The Interplay between RNA Editing Regulator ADAR1 and Immune Environment in Colorectal Cancer

An abnormality in the regulation of adenosine deaminase acting on RNA (ADAR) enzymes, which catalyzed adenosine-to-inosine (A-to-I) RNA editing, was closely associated with the highly aggressive biologic behavior and poor prognosis in many malignancies. In the present study, we aimed to investigate the relationship among transcript factors-microRNAs regulatory network, immune environment, and ADAR gene in colorectal carcinoma (CRC). The association among the expression levels of ADAR mRNA and copy number variation, methylation, and mutation status were comprehensively analyzed using cBioPortal, Wanderer, and UALCAN databases in CRC datasets. ADAR-transcript factors (TFs) and ADAR-miRNA regulation networks were constructed by Cistrome Cancer and miRWalk2.0, respectively. The full network and subnetworks for ADAR coexpression genes were constructed using the STRING database and visualized by the MCODE module of the Cytoscape app. The relationship between ADAR mRNA expression and the abundance of infiltrating immune cells in CRC patients was explored by the Tumor Immune Estimation Resource, CIBERSORT, and single-gene gene set enrichment analysis (GSEA). ADAR mRNA was elevated and was a cancer essential gene in CRC. ADAR mRNA and transcripts P110 were significantly elevated in CRC compared to normal controls. Low-level methylation in the promoter region and high copy number amplification of ADAR were responsible for high levels of ADAR mRNA expression. ADAR coexpression genes were mainly involved in immunoregulation, especially T-lymphocyte activation. Hub genes, including CD2, CD274, and FASLG, were also significantly upregulated in the ADAR-high group compared to the control group. Besides, M1 macrophages were enriched in the ADAR-high group compared to the control group. This study demonstrated that ADAR, a new essential gene, was involved in the immune regulator and was a novel immune treatment target in CRC.


Introduction
Colorectal carcinoma (CRC) was a highly prevalent malignant cancer of the lower digestive tract and was the third leading cause of cancer-related death worldwide [1]. According to the National Comprehensive Cancer Network (NCCN), the incidence and mortality of malignant tumors have been significantly increasing in China [2] but gradually decreasing in the United States [3], especially CRC. Despite the development of diagnostic and therapeutic modalities, CRC remained extremely difcult to efcient treatment and was poor survival, mainly due to high genetic heterogeneity [4,5]. Terefore, elucidating the mechanisms underlying heterogeneity was very important for CRC diagnosis and therapy.
Adenosine-to-inosine (A-to-I) RNA editing [6] was an important form of posttranscriptional processing, which alters RNA molecule modifcation by adenosine deaminases acting on RNA (ADARs) in double-stranded RNA (dsRNA). Of them, ADAR (also known as ADAR1) was a member of the adenosine deaminase family [7], which also includes three ADAR proteins (ADAR, ADARB1, and ADARB2) and two ADAR-related proteins (ADAD1 and ADAD2) according to the HUGO Gene Nomenclature Committee. Among them, ADAR [8] has higher editing efciency than ADARB1, which only edits one particular nucleotide position in the precursor, whereas ADARB2 has not been shown to be an active enzyme due to the lack of a catalytic domain. Both ADAR1 and ADAR2 were ubiquitously expressed (ADAR2 is most abundant in the brain), whereas ADAR3 expression was restrictively expressed in brain tissue [9].
Te dysregulation of RNA editing regulator ADARs has been frequently linked to human cancers including hepatocellular carcinoma [10,11], esophageal squamous cell carcinoma [12], gastric cancer [13], and breast cancer [14,15]. Unlike other cancer types, the relationship between RNA editing and CRC mostly focuses on the downstream gene AZIN1 [16][17][18]. Besides, ADAR was alternatively spliced to generate two isoforms [19], commonly known as the constitutive 110 kDa isoform ADAR1 (P110) and interferon-inducible isoform ADAR (P150), suggesting the close association of catalytic activity of ADAR with immune and infammatory responses. Te gene annotation portal BioGPS [20] has shown high expression of ADAR transcripts in T lymphocytes and other immune cells. Tese observations suggest the important roles of the regulation of RNA editing in CRC progression and prognosis. However, the abnormal alterations of ADAR, corresponding driving forces, and immune association have not been clearly elucidated in CRC.
Notably, we also constructed ADAR-transcript factors and ADAR-miRNA regulation networks to explore the upstream regulator of ADAR. Many studies showed that the TF and miRNA network might be the potential target of therapy for disease [21][22][23]. Te emergence of highthroughput technologies in recent years has provided a powerful tool to detect an increasing number of genetic and epigenetic alterations in CRC, resulting in breakthroughs in the understanding of the biological characteristics of CRC for preventing, diagnosing, and treating. Driven by the massive accumulated high-throughput dataset, we have investigated the interplay between RNA editing regulator ADAR1 and the immune environment in CRC.
In this study, we systematically analyzed ADAR mRNA expression across multiple CRC datasets to uncover the multiple roles of ADAR in CRC progression and prognosis. Our study showed the steadily elevated expression of the RNA editing regulator ADAR in CRC compared with the normal control. Functional analysis revealed signifcant involvement of ADAR coexpression genes in tumor immune checkpoints in CRC, thereby inducing a proimmunomodulatory efect. Together, the results of this study demonstrate that ADAR likely has immune regulatory roles and may serve as a novel potential biomarker and target immunotherapy for CRC.  [24], and Human Protein Atlas (HPA, https://www. proteinatlas.org/) were employed to explore the expression profle (mRNA and protein) of A to I RNA editing regulator ADAR in CRC datasets. Essential gene screening from gene efect scores derived from CRISPR knockout screens published by Broad's Achilles and Sanger's SCORE projects was analyzed by the UALCAN database. Negative scores imply cell growth inhibition and/or death following gene knockout.

Expression Pattern of Typical ADAR Transcripts in CRC.
ADAR encodes two distinct splicing isoforms: a constitutive 110 kDa isoform ADAR (p110) and an interferon-inducible isoform ADAR (p150). UCSC Xena browser (Te University of California Santa Cruz, https://xena.ucsc.edu/), a bioinformatics tool used to visualize functional genomics data from multiple sources simultaneously, was applied to assess the expression levels of ADAR transcripts (p110 and p150) mRNA in TCGA colorectal cancer based on UCSC Toil RNA seq Recompute Compendium (31) (TCGA and GTEx datasets). ADAR transcript RNA sequencing (RSEM TPM, n = 19,131) data were downloaded as log2 (TPM + 0.001) values. Te Wilcoxon signed-rank test was performed to determine the diference in the expression of ADAR-p150 and ADAR-p110 isoform between CRC and para-cancerous with the help of GraphPad Prism software 7 (San Diego, CA, USA), and P < 0.05 was considered statistically signifcant.

Exploring the Association between Genetic Alteration and mRNA Expression of ADAR Gene in CRC.
Te mutation status, methylation status, and copy number variation (CNV) of ADAR genes were analyzed with Wanderer [25], UALCAN [24], and cBioPortal databases [26] to elucidate the molecular mechanism of upregulated ADAR at the DNA level. Wanderer [25], an interactive viewer to explore DNA methylation and gene expression data, was frst used to explore the diferences in ADAR gene-wide methylation status in CRC based on TCGA cancer dataset. Te correlation between ADAR gene amplifcation, methylation, and gene expression data in TCGA CRC datasets was determined by the cBioPortal for Cancer Genomics Portal (https:// cbioportal.org).

Construction of ADAR mRNA Coexpression Network in CRC.
Genes coexpressed with ADAR were identifed using the cBioPortal for Cancer Genomics (https://www. cbioportal.org/) with multidimensional cancer genomics datasets. Spearman's correlation coefcient >0.4 and P < 0.01 were set as the cut-of criteria. Te protein-protein interaction (PPI) network was constructed with the Search Tool for the Retrieval of Interacting Genes Database (STRING) [31](https://www.string-db.org/), and a confdence score >0.4 was considered signifcant. Te hub modules of the PPI network were determined using the molecular complex detection (MCODE) module in Cytoscape with cut-of criteria: degree � 15, node score � 0.2, kcore � 2, and max depth � 100. ADAR mRNA coexpression network and subnetwork were visualized by Cytoscape software.

Identifying Association between Hub Gene Expression
Pattern and ADAR in the PPI Subnetwork. Most connectivity modules with ADAR were selected as the hub subnetwork to analyze the expression profle under ADAR high and low expression. ADAR genes were then split in TCGA dataset into low and high expression using the median value of expression profles in the genome-wide scale as a cut-of value. P < 0.05 was considered statistically signifcant.

Correlation between ADAR Expression and Infltrating
Immune Cells. Te correlations between ADAR expression and the abundance of six types of infltrating immune cells (B cells, CD4 + T cells, CD8 + T cells, neutrophils, macrophages, and dendritic cells) in CRC were estimated with the partial correlation coefcient (tumor purity) by Tumor IMmune Estimation Resource [33] (TIMER, https:// cistrome.shinyapps.io/timer/). Furthermore, we used the CIBERSORT deconvolution algorithm [34] (https:// cibersort.stanford.edu/) to estimate the abundance of 22 immune cell types under ADAR high and low expression and to evaluate the corresponding intratumoral immune cell composition. Besides, the single-gene GSEA strategy was utilized to detect the ADAR-related biological pathways (50 cancer hallmark pathways) [35] in the pan-cancer dataset.

Te ADAR mRNA Was Elevated and Was a Cancer Essential Gene in CRC.
Te fnding from the expression levels of ADAR mRNA showed statistically signifcant upregulated expressions in pan-cancer databases compared with control (P < 0.0001; Figure 1(a)), especially CRC (Figures 1(b), 1(c), and 1(e)), which was extracted from TCGA CRC dataset, CPTAC protein, and Human Protein Atlas (HPA) databases. Interestingly, we found that ADAR was a cancer essential gene in CRC, according to gene efect scores derived from CRISPR knockout screens published by Broad's Achilles and Sanger's SCORE projects (Figure 1(d)). Tis fnding inspires us to further study the ADAR.

Te ADAR Transcript p110 Was the Main Regulator of A to I RNA Editing Events in CRC.
As shown in Figures 2(a), 2(b), 2(d), and 2(e), ADAR and ADAR-p110 mRNA were highly expressed in CRC tissues compared with paracancerous tissues. However, ADAR-p150 mRNA transcript was downregulated in CRC tissues in contrast to normal colorectal tissues (Figures 2(c) and 2(f )). Tese results suggest that the ADAR transcript p110 was the main regulator of A to I RNA editing events in CRC and was involved in multiple biological functions rather than p150.

Te Association between ADAR Genetic Alterations and ADAR mRNA in CRC.
As was illustrated in Figure 3(a), the ADAR high copy number was the most signifcant genomic hallmark. As was shown in Supplementary Figure 1, higher DNA methylation in the ADAR gene and 3′-UTR was observed in both CRC and normal controls. Statistically signifcant diferences in ADAR gene promoter regions were observed, and the ADAR gene was found to be methylated at very low levels in both CRC and normal control tissues (Figures 3(b) and 3(c)). DNA methylation was negatively correlated with the expression level of ADAR (Figure 3(c)). Te results suggest that the elevated ADAR gene transcription did not result from methylation, at least not methylation of the promoter region. In addition, the ADAR gene copy number was signifcantly higher in CRC tissues compared with normal control tissues (Figures 3(d) and 3(e)), suggesting the association of the ADAR mRNA level with the copy number of deep deletions, shallow deletion, diploid, gain, and amplifcation. Tere was a signifcant positive correlation between the DNA copy number and the expression level (Figure 3(e)). In conclusion, high copy number amplifcation of DNA was the driving force for the increase in the expression level of ADAR.

ADAR Coexpression Network, Hub-Network, and Functional Enrichment.
In total, 1153 positive genes associated with ADAR mRNA expression were obtained through cBioPortal for Cancer Genomics to construct an ADARrelated coexpression gene network. As shown in Figure 5(a), 961 proteins with 8543 nodes were involved in the ADARrelated coexpression network based on the STRING database. A set of 32 proteins that interacted with at least 15 other proteins (P < 0.05, FDR <0.05) were selected in the hub network of the PPI network ( Figure 5(b)). Functional enrichment analysis showed that the genes were mainly enriched in BP of lymphocyte activation, cytokine production, infammatory response, adaptive immune response, leukocyte migration, response to interferongamma, and positive regulation of immune response ( Figure 5(c)). As shown in Figure 5(d), the results from the transcript factor analysis suggested that ADAR coexpression genes were signifcantly regulated by STAT1, IRF1, IRF9, and NFKB1. As we know, these transcription factors were closely related to immune regulation, which further suggests that ADAR was signifcantly involved in the immune regulation process of CRC.

Hub-Subnetwork Derived from the ADAR Coexpression Network and Expression Profle under the ADAR High/Low
Group. Finding from single-gene GSEA of ADAR in pancancer further confrmed that ADAR participates in the immune signal pathway, especially the interferon pathway ( Figure 6(a)). CD2, CD274, and FASLG mRNA levels were signifcantly higher expressed in the ADAR high group than in the low group (Figure 6(b)). Terefore, ADAR may exert immune regulation function through these immune-related proteins.

Te Association between ADAR Gene Expression and Immune Cell Infltration. Te correlations of ADAR with infltrating immune cells (tumor purity) including B cells,
CD4 + T cells, CD8 + T cells, neutrophils, macrophages, and dendritic cells in the gastrointestinal tumors were analyzed by TIMER. As shown in Figure 7(a), ADAR gene expression was closely correlated with dendritic cells, neutrophils, CD4 + T cells, macrophages, and CD8 + T cells in CRC, especially macrophages. Similar to the previous results (Figure 7(b)), M1 macrophages were enriched in the ADAR high group more than the low group. Te abovementioned results in the article suggest that ADAR signifcantly afects the immune regulation of M1 macrophages.

Discussion
Immune therapy has been increasingly applied in many cancers that are now characterized by their unique genomic alterations with new diagnostics. Understanding the immune environment in CRC would not only increase the therapeutic efcacy but would also provide a better treatment strategy. Previous studies have shown a signifcant correlation between increased ADAR expression and poorer survival outcomes in esophageal squamous cell carcinoma [12,36] and human hepatocellular carcinoma [10]. In this study, we demonstrated that total ADAR mRNAs and typical transcripts (p110) were highly expressed in CRC compared with adjacent paracancerous tissues. However, the p150 mRNA expression level was downregulated between CRC and normal control tissues; ADAR1-p150 was induced by interferon, whereas ADAR1-p110 was constitutively and ubiquitously expressed.
Finding from genetic alteration, the TF-ADAR-microRNA network showed that high expression of ADAR in CRC might be caused by hypermethylation of the ADAR gene body region, copy number amplifcation, positive transcription factor, and negative miRNA.
Tere are few studies on the transcriptional regulation of the RNA editing enzyme ADAR itself, most of which focus on the activity of the ADAR editing enzyme and downstream molecules. From the perspective of transcriptional regulation, our study explored the functional regulation of   editing enzymes ADAR with the help of two networks (ADAR-TFs and ADAR-miRNA network's roles). A classic example of transcription factors regulating ADAR, AR served as a transcriptional activator of the ADAR1 promoter to promote HCC tumor growth both in vitro and in vivo [37]. Te infltrated immune cells in the tumor microenvironment played a crucial function in the occurrence and development of tumors. ADAR-related coexpression network and functional enrichment analysis showed that the ADAR coexpressed proteins were mainly enriched in immunoregulation, especially T-lymphocyte activation and cytokine production. Te evaluation of immune cell infltration in CRC using the TIMER database revealed strong correlations between ADAR and tumor purity in CRC. Te expression level of ADAR was closely associated with levels of CD8 + cell, CD4 + cell, neutrophil, and dendritic cell infltration in CRC. Te correlation between ADRA expression and immune cell marker genes suggests that ADRA regulates CRC immunity through multiple immune cell populations by A-to-I dsRNA editing enzyme activity. Studies have shown the involvement of ADAR in inducing an immunomodulatory efect [38]. Our results are consistent with such reports, and these discoveries suggest that ADAR plays an important role in recruiting and governing immune responses in CRC.
Studies have revealed that CRC is diferent clinically, pathologically, and genetically and could predispose to diferent clinical assumptions regarding tumorigenesis as well as survival. Resistance to targeted drugs has become an indisputable fact, thereby resulting inefectiveness of the drug treatment. However, the precise mechanism mediating drug resistance has not been fully understood. Te understanding of the molecular and biochemical mechanisms of drug resistance will facilitate the discovery of alternative target drugs. Our study revealed several key genes with high degrees involved in the subnetwork, including CD274, CD20, and FASLG that have been widely used in clinical tumor treatment, and they were positively associated with poor prognosis in CRC. Loss of ADAR1 was demonstrated to defeat resistance to PD-1 checkpoint blockade caused by inactivation of antigen presentation by tumor cells [39], thereby limiting the efective antitumor immunity. Te fndings provide a new direction for overcoming immunotherapy resistance by regulating ADAR expression.
Although our results are encouraging, there are some limitations to the study: (1) our study was a pure bioinformatics analysis based on TCGA database and online dataset, and further biological experiments were needed to validate our results. (2) It seems to be a contradictory phenomenon that ADAR is both an oncogene and immune-active gene from the perspective of the expression level and immune association, and the follow-up experiments will focus on the role and function of ADAR in CRC.
In conclusion, ADAR expression was closely correlated with multiple immune markers in CRC. Te correlations between ADAR and the prognosis and immune cell infltration provide a foundation for further research on its immunomodulatory role in CRC diagnosis and treatment.

Data Availability
All the transcriptome sequencing data were obtained from the UCSC Xena database (https://xenabrowser.net/ datapages/), the mutational data were obtained from cBioPortal database (http://www.cbioportal.org/), other data that support the fndings of this study are available from the corresponding author upon reasonable request.

Ethical Approval
Te data were downloaded from the public database, and there was no ethical question about the data. Ethical approval was not applicable here.

Conflicts of Interest
Te authors declare that there are no conficts of interest.