Reduced A-to-I editing of endogenous Alu RNAs in severe COVID-19 disease

Severe COVID-19 disease is associated with elevated inflammatory responses. One form of Aicardi-Goutières Syndrome caused by inactivating mutations in ADAR results in reduced A-to-I editing of endogenous double-stranded RNAs (dsRNAs), induction of IFNs, interferon-stimulated genes (ISGs), other inflammatory mediators, morbidity and mortality. Alu elements, ~10% of the human genome, are the most common A-to-I editing sites. Using leukocyte whole-genome RNA sequencing data, we find reduced A-to-I editing of Alu dsRNAs in patients with severe COVID-19 disease. Unedited Alu dsRNAs, but not edited Alu dsRNAs, are potent inducers of IRF and NF-kB transcriptional responses, IL6, IL8 and ISGs. Thus, decreased A-to-I editing that may lead to accumulation of unedited Alu dsRNAs and increased inflammatory responses, is associated with severe COVID-19 disease. We employed whole-genome RNA-sequencing (RNA-seq) files from NCBI Gene Expression Omnibus (GSE149689) for analysis of A-to-I editing. We used the following workflow to identify RNA A-to-I editing sites from paired FASTQ sequencing files. The main identification tool was a python-based package called the SPRINT toolkit that accepts sequence files and produces text files with the following information for each edit site: (A) genomic location; (B) type of edit (e.g., A-to-G; T-to-C); strand (+ or -); (C) number of edits per site and total number of reads per site. Mathematica programs were developed to synthesize data: numbers of samples in groups with shared editing sites, mean numbers of total reads and edits for each editing site, and editing sites common and unique to group pairs (e.g., HC versus COV-S). This information was tied to an Alu database to annotate each site: gene locations (intronic, ncRNA, intergenic, 3UTR), and if sites were Alu or non-Alu elements. To create genome-wide A-to-I editing indices, we identified all A-to-I editing sites present in one sample and summed edit/read ratios for all editing sites across the genome for each HC, COV-M, COV-S or FLU-S sample. Data Processing: Input FASTQ RNA-seq files into SPRINT toolkit recover text files with the following information for each edit site: (A) genomic location; (B) type of edit (e.g., A-to-G; T-to-C); strand (+ or -); (C) number of edits per site and total number of reads per site. Mathematica programs were developed to synthesize data: numbers of samples in groups with shared editing sites, mean numbers of total reads and edits for each editing site, and editing sites common and unique to group pairs (e.g., HC versus COV-S). This information was tied to an Alu database to annotate each site: gene locations (intronic, ncRNA, intergenic, 3UTR), and if sites were Alu or non-Alu elements. To create genome-wide A-to-I editing indices, we identified all A-to-I editing sites present in one sample and summed edit/read ratios for all editing sites across the genome for each HC, COV-M, COV-S or FLU-S sample. assembly: hg19

重型新型冠状病毒肺炎（COVID-19）与炎症应答亢进密切相关。由腺苷脱氨酶（ADAR）失活突变引发的一种艾卡迪-古特雷斯综合征（Aicardi-Goutières Syndrome），会导致内源性双链RNA（dsRNAs）的A-to-I编辑水平降低，诱导干扰素（IFNs）、干扰素刺激基因（ISGs）及其他炎症介质的产生，并引发机体发病与死亡。Alu元件（Alu elements）约占人类基因组的10%，是最常见的A-to-I RNA编辑位点。本研究利用白细胞全基因组RNA测序数据，发现重型COVID-19患者体内Alu来源双链RNA的A-to-I编辑水平显著降低。未编辑的Alu来源双链RNA（而非已编辑的Alu来源双链RNA）可强效诱导干扰素调节因子（IRF）与核因子κB（NF-κB）的转录应答，以及白细胞介素6（IL-6）、白细胞介素8（IL-8）和ISGs的表达。因此，可能导致未编辑Alu来源双链RNA积累、炎症应答增强的A-to-I编辑水平降低，与重型COVID-19存在显著关联。

本研究采用来自美国国家生物技术信息中心（NCBI）基因表达综合数据库（Gene Expression Omnibus, GEO）的GSE149689号全基因组RNA测序（RNA-seq）文件，开展A-to-I编辑相关分析。我们采用如下流程从配对FASTQ格式测序文件中鉴定RNA的A-to-I编辑位点：核心鉴定工具为基于Python开发的SPRINT工具包（SPRINT toolkit），该工具可接收测序文件并输出包含以下信息的编辑位点文本文件：(A) 基因组位置；(B) 编辑类型（例如A-to-G、T-to-C）；链方向（正链+或负链-）；(C) 单个位点的编辑数与该位点总测序读段数。本研究开发了Mathematica软件程序以整合数据：包括共享编辑位点的组别样本量、每个编辑位点的总读段数与编辑数均值，以及组间（例如健康对照HC与重型COVID患者COV-S组）共有的与特有的编辑位点。将上述信息与Alu数据库进行关联，以注释每个编辑位点：包括基因位置（内含子区、非编码RNA区、基因间区、3'非翻译区（3'UTR））以及该位点是否属于Alu元件或非Alu元件。为构建全基因组A-to-I编辑指数，本研究首先鉴定单个样本中所有的A-to-I编辑位点，随后对HC、COV-M、COV-S及FLU-S各组样本的全基因组范围内所有编辑位点的编辑/读段比值进行求和。

数据处理流程：将FASTQ格式的RNA-seq文件导入SPRINT工具包，即可得到包含以下编辑位点信息的文本文件：(A) 基因组位置；(B) 编辑类型（例如A-to-G、T-to-C）；链方向（+或-）；(C) 单个位点的编辑数与总读段数。本研究开发了Mathematica软件程序以整合数据：包括共享编辑位点的组别样本量、每个编辑位点的总读段数与编辑数均值，以及组间（例如健康对照HC与重型COVID患者COV-S组）共有的与特有的编辑位点。将上述信息与Alu数据库进行关联，以注释每个编辑位点：包括基因位置（内含子区、非编码RNA区、基因间区、3'非翻译区（3'UTR））以及该位点是否属于Alu元件或非Alu元件。为构建全基因组A-to-I编辑指数，本研究首先鉴定单个样本中所有的A-to-I编辑位点，随后对HC、COV-M、COV-S及FLU-S各组样本的全基因组范围内所有编辑位点的编辑/读段比值进行求和。参考基因组组装版本：hg19