Potential role of LSD1 inhibitors in the treatment of sickle cell disease: a review of preclinical animal model data

Abstract

Sickle cell disease (SCD) is caused by a mutation of the β-globin gene (Ingram VM. Nature 180: 326–328, 1957), which triggers the polymerization of deoxygenated sickle hemoglobin (HbS). Approximately 100,000 SCD patients in the United States and millions worldwide (Piel FB, et al. PLoS Med 10: e1001484, 2013) suffer from chronic hemolytic anemia, painful crises, multisystem organ damage, and reduced life expectancy (Rees DC, et al. Lancet 376: 2018–2031, 2010; Serjeant GR. Cold Spring Harb Perspect Med 3: a011783, 2013). Hematopoietic stem cell transplantation can be curative, but the majority of patients do not have a suitable donor (Talano JA, Cairo MS. Eur J Haematol 94: 391–399, 2015). Advanced gene-editing technologies also offer the possibility of a cure (Goodman MA, Malik P. Ther Adv Hematol 7: 302–315, 2016; Lettre G, Bauer DE. Lancet 387: 2554–2564, 2016), but the likelihood that these strategies can be mobilized to treat the large numbers of patients residing in developing countries is remote. A pharmacological treatment to increase fetal hemoglobin (HbF) as a therapy for SCD has been a long-sought goal, because increased levels of HbF (α₂γ₂) inhibit the polymerization of HbS (Poillin WN, et al. Proc Natl Acad Sci USA 90: 5039–5043, 1993; Sunshine HR, et al. J Mol Biol 133: 435–467, 1979) and are associated with reduced symptoms and increased lifespan of SCD patients (Platt OS, et al. N Engl J Med 330: 1639–1644, 1994; Platt OS, et al. N Engl J Med 325: 11–16, 1991). Only two drugs, hydroxyurea and l-glutamine, are approved by the US Food and Drug Administration for treatment of SCD. Hydroxyurea is ineffective at HbF induction in ~50% of patients (Charache S, et al. N Engl J Med 332: 1317–1322, 1995). While polymerization of HbS has been traditionally considered the driving force in the hemolysis of SCD, the excessive reactive oxygen species generated from red blood cells, with further amplification by intravascular hemolysis, also are a major contributor to SCD pathology. This review highlights a new class of drugs, lysine-specific demethylase (LSD1) inhibitors, that induce HbF and reduce reactive oxygen species.

BACKGROUND

Sickle cell disease (SCD) is a qualitative hemoglobinopathy. In the United States, 1 of every 400 African Americans is born with SCD. Approximately 100,000 Americans have SCD (18). The causative mutation is an A-T transversion in the sixth codon of the β-globin gene, leading to the substitution of glutamic acid for valine, which results in the formation of the abnormal sickle hemoglobin (HbS) (22, 51). After deoxygenation in red blood cells (RBCs), HbS forms polymers, causing the RBCs to become deformed (sickled) and adherent, leading to vasoocclusive events.

Initially, the polymerization of HbS was considered to be the only driving force of hemolysis in SCD. However, over decades, the importance of the accumulation of reactive oxygen species (ROS) inside the RBC and in the vasculature has been established (83). The combination of vasoocclusion and ROS accumulation is involved in chronic organ damage and vasculopathies (49). Increased intracellular ROS accumulation has been linked to the exposure of phosphatidylserine (PS) and to a reduction in flippase activity, which activates the interaction of RBCs with other cells (42). Previous studies provide insights into the relative contributions of RBC ROS-induced membrane damage and biophysical alterations, as shown by a decrease in the deformability of SCD RBCs with the progression of disease severity (4). The presence of oxidized HbS and deoxygenated polymerization of HbS in the RBC create a cycle of ROS, membrane damage, hemolysis, and cellular adhesion. Membrane damage to SCD RBCs and reticulocytes by ROS causes vascular endothelial damage through abnormal cellular adhesion and the release of intravascular hemolytic products (27). SCD reticulocytes have increased adherence to endothelial cells, while ROS generated by hemolysis activate monocytes, polymorphonuclear neutrophils, and platelets in the vascular lumen, resulting in inflammation, which damages vascular endothelial cells. Damage to vascular endothelial cell leads to vascular narrowing, causing ischemia, reperfusion, infarction, and, ultimately, multiple organ failures (2, 14, 23, 62). While bone marrow transplantation can be a curative therapy for SCD, the vast majority of African Americans do not have suitable family donors. In addition, bone marrow transplantation must be performed in specialized facilities and, therefore, is not a practical option for most patients worldwide. Therefore, current treatment modalities at a majority of institutions focus on the administration of disease-modulating drugs, control of infections, and pain management (23).

Only two drugs, hydroxyurea (HU) and l-glutamine, have been approved by the US Food and Drug Administration for treatment of SCD. HU, an inhibitor of ribonucleotide reductase, has multiple effects, including induction of fetal hemoglobin (HbF), nitric oxide, and glutathione peroxidase 1 activity and alteration of RBC-endothelial cell interactions. HbF (α₂γ₂) inhibits the polymerization of HbS, and elevated levels of HbF are associated with less severe illness and longer survival (56, 78). Therefore, a pharmacological therapy for induction of HbF has long been sought for SCD therapy. It is only effective in ~50% of patients, and the distribution of HbF is not pancellular (6, 32, 76). HbF is mainly distributed in a fraction of the RBCs, F cells, which contain HbF. Other drugs that are being developed for HbF induction include inhibitors of epigenetic-modifying enzymes that exist as components of corepressor complexes involved in repression of the γ-globin gene. The targets of the pharmacological inhibitors include histone deacetylases (HDACs), DNA methyltransferase 1 (DNMT1), protein arginine N-methyltransferase 5 (PRMT5), euchromatic histone lysine methyltransferase 2 (EHMT2/G9a), and lysine-specific demethylase 1 [LSD1 (KDMA1)].

While the mechanism of l-glutamate is not completely understood, it is postulated to affect RBC ROS levels that may damage the SCD RBCs and vessels. Oxidative stress occurs when an increase in oxidants without a similar increase in antioxidants triggers a cascade of oxidative reactions that, ultimately, lead to cell death due to damage to lipids, proteins, and DNA. Several mechanisms are responsible for increased oxidative stress in SCD: 1) high levels of intracellular RBC ROS (25, 45), 2) restoration of oxygen-rich blood after ischemic injuries that generate superoxides (3, 46), and 3) the NADPH-mediated oxidative burst produced by activated neutrophils. Excessive ROS in the bloodstream leads to multiple pathophysiological outcomes, including endothelial damage (17, 26), accelerated hemolysis, hypercoagulability (73), and SCD-related vasoocclusion (50).

Multiple mechanisms are responsible for RBC intracellular generation of ROS. NADPH oxidases have been identified as a major source of ROS in SCD RBCs. The other known sources of ROS in RBCs in SCD are 1) HbS autoxidation, 2) the Fenton reaction, and 3) low levels of antioxidants, such as selenium and glutathione peroxidase 1, SOD, and catalase (9, 47). In addition, Jagadeeswaran et al. (24) recently discovered that abnormally retained mitochondria in the SCD RBC are a source of ROS. An increased proportion of RBCs that retain mitochondria was observed in patients with SCD and in a transgenic mouse model of SCD (24). ROS levels were higher in RBCs with retained mitochondria than RBCs without mitochondria. During normal terminal differentiation of RBCs, mitochondria are eliminated through a process known as “mitophagy.” Mouse models with deletions in mitophagy genes exhibited reduced survival of RBCs and severe anemia due to lack of mitochondrial clearance (12, 44, 67). Mitochondria generate ROS, primarily superoxide (O₂^·−), by the respiratory chain; O₂^·− is transformed into H₂O₂, and, then, in the presence of ferrous ions, the more damaging hydroxyl radical (OH^·) is formed (40).

Recently, targeting of intracellular ROS has been suggested as a possible therapeutic strategy for SCD. Keleku-Lukwete et al. (28) crossed a mouse model of SCD with a mouse that overexpressed the transcription factor nuclear factor erythroid 2-related factor 2 (Nrf2), a master regulator of the antioxidant cell-defense system, and showed amelioration of tissue damage. Previous studies suggest that Nrf2 could reduce the ROS levels not only by transcriptionally upregulating antioxidant enzymes and NADPH-generating pentose pathway enzymes, but also by lowering the levels of NADPH oxidase (Nox2) (30). Recent studies also showed that administration of dimethyl fumarate, a drug approved for the treatment of multiple sclerosis, to SCD mice decreased hepatic necrosis, inflammatory cytokines, and irregularly shaped RBCs and increased HbF (5).

This review focuses on LSD1 inhibitors in two preclinical models: 1) the SCD mouse model and 2) the baboon model. Testing in these models has shown that LSD1 (KDM1A) inhibitors have two key mechanisms: 1) induction of high levels of HbF in nonhuman primates and 2) reduction of the abnormal presence of mitochondria in the mature RBC of SCD mice (24b). Both the SCD mouse model and the baboon model are necessary, because each has limitations. The SCD mouse model demonstrates SCD pathology, but the human γ-globin gene is regulated as an embryonic, rather than the fetal-stage, gene in the mouse (69) and, therefore, is not efficiently induced by pharmacological agents. In contrast, nonhuman primates such as the baboon are considered to be excellent animal models to test the activity of HbF-inducing drugs, because the structure and developmental regulation of the β-like globin gene locus are highly conserved between baboon and human, and effects in the baboon are highly predictive of results in patients (1, 43). The baboon model, however, does not have HbS and, therefore, does not exhibit SCD pathology.

LSD1 (KDMA1) Plays a Crucial Role in Repression of γ-Globin

Extensive efforts to elucidate the mechanism of γ-globin repression during adult erythropoiesis have identified four trans-acting factors, the orphan nuclear receptors TR2/TR4, BCL11A, PRMT5, and LRF (ZBTB7A), that repress the γ-globin gene (35, 39, 41, 59, 68, 81, 82). TR2/TR4, the first repressor characterized, recruits a multiprotein corepressor complex, DRED, to the direct repeat (DR) elements within the γ-globin promoter (10). Biochemical purification of DRED showed that it consisted of a tetrameric core element containing TR2/TR4, DNMT1, and LSD1 with additional proteins, including HDACs, the G9A histone methyltransferase, and CoREST associated with the core to form the larger DRED complex (10). BCL11A, initially identified by genome-wide associated studies (35, 41, 68), was confirmed to be a γ-globin repressor by transgenic knockout mouse experiments that showed that deletion of BCL11A prevented repression of mouse embryonic globin and the human γ-globin gene during development (68). BCL11A also binds to the γ-globin promoter (8, 37, 38) and recruits a multiprotein corepressor complex containing HDACs, DNMT1, and LSD1 to repress γ-globin expression (85). Recruitment of corepressor complexes also facilitates γ-globin repression by ZBTB7A and PRMT5 (39, 59).

Thus recruitment of multiprotein corepressor complexes containing enzymes that modify the epigenome is essential to the mechanism of γ-globin repression. The temporal order of enzymatic action to establish repressive epigenetic modifications of the γ-globin gene during erythroid differentiation and the possible codependence of their activities is unknown. These enzymes maintain the high levels of DNA methylation, low levels of histone H3K4 methylation and histone acetylation, and high levels of histone H3K9 methylation that are characteristic of repressed genes and are targets for therapeutic interventions to increase HbF (15, 32, 79). The use of HDAC inhibitors is limited by their toxicity, while the pharmacological G9A inhibitors (7, 31, 61) developed thus far have poor bioavailability. Our work shows that inhibitors of DNMT and LSD1 are potent activators of HbF and among the most promising agents in development to increase HbF levels for therapy of SCD (19, 43, 75).

Reexpression of γ-Globin by LSD1 (KDM1A) Inhibitors

LSD1 (KDM1A) demethylates mono- and dimethylated histone H3K4 residues. Deletion of LSD1 in mice is lethal (84). The functional role of LSD1 in hematopoiesis has been investigated using tetracycline-inducible knockdown (74) and targeted deletion strategies (29). Knockdown of LSD1 expanded progenitor cells and compromised terminal hematopoietic differentiation, leading to granulocytopenia, anemia, and thrombocytopenia. In contrast, LSD1 knockdown increased the numbers of monocytes. Increased expression of key hematopoietic genes, including Gfi1b, HoxA9, and Meis1, was observed. Targeted deletion experiments also showed that LSD1 was required for the terminal differentiation of multiple blood cell lineages and impaired differentiation of stem cells. LSD1 was shown to act at both promoters and enhancers of various genes expressed in stem and progenitor cells to silence their expression during hematopoiesis (29).

Shi et al. (71) initially identified LSD1 as a therapeutic target for reactivation of HbF through experiments using RNA interference strategies and pharmacological inhibitors in cultured human erythroid progenitor cells. LSD1 knockdown in cultured human erythroid cells enhanced γ-globin mRNA synthesis in cultured human erythroid progenitor cells. Administration of tranylcypromine (TCP) increased HbF from 4.6% to 31% of total hemoglobin and γ-globin mRNA expression 9.4-fold at the highest dose tested. Induction of HbF with TCP was superior to HU and comparable to the DNMT1 inhibitor decitabine (DAC). In addition, TCP increased γ-globin expression in β-globin yeast artificial chromosome (YAC) transgenic mice (72). A timeline of the key progress in the development of LSD1 inhibitor for SCD is shown in Fig. 1A.

Fig. 1.

A: timeline of selected key events in lysine-specific demethylase (LSD1) research and development in therapeutically targeting LSD1 inhibitors in sickle cell disease (SCD). B: the LSD1 inhibitor RN-1 has 2 distinct mechanisms. One addresses sickle hemoglobin (HbS) polymerization-mediated sickling, and the other addresses red blood cell (RBC) reactive oxygen species (ROS) generation-induced hemolysis. HbF, fetal hemoglobin; Retics, reticulocytes.

Following this report, Rivers et al. (65) demonstrated that 2-(1R,2S)-2-{[4-(benzyloxy)phenyl]cyclopropylamino}-1-(4-methylpiperazin-1-yl)ethanone, HCl (RN-1), a LSD1 inhibitor with increased potency and specificity (Table 1), induced F cells and γ-globin mRNA in humanized SCD mice to levels similar to DAC, the most powerful HbF-inducing drug known, and to higher levels than either TCP or HU. F cells on day 11 were increased in mice treated with DAC (11.8 ± 2.6, P < 0.005), HU (6.2 ± 10.88, P < 0.03), and RN-1 at 2.5 mg/kg (7.76 ± 1.13, P < 0.005) and 5 mg/kg (12.5 ± 1.85, P < 0.005) compared with control mice (4.8 ± 0.6). No increase in F cell levels was observed in mice treated with TCP (4.9 ± 0.95). Increased levels of γ-globin mRNA (γ/γ + β-fold change) were observed on day 11 in mice treated with DAC (4.24 ± 1.4, P < 0.001), HU (1.8 ± 0.6, P < 0.02), and RN-1 at 2.5 mg/kg (1.78 ± 0.98, P < 0.001) and 5 mg/kg (4.34 ± 1.36, P < 0.005) compared with controls. No increase in γ-globin mRNA was observed in mice treated with TCP. On day 11, levels of F cells (P < 0.001) and γ-globin mRNA (P < 0.02) were significantly higher in mice treated with 5 mg/kg RN-1 than in those treated with HU.

Table 1.

IC₅₀ of TCP and RN-1

	IC50, µM
Compound	LSD1	MAO-A	MAO-B
TCP	2–100	0.48	4.881
RN-1	0.01–0.07	0.51	2.785

MAO, monoamine oxidase; RN-1, 2-(1R,2S)-2-{[4-(benzyloxy)phenyl]cyclopropylamino}-1-(4-methylpiperazin-1-yl)ethanone, HCl; TCP, tranylcypromine.

In an additional study in humanized SCD mice, Cui et al. (11) demonstrated similar HbF-inducing effects of RN-1 and also observed that the liver and spleen of SCD mice treated with RN-1 did not exhibit the necrotic lesions that are usually associated with SCD. They also reported reduced reticulocytosis and increased total hemoglobin and RBC lifespan in the SCD mice treated with RN-1. RN-1 (3 or 10 mg·kg body wt⁻¹·day⁻¹) was administered to SCD mice for 5 days/wk for 4 consecutive weeks. In untreated SCD mice, HbF was 0.32% of total hemoglobin, while treatment with 10 mg/kg RN-1 increased HbF to 1.2% of total hemoglobin. The increased RBC life span and decreased organ pathology were attributed to induction of mouse embryonic globin genes.

Efficacy and Safety of LSD1 Inhibitor RN-1 in Primates

While treatment of humanized SCD mice with RN-1 increased γ-globin mRNA, F cells, and F reticulocytes (11, 65), the levels of HbF were low, because the human γ-globin gene is not efficiently reactivated in this mouse model. Therefore, we tested the effect of RN-1 in baboons, considered to be the best animal model to study the effect of HbF-inducing drugs due to the conservation of the structure and pattern of developmental expression of the β-like globin genes among simian primates. Experiments performed in baboons have been directly translated to clinical trials in SCD patients (1, 43). RN-1 treatment of anemic baboons stimulated high levels of γ-globin synthesis, HbF, F cells, and F reticulocytes (63). In addition, RN-1 treatment restored high levels of HbF and caused synthesis of the individual 5′-Iγ- and 3′-Vγ-globin chains in the ratio characteristic of fetal development. Chromatin immunoprecipitation experiments showed that RN-1 treatment increased levels of histone H3K4me2, H3K4me3, and H3K9 acetylation (H3K9ac) associated with the γ-globin promoter region.

These experiments were followed by long-term (>265 days) administration of RN-1 to two normal baboons to evaluate relative safety and effectiveness (20). HbF and F cells were maintained at high levels with minimal hematological toxicity. Decreases in hemoglobin were observed during menstrual bleeding in one animal and following an accidental laceration of the perineal swelling. In vitro platelet activation assays showed that platelet function was inhibited in the treated baboons. Importantly, inhibitors of platelet function are in clinical trials in patients with SCD (48, 77). The level of inhibition of platelet function targeted by these studies appears similar to that in baboons treated with RN-1.

The LSD1 Inhibitor RN-1 Reduces Mitochondria-Containing RBCs and ROS in a SCD Mouse Model

Treatment with the LSD1 inhibitor RN-1 decreased the level of mitochondria-retaining RBCs, reduced ROS, and enhanced RBC lifespan in SCD mice. Gene expression analysis of SCD mice treated with RN-1 showed a significant (>2-fold) upregulation in the expression of key mitophagy genes, including Ulk-1, also known as ATG1, and the erythroid-specific mitophagy gene ATG7, compared with untreated SCD mice (24). ATG1 and ATG7 are key effector molecules in mitochondrial autophagy in mammalian hematopoietic cells, and their loss in erythroid cells leads to the incomplete removal of mitochondria and severe anemia in vivo (44). RN-1 reduces levels of RBCs with abnormally retaining mitochondria, thereby reducing ROS. This effect leads to an increased RBC lifespan.

LSD1 Inhibitors in Combination with Other Drugs

While relatively long-term administration of RN-1 to baboons was accomplished with minimal hematological toxicity, the therapeutic window does appear rather narrow, with increased doses producing neutropenia and thrombocytopenia due to the requirement for LSD1 activity during terminal hematopoietic differentiation. Induction of HbF by other drugs, such as DAC, can also be associated with dose-related hematological toxicities. Therefore, the best strategy may be development of combinatorial drug regimens targeting different epigenetic-modifying enzymes present in the corepressor complexes that are recruited to the γ-globin gene. This strategy will increase HbF expression in a combinatorial or synergistic manner while minimizing hematological toxicities and, thus, maximizing the therapeutic index. Although the use of combinatorial drug regimens has been proposed to increase HbF, few combinations have been tested in vivo (13, 52). We have shown that HU and RN-1 administered together induced a combinatorial increase in γ-globin expression in SCD mice (24a). In the same study, administration of RN-1 and HU in combination reduced the peripheral red blood sickle cells and spleen size in a SCD mouse model.

While other studies showed that the combination of HU and erythropoietin (Epo) administered to SCD patients was no more effective than HU alone (36), our laboratory showed that the combination of stem cell factor (SCF), Epo, and HU led to combinatorial increases in HbF synthesis in baboons (33). In baboons, administration of SCF, Epo, and HU in combination increased γ-globin chain synthesis to high levels (0.4–0.5 γ/γ + β-fold change), suggesting that the effect may be mediated by a CD117⁺ cell. We have observed that RN-1 increases the number of CD117⁺ bone marrow cells. This suggests that perturbation of erythroid differentiation by RN-1 may increase the number of HU-responsive cells. Experiments in cultured human erythroid progenitor cells showed that the KDM1A inhibitor TCP and DEC, in combination, increased HbF expression in a synergistic manner (71). Experiments in SCD mice also showed that RN-1 and DEC administered together increased HbF expression in an additive manner, although cytotoxicity was increased (24a). These results suggest that drug combinations should be further explored.

Future LSD1 Inhibitors

Recently, we showed that oral administration of an improved LSD1 inhibitor, OG-S1335, developed by Oryzon Genomics, with increased potency and specificity compared with RN-1, increased F reticulocytes and γ-globin mRNA in normal and anemic baboons (64). Another orally available LSD1 inhibitor, developed by Incyte, has entered clinical trials (ClinicalTrials.gov registry no. NCT03132324) in patients with SCD. The development of new LSD1 inhibitors with increased specificity and limited access to the central nervous system, as well as the development of targeted delivery systems to direct delivery of the drug to erythroid progenitor cells to limit adverse effects on other hematopoietic lineages, would be highly desirable.

DISCUSSION

The LSD1 inhibitor RN-1 may be an effective therapeutic agent for SCD that targets SCD pathology through two mechanisms of action. 1) RN-1 is a potent inducer of HbF in the baboon model. The effect of RN-1 on HbF levels appears to be similar to that of DAC, the most potent inducer of HbF thus far identified. The drug was safely administered for >265 days in two normal baboons at a dose that increased and maintained elevated HbF levels. 2) In a SCD mouse model, RN-1 increased RBC survival by reducing the number of RBCs with abnormal mitochondria retention, leading to reduced ROS. Further studies on mitochondrial function in SCD RBCs and related complications will provide insight into specific complications of SCD that could be ameliorated by LSD1 inhibitors. In conclusion, our studies have shown that LSD1 inhibitors may be highly effective therapeutic agents that act through a dual mechanism to attack SCD pathology. Further development of these drugs for SCD therapy should be encouraged.

Perspectives and Significance

The study of LSD1 inhibitors in SCD mice led to our discovery of abnormal mitochondrial retention in SCD. We conclude that LSD1 inhibitors increase RBC survival secondary to reduced mitochondrial ROS. An additional novel mechanism could be reduction of increased oxygen consumption by the retained mitochondria. Mitochondria are known to consume oxygen to produce ATP. Their retention in RBCs could lead to a hypoxic intracellular environment that causes HbS to deoxygenate and, subsequently, polymerize.

We have also observed that RN-1 increases expression of mitophagy regulation genes, suggesting that expression of mitophagy genes may be suppressed in SCD. These results should be addressed by future studies to compare mitophagy gene expression during normal and SCD erythropoiesis. Further studies of LSD1 effects could shift the current therapy paradigm and clinical practice to include the use of promitophagy drugs in SCD.

GRANTS

This work was funded by National Heart, Lung, and Blood Institute Grants R03 HL-135453, K01 HL-103172, and U01 HL-117658.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.R., R.J., and D.L. drafted the manuscript; R.J prepared figure; A.R., R.J., and D.L. edited manuscript; A.R., R.J., and D.L. approved a final version of the manuscript.