【作者】 杨杰；

【导师】 戴玉杰； 毕昌昊；

【作者基本信息】 天津科技大学 ， 生物工程（专业学位）， 2023， 硕士

【摘要】 新型糖基化酶碱基编辑器(Glycosylase base editor,GBE)可实现C-G的碱基颠换,但存在编辑效率较低的问题,而C-G碱基颠换被报道依赖于DNA的修复机制。因此,我们对DNA修复相关的细胞因子进行筛选,以期能提高GBE的编辑性能。首先,我们建立了稳定表达CBE编辑器的稳转HEK293T细胞系,并借由CBE诱导的基因敲除库对980个与DNA修复、代谢相关的细胞因子进行筛选。研究结果显示有20个细胞因子可能对GBE的编辑效率具有提升作用,随后,我们通过Golden Gate克隆方法将筛选的20个细胞因子融合表达在GBE的N端构建了CF(Cell Factors)-GBE。对HEK293T细胞内源性的两个位点研究结果显示,相对于GBE(20%)其中CHAF1B-GBE(44.45%)、HMGN1-GBE(32.883%)、APEX1-GBE(34.5%)、EMG1-GBE(30%)的平均编辑效率明显提升;多位点的进一步的研究结果显示,相对于GBE(12.79%),CHAF1B-GBE(27.89%)、HMGN1-GBE(26.32%)、APEX1-GBE(26.82%)、EMG1-GBE(21.86%)平均编辑效率显著提升(P<0.05),而编辑纯度没有显著改变(P>0.05)。对Indel副产物的统计结果分析显示,在RP11-177B4-3、HHEK4-2位点Indels出现显著性升高(P<0.05),在HIRA-1、HEK2-2位点Indels出现显著性降低(P<0.05);为进一步证实CF-GBE的编辑优势,我们采用慢病毒整合的方式构建了一个包含9000种g RNA及其靶位点的g RNA细胞文库,并在HEK293T细胞中进行验证。结果显示,相对于GBE,CHAF1B-GBE、HMGN1-GBE、APEX1-GBE和EMG1-GBE平均编辑效率分别提升了2.27、2.13、2.52和1.67倍,编辑窗口没有发生改变;最后,我们通过AI(Artificial Intelligence)对g RNA文库的编辑结果数据进行深度学习,针对每种CF-GBEs构建了可以预测靶位点编辑效率的卷积神经网络(Convolutional Neural Networks,CNN)模型,Pearson相关性分析结果显示CHAF1B-GBE、HMGN1-GBE、APEX1-GBE、EMG1-GBE编辑器的r值分别为0.93、0.79、0.83、0.89。随后在6个基因组位点对其预测精准度做了验证,结果显示,CNN模型可以对这些靶位点的编辑效率精确预测。本研究通过高通量碱基编辑敲除筛选的方法对影响GBE编辑性能的细胞因子进行了筛选,将筛选到的细胞因子与GBE融合表达构建了高编辑效率的CF-GBEs,并针对CF-GBEs,构建了可以精准预测编辑效率的CNN模型,极大的增加了GBE在科学研究以及基因治疗领域的应用性。更多还原

【Abstract】 A novel glycosylase base editor(GBE)has been developed to achieve C-G base conversion,but it suffers from low editing efficiency.C-G base conversion has been reported to rely on DNA repair mechanisms.Therefore,we conducted a screening of DNA repair-related cellular factors to enhance the editing performance of GBE.First,we established a stable HEK293 T cell line expressing the cytidine base editor(CBE)and used a CBE-induced gene knockout library to screen 980 cellular factors associated with DNA repair and metabolism.The research results identified 20 cellular factors that potentially enhance the editing efficiency of GBE.Subsequently,we utilized the Golden Gate cloning method to fuse the selected 20 cellular factors to the N-terminus of GBE,constructing CF(Cell Factors)-GBE.The study results on two endogenous loci in HEK293 T cells showed a significant improvement in average editing efficiency for CHAF1B-GBE(44.45%),HMGN1-GBE(32.883%),APEX1-GBE(34.5%),and EMG1-GBE(30%)compared to GBE(20%).Further investigation on multiple loci demonstrated a significant increase in average editing efficiency for CHAF1B-GBE(27.89%),HMGN1-GBE(26.32%),APEX1-GBE(26.82%),and EMG1-GBE(21.86%)compared to GBE(12.79%)(P<0.05),while the editing purity did not show significant changes(P>0.05).Statistical analysis of indel byproducts revealed a significant increase in RP11-177B4-3 and HHEK4-2 loci(P<0.05)and a significant decrease in HIRA-1 and HEK2-2 loci(P<0.05).To further validate the editing advantages of CF-GBE,we constructed a g RNA library containing 9,000 g RNAs and their target sites through lentiviral integration and performed verification in HEK293 T cells.The results showed that CHAF1B-GBE,HMGN1-GBE,APEX1-GBE,and EMG1-GBE exhibited average editing efficiency improvements of 2.27,2.13,2.52,and 1.67 times,respectively,compared to GBE,while the editing window remained unchanged.Finally,we employed Artificial Intelligence(AI)to conduct deep learning on the editing data of the g RNA library and developed Convolutional Neural Networks(CNN)models to predict target site editing efficiency for each CF-GBE.Pearson correlation analysis showed that the r-values for CHAF1B-GBE,HMGN1-GBE,APEX1-GBE,and EMG1-GBE editors were 0.93,0.79,0.83,and 0.89,respectively.Subsequently,we validated the predictive accuracy of the CNN models on six genomic loci,demonstrating precise prediction of editing efficiency for these target sites.This study employed a high-throughput base editing knockout screening method to identify cellular factors influencing GBE editing performance.The identified cellular factors were fused with GBE to construct high-efficiency CF-GBEs.Furthermore,CNN models were developed to accurately predict editing efficiency for CF-GBEs,greatly enhancing the applicability of GBE in scientific research and gene therapy fields.更多还原