碱基编辑(base editing)——一把不切断双链的手术刀 Base editing: a scalpel that never cuts the double helix
CRISPR 让人类第一次能精确地找到基因组里的某一个位置,但它接下来做的事其实相当粗暴:把 DNA 的双链一刀切断,然后指望细胞在慌乱的自我修复中,把那段序列改成我们想要的样子。问题是,人类绝大多数遗传病并不是”缺了一大段”,而只是某一个字母拼错了——一个 A 本该是 G,一个 C 本该是 T。为了改一个字母而把整条链切断,像是为了修正一个错别字而把整页纸撕掉重印。有没有可能,不切断、不撕纸,直接把那个字母擦掉重写?
断裂的代价
要理解base editing的价值,得先理解 CRISPR 的”切”到底意味着什么。当 Cas9 在目标位点切断 DNA 双链,细胞会立刻启动应急修复;而这种修复大多不精确,会在断口处随机插入或删除一小段序列。如果你的目的是”敲掉一个基因”,这种混乱反而正合心意;可如果你的目的是”把一个错误的碱基改成正确的”,这种随机的插入删除就成了灾难——你想要的精确改写被淹没在一堆意外的副产品里。更麻烦的是,double-strand break本身还会惊动细胞的 p53 警报、诱发大片段缺失甚至染色体重排。对于造血干细胞这种极其珍贵、要在体内工作一辈子的细胞,任何一点这样的隐患都值得格外警惕(这部分留到讲安全性时细说)。
换句话说,把双链切断,是 CRISPR 的力量之源,也是它最大的负担。
用化学,而不是用剪刀
真正的破局思路来自 David Liu 实验室,而且它更像化学,而不是外科。2016 年,Komor 等人在Nature上提出了”base editing”:他们不再让 Cas9 去切断 DNA,而是把它改造成一个只负责”定位、不负责切割”的向导,再在它身上焊接一个deaminase。当这套装置被引导到目标位点,deaminase会在原地对某一个碱基做一次化学改造——把胞嘧啶(C)脱氨,让它在后续复制中被读成胸腺嘧啶(T),从而完成一次 C→T 的替换。整个过程不切断双链、不需要提供修复模板,而且在大约五个核苷酸的小窗口里精确工作,插入删除的比例通常低于 1%。
第一代base editor只能做 C→T 这一种转换。真正让它成为通用工具的,是次年的第二把刀。2017 年,Gaudelli 等人同样在Nature上报告了adenine base editor(ABE):自然界原本没有能在 DNA 上给腺嘌呤脱氨的酶,他们硬是通过定向进化,把一个本来作用于 RNA 的酶改造成能在 DNA 上工作,实现了 A→G 的转换,在人类细胞里效率约达五成、产物纯度极高、插入删除通常不超过 0.1%。至此,四种最常见的点突变类型,都有了可以在不切断双链的前提下精确改写的手段。一个错别字,终于可以只擦掉那一个字母。
从化学走向治愈
对造血领域来说,这套工具最诱人的用武之地,恰恰是那些”只错了一个字母”的血液病。镰刀型贫血就是最经典的例子:它的病根是 β-珠蛋白基因上一个单碱基的突变。
2021 年,Newby 等人在Nature上给出了一个漂亮的演示。他们没有去”修正”这个致病突变,而是用一个定制的adenine base editor,把镰状突变转换成一种叫 Makassar 的天然良性变体——同样只改一个碱基,却把”致病”变成”无害”。他们把编辑器的 mRNA 送进镰刀型贫血患者的造血干/祖细胞,约八成的等位基因被成功转换;把这些细胞移植进小鼠十六周后,良性变体仍维持在约七成,红细胞在缺氧下的镰变减少了数倍。在人源化镰刀型贫血小鼠里,经过编辑的细胞让血液学指标接近正常,而二次移植证明这种改写在真正的长期造血干细胞里是持久的——只要有两成以上的等位基因被转换,就足以逆转病症。尤为关键的是,与 Cas9 核酸酶的方案相比,这种base editing没有引发 p53 激活,也没有出现那些令人担心的大片段缺失。
一个错别字被安静地改掉,整段文章的意思就变了——而这一次,没有撕纸。
换来的另一组风险
base editing并没有让风险消失,只是把风险换了一副面孔。它最特有的隐患叫”旁观者编辑”:deaminase是在一个小窗口里工作的,如果这个窗口里除了目标碱基,还恰好有别的同类碱基,它们可能被一起改掉。这就要求编辑窗口、向导 RNA、以及 Cas 蛋白能识别的序列,三者必须凑得恰到好处,有时甚至需要为一个靶点专门筛选或改造编辑器。此外,早期的base editor还被发现会在 RNA 层面产生一些脱靶改动。这些都不是”切与不切”的问题,而是”改得准不准、改得干不干净”的新课题。
所以,base editing更安全的说法需要小心限定:它规避了double-strand break那一类最粗暴的风险,但把精度的挑战推到了一个更精细的层面。
一场游戏规则的改变
回头看,base editing真正改变的,是基因编辑这件事的底层逻辑:从”切断,然后指望细胞修得符合心意”,变成”直接在原地把一个字母改写”。对那些由单碱基突变引起、又发生在造血干细胞这种不容闪失的细胞里的疾病,这种从”破坏后修复”到”化学改写”的转变,几乎是决定性的。
但它也有清晰的边界。base editing目前只能完成四种转换类型的点突变,做不了任意的替换、插入或删除——那片更大的疆域,要交给它的近亲prime editing,这是后话。而它下一段最值得期待的旅程,是与体内递送的结合:如果这把不切断双链的手术刀,有朝一日不必再把细胞取出体外、而能直接在人体骨髓里精准落下,那么它规避风险的优势,才会真正转化为更安全、也更触手可及的治愈。那把刀已经磨好,剩下的问题是,如何把它送到该去的地方。
参考文献
- 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
- Gaudelli NM, et al. Programmable base editing of A•T to G•C in genomic DNA without DNA cleavage. Nature. 2017;551(7681):464-471. DOI
- Newby GA, et al. Base editing of haematopoietic stem cells rescues sickle cell disease in mice. Nature. 2021;595(7866):295-302. DOI
CRISPR gave humanity, for the first time, the ability to find one precise spot in the genome — but what it does next is actually rather crude: it slices the DNA double strand clean through, and then counts on the cell, amid the panic of its own repair machinery, to rewrite that stretch of sequence into what we wanted. The trouble is that the vast majority of human genetic diseases are not a case of “a whole chunk missing,” but merely a single letter misspelled — an A that should have been a G, a C that should have been a T. Severing the entire strand just to fix one letter is like tearing up and reprinting a whole page to correct a single typo. Is there a way to not cut, not tear the page, but simply erase that one letter and write it over?
The price of the break
To understand the value of base editing, you first have to understand what CRISPR’s “cut” really means. When Cas9 severs the DNA double strand at the target site, the cell immediately triggers emergency repair — and that repair is mostly imprecise, randomly inserting or deleting a short stretch of sequence at the break point. If your goal is to “knock out a gene,” this chaos actually suits you fine. But if your goal is to “change one wrong base into the right one,” these random insertions and deletions become a disaster — the precise rewrite you were after gets drowned in a pile of accidental byproducts. Worse still, the double-strand break itself can set off the cell’s p53 alarm and induce large deletions or even chromosomal rearrangements. For hematopoietic stem cells — cells that are extraordinarily precious and have to work inside the body for a lifetime — any such hidden hazard warrants particular caution (we’ll save that discussion for the section on safety).
Put another way, severing the double strand is both the source of CRISPR’s power and its heaviest burden.
Chemistry, not scissors
The real breakthrough came from David Liu’s lab, and it looks more like chemistry than surgery. In 2016, Komor and colleagues proposed “base editing” in Nature: rather than letting Cas9 sever the DNA, they reengineered it into a guide responsible only for “locating, not cutting,” and welded a deaminase onto it. When this apparatus is steered to the target site, the deaminase performs one chemical modification on a single base right where it sits — deaminating cytosine (C) so that, during subsequent replication, it gets read as thymine (T), thereby completing a C→T substitution. The whole process never cuts the double strand, needs no repair template supplied, and works precisely within a small window of about five nucleotides, with insertions and deletions typically below 1%.
The first-generation base editor could perform only this one conversion, C→T. What truly made it a universal tool was the second blade, the following year. In 2017, Gaudelli and colleagues reported the adenine base editor (ABE), likewise in Nature: nature had no enzyme capable of deaminating adenine on DNA, so through directed evolution they forced an enzyme that normally acts on RNA to work on DNA instead, achieving the A→G conversion — with an efficiency of about 50% in human cells, extremely high product purity, and insertions and deletions typically no more than 0.1%. With that, all four of the most common point-mutation types had a means of being precisely rewritten without cutting the double strand. A single typo could, at last, be fixed by erasing just that one letter.
From chemistry to a cure
For the hematopoietic field, the most enticing use for this toolkit is precisely those blood diseases that are “just one letter wrong.” Sickle cell disease is the classic example: its root cause is a single-base mutation in the β-globin gene.
In 2021, Newby and colleagues gave an elegant demonstration in Nature. Rather than “correcting” the disease-causing mutation, they used a custom adenine base editor to convert the sickle mutation into a naturally occurring benign variant called Makassar — again changing just one base, yet turning “pathogenic” into “harmless.” They delivered the editor’s mRNA into the hematopoietic stem/progenitor cells of sickle cell patients, and roughly 80% of the alleles were successfully converted; sixteen weeks after transplanting these cells into mice, the benign variant still held at about 70%, and the sickling of red blood cells under low oxygen was reduced several-fold. In humanized sickle cell mice, the edited cells brought hematologic indices close to normal, and secondary transplantation proved that this rewrite is durable in genuine long-term hematopoietic stem cells — with just over 20% of alleles converted being enough to reverse the disease. Crucially, compared with the Cas9-nuclease approach, this base editing triggered no p53 activation and produced none of those worrisome large deletions.
One typo quietly corrected, and the meaning of the whole passage changes — and this time, no page was torn.
A different set of risks in exchange
Base editing does not make risk disappear; it merely gives risk a different face. Its most distinctive hazard is called “bystander editing”: the deaminase works within a small window, and if that window happens to contain, besides the target base, another base of the same kind, they may get changed together. This demands that the editing window, the guide RNA, and the sequence the Cas protein can recognize all line up just right — sometimes even requiring an editor to be specially screened or engineered for a single target. On top of that, early base editors were found to introduce some off-target changes at the RNA level. None of these are questions of “to cut or not to cut,” but new challenges of “how accurately, and how cleanly, the change is made.”
So the claim that base editing is safer needs to be carefully qualified: it sidesteps the crudest category of risk, the double-strand break, but pushes the challenge of precision to a finer level.
A change in the rules of the game
Looking back, what base editing truly changed is the underlying logic of gene editing itself: from “cut, then hope the cell repairs it the way we wanted” to “rewrite one letter in place.” For diseases caused by a single-base mutation and occurring in cells as unforgiving as hematopoietic stem cells, this shift from “break-then-repair” to “chemical rewriting” is almost decisive.
But it also has clear boundaries. Base editing can, for now, only carry out point mutations across the four conversion types; it cannot make arbitrary substitutions, insertions, or deletions — that larger territory belongs to its close relative, prime editing, a story for another time. And the next stretch of its journey most worth anticipating is its union with in vivo delivery: if this scalpel that never cuts the double strand could one day land precisely inside the human bone marrow, without having to take cells out of the body first, then its risk-avoiding advantage would truly translate into a cure that is safer and more within reach. The blade is honed; the remaining question is how to deliver it to where it needs to go.
References
- 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
- Gaudelli NM, et al. Programmable base editing of A•T to G•C in genomic DNA without DNA cleavage. Nature. 2017;551(7681):464-471. DOI
- Newby GA, et al. Base editing of haematopoietic stem cells rescues sickle cell disease in mice. Nature. 2021;595(7866):295-302. DOI