Chromosomal Integration of Adenoviral Vector DNA In Vivo

So far there has been no report of any clinical or preclinical evidence for chromosomal vector integration following adenovirus (Ad) vector-mediated gene transfer in vivo. We used liver gene transfer with high-capacity Ad vectors in the FAH^Δexon5 mouse model to analyze homologous and heterologous recombination events between vector and chromosomal DNA. Intravenous injection of Ad vectors either expressing a fumarylacetoacetate hydrolase (FAH) cDNA or carrying part of the FAH genomic locus resulted in liver nodules of FAH-expressing hepatocytes, demonstrating chromosomal vector integration. Analysis of junctions between vector and chromosomal DNA following heterologous recombination indicated integration of the vector genome through its termini. Heterologous recombination occurred with a median frequency of 6.72 × 10⁻⁵ per transduced hepatocyte, while homologous recombination occurred more rarely with a median frequency of 3.88 × 10⁻⁷. This study has established quantitative and qualitative data on recombination of adenoviral vector DNA with genomic DNA in vivo, contributing to a risk-benefit assessment of the biosafety of Ad vector-mediated gene transfer.Recombinant adenovirus (Ad) vectors are under clinical development for different applications, including tumor therapy, vaccination, and gene therapy. Today, the largest number of clinical gene transfer trials has been based on Ad vectors (http://www.wiley.co.uk/genmed/clinical). Several Ad vectors are in phase III clinical trials, and two products have already been approved in China. The occurrence of malignancies due to retroviral integration and oncogene activation in a clinical trial for the treatment of children with SCID-X1 (10) has pointed to the need for a thorough preclinical evaluation of potential genotoxic effects due to chromosomal integration of gene transfer vectors as an important part of the overall risk-benefit analysis. Detailed information on genotoxicity following gene transfer is available for vectors derived from viruses of the Retroviridae and Parvoviridae families (2, 20, 23, 26, 46). Between 60 and 75% of integrations of retrovirus, lentivirus, or adeno-associated virus (AAV)-based vectors take place in or close to genes.Chromosomal integration of Ad vector DNA following gene transfer in cell culture has been analyzed in only a few studies, and even less is known about Ad vector integration in vivo. Since the life cycle of wild-type adenovirus is extrachromosomal, Ad vectors are perceived to be nonintegrating vectors. However, in earlier studies it was observed that injection of hamsters with wild-type adenovirus type 12 (Ad12) resulted in tumor formation due to chromosomal integration of virus DNA and expression of the E1A/E1B oncoproteins (33). Recent in vitro studies with Ad vectors with E1 deletions have demonstrated the occurrence of vector integration following transduction of transformed cell lines and primary cells, with the frequencies of homologous and heterologous recombination being between 10⁻³ and 10⁻⁶ and between 10⁻³ and 10⁻⁵ per cell, respectively, depending on the conditions used (12, 14, 28, 36, 37, 42, 43). Since clinical gene transfer trials, including prophylactic vaccination of healthy volunteers against infectious diseases, are performed with large amounts of vector (in general, between 10¹⁰ and 10¹³ particles), it is possible that substantial integration of adenoviral vector DNA might also occur in vivo even if integration rates were low. However, so far there has been no attempt to experimentally address the issue of Ad vector integration in vivo. We used the FAH^Δexon5 mouse model (8) of tyrosinemia type I (MIM 27670) to analyze potential homologous and heterologous recombination events between Ad vector DNA and chromosomal DNA in vivo. Tyrosinemia type I is caused by the lack of fumarylacetoacetate hydrolase, an enzyme that is involved in the tyrosine degradation pathway and that converts fumarylacetoacetate into fumaric acid and acetoacetic acid in hepatocytes (38). Loss of fumarylacetoacetate hydrolase (FAH) activity in hepatocytes results in the accumulation of toxic and mutagenic metabolites in a cell-autonomous fashion, leading after birth to an acute hepatopathy and later in life to a chronic hepatopathy. Liver damage can be prevented both in humans and in FAH-deficient animals by the administration of 2-(2-nitro-4-trifluoromethylbenzoyl)-1,3-cyclohexanedione (NTBC), which blocks the tyrosine degradation pathway by inhibiting 4-hydroxphenyl pyruvate dioxygenase, thereby preventing the accumulation of the toxic compounds. The murine FAH gene is located on chromosome 7, contains 14 exons, and spans 20.5 kb.The autosomal recessive FAH^Δexon5 mouse model, in which exon 5 is disrupted by the insertion of a NeoR gene (8), has been a useful system to analyze chromosomal integration of AAV, retrovirus, Sleeping Beauty transposon, and plasmid DNA in hepatocytes (13, 25, 27, 31). Similar to human tyrosinemia type I patients with spontaneous reversions of point mutations (18), FAH-expressing hepatocytes have a strong growth advantage over FAH^−/− hepatocytes, and the developing nodules, consisting of FAH-positive FAH⁺] hepatocytes, can be easily distinguished in an environment of FAH^−/− hepatocytes. Following injection of an FAH-expressing Ad vector with the E1 deletion (30) into FAH^−/− mice, the development of FAH⁺ nodules in the livers of the experimental animals was observed, suggesting potential chromosomal integration of vector DNA. Since transgene expression from vectors with the E1 deletion is transient, in part due to viral toxicity and an immune response directed to viral proteins expressed from the vector, integration events and their characterization were not possible. We reasoned that the use of high-capacity Ad (HC-Ad) vectors (also called “helper-dependent” or “gutless” Ad vectors) (41) not expressing any viral proteins would allow reliable data on Ad vector integration in vivo to be obtained.