Small molecule therapeutics to treat the β-globinopathies

Structured abstract:

Purpose of review:

This review focuses on recent insights into small molecule-mediated therapeutics in development to treat the β-globinopathies.

Recent findings:

Recent investigation of fetal γ-globin gene regulation provides multiple insights into how γ-globin gene reactivation may lead to novel treatment for β-globinopathies.

Summary:

We summarize current information regarding the binding of transcription factors [TFs] that appear to be impeded or augmented by different hereditary persistence of fetal hemoglobin [HPFH] mutations. As TFs have historically proven to be difficult to target for therapeutic purposes, we next address the protein complexes associated with these HPFH mutation-affected TFs with the aim of defining proteins that might provide additional targets for chemical molecules to inactivate the co-repressors. Among the proteins associated with the TF complexes, a group of co-repressors with currently available inhibitors were initially thought to be good candidates for potential therapeutic purposes. We discuss possibilities for pharmacological inhibition of these co-repressor enzymes that might significantly re-activate fetal γ-globin gene expression. Finally, we summarize the current clinical trial data regarding the inhibition of select co-repressor proteins for the treatment of sickle cell disease and β-thalassemia.

Introduction

The β-globinopathies, sickle cell disease and Cooley’s anemia [SCD and CA], comprise the world’s most common hereditary monogenic diseases and are designated as a global health burden by the World Health Organization[1-3]. SCD was shown to be the consequence of recessive inheritance of a mutant gene among Americans of African descent by James Neel in 1949[4]; this genetic proof was followed shortly by the famous biochemical proof from the Pauling laboratory where SCD was first described as “…a molecular disease” [5]. After five more decades of study and evaluation, two key observations made it clear that reactivation of the fetal γ-globin genes in the red blood cells of children or adult patients would almost certainly play an important role in the eventual treatment of SCD. First, a detailed clinical study of a cohort of thousands of SCD patients clearly demonstrated that those with higher levels of fetal hemoglobin [HbF] had an increased life expectancy of some 25-30 years, and that “The most straightforward laboratory risk factor [for the improved longevity] was the fetal hemoglobin level” [6]. Furthermore, it was discovered, and became more clear as multiple patients with a similar phenotype were identified, that a variety of mutations in either of the γ-globin gene promoters could lead to chronically, and gene autonomously, elevated expression of the fetal globin gene bearing the mutation [referred to as HPFH mutations, for hereditary persistence of fetal hemoglobin][7]. These observations, and a subsequent deluge of confirming documentation, established the extant strategy for combating both SCD and CA [β-thalassemia major]: find a way to elevate the abundance of HbF in red blood cells to levels that can either interrupt sickle hemoglobin polymerization [for SCD] or complement the reduced levels of hemoglobin A [HbA] [in CA] to promote longer RBC lifespan! Since around 1985, several strategies have emerged. One is pharmacological, and extensive studies are ongoing to identify better and safer drugs. Two other strategies, not the subject of this review, are genetic. The first is gene therapy using lentiviral vectors to deliver the missing β-globin chains to patients with CA, and recent clinical studies show tremendous promise for this approach[8]. The second genetic approach is to employ gene editing to correct various lesions in the β-globin locus[8], but these are still in their infancy and may encounter difficult issues clinically in providing proof that off target effects of this strategy will not be harmful to patients.

To date, only two small molecule chemical therapeutics, hydroxyurea [HU] and L-glutamine, have been approved to treat SCD by the U.S. Food and Drug Administration [FDA]. Neither of these medications is effective in more than about half of all SCD patients, and thus more therapeutic options are critically required. Therefore, the purpose of this review is to highlight recent progress in developing small chemical HbF inducers that might prove to be useful in mediating HbF induction for the treatment of SCD, which will also be explored to treat CA.

Text of review

1. The pathology of SCD

Sickle cell disease is caused by the production of a mutant form of the β subunit [β^s] of hemoglobin that is incorporated with α-globin chains that, together, comprise an adult hemoglobin variant [HbS; α₂β₂^S]. Red blood cells filled with HbS exhibit a propensity under stressful conditions to polymerize within red blood cells [Figure 1A][9]. Patients with SCD suffer both acute and chronic complications that are attributed to the deoxygenated polymerization of HbS and consequent lysis of red blood cells[6].

Figure1. Primary and secondary targets of small molecule inhibitors for treating pathological symptoms of SCD.

A. Compounds [indicated by the red and white “pill”] targeting the proximal causes of SCD act to reduce sickling in RBCs either by directly reducing deoxygenated HbS polymerization [left; orange a-globin plus blue b^S- HbS tetrameric polymers] or indirectly by inducing HbF [right, orange α-globin + yellow γ-globin chains], which interrupt and terminate sickle HbS polymerization. B. Both acute and chronic complications result from increased ssRBC adherence to endothelial cells [brown] and leukocytes [grey with irregular nuclei] resulting in multicellular aggregates. These, in turn, cause cellular and molecular pathologies leading to downstream sequelae [multiorgan damage, brain infarcts, acute chest and bone pain and shortened lifespan]. These molecular pathologies are the target of therapies intended to reduce vascular occlusion and chronic injury.

Hemoglobin exists in different conformations dependent on whether or not it is bound to oxygen[10]. In its deoxygenated form, HbS polymerizes to form long, stiff strands that are capable of engendering structural changes inside red blood cells [RBCs] that result in the pathognomonic sickle cell shape[11]. Sickled RBCs [ssRBCs] behave differently than normal RBCs due to their decreased deformability and increased adhesive interactions with endothelial cells and leukocytes[12,13], the combination of which leads to vascular occlusion [VOC] in small blood vessels[14], thromboses within large blood vessels[15], and hemolytic anemia upon RBC rupture [Figure 1B][16]. The effects of each of these factors are amplified by positive feedback loops that further drive VOC. During VOC events, the supply of oxygenated blood is disrupted leading to tissue ischemia, hypoxic injury, and inflammation followed eventually by reperfusion injury[17]. The resulting effects manifest in episodes of severe pain, pulmonary/hepatic /renal injury[18-20], splenic sequestration[21], infection[22,23], and an overall diminished life expectancy[6].

Potentially curative options for sickle cell disease include bone marrow transplantation to replace the mutation-bearing hematopoietic stem cells [HSC]s in SCD patients with those of a compatible donor. As this topic has been recently reviewed elsewhere it will not be the focus of this review[24*]. Recently, there has been significant progress in utilizing gene therapy to modify HSCs and progenitors in SCD patients to treat the disease. While it is not yet ready for widespread implementation for treatment purposes, there are numerous ongoing trials assessing the efficacy of this approach for treating SCD and CA[25,26*,27]. However, given the demographics of disease distribution[28], genetic therapies likely will not be widely implemented in the near future due to the limited accessibility of such approaches for the vast majority of patients.

2. Therapeutic treatment strategies for SCD: from HbF induction and anti-sickling to symptomatic treatment

2.1. HbF induction

Both clinical and laboratory evidence indicates that increasing the levels of HbF alleviates the symptoms of SCD. This paradigm has been validated in studies where it was shown that HPFH, a natural human genetic variant in which high levels of fetal γ-globin synthesis aberrantly persist into adulthood[29], ameliorates many of the lesions associated with SCD and CA. Of note, it has been observed clinically that when an HPFH mutation is co-inherited with β-globinopathies, the elevated production of β-globin significantly mitigates disease symptoms, making increased HbF production an attractive target for the treatment of CA and SCD[30-33].

HbF does not intercalate with deoxygenated HbS polymers, thereby interrupting the conversion of RBCs to sickled RBCs when β-globin is present at sufficient concentrations[34,35]. The first drug approved for the treatment of SCD, hydroxyurea, acts to increase HbF[36]. The HbF increases associated with HU correlated with extended RBC survival[37,38], fewer acute pain crises[39], and higher quality of life[40]. Unfortunately, fewer than half of all SCD patients are responsive to HU treatment, and the responses after 2 years of treatment were modest and diminished over time[39,41]. The lack of robust and reliable compounds to interrupt HbS polymerization necessitates further study to identify novel compounds capable of efficiently inducing γ-globin for sustained, long-term use.

2.2. Anti-sickling

The sickling of RBCs is responsible for the pathophysiology of SCD, therefore considerable efforts are underway to identify compounds capable of inhibiting HbS polymerization. While HbF induction is one strategy to inhibit HbS polymerization, others have shown promising results in preventing sickling by increasing the affinity of hemoglobin for oxygen. Voxelotor [formerly GBT-440], an experimental drug that increases hemoglobin oxygen affinity, is currently under review by the FDA seeking to be the first drug approved to address inhibition of the sickling process itself within ssRBCs. Phase 3 clinical trials of Voxelotor reported a modest increase in total hemoglobin [1g/dl] amongst trial participants that was also associated with a decrease in hemolysis, although there was no discernable decrease in vascular occlusive crises[42].

2.3. Secondary treatment

While HbS polymerization is the primary cause of SCD, many of the pathological consequences of the disease can be managed by addressing the downstream effects of RBC sickling such as VOC events. While these events are multifactorial, adhesive interactions are thought to be a primary cause of acute episodes[13]. ssRBCs have been shown to abnormally adhere to endothelial cells lining the blood vessels, and the degree to which the RBC adhere to endothelial cells is commensurate with the clinical severity of disease in the SCD donor[43]. ssRBC interactions are not limited to endothelial cells, as they also bind to and activate circulating leukocytes, resulting not only in the adhesion of the activated leukocyte to endothelial cells but enabling them to recruit many more ssRBCs, further occluding blood flow[14,44]. Studies aimed at targeting the selectins that mediate these adhesive interactions have reported only limited success, without achieving the success of hydroxyurea in reducing VOC events[45-49]. These interactions are the subject of numerous ongoing studies and clinical trials, and are currently reviewed here[50,51].

There are a multitude of drugs in development that are intended to target different sequelae of HbS polymerization, including hemolysis, ssRBC adhesion molecules, vascular occlusion, oxidation, inflammation, and coagulation. Other symptomatic treatments include analgesics, blood transfusions, and stimulators of erythropoiesis. Small molecules that target other processes have been nicely reviewed recently[52**], therefore, we will focus on HbF induction as the primary purpose of the review.

3. Developmental Regulation of Human β-type Globin Gene Expression

The human β-globin locus extends over 70 kb in chromatin, and is composed of the ε- [embryonic], Gγ- and Aγ- [fetal] and δ- and β-globin [adult] structural genes, which are spatially arranged in the locus in the same order [5’ to 3’ from the locus control region [LCR]] as they are expressed during development. The embryonic β-globin gene is transcribed during the first 8 weeks of human gestation in primitive erythroid cells derived from the yolk sac, the major site of erythropoiesis in the early embryo. The first switch in β-type globin gene transcription results in the silencing of γ-globin and concomitant activation of the fetal γ-globin genes when definitive erythroid cells first emerge from the fetal liver. Gradually, at around the time of birth, a second switch from γ- to β-globin transcription occurs coordinately as the site of hematopoiesis shifts once again to the adult bone marrow. From genetic analysis of transgenic mice harboring in vitro modifications in large representations of the human β-globin loci [in bacterial artificial chromosomes [BACs] or yeast artificial chromosomes [YACs]], two nonexclusive mechanisms for globin gene "switching" have been postulated: one is regulation by sequences located in the globin gene promoters [autonomous gene control], while the other is competition among the globin genes for activation by the LCR, the super-enhancer element required for abundant expression of all of the globin genes. In the gene competition model[53,54], with all other things being equal, the gene closer to LCR should have a higher probability of interaction with the LCR “holocomplex”[55] and hence be more abundantly transcribed, unless that gene is autonomously silenced[56]. Autonomous control plays a major role in the silencing of the human embryonic ε- and fetal γ-globin genes in definitive erythroid cells[57-59]. Thus, a better understanding of the mechanism of autonomous silencing of the two γ-globin genes in adult erythroid cells may well lead to insights into novel strategies for re-activation of fetal globin transcription for the benefit of CA and SCD patients.

4. HPFH mutations and transcription factor [TF] binding

Human genetic studies lead to an early breakthrough for our understanding of how the fetal γ-globin genes are autonomously silenced. Genetic variants in three loci, including the β-type globin cluster, the HBS1L-MYB intergenic region, and B-cell lymphoma/leukemia 11A [BCL11A], explained about half of heritable HbF variation as assessed by genome-wide association study[60-64]. Decreased abundance of transcription factor c-Myb was hypothesized, and retrospectively proven[65], to be responsible for the mutant HBS1L-MYB intergenic region γ-globin inductive phenotype, and BCL11a, previously characterized as an important B cell transcription factor, was shown to play a key role in fetal globin gene repression[65-67].

In HPFH, the autonomously affected fetal γ-globin gene is abundantly transcribed in adulthood, with elevated synthesis [up to 30%] of γ-globin produced in adult erythrocytes[68]. HPFH mutations include small and large deletions in the locus as well as point mutations in both of the γ-globin gene promoters. Of the documented naturally occurring non-deletion HPFH mutations, 15 are located within either the Aγ- or Gγ-globin proximal promoters[68,69] [Figure 2A]. These mutations were thought to either interfere with repressor binding, and thereby impair normal definitive cell repression on the fetal globin genes [−567 T-to-G for GATA1 binding[70]; −202 to −195 cluster for ZBTB7A binding[71**]; −118 to −102 cluster for Bcl11a and TR2/TR4 binding[58,71**,72,73**,74]], or create ectopic activating TF binding sites for fetal globin re-activation [−198 T-to-C for KLF1[75] and Sp1 binding[76]; −175 T-to-C for TAL1 binding[69]; −113 A-to-G for GATA1 binding[77*]] [Figure 2A].

Figure 2. HPFH mutations and the TFs known or hypothesized to be affected.

A. The position of the HPFH mutations and the nucleotide changes is indicated below the identical γ-globin promoter sequences. The approximate positions and identities of the best-characterized transcription factors are shown above the sequence, while mutations that lead to ectopic binding of activating transcription factors are highlighted in dark blue lettering. B. Epigenetic enzymes associated with fetal globin proximal promoter binding TFs and the epigenetic modification of each enzyme. G [GATA1]; Z [ZBTB7A]; T [TR2/TR4]; B [Bcl11a]; L [LYAR]. M [methylation of DNA or histone, + or − indicates the activity of that enzyme on that site]; A [histone acetylation]; U [histone ubiquitination].

5. Epigenetic enzymes bound by fetal globin gene repressors

The activity of TFs has historically been very difficult to affect pharmacologically. However, following the identification of several autonomous fetal globin gene repressors, the binding of which was impaired by naturally occurring HPFH mutations [Figure 2A], the data implicitly suggested that critical epigenetic co-repressor enzymes recruited by those repressors and were required to functionally silence the locus might serve as more attractive targets for fetal globin gene re-activation. Therefore, defining the protein complexes associated with each TF could provide valuable information about the identity, and possible therapeutic potential, of those co-repressors. Among all of the epigenetic co-repressors identified to date, we focused on enzyme co-repressors for which genetic or pharmacological inhibition data indicated their ability to significantly re-activate fetal γ-globin gene expression in this review [Figure 2B].

5.1. NuRD complex and HDACs

The nucleosome remodeling deacetylase [NuRD] complex is a chromatin remodeling and histone deacetylation enzymatic machine[78] that has been shown to associated with all four HPFH mutation-affected TFs [GATA1[79,80], ZBTB7A[81], Bcl11a[82], and TR2/4[83,84**]], as well as with LYAR, which binds to, and represses, γ-globin through a GGTTAT motif located 3’ to the transcription start site[85,86] [Figure 2A], suggesting that the NuRD complex is the most extensively involved enzymatic activity in fetal globin repression. NuRD contains two catalytic subunits [ATP-dependent nucleosome remodelers CHD3 and CHD4], histone deacetylases [HDAC1 and HDAC2], methyl-CpG binding proteins [MBD2 and MBD3], scaffolding proteins [MTA1/2/3], histone binding proteins [RBBP4/7] and structural proteins [GATAD2A and GATAD2B]. Within the NuRD complex, HDAC activity may be the most promising to therapeutically target by small molecules. A recent study using dense clustered regularly interspaced short palindromic repeats (CRISPR) mutagenesis screening of all NuRD complex subunits suggested that the core subunits that repress fetal globin gene expression include HDAC2, CHD4, MBD2, MTA2, GATAD2A[87*]. In this study, the CRISPR-mediated knockout of HDAC2, but not HDAC1, re-activated γ-globin in HUDEP2 cells, a definitive erythroid cell line that expresses predominantly HbA[87*]. However, in other studies using an shRNA knockdown strategy in human primary hematopoietic stem and progenitor cell [HSPC] culture, depletion of either HDAC1 or HDAC2 induced fetal globin expression[82,88]. This discrepancy may due to the different methods used to induce loss of function or to differences in the cells examined. Notably, conditional knockout of either HDAC1 or HDAC2 in mice by [inducible, pan-hematopoietic] Mx1-Cre does not significantly impair hematopoiesis or erythropoiesis. However, compound knockout of both HDAC1 and HDAC2 conditional mutants in mice leads to severe hematopoietic and erythroid deficiencies[89,90], suggesting complementary functions for HDAC1/HDAC2 in hematopoiesis/erythropoiesis as well as indicating possibly fewer side effects if either HDAC2 [or HDAC1] alone is targeted.

5.2. DNMT1

DNA methyltransferases [DNMTs] are a group of enzymes that catalyze the transfer of methyl groups from S-adenosylmethionine to cytosine[91]. The key maintenance DNA methyltransferase, DNMT1, has been shown to associate with both Bcl11a and TR2/4 complexes[82,83,84**] [Figure 2B]. Knockdown of DNMT1 significantly re-activated fetal γ-globin expression in human primary HSPC culture[82]. Conditional knockout of the Dnmt1 gene in adult mice indicated that DNMT1 plays a key role in hematopoietic stem cell homeostasis and hematopoietic lineage decisions[92,93] by repressing the expression of key hematopoietic transcription factors, including GATA1[92,94]. However, the in vivo function of DNMT1 in adult erythroid cell progenitors, which is associated with efficient fetal globin gene induction after DNMT1 inhibition, is still unknown.

5.3. LSD1

Lysine-specific histone demethylase 1A [LSD1, aka KDM1a] is a flavin-dependent monoamine oxidase that catalyzes the removal of mono- and di-methyl groups specifically on H3K4 and H3K9[95]. LSD1 is also recovered among both the TR2/4 and Bcl11a complexes[82,83,84**] [Fig2B]. Knockdown of LSD1 has been proven to be an efficient way of inducing fetal globin gene synthesis in human primary HSPC culture[82,96]. In vivo, LSD1 loss of function experiments [examining either doxycycline-induced expression of four different targeted shRNAs or by Mx1-Cre-mediated gene deletion] showed that it promotes terminal differentiation of multiple hematopoietic lineages including erythroid cells, likely by repressing progenitor gene expression, which must be considered as a possible undesirable on target side effect of fetal globin gene induction after LSD1 inhibition[97,98].

5.4. PRC2

Polycomb repressive complex 2 [PRC2] is responsible for mono, di- and tri-methylation of H3K27 to generate a transcriptional silencing chromatin signature[99] that has been associated with both GATA1 and Bcl11a repressive complexes[82,100] [Figure 2B]. PRC2 is a multiprotein complex containing several collaborative core subunits: the lysine methyltransferases EZH1 or EZH2 and non-catalytic subunits including the zinc finger protein SUZ12 and embryonic ectoderm development [EED]. Notably, shRNA knockdown of each of these subunits re-activates fetal globin expression in human primary HSPC culture[82], suggesting that the PRC2 complex plays a role in γ-globin repression. Within the PRC2 complex, the EZH1/2 and EED are the most promising targets for fetal globin induction by potential small-molecule chemical therapeutics. Conditional knockout of PRC2 activity by Mx1-Cre-mediated EED deletion in adult mice suggested that PRC2 plays a critical role in HSC homeostasis[101]. However, constitutive disruption of PRC2 function specifically in adult erythroid cells by EPOR-Cre-mediated EED conditional deletion only slightly impaired erythropoiesis [100].

5.5. Other possible γ-globin direct transcriptional effectors

Protein arginine N-methyltransferase 5 [PRMT5] symmetrically dimethylates H2AR3, H4R3, H3R2 and H3R8[102] [only H4R3 is shown in Figure2 for simplicity] to transcriptionally repress gene expression [Figure 2B]. PRMT5 is associated with the transcription factor LYAR that binds to the GGTTAT motif downstream of the γ-globin transcriptional start site[85,86]. The knockdown of PRMT5 or LYAR modestly re-activates fetal globin expression in human primary HSPC[85,86]. Conditional deletion of PRMT5 in adult mice suggested that PRMT5 might play an important role in hematopoietic cell maintenance[103].

Euchromatic histone-lysine N-methyltransferase 2 [EHMT2 or G9a]- and Euchromatic histone-lysine N-methyltransferase 1 [EHMT1 or G9a-like protein] are histone lysine methyltransferases that catalyze mono- and di-methylation of H3K9 to generate repressive chromatin signatures. EHMT1/2 are not reported to be associated with any of the HPFH mutation-related TFs but may be recruited to the globin LCR through NF-E2[104,105]. Knockdown of both EHMT1 and EHMT2 significantly re-activated fetal globin gene expression in human primary HSPC[106]. Notably, hematopoietic deletion of EHMT2 [by Vav-Cre] displays no significant hematopoietic abnormity[107].

BRCA1 associated protein-1 [BAP1] is a nuclear deubiquitinating enzyme which, as a catalytic subunit of the polycomb repressive deubiquitinase [PR-DUB] complex, deubiquitinates H2A at lysine-119[108]. BAP1 is associated with the TR2/TR4 complex and appears to stabilize the complex by deubiquitinating its scaffold protein nuclear receptor co-repressor 1 [NCoR1][84**]. Both shRNA mediated knockdown and CRISPR-Cas9-mediated heterozygous loss of BAP1 significantly re-activated fetal globin expression in HUDEP2 cells[84**]. Conditional deletion of the Bap1 gene in the mouse hematopoietic system recapitulates human myelodysplastic syndrome or a chronic myelomonocytic leukemia phenotype[109,110], consistent with the known tumor suppressor activity of BAP1. Competitive bone marrow transplantation suggested that Bap1 deletion modestly impairs murine erythropoiesis[109].

6. Small molecule inhibitors that induce HbF

6.1. Epigenetic enzymatic inhibitors

6.1.1. HDAC inhibitors [HDI]

The observation of delayed fetal to adult globin synthesis in infants of diabetic mothers[111] encouraged Perrine and coworkers to identify the increased plasma metabolites in infants resulting in the finding that butyric acid and butyrate stimulate fetal globin expression[112,113]. Based on the hypothesis that butyrate stimulates fetal globin through HDAC inhibition, investigators subsequently confirmed the HbF inducing capacity by a variety of HDAC inhibitors (HDIs) [88,114]. HDACs are a group of enzymes that have been extensively targeted for inhibition in psychiatry, neurology and cancer[115]. Multiple HDIs have been tested in clinical trials for the relief of symptoms in both SCD and thalassemia. A phase I/II trial of HQK-1001 in SCD[116] [NCT00842088] reported a 0.2g/dL mean absolute increase of HbF and 0.83g/dL mean increase of total hemoglobin. Other phase II trial studies in SCD reported that HQK-1001 induced a mean HbF increase of 2%[117] [0.8-3.2%, NCT01322269] or 0.9%[118] [0.1-1.6%, NCT01601340]. HDI clinical potency as potential HbF inducers has been largely limited by the toxicity associated with non-specific HDAC inhibition. The clinical trial results examining HDIs for the treatment of both SCD and thalassemia are listed in Table1. In general, the performance of HDIs in clinical trials for the treatment of SCD has been somewhat disappointing.

Table1.

Small HbF induction molecules targeting epigenetic enzymes involved in fetal globin repression. To keep the length of the table reasonable, only LSD1, DNMT and HDAC inhibitors under clinical trial in SCD or thalassemia were listed. For PRC2/EZH and PRMT5 inhibitors, as there are no clinical trials in SCD or thalassemia, clinical trials in other conditions were listed.

HbF inducer	Corepressor	Molecules	Conditions in Clinical Trials	Pre-clinical	Clinical
	LSD1	INCB059872	SCD	Knockdown and inhibitor induce HbF	NCT03132324 (Phase I, Terminated duo to a business decision)
	HDAC	Vorinostat	SCD	Knockdown and inhibitor induce HbF	NCT01000155 (Phase II, Terminated duo to low enrollment)
		LBH589/Panobinos tat	SCD		NCT01245179 (Phase I, Pending for recruiting)
		HQK-1001/dimethylbut yrate	SCD		NCT00842088 (Phase I/II, 0.2g/dL HbF mean increase and 0.83g/dL total hemoglobin increase), NCT01601340 (Phase II, 0.1-1.6% HbF induction), NCT01322269 (Phase II, 0.8-3.2% HbF induction)
			Beta-thalassemia		NCT01609595 (Phase II), NCT01642758 (Phase II), NCT00790127 (Phase II)
	DNMT	Decitabine	SCD	Knockdown and inhibitor induce HbF	NCT01685515 (Phase I, 4-9% HbF induction), NCT01375608 (Phase II, 1-13.5% HbF induction)
			Thalassemia Intermedia		NCT00661726 (Phase II)
		5-Azacytidine	Beta-thalassemia		NCT00005934 (Phase II)
	PRC2/EZH	Tazemetostat	Mesothelioma	Knockdown and inhibitor induce HbF	NCT02860286 (Phase II)
			B-Cell Lymphomas or Adv Solid Tumors		NCT03010982 (Phase I)
		CPI-1205	B-Cell Lymphoma		NCT02395601 (Phase I)
	PRMT5	JNJ-64619178	Neoplasms	Knockdown induce HbF	NCT03573310 (Phase I)
		GSK3326595	Neoplasms		NCT02783300 (Phase I), NCT03614728 (Phase I)
		PF-06939999	Advanced Solid Tumors		NCT03854227 (Phase I)
	EHMT1/2	UNC0638	N/A	Knockdown and inhibitor induce HbF	N/A
Others		Hydroxyurea	SCD	Compound treatment induce HbF	Approved by FDA for SCD treatment in 1998
		Pomalidomide	SCD	Compound treatment induce HbF	NCT01522547 (Phase I, No data reported)
		Metformin	SCD/Thalassemia	Compound treatment induce HbF	NCT02981329 (Phase I, Recruiting)

6.1.2. DNMT inhibitors

DeSimone and colleagues originally hypothesized that induction of hypomethylation by 5-azacytidine [5Aza] might re-activate fetal globin gene expression based on the fact that the fetal promoters became hypermethylated in adult RBCs. They found that in vivo treatment with 5-Azacytidine induced HbF in both anemic baboons, animals that, like humans, exhibit a fetal to adult switch in β-type globin synthesis, and β-thalassemia patients[119,120]. The DNMT inhibitor 5Aza was the first chemotherapeutic agent used for HbF re-activation, however, 5Aza was later abandoned due to its toxicity[52**]. Decitabine [DEC] is a more favorable DNMT inhibitor in terms of safety when compared to 5Aza. DEC inhibition of DNMT induces up to 20-30% fetal globin expression[96,106] in human HSPC culture, and also significantly induces fetal globin synthesis in baboons[121,122]. A recent phase I clinical study [NCT01685515] indicated that the combinatorial treatment of DEC plus a stabilizer, tetrahydrouridine [THU], was safe and well-tolerated by SCD patients and induced HbF percentages of up to 4-9% with a commensurate increase in the number of healthy RBCs[123]. Another phase II study in high-risk SCD patients [NCT01375608] indicated that DEC induced up to 1-13.5% HbF with a mean value of 4.8%. The clinical trials with DNMT inhibitors in thalassemia are listed in Table1.

6.1.3. LSD1 inhibitors

The concept of employing an LSD1 inhibitor as a potential fetal globin stimulator derived from the identification of LSD1 as a key co-repressor in the direct repeat erythroid-definitive [DRED] TR2/TR4 γ-globin repressor complex[83]. Pharmacologic inhibition of LSD1 by the small molecule chemical compound tranylcypromine [TCP] led to dose-dependent increases in γ-globin synthesis of up to 30% HbF in human HSPC erythroid differentiation cultures[96]. In vivo inhibition of LSD1 by a conceptually similar small chemical inhibitor [RN-1] in SCD mice induced HbF synthesis and led to dramatic improvement of many pathological features normally associated with SCD[124]. RN-1 is also able to significantly stimulate HbF synthesis in baboons[125,126]. A phase I clinical trial for HbF induction in SCD by the LSD1 inhibitor INCB059872 was registered; however, the trial was terminated early [NCT03132324].

6.1.4. Other potential inhibitors

EHMT1/2 inhibition by UNC0638 induces HbF up to 30% in human primary HSPC culture[105,106], but no clinical trial of the EHMT1/2 inhibitor has been registered to date. Both a PRC2 inhibitor and a PRMT5 inhibitor are under study in clinical trials for cancer [Table1]. However, there are no preclinical data examining either inhibitor as way of inducing HbF. Notably, besides the majority of PRC2 inhibitors that target the catalytic domain of EZH1/2, an EED inhibitor [A-395] targeting a structural interaction within PCR2 adds to the potential diversity of PRC2 inhibition[127], which may be useful to avoid side effects attributed to catalytic inhibitors. No BAP1 inhibitors have been reported to date.

The rapid induction of new erythroid cells induced following acute anemic stress [such as bleeding or treatment with cytotoxic agents] displays different features when compared to normal steady-state erythropoiesis, and is referred to as stress erythropoiesis[128]. Human stress erythropoiesis shows some features of fetal erythroid development, such as enhanced expression of HbF[129], that are due to the fetal-skewed properties of stress erythroid progenitors. Stress erythropoiesis may be a common mechanism for fetal globin re-activation by a variety of cytotoxic HbF-inducing agents[130]. Notably, all of the epigenetic inhibitors that stimulate HbF are cytotoxic agents and may partially re-activate HbF through mechanisms that indirectly induce stress erythropoiesis.

6.2. Other small molecules that induce the production of fetal hemoglobin

Hydroxyurea was the first HbF inducing drug approved by the FDA. The mechanism of hydroxyurea-mediated response is complicated and still not fully understood: intracellular signaling pathways, epigenetic/transcriptional regulation of the γ-globin gene, post-translational regulation of key fetal globin gene repressors and induction of stress erythropoiesis may all be involved[130-132]. Other promising HbF-inducing small molecules in recent clinical trials include pomalidomide and metformin [Table1]. Pomalidomide re-activates HbF up to 30% in human primary HSPC cultures and induces HbF in myeloma patients[133]. Decreased expression of several HbF repressors [including Bcl11a] and stress erythropoiesis are associated with HbF induction mediated by pomalidomide[133]. A clinical trial of Pomalidomide in SCD has been completed but no data have been reported to date [Table1]. Metformin is an FDA approved drug known to stimulate transcription factor FOXO3 activity, a positive regulator of fetal globin gene synthesis[134]. Metformin re-activates HbF up to 30% in HSPC culture without affecting Bcl11a, MYB or KLF1 levels[134]. A clinical study of Metformin used for SCD treatment is being actively recruiting participants [Table1].

7. Small molecules targeting TFs

All of the efficient HbF-inducing small molecules that have been identified to date exhibit varying degrees of cytotoxicity because their target proteins [e.g. epigenetic co-repressor enzymes] often play many additional key roles in cell survival and differentiation in a large number, if not all, tissues. To avoid this issue, more specific inhibition of the interactions between TFs and certain co-repressors or their DNA interaction motifs, exemplified by EED inhibitor A-395[127], may prove to be promising as more refined compounds are developed[135]. Another way of targeting TFs might be through the application of PROTACs [Proteolysis Targeting Chimeras][136], bi-functional molecules that target one domain to proteins of interest [for example Bcl11a, ZBTB7A or TR2/4] fused to a second domain that can bind/recruit E3 ubiquitin ligase to shuttle the complexes for ectopic degradation to the proteasome, thereby re-activating HbF.

Conclusion

High levels of HbF significantly reduce the pathology and symptoms associated with disease in the β-globinopathies. In this review, we have summarized the known transcription factor occupants capable of recruiting corepressors responsible for silencing of the g-globin genes. The identification of chemical compounds capable of blocking the activity of these enzymatic corepressors has implications for treating β-globinopathies, with clinical trial data for selected compounds with therapeutic applications for SCD and b-thalassemia summarized herein. To date, only a limited number of druggable corepressors have been identified, necessitating a better understanding of the mechanisms governing g-globin repression in adult erythroid progenitors in the hopes of increasing the number of potential candidate target enzymes for future development of small molecule therapeutics. In contrast, structural optimization and improvement upon current small molecules should lead to the synthesis of novel compounds with potentially higher HbF induction efficiency and with lower toxicity to the benefit of millions of patients.

Key points.

1. Increased levels of fetal hemoglobin promotes red blood cell lifespan in both sickle cell disease and b- thalassemia major patients.

2. DNA binding of transcription factor repressors ZBTB7A, BCL11A, and TR2/TR4 to the γ-globin promoters is impaired by HPFH mutations.

3. Multiple epigenetic corepressors associated with these repressors can be inactivated to induce fetal globin gene expression.

4. Inhibition of the epigenetic corepressors by small molecules is a promising therapeutic option to treat the β-globinopathies.