Over the years we have worked with a number of different gene delivery vehicles, but most of our studies focus on adeno-associated viral (AAV) vectors. Engineered from a naturally occurring parvovirus, to which most of us are exposed during childhood, AAV vectors are fully deleted of viral-coding sequences, which are replaced by the therapeutic gene of interest under the control of an appropriate promoter. The field was hampered initially by the difficulty of producing high titers of purified recombinant AAV virions, but advances in manufacturing have removed this barrier, allowing clinical trials to move forward, leading to therapeutic success in some cases and permitting sharper definition of obstacles in others.
AAV vectors transduce efficiently a wide range of target cells, including hepatocytes, cells in the central nervous system, cardiomyocytes, and skeletal muscle. However, these cell types are not easily removed from the patient for genetic manipulation in the laboratory, so most AAV-mediated gene transfer is done by direct injection into the patient, i.e., in vivo gene transfer, a process that necessitates grappling at close quarters with the human immune response. From a biological perspective, this work presents the problem of analyzing the human immune response to a particle that externally resembles a virus but which lacks the virus's capacity to synthesize viral antigens, which are both targets and modulators of the host immune response. From a therapeutic perspective, the problem is to control or circumvent the host immune response, so that the viral-like particle can lead to efficient transfer of the genetic payload without invoking an immune response that harms the transduced cell or the patient. Superimposed on this is the additional complexity that the human immune response exhibits considerable heterogeneity based on the tissue in which an antigen is encountered. Thus, for clinical gene therapy, each target tissue presents a unique series of responses to be characterized and controlled.
Gene Therapy for Hemophilia B
The first trial of AAV-mediated, liver-directed gene therapy for hemophilia B, conducted by our group, led to the realization that the vector could transduce human liver and lead to the synthesis of biologically active factor IX at therapeutic levels (10–12 percent of normal) but that the duration of expression in humans was transient, that the gradual decline in factor IX levels was accompanied by a transient rise and fall in hepatic transaminases (a signal of hepatocyte damage), and that even modest titers of neutralizing antibodies to AAV prior to treatment were sufficient to block transduction completely.
Our research efforts have focused on understanding the mechanism of loss of expression. Based on data from clinical studies and in vitro models, we showed that the rise and fall in transaminases was accompanied by the expansion and contraction of a population of AAV capsid-specific CD8+ T cells. We proposed that in humans, memory CD8+ T cell responses to AAV capsid induced by a natural infection during childhood had been reactivated and had eliminated the AAV-transduced hepatocytes. The clinical data further suggested that the immune response was dependent on vector dose. This suggested a path forward, which was confirmed in a subsequent trial, in which transient immunomodulation with steroids controlled the immune response and resulted in long-term expression of the donated gene, ameliorating severe hemophilia to a mild condition that greatly reduced or in some cases abolished the need for recurrent administration of clotting factor concentrates. Currently we are focused on circumventing the other critical obstacle posed by the human immune response, preexisting antibodies to AAV that prevent transduction altogether in as many as 40 percent of adults. We have presented data in animal models on a strategy that depends on adding empty capsids to the final formulation of vector, to act as decoys to absorb antibodies.
The ability to overcome neutralizing antibodies addresses a problem of general significance affecting all gene delivery strategies in which vector is infused through the circulation, including to brain, liver, and cardiac and skeletal muscle.
Gene Transfer to the Subretinal Space for Inherited Retinal Defects
Over the past five years we have continued our successful collaboration with Jean Bennett and Albert Maguire (Perelman School of Medicine of the University of Pennsylvania) to develop a therapeutic approach for a rare form of inherited blindness, Leber's congenital amaurosis type 2. In our first phase I/II study we established the safety and efficacy of unilateral subretinal administration of an AAV2 vector encoding the gene for RPE65 into 12 children and adults between the ages of 8 and 44. In a follow-on study we showed the safety and efficacy of administration of vector to the contralateral eye in 11/12 of these subjects. Immune responses, which were carefully monitored, did not present an obstacle in this setting, where small doses are injected into an immunoprivileged site. Currently we are conducting a phase III pivotal trial in children and adults age 3 and up, who are receiving bilateral vector injection in a study with a randomized crossover design.
In Vivo Genome Editing as an Approach to Treatment of Genetic Disease
A long-term goal of the field of gene therapy has been to develop methods to correct mutations in situ rather than simply provide a wild-type copy of a mutant gene. Initial efforts to accomplish this were unsuccessful because of the very low frequency of gene-targeting events. Introduction of a double-strand break (DSB) at the locus to be corrected, and the development of synthetic molecules that could introduce a DSB at a specified locus, led to successful gene targeting in human cells in tissue culture using zinc finger nucleases (ZFNs). Potential therapeutic applications were extended by demonstration that a donor template with arms of homology, used to repair the break by homology-directed repair, could be used to introduce additional sequences, up to 8 kb, at the specified locus.
Because we are interested in treating diseases that affect the central nervous system and the liver, we have sought to determine whether we could combine our expertise in AAV-mediated gene transfer with the emerging field of ZFN therapeutics to achieve in vivo genome editing. Using ZFNs that recognize and cleave specific sites within the human F9 gene, and a suitably designed donor template, we showed, in both neonatal mice (where hepatocytes are rapidly cycling) and in adult mice (where the cells are quiescent), that injection of the donor and the ZFN vectors together resulted in reconstitution of a functional hF9 gene in a mouse carrying a single copy of a mutant hF9 gene, with circulating levels in the range of 200–350 ng/ml (4–7 percent of normal), enough to correct the hemophilic phenotype of the mouse. We are focusing now on understanding differences in the mechanism of genome editing in these two settings, with the hypothesis that the relative levels of homology-directed repair and nonhomologous end joining differ between them.
Grants from the National Institutes of Health, the Hemophilia Association of New Jersey, and the Delaware Valley Chapter of the National Hemophilia Foundation, as well as support from the Children's Hospital of Philadelphia, provided partial support for this research.
As of January 30, 2014