Forced Selection of a Human Immunodeficiency Virus Type 1 Variant That Uses a Non-Self tRNA Primer for Reverse Transcription: Involvement of Viral RNA Sequences and the Reverse Transcriptase Enzyme

Abstract

Human immunodeficiency virus type 1 uses the tRNA₃^Lys molecule as a selective primer for reverse transcription. This primer specificity is imposed by sequence complementarity between the tRNA primer and two motifs in the viral RNA genome: the primer-binding site (PBS) and the primer activation signal (PAS). In addition, there may be specific interactions between the tRNA primer and viral proteins, such as the reverse transcriptase (RT) enzyme. We constructed viruses with mutations in the PAS and PBS that were designed to employ the nonself primer tRNA^Pro or tRNA_1,2^Lys. These mutants exhibited a severe replication defect, indicating that additional adaptation of the mutant virus is required to accommodate the new tRNA primer. Multiple independent virus evolution experiments were performed to select for fast-replicating variants. Reversion to the wild-type PBS-lys3 sequence was the most frequent escape route. However, we identified one culture in which the virus gained replication capacity without reversion of the PBS. This revertant virus eventually optimized the PAS motif for interaction with the nonself primer. Interestingly, earlier evolution samples revealed a single amino acid change of an otherwise well-conserved residue in the RNase H domain of the RT enzyme, implicating this domain in selective primer usage. We demonstrate that both the PAS and RT mutations improve the replication capacity of the tRNA_1,2^Lys-using virus.

Reverse transcription is the replication step that converts a retroviral RNA genome into a DNA copy, a mechanism that is shared by retroviral elements and hepadnaviruses. Although a variety of primer molecules can be used to initiate reverse transcription, all retroviruses use a cellular tRNA primer (39, 46, 51, 52, 60). The most commonly used primers are tRNA^Pro (e.g., murine leukemia virus and human T-cell leukemia virus), tRNA^Trp (e.g., avian myeloblastosis virus), tRNA_1,2^Lys (e.g., Mason-Pfizer monkey virus), and tRNA₃^Lys (e.g., mouse mammary tumor virus and all lentiviruses, including human immunodeficiency virus type 1 [HIV-1]). Retroviruses are dedicated to the self tRNA primer, despite an excess of other tRNA molecules in the infected cell (15, 40, 63). Selective tRNA usage seems less stringent for some retroviruses (47, 48, 64).

Primer tRNA selection in HIV-1 is accomplished by multiple mechanisms. First, the primer tRNA is selectively packaged into virus particles (29, 30). All tRNA^Lys isoacceptors are packaged during particle assembly via their interaction with the Gag-Pol precursor and a protein complex composed of the cellular lysyl-tRNA synthetase and the viral Gag protein (11, 27, 28, 35, 49). Only tRNA₃^Lys is tightly associated with the viral RNA (vRNA) genome (29). Second, the vRNA genome contains an 18-nucleotide sequence that is perfectly complementary to the 3′ end of the self tRNA₃^Lys and is termed the primer-binding site (PBS) (Fig. 1). It should be noted that the PBS is not required for the selective packaging of tRNA₃^Lys into virus particles, but it is required for tight annealing of the tRNA₃^Lys primer (29). Thus, the PBS fulfills an indispensable role in the placement of the tRNA primer onto the viral genome and hence in reverse transcription (23, 45). No spontaneous mutations or more gross tRNA switches have been reported. A single point mutation that is recurrently observed at a low incidence in the HIV-1 PBS results from the infrequent usage of a low-abundance tRNA₅^Lys variant (16). In fact, the viral PBS motif is inherited from the sequence of the tRNA primer during reverse transcription, and PBS mutations can revert to the wild-type (wt) sequence within one round of replication through extension of a misaligned tRNA₃^Lys primer (14). This special reversion ability explains the almost complete conservation of the PBS sequence. HIV-1 mutants with an altered PBS identity exhibit a significant replication defect and rapidly revert to the wt PBS sequence (15, 40, 63). These results indicate that there are additional viral features that determine tRNA primer specificity. One obvious candidate is the reverse transcriptase (RT) polymerase, and a functional analysis of mutant HIV-1 virion particles revealed that RT is indeed involved in the selection and PBS annealing of the tRNA₃^Lys primer (45, 57). Biochemical studies have provided additional information on the RT-tRNA₃^Lys complex and its involvement with the RNase H domain (17, 54, 56, 58), but no high-resolution picture has emerged from these studies.

FIG. 1.

HIV-1 genome and the PAS and PBS motifs that specify tRNA primer usage. The HIV-1 DNA genome is shown at the top. The 5′ long terminal repeat (LTR) is divided into three segments (U3, R, and U5). Transcription starts at the U3-R border (arrow). A close-up of the untranslated leader of the vRNA is shown (from the transcription start site +1 to the Gag start codon AUG). The PAS and PBS are indicated. The cloverleaf structures of the self tRNA₃^Lys primer and the nonself primers tRNA_1,2^Lys and tRNA^Pro are shown below. Base modifications in the tRNA molecules are indicated according to standard nomenclature (59). R29 in tRNA_1,2^Lys indicates that this position is G in tRNA₁^Lys and A in tRNA₂^Lys, and Y41 is C in tRNA₁^Lys and U in tRNA₂^Lys The anti-PAS and anti-PBS motifs are boxed. We mutated the HIV-1 PAS and PBS motifs to complement the nonself primers. Details of these mutations are shown in Fig. 3 and 4.

The PBS is not the only RNA sequence in the retroviral genome that is important for tRNA priming. Additional contacts between the vRNA and the self tRNA have been reported for Rous sarcoma virus (1, 2, 13, 55) and subsequently for HIV-1 and HIV-2 (3, 4, 6, 20, 22, 24, 25, 44). An interaction was proposed between the U-rich anticodon loop of tRNA₃^Lys and an A-rich loop of a hairpin structure in HIV-1 RNA that is positioned upstream of the PBS (24). Based on this base-pairing interaction, it was tested whether the usage of a non-self tRNA primer can be imposed by simultaneous adaptation of the PBS and the A-rich loop. Such double mutants exhibited a replication defect, but variants that stably used the nonself primers tRNA^His, tRNA^Glu, and tRNA_1,2^Lys were selected upon prolonged culturing (18, 32, 34, 41-43, 62, 65). This type of virus reversion analysis is complicated by the fact that the retroviral genome is densely packed with replication signals. For instance, the A-rich interaction site is also an important sequence element for binding of the integrase protein during insertion of the proviral HIV-1 DNA into the host chromosome (19). Furthermore, important RNA structure motifs are encoded by this domain (3, 6, 7), and the destruction and subsequent repair of these structures may explain some of the virus reversion events (9). Despite an extensive search for second-site protein mutations that compensate for the switch of tRNA primer, no such coadaptive changes have been reported thus far (32, 65).

We recently presented evidence for an additional vRNA-tRNA contact between HIV-1 sequences that are positioned upstream of the PBS and the TΨC arm of tRNA₃^Lys (3, 6). This motif does not contribute to tRNA-PBS annealing, but it is essential for the initiation of tRNA-primed reverse transcription and was therefore termed the primer activation signal (PAS) (Fig. 1). In vitro, HIV-1 primer usage can be switched from tRNA₃^Lys to tRNA_1,2^Lys by the use of RNA templates containing PAS and PBS double mutations (4). In this study, we set out to test whether PAS-PBS double mutant viruses can stably replicate with a non-self tRNA primer.

MATERIALS AND METHODS

DNA constructs.

HIV-1 LAI viruses with mutations in the PBS were described previously (15). The PAS-PBS double mutants were made accordingly. All mutant constructs were verified by sequence analysis. The PAS-PBS-pro and -lys1,2 mutations are indicated in Fig. 3 and 4. The revertant R1 and R2 mutations (T126C and G3600A, respectively; +1 is the first nucleotide of the genomic HIV-1 RNA) were cloned via PCR mutagenesis into proviral DNA constructs. To clone the R1 mutation, we used the template pLAI-PBS_1,2^Lys (15). Oligonucleotides T7-1 and TA025 (5′-¹⁴⁰CTCTAGTTACGTGGGTCACACAACAGACGGG¹⁰⁹-3′) were used to amplify nucleotides 1 to 140 of the HIV-1 leader sequence (5). Oligonucleotides TA024 (5′-¹¹⁶GTTGTGTGACCCACGTAACTAGAGATCCCT¹⁴⁵-3′) and AD-GAG were used to amplify nucleotides 116 to 462 of the LAI-PBS_1,2^Lys genome (3). The mutated nucleotides are underlined in the primer sequences and the nucleotide positions are indicated with superscript numbers. The two PCR products were joined and amplified in a subsequent PCR with the T7-1 and AD-GAG primers. The R1 PCR fragment was sequenced, subsequently digested with HindIII and ClaI, and cloned into the Blue-5′LTR vector containing the XbaI-ClaI fragment of the pLAI proviral clone (36). The R1 mutated XbaI-ClaI fragment was finally cloned into pLAI-R37, a derivative of the full-length infectious clone pLAI (10). The mutant proviral construct was designated pLAI-R37-M-R1, in which M indicates the original PAS_1,2^Lys and PBS_1,2^Lys mutations (T128A, G129C, G190A, A192C, C193G, A194T, and A198G).

FIG. 3.

Evolution of PAS/PBS-pro mutant. SupT1 cells were transfected with 5 μg of the molecular clones. Breakthrough replication was observed in some cultures, and viruses could eventually be passaged onto fresh cells. The identity of the PBS motif is indicated as a function of the evolution time (A). The input mutant PBS is shown with open boxes, and wt revertants are shown with black boxes. (B) Culture number, day of harvest, and sequence of proviral DNA isolated from infected cells. The mutated PAS and PBS nucleotides are depicted in bold and are underlined. Nucleotide changes acquired during evolution are shown in white surrounded by a black box (“N” indicates a mixed sequence).

FIG. 4.

Evolution of PAS/PBS-lys1,2 mutant. See the legend for Fig. 3 for details. The two mutated PAS and PBS nucleotides are indicated in bold and are underlined. Mutations in the region just upstream of the PAS that were observed in some cultures may reflect G-to-A hypermutations. These transitions have been described previously for other leader revertant viruses and were therefore not analyzed further (8).

The R2 mutation (G3600A) was cloned similarly, although Blue-Pol-ApaSal was used as a template in the initial PCRs. This pBluescript-derived plasmid contains the ApaI-SalI fragment (nucleotides 1550 to 5265) of the pLAI proviral clone. This plasmid includes the 3′ part of the Gag coding sequence, all of the Pol and Vif coding sequences, and the 5′ part of the Vpr coding sequence. Oligonucleotides TA022 (5′-³⁴⁰⁰GAAATTATGGTACCAGTTAGA³⁴²⁰-3′) and TA021 (5′-³⁶¹⁵ATATTTACTTCTAATTCCGAATCCT³⁵⁹¹-3′) were used to amplify nucleotides 3400 to 3615 of the HIV-1 genome. In a separate reaction, oligonucleotides TA023 (5′-⁵³⁷²AAGGTCGACACCCAATTCTGAAATGG⁵³⁵⁰-3′) and TA020 (5′-³⁵⁹¹AGGATTCGGAATTAGAAGTAAATAT³⁶¹⁵-3′) were used to amplify nucleotides 3591 to 5372. In a final PCR, these PCR products were joined and amplified with primers TA022 and TA023. Upon sequence analysis, the PCR fragment was digested with NheI and NcoI and then cloned into Blue-Pol-ApaSal. The R2-containing BclI (2011)-SalI fragment was cloned into pLAI-R37 to obtain pLAI-wt-R2, into pLAI-R37-M to obtain pLAI-M-R2, and into pLAI-M-R1 to obtain the double revertant pLAI-M-R12 proviral construct.

Cells, virus replication, and evolution.

The SupT1 T-cell line was either transfected by electroporation with proviral DNA or infected with a virus stock as previously described in detail (3, 6). CA-p24 levels in the culture medium were determined with an enzyme-linked immunosorbent assay. The protocol used for virus evolution by a prolonged cell-free passage of virus onto fresh, uninfected SupT1 cells was described previously (37). Isolation of total cellular DNAs was performed by proteinase K treatment (3). The long terminal repeat leader region was PCR amplified with primers T7-1 and AD-GAG. The complete RT region was PCR amplified in two overlapping segments. The 5′ segment was amplified with the primers NB1 (5′-¹⁹⁶⁶AAAATGATAGGGGGAATTGG¹⁹⁸⁵-3′) and 3′RT-20 (5′-³⁰⁴³CTGCCAGTTCTAGCTCTGCTTC³⁰²²-3′), and the 3′ segment was amplified with 5′BRT (5′-²⁵⁸³GGGATGGAAAGGATCACC²⁶⁰⁰-3′) and 3′RT-22 (5′-³⁸⁹⁴AGGTTAAAATCACTAGCCATTGCTCTCC³⁸⁶⁶-3′). The PCR products were directly sequenced, thus providing the average sequence of the viral quasispecies (population sequence).

Competition and virus stability assays.

SupT1 cells were transfected as described above with a mixture of two proviral constructs (500 ng each) or were infected with the R1 or R12 virus (5 ng of CA-p24 per 5-ml culture). Competition assays were performed for at least three passages. The genetic stability of M-R1 and M-R12 was tested in cultures that were maintained for 101 to 125 days. Cell-free supernatant samples were taken at peak infection and passaged on fresh SupT1 cells every 7 to 10 days. Total cellular DNA was isolated and used as a template in a standard PCR to amplify the leader and partial RT regions. Oligonucleotides TA053 (5′-TGTAAAACGACGGCCAGTG¹GGTCTCTCTGGTTAGACCAG²²-3′) and AD-GAG were used to amplify the leader region. Primers TA052 (5′-TGTAAAACGACGGCCAGT³⁴⁷²CAGGGAGACTAAATTAGG³⁴⁸⁹-3′) and NB4 (5′-³⁸⁶⁵ATTACTGTGATATTTCTCATG³⁸⁴⁵-3′) were used to amplify nucleotides 3472 to 3845 of the RNA genome (RT region). Both PCR products were extended with M13-derived sequences present in the TA052 and TA053 primers. The products were sequenced with BigDye-labeled −21 M13 primers (5′-TGTAAAACGACGGCCAGT-3′) by use of an ABI Prism BigDye primer cycle sequencing ready reaction kit (Applied Biosystems).

RESULTS

Replication of PAS-PBS mutant viruses.

We constructed HIV-1 variants with PAS-PBS double mutations to enforce the use of the non-self primers tRNA^Pro and tRNA_1,2^Lys (Fig. 1). The mutations were introduced in the LAI molecular clone, a CXCR4-using primary HIV-1 isolate, and the two mutants were termed PAS/PBS-pro and PAS/PBS-lys1,2. These plasmids were transfected into the SupT1 T-cell line to monitor virus replication by the increase in CA-p24 production in the culture supernatant (Fig. 2). We included the wt LAI control and single PBS mutants (termed PBS-pro and PBS-lys1,2) (15). The 10-μg transfection reactions illustrated the ranking order of replication: wt > PBS-lys1,2 > PBS-pro > PAS/PBS-pro > PAS/PBS-lys1,2. The 5-μg transfection reactions confirmed the severe replication defect of all mutants except PBS-lys1,2. The relatively efficient replication of this PBS-lys1,2 single mutant was reported previously (15). The addition of the PAS-lys1,2 mutation decreased virus replication. The single PBS-pro mutant showed a significant replication defect, and inclusion of the PAS-pro mutation further weakened virus replication. The negative impact of the PAS mutation seemed less dramatic for the PBS-pro virus than for the PBS-lys1,2 virus. In any case, the PAS-PBS double mutants did not rescue the replication of the PBS single mutants.

FIG. 2.

Replication of wt HIV-1 LAI and PAS and PBS mutated viruses. SupT1 cells were transfected with 10 μg (A) or 5 μg (B) of the proviral constructs. CA-p24 production was measured in the culture medium at several days posttransfection.

Evolution of PAS-PBS mutant viruses.

We next set out to obtain faster replicating revertant viruses for the double mutants. A priori, two evolution routes could be envisaged. First, the virus could restore tRNA₃^Lys usage by reversion to the wt PBS motif. PBS reversion is a frequent event that is driven by mispriming of tRNA₃^Lys on the mutant PBS, followed by inheritance of the fully wt PBS sequence from the tRNA primer. However, priming by tRNA₃^Lys was inhibited at two levels by the PAS-PBS mutations, at the levels of tRNA annealing and initiation of reverse transcription. In contrast, the usage of the new primer was enhanced by these mutations. These combined effects may block the wt reversion route. Second, the virus could optimize replication with the new tRNA primer, and adaptive changes could thus be acquired in the viral RT enzyme or other cofactors. This latter evolutionary route is very interesting because it may reveal critical amino acid residues within RT that play an important role in selective tRNA usage.

We maintained 21 and 17 independent cultures with the PAS/PBS-pro and PAS/PBS-lys1,2 mutants, respectively, and breakthrough replication was monitored within a few weeks in 4 and 3 cultures. Viruses were passaged repeatedly. Cell samples were taken and used to amplify proviral DNAs for subsequent sequence analysis. We initially focused on the leader RNA that encodes the PAS and PBS motifs. The evolutionary routes are summarized in Fig. 3A and 4A. Partial leader sequences of the revertant PAS/PBS-pro and PAS/PBS-lys1,2 viruses are shown in Fig. 3B and 4B. Sequence analysis of the PAS/PBS-pro viral DNA revealed reversion to the wt PBS-lys3 within 47 days in three of the four positive cultures (Fig. 3, cultures P8, P11, and P18). The P18 virus contained an additional mutation immediately upstream of the PBS (G181A). Such changes in the PBS-flanking residues have been described for other viral reversion events (8, 15, 31, 34). The fourth positive culture (P14) appeared more interesting because the PAS/PBS-pro motifs were stably maintained. This culture was split at day 47 into six subcultures that were monitored over time (Fig. 3A). All samples eventually reverted back to the wt PBS-lys3 sequence at day 116 (Fig. 3B). In addition, we observed reversion to the wt PAS-lys3 by means of the G128U back mutation in two cultures (P14.3 and P14.6), and the population-based sequence showed a mixed sequence for one culture (P14.5). This finding confirms the importance of having a fully complementary PAS motif. In conclusion, we were unable to select a replicating virus that stably used the tRNA^Pro primer.

Identification of a stable virus using tRNA_1,2^Lys.

We speculated that the evolutionary jump in primer usage from tRNA₃^Lys to tRNA_1,2^Lys would be easier for HIV-1 because these primers are very similar (Fig. 1). Three cultures showed syncytia over time, indicating replication of PAS/PBS-lys1,2 revertants (Fig. 4A, cultures L3, L4, and L8). Variant L8 had reverted to wt PBS-lys3 at day 47 (Fig. 4B). The virus in culture L3 showed a mixed wt-mutant PBS sequence at day 73 and had reverted back to PBS-lys3 at day 97. The input PAS/PBS-lys1,2 motifs were maintained only in culture L4. The L4 virus continued to replicate with a tRNA_1,2^Lys primer up to day 75, but a mixed wt-mutant PBS sequence was detected at day 97. We used the day 47 sample to restart a second round of evolution by infecting six fresh SupT1 cell cultures in parallel. All six cultures became infected with L4-derived variants that stably maintained the mutant PBS-lys1,2 up to day 116. This observation suggests that the L4 viruses acquired at least one adaptive change outside the PBS motif to accommodate tRNA_1,2^Lys in the second round of evolution. An interesting change was observed within the PAS motif in five of six revertants (Fig. 4B). The mutant PAS-lys1,2 motif differed from the wt PAS-lys3 element at two nucleotide positions. These nucleotides did not revert, but an additional PAS residue was altered (U126C). This is interesting, because a U-G base pair is replaced with a stronger C-G base pair in the context of the PAS-anti-PAS interaction with tRNA_1,2^Lys.

The original L4 revertant did not yet contain the U126C PAS adaptation at day 47. Nevertheless, this virus replicated relatively efficiently, suggesting that at least one other critical mutation must be present elsewhere in the viral genome to facilitate reverse transcription primed by the non-self tRNA_1,2^Lys molecule. Replication studies confirmed the efficient replication of the L4-d47 and L4-d76 viruses (results not shown). Thus, we assumed that the L4-d47 virus has a major adaptive change elsewhere in the viral genome that allows efficient tRNA_1,2^Lys usage. Because no significant changes were present in the leader domain surrounding the PAS and PBS motifs (results not shown), we sequenced the complete RT gene. One mutation (G3600A) within the RT gene of the L4-d47 virus was identified. This nucleotide transition led to an amino acid change (G490E) in the RNase H domain of RT. The mutation was stably maintained in later samples (L4.1, L4.2, and L4.3 were sequenced). The G490 residue is absolutely conserved among virus isolates of all HIV-1 subtypes (data not shown), and the mutation was therefore further analyzed.

Role of U126C and G3600A mutations in stable tRNA_1,2^Lys primer usage.

To confirm that the PAS mutation U126C (designated R1) and the RT mutation G3600A (designated R2) increase the replication of the PAS/PBS-lys1,2 mutant virus and stabilize the usage of the tRNA_1,2^Lys primer, we cloned these mutations into the original PAS/PBS-lys1,2 mutant (further designated with an “M,” yielding M-R1, M-R2, and M-R12 [a mutant with both R1 and R2 changes]). In addition, the RT mutation was tested in a wt virus background (wt-R2). Molecular clones were constructed and transfected into SupT1 cells to monitor virus replication (Fig. 5A). The R1 PAS change significantly increased the replication of the tRNA_1,2^Lys-using virus, even in the absence of the R2 reversion in RT. No gross effect of R2 on the replication of the wt and mutant viruses could be observed. More sensitive virus competition assays were performed to measure small differences in virus replication (Table 1). From these data, we determined the following ranking order: wt ≥ wt-R2 ≫ M-R1 ≥ M-R12 > M-R2 > M. Apparently, the R2 reversion significantly improved replication of the M virus, because M-R2 efficiently outcompeted M after a single passage. Interestingly, the same R2 mutation slightly decreased the replication capacity of the wt and M-R1 viruses.

FIG. 5.

Replication of wt, PAS/PBS-lys1,2 mutant, and cloned revertant viruses. SupT1 cells were transfected with 2 μg of the proviral constructs (A) or were infected with equal amounts of viruses (1 ng of CA-p24 per 5 ml of culture) (B). CA-p24 production was measured in the culture medium for several days. M indicates the original PAS/PBS-lys1,2 mutant, R1 indicates the U126C reversion in the PAS motif, and R2 indicates the G490E reversion in the RNase H domain of RT.

TABLE 1.

Comparisons of viruses for fitness

Competing virusesa	Fittest virus	Passage no.b
wt and wt-R2	wt	4
M and M-R1	M-R1	1
M-R1 and M-R12	M-R1	2
M and M-R12	M-R12	1
M and M-R2	M-R2	1

^aMixed 1:1 at start of experiment.

^bPassage at which the fittest virus was >95% of the total virus.

To confirm these results, we monitored virus replication in SupT1 cells that were inoculated with a limiting amount of virus (Fig. 5B). The infection data confirmed the ranking order observed in the competition experiments and indicated that both the individual R1 mutation and the individual R2 mutation improve the replication of the PAS/PBS-lys1,2 mutant. The impact of the R1 mutation on the restoration of virus replication was significantly larger than that of the R2 mutation. R1 may also stabilize tRNA_1,2^Lys usage more potently than R2. Virus L4-d47, containing only R2 and not R1, still reverted to the tRNA₃^Lys-using variant L4-d97 (Fig. 4). However, L4-d47 remained stable in its tRNA_1,2^Lys usage upon acquiring R1 (L4-d116 variants 1, 2, 3, 5, and 6) in the second evolution round.

To confirm that R1 and R2 can stabilize tRNA_1,2^Lys usage, we passaged the molecularly cloned M-R1 and M-R12 viruses in six cultures per virus for a period of 101 to 125 days. The sequences of the leader RNAs and the RT genes were analyzed. Most importantly, all viruses continued to use tRNA_1,2^Lys. In a single M-R12 culture, the R2 mutation was mixed with the wt sequence after 125 days, confirming that M-R1 is a relatively fitter virus than M-R12. The R1 reversion was maintained in the entire quasispecies population in all cultures. These data suggest that R1 effectively prevents the switch to tRNA₃^Lys usage. A schematic overview of the evolution events and their effect on viral fitness is shown in Fig. 6. Back mutation to tRNA₃^Lys usage occurs frequently for the M mutant and is still possible for the M-R2 revertant but is not observed for the M-R12 revertant and the more fit M-R1 revertant.

FIG. 6.

Schematic representation of the gain in relative viral fitness during forced evolution of the PAS/PBS-lys1,2 mutant (M). Arrows indicate reversion events. The thickness of the arrows indicates the chance of reversion. The slope of the arrow indicates the gain of fitness. M indicates the original PAS/PBS-lys1,2 mutant, which predominantly reverts to a wt, tRNA₃^Lys-using virus. In the L4 culture (Fig. 4), M acquired an R2 reversion in the RT gene (G3600A). The M-R2 virus is fitter, but it can still revert back to the wt. Viral fitness increases most significantly upon acquisition of the R1 mutation in the PAS element. The M-R12 double revertant is stable in prolonged cultures and does not revert back to the wt. From competition experiments, we concluded that M-R1 replicates slightly more efficiently than M-R12. Consistent with this, the M-R12 virus lost the R2 mutation in a single culture. M-R1 stably maintained the PBS-lys1,2 sequence in all cultures, highlighting the importance of the PAS motif in tRNA_1,2^Lys usage.

DISCUSSION

To obtain an HIV-1 virus that stably uses a primer other than tRNA₃^Lys for reverse transcription, we altered the PAS and PBS sequences in the HIV-1 leader RNA. Earlier experiments from our laboratory and other laboratories had shown that altering the PBS sequence alone was not sufficient to stably switch tRNA usage (15, 40, 63). Additional studies showed that an upstream RNA PAS motif is critically involved in tRNA-primed reverse transcription (3, 6). The PAS motif exerts its function not by enhancing tRNA annealing to the PBS, but by activating the initiation of reverse transcription. The PAS motif engages in a base-pairing interaction with the complementary anti-PAS sequence in the tRNA, which likely results in the formation of a higher order RNA structure that is suitable for reverse transcription (Fig. 7). By adaptation of both the PAS and PBS motifs, the HIV-1 leader could be changed to accommodate the tRNA_1,2^Lys primer for the initiation of reverse transcription in vitro (4). We reasoned that the single PBS mutants could not optimally initiate non-self tRNA-primed reverse transcription in vivo because of the disrupted PAS-anti-PAS base-pairing interaction. We therefore constructed PAS-PBS double mutants that could properly accommodate either tRNA^Pro or tRNA_1,2^Lys as a reverse transcription primer. The replication efficiencies of these mutant viruses were compared to those of single PBS mutants. The single PBS-lys1,2 mutant replicated relatively efficiently, but the single PBS-pro mutant showed a significant replication defect. The addition of a PAS mutation further decreased virus replication, indicating that PAS adaptation does not rescue replication of the PBS single mutants. Rescue would be expected if no other viral factors are implicated in selective tRNA usage. However, several viral factors that closely interact with the tRNA primer, such as the RT enzyme, may be involved in primer selection. Thus, other incompatibility problems may not be solved by the imposed usage of a non-self tRNA primer in the PAS-PBS double mutants. The latter scenario is consistent with the experimental results.

FIG. 7.

Model for reverse transcription initiation on wt, PAS/PBS-lys1,2 mutant, and R1 revertant templates. (A) The secondary structures of the PBS region of the HIV-1 RNA genome and tRNA primers are shown schematically (black and orange lines, respectively) AC, anticodon loop; D, D loop. The tRNA primer anneals with its 3′-terminal 18 nucleotides to the PBS (the PBS and anti-PBS sequences are shown in green). An additional interaction between PAS and anti-PAS (orange) is required to activate the initiation of reverse transcription. These interactions are indicated for the wt leader with tRNA₃^Lys and for the PAS/PBS-lys1,2 mutant (M) and the M-R1 revertant with tRNA_1,2^Lys (B). The sequence differences between PAS and PBS sequences of wt, M, and R1 leader RNAs and the anti-PAS and anti-PBS sequences of tRNA₃^Lys and tRNA_1,2^Lys are indicated in bold and marked with dots. The R1 reversion (U126C; indicated by an arrowhead) stabilizes the PAS-anti-PAS interaction (replacement of a weak U-G base pair with a very stable C-G base pair).

The tRNA^Pro-using virus was not stable during prolonged passaging, despite multiple attempts. Its PBS quickly reverted to the wt tRNA₃^Lys complementary sequence. This negative result may be explained by a lack of specific packaging of tRNA^Pro into virus particles. tRNA^Lys isoacceptors are specifically packaged in wt virus particles via an interaction with Gag-Pol and the cellular lysyl-tRNA synthetase (11, 27, 35, 49, 50). Both Gag and Pol sequences are required for the formation of this tRNA-packaging complex. Therefore, it may be difficult for HIV-1 to acquire the ability to package tRNA^Pro. Adaptations in both Gag and Pol may be required to allow prolyl-tRNA synthetase and tRNA^Pro packaging and appropriate primer annealing onto PBS-pro. At present, it is unclear how the specificity of tRNA synthetase packaging can be altered.

We were more successful in switching the reverse transcription primer from tRNA₃^Lys to tRNA_1,2^Lys. Since tRNA_1,2^Lys isoacceptors are selectively packaged via the Gag-Pol-lysyl-tRNA complex, the PAS/PBS-lys1,2 virus was not expected to encounter difficulties in tRNA_1,2^Lys packaging. Nevertheless, the PAS/PBS-lys1,2 virus was severely affected in its replication efficiency and could still revert to a tRNA₃^Lys-using virus, but the acquisition of a second-site reversion in the PAS motif (U126C [R1]) stabilized the tRNA_1,2^Lys usage. In addition, a second-site reversion was located in the RNase H domain of the RT protein (G490E [R2]). The R2 reversion improved the stability of tRNA_1,2^Lys usage, although not completely. Both R1 and R2 reversions were introduced in the original PAS/PBS-lys1,2 virus and improved the replication of the PAS/PBS-lys1,2 mutant virus.

Others have described similar non-tRNA₃^Lys-using viruses, but no adaptive changes in the PAS motif or any of the viral proteins have been identified (18, 31-34, 43, 62, 65, 66). These authors did report changes in the A-rich loop located upstream of the PBS. The A-rich loop has been suggested to engage in a base-pairing interaction with the anticodon loop of the priming tRNA. The anticodon loop sequences of tRNA₃^Lys and tRNA_1,2^Lys differ by only one nucleotide (position 34; Fig. 1). For a restoration of this putative base pairing, one would expect an A170G change to occur. In our evolution studies, which lasted >3 months, we did not observe this or an equivalent adaptation of the A-rich loop to accommodate the tRNA_1,2^Lys molecule. These results indicate that full base-pairing potential between these RNA sequences is not absolutely required for tRNA selection. In general, mutant-revertant data should be interpreted with caution because this part of the viral genome encodes multiple, overlapping functions. The RNA secondary structure surrounding the A-rich loop is important for viral replication and reverse transcription (5, 7), and these sequences are required for proper integration of the provirus into the host genome (61).

The R1 reversion in the PAS motif stabilized base pairing with the anti-PAS motif in the tRNA_1,2^Lys primer (Fig. 7). This reversion increased viral replication and stabilized tRNA_1,2^Lys usage, indicating that the designed PAS-anti-PAS interaction of the original mutant was not sufficiently stable and was therefore suboptimal. The PAS motif was recently questioned by others (21), despite a wealth of experimental evidence, including the successful switch to tRNA_1,2^Lys usage in vitro by a simultaneous change of the PAS and PBS sequences (3, 4, 6, 23). Also, an equivalent interaction has been described for HIV-2 and avian sarcoma virus (1, 2, 12, 20, 55). Our results obtained with the evolved PAS/PBS-lys1,2 virus clearly confirm the importance of the PAS motif in tRNA usage in vivo. Fine-tuning of the PAS-anti-PAS interaction strength turns out to be the most decisive change for yielding a virus that stably uses tRNA_1,2^Lys. The R1 mutation also improves base-pairing potential with the natural tRNA₃^Lys primer. A previous demonstration that this mutation enhances tRNA₃^Lys usage in vitro is consistent with this notion (4).

The R2 reversion in the RT protein (G490E) specifically improved the replication and stability of the tRNA_1,2^Lys-using virus. This amino acid change did not significantly affect viral replication in a wt and R1 background. Apparently, R2 only stimulates viral replication when the tRNA-vRNA interaction is suboptimal. Glycine 490 is conserved in the RT gene of the related SIV and HIV-2 viruses that also use the tRNA₃^Lys primer. Interestingly, residue 490 (shown in yellow in Fig. 8) protrudes from the RNase H domain in the X-ray structure of the RT p51-p66 heterodimer. Thus, residue 490 seems ideally positioned to act as a “gatekeeper” for the cleft between the RNase H and RT domains. We speculate that part of the tRNA molecule binds in this cleft, which is consistent with previous cross-linking studies (54). Because residue 490 is a candidate for the discrimination between the self tRNA₃^Lys and non-self tRNA_1,2^Lys primers, we juxtaposed the most variable tRNA domain in Fig. 8 (the stem of the anticodon [AC] hairpin, shown in red). These results imply that the RNase H domain of HIV-1 RT is involved in selective tRNA binding. Further in vitro studies with recombinant RT may shed more light on the specific role of the RNase H domain in tRNA selection.

FIG. 8.

Multiple interactions between HIV-1 RT enzyme, RNA genome, and tRNA₃^Lys primer. (A) X-ray structure of the p51-p66 heterodimer of the HIV-1 RT enzyme with a double-stranded DNA duplex (26). The enzymatically active p66 domain is shown in blue, and the RNase H domain is shown in yellow. The catalytically active residues in both domains are shown as red dots. The p51 subunit is shown in green. The figure was drawn with Molscript (38) and Raster3D (53) software, using coordinates from the protein data bank (entry 1 HMI [26]). (B) Cartoon of the complex between the HIV-1 RT enzyme, the viral RNA genome, and the self tRNA₃^Lys primer. The PBS-anti-PBS base-pairing interaction is indicated. We also marked other interaction domains, including the U-rich anticodon (AC) loop of tRNA₃^Lys, an A-rich segment in the viral RNA genome (green), and the anti-PAS and PAS motifs (yellow). We selected an HIV-1 variant that switched its primer usage from tRNA₃^Lys to tRNA_1,2^Lys by simultaneous mutation of the PBS and PAS motifs. This virus acquired a point mutation in the RNase H domain (Gly490Glu; shown in yellow), which implicates this RT domain in primer binding. The sequence of the anticodon stem is the most variable region between tRNA₃^Lys and tRNA_1,2^Lys (marked in red; see Fig. 1 for details), and this tRNA domain may interact with the gatekeeper residue 490. The anticodon hairpin is proposed to dock in the cleft between the RNase H and RT domains of the p66 subunit.