A model for the integration process has been developed (32). Because of the involvement of Rep proteins, RBS and TRS, it appears likely that viral DNA replication and localized DNA replication within AAVS1 are involved in integration. AAV DNA replication involves a single strand displacement mechanism (Fig. 4) (for a review, see ref. 34). A major feature of AAV replication is the ability of the elongating strand to switch templates (35). Replication initiates from the ITR in the folded state which serves as both the ori and the primer. When the template strand is fully copied, the 3′ end of the newly synthesized strand can fold on itself and begin synthesis of a second new strand, this time using the first daughter strand as the new template (i.e., by switching templates). It is suggested that the intrinsic tendency of a Rep-mediated replication complex to switch template strands is also the underlying mechanism in the generation of aberrant AAV DNA particles and defective interfering particles.
If the first hairpin structure created by the initial priming event has not been resolved, continuing synthesis will lead to a double stranded, dimeric form of AAV DNA (in Fig. 4IIA). Resolution of the hairpin structures is achieved by Rep cleavage at TRS (Fig. 4IIB). This leads to transfer of the original hairpin sequence from parental to daughter strand and creates a 3′ OH to serve as a primer for repair of the 5′ end of the parental strand.
In vivo, AAV DNA replication requires a co-infection with a helper virus, usually adenovirus. In vitro, it is possible to observe replication of full-length AAV DNA using an extract from uninfected HeLa cells (i.e., from nonpermissive cells) which has been supplemented with purified Rep 68 (36). A significant question was whether the only important adenovirus helper effect on AAV DNA replication was to allow synthesis of sufficient amounts of Rep. The fact that in vitro AAV DNA replication is greatly enhanced by the substitution of extracts from adenovirus-infected cells for those from uninfected cells proves this hypothesis incorrect. The major consequence of using the extract from adenovirus-infected cells is to enhance the ability of the elongating DNA strand to remain on the original template. Use of uninfected cell extract leads to premature strand switching and the consequent interruption of the normal replication process with the synthesis of defective DNA molecules (35). We believe that the enhanced probability of strand switching during DNA synthesis in the absence of a helper virus coinfection plays a major role in the integration process.
A model for the integration process must take into account the following properties. (i) Involvement of RBS and TRS. (ii) Rearrangement of the AAVS1 sequences at one junction. (iii) Presence of head-to-tail AAV junctions, (iv) Despite the requirements for very distinct integration signals (RBS and TRS) a model must account for the observation that integration junctions observed are scattered within ca. 1 kb of AAVS1 downstream of RBS and TRS. A simplified model which can account for these features is shown in Fig. 5. An oligomeric complex of Rep binds to the RBS on AAVS1 and to the RBS in the AAV ITR, thus linking a circularized duplex AAV molecule to AAVS1. This represents a protein (Rep) mediated alignment of the recombination partners AAV and AAVS1, initiating the nonhomologous recombination event observed. Rep then introduces a nick into the AAVS1 TRS. DNA synthesis initiates, displacing a single strand of AAVS1. The extension of replication determines the location of a junction with AAV subsequently formed (see requirement iv mentioned above). It should be noted that the displaced single strand is circular because Rep is covalently bound to the 5′ end and presumed to be still bound to the RBS. After limited extension the elongating strand switches to the displaced single strand as the template; note that copying of the displaced circular AAVS1 sequence leads to inversion of the sequence. When it reaches the end of the displaced strand (close to the RBS as observed in the shuttle vector model system), the elongating strand again switches templates, now onto the circular AAV DNA. After synthesis proceeds on the AAV template, the elongating strand reaches RBS where Rep is bound and the strand again switches to a new template on AAVS1 (alternatively onto p220.2 in the shuttle vector system). Eventually the single-strand gap involving the inserted AAV sequence and the inverted AAVS1 sequence is repaired. Undoubtedly, this model is simplified, but we believe that it is consistent with many of the features observed in AAV integration.
The proper conjunction of RBS and TRS required for integration is present only once in the data concerning the human genome in GenBank, likely explaining the apparent presence of only a single site for specific AAV integration. However, RBS has been noted at multiple sites in the human genome; in 14/15 cases analyzed, it appears in the 5′-untranslated regions of characterized genes; therefore, indicating a cellular counterpart of the Rep protein with possible regulatory functions, which also recognizes RBS. In addition it seems likely that AAV has evolved to take advantage of one copy of this recognition signal sequence.
Our current knowledge of the requirements for site-specific AAV DNA integration, together with our proposed model for an integration mechanism, may help in the design of improved AAV-based vectors for gene therapy. At this point it can be concluded that the Rep protein is an absolute requirement for the site specificity of AAV DNA integration. Finally, we propose that the full AAV ITR may not be necessary for targeted integration. Rather, it is possible that integrity of the ITR is only required for efficient rescue of integrated proviruses, a function not necessary, or even desirable, for stable, long term gene delivery.
This work was supported by a grant from the National Institute of Allergy and Infectious Diseases (AI22251) and by a grant from the National Institute of General Medical Sciences (GM50032). R.M.L. was supported in part by a fellowship from the Norman and Rosita Winston Foundation.
1. Blacklow, N.R., Hoggan, M.D., Kapikian, A.Z., Austin, J.B. & Rowe, W.P. (1968) Am. J. Epidemiol. 88, 368–378.
2. Berns, K.I., Pinkerton, T.C., Thomas, G.F. & Hoggan, M.D. (1975) Virology 68, 556–560.
3. Cheung, A.K., Hoggan, M.D., Hauswirth, W.W. & Berns, K.I. (1980) J. Virol. 33, 739–748.
4. Podsakoff, G., Wong, K.K., Jr., & Chatterjee, S. (1994) J. Virol. 68, 5656–5666.
5. Kotin, R.M. & Berns, K.I. (1989) Virology 170, 460–467.
6. Kotin, R.M., Siniscalco, M., Samulski, R.J., Zhu, X.D., Hunter, L., Laughlin, C.A., McLaughlin, S., Muzyczka, N., Rocchi, M. & Berns, K.I. (1990) Proc. Natl. Acad. Sci. USA 87, 2211–2215.
7. Kotin, R.M., Linden, R.M. & Berns, K.I. (1992) EMBO J. 11, 5071–5078.
8. Kotin, R.M., Menninger, J.C., Ward, D.C. & Berns, K.I. (1991) Genomics 10, 831–834.
9. Samulski, R.J., Zhu, X., Xiao, X., Brook, J.D., Housman, D.E., Epstein, N. & Hunter, L.A. (1991) EMBO J. 10, 3941–3950.
10. Zhou, S.Z., Cooper, S., Kang, L.Y., Ruggieri, L., Heimfeld, S., Srivastava, A. & Broxmeyer, H.E. (1994) J. Exp. Med. 179, 1867–1875.
11. Kaplitt, M.G., Leone, P., Samulski, R.J., Xiao, X., Pfaff, D.W., O’Malley, K.L. & During, M.J. (1994) Nat. Genet. 8, 148–154.
12. Ali, R.R., Reichel, M.B., Thrasher, A.J., Levinski, R.J., Kinnon, C., Kanuga, N., Hunt, D.M. & Bhattacharya, S.S. (1996) Hum. Mol. Genet. 5, 591–594.
13. Atchison, R.W., Casto, B.C. & Hammon, W. (1965) Science 149, 754–756.
14. Hoggan, M.D., Blacklow, N.R. & Rowe, W.P. (1966) Proc. Natl. Acad. Sci. USA 55, 1467–1471.
15. Buller, R.M., Janik, J.E., Sebring, E.D. & Rose, J.A. (1981) J. Virol. 40, 241–247.
16. Weindler, F.W. & Heilbronn, R. (1991) J. Virol. 65, 2476–2483.