Yang Liu

← Primer

HSC · Healing Source Code

能不能不用化疗给 HSC 腾位置——CD117 / CD45 抗体与 ADC 预处理 Can we make room for HSCs without chemotherapy—CD117 / CD45 antibodies and ADC conditioning

2026-07-16 · conditioning, CD117, CD45, ADC, 造血干细胞移植

一个孩子被基因治疗治愈了。他的血液从此正常,不再输血,不再发作。但在那之前的一周里,他接受了足量的 busulfan——一种能把整个骨髓烧成白地的烷化剂。它清干净了病变的造血系统,也顺手清掉了别的东西:他很可能终身不育,他的肝、肺、性腺都记着这几天,他往后几十年里患第二种癌症的风险被永久地抬高了一截。这就是今天所有 HSC 基因治疗和移植的定价方式——治愈的部分越来越精准,而买单的部分,还停留在半个世纪前。于是一个朴素的问题被提了出来:骨髓里那些位置,非得用化疗来腾吗?

治愈的账单

把这笔账摊开看,它的荒谬感会更强一些。基因治疗的前半程已经精细到了不可思议的程度:从患者体内采出自体 CD34⁺ 细胞,在体外用 CRISPR 精准地切开 BCL11A 的红系enhancer,脱靶被一遍遍测序核查。然后,为了让这些价值百万美元的细胞能住进骨髓,医生给患者输了几天广谱的 DNA 烷化剂。前半程在用镊子,后半程在用推土机。

而推土机的代价是结构性的:busulfan 通过让 DNA 交联来杀细胞,它不认识”造血”这个概念,凡是分裂的细胞它都杀——肠黏膜、生殖细胞、肝窦内皮。在非人灵长类的对照实验里,研究者写得很直白:busulfan 在人体中常引起严重毒性与不育。对某些患者,这笔账根本付不起。范可尼贫血(Fanconi anemia)患者天生就缺 DNA 修复能力,给他们用 busulfan 或全身放疗,等于往一个漏水的船上再凿一个洞——他们恰恰是最需要移植、又最经不起预处理的那群人。

为什么非清不可

要拆掉推土机,得先弄明白它到底在干什么。预处理其实是三件被捆在一起做的事:杀掉残留的病变细胞(对白血病),压制宿主免疫让异体移植物不被排斥(对allogeneic移植),以及——腾龛位。

第三件事最容易被忽略,却是自体基因治疗唯一真正需要的那件。1978 年,Schofield 提出造血干细胞并不是自由漂浮的,它必须待在骨髓里特定的微环境中才能维持”干性”,他把这个位置叫做 niche。这个概念带来一个物理性的推论:niche 的数量是有限的。你把编辑好的细胞输进静脉,它们能一路归巢到骨髓,却发现所有的房间都住着人——宿主自己的 HSC。没有空房,新来的细胞就无法长期定居,engraftment 失败。

这就是清髓的真实功能:它不是在”消毒”,它是在腾房间。而如果目的只是腾房间,那么用一种把全身都推平的药,就显得极其不成比例——你为了让一位租客搬走,把整栋楼连同街区一起炸掉了。

只请一位租客离开

2007 年,斯坦福 Weissman 实验室的 Czechowicz 等人在《Science》上给出了第一个替代方案,而且优雅得近乎挑衅。他们用的不是药,是一个抗体:ACK2,它结合并阻断 c-Kit(即 CD117)。CD117 是 stem cell factor 的受体,而 HSC 的存活恰恰依赖这条信号——切断它,宿主的 HSC 就待不住了。在免疫缺陷小鼠里,单用这一个抗体,内源 HSC 被短暂清除了 98% 以上;随后输入供者 HSC,嵌合率最高达到约 90%。没有放疗,没有化疗,骨髓照样被打开了。论文结尾那句克制的展望——把这套方法外推到人,或许能带来温和而有效的预处理方案——成了此后近二十年的路线图。

但这个漂亮实验有两个隐藏条件,后来的所有工作都在还这两笔债。第一,小鼠是免疫缺陷的,不存在排斥问题;第二,裸抗体只是”劝退”,它靠阻断信号让 HSC 自己撑不下去,这份力道在免疫健全的宿主里往往不够。

给抗体装上弹头

补上力道的办法,是把抗体从劝说者变成投递员——做成 antibody-drug conjugate:抗体负责识别并被细胞内吞,弹头负责在细胞内部把它杀死。这样一来,杀伤力来自化学毒素,而选择性来自抗体识别的那个抗原。毒性从此不再是”全身广撒”,而是”谁表达这个分子,谁才中招”。

2016 年,哈佛 Scadden 实验室的 Palchaudhuri 等人在《Nature Biotechnology》上报告了第一版:把靶向 CD45 的抗体接上核糖体失活毒素 saporin。CD45 的好处是它几乎只出现在造血细胞上——身体的其余部分对这枚弹头视而不见。在免疫健全的小鼠里,单剂 CD45-saporin 让供者细胞的 engraftment 超过 90%,并完全纠正了镰刀型贫血模型。与放疗相比,它没有引起中性粒细胞减少和贫血,保住了骨髓与胸腺的微环境,T 细胞和 B 细胞迅速恢复,连抗真菌免疫都保留了下来。

三年后,Czechowicz 等人在《Nature Communications》上把同样的思路装回 CD117 这个更窄的靶点:单剂 CD117-ADC 清除了宿主 99% 以上的 HSC,而血象与免疫功能基本无恙,受试小鼠仍能正常应对病毒和真菌的攻击。

两个靶点的分工在这里显现出来,而这个差异一直延续到今天的临床:CD117 表达在 HSC 与祖细胞上,清它等于只请走干细胞这一位租客,成熟的淋巴细胞纹丝不动——免疫被完整保留,这对自体基因治疗几乎是量身定做,但它同时意味着不足以压制排斥。CD45 覆盖整个造血谱系,理论上打击面更宽,也更有希望把免疫抑制那件事一并办了。但已有的实测数据并不顺着这个直觉走:Palchaudhuri 那批免疫健全小鼠里,CD45-saporin 恰恰保住了胸腺微环境,T 与 B 细胞迅速恢复;而后文那项恒河猴的头对头试验里,CD45-ADC 同样保留了淋巴细胞。这枚弹头到底能不能压住排斥,目前还没有定论。选哪一个,取决于你到底要买上面三件事里的哪几件。

从小鼠到猴子

小鼠骨髓和人的骨髓不是一回事,这条路必须在大动物身上重走一遍。2023 年,Uchida 等人在《Nature Communications》上于恒河猴的慢病毒基因治疗模型中测试了 CD117-ADC:单次给药后,骨髓里的 CD34⁺CD90⁺CD45RA⁻ 细胞被清除 99% 以上,而淋巴细胞未见减少;基因修饰细胞的 engraftment、基因标记率与载体来源的胎儿血红蛋白诱导,都不劣于多日的 busulfan 清髓方案。最有分量的一句在后面:接受 ADC 预处理的动物保住了生育力。

2025 年,Demirci 等人在《Cell Stem Cell》上把两条线接到了一起——用 CD45-ADC 预处理的恒河猴,接受了 BCL11A enhancer 编辑的 HSPC,达到了与 busulfan 相当的 engraftment 与 HbF 重新激活,且慢病毒示踪显示重建是多克隆的,动力学与放疗或 busulfan 组相似。也就是说,本连载里那个已经上市的靶点,在动物层面已经不需要化疗来落地了。

同样在大动物里,反面的信号也已出现。2026 年,Murray 等人在《Blood Advances》上把两种携带 pyrrolobenzodiazepine 弹头的 ADC(分别靶向 CD117 与 CD45)与 busulfan 做了头对头比较:两种 ADC 都让 CRISPR 编辑过的 HSC 成功植入,造血恢复时间相仿,淋巴细胞得以保留——但长期维持下来的编辑水平低于 busulfan 组。腾出的房间够住,但也许还不够多。

走到人身上

真正走进临床的,目前是最保守的那一版:不带弹头的裸抗体。briquilimab(曾用名 JSP191、AMG 191)是一株人源化抗 CD117 单抗,机制上就是 ACK2 那个思路的人类版本——阻断 SCF 与 CD117 的结合,断掉宿主 HSC 的生存信号。它最早被用在重症联合免疫缺陷(SCID)患者身上,原因和 2007 年那批小鼠一样:这些患者本就没有免疫功能,不需要额外的免疫抑制,抗体只要负责腾房间就够了。这批数据目前主要见于 2021 年 ASH 的会议报告。

第一份完整发表的临床证据来自 2025 年,Agarwal、Bertaina 等人在《Nature Medicine》上报告了一项针对范可尼贫血伴骨髓衰竭患者的 1b 期试验:三名患者用 briquilimab 替代放疗和 busulfan,配合抗胸腺细胞球蛋白、环磷酰胺、fludarabine 与 rituximab 的免疫抑制,接受了去除 TCRαβ⁺ T 细胞与 CD19⁺ B 细胞的半相合移植。随访两年,未见治疗相关不良事件与急性 GVHD,未出现肝静脉闭塞病,中性粒细胞植入中位时间 11 天,供者嵌合维持在 99–100%,三人均存活良好,外周血淋巴细胞原有的染色体断裂异常得到纠正。三个人当然不构成证明,但对一个最经不起 DNA 损伤的病种,这已是一次颇具说服力的示范。同一株抗体也正在被加进镰刀型贫血与 β-地中海贫血的非清髓移植方案中试验。

值得注意的一个工程细节:briquilimab 在每位患者体内都在移植前自行清除完毕,不需要调整剂量。这不是锦上添花——抗体如果还留在血里,输进去的新细胞会被同一把刀迎面砍中。整个方案的可行性,系在这段药代动力学的窗口上。

还差什么

裸抗体的成功建立在一组特定的条件上:宿主免疫本就低下或已被别的药物压住,而且 HSC 足够依赖 SCF 信号。要把它推广到免疫健全、且需要高比例基因修饰细胞长期存活的场景,力道大概率不够,那就得回到 ADC。而 ADC 的账还没结清,并且已经有人替它付过一次:2023 年 1 月,Magenta Therapeutics 宣布暂停其 CD117-ADC(MGTA-117)在复发难治性 AML/MDS 中的 1/2 期剂量爬坡试验。0.08 mg/kg 剂量组最后一位接受给药的受试者出现呼吸衰竭与心跳骤停,并因此死亡;公司判定这起 5 级严重不良事件可能与 MGTA-117 本身相关,并作为可疑非预期严重不良反应(SUSAR)上报 FDA——这是该试验的第三起 SUSAR。次月,FDA 对该试验下达部分临床搁置,试验最终终止(NCT05223699)。选择性靶向并不自动等于安全——弹头一旦被内吞进错误的细胞,或者在血里提前脱落,后果和它的效力成正比。

更根本的问题还悬着几个。腾出多少 niche 才算够?大动物数据提示 ADC 组的长期编辑水平仍低于 busulfan,说明”清得干净”和”腾得够多”未必是同一件事。对恶性血液病,抗体预处理只解决了三件事里的一件,清病和免疫抑制仍要另找答案。而抗体本身的窗口、剂量与人群差异,目前的临床样本量还远不足以描出边界。

不过这条路的意义,并不取决于它能否单独走完。它真正改变的是基因治疗这件事的量级:今天它是一场完整的大手术——动员、采集、体外编辑、清髓、无菌病房、数周中性粒细胞缺乏;而每拆掉其中一根支柱,它就离”一次给药”近一分。抗体预处理拆的是最重的那一根。至于另一根——体外操作本身——正在被 in vivo 递送试图拆掉;若两者都成立,那么”腾房间”这个问题会以一种意想不到的方式消失:直接在原地改写住户的基因,就不必再请谁搬走。

半个世纪以来,我们默认要让新的造血系统住进来,得先烧掉旧的。CD117 和 CD45 这两个分子的故事说的是另一种可能:骨髓不是需要被清空的战场,而是一栋有门牌号的楼。你只需要敲对那一扇门。

参考文献

  1. Schofield R. The relationship between the spleen colony-forming cell and the haemopoietic stem cell. Blood Cells. 1978;4(1-2):7-25. PMID: 747780
  2. Czechowicz A, Kraft D, Weissman IL, Bhattacharya D. Efficient transplantation via antibody-based clearance of hematopoietic stem cell niches. Science. 2007;318(5854):1296-9. DOI
  3. Palchaudhuri R, et al. Non-genotoxic conditioning for hematopoietic stem cell transplantation using a hematopoietic-cell-specific internalizing immunotoxin. Nat Biotechnol. 2016;34(7):738-45. DOI
  4. Czechowicz A, Palchaudhuri R, Scheck A, et al. Selective hematopoietic stem cell ablation using CD117-antibody-drug-conjugates enables safe and effective transplantation with immunity preservation. Nat Commun. 2019;10:617. DOI
  5. Uchida N, Stasula U, Demirci S, et al. Fertility-preserving myeloablative conditioning using single-dose CD117 antibody-drug conjugate in a rhesus gene therapy model. Nat Commun. 2023;14:6291. DOI
  6. Demirci S, Zeng J, Palchaudhuri R, et al. BCL11A +58/+55 enhancer-editing facilitates HSPC engraftment and HbF induction in rhesus macaques conditioned with a CD45 antibody-drug conjugate. Cell Stem Cell. 2025;32(2):209-226.e8. DOI
  7. Murray J, Einhaus T, Radtke S, et al. Engraftment of gene-edited hematopoietic stem cells after antibody-drug conjugate conditioning in nonhuman primates. Blood Adv. 2026;10(4):1094-1105. DOI
  8. Agarwal R, Bertaina A, Soco C, et al. Irradiation- and busulfan-free stem cell transplantation in Fanconi anemia using an anti-CD117 antibody: a phase 1b trial. Nat Med. 2025;31:3183-3190. DOI PMID: 40696207(试验注册号 NCT04784052)
  9. Agarwal R, et al. JSP191 as a single-agent conditioning regimen results in successful engraftment, donor myeloid chimerism, and production of donor derived naïve lymphocytes in patients with severe combined immunodeficiency (SCID). Blood. 2021;138(Suppl 1):554.(ASH 会议摘要;试验注册号 NCT02963064)
  10. Addition of JSP191 (C-kit Antibody) to Nonmyeloablative Hematopoietic Cell Transplantation for Sickle Cell Disease and Beta-Thalassemia. ClinicalTrials.gov NCT05357482
  11. Magenta Therapeutics. Magenta Therapeutics Voluntarily Pauses the MGTA-117 Phase 1/2 Dose-Escalation Clinical Trial to Investigate Drug Safety.(公司公告,2023 年 1 月 25 日)

A child was cured by gene therapy. His blood is normal from now on—no more transfusions, no more crises. But in the week before that, he received a full dose of busulfan—an alkylating agent capable of burning an entire bone marrow down to bare ground. It cleared out the diseased hematopoietic system, and it cleared out some other things along the way: he is likely infertile for life, his liver, lungs, and gonads will remember those few days, and his risk of a second cancer over the coming decades has been permanently nudged upward. This is how every HSC gene therapy and transplant on offer today is priced—the curing half keeps getting more precise, while the paying half is still stuck half a century back. And so a plain question gets asked: do those spots in the bone marrow really have to be cleared with chemotherapy?

The bill for a cure

Laid out in full, the absurdity of this bill only sharpens. The first half of gene therapy has already been refined to an almost unbelievable degree: autologous CD34⁺ cells are harvested from the patient, CRISPR precisely cuts the erythroid enhancer of BCL11A ex vivo, and off-target effects are checked by round after round of sequencing. Then, so that these million-dollar cells can move into the bone marrow, the physician gives the patient several days of a broad-spectrum DNA alkylating agent. Tweezers for the first half; a bulldozer for the second.

And the bulldozer’s cost is structural: busulfan kills cells by cross-linking DNA, and it does not recognize the concept of “hematopoiesis”—it kills anything that divides: intestinal mucosa, germ cells, hepatic sinusoidal endothelium. In a controlled non-human primate experiment, the researchers put it bluntly: busulfan frequently causes severe toxicity and infertility in humans. For some patients, this bill simply cannot be paid. Patients with Fanconi anemia are born lacking DNA repair capacity; giving them busulfan or total body irradiation is like drilling another hole in a boat that is already leaking—and they are precisely the group that most needs a transplant and can least withstand the conditioning.

Why it has to be cleared

To dismantle the bulldozer, you first have to understand what it is actually doing. Conditioning is really three things bundled together: killing off residual diseased cells (for leukemia), suppressing host immunity so the allogeneic graft is not rejected (for allogeneic transplant), and—making room in the niche.

The third is the easiest to overlook, yet it is the only one that autologous gene therapy genuinely needs. In 1978, Schofield proposed that hematopoietic stem cells are not free-floating: an HSC must sit within a specific microenvironment in the bone marrow to maintain its “stemness,” and he called that spot a niche. The concept carries a physical corollary: the number of niches is finite. You infuse the edited cells into a vein, they home all the way to the bone marrow—and find that every room is occupied, by the host’s own HSCs. With no vacancy, the newcomers cannot take up long-term residence, and engraftment fails.

That is the real function of myeloablation: it is not “disinfecting,” it is freeing up rooms. And if the goal is only to free up rooms, then using a drug that flattens the entire body looks wildly out of proportion—to get one tenant to move out, you have blown up the whole building along with the block.

Asking just one tenant to leave

In 2007, Czechowicz and colleagues in Weissman’s lab at Stanford offered the first alternative in Science, and it was elegant to the point of provocation. What they used was not a drug but an antibody: ACK2, which binds and blocks c-Kit (that is, CD117). CD117 is the receptor for stem cell factor, and HSC survival depends precisely on this signal—cut it, and the host’s HSCs cannot hold their ground. In immunodeficient mice, this single antibody alone transiently depleted more than 98% of endogenous HSCs; donor HSCs were then infused, and chimerism reached as high as about 90%. No irradiation, no chemotherapy, and the bone marrow was opened up all the same. That restrained closing line—that extrapolating this approach to humans might yield a mild yet effective conditioning regimen—became the roadmap for the nearly two decades that followed.

But this beautiful experiment carried two hidden conditions, and all the work since has been repaying those two debts. First, the mice were immunodeficient, so rejection was not an issue; second, a naked antibody merely “talks the tenant out”—it relies on blocking a signal so that the HSC cannot sustain itself, and that force is often insufficient in an immunocompetent host.

Arming the antibody with a warhead

The way to supply the missing force is to turn the antibody from a persuader into a courier—making it an antibody-drug conjugate: the antibody handles recognition and internalization, the warhead kills the cell from the inside. Killing power now comes from a chemical toxin, while selectivity comes from the antigen the antibody recognizes. Toxicity is no longer “sprayed over the whole body” but rather “whoever expresses this molecule is the one who gets hit.”

In 2016, Palchaudhuri and colleagues in Scadden’s lab at Harvard reported the first version in Nature Biotechnology: an antibody targeting CD45 conjugated to the ribosome-inactivating toxin saporin. The advantage of CD45 is that it appears almost exclusively on hematopoietic cells—the rest of the body is blind to this warhead. In immunocompetent mice, a single dose of CD45-saporin achieved donor cell engraftment exceeding 90% and fully corrected a sickle cell disease model. Compared with irradiation, it did not cause neutropenia or anemia, it preserved the bone marrow and thymic microenvironments, T cells and B cells recovered rapidly, and even antifungal immunity was retained.

Three years later, Czechowicz and colleagues in Nature Communications fitted the same idea back onto the narrower target CD117: a single dose of CD117-ADC depleted more than 99% of host HSCs, while blood counts and immune function were essentially unharmed, and the treated mice could still mount normal responses to viral and fungal challenge.

The division of labor between the two targets becomes visible here, and that difference carries straight through to today’s clinic: CD117 is expressed on HSCs and progenitors, so depleting it amounts to asking only the stem cell—that one tenant—to leave, while mature lymphocytes do not budge. Immunity is preserved intact, which is almost tailor-made for autologous gene therapy, but it also means it is not enough to suppress rejection. CD45 covers the entire hematopoietic lineage, so in theory it strikes more broadly and stands a better chance of taking care of the immunosuppression piece as well. But the measured data we have do not follow that intuition: in Palchaudhuri’s immunocompetent mice, CD45-saporin in fact preserved the thymic microenvironment, with T and B cells recovering rapidly; and in the head-to-head rhesus study discussed below, CD45-ADC likewise preserved lymphocytes. Whether this warhead can actually hold rejection down remains unsettled. Which one you choose depends on which of those three things you are actually trying to buy.

From mice to monkeys

Mouse bone marrow and human bone marrow are not the same thing, and this road has to be walked again in large animals. In 2023, Uchida and colleagues in Nature Communications tested CD117-ADC in a rhesus macaque lentiviral gene therapy model: after a single dose, CD34⁺CD90⁺CD45RA⁻ cells in the bone marrow were depleted by more than 99%, while no reduction in lymphocytes was seen; engraftment of gene-modified cells, gene marking rates, and vector-derived fetal hemoglobin induction were all non-inferior to a multi-day busulfan myeloablative regimen. The weightiest sentence comes later: the animals that received ADC conditioning retained their fertility.

In 2025, Demirci and colleagues in Cell Stem Cell joined the two threads together—rhesus macaques conditioned with CD45-ADC received BCL11A enhancer-edited HSPCs and achieved engraftment and HbF reactivation comparable to busulfan, with lentiviral tracking showing that reconstitution was polyclonal and that its kinetics resembled those of the irradiation or busulfan groups. In other words, the already-approved target from earlier in this series no longer requires chemotherapy to land, at least at the animal level.

Signals pointing the other way have already appeared in large animals too. In 2026, Murray and colleagues in Blood Advances ran a head-to-head comparison of two ADCs carrying pyrrolobenzodiazepine warheads (targeting CD117 and CD45, respectively) against busulfan: both ADCs allowed CRISPR-edited HSCs to engraft successfully, hematopoietic recovery times were similar, and lymphocytes were preserved—but the level of editing maintained over the long term was lower than in the busulfan group. The rooms freed up are enough to live in, but perhaps still not enough of them.

Into humans

What has actually made it into the clinic so far is the most conservative version: the naked antibody, without a warhead. Briquilimab (formerly JSP191, AMG 191) is a humanized anti-CD117 monoclonal antibody, and mechanistically it is the human version of the ACK2 idea—blocking the binding of SCF to CD117 and cutting off the survival signal of the host’s HSCs. It was first used in patients with severe combined immunodeficiency (SCID), for the same reason as that batch of 2007 mice: these patients have no immune function to begin with, so no additional immunosuppression is needed, and the antibody only has to handle freeing up rooms. These data are so far found mainly in a 2021 ASH conference report.

The first fully published clinical evidence came in 2025, when Agarwal, Bertaina, and colleagues reported in Nature Medicine a phase 1b trial in patients with Fanconi anemia and bone marrow failure: three patients used briquilimab in place of irradiation and busulfan, together with immunosuppression consisting of anti-thymocyte globulin, cyclophosphamide, fludarabine, and rituximab, and received haploidentical transplants depleted of TCRαβ⁺ T cells and CD19⁺ B cells. Over two years of follow-up, no treatment-related adverse events and no acute GVHD were seen, no hepatic veno-occlusive disease occurred, the median time to neutrophil engraftment was 11 days, donor chimerism was maintained at 99–100%, all three were alive and well, and the pre-existing chromosomal breakage abnormalities in their peripheral blood lymphocytes were corrected. Three people are of course not proof, but for a disease that can least withstand DNA damage, this is already a rather persuasive demonstration. The same antibody is also being tested as an addition to non-myeloablative transplant regimens for sickle cell disease and β-thalassemia.

One engineering detail worth noting: briquilimab cleared itself completely from every patient before transplant, with no dose adjustment required. This is not a nice-to-have—if the antibody is still in the blood, the newly infused cells will be cut down head-on by the very same blade. The feasibility of the whole regimen hangs on that pharmacokinetic window.

What’s still missing

The success of the naked antibody rests on a particular set of conditions: host immunity is already low or has been suppressed by other drugs, and the HSCs are sufficiently dependent on SCF signaling. To extend it to settings that are immunocompetent and that require a high proportion of gene-modified cells to survive long term, the force will most likely be insufficient—and then you are back to the ADC. But the ADC’s bill has not been settled, and someone has already paid part of it: in January 2023, Magenta Therapeutics announced that it was pausing the phase 1/2 dose-escalation trial of its CD117-ADC (MGTA-117) in relapsed/refractory AML/MDS. The last subject dosed in the 0.08 mg/kg cohort developed respiratory failure and cardiac arrest, and died as a result; the company determined that this Grade 5 serious adverse event was possibly related to MGTA-117 itself, and reported it to the FDA as a Suspected Unexpected Serious Adverse Reaction (SUSAR)—the third SUSAR in this trial. The following month, the FDA placed a partial clinical hold on the trial, and the trial was ultimately terminated (NCT05223699). Selective targeting does not automatically equal safety—once a warhead is internalized into the wrong cell, or falls off prematurely in the blood, the consequences are proportional to its potency.

Several more fundamental questions are still hanging. How many niches must be freed up for it to count as enough? The large-animal data suggest that long-term editing levels in the ADC groups remain lower than with busulfan, which indicates that “cleared thoroughly” and “freed up enough rooms” are not necessarily the same thing. For malignant hematologic disease, antibody conditioning has solved only one of the three things; killing the disease and immunosuppression still need answers elsewhere. And the antibody’s own window, dosing, and population differences—current clinical sample sizes are nowhere near enough to trace the boundaries.

Still, the significance of this road does not depend on whether it can be walked to the end alone. What it truly changes is the order of magnitude of the whole undertaking: today gene therapy is a full-scale operation—mobilization, collection, ex vivo editing, myeloablation, a sterile room, weeks of neutropenia; and every pillar you remove brings it one step closer to “a single dose.” Antibody conditioning removes the heaviest one. As for another pillar—the ex vivo manipulation itself—in vivo delivery is trying to remove that one; and if both hold up, then the question of “freeing up rooms” disappears in an unexpected way: rewrite the residents’ genes in place, and no one has to move out at all.

For half a century, we have taken it for granted that to let a new hematopoietic system move in, you first have to burn down the old one. The story of these two molecules, CD117 and CD45, tells of another possibility: bone marrow is not a battlefield that needs to be emptied, but a building with door numbers. You only need to knock on the right door.

References

  1. Schofield R. The relationship between the spleen colony-forming cell and the haemopoietic stem cell. Blood Cells. 1978;4(1-2):7-25. PMID: 747780
  2. Czechowicz A, Kraft D, Weissman IL, Bhattacharya D. Efficient transplantation via antibody-based clearance of hematopoietic stem cell niches. Science. 2007;318(5854):1296-9. DOI
  3. Palchaudhuri R, et al. Non-genotoxic conditioning for hematopoietic stem cell transplantation using a hematopoietic-cell-specific internalizing immunotoxin. Nat Biotechnol. 2016;34(7):738-45. DOI
  4. Czechowicz A, Palchaudhuri R, Scheck A, et al. Selective hematopoietic stem cell ablation using CD117-antibody-drug-conjugates enables safe and effective transplantation with immunity preservation. Nat Commun. 2019;10:617. DOI
  5. Uchida N, Stasula U, Demirci S, et al. Fertility-preserving myeloablative conditioning using single-dose CD117 antibody-drug conjugate in a rhesus gene therapy model. Nat Commun. 2023;14:6291. DOI
  6. Demirci S, Zeng J, Palchaudhuri R, et al. BCL11A +58/+55 enhancer-editing facilitates HSPC engraftment and HbF induction in rhesus macaques conditioned with a CD45 antibody-drug conjugate. Cell Stem Cell. 2025;32(2):209-226.e8. DOI
  7. Murray J, Einhaus T, Radtke S, et al. Engraftment of gene-edited hematopoietic stem cells after antibody-drug conjugate conditioning in nonhuman primates. Blood Adv. 2026;10(4):1094-1105. DOI
  8. Agarwal R, Bertaina A, Soco C, et al. Irradiation- and busulfan-free stem cell transplantation in Fanconi anemia using an anti-CD117 antibody: a phase 1b trial. Nat Med. 2025;31:3183-3190. DOI PMID: 40696207 (trial registration NCT04784052)
  9. Agarwal R, et al. JSP191 as a single-agent conditioning regimen results in successful engraftment, donor myeloid chimerism, and production of donor derived naïve lymphocytes in patients with severe combined immunodeficiency (SCID). Blood. 2021;138(Suppl 1):554. (ASH conference abstract; trial registration NCT02963064)
  10. Addition of JSP191 (C-kit Antibody) to Nonmyeloablative Hematopoietic Cell Transplantation for Sickle Cell Disease and Beta-Thalassemia. ClinicalTrials.gov NCT05357482
  11. Magenta Therapeutics. Magenta Therapeutics Voluntarily Pauses the MGTA-117 Phase 1/2 Dose-Escalation Clinical Trial to Investigate Drug Safety. (Company press release, January 25, 2023)