which was then transfected into human C17 cells (293 cells which constitutively express EBNA1). Pools of hygromycin-resistant C17 clones containing the shuttle vector (hygromycin serves as a selective marker for p220.2) were isolated and grown up. On average the clones contained 50–100 copies of the shuttle vector per cell. The cloned cells were infected with AAV and after 48 hr plasmid DNA was isolated and transfected into E. coli. The only colonies to form were those containing the selectable marker of the shuttle vector (ampR which is carried by p220.2). The fraction of such colonies which hybridized to an AAV probe was considered a reflection of the frequency with which AAV had integrated into the shuttle vector. Data are summarized in Fig. 1 (24). AAV did integrate into the shuttle vector which contained the entire 8.2 kb of AAVS1. By sequential deletion analysis it was possible to map the sequences required to direct site-specific integration to the first 510 nt of AAVS1. Thus, it was possible to conclude that AAV site-specific integration was determined by the DNA sequence on 19 q and that the critical sequence was contained within the first 510 bases of the AAVS1 sequence.
To delineate the critical signal sequences within the first 510 bases of AAVS1, a brief review of the molecular genetics and biology of AAV replication is required. Within the 4.7-kb genome there are two ORFs; the one in the right half of the genome encodes the three structural proteins; the ORF in the left half of the genome encodes four regulatory proteins, Reps 78, 68, 52, and 40, with overlapping amino acid sequences. (The Rep designation is used because a frame shift mutation anywhere within the ORF blocks DNA replication) (25). There are two promoters (at map positions 5 and 19) in the left half of the genome and both spliced and unspliced forms of the two transcripts are translated to synthesize the four Rep proteins. Reps 78 and 68 have essentially identical phenotypes and are involved in all phases of the AAV life cycle. These phenotypes depend on the physiological state of the cell. In the absence of helper virus (i.e., the nonpermissive state) Rep 68/78 represses AAV gene expression and inhibits viral DNA synthesis. It is required for site-specific integration and affects expression of a number of cellular genes, most by down regulation (26). In the presence of helper virus (i.e., the permissive state) Rep 68/78 is required for AAV gene expression and transactivates expression of the structural proteins; it is also required for viral DNA replication and rescue of the integrated viral genome. Interestingly, it inhibits expression of the helper adenovirus early genes (M.A.Labow and K.I.B., unpublished data).
The AAV genome contains an ITR of 145 nt (Fig. 2). The first 125 nt constitute an overall palindrome interrupted by two smaller internal palindromes of 21 nt, one immediately on either side of the overall axis of symmetry. When folded on itself to optimize potential base pairing, the palindromic sequence forms a T-shaped structure. The long stem of the T-shaped structure contains an RBS. When Rep binds to the ITR it interacts with at least one of the cross arms of the T (or small internal palindrome) and can make a site-specific nick between nt 124 and 125 (27). After nicking, Rep is covalently bound to the 5′ side of the nick and can function as a helicase (28). The ITR is the cis-active signal in the nonpermissive state for the negative regulation of gene expression and DNA replication. In the permissive state the ITR enhances gene expression, serves as the ori for DNA replication, and is required for rescue of the viral genome from the integrated state.
Thus, it was of interest to note that the 510 nt of AAVS1 sufficient to direct site-specific integration contained both an RBS and TRS in the appropriate orientation with a comparable spacing between them. A third signal of potential interest was a hexanucleotide homologous to an enhancer of meiotic gene conversion (M26) in fission yeast (29). This sequence was also present at approximately the same position (relative to RBS and TRS) at one end of the AAV genome. Biochemical experiments had shown that Rep 68 could bind at the RBS in AAVS1 and that oligomeric complexes of Rep could link the ITR of AAV to the corresponding sequences in AAVS1 (30). It was also shown in vitro that bound Rep could nick the sequence