HIV Articles  
Back 
 
 
HIV Cure? HIV-1 Proviral DNA Excision Using an Evolved Recombinase
 
 
  Science June 29, 2007
 
Indrani Sarkar,1* Ilona Hauber,2* Joachim Hauber,2 Frank Buchholz1
 
HIV-1 integrates into the host chromosome and persists as a provirus flanked by long terminal repeats (LTRs). To date, treatment regimens primarily target the virus enzymes or virus-cell fusion, but not the integrated provirus. We report here the substrate-linked protein evolution of a tailored recombinase that recognizes an asymmetric sequence within an HIV-1 LTR. This evolved recombinase efficiently excised integrated HIV proviral DNA from the genome of infected cells. Although a long way from use in the clinic, we speculate that this type of technology might be adapted in future antiretroviral therapies, among other possible uses.
 
1 Max-Planck-Institute for Molecular Cell Biology and Genetics,
Pfotenhauerstrasse 108, D-01307 Dresden, Germany.
2 Heinrich-Pette-Institute for Experimental Virology and Immunology,
Martinistrasse 52, D-20251 Hamburg, Germany.
 
Current highly active antiretroviral therapy (HAART) targeting the viral reverse transcriptase, protease, and virus-host fusion (1, 2) has transformed HIV-1 infection into a chronic illness and curtailed the morbidity of infected individuals. Furthermore, new viral targets and novel inhibition strategies are being tested for improved control of HIV-1 (3-7). However, the current treatment strategies only suppress the viral life cycle without eradicating the infection, and new strains of HIV-1 are emerging that are resistant to suppressive treatments (8). An attractive alternative would be the specific eradication of the HIV-1 provirus.
 
Mutational and structural analyses have improved the understanding of the intricate enzymatic mechanism of site-specific recombinases and have permitted the identification of variants with altered properties [reviewed in (9) and (10)]. In particular, Cre recombinase, which has found widespread use in mouse genetics (11), has been intensively studied (12), and Cre target specificity can be altered to recognize moderately altered DNA target sites (13-15). These studies raise the possibility that new site-specific recombinases can be generated via directed evolution, which recombine more divergent target sites. More specifically, our aim was to evolve a recombinase that would recombine a sequence present within an HIV-1 LTR. Because Cre can efficiently remove genomic sequences that are flanked by two loxP sites (16), we and others have predicted that an evolved recombinase that would recombine a sequence present in the 5'-LTR and 3'-LTR of an integrated provirus could excise viral sequences from the genome (13, 17-19).
 
To start the evolutionary process, we first scanned HIV-1 LTR sequences for a sequence with similarity to the canonical loxP site. The chosen sequence belongs to the LTR of the primary HIV-1 strain TZB0003 (20) and is part of its modulatory U3 region. The selected loxLTR site is a 34-bp asymmetric sequence that has 50% sequence similarity to loxP, with four mismatches in the left element, six in the right element, and a completely different spacer (Fig. 1A). This sequence was examined in substrate-linked protein evolution in Escherichia coli (13). The loxLTR sequence was inserted into the evolution vector, and Cre and an archive of mutagenized Cre libraries (13) were tested for recombination activity (21). Recombination and subsequent polymerase chain reaction (PCR) would produce a 1.7-kb band reflecting recombination (fig. S1). However, Cre, as well as the library, failed to recombine the loxLTR sites, and no PCR product was obtained, which shows that the asymmetry and the mutations in loxLTR are too severe to result in recombination.
 
Because residual activity is required to start any directed evolution process (22), we split the original loxLTR target into two subsets. The palindromic target sites loxLTR1 and loxLTR2 were created based on the original asymmetric loxLTR sequence (Fig. 1A). However, when loxLTRs 1 and 2 were tested for recombination using either Cre or the library, no recombination was observed. Hence, the mutations in these sites were still too many for the starting library to display any activity, and this necessitated the further splitting of loxLTRs 1 and 2 by evenly dividing the half-site mutations to form four new subsets, termed loxLTRs 1a, 1b, 2a, and 2b (Fig. 1A). Splitting the mutations facilitated recognition by recombinases in the library and hence served as a starting point for subsequent directed-evolution cycles. Reiterative directed-evolution cycles resulted in enrichment of the recombinase libraries with functional candidates (Fig. 1B). The number of evolution cycles required to obtain efficient recombinases for each loxLTR varied between the subsets, but eventually efficient recombination activity of the libraries was observed for all subsets.
 
To determine whether a combinatorial approach would now allow recombination of the next higher subsets, we pooled and shuffled the libraries 1a and 1b, and 2a and 2b, and cloned the products into the evolution vectors harboring loxLTR1 and loxLTR2, respectively. The combination of mutations from the different libraries resulted in synergistic effects and led to the generation of recombinases, which now recombined loxLTRs 1 and 2 (Fig. 1B), demonstrating that an evolutionary strategy traversing through intermediates can be used to achieve a desired activity.
 
Next, we tried to address the asymmetry of the loxLTR target site. A recombinase that recombines an asymmetric target site has to recognize half-sites of varying sequence. To determine whether this task can be accomplished through substrate-linked protein evolution, we pooled and shuffled libraries from loxLTR1 and loxLTR2 and assayed for recombination in the evolution vector harboring the loxLTR sequence. Very low recombination activity was detected in the first cycles that was enriched for functional candidates in later cycles (Fig. 1B), demonstrating that symmetry in the target site is not a prerequisite for the site-specific recombination reaction.
 
After a total of 126 evolution cycles, the evolution process was halted and individual loxLTR specific recombinases were examined for their recombination properties. Fifty individual recombinases were functionally analyzed in E. coli. The most active recombinase (termed Tre) showed efficient recombination of the loxLTR site with some residual activity for loxP (Fig. 2, A and B). To quantify the target specificity of Tre, we examined its recombination properties in the different loxLTR evolution vectors. As in the reporter assay, Tre efficiently recombined the loxLTR sequence and displayed residual activity on loxP. Tre also showed efficient recombination on loxLTR2b and residual activity on loxLTR2, but no recombination was observed on loxLTR1a, loxLTR1b, loxLTR1, and loxLTR2a (Fig. 2C). This is unexpected when taking into consideration that Tre evolved from these subsets (compare figs. S2 and S3). The reason for this target specificity is currently unknown. However, this observation confirms previous findings that target specificity is regained after initial relaxation in directed evolution over many generation cycles (13, 14, 23).
 
Evolved recombinases from all subsets were sequenced to monitor the evolution process. The sequences revealed clustering of mutations arising from the different subsets that were combined through the course of evolution and complemented by novel clusters in the higher subsets (fig. S2 and table S1). In total, Tre has 19 amino acid changes when compared with Cre, with many mutations originating from different subsets (fig. S3).
 
Next, we examined the recombination properties of Tre in mammalian cells. HeLa cells were cotransfected with recombinase expression and reporter plasmids, and recombinase activity was evaluated 48 hours after transfection. As in the E. coli assays, Cre efficiently recombined the loxP reporter but did not recombine loxLTR. Tre showed efficient recombination on the loxLTR reporter and some residual activity on loxP (fig. S4, A and B). To investigate whether Tre can recombine its target in a genomic context, a stable loxLTR reporter cell line was tested for recombination after transfection with a Tre expression plasmid. PCR assays and β-galactosidase activity measurements demonstrated that Tre recombines loxLTR sequences packaged in chromatin (fig. S4, C and D).
 
To address the question of whether recombination mediated by Tre is occurring within the context of an HIV-1 LTR, reporter constructs responsive to the HIV-1 Tat transcriptional regulator were generated and tested (24) (Fig. 3A). When HeLa cells were cotransfected with a Tre expression vector along with the Tat vector and pHIV/T2/LUC, luciferase activity decreased by a factor of three (Fig. 3B). In contrast, no decrease in luciferase expression was detected when the same experiment was performed using the pHIV/T1/LUC control, containing only one loxLTR site (Fig. 3C). We performed PCR analysis to prove that the observed decrease in luciferase expression was a result of recombination and not of blocking Tat-mediated transcription from the LTR promoter by the recombinase. This experiment demonstrated that the reduction of Tat activation was indeed due to Tre-mediated excision of the luciferase cassette (Fig. 3D). Gel extraction of the PCR fragments followed by sequencing confirmed the precise excision of the loxLTR-flanked sequence.
 
To examine whether Tre can excise the provirus from the genome of HIV-1-infected human cells, we produced loxLTR containing viral pseudotypes that were used to infect HeLa cells. A virus particle-releasing cell line was cloned and stably transfected, either with a plasmid expressing Tre or with the parental control vector. The respective cell pools were monitored with respect to recombinase activity and virus production. All assays performed demonstrated the efficient deletion of the provirus from the infected cells without obvious cytotoxic effects (Fig. 4, A to E).
 
These data reveal that it is possible to evolve a recombinase to specifically target an HIV-1 LTR and that this recombinase is capable of excising the respective provirus from its chromosomal integration site. Using substrate-linked protein evolution, we demonstrated that target recognition by Cre recombinase can be adapted to a target site that is asymmetric and very remote from its original site. Given the relative ease with which we have altered Cre specificity, it is likely that additional recombinases could be generated that target other sequences present in LTRs (fig. S5). We accept that this approach is unlikely to be of immediate therapeutic use and that considerable obstacles would need to be overcome before an engineered recombinase could be practically used in any clinical setting. The most important, and likely most difficult, among these is that the enzyme would need efficient and safe means of delivery and would have to be able to function without adverse side effects in relevant target cells. Nevertheless, the results we present offer an early proof of principle for this type of approach, which we speculate might form a useful basis for the development of future HIV therapies.
 
Science 29 June 2007:
Vol. 316. no. 5833, pp. 1855 - 1857
 
Perspectives
AIDS/HIV:
A Reversal of Fortune in HIV-1 Integration
 
Alan Engelman
Department of Cancer Immunology and AIDS, Dana-Farber Cancer Institute, Boston, MA
 
Aretrovirus like HIV-1 must insert itself into the genome of an infected host cell to accomplish two things. Once integrated, viral DNA (known as the provirus) can be expressed and, importantly, the viral genome can be replicated and inherited upon host cell division. Consequently, the virus becomes inextricably linked to the host, making it virtually impossible to "cure" AIDS patients of their HIV-1 infection. On page 1912 of this issue, Sarkar et al. (1) construct an enzyme (recombinase) that can effectively excise integrated HIV-1 DNA from cultured, infected human cells. The results raise the possibility that customized enzymes might someday help to eradicate HIV-1 from the body.
 
Integrase, the viral enzyme that catalyzes HIV-1 integration, acts on short sequence elements known as attachment (att) sites located at the ends of HIV-1 DNA, called the long terminal repeats (LTRs). Because HIV-1 is a single-stranded RNA retrovirus, its genome must be transcribed by a viral enzyme called reverse transcriptase into a double-stranded DNA copy so that it can become integrated into a cell chromosome. Likely contributing to the irreversible nature of retroviral integration, att sequences must situate near a DNA end to support integrase function (2, 3). By contrast, other specialized DNA enzymes, typified by prokaryotic transposases and site-specific recombinases, efficiently excise DNA to facilitate the mobilization of genetic material (4). This has led to the hypothesis that recombinases could be modified to recognize viral sequences and perhaps excise integrated HIV-1 (5-7). Sarkar et al. test this hypothesis by constructing a modified version of Cre, a recombinase expressed by bacteriophage P1 that normally acts on DNA in prokaryotes.
 
Because of its ability to function efficiently in the absence of prokaryotic cofactors, Cre has long been used to rearrange DNA fragments in higher eukaryotes, including the deletion of relatively large segments situated between its recognition site, called loxP (8). This essential 34-base pair (bp) sequence is composed of an 8-bp core flanked by two perfect (symmetric) 13-bp inverted repeats (see the figure). A sequence that has 50% identity to this 34-bp sequence can be identified within HIV-1 long terminal repeats, aptly called loxLTR. Cre, though, is unable to rearrange DNA substrates containing these best-match loxLTR sequences (1, 5). The current breakthrough by Sarkar et al. centers on the identification of a Cre derivative, called Tre, which efficiently recognizes the loxLTR sequence and recombines loxLTR-containing DNA.
 
To circumvent the HIV-1 recombination roadblock faced by Cre, Sarkar et al. used the powerful substrate-linked protein evolution (SLiPE) (7) technique to "evolve" active mutant forms of Cre recombinase from an initial cornucopia of protein-coding sequences. To kick-start the method, they had to identify loxLTR sequences that support minimal levels of DNA recombination by Cre. Sequence changes that disrupted the perfect nature of the inverted repeats in the loxP site were initially rearranged within the loxLTR sequence, yielding hybrid substrates (see the figure) that were still not recognized by Cre. Second-generation hybrids were generated to further reduce sequence differences (mismatched bases) between the loxP and loxLTR sites. These next-generation sites each supported a low level of recombination by Cre, so they were plugged into the SLiPE assay, which selected for active recombinase enzymes from a randomly mutagenized Cre library. The technique incorporates a process known as DNA shuffling that randomly assorts DNA fragments to increase the overall extent of sequence diversity (9). Functional recombinases selected from initial SLiPE rounds were tested on first-generation loxLTR hybrid sequences, yielding enzymes that were active with substrates initially not recognized by Cre. The entire process was repeated to produce enzymes with loxLTR substrate specificity. Tre recombinase, selected from a random analysis of 50 clones, efficiently recognized, and recombined, loxLTR-containing DNAs. It retained marginal activity with loxP, yet was remarkably inactive with certain intermediary hybrids (which nonetheless drove the evolution of some of Tre's adaptive amino acid changes). Tre is an unexpected example of a Cre-based recombinase that efficiently recombines substrates harboring notable asymmetry in normally symmetric loxP site inverted repeats.
 
Excising HIV-1. Cre recombinase recognizes loxP sites in DNA, but not loxLTR sites present within HIV-1 DNA. White, pink, and blue denote sequence differences (mismatched bases) between the two sites. Sarkar et al. used multiple rounds of in vitro protein evolution to identify Tre recombinase, an enzyme derived from Cre that efficiently recognizes loxLTR. Tre can excise integrated HIV-1 from human chromosomal DNA.
 
Tre was tested for its ability to recombine loxLTR-containing substrates in cells, culminating in the demonstrable excision of integrated HIV-1 proviruses from a human cervical carcinoma cell line. Does this mean that enzymes like Tre recombinase could one day prove clinically useful? The answer is multifaceted. First, it will be important to assess the efficiency of Tre function under physiologically relevant conditions in vitro. Because the cell line used by Sarkar et al. likely harbored a few (perhaps only one) unique proviruses, the ability of Tre to eradicate HIV-1 from the multitude of chromosomal locations normally used during integration must be established. It will also be important to analyze substrates that extend beyond the specific loxLTR sequence from which Tre evolved. This region of the LTR is notably well conserved across viral clades, yet some sequences are less than 30% identical to loxP. Although SLiPE could be used to construct additional sequence-specific enzymes, the technology will be impractical if, for example, multiple enzymes are required per patient (HIV-1 diversifies to a quasispecies as patients progress toward AIDS).
 
By far the largest obstacle to potential Tre use in humans will be its safe and effective introduction into salient cell types. The very nature of retroviral integration affords HIV-1 a particularly "stealth" life-style whereby it can avoid systemic eradication by the host immune system. A small fraction of productively infected lymphocytes (CD4+ T cells) revert from an activated state to a quiescent state as a normal consequence of establishing immunological memory (10). In the resting state, these cells fail to produce virus, and so their harbored sequences escape antiretroviral therapy. Due to their natural role in the adaptive immune response, these resting T cells persist with an average half-life of ~44 months (11). Numerous attempts have been made to activate these cells, with the hope that such strategies would sensitize the accompanying viruses to antiviral drugs, leading to virus eradication. Advances with such approaches in patients have been slow to materialize, indicating that immune activation of resting CD4+ T cells as a therapy to eradicate HIV-1 is still in its infancy (10).
 
Enzymes like Tre might prove particularly useful if they could be stably expressed in this setting. Under favorable conditions, Sarkar et al. show that ~3 months is required to remove all HIV-1 traces in cultured cells, indicating that transient Tre expression is likely to provide limited benefit. Viral-based delivery systems can impart relatively long-lived trans-gene expression in nondividing cell types (12, 13). Experiments designed to test Tre in latently infected T cells alongside and perhaps in combination with immune activation therapies should reveal the effectiveness of the specialized DNA recombinase under these clinically relevant conditions. Although favorable results would represent perhaps only a baby step toward eventual use in patients, the discovery of the Tre recombinase proves that enzymatic removal of integrated HIV-1 from human chromosomes is a current-day reality.
 
References
 
1. I. Sarkar, I. Hauber, J. Hauber, F. Buchholz, Science 316, 1912 (2007).
2. L. I. Lobel, J. E. Murphy, S. P. Goff, J. Virol. 63, 2629 (1989).
3. A. D. Leavitt, R. B. Rose, H. E. Varmus, J. Virol. 66, 2359 (1992).
4. N. L. Craig, R. Craigie, M. Gellert, A. M. Lambowitz, Mobile DNA II (American Society for Microbiology, Washington, DC, 2002).
5. Y.-S. Lee, J.-S. Park, Biochem. Biophys. Res. Commun. 253, 588 (1998).
6. S.-T. Kim, G.-W. Kim, Y.-S. Lee, J.-S. Park, J. Cell. Biochem. 80, 321 (2001).
7. F. Buchholz, A. F. Stewart, Nat. Biotechnol. 19, 1047 (2001).
8. B. Sauer, Methods 14, 381 (1998).
9. W. P. Stemmer, Proc. Natl. Acad. Sci. U.S.A. 91, 10747 (1994).
10. Y. Han, M. Wind-Rotolo, H. C. Yang, J. D. Siliciano, R. F. Siliciano, Nat. Rev. Microbiol. 5, 95 (2007).
11. R. F. Siliciano, Top. HIV Med. 13, 96 (2005).
12. W. J. Swiggard et al., J. Virol. 79, 14179 (2005).
13. D. S. Strayer et al., Mol. Biotechnol. 34, 257 (2006).
10.1126/science.1145015
 
 
 
 
  icon paper stack View Older Articles   Back to Top   www.natap.org