Yang Liu

← Primer

HSC · Healing Source Code

ZFN、TALEN、CRISPR——换了三代编辑器,变的是设计门槛,没变的是那一刀 ZFN, TALEN, CRISPR — three generations of editors: the design barrier changed, the cut did not

2026-10-09 · genome editing, ZFN, TALEN, CRISPR, double-strand break, CCR5, BCL11A

要剪对地方,先得认得路。过去三十年里基因编辑的“剪刀”换了三代,真正在变的,其实只有认路的办法:第一代和第二代用蛋白质去读 DNA,第三代换成了一小段 RNA。至于落刀的那一下,三代几乎一模一样——都是把 DNA 双链切断,再看细胞怎么修。

用蛋白认路

1996 年,Kim、Cha 和 Chandrasegaran 在 PNAS 上把两种不同的 zinc finger 蛋白分别接到 FokI 的切割域上,造出了人工设计的 nuclease,在合适条件下能按序列特异地切开 DNA。2005 年,Urnov 等人在 Nature 上让这类 zinc-finger nuclease(ZFN)对准 X 连锁重症联合免疫缺陷(X-linked SCID)的 IL2RG 突变位点,不经任何筛选,就有超过 18% 的人类细胞被成功改写,约 7% 的细胞两条 X 染色体上都发生了改写。论文的思路是:切断双链,再借助 homology-directed repair,改写人类细胞里的内源基因。

可是每换一个靶点,就得重新设计一个蛋白质的 DNA 结合域。2008 年,Ramirez 等人在 Nat Methods 上发表通讯,题目即指出锌指模块化拼装的失败率出乎预料。

把密码读出来

2009 年底,Boch 等人和 Moscou 与 Bogdanove 在同一期 Science 上读出了 Xanthomonas 细菌 TAL effector 的识别密码:每个重复单元里两个高变氨基酸残基决定识别一个碱基对,一对一,后者还观察到没有明显的上下文依赖。2010 年 Christian 等人把 TALE 接到 FokI 上,做成 TALEN;2011 年 Miller 等人优化了架构,在人的 NTF3 和 CCR5 基因上做出了最高 25% 的编辑效率。设计变得可预测了,但每个靶点仍要造一个新的蛋白。

换一段 RNA

2012 年,Jinek 等人在 Science 上显示,Cas9 由 crRNA 与 tracrRNA 配对而成的双 RNA 引导去切 DNA 双链,二者还能融合成单条 RNA。2013 年,Cong 等人和 Mali 等人同期在 Science 上把它搬进了人类细胞:Cong 等人把多条 guide 编码进同一个 CRISPR array,在同一个基因组里同时编辑多个位点;Mali 等人则在 AAVS1 位点得到 10% 到 25% 的靶向率(293T 细胞),并计算出约 19 万条独特的 gRNA,覆盖约 40.5% 的人类外显子。从这些结果往回看,新增一个靶点所需的工作,从设计一个蛋白质变成了合成一段 RNA。把几代工具放在一起比较,这大概是 CRISPR 被广泛采用的一个关键原因;这是本文的概括,并非上述任何一篇论文自己下的结论。

ZFN 与 CRISPR 在造血干细胞里

本文引用的文献里,TALEN 没有造血干细胞方面的研究,这一节只谈 ZFN 与 CRISPR。其中一条线索是围绕 HIV 的受体 CCR5 展开的。2009 年,Hütter 等人在 N Engl J Med 上报告,一名白血病合并 HIV 感染的患者接受了 CCR5 Δ32/Δ32 供者的干细胞移植,停掉抗病毒治疗 20 个月后仍未出现病毒反弹。2010 年,Holt 等人在 Nat Biotechnol 上用 ZFN 破坏人 CD34⁺ 造血干/祖细胞的 CCR5,平均破坏了 17% 的等位基因;这些细胞移植到免疫缺陷小鼠体内后,在 HIV 压力下被选择性富集。2014 年 Tebas 等人在 N Engl J Med 上则把 ZFN 用在 T 细胞上:12 名患者回输了经 ZFN 修饰的自体 CD4 T 细胞(NCT00842634),其中 11% 到 28% 被修饰;主要终点是安全性,作者的结论是在该研究范围内安全。针对造血干/祖细胞的 ZFN-CCR5 试验也已注册(NCT02500849)。2019 年,Xu 等人在 N Engl J Med 上用 CRISPR 编辑了造血干/祖细胞:患者合并 HIV 与急性淋巴细胞白血病,白血病完全缓解,带 CCR5 缺失的供者细胞存活超过 19 个月,但淋巴细胞里的 CCR5 破坏率只有约 5%,作者认为这需要进一步研究(NCT03164135)。

ZFN 在造血干细胞里还留下了别的路标。2014 年,Genovese 等人在 Nature 上用 ZFN 加上 integrase-defective lentiviral vector,把纠正性 IL2RG cDNA 定点整合进人造血干细胞,小鼠中实现长期多谱系重建;作者同时指出,基因转移的低容许度和 HDR 通路的有限效率,是限制人造血干细胞基因靶向的两个因素。2015 年 Hoban 等人在 Blood 上用夹住镰刀突变的 ZFN 加供体模板,在 CD34⁺ 细胞中实现了校正,移植进小鼠后修饰水平比体外有所下降,而来自镰刀型贫血患者骨髓的细胞则产生了野生型血红蛋白四聚体。

BCL11A 是两种工具靶向同一区域的一个例子。ZFN 疗法 BIVV003 用 ZFN mRNA 破坏 BCL11A 红系 enhancer 里一个 GATAA motif,2024 年 Lessard 等人在 Sci Rep 上报告,I/II 期 PRECIZN-1(NCT03653247)的 7 名镰刀型贫血受试者耐受良好,随访超过 3 个月的 6 人中 5 人总血红蛋白与 HbF 上升、无严重血管阻塞危象。CRISPR 一侧,2021 年 Frangoul 等人在 N Engl J Med 上靶向的是 BCL11A 的红系特异 enhancer;Lessard 等人在论文里写道,exa-cel 靶向同一个位点(即 BCL11A 的红系 enhancer;本文不对 motif 层面的异同下结论),健康供者细胞中约 80% 的等位基因被修改,两名患者(NCT03655678、NCT03745287)脱离输血依赖;2023 年 12 月 8 日,FDA 批准了以此为基础的 Casgevy。这两项研究的规模、人群和随访都不同,本文并列它们,不构成对两种疗法疗效的比较。

赢家的账单

三代编辑器共享的是同一个机制:ZFN 与 TALEN 靠 DNA double-strand break 起作用,Cas9 也是。2018 年 Kosicki 等人在 Nat Biotechnol 上用长读长测序发现,Cas9 造成的断裂常修复为延伸许多 kilobase 的缺失和复杂重排,观察对象是小鼠胚胎干细胞、小鼠造血祖细胞和一个人类分化细胞系。同年,Haapaniemi 等人在永生化的视网膜色素上皮细胞中发现,Cas9 编辑引发 p53 介导的 DNA 损伤反应和细胞周期停滞,从而不利于 p53 通路功能正常的细胞;Ihry 等人在人多能干细胞中发现,Cas9 引起的双链断裂有毒性,且依赖 p53。这些研究针对的是 Cas9,所用细胞不是人造血干细胞;ZFN 与 TALEN 是否在同等程度上引起同样后果,它们没有回答,这些结论在人造血干细胞中的程度也不能由此推出。双链断裂的账,留到讲安全性时再细说。

CRISPR 另有自己的账。2013 年 Hsu 等人在 Nat Biotechnol 上发现,SpCas9 对 guide 与靶点间的错配有序列依赖的容忍,在 293T 与 293FT 细胞中,降低 SpCas9 与 sgRNA 的剂量可以减少脱靶修饰。2019 年 Charlesworth 等人在 Nat Med 上检测到,所检测的捐献者中,58% 有抗 SpCas9 抗体,67% 有抗 SpCas9 的 T 细胞,作者提示这在 CRISPR-Cas9 走向临床时应予考虑。

回头看,三代编辑器是一条线:从用蛋白读 DNA,到用 RNA 读 DNA,换了认路的方式,落刀的那一下一直没变。后面要谈的 base editing 和 prime editing,按各自首篇论文的表述,都是在不切断双链的条件下完成编辑(Komor 等人 2016 年;Anzalone 等人 2019 年)。

参考文献

  1. Kim YG, Cha J, Chandrasegaran S. Hybrid restriction enzymes: zinc finger fusions to Fok I cleavage domain. Proc Natl Acad Sci U S A. 1996;93(3):1156-60. DOI
  2. Urnov FD, et al. Highly efficient endogenous human gene correction using designed zinc-finger nucleases. Nature. 2005;435(7042):646-51. DOI
  3. Ramirez CL, et al. Unexpected failure rates for modular assembly of engineered zinc fingers. Nat Methods. 2008;5(5):374-5. DOI
  4. Boch J, et al. Breaking the code of DNA binding specificity of TAL-type III effectors. Science. 2009;326(5959):1509-12. DOI
  5. Moscou MJ, Bogdanove AJ. A simple cipher governs DNA recognition by TAL effectors. Science. 2009;326(5959):1501. DOI
  6. Christian M, et al. Targeting DNA double-strand breaks with TAL effector nucleases. Genetics. 2010;186(2):757-61. DOI
  7. Miller JC, et al. A TALE nuclease architecture for efficient genome editing. Nat Biotechnol. 2011;29(2):143-8. DOI
  8. Jinek M, et al. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science. 2012;337(6096):816-21. DOI
  9. Cong L, et al. Multiplex genome engineering using CRISPR/Cas systems. Science. 2013;339(6121):819-23. DOI
  10. Mali P, et al. RNA-guided human genome engineering via Cas9. Science. 2013;339(6121):823-6. DOI
  11. Gaj T, Gersbach CA, Barbas CF 3rd. ZFN, TALEN, and CRISPR/Cas-based methods for genome engineering. Trends Biotechnol. 2013;31(7):397-405. DOI
  12. Hütter G, et al. Long-term control of HIV by CCR5 Delta32/Delta32 stem-cell transplantation. N Engl J Med. 2009;360(7):692-8. DOI
  13. Holt N, et al. Human hematopoietic stem/progenitor cells modified by zinc-finger nucleases targeted to CCR5 control HIV-1 in vivo. Nat Biotechnol. 2010;28(8):839-47. DOI
  14. Tebas P, et al. Gene editing of CCR5 in autologous CD4 T cells of persons infected with HIV. N Engl J Med. 2014;370(10):901-10. DOI · NCT00842634
  15. ClinicalTrials.gov. Safety Study of Zinc Finger Nuclease CCR5-modified Hematopoietic Stem/Progenitor Cells in HIV-1 Infected Patients. NCT02500849
  16. Xu L, et al. CRISPR-Edited Stem Cells in a Patient with HIV and Acute Lymphocytic Leukemia. N Engl J Med. 2019;381(13):1240-1247. DOI · NCT03164135
  17. Genovese P, et al. Targeted genome editing in human repopulating haematopoietic stem cells. Nature. 2014;510(7504):235-240. DOI
  18. Hoban MD, et al. Correction of the sickle cell disease mutation in human hematopoietic stem/progenitor cells. Blood. 2015;125(17):2597-604. DOI
  19. Lessard S, et al. Zinc finger nuclease-mediated gene editing in hematopoietic stem cells results in reactivation of fetal hemoglobin in sickle cell disease. Sci Rep. 2024;14(1):24298. DOI · NCT03653247
  20. Frangoul H, et al. CRISPR-Cas9 Gene Editing for Sickle Cell Disease and β-Thalassemia. N Engl J Med. 2021;384(3):252-260. DOI · NCT03655678 · NCT03745287
  21. U.S. Food and Drug Administration. FDA approves first gene therapies to treat patients with sickle cell disease. Press announcement, 8 December 2023. Link
  22. Kosicki M, Tomberg K, Bradley A. Repair of double-strand breaks induced by CRISPR-Cas9 leads to large deletions and complex rearrangements. Nat Biotechnol. 2018;36(8):765-771. DOI
  23. Haapaniemi E, et al. CRISPR-Cas9 genome editing induces a p53-mediated DNA damage response. Nat Med. 2018;24(7):927-930. DOI
  24. Ihry RJ, et al. p53 inhibits CRISPR-Cas9 engineering in human pluripotent stem cells. Nat Med. 2018;24(7):939-946. DOI
  25. Hsu PD, et al. DNA targeting specificity of RNA-guided Cas9 nucleases. Nat Biotechnol. 2013;31(9):827-32. DOI
  26. Charlesworth CT, et al. Identification of preexisting adaptive immunity to Cas9 proteins in humans. Nat Med. 2019;25(2):249-254. DOI
  27. Komor AC, et al. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature. 2016;533(7603):420-4. DOI
  28. Anzalone AV, et al. Search-and-replace genome editing without double-strand breaks or donor DNA. Nature. 2019;576(7785):149-157. DOI

To cut in the right place, you first have to know the way. Over the past thirty years the “scissors” of gene editing have changed three times, and what has really changed is only how they find their way: the first and second generations use proteins to read DNA, the third swapped in a short stretch of RNA. The cut itself is almost the same in all three — the DNA double strand is severed, and then we watch how the cell repairs it.

Finding the way with protein

In 1996, Kim, Cha and Chandrasegaran reported in PNAS that they had fused two different zinc finger proteins to the cleavage domain of FokI, producing designed nucleases that, under suitable conditions, cut DNA in a sequence-specific manner. In 2005, Urnov and colleagues reported in Nature that a zinc-finger nuclease (ZFN) aimed at the IL2RG mutation site of X-linked severe combined immunodeficiency (X-linked SCID) rewrote more than 18% of human cells without any selection, with about 7% of cells modified on both X chromosomes. The logic of the paper: cut the double strand, then use homology-directed repair to rewrite an endogenous gene in human cells.

But every new target meant redesigning the DNA-binding domain of a protein. In 2008, Ramirez and colleagues published a correspondence in Nat Methods whose title itself points to the unexpected failure rates of modular assembly of zinc fingers.

Reading the code out

In late 2009, Boch and colleagues and Moscou and Bogdanove, in the same issue of Science, read out the recognition code of the TAL effectors of Xanthomonas bacteria: two hypervariable amino acid residues in each repeat unit determine recognition of one base pair, one to one, and the latter study also observed no obvious context dependence. In 2010, Christian and colleagues fused TALE to FokI to make TALEN; in 2011, Miller and colleagues optimized the architecture and reached editing efficiencies of up to 25% at the human NTF3 and CCR5 genes. Design had become predictable, but every target still required building a new protein.

Swapping in an RNA

In 2012, Jinek and colleagues showed in Science that Cas9 cuts the DNA double strand under the guidance of a dual RNA made by pairing crRNA with tracrRNA, and that the two could be fused into a single RNA. In 2013, Cong and colleagues and Mali and colleagues, in the same period, brought it into human cells in Science: Cong and colleagues encoded multiple guides in a single CRISPR array and edited several sites in the same genome at once; Mali and colleagues obtained 10% to 25% targeting at the AAVS1 locus (in 293T cells) and calculated about 190,000 unique gRNAs covering about 40.5% of human exons. Looking back from these results, the work needed to add a target went from designing a protein to synthesizing an RNA. Comparing the generations side by side, this is probably one key reason CRISPR was so widely adopted; that is this article’s summary, not a conclusion drawn by any of the papers above.

ZFN and CRISPR in hematopoietic stem cells

Among the literature cited here, there is no TALEN work in hematopoietic stem cells, so this section covers only ZFN and CRISPR. One thread runs around CCR5, the receptor for HIV. In 2009, Hütter and colleagues reported in N Engl J Med that a patient with leukemia and HIV infection who received a stem-cell transplant from a CCR5 Δ32/Δ32 donor showed no viral rebound 20 months after stopping antiretroviral therapy. In 2010, Holt and colleagues in Nat Biotechnol used ZFNs to disrupt CCR5 in human CD34⁺ hematopoietic stem/progenitor cells, disrupting 17% of alleles on average; after transplantation into immunodeficient mice, these cells were selectively enriched under HIV pressure. In 2014, Tebas and colleagues in N Engl J Med applied ZFNs to T cells instead: 12 patients were infused with ZFN-modified autologous CD4 T cells (NCT00842634), of which 11% to 28% were modified; the primary endpoint was safety, and the authors concluded that it was safe within the limits of that study. A ZFN-CCR5 trial in hematopoietic stem/progenitor cells has also been registered (NCT02500849). In 2019, Xu and colleagues in N Engl J Med used CRISPR to edit hematopoietic stem/progenitor cells: the patient had both HIV and acute lymphocytic leukemia, the leukemia reached complete remission, and donor cells carrying the CCR5 disruption survived for more than 19 months, but the CCR5 disruption rate in lymphocytes was only about 5%, which the authors said required further study (NCT03164135).

ZFN left other markers in hematopoietic stem cells. In 2014, Genovese and colleagues in Nature used ZFN together with an integrase-defective lentiviral vector to integrate a corrective IL2RG cDNA site-specifically into human hematopoietic stem cells, achieving long-term multilineage reconstitution in mice; the authors also noted that low tolerance for gene transfer and the limited efficiency of the HDR pathway are two factors constraining gene targeting in human hematopoietic stem cells. In 2015, Hoban and colleagues in Blood used a ZFN flanking the sickle mutation, together with a donor template, to achieve correction in CD34⁺ cells; after transplantation into mice, the modification level was lower than in vitro, while cells from the bone marrow of patients with sickle cell disease produced wild-type hemoglobin tetramers.

BCL11A is an example of two tools targeting the same region. The ZFN therapy BIVV003 uses ZFN mRNA to disrupt a GATAA motif in the BCL11A erythroid enhancer. In 2024, Lessard and colleagues reported in Sci Rep that in the phase I/II PRECIZN-1 study (NCT03653247), 7 participants with sickle cell disease tolerated it well, and among the 6 followed for more than 3 months, 5 had increases in total hemoglobin and HbF and no severe vaso-occlusive crises. On the CRISPR side, in 2021 Frangoul and colleagues in N Engl J Med targeted the erythroid-specific enhancer of BCL11A; Lessard and colleagues wrote in their paper that exa-cel targets the same site (that is, the BCL11A erythroid enhancer; this article draws no conclusion about differences or similarities at the motif level). About 80% of alleles were modified in healthy donor cells, and two patients (NCT03655678, NCT03745287) became transfusion-independent; on 8 December 2023, the FDA approved Casgevy, which is based on this approach. The two studies differ in size, population and follow-up, and this article places them side by side without comparing the efficacy of the two therapies.

The winner’s bill

The three generations share one mechanism: ZFN and TALEN work through a DNA double-strand break, and so does Cas9. In 2018, Kosicki and colleagues in Nat Biotechnol used long-read sequencing to find that breaks made by Cas9 are often repaired as deletions extending many kilobases and as complex rearrangements, in mouse embryonic stem cells, mouse hematopoietic progenitors and a human differentiated cell line. In the same year, Haapaniemi and colleagues found in immortalized retinal pigment epithelial cells that Cas9 editing triggers a p53-mediated DNA damage response and cell-cycle arrest, which acts against cells with a functional p53 pathway; Ihry and colleagues found in human pluripotent stem cells that Cas9-induced double-strand breaks are toxic, and that this depends on p53. These studies addressed Cas9, and the cells used were not human hematopoietic stem cells; whether ZFN and TALEN cause the same consequences to the same degree is something they do not answer, and the extent of these findings in human hematopoietic stem cells cannot be inferred from them. The accounting for double-strand breaks is left for the discussion of safety.

CRISPR has a bill of its own. In 2013, Hsu and colleagues found in Nat Biotechnol that SpCas9 tolerates mismatches between guide and target in a sequence-dependent way, and that in 293T and 293FT cells, lowering the doses of SpCas9 and sgRNA can reduce off-target modification. In 2019, Charlesworth and colleagues in Nat Med detected that among the donors tested, 58% had antibodies against SpCas9 and 67% had T cells against SpCas9, and the authors suggested that this should be taken into account as CRISPR-Cas9 moves toward the clinic.

Looking back, the three generations form a single line: from reading DNA with proteins to reading DNA with RNA, the way of finding the target changed, while the cut itself never did. The base editing and prime editing to be discussed later, by the accounts in their respective founding papers, both carry out editing without cutting the double strand (Komor and colleagues, 2016; Anzalone and colleagues, 2019).

References

  1. Kim YG, Cha J, Chandrasegaran S. Hybrid restriction enzymes: zinc finger fusions to Fok I cleavage domain. Proc Natl Acad Sci U S A. 1996;93(3):1156-60. DOI
  2. Urnov FD, et al. Highly efficient endogenous human gene correction using designed zinc-finger nucleases. Nature. 2005;435(7042):646-51. DOI
  3. Ramirez CL, et al. Unexpected failure rates for modular assembly of engineered zinc fingers. Nat Methods. 2008;5(5):374-5. DOI
  4. Boch J, et al. Breaking the code of DNA binding specificity of TAL-type III effectors. Science. 2009;326(5959):1509-12. DOI
  5. Moscou MJ, Bogdanove AJ. A simple cipher governs DNA recognition by TAL effectors. Science. 2009;326(5959):1501. DOI
  6. Christian M, et al. Targeting DNA double-strand breaks with TAL effector nucleases. Genetics. 2010;186(2):757-61. DOI
  7. Miller JC, et al. A TALE nuclease architecture for efficient genome editing. Nat Biotechnol. 2011;29(2):143-8. DOI
  8. Jinek M, et al. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science. 2012;337(6096):816-21. DOI
  9. Cong L, et al. Multiplex genome engineering using CRISPR/Cas systems. Science. 2013;339(6121):819-23. DOI
  10. Mali P, et al. RNA-guided human genome engineering via Cas9. Science. 2013;339(6121):823-6. DOI
  11. Gaj T, Gersbach CA, Barbas CF 3rd. ZFN, TALEN, and CRISPR/Cas-based methods for genome engineering. Trends Biotechnol. 2013;31(7):397-405. DOI
  12. Hütter G, et al. Long-term control of HIV by CCR5 Delta32/Delta32 stem-cell transplantation. N Engl J Med. 2009;360(7):692-8. DOI
  13. Holt N, et al. Human hematopoietic stem/progenitor cells modified by zinc-finger nucleases targeted to CCR5 control HIV-1 in vivo. Nat Biotechnol. 2010;28(8):839-47. DOI
  14. Tebas P, et al. Gene editing of CCR5 in autologous CD4 T cells of persons infected with HIV. N Engl J Med. 2014;370(10):901-10. DOI · NCT00842634
  15. ClinicalTrials.gov. Safety Study of Zinc Finger Nuclease CCR5-modified Hematopoietic Stem/Progenitor Cells in HIV-1 Infected Patients. NCT02500849
  16. Xu L, et al. CRISPR-Edited Stem Cells in a Patient with HIV and Acute Lymphocytic Leukemia. N Engl J Med. 2019;381(13):1240-1247. DOI · NCT03164135
  17. Genovese P, et al. Targeted genome editing in human repopulating haematopoietic stem cells. Nature. 2014;510(7504):235-240. DOI
  18. Hoban MD, et al. Correction of the sickle cell disease mutation in human hematopoietic stem/progenitor cells. Blood. 2015;125(17):2597-604. DOI
  19. Lessard S, et al. Zinc finger nuclease-mediated gene editing in hematopoietic stem cells results in reactivation of fetal hemoglobin in sickle cell disease. Sci Rep. 2024;14(1):24298. DOI · NCT03653247
  20. Frangoul H, et al. CRISPR-Cas9 Gene Editing for Sickle Cell Disease and β-Thalassemia. N Engl J Med. 2021;384(3):252-260. DOI · NCT03655678 · NCT03745287
  21. U.S. Food and Drug Administration. FDA approves first gene therapies to treat patients with sickle cell disease. Press announcement, 8 December 2023. Link
  22. Kosicki M, Tomberg K, Bradley A. Repair of double-strand breaks induced by CRISPR-Cas9 leads to large deletions and complex rearrangements. Nat Biotechnol. 2018;36(8):765-771. DOI
  23. Haapaniemi E, et al. CRISPR-Cas9 genome editing induces a p53-mediated DNA damage response. Nat Med. 2018;24(7):927-930. DOI
  24. Ihry RJ, et al. p53 inhibits CRISPR-Cas9 engineering in human pluripotent stem cells. Nat Med. 2018;24(7):939-946. DOI
  25. Hsu PD, et al. DNA targeting specificity of RNA-guided Cas9 nucleases. Nat Biotechnol. 2013;31(9):827-32. DOI
  26. Charlesworth CT, et al. Identification of preexisting adaptive immunity to Cas9 proteins in humans. Nat Med. 2019;25(2):249-254. DOI
  27. Komor AC, et al. Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature. 2016;533(7603):420-4. DOI
  28. Anzalone AV, et al. Search-and-replace genome editing without double-strand breaks or donor DNA. Nature. 2019;576(7785):149-157. DOI