo, #23227) was used to evaluate protein concentration. Tumor lysates were then examined for cytokine levels.
Cell Proliferation Assay
4.7
Cells were inoculated in culture dishes, and for cytokine stimulation, IFN‐β (R&D, #8499‐IF‐010/CF, 1000 U/mL) was added to the culture medium over 48 h. The CCK8 kit (Dojindo, #CK04) was employed to evaluate cell proliferation. Absorbance was recorded at 0, 24, 48, and 72 h utilizing a microplate reader. Colony formation assays were conducted to assess cell viability, and apoptosis was identified utilizing the Annexin V‐APC/PI Apoptosis Detection Kit (KeyGEN, #KGA1030) in accordance with the manufacturer's guidelines. The proportion of apoptotic cells was analyzed utilizing FlowJo software.
Detection of dsRNA
4.8
Immunofluorescence (IF) was conducted for detecting dsRNA in cells. A total of 1 × 10^4^ cells per well were inoculated onto 8‐well slides and incubated for 18–24 h. The cells were fixed with precooled methanol for 15 min, incubated with 0.2% Triton (Sigma) for 30 min, blocked with 5% BSA for 60 min at room temperature, and incubated with the primary antibody (clone J2, 1:100, SCICONS) for 10 h at 4°C and the secondary antibody (Alexa Fluor 555 Conjugate, 1:1000, CST) for 1 h in the absence of light at room temperature. The cells were incubated with DAPI (Invitrogen) for 5 min, covered with coverslips, and examined using a confocal microscope.
Western Blotting and Real‐Time Reverse Transcription Polymerase Chain Reaction
4.9
Total protein was extracted and quantified (lysis buffer: BeyotimeP0013C; BCA kit: Thermo Scientific, #23227). After being separated on 10% SDS‐PAGE gels, the proteins were put onto polyvinylidene fluoride (PVDF) membranes (IPVH07850, Millipore). The proteins were transferred to PVDF membranes (IPVH07850, Millipore) after separation on 10% SDS‐PAGE gels. Beta‐actin served as the internal loading control. The principal antibodies employed were listed as follows: ADAR1 (Abcam, #88574), actin (CST, #3700), PD‐L1 (CST, #13684), and IFN‐β (Abclonal, #A16223).
Total RNA was extracted using the RNA‐Quick Purification Kit (Esunbio, #ES‐RN001), and cDNA was synthesized employing a reverse transcription kit and cDNA Synthesis SuperMix (Transgen #AH341). Real‐time reverse transcription polymerase chain reaction (qRT‐PCR) was performed using SYBR Green Mix (Thermo Scientific, #4367659) to detect mRNA expression. Table S4 presents the primers employed in this study.
Chemotaxis Experiment of CD8+ T Cells In Vitro
4.10
CD8^+^ T cells were extracted from PBMCs provided by healthy adults by the cell extraction kit (Miltenyi Biotec #130‐096‐495) and then cultured in vitro. A certain quantity of CD8+ T cells was thereafter introduced into the upper compartment of the Transwell chamber of a 24‐well culture dish, while ADAR1‐knockdown cells and control cells were concurrently positioned in the lower compartment. Twenty‐four hours poststimulation, CD8^+^ T cells in the subcapsular region of both stimulated and control groups were analyzed via flow cytometry to evaluate the impact of interferon‐β on CD8^+^ T‐cell chemotaxis.
Statistical Methods
4.11
Statistical analyses were conducted applying GraphPad Prism 8 (GraphPad Software, Inc.), SPSS Statistics (version 25.0; IBM, USA), and R (version 3.6.2). The Student's t‐test was employed to compare data between groups, whereas the chi‐square test was implemented to analyze the TME. Survival curves were generated with the Kaplan–Meier method. The graphical abstract was created with BioRender.com. This study has been conducted in compliance with the ARRIVE1 guideline.
Author Contributions
Sihui Wang, Chengming Liu, and Sufei Zheng: data curation, formal analysis, and original draft preparation. Chaoqi Zhang, Xinfeng Wang, Zhanyu Wang, Yan Wang, Xiaoli Feng, and Qi Xue: performed experiments and analyzed data. Jie He and Nan Sun: project administration, supervision, reviewing and editing, and funding acquisition. All the authors have read and approved the final manuscript.
Ethics Statement
This research was conducted with the approval of the Ethics Committee of CICAMS (approval number: 20/242‐2438). The Animal Care and Use Committee of CICAMS (approval number: #NCC2020A163) thoroughly reviewed and granted approval for all mouse experimentation protocols.
Conflicts of Interest
All the authors have made significant contributions to the content of the manuscript. The authors declare no conflicts of interest.
Supporting information
FIGURE S1 (A) Sanger sequencing for detecting A‐to‐I RNA editing levels of three typical sites (AZIN1: chr8:103841636, COG3: chr13:46090371, GIPC1: chr19:14593693) in BEAS‐2B cells and six LUAD cell lines. (B) qRT‐PCR analysis of ADAR1 mRNA expression in BEAS‐2B cells and six LUAD cell lines. (C) Sanger sequencing for detecting A‐to‐I RNA editing levels of AZIN1 in BEAS‐2B cells and six LUAD cell lines. (D) Correlation of ADAR1 mRNA expression and A‐to‐I editing level of AZIN1 in BEAS‐2B cells and six LUAD cell lines. (E) Sanger sequencing for detecting A‐to‐I RNA editing levels of COG3 in BEAS‐2B cells and six LUAD cell lines. (F) Correlation of ADAR1 mRNA expression and A‐to‐I editing level of COG3 in BEAS‐2B cells and six LUAD cell lines. (G) Sanger sequencing for detecting A‐to‐I RNA editing levels of GIPC1 in BEAS‐2B cells and six LUAD cell lines. (H) Correlation of ADAR1 mRNA expression and A‐to‐I editing level of GIPC1 in BEAS‐2B cells and six LUAD cell lines. (I) Western blot demonstrating the expression of ADAR1 in ADAR1‐knockdown (Sh) cells, ADAR1‐overexpression (OE) cells, enzymatically inactive‐ADAR1 overexpression (AN) cells and control A549 (Vec) cells. (J) Sanger sequencing for detecting A‐to‐I RNA editing levels of typical site (SON: chr21: 34923319) in cell lines of J. (K) Western blot demonstrating the expression of ADAR1 in ADAR1‐knockdown (Sh) cells, ADAR1‐overexpression (OE) cells, enzymatically inactive‐ADAR1 overexpression (AN) cells and control H1975 (Vec) cells. (L) Sanger sequencing for detecting A‐to‐I RNA editing levels of typical site (SON: chr21: 34923319) in cell lines of K. (cor, Pearson correlation coefficient; **p < 0.01; ***p < 0.001; ns: not significant; ****p < 0.0001, t‐test, error bars represent the standard deviation. n = 3 independent cell culture experiments). FIGURE S2 (A) Western blot demonstrating the expression of ADAR1 in ADAR1‐knockdown LLC cells and control cells. (B). Typical IHC images of ADAR1 in ADAR1‐knockdown LLC cells and control cells. (C) IHC analysis of ADAR1 in ADAR1‐knockdown LLC cells and control cells.sh1, sh2, sh3, and con refer to ADAR1‐knockdown LLC cells constructed via shRNA‐1, shRNA‐2 or shRNA‐3 and control LLC cells, respectively. (D) Immunofluorescence analysis of dsRNA in ADAR1‐knockdown LLC cells and control cells, stained with DAPI (blue) and anti‐dsRNA antibody J2 (red). (E) Quantitative analysis of fluorescence intensity of dsRNA in ADAR1‐knockdown LLC cells and control cells. (****p < 0.0001, t‐test, error bars represent the standard deviation. n = 3 independent cell culture experiments). FIGURE S3 Typical IHC images of ADAR1, CD8, PD‐L1, IFN‐β, CXCL9, CXCL10, CXCL11. FIGURE S4 (A) Flow cytometry for analyzing the apoptosis of ADAR1‐knockdown A549 and control cells untreated or treated with IFN‐β stained with annexin V‐APC/PI. (B) Flow cytometry for analyzing the apoptosis of ADAR1‐knockdown H1975 and control cells untreated or treated with IFN‐β stained with annexin V‐APC/PI. (C) Flow cytometry for analyzing CD8^+^ T‐cells infiltration in ADAR1‐sh A549 and control cells in chemotaxis experiment, CD8 was stained with annexin APC. (D) Flow cytometry for analyzing CD8^+^ T‐cells infiltration in ADAR1‐sh H1975 and control cells in chemotaxis experiment, CD8 was stained with annexin APC. FIGURE S5 (A) IHC analysis of ADAR1 in the PR/SD and PD groups. (B) IHC analysis of CD8^+^ TILs in the PR/SD and PD groups. (C) IHC analysis of PD‐L1 in the PR/SD and PD groups.PR, SD, and PD refer to partial response, stable disease, and progressive disease, respectively (*p < 0.05, **p < 0.01, t‐test, error bars represent the standard deviation). TABLE S1 Clinical and demographic characteristics of patients. TABLE S2 Sequence of shRNA and siRNA used in LUAD cells. TABLE S3 Primers for Sanger sequencing of editing sites used in this study. TABLE S4 Primers for PCR used in this study.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1R. L. Siegel , K. D. Miller , and A. Jemal , “Cancer Statistics, 2020,” CA: A Cancer Journal for Clinicians 70, no. 1 (2020): 7–30.31912902 10.3322/caac.21590 · doi ↗ · pubmed ↗
- 2R. Zheng , S. Zhang , H. Zeng , et al., “Cancer Incidence and Mortality in China, 2016,” Journal of the National Cancer Center 2, no. 1 (2022): 1–9.39035212 10.1016/j.jncc.2022.02.002PMC 11256658 · doi ↗ · pubmed ↗
- 3Z. Chen , C. M. Fillmore , P. S. Hammerman , C. F. Kim , and K.‐K. Wong , “Non‐Small‐Cell Lung Cancers: A Heterogeneous Set of Diseases,” Nature Reviews Cancer 14, no. 8 (2014): 535–546.25056707 10.1038/nrc 3775 PMC 5712844 · doi ↗ · pubmed ↗
- 4Y. T. Lee , Y. J. Tan , and C. E. Oon , “Molecular Targeted Therapy: Treating Cancer With Specificity,” European Journal of Pharmacology 834 (2018): 188–196.30031797 10.1016/j.ejphar.2018.07.034 · doi ↗ · pubmed ↗
- 5R. Rosell , E. Carcereny , R. Gervais , et al., “Erlotinib Versus Standard Chemotherapy as First‐line Treatment for European Patients With Advanced EGFR Mutation‐Positive Non‐Small‐Cell Lung Cancer (EURTAC): A Multicentre, Open‐Label, Randomised Phase 3 Trial,” Lancet Oncology 13, no. 3 (2012): 239–246.22285168 10.1016/S 1470-2045(11)70393-X · doi ↗ · pubmed ↗
- 6M. Miller and N. Hanna , “Advances in Systemic Therapy for Non‐Small Cell Lung Cancer,” BMJ 375 (2021): n 2363.34753715 10.1136/bmj.n 2363 · doi ↗ · pubmed ↗
- 7J. E. Chaft , Y. Shyr , B. Sepesi , and P. M. Forde , “Preoperative and Postoperative Systemic Therapy for Operable Non‐Small‐Cell Lung Cancer,” Journal of Clinical Oncology: Official Journal of the American Society of Clinical Oncology 40, no. 6 (2022): 546–555.34985966 10.1200/JCO.21.01589 PMC 8853628 · doi ↗ · pubmed ↗
- 8A. J. Cooper , L. V. Sequist , and J. J. Lin , “Third‐Generation EGFR and ALK Inhibitors: Mechanisms of Resistance and Management,” Nature Reviews Clinical Oncology 19, no. 8 (2022): 499–514.10.1038/s 41571-022-00639-9PMC 962105835534623 · doi ↗ · pubmed ↗