Skip to main content
NCBI home page
Nucleic Acids Research logoLink to Nucleic Acids Research
. 2000 Feb 1;28(3):791–799. doi: 10.1093/nar/28.3.791

Selection of functional tRNA primers and primer binding site sequences from a retroviral combinatorial library: identification of new functional tRNA primers in murine leukemia virus replication

Anders H Lund 1, Mogens Duch 1, Finn Skou Pedersen 1,2,a
PMCID: PMC102544  PMID: 10637332

Abstract

Retroviral reverse transcription is initiated from a cellular tRNA molecule and all known exogenous isolates of murine leukemia virus utilise a tRNAPro molecule. While several studies suggest flexibility in murine leukemia virus primer utilisation, studies on human immunodeficiency virus and avian retroviruses have revealed evidence of molecular adaptation towards the specific tRNA isoacceptor used as replication primer. In this study, murine leukemia virus tRNA utilisation is investigated by in vivo screening of a retroviral vector combinatorial library with randomised primer binding sites. While most of the selected primer binding sites are complementary to the 3′-end of tRNAPro, we also retrieved PBS sequences matching four other tRNA molecules and demonstrate that Akv murine leukemia virus vectors may efficiently replicate using tRNAArg(CCU), tRNAPhe(GAA) and a hitherto unknown human tRNASer(CGA).

INTRODUCTION

Reverse transcription of a retroviral genome occurs in the cytoplasm of the newly infected cell. The process mediating conversion of the diploid RNA genome into proviral DNA is catalysed by the enzyme reverse transcriptase (RT) working in cooperation with other viral proteins brought along in the virion. Reverse transcription can be completed in vitro (1) and the only host-encoded component crucial to the process is the tRNA molecule acting as a primer for minus strand synthesis. The tRNA primer is derived from the producer cell and anneals to the viral genome at the primer binding site (PBS), an 18 nt sequence near the 5′-end of the viral genomic RNA. From analysis of virion tRNA content and from sequencing analysis of viral genomes it has been established that different viruses use different tRNA species as replication primers. Most mammalian C-type viruses use tRNAPro (2) as does human T cell leukemia virus (3), whereas the avian retroviruses replicate via a tRNATrp (4). Human immunodeficiency virus (HIV) (5) and mouse mammary tumor virus (6) use a tRNALys3 and the spuma (7) and visna (8) viruses utilise a tRNALys1,2 isoacceptor for initiation of minus strand DNA synthesis. Within a given virus the sequence of the PBS, and hence the use of a specific tRNA primer, is highly conserved, in part due to the conservative nature of the reverse transcription process in which the 3′ 18 nt of the tRNA is copied into DNA and serves to facilitate the second strand transfer reaction of reverse transcription (9). In some viruses specific adaptations towards the cognate tRNA primer have evolved, likely serving to increase fidelity at the initiation of reverse transcription and during the second template switch (for a recent review see 10). The existence of molecular interactions specifically recognising the primer tRNA over other tRNA species have been extensively analysed in both HIV and avian viruses through biochemical and genetic studies. In HIV, tRNALys3 molecules are up-concentrated in virions relative to the host cell cytoplasm due to interactions between tRNALys3 and the RT moiety of the Gag–Pol polyprotein (11,12). Likewise, mature HIV RT specifically binds tRNALys3, presumably through interactions with the anticodon loop (13–17). This binding furthermore induces a conformational change in the RT dimer molecule (18). While tRNA binding to the retroviral genome is primarily governed by base pairings between the PBS sequence and the 3′ 18 nt of the primer, studies on HIV have indicated the presence of a complex secondary structure in which the tRNALys3 molecule is engaged in base pairings outside the PBS sequence (19–21). For the avian viruses, tRNATrp recognition by the avian RT dimer has been shown biochemically (2,13,22,23) and base pairings between tRNATrp and the viral genome upstream from the PBS sequence have been demonstrated in biochemical and genetic studies using mutant viruses (24–27). Preferential tRNA usage in HIV and avian viruses have been confirmed through genetic studies using viruses with a modified PBS in cell culture (28–31). While initial replication with alternative tRNA primers may be demonstrated, wild-type revertant viruses with superior replication kinetics ultimatively dominate the cultures.

Interestingly, murine leukemia virus (MLV) may be less stringent in tRNA primer utilisation. Biochemical studies using MLV RT:tRNA co-sedimentation in glycerol gradients (32) or filter binding assays (13) have revealed no evidence of significant tRNAPro primer specificity. Furthermore, analysis of pol– mutants of Moloney MLV demonstrated that tRNAPro annealing to the PBS is independent of products of the pol gene (33). In addition, enzymatic probings of tRNAPro attached to Moloney MLV genomic RNA did not reveal evidence for additional base pairings between the tRNA primer and the genomic RNA outside the PBS region (34). Accordingly, we have previously shown that PBS-modified SL3-3 viruses may replicate efficiently in mice using tRNAGln1,2, tRNALys3 or tRNAArg1,2 (35). Likewise, retroviral vectors derived from Akv MLV may be efficiently transduced using either tRNAGln1,2 or tRNALys3 (36) or with a genetically engineered tRNA-like primer molecule (37). Additional indications for a less stringent tRNA primer usage in MLV come from sequencing of MLV-related viruses endogenous to the murine genome carrying PBSs matching tRNAGln (38,39).

In this study we investigate MLV tRNA utilisation through in vivo screening for functional tRNA primers and PBS sequences using a combinatorial retroviral vector library with randomised positions within the PBS. This approach allows for the simultaneous screening of all possible tRNA molecules present within a particular retroviral packaging cell for their functionality as primers in MLV replication. During reverse transcription of the retroviral genome the 3′ 18 nt of the utilised tRNA primer is copied into DNA and serves to mediate the second template shift (9). Hence, by analysing the sequence of the transduced PBS the tRNA primer originally used as a primer for reverse transcriptase may be identified. Using as a template in library generation a simple 3.4 kb construct carrying the neo gene, the combinatorial library may be synthesised using PCR-based methods only. In addition, the library is directly transfected into packaging cells, thereby omitting the diversity-limiting step of plasmid cloning. We demonstrate the feasibility of this approach by screening the BOSC 23 retroviral packaging cell line (40) for functional tRNA primers and identify, aside from the tRNAPro normally utilised by MLV, retroviral replication via four other tRNA molecules, including a hitherto unknown tRNASer(CGA) iso-acceptor.

MATERIALS AND METHODS

Library construction and verification

A combinatorial retroviral vector library was generated and amplified through a three-step PCR procedure using the replication-defective vector pPBS-xxx (41) as template DNA. In the first step, the 5′ 618 bp of the vector was amplified using primers 1 (5′-GGGAATTCTACCTTACGTTTCCCCGACCAGAGCTGATGTTCTCAG-3′), matching Akv MLV positions 7905–7924 (42), and 2 (5′-CCTGGGCGGGGGTC-TCCAANNCNNNNNNNNNNNNCCAAATGAAAGAC-3′), matching Akv MLV positions 135–181. Multiple replicate PCR reactions were performed using 100 ng pPBS-xxx template DNA/reaction. All PCR reactions in this study were performed in 100 µl PCR buffer (Stratagene) containing 25 pmol of each primer, 0.2 mM dNTP and 3 U of Taq polymerase (Stratagene). Reactions were run for 25 cycles (1.2 min at 94°C, 1.2 min at 60°C and 1 min at 73°C). This PCR introduced a specific linker in the 5′ U3 region and 14 random mutations in the PBS. The second PCR reaction using primers 3 (5′-TTGGAGACCCCCGCCCAGGGACCACC-3′), matching Akv MLV positions 163–188, and 4 (5′-GGCTCGAGAAATGACTGGCTGTCTCGAGGAGACCCTCCCAAGGAT-3′), matching Akv MLV positions 89–108, amplified the 3′ 2734 bp of the vector and introduced a specific linker in the 3′ U5 region. Reactions were run for 25 cycles with 100 ng pPBS-xxx template DNA (1.2 min at 96°C, 1.2 min at 60°C and 5 min at 73°C). The two fragments were connected and the total library amplified in a third PCR reaction using primers 5 (5′-GGGAATTCTACCTTACGTTT-3′), matching the 5′ part of PCR primer 1 and containing an EcoRI site, and 6 (5′-GGCTCGAGAAATGACTGGCT-3′), matching the 5′ part of PCR primer 4 and containing a XhoI site, specific for the introduced linker sequences. The final 3.4 kb vector library was purified on 0.7% low melting point agarose (NuSieve) followed by phenol extraction and ethanol precipitation. The library was analysed for the presence of sequence bias within the randomised positions by subcloning of 20 individual vectors in pUC19 using the introduced EcoRI and XhoI sites. The PBS region was analysed by automated sequence analysis (373A DNA Sequencer; Applied Biosystems Inc.) using primer 7 (5′-ATCGCTGGCCAGCTTACC-3′), matching Akv MLV positions 204–221. Direct sequence analysis of the vector library was done by manual dideoxy DNA sequencing according to the instructions of the manufacturer (US Biochemical) using primer 7.

Cell culture

NIH 3T3 cells were grown in DMEM supplemented with 10% newborn calf serum and 4 mM l-glutamine (Flow Laboratories). 293 cells were grown in DMEM containing 10% fetal calf serum and supplemented with either GlutaMAX (Gibco) or 4 mM l-glutamine, and the murine T-lymphocyte cell line L691 was grown in RPMI medium supplemented with 10% newborn calf serum. BOSC 23 cells were obtained from the American Type Culture Collection and passaged twice in DMEM containing either GlutaMAX or 4 mM l-glutamine, 10% dialysed fetal calf serum, 25 µg/ml mycophenolic acid (Gibco), 20 µg/ml aminopterin (Sigma), 250 µg/ml xanthine (Sigma) and 60 µg/ml thymidine (Sigma). All cell culture medium contained 100 U/ml of penicillin and 100 µg/ml of streptomycin and were incubated at 37°C in 90% relative humidity and 5.7% CO2. Prior to transfection and virus production, BOSC 23 cells were grown in DMEM supplemented with 10% fetal calf serum and either GlutaMAX or 4 mM l-glutamine. For screening of the retroviral vector library, ~8 µg of library DNA was transfected into two 80 cm2 bottles of BOSC 23 cells seeded at a density of 7 × 104 cells/cm2 the day prior to transfection. Transfections were performed using calcium phosphate as previously described (37). Virus-containing supernatants were harvested at 40 and 60 h post-transfection, filtered through a sterile 0.45 µm filter and transferred to NIH 3T3 cells (seeded at a density of 5 × 103 cells/cm2 the day before infection) in the presence of 6 µg/ml of Polybrene. Selective medium (0.6 mg/ml active G418) was added to NIH 3T3 cells 48 h after infection and individual resistant colonies were isolated and expanded after 10–12 days of selection. Genomic DNA from G418-resistant colonies was prepared as previously described (36).

Analyses of transduced vectors proviruses

To analyse the transduced primer binding site sequences in individual G418-resistant NIH 3T3 colonies, PCR amplification of 1.3 kb from integrated proviruses was performed using primers 8 (5′-TTCATAAGGCTTAGCCAGCTAACTGCAG-3′), matching Akv MLV positions 7838–7865, and 9 (5′-GGCGCCCCTGCGCTGACAGCCGGAACAC-3′), matching positions 1659–1686 of the neo gene (43). Reactions were run for 40 cycles (1.2 min at 94°C, 1.2 min at 63°C and 2.0 min at 73°C). Primer 8 was 5′ biotinylated enabling fragment purification and denaturation according to the Dynabead procedure (Dynal Inc.). Direct sequence analysis was performed using primer 7 or 10 (5′-TCCGAATCGTGGTCTCGCTGATCCTTGG-3′), matching Akv MLV positions 69–96, on an automated DNA sequencer (ABI 373A DNA sequencer; Applied Biosystems Inc.). Subclonings were performed in pGEM-T (Promega) and plasmid preparations from individual bacterial colonies were sequenced using primers 7–10.

Analysis of in vivo-selected primer binding sites

Retroviral vectors carrying specific PBS mutations matching those identified by in vivo library selection were constructed as described for library construction using the specific mutagenesis primers 11 (5′-CCTGGGCGGGGGTCTCCAAGTCCTGCTCRCAGCGCCAAATGAAAGAC-3′), introducing the PBS sequence identified in BRN2, 12 (5′-CCTGGGCGGGGGTCTCCAAGTCCCRTCTGGGGTGRCAAATGAAAGAC-3′), introducing the PBS sequence identified in BRN3, and 13 (5′-CCTGGGCGGGGGTCTCCAATCCCGGGTTTRGGCACCAAATGAAAGAC-3′), introducing the PBS sequence identified in BRN11. The vectors were cloned in pUC19 using EcoRI and XhoI. Due to the presence of variable positions within the mutagenesis primers (R = A or G) individual vector-carrying plasmids harbouring either the PBS identified by library selection or PBS sequences carrying specific marker mutations could be isolated. The structure of the resulting plasmid clones were verified by sequence analysis of the PBS and surrounding region and by restriction analysis. Analysis of vector transduction efficiencies was performed by virus titration assays on NIH 3T3 cells after transient transfection of BOSC 23 cells as previously described (37). For comparison of vector transduction efficiencies, transfections were performed in parallel with a syngeneic vector construct, pPBS-pro, carrying the wild-type PBS sequence matching tRNAPro. For analysis of marker mutation correction, genomic DNA from pooled G418-resistant colonies was subjected to PCR using primers 8 and 9 and the resulting 1.3 kb PCR fragment was cloned in pGEM-T (Promega). Plasmid DNA preparations from individual bacterial clones were sequenced using primers 7 and 10.

Northern blotting and tRNASer(CGA) identification

To analyse if the BRN2, BRN17 and BRN18 PBS sequences matched cellular tRNA molecules, northern blotting was performed using oligonucleotide probes corresponding to the selected PBS sequences, 14 matching BRN2 (5′-TGGCGCTGTGAGCAGGAT-3′), 15 matching BRN17 (TGGGGATAGAAATGCGTT-3′), 16 matching BRN18 (5′-TGGGCACT-GCGAATCGTT-3′) and as a positive control 17 (5′-TGG-GGGCTCGTCCGGGAT-3′), complementary to the 3′ 18 nt of tRNAPro. Total RNA (10 µg) from 293 cells and L691 cells was separated on denaturing 8% polyacrylamide gels and transferred to Zetaprobe membranes (Bio-Rad). Blotting was performed in 50 mM Tris–HCl (pH 8), 50 mM boric acid and 1 mM EDTA for 3 h at 40 V using a Bio-Rad transblot system. The oligonucleotide probes (100 pmol) were 5′-labelled using T4 polynucleotide kinase and [γ-32P]ATP (Amersham) and unincorporated label was removed using Microspin G-25 columns (Pharmacia). The filters were prehybridised for 1 h and hybridised overnight at 60°C in 7% SDS, 0.5% BSA, 1 mM EDTA and 0.5 M Na2HPO4 (pH 7.5). Prior to autoradiography, the filters were washed several times in 0.1× SSC (final concentration 15 mM NaCl, 1.5 mM sodium citrate) and 0.1% SDS at 60°C. As a control for RNA amount and integrity, the filters were subsequently probed using oligonucleotide 18 (5′-ATAAGAACAGATACTACACTTGA-3′), matching nucleotides 27–49 of the U2 small nuclear RNA (44). For identification of the tRNA utilised in BRN2 transduction 5 µg total RNA from 293 cells was subjected to reverse transcription using a First-strand cDNA Synthesis Kit (Pharmacia) and primer 19 (5′-biotin-AAAAATTGGCGCTGTGAGCAGGAT-3′). First strand cDNA reactions were purified using 6.7 × 106 Dynabeads and used as template in a PCR reaction with primers 19 and 20 (5′-GGGAGCTCTAGAGCTGTGAT-3′). PCR conditions were five cycles with 1 min at 94°C, 1 min at 40°C and 30 s at 73°C followed by 30 cycles with 1 min at 94°C, 1 min at 60°C and 30 s at 73°C. The resulting PCR fragment (~100 bp) was purified on Dynabeads and a partial cDNA sequence was obtained by direct automated sequencing of the PCR fragment using primers 19 and 20.

RESULTS

Library construction and verification

To investigate the tRNA primer specificity of Akv MLV a combinatorial retroviral vector library was constructed in which 14 of 18 positions in the primer binding site were randomised. Four nucleotide positions within the PBS were not altered; the 5′ trinucleotide TGG of the PBS normally base paring to the 3′ CCA tail of the tRNA primer molecule and a cytosine at PBS position 16 corresponding to a guanine conserved in murine tRNA molecules (45). The library was constructed and amplified using PCR-mediated overlap extension through a three-step PCR procedure in which the first PCR amplified the 5′ 618 bp vector fragment and introduced the PBS mutations and a linker in the 5′ U3 region (Fig. 1A). To obtain maximal sequence diversity at the randomised positions amplification products from multiple (>50) reactions were pooled. The second PCR step amplified the remaining 3′ 2734 bp of the vector and introduced a linker in the downstream U5 region, and the third PCR reaction connected the two generated amplicons and amplified the 3.4 kb vector library through primer-assisted overlap extension using primers specific for the introduced linker regions. The library was amplified from pPBS-xxx, a retroviral vector derived from Akv MLV in which the PBS had been mutated to have minimal sequence complementarity to the 3′-end of any known murine tRNA. Due to the lack of a complementary tRNA primer the replication capacity of pPBS-xxx is severely diminished, with a vector transduction efficiency 105-fold lower than that of an otherwise syngeneic construct harbouring the wild-type PBS sequence matching tRNAPro (41). A complete combinatorial 3.4 kb library containing 14 randomised positions should hold ~2.7 × 108 different retroviral vectors and could in theory be present in ~1 ng purified vector library. Due to the relatively small size of the vector the library could be synthesised and amplified using PCR only. Hence, the diversity-limiting steps associated with DNA ligation and bacterial transformation could be avoided. Prior to in vivo screening, the library was analysed for the presence of sequence bias within the randomised region. This was done by subjecting a small sample from the library to direct dideoxy sequencing (Fig. 1B), as well as by cloning and sequence analysis of 20 individual vectors (data not shown). We found no evidence of sequence bias in the randomised positions of the PBS.

Figure 1.

Figure 1

Open in a new tab

Construction and verification of a combinatorial retroviral library. (A) Library design and synthesis. The retroviral vector pPBS-xxx, carrying a non-functional PBS sequence, was used as a template for library construction. Random mutations were introduced into 14 of the 18 positions of the PBS using primers 1 and 2. Primers 3 and 4 amplified the downstream 2.7 kb of the vector and the two amplicons were connected by PCR-aided overlap extension using primers 5 and 6 specific for the linker regions inserted by primers 1 and 4 (hatched boxes). The presence of EcoRI and XhoI sites in the linkers facilitated cloning and analysis of individual library vectors. (B) Direct sequencing analysis of the PBS region of the combinatorial library. Note the presence of bands in all lanes at the randomised positions.

In vivo screening of the retroviral vector library

To screen the combinatorial library ~8 µg of library DNA was transfected into BOSC 23 packaging cells (40; Fig. 2). The BOSC 23 cell line has been selected for efficient production of recombinant retroviral vector particles after transient transfection of retroviral vector plasmid DNA and is therefore particularly useful for the transfer of retroviral vector libraries. Two replicate culture flasks were transfected with ~4 µg library DNA each. Library-containing virus particles were harvested at 40 and 60 h post-transfection and transferred to NIH 3T3 target cells. We isolated and expanded 20 G418-resistant colonies from a total population of ~30 colonies.

Figure 2.

Figure 2

Open in a new tab

Procedure for library screening. The library was transfected into BOSC23 retroviral packaging cells (40) and the resulting recombinant virus particles were used to infect NIH 3T3 fibroblast cells. Colonies obtained after selection with G418 were isolated, expanded and their proviral vector sequences amplified and analysed.

To analyse the structure of the transduced primer binding sites present in genomic DNA from G418-resistant colonies a vector-specific 1.3 kb fragment was amplified using an upstream U3-specific and a neo-specific primer. As described above, a linker sequence was inserted into the upstream U3 region as part of the library construction procedure resulting in deletion of the 5′ 67 bp. During reverse transcription the downstream U3 region is copied and translocated to the upstream part of the proviral genome, thereby reconstituting a full-length proviral vector in the target cell. Using a primer specific for the reconstituted 5′-part of the U3 region amplification of transduced proviral vectors only could be assured.

As a first approach, the sequences of the transduced PBSs were obtained by direct automated sequencing of the PCR fragments (data not shown). However, due to the presence of ambiguous positions in the sequence readouts, consistent with simultaneous sequencing of different proviral vector templates, the PCR products were cloned and individual subclones from each of the G418-resistant colonies were analysed by sequencing of an ~250 bp window encompassing the PBS sequence (Table 1). In 15 out of 20 analysed colonies only one particular PBS sequence was detected, 13 of which were complete matches to tRNAPro, one (BRN10) was complementary to the 3′ 18 nt of tRNAArg(ACG) and one colony (BRN18) contained a PBS sequence to which no match could be detected in the GenBank/EMBL nucleotide databases. In five colonies two PBS sequences were detected, one of which consistently was PBS-Pro. From the model of reverse transcription (9) it is unclear how two different PBS sequences may be found in the same G418-resistant NIH 3T3 colony unless the original target cell was infected with several particles carrying vector genomes with functional PBSs. However, given the low number of G418-resistant colonies appearing after selection (30 colonies of an initial population of >107 NIH 3T3 cells) simultaneous infections seem unlikely. One probable explanation for the findings may be that G418-resistant NIH 3T3 cells have migrated between the colonies, hence compromising the clonal integrity. Alternatively, DNA recombination among the library vectors in the packaging cells may have resulted in the formation of tandem vectors holding more than one PBS sequence. Evidence of vector transduction via tRNAArg(ACG) and tRNAPhe(GAA) were detected in BRN3 and BRN11, respectively, whereas BRN2 and BRN17 contain PBS sequences not matching any tRNA molecule in GenBank/EMBL. Aside from PBS-pro, BRN14 contains a PBS sequence with a 17/18 match to the PBS-xxx vector utilised for library generation.

Table 1. PBS sequences selected during in vivo library screening.

graphic file with name gkd163t01.jpg

Open in a new tab

aNumber of plasmid clones carrying a specific PBS sequence/number of clones analysed.

bDetected via specific PCR screening using a PBS-Phe specific primer.

cLibrary template PBS.

In the sequences flanking the transduced PBSs, colony-specific point mutations likely generated during library amplification were found, as were point mutations specific for individual subclones. The point mutations were seemingly scattered with no apparent correlation with the sequence of the PBS.

Northern blot analysis of tRNAs corresponding to selected PBS sequences

The presence of three hitherto unknown sequences among the transduced PBSs prompted us to investigate whether cellular tRNA molecules complementary to these sequences could be found. For this, northern blotting was performed using total RNA from human 293 cells and murine L691 cells. Oligonucleotides corresponding to the selected PBS sequences from BRN2, BRN17, BRN18 and PBS-Pro were used as probes (Fig. 3). The filters were subsequently re-hybridised with a probe against the U2 small nuclear RNA to check for amount and integrity of the loaded RNA. While neither the BRN17 nor the BRN18 probe resulted in bands within the expected size range of a tRNA molecule, hybridisations with the BRN2 probe gave a prominent band corresponding to a potential 85 nt tRNA molecule.

Figure 3.

Figure 3

Open in a new tab

Northern blot analysis. The existence of cellular tRNA molecules corresponding to the transduced PBS sequences was analysed by hybridisation of oligonucleotide probes to total RNA from human 293 and murine L691 cells. The probes were designed to match the 3′-end of the putative tRNAs including the CCA tail and used on independent filters. The filters were subsequently hybridised with an oligonucleotide probe against U2 small nuclear RNA to verify loading and integrity of the RNA. Filters hybridised with the BRN2, BRN17 and BRN18 probes were exposed for 19 h, while the filter hybridised with the PBS-pro probe was exposed for 4 h. M, DNA molecular size marker.

To identify the tRNA molecule complementary to the BRN2 PBS sequence a BRN2-specific oligonucleotide was used to prime cDNA synthesis of the putative tRNA. We hypothesised that the 5′ tRNA sequence is complementary to the 3′-end due to the cloverleaf structure and amplified the putative cDNA sequence of the tRNA using a primer specific for the known 3′-end of the tRNA against a primer matching the deduced 5′-end of the tRNA molecule. Sequence information obtained by direct sequence analysis of the amplified cDNA revealed a tRNA-like sequence including a trinucleotide CGA anticodon specific for the serine UCG codon (submitted to GenBank accession no. AF184043; Lund et al., unpublished results).

Functional analysis of in vivo-selected PBS sequences

To verify the results obtained from library screening a new set of retroviral vectors were constructed harbouring the in vivo-selected PBS sequences. Retroviral vectors containing the unknown PBS sequence found in BRN2, the tRNAArg(CCU)-complementary PBS sequence found in BRN3 and the tRNAPhe(GAA)-complementary sequence found in BRN11 were cloned using PCR-mediated site-directed mutagenesis. The PBS-Arg(CCU) and PBS-Phe(GAA) sequences were selected since they matched known tRNAs, and the unknown BRN2 PBS was included due to the presence of a complementary cellular RNA as evident from northern blot analysis. For each PBS-modified vector a mutant containing one or two point mutations was constructed. The introduction of point mutations within the PBS sequence results in the formation of a DNA mismatch after the second strand transfer reaction during reverse transcription in which a DNA copy of the genomic PBS is base paired to a DNA copy of the 3′ 18 nt of the utilised tRNA primer (9,46). After integration into the host cell genome the mismatch is repaired, presumably by cellular repair mechanisms, to a PBS sequence matching either the original PBS present on the vector RNA genome or to a perfect copy of the 3′ 18 nt of the tRNA primer molecule. Hence, correction of the introduced marker mutations may be taken as genetic evidence of specific tRNA primer usage (36,37).

The efficacy of the selected PBS sequences in mediating MLV vector transduction was analysed by virus titration assays after transient transfection of BOSC 23 packaging cells. All the PBS-modified vectors were able to mediate MLV vector transduction, thereby verifying the in vivo library selection procedure. Vector transfer efficiencies varied between 6 × 105 and 5 × 106 c.f.u./ml virus-containing supernatant for pPBS-BRN2 and pPBS-BRN11mb, respectively, as compared to 1 × 107 c.f.u./ml for pPBS-pro carrying the wild-type PBS matching tRNAPro (Table 2). Since titration assays per se are subject to experimental variation, we cannot from these data alone conclude if the differences in vector transfer efficiencies noted among the PBS-modified vectors are significant.

Table 2. Functional analysis of in vivo-selected PBS sequences.

graphic file with name gkd163t02.jpg

Open in a new tab

ND, not determined.

aIntroduced marker mutations relative to the selected PBS sequence are underlined.

bNucleotide positions altered during vector replication are shown in bold.

To analyse for marker mutation corrections, indicative of specific tRNA primer usage, a number of transduced PBS sequences were cloned and analysed. PCR amplification of a vector fragment containing the PBS sequence was performed using genomic DNA from pools of G418-resistant NIH 3T3 colonies (>50 colonies). The PCR fragments were cloned and the structures of the transduced primer binding sites were investigated by sequence analysis of plasmid DNA from a number of individual bacterial colonies. The pPBS-BRN2m vector contains a T→C substitution at PBS position 9 resulting in a G-T mismatch after the second strand transfer of reverse transcription. In two out of four analysed colonies the marker mutation had been corrected to a thymidine. In addition, the most 3′ nucleotide of the BRN2 PBS had been altered from cytosine to thymidine. This was also found in five analysed colonies resulting from transduction with the pPBS-BRN2 vector holding a perfect copy of the selected BRN2 PBS. Hence, the data demonstrate that MLV may efficiently replicate using a hitherto uncharacterised tRNASer(CGA). The pPBS-BRN3mm vector contains two marker mutations, both substitutions for cytosine at PBS position 3 and 13, relative to the selected BRN3 sequence matching tRNAArg(CCU). In one of six analysed colonies both marker mutations had been corrected to form a perfect match to tRNAArg(CCU), thereby verifying the functionality of tRNAArg(CCU) in MLV replication. In pPBS-BRN11ma and pPBS-BRN11mb a guanine residue at PBS position 8 was substituted for either a thymidine or a cytosine. Marker mutation correction was found in two of five analysed colonies resulting from transduction with pPBS-BRN11ma and in one of five colonies resulting from transduction with pPBS-BRN11mb. Hence, tRNAPhe(GAA) also supports efficient single cycle transduction of Akv MLV vectors.

DISCUSSION

The host cell-derived tRNA molecule plays a pivotal role in retroviral replication. This is evident from studies of viral mutants of HIV (47,48) and MLV (37,41) in which the PBS sequence has been deleted or mutated not to match a functional tRNA. Such crippled viruses have severely thwarted replication capacity with transduction efficiencies diminished 5 log compared to wild-type as measured by single cycle transductions using retroviral vector constructs (37,41). The importance of the tRNA–virus interaction renders the PBS region a good candidate motif for combinatorial library studies, since strong selection for functional PBS sequences enable identification after a single round of selection (transduction) only. Retroviral vector libraries have previously been used as expression cloning tools (49–51) and to construct libraries of genetic suppressor elements (52,53). However, while retroviral cDNA libraries normally necessitate plasmid cloning, combinatorial libraries aimed at investigating viral cis elements are relatively small in size and can thus be generated by PCR amplification and overlap extension, whereby a higher degree of library diversity can be maintained. In this study we randomised 14 of 18 PBS positions enabling potential binding of library vectors to all tRNAs present in the packaging cell. Given the amount of library generated and the length of the individual vectors, a theoretical 8000 copies of each individual PBS combination were synthesised. After in vivo library screening in BOSC 23 retroviral packaging cells 20 individual transduction events were analysed by subcloning of amplified proviral vector fragments encompassing the transduced PBS sequences.

Most of the analysed colonies contained PBS sequences matching tRNAPro. While this could be indicative of a preference in MLV for using tRNAPro, the presence of two PBS sequences in five of the analysed colonies could suggest that clonal integrity has been compromised by migration of NIH 3T3 cells between colonies. Also, transduction of tandem vectors, resulting from DNA recombination among library vectors in the BOSC cells, might explain the detection of two different PBS sequences in the analysed target cells. Using the murine Ψ-2 packaging cell line we have found that rare transduction events of the PBS-xxx vector result in proviruses of mostly the PBS-xxx and the PBS-pro type (unpublished results). Aside from PBSs matching tRNAPro, seven different PBS sequences were selected through in vivo library screening. We have previously shown that tRNAArg(ACG) sustains replication of SL3-3 MLV in mice (35) and a PBS sequence matching tRNAArg(ACG) is present in the snakehead fish retrovirus (54) and in a human endogenous retrovirus (55). The 3′-end of the tRNAPhe(GAA) molecule matches the PBS found in genomes of mouse and hamster intracisternal A particles (56,57) and the 297 retroelement of Drosophila contains a PBS sequence complementary to the fly tRNASer(AGA) isoacceptor (58). However, MLV replication via tRNAArg(CCU), tRNAPhe(GAA) or tRNASer(CGA) has not previously been reported. In this study tRNAArg(CCU), tRNAPhe(GAA) and tRNASer(CGA) were found to efficiently sustain single cycle transduction of Akv MLV vectors carrying complementary PBS sequences, thus validating the library screening procedure. From initial searches in the GenBank/EMBL databases we were unable to assign potential tRNA primers to three of the transduced PBS sequences found in BRN2, BRN17 and BRN18. Evidence of a complementary tRNA to the BRN2 PBS, later identified as tRNASer(CGA), came from northern blotting analysis (Fig. 3). While complementary tRNA molecules to the BRN17 and BRN18 PBS sequences may exist, transduction of these sequences may also have been facilitated by aberrant modes of reverse transcription (59–61). The identity of the tRNASer(CGA) molecule involved in the transduction of BRN2 was established from sequencing analysis of an amplified cDNA product. Isolation and sequencing of the corresponding human gene is the subject of a separate study (Lund et al., manuscript in preparation).

In BRN14 a 17/18 nt match to the PBS-xxx sequence of the library template was found, indicating the presence of template DNA in the library. While the usage of a replication-deficient template vector should decrease the risk of a library bias due to template contamination of the library, the formal possibility exists that the PBS-xxx sequence may selectively favour binding of a specific subset of tRNA isoacceptors.

Replication primer selectivity may be acting in several stages of virion assembly, maturation and subsequent reverse transcription. However, MLV seems to differ from avian viruses and HIV-1 in these processes. Firstly, PBS-modified SL3-3 MLV proviruses were recovered from tumors induced by injection of these mutants into newborn mice (35), whereas culturing of PBS-modified HIV-1 (28–30) or avian leukosis virus (31) results in the outgrowth of revertant viruses harbouring wild-type PBS sequences. Interestingly, a recent report from Yu and Morrow (62) studying single cycle replication of HIV-1 vectors modified to replicate via tRNAIle and tRNAHis suggested a prominent role of the PBS in tRNA replication primer selection. Secondly, products of the pol gene have been demonstrated to mediate selective tRNA incorporation into virions in HIV-1 (11,63,64). The specific mechanism responsible for selective tRNA packaging is still unclear but may involve cooperativity between the viral Gag and Gag–Pol precursor proteins (64). Although MLV isolates also package a specific subset of the cellular tRNAs (65), the underlying mechanism is likely different from that in HIV-1 since pol– mutants of Moloney MLV are largely unaffected in terms of selective tRNAPro packaging and annealing (33,66). Thirdly, while numerous biochemical studies have demonstrated specific binding of HIV-1 RT to tRNALys3 (13–15,18,67) and likewise between tRNATrp and RT from avian retroviruses (13,23,68), no such specificity has been found to exist in MLV (13,32).

In conclusion, in this study we identify, aside from transductions mediated by the tRNAPro molecule utilised by wild-type MLV, replication of Akv MLV vectors via four other tRNA molecules and provide proof of the principle of the library approach for analysing MLV tRNA utilisation in specific and viral cis elements in general. While this screening is by no means exhaustive, data from this study sustain the notion of MLV having a low stringency in tRNA primer selection and underlines the importance of the PBS sequence as the major determinant for primer selection in MLV (35,36).

Acknowledgments

ACKNOWLEDGEMENTS

The technical assistance of Ane Kjeldsen is gratefully acknowledged. We thank Florence Le Roy for help with analysis of library subclones. This work was supported by contracts CT 95-0100 (Biotechnology) and CT 95-0675 (Biomed 2) of the European Commission, the Karen Elise Jensen Foundation, the Danish Cancer Society, the Danish Biotechnology Programme and the Danish Natural Sciences and Medical Research Councils.

REFERENCES

Articles from Nucleic Acids Research are provided here courtesy of Oxford University Press

RESOURCES