tRNA Isoacceptor Preference prior to Retrovirus Gag-Pol Junction Links Primer Selection and Viral Translation

Abstract

An essential step in the replication of all retroviruses is the capture of a cellular tRNA that is used as the primer for reverse transcription. The 3′-terminal 18 nucleotides of the tRNA are complementary to the primer binding site (PBS). Moloney murine leukemia virus (MuLV) preferentially captures tRNA^Pro. To investigate the specificity of primer selection, the PBS of MuLV was altered to be complementary to different tRNAs. Analysis of the infectivity of the virus and stability of the PBS following in vitro replication revealed that MuLV prefers to select tRNA^Pro, tRNA^Gly, or tRNA^Arg. Previous studies from our laboratory have suggested that tRNA primer capture is coordinated with translation. Coincidentally, a cluster of proline, arginine, and glycine precedes the Gag-Pol junction of MuLV. Human immunodeficiency virus type 1 (HIV-1), which prefers as the primer, can be forced to utilize tRNA^Met, , tRNA^His, or tRNA^Glu, although these viruses replicate poorly. Codons for methionine, lysine, histidine, or glutamic acid are found prior to the Gag-Pol frameshift site. HIV-1 was mutated so that the 5 lysine codons prior to the Gag-Pol frameshift region were specific for . HIV-1 forced to use as the primer, with the mutation of codons specific for prior to the Gag-Pol junction, had enhanced infectivity and replicated similarly to the wild-type virus. The results demonstrate that codon preference prior to the Gag-Pol junction influences primer selection and suggest a coordination of Gag-Pol synthesis and acquisition of the tRNA primer required for retrovirus replication.

The hallmark of retrovirus replication is the process in which the viral RNA genome is converted to a DNA intermediate prior to integration into the host cell chromosome (4, 41). Reverse transcription is catalyzed by a virally encoded enzyme, reverse transcriptase (RT), which requires a cellular tRNA that acts as a primer. The 3′-terminal nucleotides of the cellular tRNA are complementary to an 18-nucleotide region on the viral RNA genome designated the primer-binding site (PBS) (35, 36, 40). The PBS is regenerated following copying of the tRNA primer used for initiation of reverse transcription during plus-strand DNA synthesis (39, 43).

The tRNA primer selected for replication differs between retroviruses (29, 30). Moloney murine leukemia virus (MuLV) selects a tRNA^Pro from the intracellular milieu as the primer for reverse transcription (12). In contrast, lentiviruses such as human immunodeficiency virus (HIV) and simian immunodeficiency virus preferentially select as the primer (29, 30). It is not clear why retroviruses prefer to select certain tRNA primers. Since the PBS is regenerated following reverse transcription, retroviruses could, in theory, select any number of tRNAs as primers. Previous studies have demonstrated for both MuLV and HIV type 1 (HIV-1) that alteration of the PBS to be complementary to different tRNAs allows these viruses to utilize these tRNAs as primers for reverse transcription (11, 20, 21, 33, 44, 47). However, even though the viruses can utilize these tRNAs, they invariably replicate more slowly than the wild-type virus, indicating a distinct preference for certain tRNAs. Recent studies with HIV-1 have suggested that the preference of HIV-1 for as the primer is due to the inclusion of lysyl-synthetase within virions (7, 18, 19). The lysyl-synthetase interacts with the HIV-1 Gag protein and is incorporated specifically into HIV-1 virions to facilitate the inclusion of . However, previous studies have also shown that is found at quantities comparable to those of in HIV-1 virions (25, 28, 48). Mutation of the HIV-1 PBS to be complementary to , though, does not result in the virus stably utilizing this tRNA (1, 20, 21). HIV-1 can be forced to use by additional mutations in U5 (A-loop) that make it complementary to the anticodon of (20, 21, 46). Previous studies using both chemical and enzymatic analysis have shown that the A-loop region of HIV-1 interacts with the anticodon region of in the initiation complex for reverse transcription (16, 17). Even with these mutations, the virus using does not replicate at levels comparable to those of the wild-type virus (31). Interestingly, the prolyl-tRNA synthetase is not found in MuLV virions, suggesting that inclusion of specific synthetases in the virion is not a universal mechanism for retrovirus primer selection (7).

To further understand the preferences for selection of specific tRNAs for retrovirus replication, we have utilized a genetic approach to generate viruses in which the PBS has been altered to be complementary to tRNAs other than that found in the wild-type genome. Previous studies from our laboratory have utilized this approach to generate HIV-1 variants that stably utilized tRNA^His, tRNA^Glu, tRNA^Met, and (11, 20-22, 44, 47). We have also reported on the alteration of the PBS of MuLV to be complementary to tRNAs other than tRNA^Pro and have found that MuLVs will utilize alternative tRNAs for replication (33). However, both MuLV and HIV-1 variants that are forced to utilize non-wild-type tRNAs replicate more slowly (with lower infectivity) than wild-type virus. Thus, even though both MuLV and HIV-1 have access to a variety of tRNAs that can be utilized for replication, they have evolved to select specific tRNAs for high-level replication. To further understand the nuances of retrovirus primer selection, we have extended our genetic approach to generate MuLV mutants with a variety of PBSs complementary to alternative tRNAs. Our results show that MuLV mutants have preferences for tRNA^Pro, tRNA^Gly, and tRNA^Arg for replication. Previous studies have suggested a link between viral translation and primer selection (23, 24, 34). A search for proline, arginine, and glycine revealed a cluster just prior to the Gag-Pol junction. HIV-1 can utilize , tRNA^His, tRNA^Met, or tRNA^Glu as an alternative primer. Coincidentally, codons for these tRNAs are located prior to the ribosomal frameshift site at the Gag-Pol junction. Alteration of the codon use in this region of HIV-1 to favor , in conjunction with the mutation in the A-loop and PBS to be complementary to , resulted in an HIV-1 variant that had infectivity comparable to that of the wild-type virus. The results of these studies suggest that the PBS and the codon usage at or near the Gag-Pol frameshift region can strongly influence the selection of the tRNA primer used in viral replication, and they support a model for coordination between Gag-Pol synthesis and the selection of the tRNA primer used for virus replication.

MATERIALS AND METHODS

Plasmid construction.

p63-2 is an infectious provirus clone of MuLV (3). p63-2(Arg), p63-2(Lys,3), and the pCR4 blunt transfer plasmid have been described previously (33). p63-2(Lys1,2), p63-2(Glu), p63-2(Cys), p63-2(Met), p63-2(Tyr), p63-2(Asp), p63-2(Phe), p63-2(Arg1), and p63-2(Arg2) are provirus clones whose PBSs exhibit complementarity to the 3′ ends of , tRNA^Glu, tRNA^Cys, elongator tRNA^Met, tRNA^Tyr, tRNA^Asp, tRNA^Phe, and two alternate tRNA isoacceptors of arginine, respectively (see Fig. 1B), and were constructed using site-directed mutagenesis by PCR overlap extension (14). Briefly, first-round PCR amplicons were generated by pairing the PBSFor primer (34) with mutagenic reverse primers and by pairing the PBSRev primer (34) with mutagenic forward primers, using Platinum Taq DNA polymerase (Invitrogen, Carlsbad, CA). The mutagenic primers are GlyF (5′-GTGCATTGGCCGGGAACGGGAGACCCCTGCCC-3′), GlyR (5′-GTTCCCGGCCAATGCACCAAATGAAAGACCCCCG-3′), Arg1F (5′-GCGAGCCAGCCAGGAGCGGGAGACCCCTGCCC-3′), Arg1R (5′-GCTCCTGGCTGGCTCGCCAAATGAAAGACCCCCG-3′), CysF (5′-GAGGGGGCACCTGGATCGGGAGACCCCTGCCC-3′), CysR (5′-GATCCAGGTGCCCCCTCCAAATGAAAGACCCCCG-3′), Arg2F (5′-GCGACTCTGGTGGGATCGGGAGACCCCTGCCC-3′), Arg2R (5′-GATCCCACCAGAGTCGCCAAATGAAAGACCCCCG-3′), Lys1,2F (5′-GCGCCCAACGTGGGGCCGGGAGACCCCTGCCC-3′), Lys1,2R (5′-GGCCCCACGTTGGGCGCCAAATGAAAGACCCCCG-3′), GluF (5′-GTTCCCTGACCGAGAACGGGAGACCCCTGCCC-3′), GluR (5′-GTTCTCGGTCAGGGAACCAAATGAAAGACCCCCG-3′), MetF (5′-GTGCCCCGTGTGAGGACGGGAGACCCCTGCCC-3′), MetR (5′-GTCCTCACACGGGGCACCAAATGAAAGACCCCCG-3′), PheF (5′-GTGCCGAAACCCGGGACGGGAGACCCCTGCCC-3′), PheR (5′-GTCCCGGGTTTCGGCACCAAATGAAAGACCCCCG-3′), TyrF (5′-GTCCTTCGAGCCGGAACGGGAGACCCCTGCCC-3′), TyrR (5′-GTTCCGGCTCGAAGGACCAAATGAAAGACCCCCG-3′), AspF (5′-GCTCCCCGTCGGGGAACGGGAGACCCCTGCCC-3′), and AspR (5′-GTTCCCCGACGGGGAGCCAAATGAAAGACCCCCG-3′).

FIG. 1.

MuLV 5′ nontranslated region and PBS mutants. (A) Schematic of the MuLV 5′ nontranslated region with a PBS that is complementary to the 3′ 18 nucleotides of tRNA^Pro. The viral genomic RNA, which is the mRNA for Gag and Gag-Pol, is shown. The PBS is located between nucleotides 146 and 163. (B) PBS mutants. The PBS of MuLV was changed to be complementary to the 3′ 18 nucleotides of the designated tRNAs (mutants).

First-round amplicons were then joined and amplified using PBSFor and PBSRev in a second round of PCRs. Second-round amplicons were digested with XmaI and AatII and subcloned into the pCR4 blunt transfer plasmid. The integrity of the inserted region was verified by sequencing. PciI/BsrGI fragments were subcloned into p63-2 to generate the full-length, PBS-mutated proviruses.

To construct HIV-1 mutants with altered codons near the Gag-Pol frameshift region, pNL4-3 (2) was partially digested with BclI. The 10,438-bp provirus-containing fragment was gel extracted and cloned into the BamHI site of pUC19. Several codons upstream of the frameshift site were changed to codons by PCR amplification of the resulting plasmid using 5′-GGATCCGGGCCCCTAGGAAGAAGGGCTGTTGGAAGTGTGGAAAGGAAGGA CACCAAATGAAGGATTG-3′ and 5′-GAATTCATTTCTGTACAAATTTCTACTAATGC-3′ as primers. The amplicon was digested with ApaI and BclI and cloned into the parent plasmid to create pNL4-3(F5). The integrity of the inserted region was verified by sequencing. PBS-containing HpaI/BssHII fragments were excised from pHXB2(Lys1,2) and pHXB2(Lys1,2AC) (21) and subcloned into the corresponding region of pNL4-3(F5) to create pNL4-3(Lys1,2F5) and pNL4-3(Lys1,2-ACF5). pNL4-3(Lys1,2) and pNL4-3(Lys1,2-AC) have been described previously (31).

Cell culture and DNA transfections.

Uninfected NIH 3T3 or Chinese hamster ovary (CHO) cells were maintained in Dulbecco's modified Eagle medium supplemented with 1% antibiotic-antimycotic and 10% fetal bovine serum (FBS). Infected cells were propagated in the absence of antimycotic. CHO cells were grown in a medium supplemented with proline to a final concentration of 100 μg/ml. 293T cells were transfected using polyethylenimine as described previously (24). CHO cells were transfected in 6-well plates at approximately 40% confluence, with complexes of 2 μg provirus DNA and 6 μl FuGENE 6 transfection reagent per well according to the manufacturer's instructions (Roche, Mannheim, Germany).

For HIV-1, 293T cells were transfected with plasmids containing different HIV provirus genomes by using methods described previously (31). Transfections were done utilizing FuGENE transfection reagent according to the manufacturer's protocol. The transfection mixtures were incubated overnight at 37°C, and the medium was then replaced with fresh Dulbecco's modified Eagle medium containing 10% FBS. After 48 h, all the supernatants were collected and stored at −80°C until further analysis. The supernatants from transfected cells were assayed for HIV-1 p24 antigen and infectious virus using the JC53-BL assay (10).

Analysis of MuLV and HIV-1 growth kinetics and infectivity.

Infectious MuLV was produced by transfecting 293T cells with provirus plasmid DNA. The medium was replaced 1 day posttransfection, and supernatants were collected the following day. Supernatants were clarified by centrifugation, treated with polybrene (final concentration, 8 μg/ml), passed through 0.45-μm-pore-size filters, and used to infect subconfluent NIH 3T3 cells in 75-cm² flasks. In several instances, infectious virus was transferred to NIH 3T3 cells by cocultivating with transfected CHO cells to prevent the need for numerous rounds of replication that might lead to evolution of the PBS. Following 1 week of cocultivation, proline supplementation was terminated to eliminate the CHO cells. Supernatants obtained from the resulting infected NIH 3T3 producer cells were used to assess viral growth kinetics (33). Infectivity was determined by counting of fused XC cells as previously described (33).

HIV-1 replication was determined using human peripheral blood mononuclear cells (PBMCs) as previously described. The PBMCs were stimulated using recombinant interleukin 2 and phytohemagglutinin (PHA) and were maintained as previously described. Infections were performed by inoculating 20 × 10⁶ PHA-stimulated PBMCs with the volume of transfection supernatant containing the desired quantity of infectious virus as determined by the JC53-BL assay, which measures infectious HIV-1; infection of JC53-BL cells, which have β-galactosidase and luciferase under the control of a lentivirus long terminal repeat, results in production of Tat, which induces transcription of β-galactosidase and luciferase (10). The virus was allowed to adsorb for 2 h at 37°C under 5% CO₂. Virus-PBMC mixtures were transferred to 25-cm² flasks and the final volumes adjusted to 10 ml with RPMI-1640 containing 15% FBS and 30 U per ml of interleukin-2 as previously described (31). Infected PBMC cultures were maintained for 4 weeks by replacing half the volume of medium every 3 to 5 days without removing PBMCs. Every 7 days, 1 ml of cell suspension was removed and centrifuged in an Eppendorf microcentrifuge at 16,000 × g for 2 min. The supernatants were analyzed for virus by an HIV p24 antigen enzyme-linked immunosorbent assay (ELISA) and a JC53-BL assay (10). Cell pellets were stored at −80°C until processing for isolation of high-molecular-weight (HMW) DNA for analysis of the PBS. Every 14 days, additional PHA-stimulated PBMCs (5 × 10⁶) were added to each culture.

Analysis of the provirus PBS region.

At specified time points during MuLV infection, HMW DNA was obtained from infected cells using the Wizard genomic DNA purification kit (Promega) according to the manufacturer's instructions. Provirus DNA encompassing the PBS region was amplified by PCR using PBSFor and PBSRev as primers and the HMW DNA as the template. PBSRev was used to sequence the PCR product directly. Sequences were analyzed using ClustalX, version 1.5b. The PBS sequences were checked against several databases: the BLAST service provided by the National Center for Biotechnology Information, the tRNA search engine provided by Bayreuth University (http://www.staff.uni-bayreuth.de/∼btc914/search/index.html), and the Mus musculus tRNA alignments provided by the University of California at Santa Cruz (http://lowelab.ucsc.edu/GtRNAdb/Mmusc/Mmusc-align.html). Note that the numbers given for tRNA^Arg (e.g., tRNA^Arg, , etc.) are used to represent differences for clarification in this paper; the 3′-terminal 18 nucleotide sequences can be found in the databases but do not contain the numbers.

Cell pellets from HIV-1-infected PBMC cultures were processed for isolation of HMW genomic DNA as previously described (31). Approximately 2 μg of each genomic DNA sample was PCR amplified using primers bracketing the U5-PBS region. Recombinant Taq polymerase was used for PCR amplification. The resulting PCR products were isolated and either directly sequenced or cloned into TA vectors prior to DNA sequencing. Analysis of individual TA clones was then used to derive PBS sequences as previously described (31).

RESULTS

Replication of MuLVs with PBSs complementary to alternative tRNAs.

The MuLV genome contains an 18-nucleotide sequence, the PBS, complementary to the 3′ nucleotides of cellular tRNA^Pro (12) (Fig. 1A). In a previous study, we showed that MuLV with a PBS made complementary to exhibited significantly reduced infectivity and delayed growth kinetics relative to those of the wild-type virus and adapted to use tRNA^Arg before reverting, through recombination, to the wild-type PBS. The results suggested that MuLV, like HIV-1, has a distinct preference for certain tRNAs (33). To further explore the preference for certain tRNAs, we made additional PBS mutants of MuLV corresponding to a variety of tRNAs (Fig. 1B).

To characterize the effects of the PBS mutations on MuLV replication, supernatants from infected NIH 3T3 producer cells were first normalized for RT activity and then used to infect fresh NIH 3T3 cells. Several patterns of replication were seen (Fig. 2A and B). In the first, we found replication similar to that of the wild-type virus but with a delay in the peak of virus production. Viruses with this pattern of replication had a PBS complementary to tRNA^Gly or . Viruses with a PBS complementary to tRNA^Arg or tRNA^Cys grew slightly more slowly and plateaued at lower levels (as determined by supernatant RT). Another replication pattern was seen for viruses with a PBS complementary to , , , or tRNA^Glu, where there was a considerable delay in replication followed by an increase in virus to levels approaching that of the wild type. Finally, viruses with a PBS complementary to tRNA^Asp, tRNA^Phe, tRNA^Met, or tRNA^Tyr did not replicate to detectable levels during the time period examined (Fig. 2), although replication was detected at extended culture times (data not shown). To further expand on the effects of the PBS mutations on virus replication, we next compared the infectivities of the viruses. Supernatants from infected NIH 3T3 producer cells were assayed for infectious units (IU) (using the XC plaque assay) and RT activity. Infectivity was calculated by dividing IU by RT activity and was expressed as a percentage of wild-type virus infectivity. The infectivities of PBS-mutated MuLV variants fell over a broad range, with the wild-type virus showing the highest infectivity and the virus with a PBS complementary to tRNA^Tyr showing the lowest, with more than a 4,000-fold reduction in infectivity (Fig. 2C).

FIG. 2.

Replication and infectivity of MuLV mutants. (A and B) Replication kinetics of wild-type and mutant MuLV. Supernatants were taken from chronically infected NIH 3T3 producer cells, normalized for RT activity, and used to initiate infection of fresh NIH 3T3 cells. Infected cells were split 1:10 every 3 days. Supernatants were harvested in conjunction with each passage, frozen at −20°C, and assayed for RT activity following termination of the cultures. The data are from duplicate cultures for each virus and are shown as an average for each time point. (C) Infectivity of wild-type and mutant MuLV. Supernatants were taken from chronically infected NIH 3T3 cells and assayed in triplicate both for RT activity and for IU as determined by the XC plaque assay (34). Infectivity was calculated by dividing the number of IU by RT activity and was expressed as a percentage of wild-type infectivity. Error bars, standard errors of the means. The identity of the samples corresponds to the complementarity of tRNA to the PBS.

Evolution of the PBS during extended culture in NIH 3T3 cells.

To further assess the preference of MuLV for different tRNAs, we propagated the PBS-mutated viruses in NIH 3T3 cultures for extended periods and monitored the genomes over time for changes in the PBS. By day 66 of culture, viruses with a PBS complementary to tRNA^Arg had seven base pair changes within the PBS, corresponding to an alternate tRNA isoacceptor of arginine. The virus with a PBS complementary to was propagated in two separate cultures. In the first, the virus adapted to the use of two alternate isoacceptors of arginine before reverting to the wild-type PBS complementary to tRNA^Pro; in the second, the virus reverted directly to the wild-type PBS. The virus with a PBS complementary to tRNA^Met adapted to the use of two distinct isoacceptors of tRNA^Arg.

The virus with a PBS complementary to tRNA^Cys was stable, although the PBS sequenced did change to an alternative isoacceptor of tRNA^Cys (data not shown). In contrast, viruses with a PBS complementary to tRNA^Glu or tRNA^Asp reverted to the use of tRNA^Pro (data not shown). The viruses with a PBS complementary to , tRNA^Gly, , , tRNA^Tyr, or tRNA^Phe were stable in NIH 3T3 cultures. However, the latter four viruses replicated slowly for the duration of the cultures, as judged by syncytia in cocultures of infected cells with XC cells at each passage. In a previous study, we reported that MuLV with a PBS complementary to adapted to the use of tRNA^Arg before reverting to the wild-type PBS through recombination with an endogenous retrovirus sequence (33). In a second trial, the virus again adapted to the use of tRNA^Arg, while in the third trial reported here, the PBS of the virus did not change. The evolution of new PBSs for MuLV probably depends both on the spontaneous selection of certain tRNAs and on the subsequent outgrowth of these mutants due to an advantage in replication.

tRNA isoacceptor preference prior to Gag-Pol junctions in MuLV and HIV.

The results of our analyses highlight the fact that, in several instances, MuLV prefers the use of certain tRNAs, particularly tRNA^Pro, tRNA^Gly, and tRNA^Arg, for replication. Since previous studies from our laboratory had linked elements of viral translation with selection of the tRNA primer, we scanned the coding sequence of MuLV for regions that could be enriched with these three amino acids. We found such a region immediately upstream of the translational readthrough required to generate the Gag-Pol polyprotein in MuLV, where we noted nine consecutive amino acids that consisted of either proline, arginine, or glycine (Fig. 3A). We also noted that several tRNAs that were not effectively used by MuLV, including tRNA^Met and tRNA^Tyr, did not have codons near this region. To determine if this could represent a possible trend among retroviruses, we next examined the HIV-1 amino acid sequence prior to the Gag-Pol frameshift region that is used to generate the Gag-Pol polyprotein (Fig. 3B). Previous studies from our laboratory had shown that HIV-1 variants that have PBS and A-loop modifications complementary to tRNA^Met, tRNA^His, , or tRNA^Glu are stable but replicate more slowly than the wild-type virus (11, 20-22, 44). Surprisingly, we noted the appearance of several codons for these amino acids in a region upstream of the frameshift site for Gag-Pol polyprotein (Fig. 3B). Of additional interest is the fact that we have reported on several HIV-1 variants in which the PBS cannot be stabilized by additional A-loop modifications, including the PBSs that are complementary to tRNA^Ile or tRNA^Ser (21, 32). Coincidentally, we did not find codons for these amino acids near the Gag-Pol junction of HIV-1.

FIG. 3.

Amino acids prior to the Gag-Pol junctions of MuLV and HIV. (A) Amino acids of MuLV Gag prior to the stop codon/readthrough required for the synthesis of the Gag-Pol polyprotein of MuLV. The region encompassing enriched proline, arginine, and glycine amino acids is marked with arrows. (B) Amino acid composition of HIV-1 Gag prior to the translation frameshift required for the synthesis of the Gag-Pol polyprotein. The amino acids methionine, histidine, and glutamic acid are asterisked. Previous studies from our laboratory have demonstrated that modification of the A-loop and PBS to be complementary to tRNA^Met, tRNA^His, or tRNA^Glu results in a virus that will stably utilize these tRNAs for replication. The tRNA isoacceptors for lysine are shown. Previous studies have shown that HIV can utilize rather than as the primer for replication if additional mutations are made within U5 as well as the PBS to correspond with the 3′-terminal nucleotides of tRNA_1,2 (21, 31). In the F5 mutation in the HIV-1 genome, the codons for three lysine amino acids were altered to use rather than (indicated by “Lys 1,2” boxes). As a result of the mutation, the five lysine codons prior to the frameshift site are now complementary to .

Effects of altering codon use prior to the Gag-Pol junction on the replication and infectivity of HIV-1 forced to use tRNA_1,2^Lys.

To further explore the possibility that codons within the region prior to the Gag-Pol junction could influence primer selection, we focused on this region in the HIV-1 genome. In preliminary studies, we found that alteration of the codons within this region to encode different amino acids, even with conservative changes (e.g., methionine to isoleucine, threonine to serine), resulted in infectious viruses that had replication kinetics altered from that of the wild-type virus (N. Ni and C. D. Morrow, unpublished data). Thus, this highly conserved region appears not to tolerate amino acid changes in the Gag-Pol polyprotein (26, 27, 37). A previous study from our laboratory demonstrated that HIV-1 could be forced to use as the primer if both the PBS and the A-loop region were altered to be complementary to the 3′-terminal 18 nucleotides and anticodon of (21). However, even though this virus maintained a PBS complementary to , it replicated more slowly than the wild-type virus as a result of lower infectivity.

We identified five codons within this region that would encode lysine; three of the five codons correspond to . We changed these sequences from AAA to AAG to correspond to the anticodon for . Additional mutants were constructed in which either the PBS alone or both the PBS and the A-loop were made complementary to , as in our previous studies (21). We then analyzed the replication of these viruses in PBMCs by measurement of p24 antigen in the culture supernatant. Viruses with the PBS complementary to (wild type) and with the codon changes in the Gag-Pol frameshift region replicated at levels similar to that of the wild-type virus. Both viruses reached peak levels of p24 antigen in the culture supernatant at approximately 21 days after the initiation of culture (Fig. 4A). The virus with a PBS and an A-loop complementary to and with a wild-type Gag-Pol region replicated more slowly than the wild-type virus; the peak amounts of p24 antigen in the culture supernatants were approximately 10-fold less than that of the wild-type virus. These results are consistent with our previous studies in which we noted that this virus had delayed replication compared to the wild-type virus; continuation of this culture to approximately 60 days allows the virus to grow continually and reach levels of p24 antigen similar to that of the wild-type virus (31) (data not shown). In contrast, we found that the virus with the A-loop and PBS complementary to and with the codon changes in the Gag-Pol frameshift region replicated rapidly within PBMCs after only a very short delay. The final levels of p24 antigen in the culture supernatant were approximately equal to those of the viruses with the PBS complementary to .

FIG. 4.

Replication and infectivity of HIV-1 with both an A-loop and a PBS complementary to and with codon alteration prior to the Gag-Pol junction (F5). (A) Replication of wild-type and mutant HIV-1 with mutation in the lysine codons prior to the Gag-Pol junction. A previous study has shown that an HIV-1 mutant with a PBS and an A-loop altered to be complementary to (Lys12-AC [indicated by “Lys 1,2”]) will stably maintain the PBS complementary to following extended culture in PBMCs. The replication of this virus (Lys 1,2-WT) and that of the virus containing the alteration of the lysine codons to be complementary to (F5 [indicated by “Lys 1,2-F5”]) was compared with those of the wild-type virus and the wild-type virus containing the F5 mutation. Infections were initiated with equal amounts of virus (as determined by the JC53 assay), and samples were taken at the designated intervals. The p24 values were measured using a solid-phase ELISA. Symbols for samples are as follows: solid circles, wild type; open circles, wild type-F5; solid squares, Lys 1,2-F5; open squares, Lys 1,2-WT. Data are representative of three independent experiments for virus growth in PBMCs. (B) Infectivities of HIV-1 and mutants. The infectivities of the released viruses from the infection for which results are shown here were determined by the JC53-BL assay. The amounts of luciferase activity at different times postinfection were determined. The peak values obtained for wild-type and mutant viruses are circled. Data are representative of three independent experiments with viruses grown in PBMCs.

We next examined the infectivities of the viruses during replication in PBMC. For these studies, we analyzed the supernatants from these cultures for infectious virus by using the JC53-BL assay (10). The viruses with the PBS complementary to , regardless of whether they contained the additional mutations in the Gag-Pol region, all had peak infectivities 14 days after the initiation of culture. Surprisingly, the virus with the A-loop and PBS complementary to and with the additional changes in Gag-Pol also showed a rapid increase in infectivity, peaking at day 14, with levels of infectious virus only slightly less than those of the wild-type virus. In contrast, the virus with the A-loop and PBS complementary to and without the additional changes in the Gag-Pol region had greatly reduced infectivity compared to that of the virus with the alterations in the Gag-Pol frameshift region. The enhanced infectivity and growth of the virus with the A-loop and PBS complementary to could have been due to the rapid reversion to utilize , as had occurred with the viruses with only the PBS complementary to . However, sequencing of PCR products from this region obtained from integrated proviruses at day 28 revealed that the PBS and A-loop were still complementary to . Sequencing was done on isolated PCR products as well as TA clones, and all sequencing revealed that the PBS was still complementary to (data not shown). DNA sequence analysis of PCR products from the mutated Gag-Pol region revealed that all of the lysine codons remained specific for . Collectively, the results of these studies demonstrate that an alteration of the codon usage near the Gag-Pol frameshift region consisting of changing the codons of three amino acids from that for to that for enhanced the replication of viruses with the A-loop and PBS complementary to . The A-loop modification was required for the continued use of by this virus, thus establishing that at least three components are important for the effective selection of the tRNA by HIV-1: the A-loop, the PBS, and codon usage near the Gag-Pol frameshift region.

DISCUSSION

In this study, we have further investigated the process of primer selection by retroviruses by using a genetic approach in which the PBS was altered to be complementary to alternative (non-wild-type) tRNAs. Analysis of the replication and stability of MuLVs with alternative PBSs revealed a preference for a PBS complementary to tRNA^Pro, tRNA^Gly, or tRNA^Arg. The selection of tRNA^Arg for MuLV was probably facilitated, in part, by the multiple isoacceptors for tRNA^Arg. The preference of MuLV for tRNA^Gly is also supported by the characterization, in a previous study, of a MuLV retrovirus DNA that contained a complete copy of tRNA^Gly, indicating that this tRNA was selected during replication and subsequently copied during plus-strand synthesis in reverse transcription (8). What might account for this selection preference for certain tRNAs? Our previous studies suggested that primer selection occurred from a pool of tRNAs that were involved in viral translation (23, 24, 34). A search for proline, glycine, and arginine revealed a 9-amino-acid cluster near the Gag-Pol frameshift junction. This correlation between amino acids near the Gag-Pol junction and tRNA primer selection prompted a reexamination of the preferences of HIV-1 to use alternative, non-wild-type () tRNAs. HIV-1 will utilize , tRNA^His, tRNA^Met, or tRNA^Glu as a primer for replication if additional mutations are made upstream in a region designated the A-loop region, which is complementary to the anticodon of the primer tRNA (11, 20-22, 44, 47). Consistent with the findings for MuLV, the codons for lysine, methionine, histidine, and glutamic acid were present near the Gag-Pol frameshift region of HIV-1, although we did not find a striking cluster of amino acids like the proline, glycine, and arginine for MuLV.

To further pursue the connection between viral translation, codon use within Gag-Pol, and tRNA primer selection, we focused on HIV-1. Previous studies have shown that viruses in which the PBS was made complementary to were unstable in cell culture and rapidly reverted to use of (20, 21, 31). However, mutation of the upstream A-loop region to include nucleotides that were complementary to the anticodon of allowed these viruses to stably utilize for replication, although the viruses replicated more slowly than the wild-type virus. Thus, even though both and are present within HIV virions, the virus has a clear and distinct preference for the use of (25, 28). By scanning of the amino acid sequence upstream of the Gag-Pol frameshift region, three of five lysine codons were found to be specific for the isoacceptor, whereas two were specific for . The three codons were targeted for mutation to give codons for . The replication of HIV-1 in which either the PBS alone or both the PBS and the A-loop were altered to be complementary to with these additional mutations in the Gag region was dramatically enhanced relative to the replication of the similar virus with the A-loop and PBS complementary to but without the changes in the lysine codons. In the current study, we have altered only three codons prior to the Gag-Pol junction to use in order to achieve this change in replication and infectivity. Codons for and are dispersed throughout HIV-1 Gag, and additional studies will be needed to determine if the five codons for lysine prior to the Gag-Pol protein are the only ones influencing primer selection or whether the overall ratio of to influences primer selection. While the replication of HIV-1 with the A-loop and PBS complementary to was enhanced by the changes in the lysine codons near the Gag-Pol frameshift, we found that its replication was still slightly lower than that of the wild type, which uses . Unique features of might also facilitate selection and use in HIV-1 reverse transcription that would account for the differences in replication when HIV-1 is forced to use (5, 6, 17).

The results of our current study support a link between primer selection and the synthesis of the Gag-Pol polyprotein. Previous studies have shown that when the translational frameshifting occurs during the synthesis of HIV-1 Gag-Pol, ribosomes stall near the Gag-Pol junction (38, 42, 45). During the pause in translation, the ribosomes could become stalled over codons on the viral mRNA (45). Aminoacylated tRNA entering the stalled ribosome would be rejected and, following disassociation from the ribosome, would become a substrate for peptidyl-tRNA hydrolase to recycle the tRNA for inclusion in the protein synthesis cycle (9, 13, 15). We postulate, then, that the ribosomal pausing that occurs during frameshifting results in a local increase of tRNAs that could be captured as primers for retrovirus replication. For HIV-1, it is possible at this time that the lysyl-synthetase is occupied mainly with , which would also facilitate the capture of this tRNA (7, 18, 19). As in the case of our Gag-Pol mutants, if we alter the codon usage to favor , the synthetase would have an overabundance of to facilitate capture. However, the mechanism by which the virus might acquire the tRNA from the synthetase is unclear and will require further study. This interpretation, though, would not account for primer capture by MuLV, which does not incorporate prolyl-synthetase in virions. An alternative possibility is that as a result of the pausing, the tRNAs that are entering and leaving the ribosome would become deaminoacylated to allow the tRNAs to reenter the translational machinery following interaction with the synthetase (9, 13). Interestingly, previous studies with prokaryotes have shown that tRNA^Lys and tRNA^Arg are especially prone to “drop off” from stalled ribosomes and are substrates for peptidyl-tRNA hydrolase. As a result of pausing, there would be a local increase in the population of deacylated tRNA, possibly as a result of transient saturation of the synthetase. During this time, a direct interaction with the A-loop and/or the PBS might occur to capture the free tRNA for use as the primer. The mechanism for capture could involve direct RNA:tRNA interaction facilitated by complementarity with the PBS for MuLV and with both the A-loop and the PBS for HIV-1. Further studies will be needed to clarify the dynamics of primer capture as well as of the egress of the primer, viral genome, and proteins from the infected cell.

In summary, the results of our studies provide a new and important insight into retrovirus preferences in primer selection. From the analysis of the preference for tRNAs of MuLV and HIV-1, we have identified a region within Gag upstream of the Gag-Pol junction that is enriched with codons for tRNAs that can be selected by MuLV or HIV-1 for replication. The results of our studies are consistent with the idea that primer selection and viral translation, in particular the synthesis of Gag-Pol, are linked and that both MuLV and HIV-1 have evolved to preferentially select certain tRNAs using similar, but not identical, mechanisms that could involve ribosomal pausing before the synthesis of the Gag-Pol polyprotein. The linking of primer selection and synthesis of the Gag-Pol polyprotein would allow the virus to coordinate these two processes during viral replication.