In the case of HIV, several sequences have been implicated by deletion studies in the encapsidation and dimerization of viral RNA. In contrast to MLV, where the Ψ sequence, located in the 5′-untranslated leader downstream of the major splice donor site, strongly contributes to RNA packaging, the role of the corresponding region in the HIV genome seems more to discriminate genomic from spliced transcripts than to promote efficient encapsidation (28–30). Additional, and possibly more important, sequences have been identified in the transcribed long terminal repeats and 5′ leader sequences upstream of the major splice donor site (31–35). A major caveat, however, is that reconstitution of the packaging function has not yet been reported by juxtaposition of the identified sequence(s) to heterologous RNA (36). Thus, the packaging signal of HIV is either highly sequence-specific or, more likely, multipartite and distributed over a rather large stretch of its 5′ sequence (37).
Several features of the HIV-derived packaging plasmid described here prevent its transfer to the target cells. The combined modifications of the 5′ end delete or disrupt all structural motifs to date implicated in RNA encapsidation and dimerization, with the possible exception of the 5′ portion of the gag sequence (30, 38, 39). Recently, McBride and Panganiban (35) reported the encapsidation efficiency of HIV-1 transcripts carrying deletions of the 5′ leader sequence; relative to the wild-type counterpart, it was reduced to <0.1 for transcripts derived from a construct comparable to pCMVΔR8.2 and to <0.02 in the presence of competing wild-type RNA (35). Furthermore, the deletion of both long terminal repeats and of the primer binding site from the packaging plasmid would prevent reverse transcription and integration of any encapsidated transcript not recombined with the vector RNA. The transducing vector, on the other hand, is endowed with a full complement of the cis-acting sequences not identified until now, which allows its proficient transfer to the target cell.
It is well recognized that the retroviral infection is an inefficient process. Once the content of virions is delivered inside the target cell, uncoating, reverse transcription, interaction with cytoplasmic chaperones and the nuclear import machinery, and maturation to an integration-competent complex take place. These events, the mechanism of which is still poorly understood, can result in degradation and arrest at a stable intermediate, as well as integration of the viral genome (40). Partial reverse transcripts have been detected in HIV and MLV virions (41–43). Recently, it was shown that viral DNA synthesis can be promoted inside intact HIV-1 particles by exposure to dNTPs and magnesium chloride and that the efficiency of the reaction can be increased by the addition of the polyamine spermine and spermidine (25, 44, 45). The resulting HIV-1 virions exhibit an increased infectivity in primary T lymphocytes infected before activation, a setting in which reverse transcription was previously demonstrated to be a rate-limiting step (46–48). The stimulation of reverse transcription inside virions was also shown to increase the transduction efficiency of MLV-based retroviral vectors in dividing targets (49). Here, we find that it significantly augments the efficacy of gene transfer mediated by the lentiviral vector. This effect was most pronounced in nondividing cells and could also be observed in vivo. Performing such in vitro reverse transcription reactions before injecting the vector may be critical for some nonproliferating targets that maintain low cytoplasmic pools of dNTPs (50, 51).
The crucial advantage of the lentiviral vector is its integration in the genome of nondividing cells. This was proven here by the dependence of the transduction on the incorporation of a functional integrase in the vector. At least in the case of the brain, the only tissue studied so far, this provides for long-term sustained expression of the transgene. No decrease in the extent of β-gal immunoreactivity was observed even 3 months after a single vector administration. Given the recent report of predominant transgene-directed immune responses in animals transduced with adenoviral vectors (52), it remains to be determined whether the brain represents an immune haven, as long suspected, or whether the adenoviral proteins expressed by the transduced cells played a critical adjuvant role.
Retroviruses are thought to select, through poorly understood mechanisms, active chromatin sites for the integration of their genome (40). This may explain why gene delivery methods based on MLV-derived vectors often suffer from the transcriptional shutoff of the transgene, as was observed in this study, when the transduced cells return to a nonproliferating status and presumably revise their pattern of chromatin expression. The ability of the lentiviral vector to integrate in nondividing cells may allow for the selection of stably open chromatin sites, thus ensuring against the transcriptional silencing of the transgene.
The high prevalence of neurons observed among the transduced cell types in the brain may be due both to the neurotropism of the envelope of the VSV, a rhabdovirus (53), and to a preferential expression of the hCMV promoter in neurons, as recently observed with transgenic animals (54). It may also reflect preferential long-term expression in nondividing cells, for the reasons discussed above.
A major issue now concerns the biosafety of the lentiviral vector. The novel feature of the packaging plasmid described in this paper precludes the generation of wild-type HIV viruses, even by unlikely rearrangement and recombination events, given the actual absence of most of HIV env sequences in all three plasmids. In the previously described plasmid pCMVΔR9, the env reading frame was blocked by insertion of a linker containing multiple stop codons. The use of a separate plasmid encoding a heterologous envelope makes it extremely unlikely that a replication-competent recombinant be generated. This would require multiple recombination events between different plasmids and/or endogenous retroviral sequences, including recombination between nonhomologous sequences. Careful scrutiny and improvement of the vector-producing system, including evaluation of the minimal set of viral genes required for efficient packaging of the vector and generation of stable packaging systems better amenable to monitoring, are now required.
We are extremely grateful to R.Pomerantz and colleagues for communicating their results before publication. We thank members of the Verma, Gage, and Trono laboratories for helpful suggestions. L.N. was supported by the American-Italian Cancer Foundation, and U.B. was supported by the Deutsche Forschungsgemeinschaft (DFG). This work is supported by grants from the National Institutes of Health and the American Cancer Society (I.M.V.), U.S. Public Health Service Grant AI37510 (D.T.); National Institutes of Health Grants AG10435 and 08514 and Hollfelder Foundation (F.H.G.); and the H.N. and Frances Berger Foundation (D.T., I.M.V., and F.H.G.). I.M.V. is an American Cancer Society Professor of Molecular Biology and D.T. is a Pew Scholar.
1. Mulligan, R.C. (1993) Science 260, 926–932.
2. Verma, I. (1994) Mol. Med. 1, 2–3.
3. Crystal, R.G. (1995) Science 270, 404–410.
4. Leiden, J.M. (1995) New Engl. J. Med. 333, 871–873.
5. Yang, Y., Li, Q., Ertl, J. & Wilson, J.M. (1994) Proc. Natl. Acad. Sci. USA 91, 4407–4411.
6. Yang, Y., Ertl, J. & Wilson, J.M. (1994) Immunity 1, 433–442.
7. Dai, Y., Schwarz, E.M., Gu, D., Zhang, W.-W., Sarvetnick, N. & Verma, I.M. (1995) Proc. Natl. Acad. Sci. USA 92, 1401–1405.
8. Knowles, R.M., Hohneker, K.W., Zhou, Z., Olsen, J.C., Noah, T.L., Hu, P.-C., Leigh, M., Engelhardt, J.F., Edwards, L.J., Jones, K.R., Grossman, M., Wilson, J.M., Johnson, L.G. & Boucher, R.C. (1995) New Engl. J. Med. 333, 823–831.
9. Kremer, E.J. & Perricaudet, M. (1995) Br. Med. Bull. 51, 31–44.
10. Miller, A.D., Miller, D.G., Garcia, J.V. & Lynch, C.M. (1993) Methods Enzymol. 217, 581–599.