Gene Therapy (2008) 15, 817–822 & 2008 Nature Publishing Group All rights reserved 0969-7128/08 $30.00 www.nature.com/gt REVIEW Adeno-associated virus integration: virus versus vector RH Smith Laboratory of Biochemical Genetics, National Heart, Lung, and Blood Institute, Bethesda, MD, USA Although a large percentage of the world population is major impetus for the development of recombinant AAV seropositive for exposure to various strains of adeno- vectors, which typically lack all AAV coding sequences. It associated virus (AAV), a human parvovirus, AAV has never was soon realized, however, that expression of at least one been identified as an etiologic agent of human disease. species of the virally encoded initiator proteins, Rep78 or Most likely contributing to the pronounced lack of pathogeni- Rep68, is necessary for targeted integration of AAV-derived city is the fact that AAV is a naturally ‘defective’ virus that DNA constructs to occur. This article will present a chronolo- requires a helper virus for productive replication of its gical outline of studies characterizing site-specific integration of genome. Another unusual aspect of wild-type AAV biology wild-type AAV sequences and the quasi-random target site is the ability of the virus to establish latent infection by selection observed with recombinant AAV vectors. preferential integration of its genome into a specific locus Gene Therapy (2008) 15, 817–822; doi:10.1038/gt.2008.55; of human chromosome 19. Site-specific integration was a published online 10 April 2008 Keywords: AAV; adeno-associated virus; integration Introduction gations into the nature of the AAV genome in latently infected cells. In 1965, Atchison et al.1 published a paper reporting the characterization of a newly discovered viral agent, which they named adeno-associated virus (AAV). AAV parti- cles were first observed by these researchers in electron Integration of wild-type AAV micrographs prepared from a stock of simian adenovirus Early work concerning the characterization of AAV that had been serially passaged in primary cultures of latency was greatly facilitated by the establishment of rhesus monkey kidney cells. Atchison et al. were able to continuous cell lines harboring AAV in a quiescent state. separate the approximately 24-nm AAV particles from Berns et al.2 reported that a human bone marrow- the larger (approximately 80-nm) adenovirus virions by derived, fibroblast-like cell line (Ruddle’s Detroit 6 cells) ultrafiltration. Upon isolation, it was noted that the that had been infected with 250 infectious units per cell partially purified AAV virions failed to replicate auto- of AAV serotype 2 (AAV-2) maintained viral sequences in nomously when inoculated onto primary cells; however, a latent state for at least 47 passages. Infection of Detroit AAV propagated readily when the cell cultures were 6 cells with significantly lower multiplicities of AAV coinfected with both AAV and its adenovirus partner. resulted in progressive loss of AAV-positive cells from Thus, AAV appeared to be a defective satellite virus that the total cell population during serial passage. Upon was dependent upon a coinfecting helper virus for cloning the infected cells, Berns et al. observed that efficient replication. In an important extension of these 2 29% of the clones were able to produce AAV after findings, Hoggan and colleagues described a collo- superinfection with adenovirus, suggesting that the quium report in which they observed that a significant establishment of AAV latency was fairly efficient at high number of human and monkey primary kidney cell lots multiplicities of infection. Solution-based hybridization harbored AAV in a quiescent (that is, latent) state analyses (that is, reassociation kinetics analysis) of and that infectious AAV particles could be recovered AAV-positive Detroit 6 cell subclones indicated a copy from the latently infected cells after superinfection number of three to five AAV genome equivalents per with adenovirus. These observations prompted investi- diploid cellular genome. To determine whether latently infected Detroit 6 cells harbored AAV sequences episo- mally or in an integrated state, Cheung et al.3 prepared Correspondence: Dr RH Smith, Laboratory of Biochemical Genetics, Hirt extracts of cellular DNA from a latently infected National, Heart, Lung, and Blood Institute, Building 10, Room Detroit 6 cell subclone and characterized the DNA 7N264, Bethesda, MD 20892, USA. E-mail: [email protected] fractions by restriction enzyme digestion and Southern Received 25 February 2008; accepted 27 February 2008; published blot analysis. It was determined that wild-type AAV online 10 April 2008 sequences were integrated into the host cell genome as AAV integration RH Smith 818 tandem, head-to-tail repeats linked to genomic DNA transcribed chromatin. A putative cyclic AMP response sequences by the viral inverted terminal repeat elements element and several recognition sequences for upstream (or ITRs). binding factor 1 were also observed within the CpG island. The last 400 basepairs of the sequenced region of AAVS1 contained a chromosome 19-abundant mini- Wild-type AAV preferentially integrates within satellite array. PCR analysis of reverse-transcribed RNA a specific region of human chromosome 19 (RT-PCR) indicated that the region of AAVS1 down- In an important contribution to our understanding of stream of the CpG island may be transcribed, although at integration site selection by wild-type AAV, Kotin and apparently low (or possibly tissue-specific) levels as Berns4 used a bacteriophage lambda-based genomic northern blot analysis did not identify mRNA transcripts DNA library derived from latently infected Detroit 6 in the cells examined. A DNase I hypersensitive site, cells to isolate cellular sequences flanking an AAV designated DHS-S1, has been mapped to the CpG island provirus. The cellular flanking sequences were used as within AAVS1.8 DHS-S1 demonstrates enhancer-like probes in Southern blot analyses of DNA extracted from properties when linked to a reporter gene in transient 22 independently derived cell lines latently infected with transfection assays and has been shown to function as a AAV,5 including clonal cell lines derived from Detroit 6, chromosomal insulator sequence.8,9 HeLa and KB cells. Sixty-eight percent of the indepen- dently derived clones displayed AAV-induced altera- AAVS1 and integration tions in the electrophoretic mobility of DNA restriction To map cis-acting elements of AAVS1 essential for site- fragments derived from the same genomic locus specific AAV integration, Giraud et al.10 utilized an (as identified by the flanking sequence probes). In Epstein–Barr virus (EBV)-derived shuttle plasmid that addition, more than half of the affected DNA bands can be stably maintained as an episome in eukaryotic were also positive for AAV genomic sequences by cells and subsequently recovered as a bacterial plasmid. Southern blot analysis. These findings suggested that in Various portions of the AAVS1 locus were cloned into the latently infected cells, wild-type AAV utilizes a common EBV shuttle plasmid and individual constructs were integration site to establish latent infection. Hybridiza- used to establish shuttle plasmid-bearing cell lines. The tion analysis of a panel of rodent–human somatic cell various cell lines were infected with AAV-2 at a multi- hybrids mapped the common AAV integration locus to plicity of 20 infectious units per cell. At 48 h post- human chromosome 19. Using an alternative approach to infection, extrachromosomal DNA was isolated and used identify AAV integration sites, Samulski et al.6 estab- to transform bacteria. Integration events were character- lished latently infected cell lines containing integrated ized by filter-based hybridization analysis using chimeric AAV genomes bearing DNA-binding sites AAV-specific probes. DNA sequences within the first recognized by the bacteriophage lambda repressor 510 bp of AAVS1 were found to be both necessary and protein. Cellular sequences flanking AAV integrants sufficient for AAV site-specific integration. In addition to were obtained by restriction digestion of genomic DNA containing a large portion of the AAVS1 CpG island, this extracted from latently infected cells and enrichment of region was found to contain a tandem GCTC repeat AAV-positive sequences by a filter-binding technique. element that can serve as a binding site for the large Rep Fragments containing cellular flanking sequences were proteins of AAV11 (Rep78 and Rep68), as well as a Rep- eluted from the filters, cloned and used to probe specific nicking site, known as a terminal resolution site Southern blots of DNA extracted from at least eight or trs. The Rep78/68 proteins are AAV-encoded non- independently derived cell lines harboring latent AAV structural proteins, which possess DNA-binding, heli- genomes. In each case, the cellular flanking sequences case and site-specific endonuclease activities that are hybridized to at least one restriction fragment that was essential for AAV DNA replication.12,13 Using the EBV also detected by an AAV-specific probe, indicating that in shuttle plasmid system, Linden et al.14 finely mapped the each cell line AAV had utilized a common genomic locus sequences of AAVS1 necessary for site-specific integra- for integration. Further hybridization analysis indicated tion to an approximately 100 bp AAVS1 region contain- that the common AAV integration site was conserved in ing the Rep-binding and nicking sites. Mutation of monkey cells,