通过EndoV‒seq对腺嘌呤单碱基编辑器进行全基 因组范围的特异性分析

[Genome-wide profiling of adenine base editor specificity by EndoV-seq](Genome-wide profiling of adenine base editor specificity by EndoV-seq)

题目：通过EndoV‒seq对腺嘌呤单碱基编辑器进行全基 因组范围的特异性分析

作者及单位：

Puping Liang, Xiaowei Xie, Shengyao Zhi, Hongwei Sun, Xiya Zhang, Yu Chen, Yuxi Chen, Yuanyan Xiong, Wenbin Ma, Dan Liu, Junjiu Huang* & Zhou Songyang*

Zhou Songyang

• The First Affiliated Hospital, Sun Yat-sen University; MOE Key Laboratory of Gene Function and Regulation, Guangzhou Key Laboratory of Healthy Aging Research, SYSU-BCM Joint Research Center, School of Life Sciences, Sun Yat-sen University, 510275, Guangzhou, China
• Key Laboratory of Reproductive Medicine of Guangdong Province, School of Life Sciences and the the First Affiliated Hospital, Sun Yat-sen University, 510275, Guangzhou, China
• Verna and Marrs Mclean Department of Biochemistry and Molecular Biology, Baylor College of Medicine, One Baylor Plaza, 77030, Houston, TX, USA
• State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangzhou, 510060, China

发表期刊及时间：

Nature Communicationsvolume 10, Article number: 67 (2019) Published: 08 January 2019

摘要：

The adenine base editor (ABE), capable of catalyzing A•T to G•C conversions, is an important gene editing toolbox. Here, we systematically evaluate genome-wide off-target deamination by ABEs using the EndoV-seq platform we developed. EndoV-seq utilizes Endonuclease V to nick the inosine-containing DNA strand of genomic DNA deaminated by ABE in vitro. The treated DNA is then whole-genome sequenced to identify off-target sites. Of the eight gRNAs we tested with ABE, 2–19 (with an average of 8.0) off-target sites are found, significantly fewer than those found for canonical Cas9 nuclease (7–320, 160.7 on average). In vivo off-target deamination is further validated through target site deep sequencing. Moreover, we demonstrated that six different ABE-gRNA complexes could be examined in a single EndoV-seq assay. Our study presents the first detection method to evaluate genome-wide off-target effects of ABE, and reveals possible similarities and differences between ABE and canonical Cas9 nuclease.

腺嘌呤碱基编辑器（ABE）能够催化A/T转换为G/C，是一种重要的基因编辑工具箱。在这里，我们使用自身 开发的EndoV‒seq平台，系统地评估ABE对全基因组的脱氨作用脱靶效应。 EndoV‒seq利用核酸内切酶V在体 外切割由ABE脱氨基因组DNA的含肌苷的DNA链。然后对处理过的DNA进行全基因组测序，从而鉴定脱靶位 点。在我们用ABE测试的8种gRNA中，发现2‒19（平均8.0）个脱靶位点，显着少于经典Cas9核酸酶产生的脱 靶位点（7‒320，均值为160.7）。体内的脱氨作用脱靶效应通过靶位点深度测序进一步验证。此外，我们证 明了可以在单个EndoV‒seq测定中检查出六种不同的ABE‒gRNA复合物。我们的研究提出了第一种检测ABE全 基因组脱靶效应的检测方法，并揭示了ABE和经典Cas9核酸酶之间可能存在的相似性和差异。

图表选摘：

Introduction

The recently developed targeted base replacement strategy using deaminases holds great promise for treating human diseases caused by pathogenic single nucleotide polymorphisms (SNPs). These RNA-directed programmable base editors can carry out single base pair conversions without inducing double strand breaks (DSBs)1. Cytosine base editors (CBEs) such as base editor 3 (BE3), which catalyze C•G to T•A base pair conversion1, have been successfully used to edit target bases in zebrafish, mouse, and human2,3,4,5,6,7. Base A deamination results in I (inosine) or X (xanthosine), where base I can pair with C and be replicated as G. Adenosine base editors (ABEs) rely on the tRNA-specific adenosine deaminase (TadA) from Escherichia coli to convert A to I on the non-complementary strand, and Cas9 nickase (nCas9) to nick the complementary strand of the target site, thus achieving A•T to G•C pair conversions8. We and others have shown efficient adenine base editing by ABEs in human cells, mouse embryos, and rat embryos8,9,10,11,12,13.

引言：

新近开发的脱氨酶靶向碱基置换策略在治疗致病性单核苷酸多态性(SNPs)引起的人类疾病方面具有广阔 的前景。这些RNA导向的可编程碱基编辑器可以在不引起双链断裂(DSBs)的情况下进行单碱基对转换。 Ctosine碱基编辑器(CBES)，如碱基编辑器3(Be 3)，催化c·g到t·a碱基对转换1，已成功地用于斑马 鱼、小鼠和人2， 3， 4， 5， 6， 7的靶碱基的编辑。碱基脱氨导致I(肌苷)或x(黄嘌呤)，其中碱基I可以与c配对，并被复制为g。腺苷碱基编辑器(ABES)依赖于来自大肠杆菌的tRNA特异性腺苷脱氨酶(Tada)将非互补链上的a转换为I，而Cas9镍酶(Ncas9)则将目标位点的互补链断裂，从而实现a.t到g.c对的转换。我 们和其他人已经证明了ABES在人类细胞、小鼠胚胎和大鼠胚胎8， 9， 10， 11， 12， 13中的有效腺嘌呤 碱基编辑。

pproximately 48% of known pathogenic SNPs may be corrected by A•T-to-G•C conversion, and >20% of these may be targetable with SpCas9-based ABEs, indicating tremendous potential for SpCas9-based ABEs in gene therapy8,14. The advent of xCas9, with its broadened PAM sequence range (5′-NGN, 5′-GAA, 5′-GAT, and 5′-CAA), promises even wider utility of ABE, as more pathogenic G•C-to-A•T SNPs may be corrected by xCas9-ABE14. However, critical questions regarding the specificity and off-target effects of ABEs remain and must be addressed before any possible ==clinical translation==（临床转化）15.

大约48%的已知致病SNPs可以通过a·t-to-g.c转换进行校正，其中>20%的SNPs可能具有基于SpCas9 Abes 的靶向性，表明基于spcase 9的abes在基因治疗方面具有巨大的潜力。 Xcas9的出现扩大了pam序列的范围 (5‘-NGN， 5’-gaa， 5‘-gat， 5’-CAA)，这使得abe的用途更加广泛，因为更具有致病力的g·c-a·t SNPs可能被xcase 9-abe 14校正。然而，关于ABES的特异性和非靶点效应的关键问题仍然存在，必须在任何 可能的临床转化之前加以解决

Digenome-seq has been developed to study genome-wide off-target effects of genome editing tools, where sequencing reads of in vitro processed genomic DNA are mapped to reference genomes with chromosomal sites scored based on DNA reads with identical 5′ or 3′ ends16,17. The method has been successfully used to evaluate genome-wide off-target effects of Cas9, Cpf1, and BE316,17,18,19,20,21,22. Because the enzymes used in these previous reports cannot cleave ABEmodified DNA22, new assays for assessing ABE activities are thus necessary.

二基因组-seq已经被开发用来研究基因组编辑工具对基因组的非目标效应，在这种工具中，体外处理的 基因组DNA的序列读取被映射到参考基因组，根据dna读取的相同5‘或3’端16， 17标记染色体位点。该 方法已成功地用于评估Cas9、 Cpf 1和316、 17、 18、 19、 20、 21、 22等基因的非靶标效应。由于以 前这些报告中使用的酶无法切割ABE修饰的dna 22，因此有必要进行新的评估ABE活性的测试

In this study, we describe a method (EndoV-seq) to investigate ABE specificity genome-wide, where in vitro deaminated genomic DNA is digested with Endonuclease V (EndoV) before being subjected to whole-genome sequencing (WGS). EndoV-seq enables us to evaluate both on-target and offtarget deamination by ABE. We further validate the results through target site deep sequencing to confirm the in vivo specificity of ABE. In addition, our findings show that EndoV-seq is amenable to multiplexing and offers clues to how ABE specificity may be improved.

在本研究中，我们描述了一种 方法(endov-seq)，在进行全基因组测序(WGS)之前，在体外用内切酶v(Endov)消化脱氨基因组DNA。 安托夫-塞克使我们能够评估ABE对目标和非目标的破坏。我们通过靶位点深度测序进一步验证了上述结 果，证实了ABE在体内的特异性。此外，我们的发现表明， endov-seq易于复用，并为如何提高Abe的 特异性提供了线索

图表选析：

image.png

Fig. 1 Using EndoV-seq to evaluate on-target deamination by ABE.

图1. 采用EndoV-seq对ABE系统的中靶效应进行评估

a A flow chart for assessing in vitro ABE off-target effects by EndoV-seq is shown, using sequences from the HEK293-2 site as an example. Genomic DNA is first incubated with recombinant ABE7.10 and the appropriate gRNA and then digested with EndoV, thereby allowing the DNA to be nicked by both nCas9 nickase (black triangle) and EndoV (red triangle, one residue downstream of base I). The cleaved DNA is subsequently fragmented and end repaired for whole-genome sequencing (WGS) with ~30–40 fold coverage.

以hek 293-2位点的序列为例，给出了用endov-seq评价体外ABE脱靶效应的流程图。基因组DNA先与重 组的abe7.10和适当的gRNA共同孵育，再与endov酶切，从而使DNA既被ncas9镍酶(黑三角)和 endov(红三角，碱基I下游的一个残基)切割。切割的DNA随后被碎裂，末端修复为全基因组测序 (WGS)，覆盖范围为30-40倍.

b Genomic DNA of 293T cells was used to PCR amplify regions spanning the HEK293-2 site. The PCR products (100 ng) were incubated with ABE7.10 (300 nM) and HEK293-2 gRNA (900 nM) for 3 h before EndoV (1U) incubation (30 min). The treated products were resolved by agarose gel electrophoresis. Recombinant Cas9 was used as a positive control for DNA cleavage. Molecular weight marker size is in base pairs. Source data are provided as a Source Data file.

用293 t细胞基因组DNA进行PCR扩增，扩增出跨越HEK 293-2位点的区域。 PCR产物(10 0 ng)与 abe7.10(30 0 nm)和hek 2 93-2 gRNA(90 0 nm)共同孵育3 h后(1U)孵育30 min。产物经琼脂糖凝胶电泳 鉴定。以重组Cas9作为DNA裂解的阳性对照。分子量标记大小为碱基对。源数据作为源数据文件提 供。

c Sanger sequencing chromatograms of PCR products amplified from the HEK293-2 gRNA target site using genomic DNA (10 µg) treated with ABE7.10 (300 nM, 8 h) ± EndoV (8U, 3 h). Mock treated genomic DNA served as a control. PAM, blue. Target base A, red and highlighted with red arrow. Peaks on the chromatograph, green for A, red for T, blue for C, and black for G.

用APE7.10(30 0 nm， 8 h) ± endov(8u， 3 h)处理hk 293-2 gRNA靶位点的Sanger测序图谱。模拟处理 的基因组DNA作为对照。帕姆蓝色。目标基地a，红色和突出显示红色箭头。色谱峰， a为绿色， t为红 色， c为蓝色， g为黑色。

d PCR products from cwere deep sequenced. The frequency of each allele is shown on the right. PAM, blue. Target base A, red.

对c基因的PCR产物进行了深度测序。每个等位基因的频率显示在右 边。帕姆蓝色。目标基地a，红色。

e Alignment of whole-genome sequencing reads of the HEK293-2 gRNA target region as visualized by the Integrative Genomics Viewer (IGV). Target base A, red. PAM, blue

整合基因组观察器(IGV)对 hek293-2gRNA靶区进行全基因组测序。目标基地a，红色。帕姆蓝

image.png

Fig. 2 Using EndoV-seq to profile genome-wide off-target deamination by ABE.

图2. 利用EndoV‒seq 通过ABE工具分析全基因组脱氨脱靶效应。

a Genome-wide cleavage scores (cutoff score of >2.5) of genomic DNA treated with Cas9 (blue), BE3 (yellow), or ABE7.10 (coral) using human HBG, VEGFA3, HEK293-2, or mouse Dmd gRNAs. Untreated genomic DNA (gray) served as controls. Red arrows, on-target sites. b Sequence logos of EndoV-captured (ABE7.10) and Digenome-captured (Cas9 and BE3) off-target (with scores of >2.5) and on-target sites of the listed gRNAs. Target sequences are shown with PAM in blue. Note: The length of Dmd gRNA is 19-nt. c Venn diagrams that compare Digenome-captured sites for Cas9 and BE3 with EndoV-seq captured sites of ABE7.10 (score of >0.1 for ABE7.10 and BE3, score of >2.5 for Cas9) are shown for the target sites listed. d HEK-293T cells were co-transfected with vectors encoding ABE7.10 together with HBG gRNA (that targets both HBG1 and HBG2) and VEGFA3 gRNA. At 48 h after transfection, genomic DNA was extracted for PCR amplification and deep sequencing. GFP-transfected cells were used as controls. Error bars represent SEM (n = 3). Statistical significance was calculated using a two-tailed unpaired t-test (p* < 0.001). OT, off-target. OT10 of VEGFA3 failed to be amplified by PCR. Source data are provided as a Source Data file

a 使用人类HBG、VEGFA3、HEK293‒2细胞，或小鼠Dmd gRNA ，用Cas9（蓝色）、BE3（黄色）或 ABE7.10（珊瑚色）处理的基因组DNA 的全基因组切割评分（截止分数> 2.5）。未处理的基因组DNA（灰 色）作为对照。红色箭头表示在靶位点。b EndoV捕获位点（ABE7.10）、双基因组捕获位点（Cas9和 BE3）、脱靶位点（分数> 2.5）和所列gRNA的靶上位点的序列标识。靶序列和PAM序列用蓝色显示。注意： Dmd gRNA 的长度为19‒nt。c 对比Cas9和BE3的双基因组捕获位点与EndoV‒seq捕获的ABE7.10位点 （ABE7.10和BE3的得分> 0.1，Cas9的得分> 2.5）的维恩图显示了列出的目标位点。d 含有编码ABE7.10 基因的载体与HBG gRNA（靶向HBG1和HBG2）和VEGFA3 gRNA 分别共转染进HEK‒293T细胞当中。在转 染后48小时，提取基因组DNA用于PCR扩增和深度测序。GFP转染的细胞用作对照。误差柱代表标准误（n  = 3）。使用双尾非配对t检验计算统计显着性（*** p <0.001）。OT为脱靶（off‒target）。VEGFA3的OT10 未能通过PCR扩增。源数据作为源数据文件提供。

image.png

Fig. 3 Using multiplex EndoV-seq to profile off-target effects of ABE.

图3使用多元EndoV-seq来分析ABE的脱靶效应。

a Human genomic DNA (10 μg) was treated with a mixture of 300 nM ABE7.10 and six gRNAs (200 nM each) (HEK293-2, EMX1, FANCF, HBB-28 (T > C), RNF2 and HBG) for 8 h and then with EndoV (8 U) for 3 h. The treated gnomic DNA was subsequently sequenced with ~30–40-fold coverage and genome-wide cleavage scores calculated. Sites with cleavage score of >2.5 are plotted here. Orange, genomic DNA treated with ABE7.10 and multiplex gRNAs. Gray, untreated genomic DNA. b Genome-wide cleavage scores (cutoff > 2.5) of untreated (gray), monoplex EndoV-seq (blue), and multiplex EndoV-seq (orange) for the indicated gRNAs are plotted. Red arrows, on-target sites. c Sequence logos (by WebLogo) of multiplex EndoV-seq captured off-target (DNA cleavage scores of >2.5) and on-target sites. d A comparison of monoplex vs. multiplex EndoV-seq captured off-target sites (DNA cleavage scores of >2.5, on-target sites not shown)

用300nM ABE7.10和6个gRNA（每个200nM）（HEK293-2，EMX1，FANCF，HBB-28（T> C），RNF2和HBG）的混合物处理人基因组DNA（10μg）。 8小时，然后使用EndoV（8 U）3小时。随后对处理过的基因组 DNA进行测序，覆盖率为~30-40倍，并计算全基因组切割分数。这里绘制了具有> 2.5的解理分数的位点。用ABE7.10和多重gRNA处理的橙色基因组DNA。灰色，未经处理的基因组DNA。 b绘制了针对指定的gRNA的未处理（灰色），monoplex EndoV-seq（蓝色）和多重EndoV-seq（橙色）的全基因组切割评分（截止值> 2.5）。红色箭头，目标网站。 c序列标识（通过WebLogo）多重EndoV-seq捕获的脱靶（DNA切割分数> 2.5）和目标位点。 d monoplex与多重EndoV-seq捕获的脱靶位点的比较（DNA切割得分> 2.5，未显示的靶位点）

image.png

image.png

Fig. 4 The length of gRNAs affects ABE7.10 specificity. a HBG and VEGFA3 gRNAs of different length were designed based on the 20-mer gRNA (GX19) validated in Fig. 3d. Mismatched bases are in lower case. Target bases are in red. PAM sequences are in blue. b HEK-293T cells were co-transfected with the ABE7.10 expression vector and individual HBG or VEGFA3 gRNAs from a. Genomic DNA was then extracted for target site deep sequencing. GFP-transfected cells were used as controls. The frequencies of A-to-G conversion for each gRNA at both on-target and top potential off-target (OT) sites were calculated. OT10 of VEGFA3 was failed to be amplified. Error bars represent SEM (n = 3). Statistical significance was calculated using a two-tailed unpaired t-test. *p < 0.05; *p < 0.01; and ****p < 0.001. Source data are provided as a Source Data file. c The relative activity at each site was calculated by normalizing the A-to-G conversion frequency at that site to the on-target frequency of the GX19 gRNA. The ratios are presented as heat maps where higher values correspond to higher activities

图4. gRNAs 的长度影响 ABE7.10 特异性。 a 基于图 3d 中验证的 20-mer gRNA (GX19)设计不 同长度的 HBG 和 VEGFA3 gRNA。 不匹配的碱基用小写字母表示。目标碱基是红色的。 PAM 序列是蓝色的。 b HEK-293T 细胞与 ABE7.10 表达载体和 a 中的单个 HBG 或 VEGFA3 gRNAs 共转染。 提取基因组 DNA 进行目标位点深度测序。以转染 GFP 的细胞 作为对照。 计算了各 gRNA 在靶上和靶外(OT)点的 A-to-G 转换频率。 VEGFA3 的 OT10 扩增失败。 误差条表示 SEM (n = 3)。统计学显著性采用双尾不配对 t 检验。 * p < 0.05;* * p < 0.01;p < 0.001。源数据作为源数据文件提供。 c 通过将该位点的 A-to-G 转换频率归一 化为 GX19 gRNA 的中靶频率来计算各位点的相对活性。比率以热图的形式表示，其中较 高的值对应较高的活动。

翻译小组：

王俊豪、叶名琛、陈志荣、黄子亮、邓峻玮、郑凌伶