HSC transfer and permanent high level erythroid expression of a γ-globin gene is an attractive approach for gene therapy of SCD[11
]. This is the first report to establish that this approach can successfully treat an exceptionally severe murine model of SCD[30
]. Although it is not known what degree of myeloablative conditioning in humans will be needed to obtain sufficient engraftment of genetically modified HSCs for a therapeutic effect, a subablative regimen would be preferable since many SCD patients have pre-existing multi-organ damage. One advantage of using γ-globin, rather than a mutant β-globin molecule, is that additional chemotherapy or irradiation to achieve immunoablation would not be required since γ-globin is endogenously expressed in SCD patients. Numerous examples of self immune responses to cells expressing novel, mutated endogenous antigens have been previously documented [42
], raising the possibility that expression of an altered version of β-globin in developing BM erythroid precursors could be problematic.. The potential need for immunoablation in the conditioning regimen with this approach would add significant risk to the gene therapy procedure for patients with SCD.
We observed an increase in Hb per vector copy of 2.6 g/dL with the V5 vector, similar to what we found in a previous gene therapy study involving β-thalassemia mice[27
]. The V5m3 vector, containing the β-globin 3’UTR in lieu of its γ-globin counterpart, produced a more robust improvement in Hb, with a rise of 4.1 g/dL per vector copy. These values exceed those observed by others (1.9–2.1 g/dL per copy) using a mutant β-globin vector in the same or similar SCD mouse models[28
]. Mice receiving transplants of cells transduced with the V5m3 vector, relative to the V5 vector, had a significantly better correction of their anemia as evidenced by their higher Hb and RBC counts. Extramedullary hematopoiesis, as judged by spleen size and pathology, was also more completely resolved in the V5m3 group than the V5 group. This improved performance of the V5m3 vector correlated with the higher levels of γ-globin mRNA per vector copy, relative to that observed in animals transplanted with V5-transduced cells. Recent studies indicate that, in addition to the normal developmental transcriptional down-regulation of γ-globin, there is also a post-transcriptional silencing mechanism affecting γ-globin mRNA stability in adult erythroid cells[37
]. Russell has proposed a model based on data in transgenic mice wherein limiting quantities of α-CP bind preferentially to the 3’UTR of “adult” globin mRNAs, leaving stage inappropriate globin mRNAs such as γ-globin destabilized. Consistent with this model, others previously showed that accumulation of the native γ-globin transcript occurs most optimally in the absence of or with reduced β-globin expression[36
] Our data are also in line with these results as we observed that removal of the γ-globin 3’UTR from the V5 vector and replacement with its β-globin counterpart led to a significant 2.5-fold increase in the relative amount of steady-state, γ-globin mRNA per vector copy. It appears that the native γ-globin transcript, which does not contain the unique β-globin 3’UTR stem-loop motif that mediates the specific protein binding necessary for β-globin transcript stability [35
], is not optimal for high level γ-globin protein expression in the adult erythroid environment.. In addition to potentially enhancing γ-globin transgene mRNA stability, it is also possible that the 545 bp of β-globin 3’ sequences that extend beyond the 3’UTR and are included in the V5m3 vector might play a role in enhancing transgene mRNA levels. For instance, these sequences might enhance pre-mRNA processing, including polyadenylation. Further studies may allow delineation of the mechanism(s) that facilitate higher globin mRNA production by the V5m3 vector.
Despite the higher levels of γ-globin transgene mRNA in the PB of V5m3 mice compared to V5 mice, which correlated with a more complete correction of all aspects of the SCD phenotype that we measured, the differences in the mean and variance of HbF level of the two groups did not achieve statistical significance. This is most likely due to the limited number of animals analyzed. However, we cannot rule out that there is a steep response curve, as a function of HbF level, for achieving increasingly more complete phenotypic correction in this model.
Transplantation of BM from individual V5m3 primary recipients into secondary recipients showed continued, high levels of γ-globin expression, similar to those observed in the primary donors. This not only suggests effective HSC gene transfer but also is consistent with a lack of significant vector silencing. Despite this, we did observe evidence of position effect variegation of vector transgene expression. The correlation between BM VCN and the amount of HbF was not copy number dependent (r2
=0.37 and data not shown). In some mice, lower VCN gave equivalent HbF levels as those observed in some mice with higher VCN (data not shown). These results are consistent with our previous work and that of others highlighting the impact of position effect on globin lentiviral vector expression[25
]. This reflects the variability of expression of an integrated transgene at ectopic genomic locations that do not reproduce the transcriptional environment of the native locus. A recent study indicates that detrimental position effects can be reduced by use of the 1.2 kb insulator element from the chicken β-globin locus[46
]. However, inclusion of the 1.2 kb fragment in lentiviral vectors, particularly globin vectors, can significantly reduce titer[47
](Hargrove, Hanawa, and Persons, unpublished data). One potential solution would be to utilize a 400 bp subfragment that seems to retain significant protection against position effects [48
]. Further studies will be required to identify the optimal insulator element which retains functional activity while having minimal impact on globin vector titer.
In addition to their sickle cell phenotype, the BERK mouse model used in these studies exhibits a mild β-thalassemic phenotype, with an α/β globin chain ratio of 0.79–0.82[21
]. The relative deficiency of βS
chains in this model closely represents the situation of humans with sickle/β0
-thalassemia. Given our data and previous studies showing that SCD transgenic mice with a relative deficiency of βS
chains have a reduction in the HbF level therapeutic threshold [21
], patients with sickle/β0
-thalassemia may be an ideal population for initial γ-globin gene therapy clinical trials. Similarly, incorporation of an shRNA targeting βS
-globin into the γ-globin vector, as proposed by Sadelain and colleagues, may be beneficial to improve the therapeutic efficacy for SCD patients with balanced globin chain synthesis[49
Therapeutic protein production was obtained in the majority of the mice in this study with an average VCN ranging from 0.3 to 1.8 copies, less than in previous studies[28
]. One of the goals in HSC-targeted gene therapy is to obtain functional cell correction with the lowest number of vector copies per cell since the risk of genotoxicity likely increases with increasing copy number[50
]. Although this work was not a safety study, we observed no evidence of myelodysplasia or leukemia in the different groups of primary and secondary transplant recipients. This is consistent with our previous work that showed that although globin vector integrations can lead to cellular gene dysregulation in erythroid cells, even at great distances, no functional consequences in terms of clonal dominance or leukemia occurred. Using a primate autologous transplant model, we have recently found that lentiviral vectors have a more favorable vector insertion profile than that associated with γ-retroviral vectors (Persons and Dunbar, manuscript submitted). We therefore believe that use of a γ-globin lentiviral vector for gene therapy of human SCD will have an acceptable benefit-to-risk ratio. With continued focused effort, successful gene therapy for SCD may become a reality in the near future.