Abstract
What is the rationale for FGT over postnatal therapies? The early gestation fetal environment is unique in many ways that may facilitate efficient gene transfer and overcome some of the major obstacles that continue to limit the success of adult gene therapy trials. First, the fetus is very small, weighing only 23 g at the beginning of the second trimester (∼13 weeks of gestation). Compared with an adult, significantly smaller volumes of vector are required to treat the fetus and, on a per-kilogram basis, far higher doses of vector can be administered. Second, there is an exponential expansion in organ growth during fetal development. As a point of reference, fetal body weight increases more than 150-fold between 13 weeks of gestation and full term. Accordingly, there is a greater proportion of resident stem/progenitor cells present during early fetal development that, if transduced with genome-integrating vector, can profoundly increase transduction efficiency and provide lifelong therapeutic transgene expression (Endo et al., 2010). Third, during early gestation while the thymus is processing self-antigen, the fetal immune system can be made tolerant to foreign antigen. During this period, appropriate presentation of foreign antigen (i.e., vector, transgene) to the fetal thymus during early development can induce lifelong tolerance to antigen (Tran et al., 2001; Waddington et al., 2004; Sabatino et al., 2007). Thus, FGT could theoretically be used to induce immunological tolerance to vector/transgene before birth and enable repeated administration of vector/transgene after birth while avoiding seroconversion. At present, there is a paucity of information on which antigens will or will not induce a fetal immune response and when a response will appear during gestation. The last, and probably most compelling, rationale for FGT over postnatal treatment is the ability to treat or cure disease before the onset of irreversible organ damage. It could be argued that the risk-to-benefit ratio of FGT is too high for genetic diseases that are not prenatally lethal and for which the onset of disease pathology occurs well after birth. Why risk maternal–fetal health given the potential success of postnatal gene therapies? The counterargument is that FGT, because of the biological advantages cited previously, may facilitate gene transfer that will never be possible later in life. In the example of cystic fibrosis (CF), results of pulmonary gene transfer trials using adeno-associated virus serotype 2 (AAV2) have been, in general, of low efficiency due to physical and immunological barriers impairing gene transfer. Physical barriers including airway mucous and cilia, extracellular glycoproteins (i.e., glycocalyx), and pulmonary macrophages, which prevent vector access to the epithelial surface. Compounding those problems is the adult immune response, which prevents gene transfer by repeated administration of the same gene transfer vector serotype (Moss et al., 2004). Many of these barriers are not present during early fetal development and, with access to a greater number of airway stem/progenitor cells, it may ultimately be more efficient to treat CF in utero.
The choice of vector system for FGT is critical for the successful achievement of therapeutic goals. Viral vector systems have proven to be most effective for transferring genes to the developing fetus, and the merits and challenges of each are briefly summarized. Adenovirus-based gene transfer vectors are unlikely candidates for long-term corrective gene therapy because of their short-lived (i.e., weeks) transgene expression. However, these vectors may be extremely useful in prenatal management of developmental conditions requiring transient gene expression, and the detrimental inflammatory effects seen postnatally do not occur early in gestation. AAVs have a low immunogenic profile and transduce nondividing cells with only low-frequency integration into the host genome. Because AAV genomes are not replicated during cell division, which occurs at a rapid rate during early fetal/postnatal development, one would anticipate a rapid decline in transgene expression (Boyle et al., 2001). HIV-1-based lentiviral vectors (LVs) are able to integrate into the host genome to provide lifelong stable transgene expression (Endo et al., 2010; Stitelman et al., 2010). An undesirable consequence of LVs for gene therapy is the risk of activation of oncogenes and/or insertional mutagenesis, which may be heightened for FGT because of rapid fetal growth. Germline integration is another serious concern for LV gene therapy, but experimental data thus far do not support a high risk of germline alteration with FGT.
If the goal of human FGT is to be achieved, proof-of-principle studies performed primarily in rodents must be translated into preclinical models. In this issue of Human Gene Therapy, David and colleagues (2011) provide an eloquent report of long-term exogenous gene expression after liver-directed gene transfer to pre- and postimmune fetal sheep. The ovine model represents an important preclinical model for addressing the safety and efficacy of FGT, as it shares many important physiological, developmental, and immunological characteristics with humans, and fetal lambs are of sufficient size to perform early gestational gene transfer by minimally invasive techniques (Porada et al., 2004). Under ultrasound guidance, fetal sheep were intraperitoneally injected with AAV2/8 expressing human factor IX (hFIX) under the control of the LP1 promoter during early (60–65 days) or late (97 and 105 days) gestation (term, ∼145 days). Initial analysis of fetal plasma hFIX levels at ∼3 weeks after gene transfer demonstrated that, despite a more than 10-fold higher vector dose per kilogram administered to younger fetuses, serum hFIX levels were highest in fetuses injected at late gestation. Regardless of timing of injection, hFIX levels gradually declined during postnatal development and by adult life were subtherapeutic or undetectable. As discussed by the authors, this result would be expected given the nonintegrating nature of the AAV vector and rapid growth of the liver over the study period. Liver weight in sheep increases more that 120-fold between 60 days of gestation and adulthood (Thurley, 1973; Upton, 2008). In contrast to the results obtained by David and colleagues using AAV, similar studies using integrating viral vectors demonstrate that the transduction efficiency of fetal liver is inversely proportional to gestational age at injection (Porada et al., 2005; Endo et al., 2010). Taken together, these studies support the premise that compared with AAVs, integrating viral vectors may be better suited to achieve stable transgene expression in fetal organs with high rates of proliferation.
Strong antibody responses to hFIX challenge were observed in lambs irrespective of timing of fetal vector administration (i.e., early or late), indicating that permanent tolerance to foreign antigen was not achieved under these experimental conditions. This result may be explained by too late an administration of vector to achieve tolerance in the late administration group, and insufficient sustained levels of circulating hFIX during fetal development, and/or inappropriate presentation of antigen during fetal thymic processing of self-antigen in the early administration group. Successful induction of fetal tolerance to transgene has been demonstrated in mice using AAV vector (Sabatino et al., 2007) and in sheep using LV (Tran et al., 2001). This study emphasizes the need for better understanding of what the requirements are for tolerance induction in fetal systems. Clearly, the timing, mode of administration, and choice of vector all likely impact the likelihood of inducing tolerance.
Last, the mode of vector administration is an important consideration for future clinical gene therapy strategies. Intravenous delivery of AAV via the umbilical circulation, an alternative strategy to that used in this study, would achieve more direct delivery of viral particles to hepatocytes via the liver sinusoids than could be achieved by intraperitoneal injection. Such a strategy would, theoretically, increase transduction efficiency but would carry an increased risk of fetal loss if performed during early gestation (Orlandi et al., 1990). Widespread distribution of transgene to other organs may be anticipated after intravenous administration of AAV, although as clearly demonstrated by David and colleagues, expression of therapeutic protein could be restricted to the liver by incorporating cell-specific promoter elements within the transgene. A greater understanding of in vivo fetal tissue specificity for tropism-modified viral vectors will be an important step in achieving safe and targeted gene delivery to the fetus. As a final comment, we wish to congratulate Dr. David and colleagues on completing a well-designed and technically challenging study.