Yang Liu

← Primer

HSC · Healing Source Code

最安全的突变是人群已经试过的——HBG 启动子编辑，以及它写在两处的代价 The safest mutation is one a population has already tested — HBG promoter editing, and the price of a sequence written twice

2026-09-27 · HBG1, HBG2, HPFH, 启动子编辑, 碱基编辑, 胎儿血红蛋白, BCL11A, ZBTB7A

一个做基因编辑的人最怕被问到的问题，不是「你能不能改这个碱基」，而是「改完之后，这个人往后六十年会怎样」。前者是技术问题，实验台上能回答；后者是流行病学问题，实验台上永远回答不了——你没有六十年，也没有一万个受试者。

所以有一类靶点选择显得格外聪明：不去发明一个新的改动，而是去抄一个人群里已经存在的。

遗传性胎儿血红蛋白持续症（hereditary persistence of fetal hemoglobin，HPFH）就是这样一份现成的答卷。正常人出生后，两个 γ-globin 基因 HBG1 与 HBG2 会被逐步关掉，成人型 β-globin 接手，红细胞里的血红蛋白从胎儿型（HbF）换成成人型（HbA）。HPFH 携带者身上这个切换没有完成，他们成年后仍然大量表达 γ-globin，而且——这是关键——基本没事。更有意思的是，当 HPFH 与镰刀型贫血或 β-thalassemia 的致病突变同时遗传给一个人时，后者的临床表现会被明显减轻。2016 年 Traxler 等在《Nature Medicine》上把这一层逻辑写得很直白：这些疾病在出生后才发病，正是因为 HbF 退场、HbA 登场；只要把 HbF 留住，症状就会缓解。

于是问题变成一个纯工程问题：HPFH 携带者的基因组里，那个「留住 HbF」的改动具体写在哪个位置？把同一个改动写进病人的造血干细胞，能不能复现同一个良性表型？

抄写之前，先要知道抄哪几个字母

HPFH 有好几种，有的是大片段缺失，有的只是启动子上换了一个碱基。落到 HBG1/HBG2 启动子上的那一类，点突变并不散落，而是明显地聚成两簇：一簇在转录起始位点上游约 115 bp 处，一簇在约 200 bp 处。

为什么恰好是这两处，2018 年有两组人几乎同时给出了机制层面的答案，而且答案彼此咬合。

Martyn、Wienert 等人与 Crossley 实验室在《Nature Genetics》上报告，胎儿血红蛋白的两个主要抑制因子 BCL11A 与 ZBTB7A（又名 LRF）分别直接结合在 −115 和 −200 这两个位置；用 CRISPR-Cas9 把天然存在的 HPFH 突变引入红系细胞，抑制因子的结合被破坏，γ-globin 表达随之升高。换句话说，HPFH 的点突变不是玄学，它们是抑制因子落脚点上的几个错字。

同一年，Liu 等人在 Orkin 实验室完成的工作从另一头切进来，发表在《Cell》上。他们先用功能筛选和蛋白结合微阵列确定 BCL11A 的偏好识别序列是 TGACCA，然后发现一件很漂亮的事：这个基序出现在人和小鼠所有胚胎型、胎儿型 globin 基因的启动子里，而成人型基因的启动子里一个都没有——而且在 γ-globin 启动子里，它出现了两次，位于 −118 到 −113 与 −91 到 −86。用 CUT&RUN 在红系细胞里画出结合位置，BCL11A 偏好占据的是靠远端的那一个。而恰恰是这个远端基序，在罕见的 HPFH 个体身上被 −117、−114 的单碱基替换、或者一段 13 bp 的缺失改掉了。

这两篇一起把「BCL11A 是通过增强子间接调控」这个含混的印象收紧成了一句更强的结论：BCL11A 直接坐在 γ-globin 启动子上。这句话对治疗设计的意义是直接的——你不必绕远路去动那个转录因子本身，可以就在目标基因自己的启动子上动手。

从鉴定到复刻

有了坐标，复刻工作在几年里沿着两条精度不同的路推进。

先是核酸酶。Traxler 等人用 CRISPR-Cas9 在人血液祖细胞里破坏那段同时存在于 HBG1 与 HBG2 启动子中的 13 nt 序列，复刻天然的 HPFH 突变；编辑后的祖细胞分化出的红细胞 HbF 升高，而且升高的幅度足以抑制镰刀型贫血细胞在低氧下出现的病理形态。同一年，Ye 等人在 Kan 实验室把缺失型 HPFH 基因型做进人的造血干/祖细胞，由此分化的红系细胞 γ-globin 表达显著升高。这两篇都是细胞层面的概念验证，但方向已经很清楚。

2019 年，Métais 等人在《Blood Advances》上把这件事推到了干细胞层面：他们在造血干细胞里破坏 HBG1/HBG2 启动子上 −118 到 −114 的 BCL11A 结合位点，把编辑后的细胞移植进小鼠，16 周后既没看到造血受损，也没检出脱靶突变，重建出的红细胞 HbF 达到了被认为有治疗意义的量级。

Miccio 实验室走的是另一簇坐标。2020 年 Weber 等人在《Science Advances》上报告，他们用 CRISPR-Cas9 在 HBG 启动子上制造插入与缺失，模拟 HPFH 突变，破坏已知与推测的抑制因子结合位点；其中破坏 LRF 结合位点这一种，在病人来源的造血干/祖细胞里让 γ-globin 去抑制，并纠正了镰刀化表型，而且异种移植实验显示，真正有重建能力的那部分干细胞里编辑效率也很高。「LRF 位点是一个有力靶点」这个结论，是从这里立住的。

抄到后来，就不只是抄了

碱基编辑进场之后，事情变得更有意思。核酸酶复刻 HPFH 的方式其实很粗——它靠双链断裂后的修复留下一串长度不定的插入缺失，只是碰巧把那个结合位点打坏了。而 HPFH 的天然突变往往是单个碱基的替换，要真正做到「照抄」，需要的是精确的碱基改写。

2022 年 Antoniou 等人在《Nature Communications》上用碱基编辑在 HBG 启动子的 −200 区造出了一系列突变，包括把四到八个诱导 γ-globin 的突变组合在一起。他们的结论里有一条超出了「照抄」的范畴：在这个区域里，最有效的策略不是破坏 LRF 的结合，而是造出一个 KLF1 激活因子的结合位点——而这个效果在长期重建型造血干/祖细胞里同样成立。

2023 年，Mayuranathan 等人在《Nature Genetics》上把这条思路推到了另一个位点：用腺嘌呤碱基编辑器改写 γ-globin 启动子 −175 位，生成一个 TAL1–GATA1 结合基序，从而促进启动子与增强子的互作。在纯合编辑的红系克隆里，HbF 占比达到 81% ± 7%，而未编辑对照是 17% ± 11%。

到这一步，「抄自然」这个说法需要被修订。HPFH 提供的不是一份必须逐字照抄的答案，而是一个已经被人群体验过的邻域：这一小段序列可以被改动，改动之后的表型是良性的。工程上真正在做的，是在这个被验证过是安全的邻域里，挑一个比天然突变更有效的写法。这与「完全新发明一个改动」仍然有本质区别，但也不再等同于「人群已经替我们跑完了安全性试验」——待核的那一部分，随着改动离天然版本越远，又长回来了一点。

这笔交易的价格：同一句话被写了两遍

HBG 启动子路线最硬的一道边界，不在效率上，而在基因组结构上。

HBG1 与 HBG2 是一对高度同源的旁系同源基因，两者启动子上的那段目标序列几乎一模一样——这正是 Traxler 等人当年可以用一个 13 nt 靶点同时覆盖两个基因的原因。便利的另一面是：一条 gRNA 切下去，切的是两处。两个断裂之间的片段如果在修复中被整段丢掉，结果就是一个 4.9 kb 的缺失，把 HBG2 整个抹掉。Antoniou 等人在同一篇文章里明确记录了这一点，并且把它当作核酸酶路线与碱基编辑路线的分水岭：4.9 kb 缺失在 Cas9 处理的细胞里出现，在以 RNA 递送的碱基编辑样本里频率可以忽略——因为碱基编辑不制造那个双链断裂。

这不是一句「还需要进一步研究」可以带过的注脚。它至少带来三层麻烦。其一，被删掉的不是随机序列，而是一个本来就要表达的 γ-globin 拷贝，所以「删除」与「治疗」在同一个位点上纠缠：你想留住 HbF，却可能删掉产生 HbF 的一个基因。其二，两个拷贝的存在让「编辑效率」这个数字变得难读——等位基因层面的编辑状态有多种组合，而不同组合的 HbF 产出并不相同，这也是 Mayuranathan 等人观察到的「Cas9 策略下 HbF 水平更低且克隆间差异更大」的部分来源。其三，重复序列之间的断裂修复，本身就是产生大片段重排的经典温床，而移植后这些细胞要在人体内自我更新几十年，任何大结构变化的克隆行为都需要被长期追踪，而不是靠 16 周的小鼠数据结案。

顺着这条逻辑，启动子路线与敲掉 BCL11A 红系增强子这条更早成熟的路线，取舍就清楚了。增强子路线动的是一个转录因子在红系里的表达量，序列改动本身落在一个非重复区域，干净；但被下调的是一个有多种功能的调控蛋白，影响面在 γ-globin 之外还有多少，需要靠证据一点点圈定。启动子路线动的是目标基因自己的顺式元件，改动的后果原则上被限制在 β-globin 位点内部，而且有 HPFH 人群作为表型参照；代价是这个元件恰好坐在一段重复序列里。一边是「改动很干净，但下游影响面要证明」，另一边是「下游影响面很窄，但改动本身不干净」。Mayuranathan 等人把两条路并排比较后给出的判断是：HBG 启动子上的碱基编辑诱导 HbF 更强、克隆间更均一，强于针对启动子 BCL11A 基序的 Cas9 策略，也强于针对 BCL11A 红系增强子的 Cas9 策略。

人体这一关，现在有三个版本同时在过

2023 年，Sharma 等人在《New England Journal of Medicine》上报告了用 CRISPR-Cas9 编辑 HBG1/HBG2 启动子治疗镰刀型贫血的首批临床结果。此后两年，同一个靶点上又开出两条不同工具的路：2026 年 Hanna 等人报告了用 CRISPR-Cas12a 编辑 HBG1/HBG2 启动子的结果，Gupta 等人则报告了碱基编辑版本——后者纳入 31 名 12 至 35 岁有重型病史的患者，移植后约三周完成造血重建，在最后一次红细胞输注 60 天之后，没有再出现研究者报告的重度血管阻塞危机；不良事件在参与者中普遍存在。

三篇论文，同一段启动子，三种写字的工具。这件事本身就说明这个靶点的选择逻辑站住了：当多个团队拿着不同的酶去改同一段 200 bp 的序列时，他们真正共享的不是工具，而是 HPFH 携带者提供的那份表型保证。

把未知的安全性换成一份已经被人群携带了几代的安全性，是基因治疗里少有的、可以先看答案再动手的机会。但这份答案被写在了一段重复两遍的序列上——你照抄的时候，笔尖同时落在两处。这笔交易有它自己的价格，而价格恰好藏在让这笔交易变得方便的那个特征里。

参考文献

  1. Martyn GE, Wienert B, Yang L, Shah M, Norton LJ, Burdach J, Kurita R, Nakamura Y, Pearson RCM, Funnell APW, Quinlan KGR, Crossley M. Natural regulatory mutations elevate the fetal globin gene via disruption of BCL11A or ZBTB7A binding. Nat Genet. 2018;50(4):498–503. DOI
  2. Liu N, Hargreaves VV, Zhu Q, Kurland JV, Hong J, Kim W, Sher F, Macias-Trevino C, Rogers JM, Kurita R, Nakamura Y, Yuan GC, Bauer DE, Xu J, Bulyk ML, Orkin SH. Direct promoter repression by BCL11A controls the fetal to adult hemoglobin switch. Cell. 2018;173(2):430–442.e17. DOI
  3. Traxler EA, Yao Y, Wang YD, Woodard KJ, Kurita R, Nakamura Y, Hughes JR, Hardison RC, Blobel GA, Li C, Weiss MJ. A genome-editing strategy to treat β-hemoglobinopathies that recapitulates a mutation associated with a benign genetic condition. Nat Med. 2016;22(9):987–990. DOI
  4. Ye L, Wang J, Tan Y, Beyer AI, Xie F, Muench MO, Kan YW. Genome editing using CRISPR-Cas9 to create the HPFH genotype in HSPCs: an approach for treating sickle cell disease and β-thalassemia. Proc Natl Acad Sci U S A. 2016;113(38):10661–10665. DOI
  5. Métais JY, Doerfler PA, Mayuranathan T, Bauer DE, Fowler SC, Hsieh MM, et al. Genome editing of HBG1 and HBG2 to induce fetal hemoglobin. Blood Adv. 2019;3(21):3379–3392. DOI
  6. Weber L, Frati G, Felix T, Hardouin G, Casini A, Wollenschlaeger C, et al. Editing a γ-globin repressor binding site restores fetal hemoglobin synthesis and corrects the sickle cell disease phenotype. Sci Adv. 2020;6(7):eaay9392. DOI
  7. Antoniou P, Hardouin G, Martinucci P, Frati G, Felix T, Chalumeau A, et al. Base-editing-mediated dissection of a γ-globin cis-regulatory element for the therapeutic reactivation of fetal hemoglobin expression. Nat Commun. 2022;13:6618. DOI
  8. Mayuranathan T, Newby GA, Feng R, Yao Y, Mayberry KD, Lazzarotto CR, et al. Potent and uniform fetal hemoglobin induction via base editing. Nat Genet. 2023;55(7):1210–1220. DOI
  9. Sharma A, Boelens JJ, Cancio M, Hankins JS, Bhad P, Azizy M, et al. CRISPR-Cas9 editing of the HBG1 and HBG2 promoters to treat sickle cell disease. N Engl J Med. 2023;389(9):820–832. DOI
  10. Hanna R, Frangoul H, Pineiro L, McKinney C, Mapara M, Dalal J, et al. CRISPR-Cas12a gene editing of HBG1 and HBG2 promoters to treat sickle cell disease. N Engl J Med. 2026;394(13):1281–1291. DOI
  11. Gupta AO, Sharma A, Frangoul H, Kanter J, Mapara MY, Dalal J, et al. Base editing of HBG1 and HBG2 promoters for sickle cell disease. N Engl J Med. 2026;394(18):1824–1835. DOI

The question a genome editor least wants to be asked is not “can you change that base.” That one can be answered at the bench. The question is “what happens to this person over the next sixty years” — and that is an epidemiological question, not a bench question. You do not have sixty years, and you do not have ten thousand subjects.

Which is why one particular style of target selection looks so shrewd: rather than inventing a new change, copy one that a human population already carries.

Hereditary persistence of fetal hemoglobin (HPFH) is exactly that kind of ready-made answer. After birth, the two γ-globin genes HBG1 and HBG2 are progressively shut off, adult β-globin takes over, and the hemoglobin inside red cells switches from the fetal form (HbF) to the adult form (HbA). In people with HPFH that switch never completes: they keep expressing γ-globin into adult life, and — this is the crucial part — they are essentially healthy. Better still, when HPFH is co-inherited with a sickle cell or β-thalassemia mutation, the clinical severity of the hemoglobinopathy is measurably reduced. Traxler and colleagues laid the logic out plainly in Nature Medicine in 2016: these disorders become symptomatic postnatally precisely because HbF gives way to HbA, and they are alleviated when γ-globin expression is maintained.

That reframes the therapeutic problem as a purely engineering one. Where, exactly, in an HPFH carrier’s genome is the change that keeps HbF switched on? And if you write the same change into a patient’s hematopoietic stem cells, do you reproduce the same benign phenotype?

Knowing which letters to copy

HPFH comes in several flavors. Some forms are large deletions; others are single base substitutions in a promoter. The promoter class is striking because the point mutations are not scattered — they cluster in two places, roughly 115 and roughly 200 base pairs upstream of the γ-globin transcription start sites.

Why those two spots in particular got mechanistic answers in 2018, from two groups working almost simultaneously, and the two answers interlock.

Martyn, Wienert and colleagues in the Crossley lab reported in Nature Genetics that the two major fetal globin repressors, BCL11A and ZBTB7A (also called LRF), bind directly at the −115 and −200 sites respectively. Introducing the naturally occurring HPFH mutations into erythroid cells by CRISPR-Cas9 disrupted repressor binding and raised γ-globin expression. The HPFH point mutations, in other words, are not mysterious: they are typos in a repressor’s footprint.

The same year, Liu and colleagues in the Orkin lab came at it from the protein side, in Cell. Using a functional assay and a protein binding microarray they established that BCL11A’s preferred recognition sequence is TGACCA — and then found something elegant. That motif appears in the promoters of every embryonic and fetal-expressed globin gene in humans and mice and in none of the adult-expressed ones. In the γ-globin promoters it appears twice, at −118 to −113 and at −91 to −86. CUT&RUN mapping in erythroid cells showed BCL11A occupying preferentially the more distal of the two. And it is exactly that distal motif that is altered in rare HPFH individuals, by single base substitutions at −117 and −114 or by a 13-bp deletion.

Together the two papers tightened a vague impression — that BCL11A acts indirectly, through enhancers — into a sharper claim: BCL11A sits directly on the γ-globin promoter. For therapeutic design the consequence is immediate. You do not have to take the long route through the transcription factor itself; you can work on the target gene’s own promoter.

From mapping to copying

With coordinates in hand, the copying proceeded along two tracks of increasing precision.

Nucleases came first. Traxler and colleagues used CRISPR-Cas9 in human blood progenitors to mutate the 13-nucleotide sequence present in both the HBG1 and HBG2 promoters, recapitulating the natural HPFH mutation. Edited progenitors produced red cells with elevated HbF, and the elevation was enough to inhibit the pathological hypoxia-induced sickling morphology. The same year, Ye and colleagues in the Kan lab engineered the deletional HPFH genotype into human hematopoietic stem and progenitor cells; erythroid cells derived from them showed significantly higher γ-globin expression. Both were cell-level proofs of concept, but the direction was clear.

In 2019, Métais and colleagues pushed it to the stem cell level in Blood Advances. They disrupted the BCL11A binding site at −118 to −114 in the HBG1/HBG2 promoters of hematopoietic stem cells and transplanted the edited cells into mice; sixteen weeks later there was no detectable impairment of hematopoiesis and no off-target mutations, and red cells derived from the edited stem cells carried HbF at levels considered potentially therapeutic.

The Miccio lab worked the other cluster. In 2020, Weber and colleagues reported in Science Advances that they used CRISPR-Cas9 to generate insertions and deletions in the HBG promoters, mimicking HPFH mutations and disrupting known and putative repressor binding sites. Disrupting the LRF binding site in patient-derived HSPCs derepressed γ-globin and corrected the sickling phenotype, and xenotransplantation showed high editing efficiency in the repopulating stem cell compartment. The conclusion that the LRF site is a potent target was established there.

At some point it stops being copying

Base editing made the story more interesting. Nuclease-based copying is in fact quite crude: it relies on double-strand break repair to leave a variable-length indel that happens to wreck the binding site. Natural HPFH mutations are usually single base substitutions. Literal copying requires literal base rewriting.

In 2022, Antoniou and colleagues used base editing in Nature Communications to generate a panel of mutations in the −200 region of the HBG promoters, including combinations of four to eight γ-globin-inducing mutations. One of their conclusions steps outside the copying frame entirely: in that region, the most potent strategy was not disrupting LRF binding but creating a binding site for the KLF1 activator — and the effect held even in long-term repopulating HSPCs.

In 2023, Mayuranathan and colleagues carried the idea to a different position in Nature Genetics. Adenine base editing rewrote γ-globin promoter position −175 (A>G), generating a TAL1–GATA1 binding motif that stimulates promoter–enhancer interactions. Homozygous −175A>G edited erythroid colonies expressed 81 ± 7% HbF, against 17 ± 11% in unedited controls.

At that point “copying nature” needs revising. HPFH does not supply an answer to be transcribed verbatim; it supplies a neighborhood that a population has already lived in — a short stretch of sequence that can be altered, with benign consequences. What engineering actually does is pick, inside that validated neighborhood, a rewrite more potent than the natural variant. That is still categorically different from inventing a change from scratch. But it is no longer the same as saying the population already ran the safety trial for you. The further the edit sits from the natural version, the more of that unknown safety grows back.

The price: the same sentence written twice

The hardest boundary on the HBG promoter route is not efficiency. It is genome architecture.

HBG1 and HBG2 are highly homologous paralogs, and the targeted stretch of their promoters is nearly identical — which is exactly why Traxler and colleagues could cover both genes with one 13-nucleotide target. The other face of that convenience: one guide RNA cuts in two places. If the fragment between the two breaks is lost during repair, the result is a 4.9-kb deletion that encompasses the HBG2 gene. Antoniou and colleagues documented this explicitly and treated it as the watershed between nuclease and base editing approaches: the 4.9-kb deletion was detected in Cas9-treated cells, while in RNA-delivered base-edited samples its frequency was negligible — because base editors do not make that double-strand break.

This is not a footnote to be waved off as needing further study. It carries at least three kinds of trouble. First, what gets deleted is not filler sequence but one of the two γ-globin copies you are trying to switch on, so deletion and therapy are entangled at the same locus. Second, having two copies makes “editing efficiency” a hard number to read: alleles can end up in several combinations, and different combinations do not yield the same HbF — part of the reason Mayuranathan and colleagues found HbF lower and more variable under Cas9 strategies. Third, break repair between repeated sequences is a classic source of large rearrangements, and these cells have to self-renew in a person for decades; the clonal behavior of any large structural change needs long-term tracking, not a sixteen-week mouse readout.

Follow that logic and the trade against knocking out the BCL11A erythroid enhancer becomes legible. The enhancer route changes how much of a transcription factor erythroid cells make; the sequence change itself lands in non-repetitive DNA, which is clean, but what is being turned down is a regulator with more than one job, and its reach beyond γ-globin has to be circumscribed by evidence. The promoter route changes a cis-regulatory element belonging to the target gene, so the consequences are in principle confined to the β-globin locus, with HPFH carriers as a phenotypic reference — at the price that the element happens to sit inside a duplicated sequence. One side is a clean edit with a downstream footprint to be proven; the other is a narrow footprint with an unclean edit. Comparing them side by side, Mayuranathan and colleagues found that base editing at the HBG promoter induced HbF more potently and with less clonal variability than Cas9 strategies aimed either at the promoter’s BCL11A motif or at the BCL11A erythroid enhancer.

Three versions now clearing the human bar

In 2023, Sharma and colleagues reported in the New England Journal of Medicine the first clinical results from CRISPR-Cas9 editing of the HBG1 and HBG2 promoters in sickle cell disease. Two years later the same target had two more routes running on different enzymes: in 2026, Hanna and colleagues reported CRISPR-Cas12a editing of the HBG1/HBG2 promoters, and Gupta and colleagues reported the base editing version — the latter in 31 patients aged 12 to 35 with a history of severe disease, with hematopoietic reconstitution in roughly three weeks and no investigator-reported severe vaso-occlusive crises later than 60 days after the last red-cell transfusion, alongside adverse events that were common across participants.

Three papers, one stretch of promoter, three different writing instruments. That convergence is itself evidence that the target selection holds up: when several groups bring different enzymes to the same 200 base pairs, what they share is not the tool but the phenotypic guarantee that HPFH carriers provide.

Trading unknown safety for safety a population has carried for generations is one of the rare chances in gene therapy to see the answer before you act. But this answer was written into a sequence that appears twice — copy it, and the pen lands in both places at once. The trade has its own price, and the price is hidden inside the very feature that made the trade convenient.

References

  1. Martyn GE, Wienert B, Yang L, Shah M, Norton LJ, Burdach J, Kurita R, Nakamura Y, Pearson RCM, Funnell APW, Quinlan KGR, Crossley M. Natural regulatory mutations elevate the fetal globin gene via disruption of BCL11A or ZBTB7A binding. Nat Genet. 2018;50(4):498–503. DOI
  2. Liu N, Hargreaves VV, Zhu Q, Kurland JV, Hong J, Kim W, Sher F, Macias-Trevino C, Rogers JM, Kurita R, Nakamura Y, Yuan GC, Bauer DE, Xu J, Bulyk ML, Orkin SH. Direct promoter repression by BCL11A controls the fetal to adult hemoglobin switch. Cell. 2018;173(2):430–442.e17. DOI
  3. Traxler EA, Yao Y, Wang YD, Woodard KJ, Kurita R, Nakamura Y, Hughes JR, Hardison RC, Blobel GA, Li C, Weiss MJ. A genome-editing strategy to treat β-hemoglobinopathies that recapitulates a mutation associated with a benign genetic condition. Nat Med. 2016;22(9):987–990. DOI
  4. Ye L, Wang J, Tan Y, Beyer AI, Xie F, Muench MO, Kan YW. Genome editing using CRISPR-Cas9 to create the HPFH genotype in HSPCs: an approach for treating sickle cell disease and β-thalassemia. Proc Natl Acad Sci U S A. 2016;113(38):10661–10665. DOI
  5. Métais JY, Doerfler PA, Mayuranathan T, Bauer DE, Fowler SC, Hsieh MM, et al. Genome editing of HBG1 and HBG2 to induce fetal hemoglobin. Blood Adv. 2019;3(21):3379–3392. DOI
  6. Weber L, Frati G, Felix T, Hardouin G, Casini A, Wollenschlaeger C, et al. Editing a γ-globin repressor binding site restores fetal hemoglobin synthesis and corrects the sickle cell disease phenotype. Sci Adv. 2020;6(7):eaay9392. DOI
  7. Antoniou P, Hardouin G, Martinucci P, Frati G, Felix T, Chalumeau A, et al. Base-editing-mediated dissection of a γ-globin cis-regulatory element for the therapeutic reactivation of fetal hemoglobin expression. Nat Commun. 2022;13:6618. DOI
  8. Mayuranathan T, Newby GA, Feng R, Yao Y, Mayberry KD, Lazzarotto CR, et al. Potent and uniform fetal hemoglobin induction via base editing. Nat Genet. 2023;55(7):1210–1220. DOI
  9. Sharma A, Boelens JJ, Cancio M, Hankins JS, Bhad P, Azizy M, et al. CRISPR-Cas9 editing of the HBG1 and HBG2 promoters to treat sickle cell disease. N Engl J Med. 2023;389(9):820–832. DOI
  10. Hanna R, Frangoul H, Pineiro L, McKinney C, Mapara M, Dalal J, et al. CRISPR-Cas12a gene editing of HBG1 and HBG2 promoters to treat sickle cell disease. N Engl J Med. 2026;394(13):1281–1291. DOI
  11. Gupta AO, Sharma A, Frangoul H, Kanter J, Mapara MY, Dalal J, et al. Base editing of HBG1 and HBG2 promoters for sickle cell disease. N Engl J Med. 2026;394(18):1824–1835. DOI