Gene transfer to HSCs utilizing lentiviral vectors is a promising modality for the treatment of human diseases, and large-animal models remain important to inform the clinical effort. Among large-animal models, the rhesus macaque has proven well suited for preclinical safety and efficacy studies. Recently in a human trial of gene therapy for X-linked severe combined immunodeficiency using an MLV vector to transduce bone marrow-derived HSCs, four patients developed T-cell-type acute lymphoblastic leukemia caused by insertional mutagenesis (11
), and additional events have heightened the need to better understand the underlying mechanism eliciting these events (28
). New vector systems including SIN vectors have emerged as an alternative to MLV vectors. However, several groups have been unsuccessful in developing HIV-1 vectors for preclinical testing in rhesus models (29
), while others have achieved long-term marking of only 5% or less (2
). These events create a mandate to develop such relevant large-animal gene therapy models for preclinical testing of HIV-1 vectors.
TRIM5α and APOBEC3G are species-specific restriction factors that block retroviral infection (22
). Rhesus TRIM5α can bind to the hCA, mediate HIV-1 RNA degradation, and block HIV-1 infection, while SIV (containing sCA) can escape rhesus TRIM5α-mediated viral degradation. These restriction factors have important implications in gene transfer experiments, and several strategies to modulate these factors have been employed to improve transduction by viral vectors. One such approach is to overwhelm all available TRIM5α by transducing at high MOIs. This strategy was demonstrated to be somewhat effective in our study (Fig. ), where the conventional HIV-1 vector at MOIs of 10 or higher were shown to transduce rhesus cells in vitro. This method, however, remains inferior to the χHIV vector and cannot promote efficient transduction of rhesus hematopoietic cells in vivo (Fig. ).
HIV-1 vectors that include mutations in the CypA-binding domain of the capsid sequence have been developed as another strategy to overcome the resistance to transduction in old world monkeys. These modified vectors have been reported to transduce both human and baboon CD34+
) but are not sufficient to allow escape from rhesus TRIM5α, and transduction rates among rhesus CD34+
cells remained suboptimal (17
). By exchanging the entire hCA sequence with that of the sCA, we found that the χHIV vector efficiently transduces both human and rhesus CD34+
cells in vitro, suggesting that sCA escapes from rhesus TRIM5α binding. Our proof-of-concept results are further confirmed by a recent study which demonstrated that an experimental HIV-1 virus encoding sCA and SIV-Vif could replicate in rhesus peripheral blood cells through escape from rhesus TRIM5α and APOBEC3G, respectively (16
). Our χHIV vector which includes sCA showed transduction efficiency for both human and rhesus blood cell lines superior to those of the alternative vectors.
As an additional strategy, we sought to incorporate Vif into our vector system in order to allow binding to APOBEC3G and enhance viral transduction. Our inclusion of SIV or HIV-1 Vif did not demonstrate the intended results and modestly decreased the viral titer during vector production. Although APOBEC3G plays no direct role in inhibiting transduction, several reasons explain why the addition of SIV Vif did not further improve transduction rates in rhesus blood cells. First, our lentiviral vectors do not contain the APOBEC3G protein in the virion particles because 293T cells do not express APOBEC3G in our vector production system (data not shown). Therefore, SIN lentiviral vectors that transduce only once theoretically do not require the SIV Vif gene to transduce rhesus HSCs. Second, when APOBEC3G is incorporated into HIV-1 virions during virus production, it can dramatically reduce viral infectivity in the absence of Vif (20
). However, when there is a species mismatch between APOBEC3G and the genetically inserted Vif, the APOBEC3G-Vif complexes do not form, and APOBEC3G (produced by rhesus or human host cells) is allowed to suppress viral infection (5
). Additionally, APOBEC3G has a Vif-independent pathway for inhibiting viral replication (7
). This prior observation is consistent with our escalating viral vector exposure experiments with primary CD34+
cells, where transduction rates plateaued well before 100%. Collectively, these observations suggest gene transfer to CD34+
cells is affected by restriction factors other than TRIM5α and APOBEC3G and thus warrants further exploration.
Though we succeeded in achieving high-level gene transfer to bulk CD34+
cells, clonogenic progenitors, and myeloid and erythroid cells differentiated in liquid culture in vitro, demonstration of transduction of the true HSCs requires transplantation experiments at a minimum. Recently it has been shown that HIV-1-based vectors efficiently transduce repopulating HSCs of the pigtailed macaque, in which TRIM5α undergoes alternative splicing (26
). While higher-level marking is feasible in this model, vector integration under the control of an alternatively spliced TRIM5α (26
) may not be reliably predicted. In our rhesus competitive repopulation assay, preliminary gene marking levels in all blood lineages derived from the χHIV vector were significantly higher than that from the conventional HIV-1 vector and plateaued at levels of 15 to 30% 3 to 7 months after transplantation, suggesting that this high-level marking in peripheral blood cells originates from HSCs (32
). Granulocytes and platelets, which have short half-lives, arrived at plateau levels of gene marking earlier (about 1 month), while lymphocytes and red blood cells, which have longer half-lives, reached a plateau in marking levels later (2 to 3 months). These early results thus suggest transduction of the true HSCs by our χHIV vector at potentially clinically relevant levels.
An important advantage of our vector system is that HIV-1 helper plasmids can be combined with the SIV vector plasmid to prepare vectors, while the HIV-1 vector plasmid could not be combined with SIV helper plasmids except by the exchanging of the capsid sequences only. Using our system, existing HIV-1-based therapeutic vector plasmids can therefore be utilized to produce χHIV vectors by simply replacing HIV-1 Gag/Pol plasmid with that of χHIV Gag/Pol (HIV1+sCA). Moreover, the χHIV system can produce high-titer vectors comparable to conventional HIV-1 vectors, with levels as high as 3 × 109 IU/ml following concentration by ultracentrifugation. Our vector system should thus allow testing of a variety of therapeutic vectors using conventional HIV-1 plasmids, already constructed in the large-animal rhesus model.
In summary, we developed an HIV-1 vector system that allows efficient transduction of both human and rhesus HSCs. We believe this vector system will prove invaluable for comprehensive preclinical safety and efficacy testing of a variety of potentially therapeutic vectors in the rhesus macaque model.