Search

In this work we provide preclinical data to support initiation of a first-in-human trial for sickle cell disease (SCD) using an approach that relies on reversal of the developmental fetal-to-adult hemoglobin switch. Erythroid-specific knockdown of BCL11A via a lentiviral-encoded microRNA-adapted short hairpin RNA (shRNAmiR) leads to reactivation of the gamma-globin gene while simultaneously reducing expression of the pathogenic adult sickle β-globin. We generated a refined lentiviral vector (LVV) BCH-BB694 that was developed to overcome poor vector titers observed in the manufacturing scale-up of the original research-grade LVV. Healthy or sickle cell donor CD34+ cells transduced with Good Manufacturing Practices (GMP)-grade BCH-BB694 LVV achieved high vector copy numbers (VCNs) >5 and gene marking of >80%, resulting in a 3- to 5-fold induction of fetal hemoglobin (HbF) compared with mock-transduced cells without affecting growth, differentiation, and engraftment of gene-modified cells in vitro or in vivo. In vitro immortalization assays, which are designed to measure vector-mediated genotoxicity, showed no increased immortalization compared with mock-transduced cells. Together these data demonstrate that BCH-BB694 LVV is non-toxic and efficacious in preclinical studies, and can be generated at a clinically relevant scale in a GMP setting at high titer to support clinical testing for the treatment of SCD.

RNA interference (RNAi) technology using short hairpin RNAs (shRNAs) expressed via RNA polymerase (pol) III promoters has been widely exploited to modulate gene expression in a variety of mammalian cell types. For certain applications, such as lineage-specific knockdown, embedding targeting sequences into pol II-driven microRNA (miRNA) architecture is required. Here, using the potential therapeutic target BCL11A, we demonstrate that pol III-driven shRNAs lead to significantly increased knockdown but also increased cytotoxcity in comparison to pol II-driven miRNA adapted shRNAs (shRNA(miR)) in multiple hematopoietic cell lines. We show that the two expression systems yield mature guide strand sequences that differ by a 4 bp shift. This results in alternate seed sequences and consequently influences the efficacy of target gene knockdown. Incorporating a corresponding 4 bp shift into the guide strand of shRNA(miR)s resulted in improved knockdown efficiency of BCL11A. This was associated with a significant de-repression of the hemoglobin target of BCL11A, human γ-globin or the murine homolog Hbb-y. Our results suggest the requirement for optimization of shRNA sequences upon incorporation into a miRNA backbone. These findings have important implications in future design of shRNA(miR)s for RNAi-based therapy in hemoglobinopathies and other diseases requiring lineage-specific expression of gene silencing sequences.

Reducing expression of the fetal hemoglobin (HbF) repressor BCL11A leads to a simultaneous increase in γ-globin expression and reduction in β-globin expression. Thus, there is interest in targeting BCL11A as a treatment for β-hemoglobinopathies, including sickle cell disease (SCD) and β-thalassemia. Here, we found that using optimized shRNAs embedded within an miRNA (shRNAmiR) architecture to achieve ubiquitous knockdown of BCL11A profoundly impaired long-term engraftment of both human and mouse hematopoietic stem cells (HSCs) despite a reduction in nonspecific cellular toxicities. BCL11A knockdown was associated with a substantial increase in S/G2-phase human HSCs after engraftment into immunodeficient (NSG) mice, a phenotype that is associated with HSC exhaustion. Lineage-specific, shRNAmiR-mediated suppression of BCL11A in erythroid cells led to stable long-term engraftment of gene-modified cells. Transduced primary normal or SCD human HSCs expressing the lineage-specific BCL11A shRNAmiR gave rise to erythroid cells with up to 90% reduction of BCL11A protein. These erythrocytes demonstrated 60%-70% γ-chain expression (vs. < 10% for negative control) and a corresponding increase in HbF. Transplantation of gene-modified murine HSCs from Berkeley sickle cell mice led to a substantial improvement of sickle-associated hemolytic anemia and reticulocytosis, key pathophysiological biomarkers of SCD. These data form the basis for a clinical trial application for treating sickle cell disease.

The fetal hemoglobin repressor BCL11A represents a therapeutic target for β-hemoglobinopathies as reduced expression of BCL11A leads to simultaneous increased γ-globin and reduced β-globin expression. Reversing this hemoglobin switch is particularly relevant in sickle cell disease to reduce the βS concentration and increase expression of the protective fetal hemoglobin (HbF, α2γ2). Here we show that despite use of optimized shRNAs embedded into a miRNA (shRNAmiR) architecture to reduce non-specific cellular toxicities (Guda et al. MT, 2015), the knockdown of BCL11A profoundly and specifically impaired long-term engraftment of both human and mouse hematopoietic stem cells (HSCs). In competitive transplantation assays cells transduced with shRNAmiRs targeting the BCL11A mRNA were consistently underrepresented after hematopoietic reconstitution with gene modified HSCs. Although this effect was particularly pronounced in the B-cell compartment, it was also the case for all other assessed hematopoietic lineages. Mechanistically, while knock-down of BCL11A did not lead to a detectable phenotype in terms of apoptosis, growth or differentiation in human or mouse HSCs in vitro, a significant increase in S/G2-phase human HSCs after engraftment into NSG mice was found, a phenotype associated with stem cell exhaustion. To avoid this BCL11A-specific HSC toxicity, we suppressed BCL11A in erythroid cells in a lineage-specific fashion by using transcriptional regulatory elements derived from the β-globin locus. Utilizing this approach for the expression of the optimized shRNAmiRs led to stable long-term engraftment of mouse and human gene modified cells in congenic or NSG-mice, respectively. Transduced primary normal or sickle cell disease (SCD) human HSCs gave rise to erythroid cells with up to 90% reduction of BCL11A protein. These erythrocytes demonstrated 60-70% γ-chain expression and a corresponding increase in HbF at low vector copy numbers per cell (VCN

Repression of the BCL11A protein could represent a therapeutic target for beta-hemoglobinopathies, as its knock-down has been shown to induce the expression of the fetal HBG (y-globin) gene ultimately leading to increased levels of the fetal hemoglobin tetramer (HbF, a2y2). In mice, Bcl11a is a key repressor of the murine HBG homolog Hbb-y. RNA interference (RNAi) technology using short hairpin RNAs (shRNAs) expressed via pol III promoters has been used to modulate gene expression in a variety of mammalian cell types. However, we found a negative impact of Bcl11a knockdown on hematopoietic stem cells (HSCs), limiting the repopulation efficiency and long-term engraftment after genetic modification, which is a major impediment for its translation into human therapeutic applications. To achieve lineage-specific targeting of mRNAs in an attempt to reduce HSC toxicity, expression of shRNAs via pol II promoters is required, necessitating embedding the shRNA in mammalian microRNA (shRNAmir) sequences for expression and processing. To achieve optimal knockdown of the Bcl11a transcription factor in erythroid progenitor and precursor cells, we first compared the efficiency of mRNA modulation via pol III (U6-promoter) vs pol II (SFFV-promoter) based lentiviral vectors. We demonstrate a 100-1000 fold lower Hbb-y induction using shRNAmir vs pol III mediated shRNA vector backbones due to reduced Bcl11a knockdown efficiency. In order to understand the molecular basis for these differences, small RNA sequence analysis was performed on murine erythroleukemia cells (MEL) cells transduced by multiple shRNA–shRNAmir pairs. We show that shRNAs expressed via a U6 promoter yield guide strand sequences which differ by a 2-4 nt shift compared to pol II driven (shRNAmir) mature guide strand sequences. RNA sequencing demonstrated that the stretch of uridines making up part of the pol III termination signal is transcribed and included at the 3’ end of the shRNA. This results in the generation of mature guide strand sequences with an alternative seed sequence compared to the predicted sequence and compared to miRNA embedded shRNAs. The difference in the seed sequences between the two expression systems strongly influences the efficacy of target gene knockdown, leading to reduced knockdown in pol II based vectors. We engineered a 4bp shift into guide strands of shRNAmirs that resulted in a faithfully processed shRNA sequence (a mature guide strand sequence identical to U6-driven sh-RNAs) and improved knock-down efficiency of Bcl11a at the protein level in most cases. The improved knockdown of Bcl11a was associated with a 100-300-fold enhancement of Hbb-y induction in MEL cells. Based on these results, we propose a modified strategy for the prospective design of shRNAmirs derived from shRNA screens in pol III vector backbones to achieve lineage-specific regulation of target genes. Targeted expression of shRNAmiRs to the erythroid compartment driven by a b-globin promoter/LCR element circumvented the detrimental effect on HSC engraftment, while still mediating efficient BCL11A knockdown, leading to high y-globin induction and formation of substantial amounts of fetal hemoglobin in human CD34-derived erythroid cells in vitro. Disclosures No relevant conflicts of interest to declare.

Abstract 753 Sickle cell disease (SCD) is caused by a mutation in the β-globin protein, leading to the polymerization of hemoglobin in deoxygenated conditions. The transcription factor BCL11A is a key regulator of developmental silencing of human fetal (γ-) globin, and also critical to repressing γ-globin in adult erythroid cells. A Bcl11a null mouse model carrying a transgenic YAC with a humanized β-globin locus (β-YAC) displays increased levels of fetal hemoglobin (HbF) in adult erythrocytes, and crossing these animals with a SCD murine model abolishes the SCD phenotype. BCL11A therefore constitutes a genetically validated target to induce HbF and reduce erythrocyte “sickling”, which would be predicted to ameliorate the phenotype of SCD patients. However, defective lymphoid development has been observed in Bcl11a genetic null mice, suggesting potential toxicities of BCL11A knockdown. We generated self-inactivating lentiviral vectors (LV) integrating miR-223 microRNA-based inhibitory shRNAs against BCL11A/Bcl11a. Since future clinical applications will need to balance efficacy and potential side effects, LVs were engineered to allow the comparison of effects of high level and ubiquitous versus erythroid lineage-restricted versus inducible expression of miRNA targeting BCL11A. LV backbones therefore included either a strong, viral LTR promoter/enhancer (SFFV-LV), a β-globin locus control region with the endogenous β-globin promoter (LCR-LV), or a tetracycline-inducible promoter (TET-LV). We performed assays to quantify transgenic miRNA expression and demonstrated that the BCL11A knockdown and induction of fetal globin gene output correlated with the expression of targeting miR223-based shRNA. Transduction at low MOI (=2) of murine hematopoietic stem cells (HSC) with LVs carrying the abovementioned regulatory elements leads to long-term engraftment and transgene expression in-vivo. Mice transplanted with SFFV-LV show fluorescent marking up to 70% across myeloid, lymphoid and erythroid lineages. The maximal BCL11A/Bcl11a mRNA and protein knock-down observed in primary hematopoietic cells in-vitro and in-vivo was 70%. This was confirmed in FACS-sorted bone marrow B-lymphoid (B220+) and erythroid progenitors (Terr119+/CD71+) and peripheral blood leukocytes at 4 months post-transplant. BCL11A/Bcl11a knockdown induced fetal globin gene expression depending on the vector backbone and targeting shRNA sequence employed. With SFFV-LV, we observed a 5–20 fold upregulation of fetal globin gene (γ/(ϵ+γ+β)) output in mice transplanted with HSCs containing the humanized β-YAC transgene. With TET-LV, the induction was dose-dependent and maximally caused a 150-fold increase in murine ϵγ-globin gene expression in-vitro. Human HSC transduced (MOI=2) with the LCR-LV and differentiated in-vitro resulted in a 3-fold increase of γ-globin mRNA in erythrocytes. SCD patient-derived HSC, which were transduced with LCR-LV (MOI=5) and transplanted into immunodeficient NSG mice, resulted in peripheral human erythrocytes that showed a reversal of the hemoglobin switch with a maximal induction 10% HbF as measured by flow cytometry. In a human ex-vivo B-lymphoid differentiation assay, SFFV-LV transduced (MOI=2) HSC populations with a 70% BCL11A knock-down showed no difference versus control in total cell numbers or in the sequential acquisition of CD43, CD19 and IgM (corresponding to physiological differentiation from common lymphoid progenitor to immature B-lymphocyte), thus showing no evidence for a differentiation block. Since IFN-response gene activation has been described with shRNA silencing and could potentially lead to HSC exhaustion, we quantified ISG20, ISG56 and OAS1 mRNA levels in human HSCs after miR-223-based SFFV-LV transduction (MOI=2). We observed less IFN-response gene activation in miR223-based SFFV-LV transduced HSC than in non-miRNA-based shRNA SFFV-LV transduced controls. In summary, our pre-clinical data demonstrates the potential efficacy of hematopoietic miRNA-mediated BCL11A/Bcl11a silencing to induce the expression of fetal hemoglobin in murine and human model systems, including primary cells. At the levels of BCL11A knock-down obtained, we did not observe any B-lymphoid toxicity. These results support the translation of LV-based miRNA-mediated BCL11A silencing into the clinical setting. Disclosures: No relevant conflicts of interest to declare.