Yang Liu

← Primer

HSC · Healing Source Code

一个 32 碱基的缺失治愈了 HIV——但治愈的原因,可能不只是它 A 32-Base Deletion Cured HIV — But That May Not Be the Whole Reason

2026-07-25 · HIV, CCR5, 基因编辑, 造血干细胞移植, 功能性治愈

柏林的一位血液科医生面对过一份两头为难的病历:一个感染 HIV 多年的男人,又得了急性髓系白血病。白血病要治,只能做异体造血干细胞移植。而他没有在配型合适的候选者里随手挑一个,他想找一个基因组里恰好缺了 32 个碱基的人。那个缺口本身平平无奇,唯一的后果是让一个叫 CCR5 的受体没能出现在细胞表面,而 HIV 想进入 CD4 细胞,通常得先握住这只手。这个念头后来把一个人从终身服药里放了出来,也把整个领域拽进一场持续近二十年的追问:那次治愈,究竟是这 32 个碱基的功劳,还是别的什么?

人群里天生带着一个漏洞

线索不是从治疗里来的,是从一群”该被感染却没被感染”的人身上来的。1996 年,Liu 等人在 Cell 上报告了两个反复暴露于 HIV 却始终未感染的个体,代号 EU2 和 EU3:他们的 CD4 细胞在体外高度抵抗巨噬细胞嗜性病毒进入,原因是两人纯合携带 CCR5 基因上一段 32 个碱基对的缺失,编码出的蛋白严重截短,在细胞表面根本检测不到。论文里有一句当时几乎顺带写下的话,日后成了整条治疗路线的许可证——这个缺陷”在携带者身上没有明显的表型”。

同一年,Samson 等人在 Nature 上把它放进人群尺度:这个等位基因在欧洲裔人群中的频率约为 0.092,在西非、中非和日本人群中则完全检测不到。它显然不是为对抗 HIV 演化出来的;至于它因何而来,至今没有定论。2005 年,Novembre 等人在 PLoS Biology 上用一个空间显式的扩散模型论证,Δ32 今天的地理分布与”它在 HIV 出现之前很久就已承受强选择”是相容的;而同一期上,Sabeti 等人用密得多的遗传图谱重估同一批证据,发现这个位点周边的变异模式并不比基因组别处更异常,该等位基因很可能在五千多年前就已出现,用中性演化也解释得通。争论没有收口,但有一件事不受影响:有些人天生没有 CCR5,看上去健康,却对主流 HIV 毒株近乎免疫。

柏林

2009 年,Hütter 等人在 N Engl J Med 上给出了答案的第一半:一名同时患有急性髓系白血病和 HIV 感染的患者接受了 CCR5Δ32 纯合供者的造血干细胞,在移植并停用抗逆转录病毒治疗之后的 20 个月里,病毒没有反弹。

让”缓解”变成”治愈”的是随后的追踪。2011 年,Allers 等人在 Blood 上报告,这名患者的 CD4 T 细胞不仅在外周血、也在肠道黏膜免疫系统里完成了重建;更要紧的是,重建起来的细胞含有很高比例的活化记忆 CD4 T 细胞——正是 HIV 最偏爱的靶子——而且在体外对 CXCR4 嗜性的 HIV 仍然可以被感染。保护并非来自”病毒无处可去”,而是来自”能用的那扇门被拆了”。

这位患者的名字后来公开了,Timothy Ray Brown。他在 2020 年 9 月因白血病复发去世;至去世时,他身上一直没有再出现 HIV 感染的迹象。

复制它,比想象中难,也比想象中容易

一个 n=1 的奇迹最怕无法重复,而这条路上先出现的是一个刺眼的失败。2014 年,Kordelas 等人在 N Engl J Med 上以通讯形式报告了一例接受 CCR5Δ32/Δ32 移植后仍然反弹的患者;后续病毒学分析指出了原因——反弹的是一株复制能力很强的 CXCR4 嗜性变异体,在移植之前就已存在于患者体内。CCR5 这扇门被拆掉了,病毒手里还有另一把钥匙。此后的病例报告,会先交代移植前的病毒 tropism。

真正的第二例出现在 2019 年。Gupta 等人在 Nature 上报告的”伦敦病人”因霍奇金淋巴瘤接受 CCR5Δ32/Δ32 移植,只做了一次、用减低强度的 conditioning、没有全身照射——柏林病人则做了两次,且每次都配合全身照射。停药 18 个月后,他的血浆 HIV-1 RNA 低于每毫升 1 拷贝,用总计 2400 万个静息 CD4 T 细胞也培养不出可复制的病毒。次年,同一团队在 Lancet HIV 上把随访推到停药后 30 个月,取样一路铺到精液、脑脊液、肠道和淋巴结:腋窝淋巴结还能测到低水平的 LTR 与 env 信号,但完整前病毒检测阴性;模型推算,在八成供者嵌合的前提下终身缓解的概率为 98%,九成嵌合时超过 99%——他们用了 cure 这个词。

此后案例接连出现:杜塞尔多夫那例停药后 4 年,体外与人源化小鼠体内的扩增实验都找不出可复制的病毒(Jensen 等,Nat Med,2023);纽约那例是一名混血女性,也是第一例靠 CCR5Δ32/Δ32 脐带血联合成人半相合细胞取得缓解的病例(Hsu 等,Cell,2023);City of Hope 那例已 63 岁、感染 HIV 三十一年,同样用减低强度 conditioning(Dickter 等,N Engl J Med,2024);奥斯陆那例连作为主要病毒储库的肠道都达成了完全的供者嵌合(Myhre 等,Nat Microbiol,2026)。

一个让整条逻辑松动的反例

如果故事到此为止,结论会非常干净:去掉 CCR5,治愈 HIV。2024 年,Sáez-Cirión 等人在 Nat Med 上报告的日内瓦病例撬开了一道缝。这名患者因髓系肿瘤接受异体移植,供者是一位非亲属的野生型 CCR5 携带者——按理说,重建起来的免疫系统对 HIV 门户洞开。可停药后 32 个月,血浆病毒载量始终检测不到,而他的 CD4 T 细胞在体外仍然可以被 HIV 感染。整个期间,他因慢性移植物抗宿主病一直在服用 ruxolitinib。2026 年,Gaebler 等人在 Nature 上报告的”第二位柏林病人”把这道缝撑得更开:供者与患者都只是 CCR5 野生型/Δ32 杂合,CCR5 功能是在的,而这名患者在移植三年后停药,至今维持超过 6 年的缓解。他们的结论分量很重:CCR5Δ32 介导的病毒抵抗,并不是持久缓解的必要条件。

另一半功劳属于谁,线索早就有。2014 年,Henrich 等人在 Ann Intern Med 上报告的两名波士顿患者接受的是野生型供者细胞:储库至少下降了 3 个数量级,但停药后 12 周和 32 周,病毒仍然反弹。而 2024 年,Salgado 等人在 Lancet HIV 上报告的 IciStem 队列给出了系统答案:无论供者 CCR5 基因型如何,一旦达成完全供者嵌合,外周血病毒储库都会迅速塌缩;他们的模型指向的机制是——conditioning 化疗先做了大规模清场,真正把残余储库扫干净的是供者细胞介导的同种异体免疫,潜伏感染细胞的半衰期由 44 个月缩短到 1.5 个月。

于是柏林病人的治愈更像两件事叠加:一次把宿主免疫系统整体换掉的移植,加上一道让病毒无法重新立足的门锁。Δ32 是那道锁;但清场足够彻底时,锁也许可以不那么严。

为什么这条路走不出移植室

这套方案有一个几乎无解的约束:它要求一个人同时满足两个互不相干的条件——既感染 HIV,又恰好得一种需要异体造血干细胞移植的血液系统恶性肿瘤——然后还要在配型库里找到一个 CCR5Δ32 纯合的供者。这个基因型有多稀有,Glass 等人 2006 年在 J Exp Med 里给过一个直接的数字:1318 名美国欧洲裔健康献血者中,Δ32 纯合子占 1.0%。而异体移植本身带着实打实的死亡风险,抗逆转录病毒治疗却已经能把病毒压到检测不到——这条路只能搭肿瘤指征的顺风车。IciStem 队列说明了尺度:横跨八个国家、从 2009 到 2019 年,总共 30 例,其中 10 例用的是 Δ32 供者细胞;而 Gaebler 等人给出了对照的分母——自这场疫情开始以来估计有 8800 万人感染 HIV,被记录在案的治愈仅六例。

于是有人想把那个缺失做出来

自然界的 Δ32 供者太少,那就人工制造。CCR5 因此成为最早的临床基因编辑靶点之一:它不需要修复什么,只需要破坏——把一个基因敲坏,恰恰是核酸酶最擅长的事。2014 年,Tebas 等人在 N Engl J Med 上给 12 名感染者回输了用 zinc-finger nuclease 编辑过 CCR5 的自体 CD4 T 细胞(NCT00842634),安全性过关。但 T 细胞终究是终末产物,真正的源头在骨髓里。

2010 年,Holt 等人在 Nat Biotechnol 上把 zinc-finger nuclease 用在人 CD34⁺ 造血干/祖细胞上,平均编辑掉 17% 的等位基因;这些细胞仍能在免疫缺陷小鼠里 engraftment,而在 CCR5 嗜性 HIV 攻击下,小鼠体内迅速富集出 CCR5 双等位缺失的细胞、病毒载量显著更低——少数编辑细胞就足以撑起一个抗病毒的造血系统,至少在小鼠里是这样。

大动物给出的数字冷静得多。2016 年,Peterson 等人在 Blood 上把这套编辑放进猪尾猴的自体移植:ex vivo 编辑效率可达 64%,移植早期在体内还有 40%,但六个月后,真正长期重建造血的细胞里只剩 3% 到 5%——编辑效率和长期植入效率是两条完全不同的曲线。人体上的第一次尝试印证了这点:2019 年,Xu 等人在 N Engl J Med 上报告了一名同时患有 HIV 感染和急性淋巴细胞白血病的患者(NCT03164135),他接受的是经 CRISPR 敲除 CCR5 的供者 CD34⁺ 细胞,这些细胞在体内存活超过 19 个月,未见与基因编辑相关的不良事件。但淋巴细胞中 CCR5 被破坏的比例只有约 5%,作者自己写明,这提示该路径还需要进一步研究。

5% 意味着什么,后来有了定量的回答。2025 年,Claiborne 等人在 Nat Commun 上把 CRISPR-Cas9 在动员的人 CD34⁺ 细胞里推到超过 90% 的编辑率,移植后的小鼠对 HIV 感染表现出抵抗;而梯度实验画出的是一条逐级下降的曲线:编辑到 96% 的那组,连续八次攻毒无一感染;69% 时每次攻毒的感染风险下降约 82%,具统计学显著性;54% 时降幅约 45%,未达显著;到 26% 时已与完全不编辑没有差别。作者据此提出,要完全阻断感染,成熟移植物里的编辑比例需要在 90% 以上;低于这条线并非白做,只是保护逐级衰减,到 54% 以下迅速逼近于无。

还没解决的部分

三道坎很清楚:编辑率与长期植入率的乘积必须够高,而两者在大动物身上互相拉扯;走自体 ex vivo 基因治疗这条路,编辑过的细胞要植入,仍得先用 conditioning 腾出骨髓龛位,而这个慢性感染已被药物控制住,风险与收益并不匹配;递送则决定成本,只有编辑能直接在体内完成,整个结构才会改变。2026 年,Anderson 等人在 Mol Ther 上用携带 base editor 的辅助依赖型腺病毒载体在人源化小鼠体内编辑造血干细胞,配合体内选择,在骨髓单个核细胞中达到约 50% 的 CCR5 碱基编辑,攻毒后血浆病毒滴度比对照低约 12 倍。这是小鼠不是人,但方向是把 ex vivo 那一整套压缩成一次给药。

还有一件事不该跳过:CCR5 并不是白拿的。Glass 等人那篇论文的正题其实是西尼罗病毒:在两个独立的有症状感染队列里,Δ32 纯合子的比例分别达到 4.2% 和 8.3%,远高于对照人群的 1.0%,其中一个队列还与死亡结局显著相关。这个靶点后来还经历过一次更尖锐的争论:2019 年 Nat Med 上一篇论文声称 Δ32 纯合在人类中有害,该文当年即被撤稿;次年,Maier 等人在同一刊物上给出结论,UK Biobank 队列中没有统计学证据支持 Δ32 影响寿命。一个被反复检验、被撤回、又被重新厘清的”不影响寿命”,比一个从未被质疑过的”无害”结实得多。(2018 年那起以 CCR5 为对象的人类胚胎编辑事件属于另一回事,留到讲生殖系编辑时再说。)

回到那 32 个碱基。它最初只是人群里一个无关紧要的缺口,却让我们知道 HIV 进门要握哪只手,也第一次把终身感染变成一段可以结束的病史。但接连出现的野生型与杂合供者病例正在提醒:那扇被拆掉的门也许不是全部答案,把整套免疫系统连根换掉、再由供者的同种异体免疫把残余储库一路扫清,可能才是这几例治愈共同的底色。这对基因治疗其实是个更难的消息——编辑能造出那扇门,却造不出那场清场。要让 CCR5 这条线走出移植室,接下来要解决的恰恰是:如何在不做移植的前提下,把移植做过的事做完。

参考文献

  1. Liu R, Paxton WA, Choe S, et al. Homozygous defect in HIV-1 coreceptor accounts for resistance of some multiply-exposed individuals to HIV-1 infection. Cell. 1996;86(3):367-377. DOI
  2. Samson M, Libert F, Doranz BJ, et al. Resistance to HIV-1 infection in caucasian individuals bearing mutant alleles of the CCR-5 chemokine receptor gene. Nature. 1996;382(6593):722-725. DOI
  3. Novembre J, Galvani AP, Slatkin M. The geographic spread of the CCR5 Delta32 HIV-resistance allele. PLoS Biol. 2005;3(11):e339. DOI
  4. Sabeti PC, Walsh E, Schaffner SF, et al. The case for selection at CCR5-Delta32. PLoS Biol. 2005;3(11):e378. DOI
  5. Hütter G, Nowak D, Mossner M, et al. Long-term control of HIV by CCR5 Delta32/Delta32 stem-cell transplantation. N Engl J Med. 2009;360(7):692-698. DOI
  6. Allers K, Hütter G, Hofmann J, et al. Evidence for the cure of HIV infection by CCR5Δ32/Δ32 stem cell transplantation. Blood. 2011;117(10):2791-2799. DOI
  7. Watts G. Timothy Ray Brown. Lancet. 2020;396(10259):1327. DOI
  8. Kordelas L, Verheyen J, Beelen DW, et al. Shift of HIV tropism in stem-cell transplantation with CCR5 Delta32 mutation. N Engl J Med. 2014;371(9):880-882. DOI
  9. Verheyen J, Thielen A, Lübke N, et al. Rapid rebound of a preexisting CXCR4-tropic human immunodeficiency virus variant after allogeneic transplantation with CCR5 Δ32 homozygous stem cells. Clin Infect Dis. 2019;68(4):684-687. DOI
  10. Gupta RK, Abdul-Jawad S, McCoy LE, et al. HIV-1 remission following CCR5Δ32/Δ32 haematopoietic stem-cell transplantation. Nature. 2019;568(7751):244-248. DOI
  11. Gupta RK, Peppa D, Hill AL, et al. Evidence for HIV-1 cure after CCR5Δ32/Δ32 allogeneic haemopoietic stem-cell transplantation 30 months post analytical treatment interruption: a case report. Lancet HIV. 2020;7(5):e340-e347. DOI
  12. Jensen BO, Knops E, Cords L, et al. In-depth virological and immunological characterization of HIV-1 cure after CCR5Δ32/Δ32 allogeneic hematopoietic stem cell transplantation. Nat Med. 2023;29(3):583-587. DOI
  13. Hsu J, Van Besien K, Glesby MJ, et al. HIV-1 remission and possible cure in a woman after haplo-cord blood transplant. Cell. 2023;186(6):1115-1126.e8. DOI
  14. Dickter JK, Aribi A, Cardoso AA, et al. HIV-1 remission after allogeneic hematopoietic-cell transplantation. N Engl J Med. 2024;390(7):669-671. DOI
  15. Myhre AE, Meyer-Myklestad MH, Gullaksen HH, et al. Long-term HIV-1 remission achieved through allogeneic haematopoietic stem cell transplant from a CCR5Δ32/Δ32 sibling donor. Nat Microbiol. 2026;11(5):1374-1386. DOI
  16. Sáez-Cirión A, Mamez AC, Avettand-Fenoel V, et al. Sustained HIV remission after allogeneic hematopoietic stem cell transplantation with wild-type CCR5 donor cells. Nat Med. 2024;30(12):3544-3554. DOI
  17. Gaebler C, Kor S, Allers K, et al. Sustained HIV-1 remission after heterozygous CCR5Δ32 stem cell transplantation. Nature. 2026;650(8102):701-709. DOI
  18. Henrich TJ, Hanhauser E, Marty FM, et al. Antiretroviral-free HIV-1 remission and viral rebound after allogeneic stem cell transplantation: report of 2 cases. Ann Intern Med. 2014;161(5):319-327. DOI
  19. Salgado M, Gálvez C, Nijhuis M, et al. Dynamics of virological and immunological markers of HIV persistence after allogeneic haematopoietic stem-cell transplantation in the IciStem cohort: a prospective observational cohort study. Lancet HIV. 2024;11(6):e389-e405. DOI
  20. Glass WG, McDermott DH, Lim JK, et al. CCR5 deficiency increases risk of symptomatic West Nile virus infection. J Exp Med. 2006;203(1):35-40. DOI
  21. Tebas P, Stein D, Tang WW, et al. Gene editing of CCR5 in autologous CD4 T cells of persons infected with HIV. N Engl J Med. 2014;370(10):901-910. DOI
  22. Holt N, Wang J, Kim K, 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):839-847. DOI
  23. Peterson CW, Wang J, Norman KK, et al. Long-term multilineage engraftment of autologous genome-edited hematopoietic stem cells in nonhuman primates. Blood. 2016;127(20):2416-2426. DOI
  24. Xu L, Wang J, Liu Y, 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
  25. Claiborne DT, Detwiler Z, Docken SS, et al. High frequency CCR5 editing in human hematopoietic stem progenitor cells protects xenograft mice from HIV infection. Nat Commun. 2025;16(1):446. DOI
  26. Anderson AK, Georgakopoulou A, Kuhlmann AS, et al. In vitro and in vivo base editing of CCR5 in hematopoietic stem cells confers HIV-1 resistance. Mol Ther. 2026;34(7):4084-4103. DOI
  27. Wei X, Nielsen R. CCR5-∆32 is deleterious in the homozygous state in humans. Nat Med. 2019;25(6):909-910(已于 2019 年撤稿:Nat Med. 2019;25(11):1796). DOI ／ 撤稿声明 DOI
  28. Maier R, Akbari A, Wei X, et al. No statistical evidence for an effect of CCR5-∆32 on lifespan in the UK Biobank cohort. Nat Med. 2020;26(2):178-180. DOI

A haematologist in Berlin once faced a chart with no good side to it: a man who had lived with HIV for years had now developed acute myeloid leukaemia. The leukaemia had to be treated, and that meant an allogeneic haematopoietic stem cell transplant. Rather than pick any well-matched candidate from the registry, he went looking for someone whose genome happened to be missing 32 bases. The gap itself is unremarkable; its only consequence is that a receptor called CCR5 never reaches the cell surface — and HIV, to get into a CD4 cell, usually has to take that hand first. The idea eventually released one man from lifelong medication, and pulled an entire field into a question that has now run for nearly two decades: was that cure the work of those 32 bases, or of something else?

A hole the population was born with

The clue did not come from treatment. It came from people who should have been infected and were not. In 1996, Liu and colleagues reported in Cell on two individuals, designated EU2 and EU3, who had been repeatedly exposed to HIV and never infected: their CD4 cells were highly resistant in vitro to entry by macrophage-tropic virus, because both carried, in homozygous form, a 32-base-pair deletion in the CCR5 gene; the encoded protein was severely truncated and could not be detected at the cell surface. One almost offhand line in that paper later became the licence for an entire therapeutic route — the defect “has no obvious phenotype in the affected individuals.”

That same year, Samson and colleagues put it on a population scale in Nature: the allele has a frequency of about 0.092 in Caucasian populations and is absent from black populations of western and central Africa and from Japanese populations. It plainly did not evolve to counter HIV; why it is there at all remains unsettled. In 2005, Novembre and colleagues argued in PLoS Biology, using a spatially explicit model of the allele’s spread, that the present-day geographic distribution of Δ32 is consistent with its having been under intense selection long before HIV appeared; in the same issue, Sabeti and colleagues re-examined the same evidence with much denser genetic maps and found that the pattern of variation at the locus does not stand out as exceptional relative to other loci across the genome — the allele is likely to have arisen more than 5,000 years ago, and neutral evolution accounts for it just as well. The argument has not closed. One thing is unaffected either way: some people are born without CCR5, look healthy, and are close to immune to the dominant strains of HIV.

Berlin

In 2009, Hütter and colleagues supplied the first half of the answer in N Engl J Med: a patient with both acute myeloid leukaemia and HIV infection received haematopoietic stem cells from a CCR5Δ32 homozygous donor, and for the 20 months after transplantation and discontinuation of antiretroviral therapy the virus did not rebound.

What turned remission into cure was the follow-up. In 2011, Allers and colleagues reported in Blood that this patient’s CD4 T cells had reconstituted not only in peripheral blood but in the gut mucosal immune system; more to the point, the reconstituted compartment contained a high proportion of activated memory CD4 T cells — exactly what HIV prefers — and in vitro those cells remained susceptible to productive infection with CXCR4-tropic HIV. The protection did not come from the virus having nowhere to go. It came from the door it uses having been taken out.

The patient’s name later became public: Timothy Ray Brown. He died in September 2020 following a relapse of his leukaemia; to the end, no sign of HIV infection had returned.

Repeating it is harder than it looks, and easier

An n=1 miracle is most vulnerable to not repeating, and what came first on this road was a conspicuous failure. In 2014, Kordelas and colleagues reported as a letter in N Engl J Med a patient who rebounded despite a CCR5Δ32/Δ32 transplant; the virological analysis that followed attributed the rebound to a highly replicative CXCR4-tropic variant that could already be detected before the transplant. The CCR5 door had been removed; the virus still held another key. Case reports since then tend to begin by accounting for the virus’s tropism before transplantation.

The real second case came in 2019. The “London patient”, reported by Gupta and colleagues in Nature, received a CCR5Δ32/Δ32 transplant for Hodgkin’s lymphoma — a single procedure, with reduced-intensity conditioning and no irradiation; the Berlin patient had undergone two, each with total body irradiation. Eighteen months after treatment interruption his plasma HIV-1 RNA was below one copy per millilitre, and a total of 24 million resting CD4 T cells yielded no replication-competent virus. The following year the same team extended follow-up in Lancet HIV to 30 months post-interruption, sampling semen, cerebrospinal fluid, gut and lymph node: axillary lymph node tissue still gave a low-level positive signal for the long-terminal repeat and env, but was negative by the intact proviral DNA assay; modelling put the probability of remission for life at 98% in the context of 80% donor chimerism, and above 99% with 90% donor chimerism. They used the word cure.

More cases followed: the Düsseldorf patient, four years after treatment interruption, with neither ex vivo assays nor in vivo outgrowth assays in humanized mice yielding replication-competent virus (Jensen et al., Nat Med, 2023); the New York patient, a mixed-race woman and the first remission achieved with CCR5Δ32/Δ32 cord blood combined with haploidentical adult cells (Hsu et al., Cell, 2023); the City of Hope patient, 63 years old and 31 years into HIV infection, likewise on reduced-intensity conditioning (Dickter et al., N Engl J Med, 2024); and the Oslo patient, in whom even the gut — the primary viral reservoir — reached full donor chimerism (Myhre et al., Nat Microbiol, 2026).

The counterexample that loosens the whole logic

If the story ended there the conclusion would be very clean: remove CCR5, cure HIV. In 2024, the Geneva case reported by Sáez-Cirión and colleagues in Nat Med prised open a gap. This patient received an allogeneic transplant for a myeloid malignancy from an unrelated donor with wild-type CCR5 — by rights, the reconstituted immune system should have been wide open to HIV. Yet 32 months after treatment interruption the plasma viral load remained undetectable, while his CD4 T cells remained susceptible to HIV infection in vitro. Throughout, he stayed on ruxolitinib for chronic graft-versus-host disease. In 2026, the “second Berlin patient” reported by Gaebler and colleagues in Nature widened the gap: donor and patient were both merely CCR5 wild-type/Δ32 heterozygous, so CCR5 function was present, and this patient stopped antiretroviral therapy three years after the transplant and has now sustained remission for more than six years. Their conclusion carries weight: CCR5Δ32-mediated HIV resistance is not essential for durable remission.

Where the other half of the credit lies has been hinted at for a long time. In 2014, Henrich and colleagues reported in Ann Intern Med on two Boston patients who received wild-type donor cells: the reservoir fell by at least three orders of magnitude, yet the virus rebounded 12 and 32 weeks after antiretroviral cessation. And in 2024, the IciStem cohort reported by Salgado and colleagues in Lancet HIV gave a systematic answer: regardless of donor CCR5 genotype, HIV reservoirs in peripheral blood collapsed as soon as full donor chimerism was achieved. Their model points to the mechanism — conditioning chemotherapy performs the massive clearing first, and what actually sweeps out the residual reservoir is allogeneic immunity mediated by donor cells, with the half-life of latently infected replication-competent cells falling from 44 months to 1.5 months.

So the Berlin cure looks more like two things stacked: a transplant that replaced the host immune system wholesale, plus a lock that kept the virus from re-establishing itself. Δ32 is the lock; but when the clearing is thorough enough, the lock may not need to be so tight.

Why this route cannot leave the transplant ward

The approach carries a nearly insoluble constraint: it requires one person to satisfy two unrelated conditions — to be infected with HIV, and to develop precisely the kind of haematological malignancy that calls for an allogeneic haematopoietic stem cell transplant — and then to find a CCR5Δ32 homozygous donor in the registry. How rare that genotype is, Glass and colleagues gave a direct number for in J Exp Med in 2006: among 1,318 healthy Caucasian American blood donors, Δ32 homozygotes made up 1.0%. Allogeneic transplantation carries a real risk of death, while antiretroviral therapy can already push the virus below detection — so this route can only ride along with an oncological indication. The IciStem cohort shows the scale: across eight countries, from 2009 to 2019, 30 cases in total, ten of them with Δ32 donor cells. And Gaebler and colleagues supply the denominator to read that against — an estimated 88 million people have acquired HIV since the onset of the epidemic, and six cures are documented.

So someone set out to manufacture the deletion

Natural Δ32 donors are too scarce; make them instead. That is why CCR5 became one of the earliest clinical gene-editing targets: nothing has to be repaired, only broken — and breaking a gene is exactly what a nuclease does best. In 2014, Tebas and colleagues in N Engl J Med infused autologous CD4 T cells whose CCR5 had been edited with a zinc-finger nuclease into 12 infected people (NCT00842634); it cleared safety. But T cells are, in the end, a terminal product; the real source is in the bone marrow.

In 2010, Holt and colleagues applied zinc-finger nucleases to human CD34⁺ haematopoietic stem/progenitor cells in Nat Biotechnol, disrupting CCR5 at a mean frequency of 17% of total alleles; these cells still engrafted in immunodeficient mice, and under challenge with CCR5-tropic HIV the mice rapidly selected for CCR5-null cells and had significantly lower viral loads — a minority of edited cells is enough to carry an HIV-resistant haematopoietic system, at least in mice.

The large-animal numbers are far more sober. In 2016, Peterson and colleagues put the same editing into autologous transplantation in pigtailed macaques in Blood: CCR5 disruption reached up to 64% ex vivo and 40% in vivo early after transplant, but at six months only 3% to 5% remained in the cells doing the long-term repopulating — editing efficiency and long-term engraftment are two entirely different curves. The first attempt in a human bore this out: in 2019, Xu and colleagues reported in N Engl J Med a patient with both HIV infection and acute lymphoblastic leukaemia (NCT03164135) who received donor CD34⁺ cells in which CCR5 had been ablated by CRISPR; those cells persisted for more than 19 months without gene-editing-related adverse events. But CCR5 disruption in lymphocytes reached only about 5%, which, as the authors themselves wrote, indicates the need for further research into this approach.

What 5% means got a quantitative answer later. In 2025, Claiborne and colleagues in Nat Commun pushed CRISPR-Cas9 above 90% CCR5 editing in mobilized human CD34⁺ cells, and the transplanted mice were resistant to HIV infection; a titration experiment then drew a curve of steadily decreasing benefit: in the group at 96% editing, none of the mice was infected after eight consecutive challenges; at 69%, the per-challenge risk of infection fell by about 82%, a statistically significant effect; at 54%, the reduction was about 45% and did not reach significance; and at 26% it was no different from no editing at all. From this the authors propose that complete abrogation of productive HIV infection requires editing above 90% in the mature graft — below that line the effort is not wasted, but protection decays step by step, and beneath 54% it approaches nothing.

What is still unsolved

Three obstacles are clear. The product of editing rate and long-term engraftment has to be high enough, and in large animals the two pull against each other. On the autologous ex vivo gene-therapy route, edited cells still cannot engraft without conditioning to clear marrow niches first — and against a chronic infection that drugs already control, that risk does not match the benefit. Delivery, in turn, determines cost: only if editing can be done directly in the body does the whole structure change. In 2026, Anderson and colleagues in Mol Ther used a helper-dependent adenoviral vector carrying a base editor to edit haematopoietic stem cells in vivo in a humanized mouse model; with in vivo selection they reached about 50% CCR5 base editing in bone marrow mononuclear cells, and after challenge plasma viral titres were about 12-fold lower than in controls. This is mice, not humans — but the direction it demonstrates is compressing the entire ex vivo procedure into a single dose.

One more thing should not be skipped: CCR5 is not free. The actual subject of that Glass paper was West Nile virus: in two independent cohorts of symptomatic infection, Δ32 homozygotes reached 4.2% and 8.3%, far above the 1.0% of the control population, and in one of the cohorts homozygosity was significantly associated with fatal outcome. The target later went through a sharper argument still: a 2019 paper in Nat Med claimed Δ32 homozygosity is deleterious in humans, and was retracted that same year; the following year Maier and colleagues concluded in the same journal that there is no statistical evidence for an effect of Δ32 on lifespan in the UK Biobank cohort. A “no effect on lifespan” that has been tested, retracted and re-established is sturdier than a “harmless” that was never questioned. (The 2018 human embryo editing episode aimed at CCR5 belongs to another discussion altogether, and is left to the piece on germline editing.)

Back to those 32 bases. It began as an inconsequential gap in the population, and it told us which hand HIV has to take on the way in, and turned lifelong infection, for the first time, into a medical history that could end. But the wild-type and heterozygous donor cases now accumulating are a reminder: the door that was taken out may not be the whole answer, and replacing the entire immune system, then letting donor-mediated allogeneic immunity sweep out what remains of the reservoir, may be what these cures actually have in common. For gene therapy that is the harder news — editing can build the door, but not the clearing-out. To get the CCR5 route out of the transplant ward, what has to be solved next is exactly this: how to finish what a transplant does, without doing the transplant.

References

  1. Liu R, Paxton WA, Choe S, et al. Homozygous defect in HIV-1 coreceptor accounts for resistance of some multiply-exposed individuals to HIV-1 infection. Cell. 1996;86(3):367-377. DOI
  2. Samson M, Libert F, Doranz BJ, et al. Resistance to HIV-1 infection in caucasian individuals bearing mutant alleles of the CCR-5 chemokine receptor gene. Nature. 1996;382(6593):722-725. DOI
  3. Novembre J, Galvani AP, Slatkin M. The geographic spread of the CCR5 Delta32 HIV-resistance allele. PLoS Biol. 2005;3(11):e339. DOI
  4. Sabeti PC, Walsh E, Schaffner SF, et al. The case for selection at CCR5-Delta32. PLoS Biol. 2005;3(11):e378. DOI
  5. Hütter G, Nowak D, Mossner M, et al. Long-term control of HIV by CCR5 Delta32/Delta32 stem-cell transplantation. N Engl J Med. 2009;360(7):692-698. DOI
  6. Allers K, Hütter G, Hofmann J, et al. Evidence for the cure of HIV infection by CCR5Δ32/Δ32 stem cell transplantation. Blood. 2011;117(10):2791-2799. DOI
  7. Watts G. Timothy Ray Brown. Lancet. 2020;396(10259):1327. DOI
  8. Kordelas L, Verheyen J, Beelen DW, et al. Shift of HIV tropism in stem-cell transplantation with CCR5 Delta32 mutation. N Engl J Med. 2014;371(9):880-882. DOI
  9. Verheyen J, Thielen A, Lübke N, et al. Rapid rebound of a preexisting CXCR4-tropic human immunodeficiency virus variant after allogeneic transplantation with CCR5 Δ32 homozygous stem cells. Clin Infect Dis. 2019;68(4):684-687. DOI
  10. Gupta RK, Abdul-Jawad S, McCoy LE, et al. HIV-1 remission following CCR5Δ32/Δ32 haematopoietic stem-cell transplantation. Nature. 2019;568(7751):244-248. DOI
  11. Gupta RK, Peppa D, Hill AL, et al. Evidence for HIV-1 cure after CCR5Δ32/Δ32 allogeneic haemopoietic stem-cell transplantation 30 months post analytical treatment interruption: a case report. Lancet HIV. 2020;7(5):e340-e347. DOI
  12. Jensen BO, Knops E, Cords L, et al. In-depth virological and immunological characterization of HIV-1 cure after CCR5Δ32/Δ32 allogeneic hematopoietic stem cell transplantation. Nat Med. 2023;29(3):583-587. DOI
  13. Hsu J, Van Besien K, Glesby MJ, et al. HIV-1 remission and possible cure in a woman after haplo-cord blood transplant. Cell. 2023;186(6):1115-1126.e8. DOI
  14. Dickter JK, Aribi A, Cardoso AA, et al. HIV-1 remission after allogeneic hematopoietic-cell transplantation. N Engl J Med. 2024;390(7):669-671. DOI
  15. Myhre AE, Meyer-Myklestad MH, Gullaksen HH, et al. Long-term HIV-1 remission achieved through allogeneic haematopoietic stem cell transplant from a CCR5Δ32/Δ32 sibling donor. Nat Microbiol. 2026;11(5):1374-1386. DOI
  16. Sáez-Cirión A, Mamez AC, Avettand-Fenoel V, et al. Sustained HIV remission after allogeneic hematopoietic stem cell transplantation with wild-type CCR5 donor cells. Nat Med. 2024;30(12):3544-3554. DOI
  17. Gaebler C, Kor S, Allers K, et al. Sustained HIV-1 remission after heterozygous CCR5Δ32 stem cell transplantation. Nature. 2026;650(8102):701-709. DOI
  18. Henrich TJ, Hanhauser E, Marty FM, et al. Antiretroviral-free HIV-1 remission and viral rebound after allogeneic stem cell transplantation: report of 2 cases. Ann Intern Med. 2014;161(5):319-327. DOI
  19. Salgado M, Gálvez C, Nijhuis M, et al. Dynamics of virological and immunological markers of HIV persistence after allogeneic haematopoietic stem-cell transplantation in the IciStem cohort: a prospective observational cohort study. Lancet HIV. 2024;11(6):e389-e405. DOI
  20. Glass WG, McDermott DH, Lim JK, et al. CCR5 deficiency increases risk of symptomatic West Nile virus infection. J Exp Med. 2006;203(1):35-40. DOI
  21. Tebas P, Stein D, Tang WW, et al. Gene editing of CCR5 in autologous CD4 T cells of persons infected with HIV. N Engl J Med. 2014;370(10):901-910. DOI
  22. Holt N, Wang J, Kim K, 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):839-847. DOI
  23. Peterson CW, Wang J, Norman KK, et al. Long-term multilineage engraftment of autologous genome-edited hematopoietic stem cells in nonhuman primates. Blood. 2016;127(20):2416-2426. DOI
  24. Xu L, Wang J, Liu Y, 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
  25. Claiborne DT, Detwiler Z, Docken SS, et al. High frequency CCR5 editing in human hematopoietic stem progenitor cells protects xenograft mice from HIV infection. Nat Commun. 2025;16(1):446. DOI
  26. Anderson AK, Georgakopoulou A, Kuhlmann AS, et al. In vitro and in vivo base editing of CCR5 in hematopoietic stem cells confers HIV-1 resistance. Mol Ther. 2026;34(7):4084-4103. DOI
  27. Wei X, Nielsen R. CCR5-∆32 is deleterious in the homozygous state in humans. Nat Med. 2019;25(6):909-910 (retracted 2019: Nat Med. 2019;25(11):1796). DOI / Retraction notice DOI
  28. Maier R, Akbari A, Wei X, et al. No statistical evidence for an effect of CCR5-∆32 on lifespan in the UK Biobank cohort. Nat Med. 2020;26(2):178-180. DOI