Yang Liu

← Primer

HSC · Healing Source Code

重启胎儿血红蛋白——这条治疗逻辑是病人先演示出来的 Reactivating fetal hemoglobin — the therapeutic logic patients demonstrated first

2026-09-27 · HbF, 血红蛋白病, 人群遗传学, 镰刀型贫血, 表型修饰

一个在出生前后被关掉的开关

人在胎儿期造的血红蛋白和出生以后造的不是同一种。胎儿血红蛋白（fetal hemoglobin，HbF）由两条 α 链配两条 γ 链组成，γ 链来自 HBB 基因簇里的 HBG1 和 HBG2；出生前后几个月，γ 链的产量迅速让位给 β 链，到一岁左右，HbF 在多数人血里已经降到百分之一上下。γ 基因并没有被删掉，它只是被沉默了——一个完整、可用、却被按下静音的基因。

这在很长时间里只是发育生物学的好题目。把它变成治疗问题的是另一个观察：镰刀型贫血那个让 β 链第六位谷氨酸变成缬氨酸的点突变只落在 β 链上，β-地中海贫血的绝大多数突变也是如此。γ 链完好无损。换句话说，这两类病人体内都躺着一个没坏的替补，而且这个替补造出来的四聚体不参与 HbS 的聚合，也能把 α 链配平。

于是问题变得非常具体：能不能把静音解除？这条思路听起来像是实验室里先算出来、再去人身上验证的。事实恰好相反。

良性的例外

遗传性胎儿血红蛋白持续症（hereditary persistence of fetal hemoglobin，HPFH）是成年后仍持续表达 γ-globin 的状态。它本身通常不构成疾病，携带者往往是在家系筛查里被顺带发现的。真正有分量的是另一种组合——当 HPFH 与镰刀型突变同时遗传给同一个人。

Ngo 等人在 2012 年的《British Journal of Haematology》上报告了 30 例经分子诊断确认的 HbS-HPFH。这些人的 HbF 在婴儿期高达 50%–90%，随后陡降，到三至五岁之间稳定下来；五岁以上者的 HbF 平均为 31%±3%，血红蛋白浓度平均 130±10 g/L，平均红细胞体积 75±4 fl。作者在摘要第一句里就写下了这项研究最要紧的事实：这些人「即使有，也极少出现镰刀型贫血相关的并发症」。

值得停下来想一想这句话的分量：一群人携带着与重症患者完全相同的致病突变，却基本不生病。这不是细胞模型，不是小鼠，是人；而这个「实验」的设计者不是任何一位科学家，是遗传重组本身。

一旦这类家系被看见，下一个问题就自然浮出来：需要多少 HbF 才够？Powars 等人在 1984 年干脆把问题写成了《Blood》上的论文标题——「镰刀型贫血中是否存在一个能减轻病情的胎儿血红蛋白阈值？」这个问句四十年后依然没有普适答案，但它把 HbF 从一个定性的「保护因素」推向了一个需要定量的临床变量。

一个可以测量、而且高度遗传的性状

要定量，先得承认 HbF 在成人身上并不均匀铺开。它集中在一小群被称为 F 细胞（F cells）的红细胞里，其余细胞几乎不含 HbF；F 细胞比例因此成了比总 HbF 更稳定的表型指标。

2000 年，Garner 等在《Blood》上发表了一项双生子研究：264 对同卵、511 对同性别异卵双生子，全部来自未经筛选的普通人群。结果是 F 细胞水平的变异有 89% 可归因于遗传因素。

这个数字改变了研究的性质。它意味着 HbF 不是随机的生理波动，而是一个受强遗传控制的数量性状；而数量性状是可以被定位的。从这一刻起，「怎么重启 HbF」这个药理学问题被翻译成了一个人群遗传学问题：找出那些让某些人天生 HbF 偏高的位点。

把信号钉到染色体上

定位工作走了将近二十年，最终收敛到三个位点。

最早浮现的信号在 HBB 基因簇自己身上。1985 年，Labie 等在《PNAS》上报告，β-地中海贫血与镰刀型贫血患者中，高 Gγ-globin 表达与高 HbF 水平依赖于一个共同的单倍型；同年，Gilman 与 Huisman 在《Blood》上把这个关联落到了 HBG2 上游的一处序列变异。这个位点后来在文献里被固定称作 Xmn1-HBG2——以限制酶切位点命名，因为在那个年代这就是能拿到的最精细分辨率。

第二个信号来自家系连锁分析。1996 年，Craig 等在《Nature Genetics》上用回归式方法同时解剖了 11p 与 6q 上控制 HbF 的位点，把一个显著信号指向 6 号染色体长臂。这个信号在原地放了十一年，直到 2007 年 Thein 等在《PNAS》上完成精细定位：真正负责的变异位于 HBS1L 与 MYB 之间的基因间区——不在任何一个基因的编码序列里，而在中间那段过去会被跳过的 DNA 上。

第三个信号出现在同一年。Menzel 等在《Nature Genetics》上报告，一个影响 F 细胞产量的数量性状位点定位到 2p15 上「一个编码锌指蛋白的基因」。论文标题里没有写出这个基因的名字，因为当时还没人知道它在红细胞里做什么。它叫 BCL11A。次年，Uda 等在《PNAS》上把它带进了临床表型：在撒丁岛的 β-地中海贫血人群中，BCL11A 的变异同时关联着 HbF 持续存在和病情减轻。

三个位点凑齐之后，它们合起来能解释多少？2008 年，Lettre 等在《PNAS》上把 BCL11A、HBS1L-MYB 与 β-globin 位点的多态性放在一起分析，样本是 1275 例来自美国镰刀型贫血合作研究（CSSCD）的患者与 350 例巴西队列患者。五个 SNP 合计解释了 HbF 水平变异的 20% 以上，其中单个影响最大的变异（BCL11A 的 rs4671393）在 CSSCD 队列中就占 14.1%。

二十个百分点是双面的：对一个复杂性状这已是极高的解释力，但它同时说明 HbF 变异的大部分仍然没有归属。

高 HbF 是否真的换来更好的结局

遗传学指出了一个开关，但要让它成为治疗靶点还需要一层证据：HbF 水平与临床结局的关系必须在足够大的前瞻队列里成立。

CSSCD 提供了这层证据。1991 年，Platt 等在《NEJM》上报告了对 3578 名患者的前瞻性随访，系统刻画了疼痛发作的发生率与风险因素——这是第一次有人把「危机」这个临床印象变成可计数的率。1994 年，同一组人报告了这个队列的死亡率与预期寿命。按后续文献对该文的通行读法，在前羟基脲时代的 CSSCD 患者中，HbF 高于约 8.6% 与更好的生存相关。

这个数字值得小心对待。它来自一个特定人群、一个特定年代、一种特定的测量方式，而不是一条生物学常数。Lettre 等在 2008 年的分析里补上了一个旁证：与 HbF 相关的那些 SNP 所携带的疼痛发作率信息，超出了它们通过单次稳态 HbF 测量所能解释的部分。也就是说，一次抽血测到的 HbF 只是终生 HbF 暴露的一个快照，基因型比快照更接近真实的保护程度。

同样要说清的是，HbF 不是一条线性的保护曲线。它在细胞间的分布决定了有多少红细胞真正落在聚合阈值之下——同样的总 HbF，若集中在少数细胞里，剩下那些照旧会镰变。阈值也随单倍型移动：在阿拉伯-印度单倍型人群中，BCL11A 与 MYB 合起来只解释 HbF 变异的 8.8%，与非洲裔人群里的图景明显不同。

药物先替这条通路做了一次可治性验证

在有任何基因编辑手段之前，「升高 HbF 能改善病情」已经被药理学验证过了。

1982 年，Ley 等在《NEJM》上报告，5-azacytidine 在一名 β⁺-地中海贫血患者身上选择性地提高了 γ-globin 合成。这是单个病人的观察，分量有限，但它第一次说明成人的 γ 基因可以被药物重新唤醒。1984 年，Platt 等在《Journal of Clinical Investigation》上报告，羟基脲（hydroxyurea）能提高镰刀型贫血患者的胎儿血红蛋白产量。

十一年后，这条线走到了随机对照试验。1995 年，Charache 等在《NEJM》上报告了多中心羟基脲研究（MSH，NCT00000586）的结果：299 名成年患者随机接受羟基脲或安慰剂，羟基脲组每年疼痛发作 2.5 次，安慰剂组 4.5 次，试验因疗效明确而提前终止。2003 年，Steinberg 等在《JAMA》上报告了该队列的长期随访，把羟基脲的使用与死亡率和并发症联系起来。

需要分清的是，羟基脲的获益不能全部记在 HbF 账上。它同时降低白细胞与网织红细胞计数、改变细胞黏附、并可能释放一氧化氮，这些机制彼此纠缠；而且它升高 HbF 的幅度个体差异极大，部分依赖于一个仍有旺盛增殖能力的红系骨髓。羟基脲证明的不是「HbF 是唯一的答案」，而是「这条通路可以被撬动，而且撬动它能改变临床结局」。对于一个还没有任何靶向手段的时代，这已经是最要紧的一句话。

家系早就把分子标好了

最后一步是把人群遗传学的线索还原成分子事件，而这一步的收尾漂亮得近乎巧合。

有一类 HPFH 不是由大片段缺失引起的，而是 γ-globin 基因启动子上的点突变造成的，位置集中在转录起始位点上游约 115 bp 和 200 bp 两处。2018 年，Martyn 等在《Nature Genetics》上报告，这两处正是两个主要胎儿球蛋白抑制因子的结合位点：BCL11A 直接结合 −115 处，ZBTB7A（又称 LRF）直接结合 −200 处。他们进一步用 CRISPR-Cas9 把这些天然的 HPFH 突变引入红系细胞，观察到抑制因子结合被破坏、γ-globin 表达升高。这一步是在细胞层面完成的，但它的意义在别处：那些在家系里被标注为「良性变异」的位点，其实是人群在机制被讲清楚之前几十年，就已经替研究者把抑制因子的落脚点圈了出来。

这条链条到这里闭合了。2008 年 Sankaran 等在《Science》上确立了 BCL11A 作为发育阶段特异性抑制因子的身份，把 2p15 上那个「编码锌指蛋白的基因」变成了一个有机制的靶点；今天临床上真正在做的两类操作——破坏 BCL11A 的红系增强子，或者直接在 γ-globin 启动子上复刻一个 HPFH 突变——都是在这张由家系、双生子和队列共同画出来的地图上选落点。这些路线之间的取舍，以及 BCL11A 本身的故事，各有各的篇幅可说。

还没有答案的部分

HbF 变异中仍有大半找不到归属，意味着调控网络里还有没被定位的节点。阈值问题也没有解决：不同单倍型背景下同样的 HbF 百分比换来的保护并不相同，「多少才够」在个体层面仍是经验判断。至于把 γ 基因长期开着是否有代价——它毕竟是被进化按下静音的基因——目前只有随访有限的人体数据可以回答。

但这条逻辑链本身有一种少见的完整性。它从一群不生病的病人开始，经过双生子的血涂片、家系的连锁分析、大队列的疼痛日记和一次提前终止的随机对照试验，最后落在两个抑制因子距转录起始位点一百多个碱基的结合位点上。整条链上没有哪一步是靠聪明的机制推演跳过去的。

出生后那几个月里被关掉的开关从来没有被拆掉。它只是被盖住了，而人群里一直有人的盖子没盖严——是这些人，而不是任何一个实验，先告诉了医学：这个开关值得去找。

参考文献

  1. Powars DR, Weiss JN, Chan LS, Schroeder WA. Is there a threshold level of fetal hemoglobin that ameliorates morbidity in sickle cell anemia? Blood. 1984;63(4):921-926. DOI

  2. Labie D, Pagnier J, Lapoumeroulie C, et al. Common haplotype dependency of high G gamma-globin gene expression and high Hb F levels in beta-thalassemia and sickle cell anemia patients. Proc Natl Acad Sci USA. 1985;82(7):2111-2114. DOI

  3. Gilman JG, Huisman TH. DNA sequence variation associated with elevated fetal G gamma globin production. Blood. 1985;66(4):783-787. DOI

  4. Platt OS, Thorington BD, Brambilla DJ, et al. Pain in sickle cell disease: rates and risk factors. N Engl J Med. 1991;325(1):11-16. DOI

  5. Platt OS, Brambilla DJ, Rosse WF, et al. Mortality in sickle cell disease: life expectancy and risk factors for early death. N Engl J Med. 1994;330(23):1639-1644. DOI

  6. Craig JE, Rochette J, Fisher CA, et al. Dissecting the loci controlling fetal haemoglobin production on chromosomes 11p and 6q by the regressive approach. Nat Genet. 1996;12(1):58-64. DOI

  7. Garner C, Tatu T, Reittie JE, et al. Genetic influences on F cells and other hematologic variables: a twin heritability study. Blood. 2000;95(1):342-346. DOI

  8. Thein SL, Menzel S, Peng X, et al. Intergenic variants of HBS1L-MYB are responsible for a major quantitative trait locus on chromosome 6q23 influencing fetal hemoglobin levels in adults. Proc Natl Acad Sci USA. 2007;104(27):11346-11351. DOI

  9. Menzel S, Garner C, Gut I, et al. A QTL influencing F cell production maps to a gene encoding a zinc-finger protein on chromosome 2p15. Nat Genet. 2007;39(10):1197-1199. DOI

  10. Uda M, Galanello R, Sanna S, et al. Genome-wide association study shows BCL11A associated with persistent fetal hemoglobin and amelioration of the phenotype of β-thalassemia. Proc Natl Acad Sci USA. 2008;105(5):1620-1625. DOI

  11. Lettre G, Sankaran VG, Bezerra MAC, et al. DNA polymorphisms at the BCL11A, HBS1L-MYB, and β-globin loci associate with fetal hemoglobin levels and pain crises in sickle cell disease. Proc Natl Acad Sci USA. 2008;105(33):11869-11874. DOI

  12. Sankaran VG, Menne TF, Xu J, et al. Human fetal hemoglobin expression is regulated by the developmental stage-specific repressor BCL11A. Science. 2008;322(5909):1839-1842. DOI

  13. Ley TJ, DeSimone J, Anagnou NP, et al. 5-Azacytidine selectively increases γ-globin synthesis in a patient with β⁺ thalassemia. N Engl J Med. 1982;307(24):1469-1475. DOI

  14. Platt OS, Orkin SH, Dover G, Beardsley GP, Miller B, Nathan DG. Hydroxyurea enhances fetal hemoglobin production in sickle cell anemia. J Clin Invest. 1984;74(2):652-656. DOI

  15. Charache S, Terrin ML, Moore RD, et al. Effect of hydroxyurea on the frequency of painful crises in sickle cell anemia. N Engl J Med. 1995;332(20):1317-1322. DOI（MSH，ClinicalTrials.gov NCT00000586）

  16. Steinberg MH, Barton F, Castro O, et al. Effect of hydroxyurea on mortality and morbidity in adult sickle cell anemia. JAMA. 2003;289(13):1645-1651. DOI

  17. Ngo DA, Aygun B, Akinsheye I, et al. Fetal haemoglobin levels and haematological characteristics of compound heterozygotes for haemoglobin S and deletional hereditary persistence of fetal haemoglobin. Br J Haematol. 2012;156(2):259-264. DOI

  18. Martyn GE, Wienert B, Yang L, et al. Natural regulatory mutations elevate the fetal globin gene via disruption of BCL11A or ZBTB7A binding. Nat Genet. 2018;50(4):498-503. DOI

A switch closed around the time of birth

The hemoglobin a fetus makes is not the hemoglobin an adult makes. Fetal hemoglobin (HbF) pairs two α chains with two γ chains, the γ chains coming from HBG1 and HBG2 in the HBB cluster. Over the months surrounding birth, γ production gives way to β, and by about a year of age HbF has fallen to roughly one percent of the total in most people. The γ genes are not deleted. They are silenced — intact, functional, muted.

For a long time this was a good developmental biology problem and nothing more. What turned it into a therapeutic one was a second observation: the point mutation that turns the sixth residue of β-globin from glutamate to valine lands only on the β chain, and so do nearly all β-thalassemia mutations. The γ chain is untouched. Both groups of patients therefore carry an undamaged substitute — one whose tetramers do not join HbS polymer and can still balance the α chains being made.

The question becomes very concrete: can the mute button be released? This sounds like a line of reasoning worked out in a laboratory and then tested in people. It ran the other way.

The benign exception

Hereditary persistence of fetal hemoglobin (HPFH) is a state in which γ-globin expression continues into adulthood. On its own it is usually not a disease; carriers are often found incidentally during family studies. What carries real weight is the combination — HPFH and the sickle mutation inherited by the same person.

In 2012, Ngo and colleagues reported in the British Journal of Haematology on 30 molecularly confirmed cases of HbS-HPFH. HbF ran between 50% and 90% in infancy, fell steeply, and stabilized between ages three and five; in those aged five or older it averaged 31%±3%, with hemoglobin concentration averaging 130±10 g/L and mean corpuscular volume 75±4 fl. The abstract’s opening sentence states the finding that matters most: these individuals have “few, if any, sickle cell disease-related complications.”

It is worth pausing on that. A group of people carrying exactly the mutation that devastates others, and largely not getting sick. Not a cell model, not a mouse — people. And the experiment was designed by genetic recombination, not by any scientist.

Once such families are recognized, the next question arrives on its own: how much HbF is enough? In 1984 Powars and colleagues simply made the question their title in Blood — “Is there a threshold level of fetal hemoglobin that ameliorates morbidity in sickle cell anemia?” Four decades later there is still no universal answer, but the question moved HbF from a qualitative protective factor to a clinical variable someone had to quantify.

A trait that can be measured, and is strongly heritable

Quantifying it means first accepting that HbF is not spread evenly across adult red cells. It concentrates in a minority population called F cells, while the rest contain almost none. The F-cell fraction is therefore a steadier phenotype than total HbF.

In 2000, Garner and colleagues published a twin study in Blood: 264 monozygotic and 511 same-sex dizygotic pairs, drawn from unselected twins. Eighty-nine percent of the variation in F-cell levels was attributable to genetic factors.

That number changed the nature of the problem. It meant HbF was not physiological noise but a quantitative trait under strong genetic control — and quantitative traits can be mapped. From that point on, the pharmacological question of how to reactivate HbF had been translated into a population-genetics question: find the loci that make some people naturally high.

Pinning the signal to chromosomes

The mapping took nearly two decades and converged on three loci.

The first signal came from inside the HBB cluster itself. In 1985, Labie and colleagues reported in PNAS that in both β-thalassemia and sickle cell anemia, high Gγ-globin expression and high HbF depended on a shared haplotype; the same year, Gilman and Huisman localized the association in Blood to a sequence variation upstream of HBG2. The locus has since been known in the literature as Xmn1-HBG2 — named for a restriction site, because in that era a restriction site was the finest resolution available.

The second came from family linkage analysis. In 1996, Craig and colleagues used a regressive approach in Nature Genetics to dissect HbF-controlling loci on 11p and 6q simultaneously, pointing a significant signal at the long arm of chromosome 6. It sat there for eleven years, until Thein and colleagues completed the fine-mapping in PNAS in 2007: the responsible variants lie in the intergenic region between HBS1L and MYB — in neither gene’s coding sequence, but in the stretch of DNA between them that earlier work would have skipped over.

The third appeared the same year. Menzel and colleagues reported in Nature Genetics that a QTL influencing F-cell production mapped to “a gene encoding a zinc-finger protein on chromosome 2p15.” The title does not name the gene, because at the time nobody knew what it did in red cells. It was BCL11A. The following year Uda and colleagues carried it into clinical phenotype in PNAS: in the Sardinian β-thalassemia population, BCL11A variation was associated both with persistent HbF and with amelioration of disease.

With three loci in hand, how much do they explain together? In 2008, Lettre and colleagues analyzed polymorphisms at BCL11A, HBS1L-MYB and the β-globin locus jointly in 1,275 patients from the Cooperative Study of Sickle Cell Disease (CSSCD) and 350 from a Brazilian cohort. Five SNPs together accounted for more than 20% of the variation in HbF levels, with the single strongest variant, BCL11A rs4671393, accounting for 14.1% in CSSCD.

Twenty percent cuts both ways. For a complex trait it is a very high figure — and it says just as plainly that most of the variance in HbF remains unaccounted for.

Does more HbF actually buy better outcomes

Genetics had identified a switch. Turning it into a therapeutic target required one more layer: the relationship between HbF level and clinical outcome had to hold up in a cohort large enough to trust.

CSSCD supplied that layer. In 1991, Platt and colleagues reported in the New England Journal of Medicine on prospective follow-up of 3,578 patients, characterizing the rates of and risk factors for painful episodes — the first time the clinical impression of “crisis” became a countable rate. In 1994 the same group reported the cohort’s mortality and life expectancy. On the reading of that paper that the subsequent literature has settled on, HbF above roughly 8.6% was associated with improved survival among CSSCD patients in the pre-hydroxyurea era.

That figure deserves care. It comes from one population, one era, one way of measuring — not from a biological constant. Lettre’s 2008 analysis added a revealing corollary: the HbF-associated SNPs carried information about pain rate beyond what they explained through a single steady-state HbF measurement. A single blood draw is a snapshot of a lifetime of HbF exposure, and genotype sits closer to the real degree of protection than the snapshot does.

It also has to be said that HbF is not a linear protective curve. Its distribution across cells determines how many red cells actually fall below the polymerization threshold; the same total HbF, concentrated in a minority of cells, leaves the rest to sickle as before. And the threshold moves with haplotype: in the Arab-Indian haplotype population, BCL11A and MYB together account for only 8.8% of HbF variance, a markedly different picture from that in populations of African descent.

Drugs established that the pathway was actionable

Before any gene-editing tool existed, the proposition that raising HbF improves disease had already been validated pharmacologically.

In 1982, Ley and colleagues reported in the New England Journal of Medicine that 5-azacytidine selectively increased γ-globin synthesis in a patient with β⁺ thalassemia. A single patient carries limited weight, but it was the first demonstration that adult γ genes could be reawakened by a drug. In 1984, Platt and colleagues reported in the Journal of Clinical Investigation that hydroxyurea enhanced fetal hemoglobin production in sickle cell anemia.

Eleven years later the line reached a randomized trial. In 1995, Charache and colleagues reported the Multicenter Study of Hydroxyurea (MSH, NCT00000586) in the New England Journal of Medicine: 299 adults randomized to hydroxyurea or placebo, with 2.5 painful crises per year on hydroxyurea against 4.5 on placebo, and the trial stopped early for efficacy. In 2003, Steinberg and colleagues reported long-term follow-up of the cohort in JAMA, linking hydroxyurea use to mortality and morbidity.

The distinction to keep clear is that hydroxyurea’s benefit cannot all be booked to HbF. It simultaneously lowers white cell and reticulocyte counts, alters cell adhesion, and may release nitric oxide — mechanisms that are entangled with one another. Its effect on HbF also varies enormously between individuals and depends in part on an erythroid marrow still capable of vigorous proliferation. What hydroxyurea proved was not that HbF is the whole answer, but that the pathway can be moved, and that moving it changes clinical outcomes. For an era with no targeted option at all, that was the sentence that mattered.

The families had already marked the molecules

The last step was reducing the population-genetics clue to a molecular event, and the ending is almost too neat.

One class of HPFH is caused not by large deletions but by point mutations in the γ-globin promoter, clustered around 115 and 200 base pairs upstream of the transcription start site. In 2018, Martyn and colleagues reported in Nature Genetics that these are precisely the binding sites of the two major fetal globin repressors: BCL11A binds directly at −115, ZBTB7A (also known as LRF) at −200. They then used CRISPR–Cas9 to introduce the naturally occurring HPFH mutations into erythroid cells and saw repressor binding disrupted and γ-globin expression rise. That step was done at the cell level, but its significance lies elsewhere: the positions catalogued in family studies as “benign variants” were, decades before the mechanism was described, the population’s own annotation of where the repressors sit.

The chain closes there. In 2008, Sankaran and colleagues established in Science that BCL11A is a developmental-stage-specific repressor, turning that “gene encoding a zinc-finger protein” on 2p15 into a target with a mechanism. The two operations now being performed clinically — disrupting the erythroid enhancer of BCL11A, or recreating an HPFH mutation directly in the γ-globin promoter — both choose their landing sites on a map drawn jointly by pedigrees, twins and cohorts. The trade-offs among those routes, and the story of BCL11A as a molecule in its own right, each deserve their own treatment.

What is still open

Most of the variance in HbF remains unassigned, which means the regulatory network still contains unmapped nodes. The threshold problem is unsolved: the same HbF percentage buys different protection against different haplotype backgrounds, and “how much is enough” remains a judgment rather than a calculation at the individual level. Whether keeping the γ genes switched on for decades carries a cost — they were, after all, muted by evolution — can so far only be answered by human data with limited follow-up.

But the logic itself has an unusual completeness. It begins with patients who do not get sick, runs through twins’ blood films, family linkage analysis, a large cohort’s pain diaries and a randomized trial stopped early, and lands on two repressor binding sites a hundred-odd bases from a transcription start site. Not one step of it was skipped by clever mechanistic deduction.

The switch closed in the months around birth was never removed. It was only covered — and in every population there have been people whose cover never quite sealed. It was those people, not any experiment, who first told medicine that the switch was worth looking for.

References

  1. Powars DR, Weiss JN, Chan LS, Schroeder WA. Is there a threshold level of fetal hemoglobin that ameliorates morbidity in sickle cell anemia? Blood. 1984;63(4):921-926. DOI

  2. Labie D, Pagnier J, Lapoumeroulie C, et al. Common haplotype dependency of high G gamma-globin gene expression and high Hb F levels in beta-thalassemia and sickle cell anemia patients. Proc Natl Acad Sci USA. 1985;82(7):2111-2114. DOI

  3. Gilman JG, Huisman TH. DNA sequence variation associated with elevated fetal G gamma globin production. Blood. 1985;66(4):783-787. DOI

  4. Platt OS, Thorington BD, Brambilla DJ, et al. Pain in sickle cell disease: rates and risk factors. N Engl J Med. 1991;325(1):11-16. DOI

  5. Platt OS, Brambilla DJ, Rosse WF, et al. Mortality in sickle cell disease: life expectancy and risk factors for early death. N Engl J Med. 1994;330(23):1639-1644. DOI

  6. Craig JE, Rochette J, Fisher CA, et al. Dissecting the loci controlling fetal haemoglobin production on chromosomes 11p and 6q by the regressive approach. Nat Genet. 1996;12(1):58-64. DOI

  7. Garner C, Tatu T, Reittie JE, et al. Genetic influences on F cells and other hematologic variables: a twin heritability study. Blood. 2000;95(1):342-346. DOI

  8. Thein SL, Menzel S, Peng X, et al. Intergenic variants of HBS1L-MYB are responsible for a major quantitative trait locus on chromosome 6q23 influencing fetal hemoglobin levels in adults. Proc Natl Acad Sci USA. 2007;104(27):11346-11351. DOI

  9. Menzel S, Garner C, Gut I, et al. A QTL influencing F cell production maps to a gene encoding a zinc-finger protein on chromosome 2p15. Nat Genet. 2007;39(10):1197-1199. DOI

  10. Uda M, Galanello R, Sanna S, et al. Genome-wide association study shows BCL11A associated with persistent fetal hemoglobin and amelioration of the phenotype of β-thalassemia. Proc Natl Acad Sci USA. 2008;105(5):1620-1625. DOI

  11. Lettre G, Sankaran VG, Bezerra MAC, et al. DNA polymorphisms at the BCL11A, HBS1L-MYB, and β-globin loci associate with fetal hemoglobin levels and pain crises in sickle cell disease. Proc Natl Acad Sci USA. 2008;105(33):11869-11874. DOI

  12. Sankaran VG, Menne TF, Xu J, et al. Human fetal hemoglobin expression is regulated by the developmental stage-specific repressor BCL11A. Science. 2008;322(5909):1839-1842. DOI

  13. Ley TJ, DeSimone J, Anagnou NP, et al. 5-Azacytidine selectively increases γ-globin synthesis in a patient with β⁺ thalassemia. N Engl J Med. 1982;307(24):1469-1475. DOI

  14. Platt OS, Orkin SH, Dover G, Beardsley GP, Miller B, Nathan DG. Hydroxyurea enhances fetal hemoglobin production in sickle cell anemia. J Clin Invest. 1984;74(2):652-656. DOI

  15. Charache S, Terrin ML, Moore RD, et al. Effect of hydroxyurea on the frequency of painful crises in sickle cell anemia. N Engl J Med. 1995;332(20):1317-1322. DOI (MSH; ClinicalTrials.gov NCT00000586)

  16. Steinberg MH, Barton F, Castro O, et al. Effect of hydroxyurea on mortality and morbidity in adult sickle cell anemia. JAMA. 2003;289(13):1645-1651. DOI

  17. Ngo DA, Aygun B, Akinsheye I, et al. Fetal haemoglobin levels and haematological characteristics of compound heterozygotes for haemoglobin S and deletional hereditary persistence of fetal haemoglobin. Br J Haematol. 2012;156(2):259-264. DOI

  18. Martyn GE, Wienert B, Yang L, et al. Natural regulatory mutations elevate the fetal globin gene via disruption of BCL11A or ZBTB7A binding. Nat Genet. 2018;50(4):498-503. DOI