INTRODUCTION

Retroviruses contain two copies of an RNA genome which replicates via a DNA intermediate (1). Transcription of the RNA genome into the DNA intermediate is performed by the viral enzyme reverse transcriptase (RT).¹ The primer for reverse transcription of the retroviral genome is a cellular transfer RNA (tRNA). tRNA is also crucial in several steps of the reverse transcription process subsequent to priming, such as the strand-transfer reactions. Human immunodeficiency viruses (HIV) and their simian counterparts utilize tRNA^Lys-3 as their primer (2, 3). HIV-1 virions appear to contain approximately eight tRNA^Lys-3 molecules per two copies of viral genome (4).

The reverse transcriptase protein of HIV-1 has been shown to directly bind the tRNA^Lys-3 primer in vitro (5-8). The interaction between tRNA^Lys-3 and HIV-1 reverse transcriptase is complex. At least four cellular tRNAs have been isolated from HIV-1 virions including tRNA^Lys-3, tRNA^Lys-1, tRNA^Lys-2, and tRNA^Ile although tRNA^Lys-3 is the predominant species isolated (3). Several interactions have been demonstrated in vitro between HIV-1 RT and tRNA^Lys-3. Barat et al. (5) have demonstrated specific contacts with the anticodon loop. Nuclease footprinting suggests partial protection of the so-called D-loop, TC loop, and anticodon loops by RT (6).

Mutations in the TC loop, and particularly the D-loop of tRNA^Lys-3 were found to significantly impair binding to HIV-1 RT in gel shift assays (8). Consensus RNA motifs selected in vitro by the SELEX procedure demonstrate that a consensus RNA pseudoknot with significant secondary and tertiary homology to the D-loop of tRNA^Lys-3 is a preferred binding motif to RT (9). Additionally, RT facilitates annealing of the tRNA^Lys-3 to the PBS sequence in vitro suggesting RT also plays a role in unwinding of the tRNA^Lys-3 acceptor stem (8). Finally, while RT-deficient virions contain a viral genome, the tRNA^Lys-3 primer is not packaged, indicating that tRNA^Lys-3 packaging is dependent upon the presence of RT (8). In addition to the primary tRNA-RT interaction required for packaging, RT also recognizes preformed tRNA-HIV-1 template complexes in vitro suggesting a second level of interaction likely to be important in initiation of cDNA synthesis (8).

Selective incorporation of tRNA^Lys-3 is thought to be mediated by the Pr160^gag-pol precursor. Precursor processing is not required for selective incorporation (3). Incorporation is also not dependent on the primer-binding site. Interactions have been described between the p66 as well as the p51 subunits of RT and the 5 end of tRNA by platinum cross-linking (10). Preincubation of heterodimeric RT (p66/p51) was noted to increase catalytic activity on a poly(A)-oligo(dT) template (11) suggesting that tRNA binding may also lead to "activation" of the p66/p51 heterodimer. Nuclease footprinting of HIV-1 RT-tRNA^Lys-3 complexes demonstrated protection of bases within the anticodon and D-loops, and potential TC-D-loop interactions which are disrupted in the presence of HIV-1 RT (6). Other investigators, using pancreatic RNase A footprinting, showed protection of both the tRNA^Lys-3 anticodon loop and D-loop but not the acceptor stem, confirming a strong HIV-1 RT interaction with several regions of tRNA (7).

In addition to interaction between the tRNA primer and HIV-1 RT, other interactions involving the tRNA primer include annealing of the 3-terminal 18 base pairs of tRNA^Lys-3 homologous to the HIV-1 PBS and annealing of various loops of tRNA^Lys-3 to regions of the HIV-1 genome outside the PBS (12-18). HIV-1 particles containing viral RNA sequences lacking the PBS continue to preferentially package tRNA^Lys-3, suggesting that PBS-tRNA annealing alone does not determine the packaging preference for the tRNA^Lys-3 primer (3).

Anti-retroviral strategies which make use of the unique properties of tRNAs have recently been proposed. Rossi and co-workers (19) described a chimeric tRNA in which anti-HIV-1 ribozyme was fused to tRNA^Lys-3. Because this chimeric RNA contained tRNA^Lys-3 coding sequences, it was expressed at high levels via RNA polymerase III. In addition, this ribozyme/tRNA hybrid was specifically targeted to HIV-1 virions based on the affinity of tRNA^Lys-3 for RT. Another study described a chimeric tRNA in which an anti-HIV-1 antisense RNA was linked to tRNA^Pro (20). In that case, the presence of tRNA^Pro sequences increased stability of the chimeric RNA in the intracellular environment (20).

We have developed a strategy for inhibition of the HIV-1 replication cycle via interference with priming of reverse transcription. To accomplish this, we designed a tRNA^Lys-3-derived mutant tRNA with a modified acceptor stem. This mutant tRNA, designated tRNAtarD, was shown to inhibit HIV-1 replication when expressed in human CD4 lymphocytes. The results presented here, together with available published information, indicate that potent inhibition of retroviral replication can be achieved by targeting priming of reverse transcription.

Inhibition of retroviral replication through interference with the normal process of priming constitutes a novel anti-retroviral approach and also provides a powerful tool for dissecting molecular aspects of priming.

EXPERIMENTAL PROCEDURES

The tRNA^Lys-3-UUU gene (21) with 5- and 3-flanking sequences was obtained by amplification of DNA extracted from HeLa cells using the primers: 5- TCGCCGAGATAAGCTTCAGCCTCTACTATGGTACAG-3 and 5-ATAATAGCACAAGCTTTATTACCCTCCACCGTCGTT-3 and Vent DNA polymerase (New England Biolabs) according to the manufacturer's instructions. The PCR fragment was then digested with HindIII and cloned into the HindIII site of pBluescript-KS vector, to yield pM13^Lys-3. The tRNAtarD gene, including 20 nucleotides upstream and 130 nucleotides downstream flanking the human tRNA^Lys-3-UUU gene, was obtained by PCR amplification using two primer pairs: 1L, 5-GTAAAGCTCTCGTGAAGATAGACCATAGCTCAGTCGGTAGAGC-3 and 1R, 5-CCAAAAGCAAAGACATGCCGCTTAGACCACAGGGACTTGAACCCTGGAC-3; and 2L, 5-GTCCAGGGTTCAAGTCCCTGTGGTCTAAGCGGCATGTCTTTGCTTTTGG-3 and 2R, 5-ATAATAGCACAAGCTTTATTACCCTCCACCGTCGTT-3. The first amplification was done using the pM13^Lys-3 DNA template and primers 1L and 1R, resulting in fragment A. A second amplification was carried out using pM13^Lys-3 and primers 2L and 2R, giving rise to fragment B. Both A and B fragments were then agarose-gel purified. A third PCR reaction was performed using a mixture of purified A and B fragments as templates. Primers 1L and 2R were used to generate a tRNAtarD sequence containing 5- and 3-flanking sequences required for correct transcription and processing by polymerase III. The final PCR fragment was cloned into the blunted XhoI site of pAC7, an AAV-derived vector (22), resulting in pACtarD. The sequence was then verified. tRNAtarD coding sequences were excised from pACtarD by HindIII and NcoI digestion. After complete filling of the ends formed by HindIII and NcoI, the fragment was subcloned into the SnaBI site of the vector pN2A (23), resulting in the retroviral vector pN2A-tarD (Fig. 1C), and sequence and orientation were confirmed by sequencing.

The vectors pUClys and pUCtarD contain coding sequences for both tRNA^Lys-3 and tRNAtarD downstream of the T7 promoter and were used for in vitro transcription. Coding sequences for tRNA^Lys-3 were synthesized by annealing and extending two oligonucleotides: left primer 5-AATTGCTGCAGTAATACGACTCACTATAGCCCGGATAGCTCAGTCGGTAGAGCATCAGACTTTTAATCTGAGGGTCCAGGGTTCAAG-3; and right primer 5-CTAAGCTGCAGATGCATGGCGCCCGAACAGGGACTTGAACCCTGGACCCT-3 at 70 °C for 15 min followed by 37 °C for 30 min in 1 × PCR buffer using Vent polymerase (New England Biolabs). The resulting PCR fragment was digested with PstI and cloned into the PstI site of pUC120 to yield pUClys. The tRNAtarD sequences were generated by amplification of pUClys using the primer pair: left 5-AATTGCTGCAGTAATACGACTCACTATATAGACCATAGCTCAGTCGGTAGAGCA and right 5-CTAAGCTGCAGATGCATGGTTAGACCACAGGGACTTGAACCCTGGACCCT. The resultant PCR fragment was then cloned into the SmaI site of pUC18 to yield pUCtarD.

The vector pSP-PBS was constructed by subcloning a 0.9-kilobase pair, ScaI to NsiI fragment from the HIV-1 molecular clone pNL4-3 (24), into the bacterial vector, pSP-73 (Promega Corp., Madison, WI), previously digested with PvuII and NsiI. The restriction fragment from pNL4-3 includes nucleotides 133 to +386 of the flanking sequences and 5-LTR region of HIV-1-NL4-3 (24).

pUClys and pUCtarD were linearized by NsiI digestion to generate a 3 CCA end, and transcribed by T7 polymerase using reagents and conditions supplied with the Promega T7 transcription kit in the presence of 10 µCi of [³²P]CTP. Radiolabeled RNAs were gel purified according to the method described previously (25). Competition analysis was performed by standard methods (26).

The GP+E-86 ecotropic packaging cell line was a generous gift from A. Bank (Columbia University, New York). The PA317 amphotropic packaging cell line and NIH3T3 fibroblasts were obtained from the American Type Culture Collection. The T-cell line MT-2 was obtained from the AIDS Research and Reference Reagent Program of the National Institutes of Health (Rockville, MD). The vector pN2A-tarD was transfected into the ecotropic packaging cell line GP+E-86 using lipofection (Life Technologies, Inc.). Transfected cells were selected in 0.5 mg/ml G418 (Geneticin, Life Technologies, Inc.). Virus-containing supernatant from the GP+E-86 cells was collected and used to transduce PA317 amphotropic packaging cells, and the cells selected in 0.5 mg/ml G418. High-titer clones were identified by measuring the transfer of G418 resistance using serially diluted supernatant to NIH3T3 fibroblasts, and supernatant from these clones was used to transduce MT-2 cells and CEM cells (ATCC).

The control vector, pLN, a MoMuLV-based retroviral vector containing the neomycin resistance gene expressed from the MoMuLV-LTR, was constructed and packaged in the laboratory of A. Dusty Miller (Fred Hutchinson Cancer Center, Seattle, WA).

To transduce MT-2 cells, LN and N2A-tarD vector-producing PA317 cells were irradiated at 40 gray and plated at 2 × 10⁶ cells/100-cm² plate 24 h before the addition of 1 × 10⁶ MT-2 cells. Co-cultivation was carried out in the presence of 8 µg/ml Polybrene for 48 h. Non-adherent MT-2 cells were collected, and a second round of co-cultivation over irradiated vector-producing fibroblasts was performed. Transduced MT-2 cells were selected in 0.5 mg/ml G418, and subcloned by limiting dilution in 96-well plates, to yield MT2/LN (transduced with LN) and MT2/TARD (transduced with N2A-tarD) clones.

Total RNA was extracted from parental MT2 and MT2/TARD cells using the acid guanidinium isothiocyanate/phenol-chloroform method (27). The reverse transcriptase reaction was carried out using MoMuLV reverse transcriptase (Life Technologies, Inc.) following the manufacturer's recommendations for first strand synthesis, using 100 ng of the antisense primers described below. One µl of reverse transcribed cDNA was amplified using 1 µl of RT reaction product as template. Each PCR reaction was conducted in a final volume of 50 µl containing 1.25 unit of Taq DNA polymerase and a total of 300 ng of tRNAtarD-specific primers (sense, 5-TAGACCATAGCTCAGTCGGT-3; and antisense, 5-TGGTTAGACCACAGGGACTT-3) or 300 ng of tRNA^Lys-3 specific primers (sense, 5-GCCCGGATAGCTCAGTCGGT-3; and antisense, 5-TGGCGCCCGAACAGGGACTT-3). Thermocycling temperatures used were: initial denaturation at 94 °C for 5 min, annealing at 60 °C for 2 min, and extension at 72 °C for 2 min, followed by cycling at 94 °C for 1 min, 60 °C for 2 min, and 72 °C for 2 min for a total of 30 cycles. 10 µl of the amplified PCR product was subjected to electrophoresis in a 2% agarose gel.

Parental MT-2, MT2/LN cells which were transfected with retroviral vector alone (no tRNAtarD expression), and MT2/TARD were compared for ability to support HIV-1 replication. Cells were routinely grown in Iscove's medium (Life Technologies, Inc.) containing 10% fetal bovine serum and 1 × of Penn/Strep at 37 °C. The virus strain used in this study was HIV-IIIB, which was prepared and titered in MT-2 cells.

Cells were seeded into 96-well plates at 5 × 10³ cells/well. HIV-IIIB virus stock was serially diluted 10-fold in Iscove's medium containing no fetal bovine serum. Using 4-6 wells per dilution for each cell line, 0.1 ml/well of each virus dilution was inoculated. Control wells received the same amount of medium. Infected cultures were examined daily for syncytia formation and viral titers determined at days 5 and 10 postinfection according to the method described by Reed and Muench (28).

For each infection, a total of 1 × 10⁴ cells (MT-2, MT2/TARD, or MT2/LN) in exponential growth phase were harvested, washed once with medium, and pelleted. The cell pellet was then resuspended in 1 ml of diluted HIV-IIIB stock containing 10 TCID₅₀ units of virus. After adsorption at 37 °C for 2 h, 10 ml of medium was added, and the cells pelleted by centrifugation, resuspended in 15 ml of Iscove's and 10% fetal calf serum medium, and transferred into a 25-cm² flask. Duplicate infections per cell line were employed in each challenge assay, and the infected cultures were incubated at 37 °C. Every other day beginning from day 2 postinfection, 0.5 ml of culture supernatant was removed from the flasks, and virus replication was monitored by measuring the production of p24 viral antigen in culture supernatant using an HIV-1 p24 antigen capture enzyme-linked immunosorbent assay (Coulter Immunology, Hialeah, FL).

Template RNAs comprising HIV-1 genome sequences for RNA-dependent DNA polymerase assays in vitro were generated by in vitro transcription from the plasmid clone, pSP-PBS (see "Vector Construction" above). Template RNA 1, 545 nucleotides long, was generated by transcription in vitro of an XmnI restriction fragment pSP-PBS using T7 RNA polymerase. Template RNA 1 encompassed nucleotides 133 to +386 of the flanking sequences and 5-LTR region of pNL4-3 (24). In addition, its 5 end contained 26 nucleotides from the multiple cloning site of pSP-73. Template RNA 2 was generated following restriction digestion of the plasmid pSP-PBS with HindIII followed by transcription in vitro using T7 RNA polymerase. Template RNA 2 contained the region of the genomic RNA from 133 to +81, 240 nucleotides in length. This sequence does not include the PBS. DNA primers 1, 2, and 3 were complementary to regions of the template RNAs corresponding to HIV-1 genomic positions +191 to +209, +59 to +78, and +22 to +41, respectively. Conditions for annealing to and DNA polymerization from RNA templates with various primers were as described previously (29). The RNA template and the respective primers were annealed in 10 mM Tris-HCl (pH 8.0), 1 mM EDTA, and 80 mM KCl with a 10:1 molar ratio of primer to template. Similarly, RNA-dependent DNA polymerase assays were carried out as before with slight modifications (29). Final reaction mixtures contained 50 mM Tris-HCl (pH 8.0), 1 mM dithiothreitol, 1.0 mM EDTA, 34 mM KCl, 6 mM MgCl₂, 2 nM substrate, 50 mM unlabeled dNTPs, and 4 units of recombinant purified HIV-1 RT. Recombinant HIV-1 RT was graciously provided to us by Genetics Institute (Cambridge, MA). This RT has native heterodimeric structure, is free of detectable nuclease and polymerase contaminants, and has a specific activity of 40,000 units/mg (29). The RT was preincubated with the substrate for 5 min at 37 °C. The reaction was initiated by the addition of MgCl₂ and dNTPs, incubated for 15 min and terminated with 25 ml of termination mixture (90% formamide (v/v), 10 mM EDTA (pH 8.0), and 0.1% each of xylene cyanol and bromphenol blue). For some experiments, DNA synthesis was performed in the presence of [-³²P]dCTP. When radiolabeled tRNA primers were employed, unlabeled dNTPs were used in the synthesis reaction. Reaction products were resolved on 6% urea polyacrylamide denaturing gels followed by autoradiography.

Cells were harvested by low speed centrifugation (3,000 rpm, 30 min). Recovered supernatant was filtered through a 0.22-mm filter, and virus present in the filtrate was concentrated and purified by sucrose gradient centrifugation prior to isolation of viral RNA. HIV-1-IIIB produced from MT2/LN cells was similarly prepared for control analysis. Total viral RNA was extracted from purified viral pellets using the acid guanidinium isothiocyanate/phenol chloroform method (27). RNA concentration was determined by A₂₆₀, and RNA integrity was verified by electrophoresis on a 1% agarose gel. Analysis of viral RNAs for mutant tRNAtarD was conducted by RT-PCR (see above). For each RT-PCR reaction, 0.2 µg of viral RNA was amplified by the tRNAtarD-specific primers, and amplified products were subjected to gel electrophoresis. Amplification of the viral RNAs with tRNA^Lys-3-specific primers was performed as a control.

Infected cell DNA was extracted by the urea lysis method (30), phenol-extracted, ethanol-precipitated, and resuspended in TE buffer to a final concentration of 1.4 µg/µl. Target cell DNA was diluted to contain 1 copy of proviral DNA per µl, and 1 µl of this dilution was amplified 11 independent times with Pfu polymerase (Stratagene, La Jolla, CA) following the manufacturer's recommendations. PCR products thus obtained were purified on ion exchange mini-columns (QIAGEN Inc., Chatsworth, CA) and sequenced in the sense and antisense directions (AmpliTaq, Perkin-Elmer). Primers for amplification and sequencing were as follows: sense: 5-GGTCTCTCTGGTTAGACCAG-3 and antisense: 5-CGTCGAGAGATCTCCTCTG-3. Amplification conditions were 28 cycles, where each cycle consisted of the following steps: 94 °C for 1 min followed by 55 °C for 1 min followed by 72 °C for 30 s.

RESULTS

Design and Construction of a Mutant tRNA with Modified 3-Terminal Sequences

We wished to synthesize a mutant tRNA in which recognition of the PBS would be redirected to a region of the viral genome different from PBS. We constructed a gene encoding a tRNA^Lys-3-derived mutant in which 3 primer-binding sequences in the acceptor stem were modified to be complementary to the highly conserved TAR stem-loop region of HIV-1. This mutant was designated tRNAtarD (Fig. 1, panel A). This modification was introduced with the purpose of triggering aberrant priming of reverse transcription at a site upstream from the natural PBS. We chose the TAR stem-loop sequence as a target for this aberrant priming by mutant tRNA because it is a conserved sequence and is important in the reverse transcription process itself (31). Priming of reverse transcriptase at the TAR site would be predicted to interfere with the retroviral life cycle by causing premature priming upstream of the PBS.

We substituted seven nucleotides of the 3 sequence of tRNA^Lys-3 to produce a sequence complementary to the conserved TAR region of HIV-1, and made six additional substitutions in 5 sequences complementary to the 3 mutations to restore the tRNA secondary structure (Fig. 1, panel A), in the acceptor stem such that the mutant tRNA would retain predicted secondary and tertiary tRNA structure and ability to bind RT (32, 33). Care was taken not to modify other regions of the tRNA which are important for polymerase III transcription such as "A" and "B" boxes (34), for stability, and for interaction with HIV-1 RT (5, 6, 8).

HIV-1 RT has been reported to specifically recognize the tRNA^Lys-3 anticodon region (5, 35, 36). Mutant tRNAtarD retains wild-type anticodon stem-loop sequences (Fig. 1). To determine whether substitutions within either the 3 primer binding sequences or the structure of the acceptor stem of tRNA would affect the affinity of RT for mutant tRNAtarD, the previous tRNA and wild-type tRNA^Lys-3 were transcribed in vitro and assayed for binding by gel retardation assay in the presence of cold competitor tRNAs (Fig. 2). As shown in Fig. 2, binding of tRNAtarD to HIV-1 RT could be effectively competed by increasing molar concentrations of tRNA^Lys-3 (lanes 1-6). Conversely, binding of tRNA^Lys-3 to HIV-1 RT could be effectively competed by increasing molar concentrations of tRNAtarD (lanes 9-14). The degree of competition that tRNAtarD and tRNA^Lys-3 exert on each other is identical and therefore, we conclude that tRNAtarD binds to purified HIV-1 RT with similar affinity to that of tRNA^Lys-3 in our assays. Additional gel-shift experiments showed that tRNAtarD bound efficiently to HIV-1 RT, but not to MoMuLV RT (data not shown). The ability of tRNA^Lys-3-derived mutant tRNAtarD to bind RT is likely retained due to the preserved anticodon region of tRNA, which is important in recognition by HIV-1 RT (36).

We transduced the MT-2 T-cell line with the retroviral vector, N2A-tarD (Fig. 1, panel B) containing tRNAtarD coding sequences. Assays for polymerase III-directed transcription in vitro using Jurkat-derived T-cell extracts, and the N2A-tarD vector coding sequences, resulted in transcription and correct processing of tRNAtarD (data not shown). Following G418 selection of transduced cells, DNA and RNA were extracted from bulk G418-selected cloned cell lines, and mutant tRNA sequences were detected using RT-PCR. A band with the size of mutant tRNAtarD (76 nucleotides) was detected in MT2/TARD cells (Fig. 3, panel A, lanes 1, 3, 5, and 7) but was not observed in control MT2/LN cells (Fig. 3, panel A, lane 9). No band was detected in the MT2/TARD cells when PCR was carried out in the absence of RT (lanes 2, 4, 6, and 8), indicating that the target band obtained in the MT2/TARD cells was derived from tRNA, and not residual DNA contamination. In contrast, a characteristic band resulting from tRNA^Lys-3 was detected with similar intensity in both parental MT-2 and MT2/TARD cells, when the RNA samples were amplified with tRNA^Lys-3 primers (data not shown). Northern blotting and hybridization with a radiolabeled tRNAtarD-specific probe confirmed tRNAtarD expression (data not shown).

To examine the relative sensitivity of the transduced MT2/TARD cells to HIV-1, an infectious stock of HIV-IIIB was simultaneously titered in both control MT2/LN cells and G418-resistant MT2/TARD cell clones (Fig. 3, panel B) and bulk-selected G418-resistant MT2/TARD cells. The relative ability of these subclones to support HIV-1 replication was determined by the TCID₅₀ assay. Although relative titers varied among different subclones, all subclones tested showed decreased TCID₅₀ titers compared with control MT-2 or MT2/LN cells. As shown in Fig. 3, panel B, the TCID₅₀ titers measured for subclones 5 and 12 were 10^4.8 and 10^5.0, respectively. These two subclones were the most refractory to HIV-1 replication, showing a marked decrease (>1.5 log) in virus titers compared with control MT-2 cells (10^6.4) or MT2/LN (10^6.5). Subclones 3 and 9, representing the least resistant clones, exhibited a lower but detectable drop (<0.3 log) in virus titers compared with control cells.

Similar data were obtained with uncloned, G418-selected MT2/TARD cells. The frequency of syncytia formation in pooled G418-resistant MT2/TARD cells was significantly lower (Fig. 5, panel A) than that in parental MT-2 cells (Fig. 5, panel B) when using an identical viral inoculum. The TCID₅₀ titer obtained at day 5 post-infection was 10^4.5 in MT-2 cells as compared with 10^3.0 in MT2/TARD cells. The relative estimated titers increased with time and reached 10^6.0 in parental MT-2 cells and 10^4.8 in the MT2/TARD cells, respectively, at day 10 post-infection. Thus, the estimated viral titer in pooled G418-resistant MT2/TARD cells was reproducibly 1.0-1.5 log lower than that of control MT2/LN cells, or MT2/LN cells not expressing mutant tRNAtarD. The estimated viral titers did not change significantly with extended incubation of up to 15 days.

The above results, taken together with those described in Fig. 3, panel A, demonstrate that the level of tRNAtarD expression correlated with the efficiency of HIV-1 inhibition in individual cell clones. This assay was repeated several times with similar results.

Inhibition of viral replication in cells transduced with tRNAtarD could be due to changes induced in cells during the transduction, drug selection, or cloning procedures. To rule out these possibilities, we studied the rate of growth of MT-2 and MT2/TARD cells. Comparison of growth kinetics indicated that the growth rate of MT2/TARD was indistinguishable from that observed in parental MT-2 cells and MT2/LN cells (data not shown). It was also possible that lower infectivity of MT2/TARD cells was caused by decreased expression of CD4, the primary receptor for HIV-1. We performed flow cytometric analysis in all the cell clones described, and found no appreciable differences in CD4 expression with respect to MT-2 or MT2/LN cells (data not shown). Therefore, we conclude that reduction in levels of viral replication is not due to changes in growth rate or CD4 antigen expression.

To further assess the inhibitory effect of tRNAtarD on HIV-1 replication, MT2/TARD subclone 5 was tested for its ability to support HIV-1 replication. Following HIV-1 infection, HIV-1 p24 viral antigen was initially detected on day 9 in the control MT2/LN cultures, and reached a peak on day 17 (Fig. 4). In contrast, p24 antigen was not detected until day 17 in MT2/TARD cells, demonstrating a delay of approximately 8 days in p24 production compared with control cells. Visible syncytia appeared in the control cultures at day 12 post-infection, and involved most of the cells by day 17. In contrast, very few syncytia were observed even after 27 days post-infection in MT2/TARD cells. Representative photomicrographs of infected MT2/TARD and MT2/LN taken at day 15 post-infection are shown in Fig. 5. Qualitatively similar results were also observed in CEM cells transduced with N2A-tarD relative to parental CEM cells (data not shown). These measurements demonstrated that tRNAtarD expression resulted in decreased HIV-1 infection as assayed by either p24 production or relative titer of the same HIV-1 stock.

We next determined whether mutant tRNAtarD was incorporated into HIV-1 virions. Total viral RNA from 10⁹ TCID₅₀ of infectious virus was extracted from HIV-IIIB virus propagated in MT2/TARD cells, or from virus produced in MT2/LN cells. The results of RT-PCR amplification of viral RNAs, designed to detect the tRNAtarD sequence, is shown in Fig. 6. When viral RNAs were probed with tRNAtarD-specific primers, a 76-base pair tRNAtarD product was detected in the viral RNA sample from HIV-IIIB propagated in MT2/TARD cells (Fig. 6, lane 2), but was not seen in viral RNA prepared from control virions (Fig. 6, lane 1). No PCR products were observed in controls in which either viral RNA or RT was absent (Fig. 6, lanes 3 and 4). PCR products were detected in both control and target viral RNAs when the tRNA^Lys-3 specific primers were used (lanes 5 and 6). These results indicated that mutant tRNAtarD was incorporated into HIV-IIIB virions prepared from MT2/TARD cells.

A previous report demonstrated that the TAR loop was a suitable target for inhibition of HIV-1 replication using antisense RNA (37). It is possible that tRNAtarD inhibits HIV replication by binding to the TAR loop in viral mRNA, therefore interfering with Tat-mediated transactivation. A second possibility is that binding of tRNAtarD to the TAR loop on viral mRNAs would prevent their efficient expression. We tested these possibilities by measuring basal and Tat-induced transcriptional activities of the HIV-1 LTR in the presence or the absence of tRNAtarD (Table I). MT-2 cells were stably transfected with the LN (MT2/LN) or N2A-tarD (MT2/TARD) vectors. These cells were then transiently co-transfected with a construct in which the chloramphenicol acetyltransferase reporter gene was driven by the HIV-1 LTR, and with a Tat expression vector (pSV-Tat) or control plasmid. Basal LTR activity in the absence of Tat was about 2-fold higher in MT2/LN than in MT2/TARD cells, while Tat transactivation efficiencies were 23.4-fold in MT2/LN cells and 21.6-fold in MT2/TARD cells. The above experiments suggest that expression of tRNAtarD did not markedly influence in the ability of Tat to transactivate the LTR. However, we cannot exclude a moderate inhibitory effect on LTR-directed transcription. Since the expression of tRNAtarD can inhibit viral production by several orders of magnitude, we conclude that potential antisense effects exerted by tRNAtarD may account only in part for the overall virus inhibition we observed.

Table I. Mechanism of inhibition of viral replication by tRNAtarD Potential antisense effect of tRNAtarD on viral gene expression. MT2/LN and MT2/tarD cells were transfected with the plasmids pHIVLTRCAT, pSV-Tat, and control plasmid as indicated, and chloramphenicol acetyltransferase (CAT) activity was measured 48 h post-transfection. Tat transactivation efficiencies were calculated by dividing chloramphenicol acetyltransferase activity with pSV-Tat by chloramphenicol acetyltransferase activity obtained with control plasmid transfection. Cell type Plasmids CAT activity HIV-LTR-CAT pSV-Tat cpm MT2/LN +   4,807 +   4,990 + + 118,024 + + 110,719 MT2/TARD pool +   3,408 +   1,679 + + 46,578 + + 63,253

We anticipated that tRNAtarD would interfere with normal initiation of DNA synthesis. Consequently, we measured the ability of tRNAtarD to prime HIV-1 RT-directed reverse transcription in vitro from a segment of HIV-1 genomic RNA (RNA 1). The RNA 1 segment used includes sequences from the PBS to the 5 end of the viral RNA. Reactions were carried out in vitro with highly purified HIV-1 RT, using in vitro synthesized tRNAtarD or tRNA^Lys-3 as primers. Control reactions were also carried out using three DNA primers (Fig. 7, panel A). Expected extension products of lengths 368, 237, and 200 from DNA primers 1, 2, and 3, respectively, were observed (Fig. 7, panel B, lanes 4, 3, and 2). Extension of the control primers to correct lengths demonstrates that RNA 1 can serve as an effective template for DNA synthesis primed from a variety of locations. Extension of the tRNA primer at the PBS would be expected to yield a product 416 nucleotides in length, whereas priming at the TAR loop homology site would generate a 243-nucleotide-long product. As expected, priming with tRNA^Lys-3 produced a 416-nucleotide-long product (Fig. 8, panel B, lane 5). Priming with tRNAtarD also produced a 416-nucleotide band, indicating that, despite the introduction of mutations in tRNAtarD, the residual complementarity was sufficient to prime reverse transcription at the PBS (Fig. 7, panel B, lane 6). A band is also visible at about position 243, consistent with extension of tRNAtarD from its intended binding site on the TAR stem-loop. However, the observed band could also have been caused by normal pausing of polymerization during synthesis of the 416-nucleotide product. One potential effect of the observed priming by tRNAtarD at the PBS would be to introduce foreign sequences into the PBS of viral progeny.

To improve our ability to see potential products of priming from the intended tRNAtarD-binding site, we measured synthesis on a shorter viral RNA template containing the TAR loop sequences but not the PBS (RNA 2). We observed a 243-nucleotide-long product consistent with priming within the stem of the TAR hairpin (Fig. 8, panel D, lane 2). No extension products were observed in control experiments containing either tRNA^Lys-3 or DNA primer 1 plus RNA 2, since neither primer was expected to bind the template (Fig. 7, panel D, lane 1 and panel C, lane 1). Products of expected lengths were observed using DNA primers 2 and 3 on RNA 2 (Fig. 7, panel C, lanes 3 and 2). This result demonstrates that RT-directed priming can also occur from a site outside of the PBS. During infection, this would produce a short reverse transcript, likely to disrupt viral replication.

We demonstrated that exponential growth of HIV-1 in tRNAtarD-expressing cells was delayed by 7-14 days with respect to viral growth in parental MT-2 cells (Fig. 4). Although this delay is a clear indication of the inhibitory activity of the mutant tRNAtarD, significant levels of viral burden were ultimately achieved in these cultures (Fig. 4). The rise in p24 levels observed at day 17 in MT2/TARD cells, to which we will refer as "breakthrough virus," may be due to the generation of escape mutants. An alternative explanation is that viral amplification beyond a certain threshold may overwhelm the inhibitory capacity of the mutant tRNA.

We examined the possibility of mutations in the PBS and PBS-proximal areas reported to be involved in interactions with tRNA^Lys-3 (12-18) in breakthrough viruses. Breakthrough virus was generated by infecting MT2/TARD with a HIV-1 NL4-3, a previously described molecular clone (24) at a low multiplicity of infection (0.01) and passaged until breakthrough was observed. The growth kinetics of this infection was compared with that of unmodified MT-2 cells as was described in Fig. 4. Kinetics of viral replication of HIV-1 in MT2 and MT2/TARD cells in this experiment paralleled our previous observations (Fig. 4). After 5 weeks of passage, a cell-free virus stock from the MT2/TARD cell culture was prepared. This virus stock was used to infect a fresh culture of MT2/TARD cells, this time at high multiplicity of infection (1.0) and cellular DNA was extracted 48 h later and used as a substrate for limiting-dilution PCR and DNA sequencing.

Sequence for 11 PCR amplification products analyzed comprised nucleotides 500-685 (see Fig. 8) of the HIV-1 NL4-3 proviral sequence (24). This region included the PBS and short sequences outside the PBS proposed to interact with tRNA^Lys-3 (12-18). Sequences for 11 independent PCR clones showed no mutations when compared with the sequence of the virus used as inoculum. Therefore, we conclude that the ability of HIV-1 to replicate actively after long-term passage in MT2/TARD cells is not due to mutations in the PBS or PBS-proximal areas. It is therefore possible that mutations elsewhere in the genome confer resistance to the inhibitor tRNAtarD. However, it is also possible that virus amplification beyond the inhibitory capacity of tRNAtarD occurs after several weeks of passage, in the absence of mutations. Finally, mutations in the RT enzyme which decrease affinity for mutant tRNAtarD have not been ruled out.

DISCUSSION

tRNA functions in living cells mainly as a vehicle to translate genetic information stored in mRNA into amino acid sequence in proteins. Cellular tRNAs are recognized by many cellular proteins including 5 and 3 tRNA processing enzymes (38) and tRNA aminoacyl transferases (aa-tRNA synthetase) (32, 33). Most of these enzymes recognize both the anticodon region and specific features of the tridimensional structure of tRNA (32). There are two different human tRNA^Lys genes. One of these, tRNAlys-UUU (21, 39), has complementary sequence to the HIV-1 RNA genome, and is the primer for HIV-1 reverse transcription.

We have introduced mutations into tRNAlys-UUU (tRNA^Lys-3) designed to alter PBS binding specificity, while maintaining conserved sequence in the so-called A and B boxes needed for polymerase III-directed transcription (34), and maintaining integrity of the anticodon region. We have substituted sequences in the acceptor stem, to make a 3 domain complementary to the conserved HIV-1 TAR region. To ensure efficient transcription of this mutated gene, we included 5-flanking sequences which may contain enhancers for polymerase III-directed transcription. Also included were 3-flanking sequences consisting of the stop signal for polymerase III, and nucleotides required for processing, derived from the most efficiently expressed of the three cellular tRNAlys-UUU loci (21, 40). The mutant tRNA was also designed to maintain essential features, in regions away from the 3 end, required for interaction with HIV-1 RT and NCp7 (5, 36). Expression of mutant tRNAtarD in cultured T-cells did not alter cellular morphology, the rate of cell division, or CD4 antigen expression, suggesting it did not interfere with the function of the cellular tRNA^Lys-3.

Mutant tRNAtarD expression results in decreased HIV-1 replication, as assayed by p24 levels, or by relative TCID₅₀ titer in the MT-2 and CEM T-cell lines. Analysis of multiple transduced subclones showed a correlation between levels of tRNAtarD expression and HIV-1 inhibition.

Although the exact mechanism by which mutant tRNA expression protects cells against HIV-1 challenge is not yet known, several possibilities exist. Several reports indicated that p66 of the HIV-1 RT p51/p66 heterodimer recognizes and binds to the tRNA^Lys-3 anticodon region (7, 18, 41, 42) and may help unwind the acceptor stem (18) in the presence of NCp7 protein (43). Another report demonstrated that excess wild-type tRNA^Lys-3 primer inhibited the DNA polymerase activity of a recombinant HIV-1 RT, p66/p51 heterodimeric form (44). This effect was ascribed to the anticodon region of tRNA^Lys-3 primer (44). As mutant tRNAtarD levels appear to be lower than endogenous tRNA^Lys-3, direct competition is not likely to be the primary mechanism for inhibition of viral replication.

Although tRNAtarD can bind to RT, we had originally expected that it would not prime reverse transcription from the correct PBS since tRNAtarD lacks significant complementary sequences to the 5 region of the HIV-1 PBS (Fig. 9). However, results of priming assays carried out in vitro using purified HIV-1 RT demonstrate that the tRNAtarD can prime at either the PBS or the intended binding site in the TAR stem-loop. Significantly, the three 3-most nucleotides in tRNAtarD (5-CCA-3), which are post-transcriptionally added to all tRNAs, are complementary to the corresponding nucleotides in the PBS. Alternatively, tRNAtarD-directed priming from the PBS would introduce a foreign sequence into the PBS which could interfere with the second strand transfer step of viral replication. In this step, the sequences copied into DNA from the modified region of tRNAtarD have to bind to sequences complementary to the normal PBS sequence to complete synthesis of the double stranded DNA intermediate of viral replication. Introduction of a foreign sequence into the PBS could have additional effects, such as lowering the efficiency of priming by normal tRNA^Lys-3 in the altered virus. This could affect both replication in infected cells, and the ability of the virus to infect new cells. tRNAtarD-directed priming from the TAR sequence would make a viral minus strand DNA that is too short to complete synthesis. Aberrant cDNA synthesis could also lead to premature viral RNA template degradation by the RNase H activity of HIV-1 RT, in theory leading to production of integration-defective proviral DNA.

Inhibitory effects may be enhanced further by the selective packaging of this mutant tRNA into virions, since the PBS is not thought to be required for selective incorporation of tRNA^Lys-3 (42, 45) and since incorporation into the virions is thought to be mediated by interaction of Pr160-polymerase precursor with the tRNA anticodon region (3). We demonstrated the incorporation of mutant tRNAtarD into virions. We also observed decreased infectivity of virus particles produced in cells expressing mutant tRNAtarD compared with wild-type virions, when normalized for p24 levels (data not shown). Mutant tRNAs with homology to sequences other than the PBS may result in the production of defective virions.

Potential mutations leading to adaptation to growth in MT2/TARD cells include alterations in the PBS to acquire complementarity to TAR, mutations in TAR to reduce complementarity to mutant tRNAtarD, or mutations within RT to decrease affinity for mutant tRNA. Because results in vitro suggested that tRNAtarD was preferentially priming DNA synthesis at the PBS (Fig. 7), we wanted to ascertain whether the appearance of breakthrough viruses might be a result of mutations in the PBS. Sequence analysis of 11 independent PCR clones showed absence of mutations in the PBS and adjacent areas. This result may have several explanations. First, mutations elsewhere in the genome (i.e. the TAR region, RT coding sequence) may allow viruses to escape the inhibition by tRNAtarD. It is also possible that the production of such mutations does not occur, and the high levels of p24 production in MT2/TARD cells at late time points is a result of virus amplification beyond the inhibitory capacity of tRNAtarD. In this case, one would conclude that the inhibition of viral replication by tRNAtarD is simply more effective at low virus titers (i.e. early in infection) than at high virus titers.

We did not observe adverse effects of tRNAtarD expression on human T-cell lines as assayed by morphologic examination, CD4⁺ expression, or changes in growth kinetics. Although we believe the alterations we have introduced will preclude interactions with cellular aminoacyl transferase, the safety of this approach remains to be established. Rare neurologic syndromes have been described in patients with mutations in mitochondrial tRNA^Lys-3 (46-48). Whether hematopoietic or lymphoid survival will be negatively affected in vivo by introduction of tRNAtarD is not known. However, preliminary results in T-cell lines suggest that there is no apparent toxicity, and that tRNAtarD may be useful as an anti-HIV-1 therapeutic strategy.

We have observed a major inhibition of HIV-1 replication in cells expressing mutant tRNAtarD. This strategy may offer advantages relative to conventional antisense because of the specific interaction of HIV-RT with tRNA^Lys-3 derivative molecules, and the apparent ability of modified tRNAs to interfere with reverse transcription. The use of tRNA^Lys-3 mutants with altered primer binding specificity to target HIV-1 replication may represent a novel gene therapy approach for acquired immunodeficiency syndrome.