Improving efficacy of in vivo therapy of sickle cell disease by hijacking natural biology of hematopoietic stem cells

Main text

In this issue of Molecular Therapy, Li et al. describe the amelioration of sickle cell disease (SCD) pathology following intravenous administration of a viral vector in mouse model of disease.¹ This vector delivered a base editor to hematopoietic stem cells (HSCs), correcting the sickle point mutation in the β-globin gene to a naturally occurring G-Makassar variant. The availability of a safe, effective, and broadly accessible in vivo gene therapy for SCD would dramatically improve the quality of life and extend the productive lifespan of millions of people living with SCD worldwide. Despite major advances in the treatment of SCD, all currently available pharmacologic therapies remain palliative. Two ex vivo cell therapies for SCD recently approved by the US FDA, Casgevy and Lyfgenia, are now available for clinical use; however, the complexity of ex vivo manipulation of autologous HSCs, the need for extensive chemotherapy conditioning with its potential for loss of fertility, and the extreme cost of these cell therapies at millions of dollars per patient significantly limit their equitable and widespread adoption.

The therapeutic “editing” of disease-causing mutations in HSCs in vivo, without the need for extracting and reintroducing HSCs back into the patient, has the potential to overcome many of the limitations of ex vivo therapy approaches. However, reaching and modifying genes in HSCs in vivo in numbers sufficient to alleviate pathology after intravenous administration of current therapeutic modalities is challenging due to several fundamental features of natural HSC biology. Although HSCs replenish all blood cell lineages for the duration of one’s life, they are rare and largely reside in bone marrow niches in a “dormant,” quiescent state. An earlier study estimated that the HSC pool in humans is only about 11,000 cells and that at any given time, 90% of the HSCs are in a dormant, quiescent state.² In a later study, using spontaneous mutations as clonal markers for quantification of the size of the active HSC pool, it was estimated that the number of HSCs contributing to hematopoiesis in humans is 50,000–200,000 cells per individual.³ Both studies agreed, however, that human HSCs divide extremely slowly, only once in 2–20 months.^2,3 The quiescent phenotype of HSCs represents a significant barrier for precise gene editing technologies due, in part, to the low levels of dNTPs in non-dividing cells, which are necessary for the synthesis of corrected DNA strand templates for gene modification with prime editing platforms.⁴

Another aspect of natural HSC biology that represents a barrier to efficient gene editing is the high sensitivity of HSCs to genotoxic stressors, which activate the p53-dependent DNA damage response, leading to cell-cycle arrest, senescence, and cell death. Ex vivo transduction of hematopoietic stem and progenitor cells (HSPCs) with adeno-associated viral (AAV) vectors for gene correction through homology-directed repair (HDR) mechanisms triggers a p53-mediated DNA damage response upon binding of the cellular MRE11-RAD50-NBS1 complex to the AAV vector genome inverted terminal repeates (ITRs), leading to cell-cycle arrest and a reduction in engraftment capacity in xenotransplant models.⁵ Although it is not surprising that the delivery of Cas9 DNA endonucleases to HSPCs (introducing targeted DNA double-strand breaks in the cell genome) activated the p53-pathway-dependent DNA damage response, this pathway was also activated after electroporation of human HSPCs with both base editors and prime editors, which do not require the introduction of double-strand DNA breaks in order to correct mutated genes.⁶ Optimization of HSPC manipulation protocols, the donor DNA platform for HDR-mediated repair, and the type of gene editing enzymes can significantly reduce the genotoxic stress response ex vivo and preserve the long-term engraftment capacity of edited HSPCs in mouse models.^5,6

In the current study, Li et al. used a non-replicating, high-capacity adenovirus vector platform devoid of all viral genes for delivery of the ABE8e-NRCH base editor to HSCs after intravenous vector administration to CD46/Townes mice, a well characterized mouse model of SCD.¹ Due to the large transgene capacity of this adenovirus-based gene delivery platform, in addition to the single-guided RNA (sgRNA) expression cassette (targets the β-globin gene containing the sickle mutation), the investigators also incorporated a transgene expressing the human MGMT^P140K variant, which provides a survival advantage to transduced HSCs that express it, when exposed to O6BG/BCNU chemotherapy conditioning.⁷ Use of this type of adenovirus vector for in vivo gene delivery to HSCs may be advantageous compared to other viral vector platforms because linear, double-stranded adenovirus genomic DNA is “capped” on both ends by a covalently bound viral terminal protein, TP,⁸ which prevents recognition of the linear vector DNA by the cellular DNA repair machinery, thus avoiding activation of the DNA damage response in transduced cells.

However, the most remarkable insight that the investigators reveal through their study is a demonstration that the quiescent nature of HSCs can be exploited to achieve phenotypic correction of SCD following in vivo expansion of corrected HSCs that transiently harbor a non-integrated episomal adenoviral vector genome expressing a chemotherapy resistance transgene (Figure 1). First, the authors used granulocyte colony-stimulating factor (G-SCF) and plerixafor to mobilize HSCs and HSPCs from the bone marrow into the blood and then injected the adenoviral vector into mice intravenously. At 6, 19, and 33 days post-vector administration, mice then received O6BG/BCNU chemotherapy.¹ Because committed progenitor and short-term repopulating cells divide rapidly, the episomal adenovirus vector DNA is lost in these cell populations. In contrast, in slowly dividing and quiescent HSCs and long-term repopulating cells, the episomal vector genome is transiently retained, and due to the expression of the MGMT^P140K, the cells are resistant to chemotherapy conditioning (Figure 1). Using this approach, the authors showed that whereas in vivo correction of the sickle mutation in peripheral blood mononuclear cells was 3% in mice without chemotherapy conditioning, in animals that were treated with the adenoviral vector followed by delayed chemotherapy conditioning, the proportion of peripheral blood mononuclear cells with corrected sickle mutation reached 23.8%.¹ The proportion of corrected cells in the periphery was stably maintained for up to 16 weeks post-vector administration, as well as in secondary recipients that were transplanted with Lin⁻ cells from the bone marrow of mice that received in vivo adenoviral vectortherapy. Importantly, this level of sickle mutation correction was sufficient to alleviate pathology associated with SCD, including normalization of hemoglobin levels, reduction in the percentage of reticulocytes and sickle cells in peripheral blood, and significant reduction in splenomegaly.¹

Figure 1.

Schematic diagram of a strategy for in vivo expansion of in-vivo-corrected HSCs based on transient expression of a chemotherapy resistance transgene from an episomal adenovirus vector genome in quiescent HSCs following in vivo transduction with a therapeutic adenovirus vector

LT-HSCs, long-term repopulating hematopoietic stem cells; HPCs, hematopoietic progenitor cells; TP, terminal protein; I.V. administration, intravenous administration; Ad, adenovirus.

Although highly encouraging, therapeutic application of the approach that Li et al. demonstrated in mice requires further refinement prior to use in patients with SCD. Because administration of G-CSF is contraindicated for patients with SCD due to toxicity, alternate mobilizing agents may be needed to achieve efficient in vivo transduction of HSCs after intravenous administration of adenovirus vectors to patients with SCD. In addition, HSCs in patients with SCD are very sensitive to genotoxic drugs, and the O6BG/BCNU conditioning that was used in mice cannot be used to improve the outcomes of the in vivo SCD therapy in patients. Several options for non-genotoxic conditioning are currently being developed that may allow for the in vivo selection of transduced HSCs, including epitope-edited CD123- or CD33-targeted antibodies^9,10 or a truncated EPO receptor.¹¹ It is also noteworthy that while red blood cells with a homozygous G-Makassar genotype (expressing HbGG hemoglobin) have no clinically significant pathology, red blood cells containing HbGS hemoglobin are dehydrated, poorly deformable, and still sickle under hypoxia.¹² These data indicate that to achieve a clinically meaningful improvement in SCD pathology in patients using base editors to create a G-Makassar β-globin variant, a bi-allelic correction of the sickle mutation in SCD HSCs is likely required. Lastly, accumulating evidence suggests that the efficacy of gene editing technologies observed in mouse models may not be easily replicated in large animal models, including non-human primates, and ultimately human patients. It is plausible that due to a faster HSC replication rate in mice (once in 2.5 weeks), compared to non-human primates and humans (once in ∼25 and ∼40 weeks, respectively²), the mouse HSCs are more permissive to gene editing compared to HSCs in larger species. To this end, a recent report on the impact of gene editing on HSPC dynamics in macaques showed that after efficient ex vivo editing of HSPCs with adenine base editing (ABE) and subsequent transplantation, the cells engrafted more slowly. High editing rates correlated with reduced long-term engraftment, suggesting that base editing has a net-negative impact on HSCs and that the majority of ex vivo base editing events occur in more rapidly dividing, short-term repopulating cells.¹³ Although there is clearly much work to be done before safe, effective, and accessible therapy can reach patients with SCD, the study of Li et al. represents a significant advance toward this goal and demonstrates that the efficacy of in vivo therapy through gene editing in the HSC compartment can be improved by capitalizing on the non-integrating episomal nature of the therapeutic adenoviral vector genome and on the quiescent nature of genuine HSCs.

Declaration of interests

D.M.S. is listed as inventor on patents and patent applications related to adenovirus vector technologies. D.M.S. is a co-founder and shareholder of AdCure Bio, which develops adenovirus technologies for therapeutic use.