为什么 in vivo 是圣杯——把手术级移植压缩成一针 Why in vivo is the holy grail: compressing a surgical transplant into a single shot
一款已经获批的造血干细胞基因疗法,听上去像是”打一针就好了”,实际经历却更接近一场大手术。医生要先用药物把患者的造血干细胞从骨髓里动员到血液,再通过单采把它们收集起来,送进工厂在体外完成基因编辑;与此同时,患者要接受清髓化疗,把自己原有的骨髓几乎清空,为编辑后的细胞腾出位置;最后细胞被回输,患者在医院里住上数周,等待新的造血系统重新长起来。整个流程耗时数月、花费以百万美元计,只有极少数中心做得了。于是一个显而易见的问题浮现出来:这一切,能不能省掉——能不能真的就打一针?
圣杯的定义
所谓体内(in vivo)基因治疗,就是不再把细胞取出体外,而是把基因编辑的工具直接注射进人体,让它在骨髓里就地找到造血干细胞、完成改写。它之所以被称为这个领域的圣杯,不是因为编辑本身更强,而是因为它有望一次性拆掉现行方案里最沉重的三块负担:不必动员和采集细胞,不必在工厂里做复杂而昂贵的体外制备,甚至有望免去那场几乎要人命的清髓化疗。
这三块负担,恰恰是把”治愈”挡在绝大多数患者门外的真正原因。当一款疗法从”手术级工程”退化成”一次注射”,它的成本、风险和门槛才可能整体坍缩到一个普通医院、乃至一个资源有限的地区也能承受的量级。换句话说,体内化不是让疗法更炫,而是让它能真正抵达那些最需要它的人。
为什么这么难
如果体内化的好处如此明显,为什么它至今仍主要停留在动物实验里?答案几乎全部压在两个字上:递送。
造血干细胞藏身于骨髓深处,那里血流缓慢,还隔着一道血-骨髓屏障,药物很难高浓度地抵达。更棘手的是,当你把大多数纳米颗粒注入血液,它们会不由自主地涌向肝脏——肝脏几乎是所有全身递送系统的默认目的地。要让编辑工具绕开肝脏这个”黑洞”,穿过骨髓的屏障,再在成千上万种细胞里精准认出造血干细胞并只对它下手,这是一道叠了好几层的难题。长期以来,这个领域缺的从来不是编辑器,而是一辆能把编辑器准确开进骨髓的车。
已经走出的几步
尽管困难,过去几年里,这辆”车”的雏形已经在实验室里被一点点造出来,而且思路各不相同。
一条路是让递送载体对骨骼本身产生亲和力。2022 年,Xue 等人在 JACS(Journal of the American Chemical Society)上报告了一类带双膦酸结构的脂质材料——双膦酸对骨矿物质有天然的高亲和力,他们据此构建的脂质纳米颗粒能在小鼠体内显著提高 mRNA 在骨微环境中的表达,等于给递送车装上了”认路去骨头”的导航。
更直接针对造血干细胞的进展来自 2024 年。Lian 等人在 Nature Nanotechnology 上报告了一系列”骨髓归巢”的脂质纳米颗粒,能把 mRNA 递送给骨髓里至少十四种不同的细胞,其中就包括健康的和患病的造血干细胞;在一个表达人类镰刀型贫血表型的小鼠模型里,他们用这套系统在体内实现了 CRISPR 编辑和base editing,用于重启胎儿血红蛋白、以及把镰状等位基因转换成非镰状。几乎同期,Palanki 等人在 PNAS 上走了另一条更巧的路:他们利用造血干细胞在胎儿发育期会暂居于容易抵达的肝脏这一窗口,设计出靶向 HSC 表面 CD45 受体的脂质纳米颗粒,通过一次子宫内注射,就在胎儿的造血干细胞里实现了概念验证性的基因编辑。
而把矛头最直接地对准造血干细胞本身的,是 2023 年 Breda 等人发表在 Science 上的工作。他们既不绕道骨骼,也不借助胎儿肝脏的窗口,而是盯住造血干细胞表面的一个身份标签——CD117,也就是干细胞因子受体;他们据此造出一种靶向 CD117 的脂质纳米颗粒来装载 mRNA,在体内把工具直接递到造血干细胞手里。用这辆车运送靶向人 CD117 的编辑系统,他们在人源镰状细胞里实现了近乎完全的纠正;更有意思的是,当同一辆车改去运送促凋亡的 PUMA mRNA 时,造血干细胞的功能被有选择地削弱,等于在不动用清髓化疗的前提下为移植腾出了位置——同一套递送既能送编辑器,也能送”预处理”。这项工作同样做在小鼠与体外的人源细胞上,仍属临床前证据;但它把”在体内认出并改造造血干细胞”这件事,从一个愿望变成了一个可以照着设计的分子问题。
需要清醒地指出:这些成果目前几乎都停留在小鼠和临床前阶段,离人体应用还有相当距离。但它们共同证明了一件此前悬而未决的事——体内、原位地编辑造血干细胞,在原理上是可行的。
还差什么
从”小鼠里能做”到”人身上能治”,中间隔着的仍是一串硬骨头。编辑效率要足够高、且要稳定地落在真正的长期造血干细胞上,而不是只改到那些很快就被淘汰的短命细胞;靶向要足够专一,否则编辑器在体内四处误伤,后果无法像体外那样”先筛一遍再回输”地被拦下——这正是体内方案失去的那张安全网:你没有机会在把细胞放回人体之前,先检查它们被改成了什么样。此外,如何在不清髓的前提下让编辑过的细胞获得足够的立足之地,以及递送载体反复给药时的免疫反应,都还是敞开的问题。
这些困难说明,体内化虽然拆掉了旧流程的重负,却也把安全和精准的责任,整体前移到了”注射进去的那一刻”。
通向那一针的路
把这些线索连起来,体内造血干细胞基因治疗的图景就清晰了:它是这个领域用来解决成本与可及性的最终答案,今天还很早,但方向明确、且正在快速推进。它也几乎注定要和更温和的编辑方式携手——一把不切断双链、副产品更少的base editing刀,配上一辆能精准开进骨髓的递送车,这样的组合远比”体内做一次double-strand break”更让人安心。
回到开篇那场耗时数月的”大手术”:它今天之所以还是大手术,并不是因为编辑本身有多复杂,而是因为我们还不得不把细胞取出、清空骨髓、再放回去。当递送这道难关被真正攻克的那一天,所有这些环节都可能被折叠进一支针管——那时,这个领域才算真正兑现了它对患者许下的承诺。
参考文献
- Xue L, et al. Rational Design of Bisphosphonate Lipid-like Materials for mRNA Delivery to the Bone Microenvironment. J Am Chem Soc. 2022;144(22):9926-9937. DOI
- Lian X, et al. Bone-marrow-homing lipid nanoparticles for genome editing in diseased and malignant haematopoietic stem cells. Nat Nanotechnol. 2024;19(9):1409-1417. DOI
- Palanki R, et al. In utero delivery of targeted ionizable lipid nanoparticles facilitates in vivo gene editing of hematopoietic stem cells. Proc Natl Acad Sci U S A. 2024;121(32):e2400783121. DOI
- Breda L, et al. In vivo hematopoietic stem cell modification by mRNA delivery. Science. 2023;381(6656):436-443. DOI
An already-approved hematopoietic stem cell gene therapy sounds like “one shot and you’re cured,” but the actual experience is closer to major surgery. Physicians first use drugs to mobilize the patient’s hematopoietic stem cells out of the bone marrow and into the blood, then collect them by apheresis and ship them to a facility for gene editing ex vivo. Meanwhile the patient undergoes myeloablative chemotherapy that all but empties their existing bone marrow to make room for the edited cells. Finally the cells are infused back, and the patient stays in the hospital for weeks, waiting for a new blood-forming system to grow. The whole process takes months, costs on the order of millions of dollars, and can only be done at a handful of centers. So an obvious question surfaces: can all of this be spared — can it really become just one shot?
Defining the holy grail
In vivo gene therapy means no longer taking cells out of the body, but instead injecting the gene-editing tools directly into the patient and letting them find hematopoietic stem cells in place, inside the bone marrow, and rewrite them there. It is called the holy grail of this field not because the editing itself is more powerful, but because it promises to dismantle, in a single stroke, the three heaviest burdens of the current approach: no need to mobilize and collect cells, no need for complex and expensive ex vivo manufacturing in a facility, and potentially even no need for the near-lethal myeloablative chemotherapy.
These three burdens are precisely the real reasons a “cure” remains out of reach for the vast majority of patients. When a therapy degrades from “surgery-grade engineering” back into “a single injection,” its cost, risk, and barrier to entry can collapse together to a level that an ordinary hospital — even a region with limited resources — can bear. In other words, going in vivo isn’t about making the therapy flashier; it’s about letting it actually reach the people who need it most.
Why it is so hard
If the benefits of going in vivo are so obvious, why does it still remain largely confined to animal experiments? The answer rests almost entirely on one word: delivery.
Hematopoietic stem cells hide deep in the bone marrow, where blood flow is slow and a blood–bone-marrow barrier stands in the way, making it hard for drugs to reach in high concentration. More troublesome still, when you inject most nanoparticles into the blood, they drift irresistibly toward the liver — the liver is the default destination of nearly every systemic delivery system. Getting the editing tools to skirt this “black hole” of the liver, cross the bone-marrow barrier, and then, among thousands of cell types, precisely recognize hematopoietic stem cells and act only on them — this is a problem stacked several layers deep. For a long time, what this field lacked was never the editor, but a vehicle that could drive the editor accurately into the bone marrow.
The steps already taken
Difficult as it is, over the past few years the rough form of this “vehicle” has been built up piece by piece in the lab, along several distinct lines of thinking.
One path is to give the delivery carrier an affinity for bone itself. In 2022, Xue and colleagues reported in JACS (Journal of the American Chemical Society) a class of lipid-like materials bearing a bisphosphonate structure — bisphosphonates have a natural high affinity for bone mineral — and the lipid nanoparticles they built on this basis significantly raised mRNA expression in the bone microenvironment in mice, in effect equipping the delivery vehicle with a “navigate-to-bone” guidance system.
Progress aimed more directly at hematopoietic stem cells came in 2024. Lian and colleagues reported in Nature Nanotechnology a series of “bone-marrow-homing” lipid nanoparticles able to deliver mRNA to at least fourteen different cell types in the bone marrow, including healthy and diseased hematopoietic stem cells; in a mouse model expressing a human sickle cell phenotype, they used this system to achieve CRISPR editing and base editing in vivo, to reactivate fetal hemoglobin and to convert the sickle allele to a non-sickle one. At nearly the same time, Palanki and colleagues took another, cleverer path in PNAS: exploiting the developmental window in which hematopoietic stem cells transiently reside in the more accessible fetal liver, they designed lipid nanoparticles targeting the CD45 receptor on the HSC surface and, with a single in utero injection, achieved proof-of-concept gene editing in fetal hematopoietic stem cells.
The line aimed most directly at hematopoietic stem cells themselves is the 2023 work of Breda and colleagues in Science. They neither detoured through bone nor exploited the fetal liver window; instead they fixed on an identity tag on the HSC surface — CD117, the stem cell factor receptor — and built a CD117-targeted lipid nanoparticle to carry mRNA, handing the tools directly to hematopoietic stem cells in vivo. Using this vehicle to deliver an anti-human CD117 editing system, they achieved near-complete correction of human sickle cells; more intriguingly, when the same vehicle was redirected to carry pro-apoptotic PUMA mRNA, HSC function was selectively blunted — in effect making room for a transplant without resorting to myeloablative chemotherapy, so that one delivery platform can ship both the editor and the “conditioning.” This work too was done in mice and in human cells in vitro, and so remains preclinical evidence; but it turned “recognizing and modifying hematopoietic stem cells in vivo” from a wish into a molecular problem one can deliberately design against.
It must be said soberly: these results at present nearly all remain at the mouse and preclinical stage, still a considerable distance from human application. But together they prove something that had until now been an open question — that editing hematopoietic stem cells in vivo, in place, is feasible in principle.
What is still missing
Between “it works in mice” and “it can treat a person” lies a string of hard problems still to be cracked. Editing efficiency has to be high enough, and it has to land stably on the true long-term hematopoietic stem cells rather than only on the short-lived cells that are soon discarded. Targeting has to be specific enough — otherwise the editor injures things at random throughout the body, and the consequences cannot be intercepted the way they can ex vivo by “screening first, then infusing.” This is exactly the safety net the in vivo approach gives up: you have no chance to inspect what the cells have been turned into before putting them back into the body. On top of this, how to give the edited cells enough of a foothold without myeloablation, and the immune response to repeated dosing of the delivery carrier, both remain open questions.
These difficulties show that while going in vivo dismantles the heavy burdens of the old process, it also shifts the entire responsibility for safety and precision forward, onto “the very moment of injection.”
The road to that one shot
Tie these threads together and the picture of in vivo hematopoietic stem cell gene therapy comes into focus: it is this field’s ultimate answer to cost and accessibility, still very early today, but clear in direction and advancing fast. It is also all but destined to go hand in hand with gentler editing methods — a base editing blade that does not sever the double strand and leaves fewer byproducts, paired with a delivery vehicle that can drive precisely into the bone marrow, is a combination far more reassuring than “making a double-strand break in vivo.”
Return to that months-long “major surgery” from the opening: the reason it is still major surgery today is not that the editing itself is so complex, but that we still have to take the cells out, empty the bone marrow, and put them back. On the day the delivery obstacle is truly overcome, all of these steps may be folded into a single syringe — and only then will this field have genuinely delivered on the promise it made to patients.
References
- Xue L, et al. Rational Design of Bisphosphonate Lipid-like Materials for mRNA Delivery to the Bone Microenvironment. J Am Chem Soc. 2022;144(22):9926-9937. DOI
- Lian X, et al. Bone-marrow-homing lipid nanoparticles for genome editing in diseased and malignant haematopoietic stem cells. Nat Nanotechnol. 2024;19(9):1409-1417. DOI
- Palanki R, et al. In utero delivery of targeted ionizable lipid nanoparticles facilitates in vivo gene editing of hematopoietic stem cells. Proc Natl Acad Sci U S A. 2024;121(32):e2400783121. DOI
- Breda L, et al. In vivo hematopoietic stem cell modification by mRNA delivery. Science. 2023;381(6656):436-443. DOI