Despite several set-backs, the development of suicide vector systems can still be a promising approach for gene therapy of HIV-1 infection [5, 41]. They can be used to reduce viral burden by killing infected, virus-producing cells [9, 21] and to target and eliminate the reservoir of latent HIV-1 . They might even be used to modify stem cells prior to their transplantation into the patient [42, 43]. These booby-trapped hematopoietic cells would commit suicide upon infection by HIV-1, thereby interrupting the viral replicative cycle and reducing viral load. However, for such an approach to function, several requirements have to be fulfilled. First, the suicide vector must be strongly HIV-1 dependent without compromising the rapid induction of the suicide gene by the viral proteins. Second, the suicide gene itself has to be fast and efficient in its induction of cell death, to ensure target cell elimination before infectious virus particles are generated. Third, the suicide gene should not be immunogenic. Fourth, the suicide system has to be delivered efficiently to the HIV-1 target cells. So far, no published system fulfills all of these necessities. In our proof-of-concept study, we addressed the first three requirements.
The suicide vector pLtBid(INS)2R contains the HIV-1 LTRs, thus, expression of tBid depends on HIV-1 Tat. However, the HIV-1 LTR alone would lead to unspecific expression of tBid as frequently observed using solely HIV-1 LTR-based vector systems [21, 32, 44]. The RRE present in the suicide vector pLtBid(INS)2R reduces this leakiness, because expression now depends on both Tat and Rev [9, 33, 34, 45]. In addition, the suicide vector pLtBid(INS)2R comprises two INS-containing regions of the HIV-1 gag gene. They increase the dependency on Rev, thereby reducing basal activity even further [40, 46, 47]. We tested the control vector pLRed(INS)2R in HeLa SS6 cells showing that the expression of dsRed is strongly dependent on the simultaneous presence of Tat and Rev. We observed no leakiness of the suicide vector pLtBid(INS)2R in the very sensitive cell line HeLa SS6 transfected only with pLtBid(INS)2R or co-transfected with pCMV-rev. The presence of Tat alone slightly induced the expression of tBid followed by induction of cell death in a small number of cells. However, only the presence of both Tat and Rev significantly enhanced the induction of cell death introduced by tBid showing that the system is not leaky and that it strongly depends on the HIV-1 proteins Tat and Rev. It is promising that no leakiness was observed in cells stably transfected with the control vector pLRed(INS)2R for more than 3 months. In addition, dsRed expression was still inducible by HIV-1 Tat and Rev after 3 months .
The human pro-apoptotic protein tBid is the effector molecule of our suicide vector pLtBid(INS)2R. tBid fulfills several qualifications of a suitable suicide protein: i) cell death is induced within hours after expression of tBid, ii) tBid is efficient in very low concentrations , iii) it is not immunogenic as, for instance, viral suicide proteins [11, 32], and iv) it has been successfully applied as suicide protein in a gene therapeutic approach where breast cancer cells were specifically killed by tBid in vitro . We observed rapid induction of cell death in HeLa SS6 cells within 24 hours after transfection of pCMV-tBid. Similar kinetics and efficacies had previously been shown in several cell lines [23, 25, 49] including Jurkat T lymphocytes .
We tested the cell death inducing properties of our suicide vector pLtBid(INS)2R with respect to HIV-1 using different experimental approaches. First, we co-transfected cells with the suicide vector pLtBid(INS)2R and the full-length HIV-1 clone pNL4-3/GFP. A few hours after transfection, before infectious HIV-1 particles were generated, the cells started to die. Here, no release of virus particles was observed as revealed by p24 ELISA. Second, cells were transfected with the suicide vector pLtBid(INS)2R-CD4 and infected with infectious HIV-1 NL4-3. Virus replication was strongly reduced and the amount of dead cells increased compared to HeLa SS6 cells transfected with the control vector pLRed(INS)2R-CD4. Although we have not measured tBid directly in dying cells due to the very low amounts of tBid sufficient to induce apoptosis [23, 49], which makes it very difficult to detect tBid with conventional detection methods, we are convinced that tBid expression induced cell death. In all our experiments, we used the isogenic vector pLRed(INS)2R as control and we never observed any indication of increased cell death induced by dsRed.
In contrast to the co-transfection experiments using pNL4-3/GFP, the inhibition of virus particle production was incomplete when pLtBid(INS)2R-CD4 transfected HeLa SS6 cells were infected with HIV-1 NL4-3. This is not due to the replacement of nef by gfp in pNL4-3/GFP, because we obtained similar results using pNL4-3 (data not shown). Several possible reasons could explain this observation. Residual replication might be due to the long CD4 half-life of approximately 20-24 hours on the cell surface [51, 52] when rapidly dividing cells lose the transiently transfected suicide vector pLtBid(INS)2R-CD4. We have observed that transiently transfected plasmids were not detectable anymore after a few cell divisions in HeLa SS6 cells (unpublished observations). This would lead to the presence of cells which do not contain the suicide vector pLtBid(INS)2R-CD4, but still exhibit sufficient amounts of CD4 to promote HIV-1 entry and replication. It is known for a T cell-line adapted HIV-1 strain like NL4-3 that low CD4 expression does not impair virus replication . It has also been described that cells lacking the CCR5 or CXCR4 receptor can be rendered CCR5+ or CXCR4+ by uptake of membrane-derived microparticles from CCR5- or CXCR4-positive cells . There is evidence that CD4+ cells undergoing programmed cell death shed microparticles that carry CD4 , offering another possibility of how co-cultured cells not expressing the CD4 antigen themselves could become infected after CD4 uptake.
The incomplete replication suppression might also be caused by differences in the levels and kinetics of HIV-1 Tat and Rev expression in the more physiological conditions using infectious virus particles [56, 57]. A critical threshold level of Rev is required for highly productive HIV-1 infection . In contrast to the HIV-1 genome, pLtBid(INS)2R-CD4 contains a duplication of the INS region. In the absence of Rev, gene expression of gag/pol and env mRNA is not completely inhibited in the context of wild-type HIV-1 , but this INS duplication results in enhanced suppression of gene expression. Thus, it might be that a threshold level of Rev for sufficient expression of tBid was not reached in all cells by the infection. To address this concern, we are currently focusing on improving the pro-apoptotic properties of tBid by (i) introducing mutations that eliminate ubiquitin acceptor sites and thereby improve tBid's intracellular stability  and by (ii) modifying its N-terminal sequence to allow myristoylation of tBid, which promotes targeting to mitochondria and enhances pro-apoptotic activity in cell culture . A similar low efficiency was recently described for replication-competent wild-type HIV-1 virions compared to single-cycle VSV-G pseudotyped HIV-1 virus particles in another investigation of a potential Rev-dependent suicide vector .