Plerixafor+G-CSF–mobilized CD34⁺ cells represent an optimal graft source for thalassemia gene therapy

β-Thalassemias are monogenic disorders causing partial or complete elimination of β-globin expression.¹  Allogeneic hematopoietic stem cell transplantation (allo-HSCT), although curative, is limited by the lack of matched donor availability.^2,3  Thus, most patients undergo palliative, lifelong transfusions and chelation¹  that compromise quality and duration of life.⁴  Gene therapy (GT), by introducing the normal globin gene into patients’ own CD34⁺ cells, is devoid of the complications usually accompanying allo-HSCT, and it is anticipated to permanently correct erythropoiesis, rendering patients transfusion free.^5,6  Mobilized blood represents the preferable source for HSCT and stem cell GT due to the high numbers of CD34⁺ cells provided through a minimally invasive procedure.^7,8  We previously defined Plerixafor+granulocyte colony-stimulating factor (G-CSF) as the optimal mobilization approach for thalassemia GT over other strategies (G-CSF, hydroxyurea [HU]+G-CSF, Plerixafor) in 2 clinical trials.^9,10  Because differently mobilized cells may possess distinct intrinsic characteristics, we compare here their potency regarding gene transfer and vector-encoded β-globin expression in vitro and in xenografts to define the optimal graft for thalassemia GT.

The 2 mobilization trials in adults with thalassemia were previously described.^9,10  The bulk of the purified CD34⁺ cells was cryopreserved for future clinical-grade gene transfer, whereas small aliquots were frozen for the preclinical studies, described herein.

CD34⁺ cell aliquots with >85% purity, from 31 differently mobilized thalassemic patients (HU+G-CSF: n = 7, G-CSF: n = 7, Plerixafor: n = 13 and Plerixafor+G-CSF: n = 4; supplemental Table 1 available on the Blood Web site) were thawed, prestimulated overnight, and subjected to two 24-hour transductions (multiplicity of infection 25 × 2) with the research-grade TNS9.3.55 β-globin lenti-vector,¹¹  currently being clinically tested (#NCT01639690).¹²  Transduced cells were seeded in clonogenic assays and erythroid differentiation culture¹²  and, if in sufficient numbers, xeno-transplanted (supplemental Figure 1) to comparatively assess gene transfer rates, β-globin expression, and engraftment. Integration site (IS) analysis was performed on genomic DNA from erythroid differentiation liquid cultures derived from transduced CD34⁺ cells.

IL2Rγ^null mice were conditioned with 100 mg/kg Busulfan, corresponding to the human equivalent, nonmyeloablative dose of 8 mg/kg (www.accessdata.fda.gov/scripts/cder/onctools/animalquery.cfm), and infused with 1 × 10⁶ transduced CD34⁺ cells, under a protocol approved by the institutional animal care and use committee. For more information, see supplemental Methods.

TNS9.3.55-transduced and untransduced CD34⁺ cells provided similar clonogenic potential and expanded and differentiated successfully in erythroid liquid culture (supplemental Figure 2A-C), irrespective of the mobilization approach. The overall mean gene transfer in clonogenic assays was 48.4 ± 2.8% and the mean vector copy number (VCN) reached 1.91 ± 0.04 (Figure 1A-B, left). Plerixafor-mobilized cells exhibited higher colony transduction rates (Figure 1A, right) compared with Plerixafor+G-CSF (P = .04). This might be causally linked to the cycling status of Plerixafor-mobilized cells, as they comprised a greater fraction of cells in the G1 phase (P = .01; supplemental Figure 2D).^13,14  Interestingly, and apparently not linked to quiescent cell cycle kinetics (supplemental Figure 2D), Plerixafor+G-CSF–mobilized CD34⁺ cells displayed significantly lower colony VCN compared with all other mobilized cells (P < .0001; Figure 1B, right). We showed that murine thalassemic Plerixafor+G-CSF–mobilized cells are immunophenotypically (CD150⁺/CD48⁻/lin⁻/sca-1⁺/c-kit⁺) and functionally more primitive than differently mobilized cells,¹³  and Genovese et al¹⁵  demonstrated that the most primitive, and with the highest long-term repopulation potential, CD34⁺ cell subset (CD34⁺/CD133⁺/CD90⁺) displayed the lowest transduction efficiency; these findings may tie in with the comparably lower transducibility οf Plerixafor+G-CSF cells in the current study, although the readout from our assays derives from progenitor cells, as it is usually the case with globin vectors.

The mean β-globin expression by high-performance liquid chromatography, either in terminally differentiated bulk erythroid suspension cultures or in pools of erythroid bursts (BFU-E), in individually analyzed samples from all the differently mobilized CD34⁺ cells, approximated the normal hemizygous expression with an average of 1.74 and 0.67 VCN/cell, respectively (Table 1). Interestingly, although globin expression after transduction did not differ among the 3 different genetic backgrounds (β⁰/β⁰, β⁰/β⁺, and β⁺/β⁺; supplemental Figure 3A-B), the net increment of β/α ratio provided by the vector was highly different among genotypes; the β⁰/β⁰ background induced the highest gain in expression (P < .0001), whereas the β⁺ allele seemed to partially compromise the transgene-derived β-chain synthesis (Table 1). The mean overall net increase of β/α ratio after transduction was 0.33, and the normalized per VCN β-globin gain in expression was 0.21 in bulk and 0.52 in BFU-E (Table 1). In fact, the vector-encoded β-globin expression/VCN is underestimated in bulk culture, because the TNS9.3.55 lots were derived from research-grade preparations and the likely transient presence of plasmid may lead to an overestimation of VCN. These expression levels, albeit measured in vitro, allow us to anticipate that less than 2 VCN/cell will probably suffice to provide significant clinical benefit to both β⁰- and β⁺-thalassemia. Interestingly, and despite their lower transduction efficiency, Plerixafor+G-CSF–mobilized cells demonstrated higher β-globin expression per VCN compared with the other mobilization groups (bulk: P = .05; BFU-E: P = .03; Figure 1C-D). This difference is neither due to a confounding effect of the genotype, as it was not associated with a higher frequency of β⁰/β⁰ genotypes in the Plerixafor+G-CSF group (25% vs 67%, 60%, and 25% of the HU+G-CSF, G-CSF, and Plerixafor groups, respectively; supplemental Table 1) or a higher erythroid lineage transduction in Plerixafor+G-CSF–mobilized CD34⁺ cells over the differently mobilized cells (supplemental Table 2). Differential gene expression and chromatin remodeling events may, for different mobilization treatments, bring about accessibility to alternative genomic regions for vector integration, and in the case of Plerixafor+G-CSF mobilization, select IS that favor high levels of expression. IS analysis demonstrated a highly polyclonal global integration pattern with similar distribution across all mobilization groups (supplemental Table 3). Plerixafor+G-CSF cell IS were slightly more enriched within genes and near gene promoters associated with globin expression (supplemental Table 4), but not significantly overrepresented. Gene ontology analysis negatively correlated Plerixafor+G-CSF cell IS with gene silencing and positively correlated with positive regulation of the cell cycle and telomere maintenance and organization (supplemental Table 5; supplemental Results).

IL2Rγ^null mice conditioned with a nonmyeloablative dose of Busulfan and infused with equal numbers of the differently mobilized CD34⁺ cells demonstrated long-term hematopoietic repopulation (Figure 1E-I; supplemental Figure 4A). Gene marking in terms of VCN in blood, percent vector-positive colonies, and colony VCN per cell, as well as β-globin expression derived from ex vivo expanded and terminally differentiated bone marrow cells, was persistent long-term (supplemental Figure 4B-C) and similar to that measured in the preinfusion product (supplemental Table 6). Plerixafor+G-CSF cells displayed enhanced short-term engraftment over all other differently mobilized grafts (Figure 1E-H), a feature that may be clinically translatable into faster hematologic reconstitution and abbreviated hospitalizations and therefore significantly reducing early transplant-related morbidity/mortality. Admittedly, our long-term engraftment data are limited by the small number of surviving mice; thus, we cannot draw conclusions in the xenograft model. Nevertheless, we previously demonstrated, in the less vulnerable to mortality Hbb^th-3 mouse model, a clear superiority in the long-term engraftment capacity of Plerixafor+G-CSF cells over differently mobilized cells.¹³  Genovese et al¹⁵  showed that by modifying the culture process, the transduction rates of the CD34⁺ cell subset with the highest long-term repopulation potential but the lowest transduction efficiency were significantly increased, yet with preservation of the primitive cell nature. Similar manipulation for Plerixafor+G-CSF–mobilized cells might thus boost VCN, maintaining their high β-globin expression/VCN and their long-term repopulating potential.

Overall, the high CD34⁺ cell numbers obtained by Plerixafor+G-CSF mobilization through single collections, even in hard-to-mobilize thalassemic patients,¹⁰  coupled with the superior short-term engraftment and increased β-globin production/VCN of Plerixafor+G-CSF cells, render them an optimal graft for thalassemia GT. Importantly, a given required expression level might be achieved by Plerixafor+G-CSF–mobilized cells at a lower VCN and, subsequently, higher biosafety for clinical applications. In summary, we provide data on the quality and preclinical efficacy of mobilized transduced HSCs from patients with thalassemia that may prove critical for the optimization of clinical GT protocols. Additional information can be found in supplemental Results and Discussion.

Contribution: G.K. designed and conducted experiments, collected and assembled data, performed data analysis and interpretation, and wrote the manuscript; N.P., J.E.A., F.Z., and N.Z. conducted experiments or/and data collection; C.W.L. conducted experiments and data collection and edited the manuscript; M.K., C.T. and A.A. provided administrative or/and financial support; M.S. and I.R. supplied study materials and provided administrative support; G.S. provided financial and administrative support and gave final approval of manuscript; and E.Y. conceived and designed the study, interpreted data, provided financial and administrative support, wrote the manuscript, and provided final approval of manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Evangelia Yannaki, Gene and Cell Therapy Center, Hematology Department-Bone Marrow Transplantation Unit, George Papanicolaou Hospital, Thessaloniki 57010, Greece; e-mail: eyannaki@u.washington.edu.

The authors gratefully acknowledge Dr Thalia Papayannopoulou for critically reading and commenting on the manuscript and thank student Penelope Papayanni for providing technical assistance.

This work was supported by National Institutes of Health, National Heart, Lung, and Blood Institute grant P01 HL053750-19, “Cooperation-Action I” National Strategic Reference Framework (ESPA) Program 09SYN-12-1159, and EU FP7 grant agreement 306201 (THALAMOSS).