Our research, both in animals and humans, has focused on determining the basis for defects in coagulation and on developing novel genetic strategies for their treatment. The most common of these defects, hemophilia, is caused by the absence or malfunction of genes that code for the production of proteins essential for clotting. Most of the approximately 20,000 people in the United States who have hemophilia lack clotting factor VIII (hemophilia A). The remainder (approximately 15 percent) lack properly functioning factor IX (hemophilia B, or Christmas disease).
Hemophilia is an attractive target for studies of novel genetic therapies for several reasons. Factor IX is encoded by a single gene on the X chromosome that has been isolated and characterized; therefore it is possible to identify the mutation in an individual with hemophilia and to supply a working copy of the gene to that individual's cells, for standard gene addition approaches. Even very small increases in clotting factor can improve a patient's clinical symptoms and provide an easily identifiable endpoint for study. Clotting factor can be produced in almost any tissue, as long as it gains access to the circulation, which allows for a variety of tissues to be targeted for transduction. Furthermore, there are large- and small-animal models of the disease, so that promising approaches can be assessed for efficacy before being tested in humans.
A major focus of our program is in vivo gene transfer using recombinant adeno-associated viral (rAAV) vectors. These have a number of advantages as gene delivery vehicles in the setting of genetic disease. They mediate long-term expression of the donated gene in a variety of animal models; they readily transduce a wide range of nondividing target cells, including muscle, liver, and cells of the central nervous system; and the vector DNA is stabilized primarily in an episomal form, reducing risks related to insertional mutagenesis. We have developed rAAV vectors to carry the DNA sequence that directs the production of the blood-clotting protein, factor IX, and more recently of the larger factor VIII cDNA sequence as well.
We have introduced these vectors into skeletal muscle and liver of mice and dogs with hemophilia B and have shown long-term correction of bleeding abnormalities. In our initial clinical studies, we introduced rAAV into leg muscles of adult patients with severe hemophilia B. Muscle biopsies of injected sites showed evidence of gene transfer and expression, even as long as three years after vector injection, but circulating levels of factor IX were generally less than 1 percent, failing to reach the desired target of 3–10 percent. More recent animal studies using intravascular delivery to skeletal muscle have produced circulating levels of factor IX of ~10 percent in hemophilia B dogs, a level of clotting factor that would be therapeutic if achieved in humans.
We have also been working to develop a liver-directed approach. Reengineering of the gene cassette within the vector has increased the efficiency of our vectors, resulting in higher levels of clotting factor from a given dose of vector. We produced circulating factor IX levels of 4–12 percent in hemophilia B dogs treated with vector doses lower than those administered in the clinical study in skeletal muscle. These levels have persisted for at least five years, with the experiment still ongoing.
After extensive safety studies in mice, rats, hemophilic dogs, and nonhuman primates, we began a phase I study of a liver-directed approach aimed at achieving gene transfer in hemophilia B using rAAV vectors. This dose-escalation study established a number of important findings. First, it showed that vector infusion into the liver was well-tolerated by all subjects, with no acute or long-term sequelae documented in up to six years of follow-up (observation ongoing). Second, it showed that we could achieve therapeutic levels of factor IX in humans by infusing the vector, and that the dose required for this was accurately predicted by studies in hemophilic dogs. Third, however, the study documented a finding that had not been seen in any animal models, specifically that factor IX levels, after remaining in a therapeutic range for a period of four weeks, slowly returned to baseline, with a concomitant asymptomatic rise in liver enzymes that resolved without medical intervention. Careful follow-up studies in a subsequent patient showed that the rise and fall in liver enzymes was accompanied by the expansion and then contraction of a population of AAV capsid-specific CD8+ T cells, but showed no evidence of a CD8+ T cell response to factor IX.
The most parsimonious hypothesis to account for these findings is a CD8+ T cell response to AAV capsid that recognized and destroyed transduced hepatocytes, leading to loss of factor IX expression and a transient rise in liver enzymes. Our working hypothesis to explain why this was observed in human subjects but not in animals is that prior exposure to AAV in humans, the only natural hosts for wild-type AAV, probably underlies the difference. Wild-type AAV is a Dependovirus that cannot replicate except in the presence of a helper virus such as adenovirus; thus, although AAV on its own fails to induce the inflammatory reactions needed for a maximal adaptive immune response, in the setting of a natural infection, it is likely that the helper virus, which causes activation of the innate immune response, promotes induction of CD8+ T cells directed to the antigens of both the helper and AAV. These give rise to a small pool of memory T cells, which, on reexposure to capsid, are activated and eliminate the capsid-harboring cells (transduced hepatocytes).
In response to these findings, we have initiated several new lines of investigation, including analysis of T cell responses to AAV capsid in normal and vector-infused human subjects; tracking of the intracellular fate of AAV capsid in the transduced cell, to determine the duration of display of capsid-derived peptides on the surface after transduction; development of an animal model of loss of transgene expression and transaminase elevation; and amendment of the clinical protocol to block the T cell–mediated destruction of transduced hepatocytes through short-term immunosuppression. Preclinical studies in a nonhuman primate model have shown that coadministration of AAV with an immunosuppressive regimen consisting of 10 weeks of sirolimus and mycophenolate mofetil (MMF) is safe and does not affect efficiency of AAV transduction of liver. This regimen will be tested in an upcoming clinical study.
Although translational studies with AAV remain a central part of our program, we are also pursuing several other innovative strategies for gene correction that are at earlier stages of development. Our studies are currently focused on hemophilia, but these strategies may be more widely applicable, and our location at a major children's hospital affords access to large populations of patients with genetic disorders. These strategies include the use of small-molecule, orally bioavailable drugs that can "read through" premature termination codons, and the use of zinc finger nucleases to effect gene correction in situ for mutations in the F9 gene.