An RNA secondary structure bias for non-homologous reverse transcripta
http://www.100md.com
《核酸研究医学期刊》
1 Department of Molecular Biology and 2 Department of Medical Microbiology and Immunology, University of Aarhus, C.F. Mollers Allé, Building 130, DK-8000 Aarhus, Denmark
*To whom correspondence should be addressed at Department of Molecular Biology, University of Aarhus, DK-8000 Aarhus, Denmark. Tel: +45 8 9421111, direct, +45 8 9422614; Fax: +45 8 6196500; Email: fsp@mb.au.dk
Present addresses:
Thomas Jespersen, Department of Medical Physiology, The Panum Institue, University of Copenhagen, Blegdamsvej 3, DK-2200 Copenhagen N., Denmark
Lars Aagaard, Bioinformatics Research Center, Department of Computer Science, Aarhus University, DK-8000 Aarhus, Denmark
ABSTRACT
Murine leukemia viruses harboring an internal ribosome entry site (IRES)-directed translational cassette are able to replicate, but undergo loss of heterologous sequences upon continued passage. While complete loss of heterologous sequences is favored when these are flanked by a direct repeat, deletion mutants with junction sites within the heterologous cassette may also be retrieved, in particular from vectors without flanking repeats. Such deletion mutants were here used to investigate determinants of reverse transcriptase-mediated non-homologous recombination. Based upon previous structural analysis the individual recombination sites within the IRES could be assigned to either base-paired or unpaired regions of RNA. This assignment showed a significant bias (P = 0.000082) towards recombination within unpaired regions of the IRES. We propose that the events observed in this in vivo system result from template switching during first-strand cDNA synthesis and that the choice of acceptor sites for non-homologous recombination are guided by non-paired regions. Our results may have implications for recombination events taking place within structured regions of retroviral RNA genomes, especially in the absence of longer stretches of sequence similarity.
INTRODUCTION
Template switching during reverse transcription contributes to genetic recombination in retroviruses (1,2). Although recombination may also take place during plus-strand synthesis, recombination by template switching during minus-strand synthesis is probably more frequent and has been intensively studied (3). By intermolecular template-switching reverse transcriptase (RT) switches between the two complete RNA genomes that are co-packaged in a retroviral particle (4). Template switching can also take place intramolecularly between two positions on the same RNA (5). Analysis of the requirements for RT template switching in vitro and in vivo has revealed that sequence similarity at the donor and acceptor site greatly facilitates recombination (6,7). However, non-homologous recombination events between sites without any sequence identity occur at low frequencies (7).
The original forced copy-choice model proposed that recombination occurs when RT encounters a break in the RNA template during minus-strand synthesis (8). This model has been extended to include events not dependent upon template RNA breakage. In a further development it was recently shown that the RT-associated RNase H activity plays a critical role for minus-strand recombination in vivo. This dynamic copy choice model (9) states that the balance between the rate of DNA polymerization and the rate of RNA template degradation behind the RT determines the amount of nascent DNA that is free to pair with an acceptor template.
Features of the donor site such as secondary structures have been found to influence the rate of template switching (10–13). Such structures have also recently been reported to play a role at the acceptor site (7,14,15). In the present work, we used a heterologous sequence including the encephalomyocarditis virus (EMCV) internal ribosome entry site (IRES) element as a target sequence for deletions in an in vivo assay. This IRES element was first described as mediating internal ribosome entry on polycistronic RNA of the EMCV (16). The structure has been extensively characterized (17,18) and correct folding of the RNA in a heterologous context can furthermore be judged by the ability of the element to direct translation of a reporter protein such as the enhanced green fluorescence protein (EGFP).
By using a heterologous cassette inserted into a complete viral genome the selective pressure upon the resultant vector is reduced to the gain in replication fitness acquired by deletions in the insert. Moreover, since there is no positive selection for regeneration of specific sequences all deletions can be assumed to result in replication competence, at least as long as they are confined to the heterologous cassette.
Some deletions will result in complete loss, others in only partial loss of the cassette. In this report we look only at the latter type with one or both deletion site junctions within an IRES-EGFP cassette containing internal direct repeats of a maximum of 9 nt. We note that the deletion site junctions lack sequence similarity, and, looking only at junctions within the IRES, we find a significant bias towards sites predicted to be in non-paired regions of the IRES RNA.
MATERIALS AND METHODS
Plasmids
The plasmid-cloned replication-competent vectors are AkvU3-EGFP, AkvSL33U3-EGFP and AkvEn-EGFP. AkvU3-EGFP has been described previously (19). It harbors an IRES-EGFP insert in the Cel II site of the 3' long terminal repeat (LTR) of the Akv virus. The linker (20) between the 3' end of the EMCV IRES element and the start of the EGFP gene in AkvU3-EGFP ensures 10-fold higher translation of the EGFP gene as compared to other linkers (21,22).
The AkvSL33U3-EGFP will be described elsewhere (M. L. Carrasco, M. Duch and F.S. Pedersen, submitted for publication). In short, it was constructed by replacing the 3' LTR of AkvU3-EGFP downstream of the IRES-EGFP insert with a polymerase chain reaction (PCR) amplified fragment of the corresponding sequence from SL3-3 (23). The primers located in the 5' end of the LTR were designed to delete the 3' Cel II site flanking the insert in AkvU3-EGFP. The resulting AkvSL33U3-EGFP construct has the following sequence with no direct repeat flanking the insert: 5'-GCTTAGC-TGCAGATGCATGGCCCATGCGGCCGC-IRES-EGFP-CAG CTAACT-3'. Underlined nucleotides are derived from the LTR whereas letters in italics denote the heterologous polylinker insert.
The AkvEn-EGFP construct (24) was derived from pAkv (21) where the IRES-EGFP cassette has been inserted 3 nt after the termination codon of the envelope gene followed by a duplicated sequence of the C-terminal region of the envelope gene upstream of the polypurine tract (PPT) resulting in a repeated structure flanking the insert. In the coding sequence for the C-terminal part of the envelope gene a CGT codon was exchanged to another arginine codon CGG in the C-terminal part of the envelope gene. The T and the G residues are shown in bold in Figure 1A.
Figure 1. Experimental design. (A) Schematic representation of the three replication-competent retroviral vectors used. Letters in bold are not part of a repeat. Figures are not drawn to scale. PSI, packaging signal. (B) Outline of experimental procedure for isolation of deletion mutants.
Cell culture, transfection and transductions
NIH3T3 cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) containing Glutamax-1 (Gibco BRL) supplemented with 10% newborn calf serum (NCS) (Life Technologies, Inc.), 1 unit of penicillin and 1 μg of streptomycin per 1 ml of final medium. BOSC 23 cells were grown as described for NIH3T3 cells except that NCS was exchanged with fetal calf serum (FCS). Prior to usage, the BOSC 23 cells were selected in HAT medium as described (21). For generation of replication-competent viruses, BOSC 23 cells were seeded at a density of 7 x 104 cells/cm2 1 day prior to transfection. The cells were transfected with 10 μg of plasmid DNA by the calcium phosphate method without a glycerol shock as described (25). One day post-transfection the medium was renewed on the BOSC 23 cells and fresh NIH3T3 cells were seeded in 75 cm2 flasks at a density of 5 x 103 cells/cm2. Two days post-transfection undiluted virus-containing medium was transferred from the BOSC 23 cells to the NIH3T3 cells, supplemented with 6 μg/ml Polybrene (Aldrich Chemical Co. Inc.) and cultured for 1 day. Twenty-four hours post-infection the medium was exchanged with fresh medium containing 2 μg/ml Polybrene and the cultures were passaged for 1 week. For consecutive rounds of infection (two to six), the same scheme was followed (Fig. 1B).
DNA isolation, PCR amplification and subcloning
DNA was isolated using the DNAzol genomic DNA isolation reagent according to the manufacturer’s protocol (Molecular Research Center, Inc.). DNA fragments containing deletion mutants were amplified using standard PCR techniques. In brief, PCR amplification was performed in 100 μl reactions containing 1 μg of purified DNA, 25 pmol of each primer, 0.2 mM dNTPs and 2.5 units of AmpliTaq Gold DNA polymerase (Perkin-Elmer Cetus). After an initial activation step of 10 min at 94°C, the reaction was followed by 40 cycles of 1.2 min at 94°C, 1.2 min at 60°C and 3 min at 72°C. The reaction was terminated with a final incubation of 7 min at 72°C. The oligonucleotides used as primers are indicated in Figure 2A and are as follows: 1, 5'-CAACAAGGGTGGT TTGAAGGGCTGTTTAA-3'; 2, 5'-GAATTCGATATCGA TCCCCGGTCATCTGGG-3'; 3, 5'-GATCGCTTAGCTGC AGATGCATGGCCCATGCGGCCGCCCCCTAAC-3'; and 4, 5'-AGACCTGGATCCCGCTTTACTTGTACAGCTCG TCCATGC-3'. The PCR products were electrophoresed on agarose gels and bands migrating between the products amplified from a non-reverted replication-competent vector and a complete revertant, were isolated using a GFX purification kit (Amersham Pharmacia Biotech Inc.) according to instructions from the manufacturer. The resulting products were cloned in pGEM-Teasy vector (Promega) for subsequent sequencing. Bacterial clones were screened by PCR using the same primers as used in the initial PCR amplification of the deletion mutants from the genomic DNA and positive clones picked for mini-preparation and sequence analysis of plasmid DNA. Primers used for sequencing were located in the pGEM-Teasy vector outside the cloning site, either 5'-GTTTTCCCAGTCACG-3' or 5'-CAGGAAACAGCTATG-3'. The sequencing reaction was performed using the Thermo Sequenase II dye terminator cycle sequencing kit (Amersham Pharmacia Biotech Inc.) and Applied Biosystems 373A DNA sequencer.
Figure 2. Retrieval of deletion site junctions. (A) Overview of positions of oligonucleotide primers for PCR amplification of the IRES-EGFP deletion mutants. (B) Example of an agarose gel analysis of the resultant PCR product from AkvEn-EGFP infected NIH3T3 cells using primers 3 and 2. Boxed bands/smear represent possible deletion mutants. DNA from boxed area was isolated, subcloned and sequenced. From left to right: lane 1, DNA marker; lane 2, 4th passage; lane 3, 5th passage; lane 4, 6th passage; lanes 5 and 6, DNA markers.
RESULTS
Replication-competent retroviruses with IRES sequence
We used three replication-competent murine leukemia viruses (MLV) where an IRES-EGFP cassette was inserted into either Akv or a chimera of Akv and SL3-3 viruses (19,24 and M. L. Carrasco, M. Duch and F. S. Pedersen, submitted for publication). Two insertion sites were used, one was located between the envelope gene and the PPT resulting in only one copy of the IRES-EGFP element at both RNA and DNA level. This construct (AkvEn-EGFP) was based upon the Akv virus where the IRES-EGFP element was flanked by a 38 out of 39 nt direct repeat due to duplication of sequences upstream of the PPT (see Fig. 1A).
The second site used for insertion was situated 20 nt 3' of the start of the U3 region in the downstream LTR. Being located at this position, the IRES-EGFP cassette will be copied to both the upstream and the downstream LTR during reverse transcription of the genomic RNA to double-stranded DNA. Only the IRES-EGFP cassette placed in the downstream LTR will be transcribed into viral RNA and thereby participate in further cycles of reverse transcription. In these two constructs where the IRES-EGFP element was inserted into the LTR, the Cel II site located in the upstream part of the U3 region was used. The first construct, AkvSL33U3-EGFP, was based upon a chimeric virus where the LTR of the Akv virus was exchanged with the LTR of the SL3-3 virus. This construct contained no direct repeat flanking the IRES-EGFP element. In the second construct, AkvU3-EGFP, the insert was flanked by a duplicated Cel II site giving rise to a perfect direct repeat of 8 out of 8 nt residing within an imperfect direct repeat of 13 out of 14 nt flanking the IRES-EGFP element (Fig. 1A).
Experimental procedure
In separate work we have analyzed the genomic stability of replication-competent vectors during repeated passages in cell culture (21,24). In those studies, BOSC 23 packaging cells were transiently transfected with a replication-competent vector construct and undiluted media transferred to NIH3T3 cells 48 h post-transfection. These infected NIH3T3 cells were passaged for 1 week, whereafter undiluted supernatant was transferred to fresh NIH3T3 cells for a second round of infection events (Fig. 1B). Following this experimental procedure, revertants started to appear between rounds two and five as analyzed by northern blotting (21,24). Each round of transfer of fresh NIH3T3 cells probably represents several rounds of viral infection events as only 40% of cells were infected 48 h post-transduction, whereas after 1 week more than 99% of the cells in the population were infected, as analyzed for EGFP expression by fluorescence microscopy.
In the present study, we were interested only in those recombination events which had resulted in deletions with one or both junctions located within the heterologous cassette. According to this criterion, the analysis does not include mutants where the complete transgene cassette is lost.
For the AkvU3-EGFP and AkvEn-EGFP vectors the vast majority of such complete cassette deletions are driven by the flanking direct repeats (24). For optimal detection of partial cassette deletions, DNA from infection rounds four to six was analyzed by PCR amplification using the oligonucleotide primers (Fig. 2A) in various combinations. One example of such a PCR amplification of DNA from round four to six for AkvU3-EGFP is shown in Figure 2A. Using primers located outside the insert (primers 1 and 2), complete reversion to wild type could be detected for all three constructs, as verified by PCR amplification and sequencing (24).
Mapping sites of recombination
This population of bands, as exemplified by the boxed areas in Figure 2B, was isolated, subcloned and sequenced. In several cases the same deletion mutant was isolated more than once from different rounds of the same culture, sometimes from all three rounds, most likely resulting from a single primary deletion event and counted as such in this analysis. Notably, none of the junctions arising from deletions within the IRES-EGFP cassette were independently isolated from the three different viral constructs used. Figure 3 is a graphical representation of all the deletion sites. Some of the deletion sites identified in this set-up reach into the retroviral sequences of the vector as shown in Figure 3, which also depicts critical cis elements for viral replication next to the insert. These are the PPT and integration site signals at the upstream U3 border located 3' to the heterologous cassette in AkvEn-EGFP and 5' to the cassette in AkvSL33U3-EGFP and AkvU3-EGFP. We note that deletion junctions found for AkvEn-EGFP extend into viral sequences at the 5' site but do not extend beyond the cis restriction at the 3' site. For the AkvSL33U3-EGFP construct, deletion sites do not map beyond the cis restriction at the 5' end of the LTR, whereas the 3' deletion junctions extend 200 nt into the U3 sequences where no critical cis elements are located. The few non-complete junction site revertants that were retrieved from the vast majority of complete revertants from the AkvU3-EGFP construct, conform to the same patterns. Altogether, the distribution of deletion borders, as depicted in Figure 3, suggests that the maintenance of viral cis elements is crucial, making it unlikely that the observed pattern is caused by artefacts generated during the post-replication steps (PCR, subcloning and sequencing).
Figure 3. Bar chart of deletion junctions. Indicated below the bar chart are the functional domains of the replication-competent vectors. Small vertical bars represent stretches of homology. Arrows point to borders of cis element restrictions. Numbers at the axes refer to the nucleotide positions in the corresponding plasmids of the replication-competent vectors.
Analysis of sequence similarity at the deletion site
In order to determine if any correlation between RNA primary or secondary structure and deletion junction site distribution existed, only clones where one of the deletion site junctions was located within the IRES element were used for further analysis. Solely 5' recombination sites were recovered from the IRES element.
Depicted in Figure 4 are 10 nt of the upstream and downstream deletion site junctions. The sequences were analyzed for direct repeats at the junction sites potentially used for homologous recombination as well as additional inserts of nucleotides at the deletion site. For each recombinant viral vector, the size of the deleted sequence is given in nucleotides, ranging from 612 to 1374 nt. The lack of smaller deletions may relate to the selective forces operating during selection for replication fitness as well as to the gel excision procedure. Again, all revertants resulting in complete loss of the heterologous cassette were excluded from the analysis. Inspection of Figure 4 reveals that the deletion site junctions harbor no or only short repeats. This indicates that sequence similarity is not the major determinant of deletion site selection in this set-up.
Figure 4. Deletion site junction sequences. The numbers at the extreme left specify the reference numbers of the isolated clones. Numbers internal in the sequence indicate the size of the deletion. Left, semi-boxed sequences represent nucleotides at the 5' site maintained after recombination. Right, semi-boxed sequences represent 3' nucleotides maintained after recombination. Internal boxed sequence represents deleted sequences. Sequences in bold (and italics) represent nucleotide insertions at the deletion site. Sequences in bold and underlined represent direct repeats at the deletion junctions.
Correlation between recombination site distribution and RNA secondary structure
The IRES element from EMCV has been extensively characterized (16) and the secondary structure given in Figure 5 was unravelled by Pilipenko et al. (17) and Jang and Wimmer (18). The deletion junction sites were depicted on this RNA structure to assess a possible correlation between RNA structure and recombination site distribution. Also shown are the cases of sequence identity at the recombination site identified in Figure 4. For the accumulated data set of all constructs we note a clustering of deletion site junctions in the same unpaired region in the beginning of the IRES element, position 50–57 in Figure 5.
Figure 5. Schematic representation of the upstream deletion site junctions in the RNA secondary structure of the EMCV IRES element. The model is based upon the published secondary structure (17,18). The brown lines represent data from Logg et al. (28) while the remaining three colors represent the various constructs used in the present work. The length of the lines indicate that the recombination has occurred somewhere between these nucleotides. The length varies due to the different size of direct repeat at the recombination site. The two deletion events involving extra nucleotides (Fig. 4) have been depicted as if there were no extra nucleotides. Letters in roman and nucleotides in pink represent stretches of sequence homology 8 internal in the IRES-EGFP insert described in Figure 6.
The IRES structure was divided into paired or unpaired nucleotides giving a total of 334 paired as compared to 224 unpaired nucleotides, i.e. 40% unpaired nucleotides (Fig. 5 and Tables 1 and 2). From the experimental data set only recombination events occurring exclusively in either an unpaired or paired region were used to determine the distribution of recombination sites. From a total of 39 recombination events (Fig. 4), 34 could conclusively be ascribed to either an unpaired or a paired region of the IRES element (Fig. 5). Of these 34, 25 had occurred in an unpaired region and nine had occurred in a paired region. The accumulated likelihood of observing 25 or more events of recombination involving unpaired regions as acceptor out of a total of 34 events is 0.0082%, as evaluated in a binomial distribution with a probability parameter of 40% for unpaired events (Tables 1 and 2).
Table 1. Distribution of recombination sites between unpaired and paired regions of the IRES sequence
Table 2. Likelihood of recombination events occurring in an unpaired region
DISCUSSION
We have used deletions within a heterologous sequence cassette in a replication-competent vector to study features of a template that favors retroviral recombination. In this set-up, it may be assumed that the increased replication fitness associated with deletion of non-viral sequences is the sole force driving the increased prevalence of deletion mutants upon multiple passages. The presence of a repeat flanking the insert will facilitate complete deletion of the heterologous cassette. We may also assume that all deletions within the heterologous insert will be compatible with continued replication; however, it is not unlikely that individual parts of the insert may affect replication fitness differently. In compliance with the expected continued replication of deletion mutants we have many examples of repeated isolation of the same mutant at different passages, some deletion mutants were even isolated from all three passages analyzed. It cannot be excluded that some of the deletion mutants retrieved result from consecutive deletion events rather than a single event. However, we have observed no case of two or more separate deletions within the same isolate.
In two of the vectors used, a direct repeat sequence flanks the heterologous insert (see Fig. 1A). Such repeats favor the perfect excision of the insert by homologous recombination and result in the regeneration of replication-competent viruses without insert. In spite of this favored process of complete excision in these vectors, a small number of deletions using other sites with little sequence identity at the deletion borders may be retrieved (Fig. 4).
We note that the repeated isolation of mutants with exactly the same deletion junctions from different passages argues that the number of incomplete cassette deletion mutants within a given pool is limited. On the other hand, the fact that the junctions retrieved from separate experiments are different argues in favor of a large number of potential deletion junctions.
Looking at 10 nt on either site of the deletions in the analysis (Fig. 4), we do not observe a clear pattern of direct and inverted repeats at the junctions. This indicates that a majority of the deletions result from non-homologous recombination events. Most of the events resulted in a perfect junction, but in two out of the 40 cases analyzed 1 or 2 nt had been inserted at the junction, presumably by non-templated nucleotide addition (26,27). We looked for candidate target sequences for homologous recombination to further assess the lack of sequence identity at the retrieved junction sites. Scanning in windows of 8 nt (Fig. 6) revealed stretches of sequence identity of up to only 9 nt, i.e. below the sequence identity windows of 14 nt or more found to be critical for homologous recombination in a previous study (7). The numbers in roman letters at the identity stretches correspond to the sequences given in Figure 5. We note that none of these putative homologous recombination events are identified in the present screens, even though they would result in deletions within the observed size ranges.
Figure 6. A dot matrix analysis of internal stretches of sequence identity in the IRES-EGFP insert. Each dot represents an 8 out of 8 nt direct repeat. Stretches of sequence identity marked with roman letters are marked in pink in Figure 5.
Using the published structure of a functional EMCV IRES element we can localize deletion site junctions on a pattern of paired and unpaired regions (Fig. 5). This analysis results in a highly significant bias for unpaired regions. We should stress that this overall pattern is found using three vectors harboring an EGFP cassette and that the preferred acceptor sites are found combined with different donor sites. Our study indicates that secondary structure may have a higher influence than sequence identity at or below 8–9 nt. It is therefore a possibility that molecular interactions not involving base-pairing may play an important role. A straightforward model of structure-determined acceptor template use proposes that unpaired regions provide better accessibility for the priming complex of RT and DNA. As an alternative possibility, the acceptor function may be facilitated by selective cleavage of non-paired RNA within the viral particle or nucleoprotein complex (2).
Logg et al. (28) have reported on deletion junction sites involving an EMCV IRES cassette isolated from a replication-competent MLV vector of a slightly different design. Among the six recombination events with a 5' deletion junction site within the IRES sequence described by Logg et al. (28), four are in unpaired regions, one event in a paired region and one event could not be conclusively ascribed to either paired or unpaired regions as depicted in Figure 5. Combining this data set with ours gives a bias for unpaired regions with a P-value of 0.0015%. We note that there seems to be a preference for the 5' unpaired region as one of the recombination partners. Whether this reflects a specific preference for this particular unpaired region or if it reflects generation of the largest deletion of the heterologous insert we cannot determine by the present experimental approach.
Although structural studies of retroviral RNAs have focused upon the 5' leader region, the complete retroviral RNA and not only the leader sequence may be highly structured in the dimer (29). In our study the heterologous RNA may be exempt from participating in this overall dimer structure and be accessible on the basis of its own secondary structure. The fact that the IRES element functions to direct marker protein expression within the retroviral RNA context provides support that it has retained its structure, when functioning as mRNA. Within the nucleoprotein complex where reverse transcription takes place, the viral RNA is coated with nucleocapsid (NC) protein, believed to act as a chaperone by lowering the activation energy for transition between structural conformations. Another feature of the NC protein is to assist the integrase in recognizing and processing the integration signals at the termini of the unintegrated double-stranded DNA. How the RNA structure of the IRES element interacts with the NC protein and whether this will have any effect on the single, quite stable secondary structure of this RNA remain uncertain.
Studies of recombination within retroviral sequences have identified cases where recombination frequencies are determined by a combination of structural features and sequence homologies. Such work has focused upon template shifts between homologous stem–loop structures such as the proposed kissing-loop dimerization structure within the leader region of MLV (30) and the TAR structure near the 5' end of the human immunodeficiency virus RNA (31). In these cases of recombination between identical or closely related sequences it could not be concluded if the determining structural features reside within the donor or acceptor sequence or in an overall dimeric structure. However, using an human immunodeficiency virus (HIV-1) derived in vitro homologous recombination system, Moumen et al. (15) recently acquired evidence that secondary structure at the acceptor was the determining factor for template switching. This feature applied under conditions both with and without added NC protein. Interestingly, this work mapped the preferred acceptor sites to hairpin structures, but, due to the sequence identity stretches at donor and acceptor templates, exact switch positions could not be defined. In our case, however, looking at non-homologous recombination, it was possible to assign transfer events more precisely to a single or a few nucleotides. By such analysis we found a statistically significant preference for unpaired versus paired acceptor nucleotides, i.e. the loop or bulge parts of the stem–loop structures. Although it is intriguing to compare the in vitro results from Moumen et al. (15) on homologous recombination with our in vivo data on non-homologous recombination, further work will be required to assess if structural determinants are the same for the two types of events.
Further studies will have to reveal if such structural determinants also play major roles in the recombination between natural retroviral sequences and therefore are important in processes that generate genetic diversity of retroviruses.
ACKNOWLEDGEMENTS
We thank Morten Frydenberg for statistical advice and Lene Svinth J?hnke for expert technical assistance. This work was supported by Karen Elise Jensen’s Fund, the Danish Cancer Society and the Danish Research Agency.
REFERENCES
Hu,W.S. and Temin,H.M. (1990) Genetic consequences of packaging two RNA genomes in one retroviral particle: pseudodiploidy and high rate of genetic recombination. Proc. Natl Acad. Sci. USA, 87, 1556–1560.
Negroni,M. and Buc,H. (2001) Mechanisms of retroviral recombination. Annu. Rev. Genet., 35, 275–302.
Zhang,J., Tang,L.Y., Li,T., Ma,Y. and Sapp,C.M. (2000) Most retroviral recombinations occur during minus-strand DNA synthesis. J. Virol., 74, 2313–2322.
Hu,W.S. and Temin,H.M. (1990) Retroviral recombination and reverse transcription. Science, 250, 1227–1233.
Pathak,V.K. and Temin,H.M. (1990) Broad spectrum of in vivo forward mutations, hypermutations and mutational hotspots in a retroviral shuttle vector after a single replication cycle: deletions and deletions with insertions. Proc. Natl. Acad. Sci. USA, 87, 6024–6028.
An,W. and Telesnitsky,A. (2002) Effects of varying sequence similarity on the frequency of repeat deletion during reverse transcription of a human immunodeficiency virus type 1 vector. J. Virol., 76, 7897–7902.
Pfeiffer,J.K. and Telesnitsky,A. (2001) Effects of limiting homology at the site of intermolecular recombinogenic template switching during Moloney murine leukemia virus replication. J. Virol., 75, 11263–11274.
Coffin,J.M. (1979) Structure, replication and recombination of retrovirus genomes: some unifying hypotheses. J. Gen. Virol., 42, 1–26.
Hwang,C.K., Svarovskaia,E.S. and Pathak,V.K. (2001) Dynamic copy choice: steady state between murine leukemia virus polymerase and polymerase-dependent RNase H activity determines frequency of in vivo template switching. Proc. Natl Acad. Sci. USA, 98, 12209–12214.
Delviks,K.A. and Pathak,V.K. (1999) Effect of distance between homologous sequences and 3' homology on the frequency of retroviral reverse transcriptase template switching. J. Virol., 73, 7923–7932.
Pfeiffer,J.K., Georgiadis,M.M. and Telesnitsky,A. (2000) Structure-based moloney murine leukemia virus reverse transcriptase mutants with altered intracellular direct-repeat deletion frequencies. J. Virol., 74, 9629–9636.
Pfeiffer,J.K., Topping,R.S., Shin,N.H. and Telesnitsky,A. (1999) Altering the intracellular environment increases the frequency of tandem repeat deletion during Moloney murine leukemia virus reverse transcription. J. Virol., 73, 8441–8447.
Svarovskaia,E.S., Delviks,K.A., Hwang,C.P. and Pathak,V.K. (2000) Structural determinants of murine leukemia virus reverse transcriptase that affect the frequency of template switching. J. Virol., 74, 7171–7178.
Moumen,A., Polomack,L., Roques,B., Buc,H. and Negroni,M. (2001) The HIV-1 repeated sequence R as a robust hot-spot for copy-choice recombination. Nucleic Acids Res., 29, 3814–3821.
Moumen,A., Polomack,L., Unge,T., Veron,M., Buc,H. and Negroni,M. (2003) Evidence for a mechanism of recombination during reverse transcription dependent on the structure of the acceptor RNA. J. Biol. Chem., 278, 15973–15982.
Jang,S.K., Krausslich,H.G., Nicklin,M.J., Duke,G.M., Palmenberg,A.C. and Wimmer,E. (1988) A segment of the 5' nontranslated region of encephalomyocarditis virus RNA directs internal entry of ribosomes during in vitro translation. J. Virol., 62, 2636–2643.
Pilipenko,E.V., Blinov,V.M., Chernov,B.K., Dmitrieva,T.M. and Agol,V.I. (1989) Conservation of the secondary structure elements of the 5'-untranslated region of cardio- and aphthovirus RNAs. Nucleic Acids Res., 17, 5701–5711.
Jang,S.K. and Wimmer,E. (1990) Cap-independent translation of encephalomyocarditis virus RNA: structural elements of the internal ribosomal entry site and involvement of a cellular 57-kD RNA-binding protein. Genes Dev., 4, 1560–1572.
Bachrach,E., Pelegrin,M., Piechaczyk,M., Pedersen,F.S. and Duch,M. (2002) Efficient gene transfer into spleen cells of newborn mice by a replication-competent retroviral vector. Virology, 293, 328–334.
Morgan,R.A., Couture,L., Elroy-Stein,O., Ragheb,J., Moss,B. and Anderson,W.F. (1992) Retroviral vectors containing putative internal ribosome entry sites: development of a polycistronic gene transfer system and applications to human gene therapy. Nucleic Acids Res., 20, 1293–1299.
Jespersen,T., Duch,M., Carrasco,M.L., Warming,S. and Pedersen,F.S. (1999) Expression of heterologous genes from an IRES translational cassette in replication competent murine leukemia virus vectors. Gene, 239, 227–235.
Koo,H.M., Brown,A.M., Kaufman,R.J., Prorock,C.M., Ron,Y. and Dougherty,J.P. (1992) A spleen necrosis virus-based retroviral vector which expresses two genes from a dicistronic mRNA. Virology, 186, 669–675.
Lund,A.H. and Pedersen,F.S. (1999) The nucleotide sequence of the high-leukemogenic murine retrovirus SL3-3 reveals a patch of mink cell focus forming-like sequences upstream of the ecotropic envelope gene. Arch. Virol., 144, 2207–2212.
Duch,M., Carrasco,M.L., Jespersen,T., Hansen,B.D. and Pedersen,F.S. (2004) Transgene stability for three replication competent murine leukemia virus vectors. Gene, 329, 61–69.
Graham,F.L. and van der Eb,A.J. (1973) A new technique for the assay of infectivity of human adenovirus 5 DNA. Virology, 52, 456–467.
Diaz,L., Cristofaro,J.V. and DeStefano,J.J. (2000) Human immunodeficiency virus reverse transcriptase base misincorporations can promote strand transfer. Arch. Virol., 145, 1117–1131.
Kulpa,D., Topping,R. and Telesnitsky,A. (1997) Determination of the site of first strand transfer during Moloney murine leukemia virus reverse transcription and identification of strand transfer-associated reverse transcriptase errors. EMBO J., 16, 856–865.
Logg,C.R., Logg,A., Tai,C.K., Cannon,P.M. and Kasahara,N. (2001) Genomic stability of murine leukemia viruses containing insertions at the Env-3' untranslated region boundary. J. Virol., 75, 6989–6998.
Huthoff,H. and Berkhout,B. (2002) Multiple secondary structure rearrangements during HIV-1 RNA dimerization. Biochemistry, 41, 10439–10445.
Mikkelsen,J.G., Lund,A.H., Duch,M. and Pedersen,F.S. (2000) Mutations of the kissing-loop dimerization sequence influence the site specificity of murine leukemia virus recombination in vivo. J. Virol., 74, 600–610.
Kim,J.K., Palaniappan,C., Wu,W., Fay,P.J. and Bambara,R.A. (1997) Evidence for a unique mechanism of strand transfer from the transactivation response region of HIV-1. J. Biol. Chem., 272, 16769–16777.(Mogens Duch1, Maria L. Carrasco1, Thomas)
*To whom correspondence should be addressed at Department of Molecular Biology, University of Aarhus, DK-8000 Aarhus, Denmark. Tel: +45 8 9421111, direct, +45 8 9422614; Fax: +45 8 6196500; Email: fsp@mb.au.dk
Present addresses:
Thomas Jespersen, Department of Medical Physiology, The Panum Institue, University of Copenhagen, Blegdamsvej 3, DK-2200 Copenhagen N., Denmark
Lars Aagaard, Bioinformatics Research Center, Department of Computer Science, Aarhus University, DK-8000 Aarhus, Denmark
ABSTRACT
Murine leukemia viruses harboring an internal ribosome entry site (IRES)-directed translational cassette are able to replicate, but undergo loss of heterologous sequences upon continued passage. While complete loss of heterologous sequences is favored when these are flanked by a direct repeat, deletion mutants with junction sites within the heterologous cassette may also be retrieved, in particular from vectors without flanking repeats. Such deletion mutants were here used to investigate determinants of reverse transcriptase-mediated non-homologous recombination. Based upon previous structural analysis the individual recombination sites within the IRES could be assigned to either base-paired or unpaired regions of RNA. This assignment showed a significant bias (P = 0.000082) towards recombination within unpaired regions of the IRES. We propose that the events observed in this in vivo system result from template switching during first-strand cDNA synthesis and that the choice of acceptor sites for non-homologous recombination are guided by non-paired regions. Our results may have implications for recombination events taking place within structured regions of retroviral RNA genomes, especially in the absence of longer stretches of sequence similarity.
INTRODUCTION
Template switching during reverse transcription contributes to genetic recombination in retroviruses (1,2). Although recombination may also take place during plus-strand synthesis, recombination by template switching during minus-strand synthesis is probably more frequent and has been intensively studied (3). By intermolecular template-switching reverse transcriptase (RT) switches between the two complete RNA genomes that are co-packaged in a retroviral particle (4). Template switching can also take place intramolecularly between two positions on the same RNA (5). Analysis of the requirements for RT template switching in vitro and in vivo has revealed that sequence similarity at the donor and acceptor site greatly facilitates recombination (6,7). However, non-homologous recombination events between sites without any sequence identity occur at low frequencies (7).
The original forced copy-choice model proposed that recombination occurs when RT encounters a break in the RNA template during minus-strand synthesis (8). This model has been extended to include events not dependent upon template RNA breakage. In a further development it was recently shown that the RT-associated RNase H activity plays a critical role for minus-strand recombination in vivo. This dynamic copy choice model (9) states that the balance between the rate of DNA polymerization and the rate of RNA template degradation behind the RT determines the amount of nascent DNA that is free to pair with an acceptor template.
Features of the donor site such as secondary structures have been found to influence the rate of template switching (10–13). Such structures have also recently been reported to play a role at the acceptor site (7,14,15). In the present work, we used a heterologous sequence including the encephalomyocarditis virus (EMCV) internal ribosome entry site (IRES) element as a target sequence for deletions in an in vivo assay. This IRES element was first described as mediating internal ribosome entry on polycistronic RNA of the EMCV (16). The structure has been extensively characterized (17,18) and correct folding of the RNA in a heterologous context can furthermore be judged by the ability of the element to direct translation of a reporter protein such as the enhanced green fluorescence protein (EGFP).
By using a heterologous cassette inserted into a complete viral genome the selective pressure upon the resultant vector is reduced to the gain in replication fitness acquired by deletions in the insert. Moreover, since there is no positive selection for regeneration of specific sequences all deletions can be assumed to result in replication competence, at least as long as they are confined to the heterologous cassette.
Some deletions will result in complete loss, others in only partial loss of the cassette. In this report we look only at the latter type with one or both deletion site junctions within an IRES-EGFP cassette containing internal direct repeats of a maximum of 9 nt. We note that the deletion site junctions lack sequence similarity, and, looking only at junctions within the IRES, we find a significant bias towards sites predicted to be in non-paired regions of the IRES RNA.
MATERIALS AND METHODS
Plasmids
The plasmid-cloned replication-competent vectors are AkvU3-EGFP, AkvSL33U3-EGFP and AkvEn-EGFP. AkvU3-EGFP has been described previously (19). It harbors an IRES-EGFP insert in the Cel II site of the 3' long terminal repeat (LTR) of the Akv virus. The linker (20) between the 3' end of the EMCV IRES element and the start of the EGFP gene in AkvU3-EGFP ensures 10-fold higher translation of the EGFP gene as compared to other linkers (21,22).
The AkvSL33U3-EGFP will be described elsewhere (M. L. Carrasco, M. Duch and F.S. Pedersen, submitted for publication). In short, it was constructed by replacing the 3' LTR of AkvU3-EGFP downstream of the IRES-EGFP insert with a polymerase chain reaction (PCR) amplified fragment of the corresponding sequence from SL3-3 (23). The primers located in the 5' end of the LTR were designed to delete the 3' Cel II site flanking the insert in AkvU3-EGFP. The resulting AkvSL33U3-EGFP construct has the following sequence with no direct repeat flanking the insert: 5'-GCTTAGC-TGCAGATGCATGGCCCATGCGGCCGC-IRES-EGFP-CAG CTAACT-3'. Underlined nucleotides are derived from the LTR whereas letters in italics denote the heterologous polylinker insert.
The AkvEn-EGFP construct (24) was derived from pAkv (21) where the IRES-EGFP cassette has been inserted 3 nt after the termination codon of the envelope gene followed by a duplicated sequence of the C-terminal region of the envelope gene upstream of the polypurine tract (PPT) resulting in a repeated structure flanking the insert. In the coding sequence for the C-terminal part of the envelope gene a CGT codon was exchanged to another arginine codon CGG in the C-terminal part of the envelope gene. The T and the G residues are shown in bold in Figure 1A.
Figure 1. Experimental design. (A) Schematic representation of the three replication-competent retroviral vectors used. Letters in bold are not part of a repeat. Figures are not drawn to scale. PSI, packaging signal. (B) Outline of experimental procedure for isolation of deletion mutants.
Cell culture, transfection and transductions
NIH3T3 cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) containing Glutamax-1 (Gibco BRL) supplemented with 10% newborn calf serum (NCS) (Life Technologies, Inc.), 1 unit of penicillin and 1 μg of streptomycin per 1 ml of final medium. BOSC 23 cells were grown as described for NIH3T3 cells except that NCS was exchanged with fetal calf serum (FCS). Prior to usage, the BOSC 23 cells were selected in HAT medium as described (21). For generation of replication-competent viruses, BOSC 23 cells were seeded at a density of 7 x 104 cells/cm2 1 day prior to transfection. The cells were transfected with 10 μg of plasmid DNA by the calcium phosphate method without a glycerol shock as described (25). One day post-transfection the medium was renewed on the BOSC 23 cells and fresh NIH3T3 cells were seeded in 75 cm2 flasks at a density of 5 x 103 cells/cm2. Two days post-transfection undiluted virus-containing medium was transferred from the BOSC 23 cells to the NIH3T3 cells, supplemented with 6 μg/ml Polybrene (Aldrich Chemical Co. Inc.) and cultured for 1 day. Twenty-four hours post-infection the medium was exchanged with fresh medium containing 2 μg/ml Polybrene and the cultures were passaged for 1 week. For consecutive rounds of infection (two to six), the same scheme was followed (Fig. 1B).
DNA isolation, PCR amplification and subcloning
DNA was isolated using the DNAzol genomic DNA isolation reagent according to the manufacturer’s protocol (Molecular Research Center, Inc.). DNA fragments containing deletion mutants were amplified using standard PCR techniques. In brief, PCR amplification was performed in 100 μl reactions containing 1 μg of purified DNA, 25 pmol of each primer, 0.2 mM dNTPs and 2.5 units of AmpliTaq Gold DNA polymerase (Perkin-Elmer Cetus). After an initial activation step of 10 min at 94°C, the reaction was followed by 40 cycles of 1.2 min at 94°C, 1.2 min at 60°C and 3 min at 72°C. The reaction was terminated with a final incubation of 7 min at 72°C. The oligonucleotides used as primers are indicated in Figure 2A and are as follows: 1, 5'-CAACAAGGGTGGT TTGAAGGGCTGTTTAA-3'; 2, 5'-GAATTCGATATCGA TCCCCGGTCATCTGGG-3'; 3, 5'-GATCGCTTAGCTGC AGATGCATGGCCCATGCGGCCGCCCCCTAAC-3'; and 4, 5'-AGACCTGGATCCCGCTTTACTTGTACAGCTCG TCCATGC-3'. The PCR products were electrophoresed on agarose gels and bands migrating between the products amplified from a non-reverted replication-competent vector and a complete revertant, were isolated using a GFX purification kit (Amersham Pharmacia Biotech Inc.) according to instructions from the manufacturer. The resulting products were cloned in pGEM-Teasy vector (Promega) for subsequent sequencing. Bacterial clones were screened by PCR using the same primers as used in the initial PCR amplification of the deletion mutants from the genomic DNA and positive clones picked for mini-preparation and sequence analysis of plasmid DNA. Primers used for sequencing were located in the pGEM-Teasy vector outside the cloning site, either 5'-GTTTTCCCAGTCACG-3' or 5'-CAGGAAACAGCTATG-3'. The sequencing reaction was performed using the Thermo Sequenase II dye terminator cycle sequencing kit (Amersham Pharmacia Biotech Inc.) and Applied Biosystems 373A DNA sequencer.
Figure 2. Retrieval of deletion site junctions. (A) Overview of positions of oligonucleotide primers for PCR amplification of the IRES-EGFP deletion mutants. (B) Example of an agarose gel analysis of the resultant PCR product from AkvEn-EGFP infected NIH3T3 cells using primers 3 and 2. Boxed bands/smear represent possible deletion mutants. DNA from boxed area was isolated, subcloned and sequenced. From left to right: lane 1, DNA marker; lane 2, 4th passage; lane 3, 5th passage; lane 4, 6th passage; lanes 5 and 6, DNA markers.
RESULTS
Replication-competent retroviruses with IRES sequence
We used three replication-competent murine leukemia viruses (MLV) where an IRES-EGFP cassette was inserted into either Akv or a chimera of Akv and SL3-3 viruses (19,24 and M. L. Carrasco, M. Duch and F. S. Pedersen, submitted for publication). Two insertion sites were used, one was located between the envelope gene and the PPT resulting in only one copy of the IRES-EGFP element at both RNA and DNA level. This construct (AkvEn-EGFP) was based upon the Akv virus where the IRES-EGFP element was flanked by a 38 out of 39 nt direct repeat due to duplication of sequences upstream of the PPT (see Fig. 1A).
The second site used for insertion was situated 20 nt 3' of the start of the U3 region in the downstream LTR. Being located at this position, the IRES-EGFP cassette will be copied to both the upstream and the downstream LTR during reverse transcription of the genomic RNA to double-stranded DNA. Only the IRES-EGFP cassette placed in the downstream LTR will be transcribed into viral RNA and thereby participate in further cycles of reverse transcription. In these two constructs where the IRES-EGFP element was inserted into the LTR, the Cel II site located in the upstream part of the U3 region was used. The first construct, AkvSL33U3-EGFP, was based upon a chimeric virus where the LTR of the Akv virus was exchanged with the LTR of the SL3-3 virus. This construct contained no direct repeat flanking the IRES-EGFP element. In the second construct, AkvU3-EGFP, the insert was flanked by a duplicated Cel II site giving rise to a perfect direct repeat of 8 out of 8 nt residing within an imperfect direct repeat of 13 out of 14 nt flanking the IRES-EGFP element (Fig. 1A).
Experimental procedure
In separate work we have analyzed the genomic stability of replication-competent vectors during repeated passages in cell culture (21,24). In those studies, BOSC 23 packaging cells were transiently transfected with a replication-competent vector construct and undiluted media transferred to NIH3T3 cells 48 h post-transfection. These infected NIH3T3 cells were passaged for 1 week, whereafter undiluted supernatant was transferred to fresh NIH3T3 cells for a second round of infection events (Fig. 1B). Following this experimental procedure, revertants started to appear between rounds two and five as analyzed by northern blotting (21,24). Each round of transfer of fresh NIH3T3 cells probably represents several rounds of viral infection events as only 40% of cells were infected 48 h post-transduction, whereas after 1 week more than 99% of the cells in the population were infected, as analyzed for EGFP expression by fluorescence microscopy.
In the present study, we were interested only in those recombination events which had resulted in deletions with one or both junctions located within the heterologous cassette. According to this criterion, the analysis does not include mutants where the complete transgene cassette is lost.
For the AkvU3-EGFP and AkvEn-EGFP vectors the vast majority of such complete cassette deletions are driven by the flanking direct repeats (24). For optimal detection of partial cassette deletions, DNA from infection rounds four to six was analyzed by PCR amplification using the oligonucleotide primers (Fig. 2A) in various combinations. One example of such a PCR amplification of DNA from round four to six for AkvU3-EGFP is shown in Figure 2A. Using primers located outside the insert (primers 1 and 2), complete reversion to wild type could be detected for all three constructs, as verified by PCR amplification and sequencing (24).
Mapping sites of recombination
This population of bands, as exemplified by the boxed areas in Figure 2B, was isolated, subcloned and sequenced. In several cases the same deletion mutant was isolated more than once from different rounds of the same culture, sometimes from all three rounds, most likely resulting from a single primary deletion event and counted as such in this analysis. Notably, none of the junctions arising from deletions within the IRES-EGFP cassette were independently isolated from the three different viral constructs used. Figure 3 is a graphical representation of all the deletion sites. Some of the deletion sites identified in this set-up reach into the retroviral sequences of the vector as shown in Figure 3, which also depicts critical cis elements for viral replication next to the insert. These are the PPT and integration site signals at the upstream U3 border located 3' to the heterologous cassette in AkvEn-EGFP and 5' to the cassette in AkvSL33U3-EGFP and AkvU3-EGFP. We note that deletion junctions found for AkvEn-EGFP extend into viral sequences at the 5' site but do not extend beyond the cis restriction at the 3' site. For the AkvSL33U3-EGFP construct, deletion sites do not map beyond the cis restriction at the 5' end of the LTR, whereas the 3' deletion junctions extend 200 nt into the U3 sequences where no critical cis elements are located. The few non-complete junction site revertants that were retrieved from the vast majority of complete revertants from the AkvU3-EGFP construct, conform to the same patterns. Altogether, the distribution of deletion borders, as depicted in Figure 3, suggests that the maintenance of viral cis elements is crucial, making it unlikely that the observed pattern is caused by artefacts generated during the post-replication steps (PCR, subcloning and sequencing).
Figure 3. Bar chart of deletion junctions. Indicated below the bar chart are the functional domains of the replication-competent vectors. Small vertical bars represent stretches of homology. Arrows point to borders of cis element restrictions. Numbers at the axes refer to the nucleotide positions in the corresponding plasmids of the replication-competent vectors.
Analysis of sequence similarity at the deletion site
In order to determine if any correlation between RNA primary or secondary structure and deletion junction site distribution existed, only clones where one of the deletion site junctions was located within the IRES element were used for further analysis. Solely 5' recombination sites were recovered from the IRES element.
Depicted in Figure 4 are 10 nt of the upstream and downstream deletion site junctions. The sequences were analyzed for direct repeats at the junction sites potentially used for homologous recombination as well as additional inserts of nucleotides at the deletion site. For each recombinant viral vector, the size of the deleted sequence is given in nucleotides, ranging from 612 to 1374 nt. The lack of smaller deletions may relate to the selective forces operating during selection for replication fitness as well as to the gel excision procedure. Again, all revertants resulting in complete loss of the heterologous cassette were excluded from the analysis. Inspection of Figure 4 reveals that the deletion site junctions harbor no or only short repeats. This indicates that sequence similarity is not the major determinant of deletion site selection in this set-up.
Figure 4. Deletion site junction sequences. The numbers at the extreme left specify the reference numbers of the isolated clones. Numbers internal in the sequence indicate the size of the deletion. Left, semi-boxed sequences represent nucleotides at the 5' site maintained after recombination. Right, semi-boxed sequences represent 3' nucleotides maintained after recombination. Internal boxed sequence represents deleted sequences. Sequences in bold (and italics) represent nucleotide insertions at the deletion site. Sequences in bold and underlined represent direct repeats at the deletion junctions.
Correlation between recombination site distribution and RNA secondary structure
The IRES element from EMCV has been extensively characterized (16) and the secondary structure given in Figure 5 was unravelled by Pilipenko et al. (17) and Jang and Wimmer (18). The deletion junction sites were depicted on this RNA structure to assess a possible correlation between RNA structure and recombination site distribution. Also shown are the cases of sequence identity at the recombination site identified in Figure 4. For the accumulated data set of all constructs we note a clustering of deletion site junctions in the same unpaired region in the beginning of the IRES element, position 50–57 in Figure 5.
Figure 5. Schematic representation of the upstream deletion site junctions in the RNA secondary structure of the EMCV IRES element. The model is based upon the published secondary structure (17,18). The brown lines represent data from Logg et al. (28) while the remaining three colors represent the various constructs used in the present work. The length of the lines indicate that the recombination has occurred somewhere between these nucleotides. The length varies due to the different size of direct repeat at the recombination site. The two deletion events involving extra nucleotides (Fig. 4) have been depicted as if there were no extra nucleotides. Letters in roman and nucleotides in pink represent stretches of sequence homology 8 internal in the IRES-EGFP insert described in Figure 6.
The IRES structure was divided into paired or unpaired nucleotides giving a total of 334 paired as compared to 224 unpaired nucleotides, i.e. 40% unpaired nucleotides (Fig. 5 and Tables 1 and 2). From the experimental data set only recombination events occurring exclusively in either an unpaired or paired region were used to determine the distribution of recombination sites. From a total of 39 recombination events (Fig. 4), 34 could conclusively be ascribed to either an unpaired or a paired region of the IRES element (Fig. 5). Of these 34, 25 had occurred in an unpaired region and nine had occurred in a paired region. The accumulated likelihood of observing 25 or more events of recombination involving unpaired regions as acceptor out of a total of 34 events is 0.0082%, as evaluated in a binomial distribution with a probability parameter of 40% for unpaired events (Tables 1 and 2).
Table 1. Distribution of recombination sites between unpaired and paired regions of the IRES sequence
Table 2. Likelihood of recombination events occurring in an unpaired region
DISCUSSION
We have used deletions within a heterologous sequence cassette in a replication-competent vector to study features of a template that favors retroviral recombination. In this set-up, it may be assumed that the increased replication fitness associated with deletion of non-viral sequences is the sole force driving the increased prevalence of deletion mutants upon multiple passages. The presence of a repeat flanking the insert will facilitate complete deletion of the heterologous cassette. We may also assume that all deletions within the heterologous insert will be compatible with continued replication; however, it is not unlikely that individual parts of the insert may affect replication fitness differently. In compliance with the expected continued replication of deletion mutants we have many examples of repeated isolation of the same mutant at different passages, some deletion mutants were even isolated from all three passages analyzed. It cannot be excluded that some of the deletion mutants retrieved result from consecutive deletion events rather than a single event. However, we have observed no case of two or more separate deletions within the same isolate.
In two of the vectors used, a direct repeat sequence flanks the heterologous insert (see Fig. 1A). Such repeats favor the perfect excision of the insert by homologous recombination and result in the regeneration of replication-competent viruses without insert. In spite of this favored process of complete excision in these vectors, a small number of deletions using other sites with little sequence identity at the deletion borders may be retrieved (Fig. 4).
We note that the repeated isolation of mutants with exactly the same deletion junctions from different passages argues that the number of incomplete cassette deletion mutants within a given pool is limited. On the other hand, the fact that the junctions retrieved from separate experiments are different argues in favor of a large number of potential deletion junctions.
Looking at 10 nt on either site of the deletions in the analysis (Fig. 4), we do not observe a clear pattern of direct and inverted repeats at the junctions. This indicates that a majority of the deletions result from non-homologous recombination events. Most of the events resulted in a perfect junction, but in two out of the 40 cases analyzed 1 or 2 nt had been inserted at the junction, presumably by non-templated nucleotide addition (26,27). We looked for candidate target sequences for homologous recombination to further assess the lack of sequence identity at the retrieved junction sites. Scanning in windows of 8 nt (Fig. 6) revealed stretches of sequence identity of up to only 9 nt, i.e. below the sequence identity windows of 14 nt or more found to be critical for homologous recombination in a previous study (7). The numbers in roman letters at the identity stretches correspond to the sequences given in Figure 5. We note that none of these putative homologous recombination events are identified in the present screens, even though they would result in deletions within the observed size ranges.
Figure 6. A dot matrix analysis of internal stretches of sequence identity in the IRES-EGFP insert. Each dot represents an 8 out of 8 nt direct repeat. Stretches of sequence identity marked with roman letters are marked in pink in Figure 5.
Using the published structure of a functional EMCV IRES element we can localize deletion site junctions on a pattern of paired and unpaired regions (Fig. 5). This analysis results in a highly significant bias for unpaired regions. We should stress that this overall pattern is found using three vectors harboring an EGFP cassette and that the preferred acceptor sites are found combined with different donor sites. Our study indicates that secondary structure may have a higher influence than sequence identity at or below 8–9 nt. It is therefore a possibility that molecular interactions not involving base-pairing may play an important role. A straightforward model of structure-determined acceptor template use proposes that unpaired regions provide better accessibility for the priming complex of RT and DNA. As an alternative possibility, the acceptor function may be facilitated by selective cleavage of non-paired RNA within the viral particle or nucleoprotein complex (2).
Logg et al. (28) have reported on deletion junction sites involving an EMCV IRES cassette isolated from a replication-competent MLV vector of a slightly different design. Among the six recombination events with a 5' deletion junction site within the IRES sequence described by Logg et al. (28), four are in unpaired regions, one event in a paired region and one event could not be conclusively ascribed to either paired or unpaired regions as depicted in Figure 5. Combining this data set with ours gives a bias for unpaired regions with a P-value of 0.0015%. We note that there seems to be a preference for the 5' unpaired region as one of the recombination partners. Whether this reflects a specific preference for this particular unpaired region or if it reflects generation of the largest deletion of the heterologous insert we cannot determine by the present experimental approach.
Although structural studies of retroviral RNAs have focused upon the 5' leader region, the complete retroviral RNA and not only the leader sequence may be highly structured in the dimer (29). In our study the heterologous RNA may be exempt from participating in this overall dimer structure and be accessible on the basis of its own secondary structure. The fact that the IRES element functions to direct marker protein expression within the retroviral RNA context provides support that it has retained its structure, when functioning as mRNA. Within the nucleoprotein complex where reverse transcription takes place, the viral RNA is coated with nucleocapsid (NC) protein, believed to act as a chaperone by lowering the activation energy for transition between structural conformations. Another feature of the NC protein is to assist the integrase in recognizing and processing the integration signals at the termini of the unintegrated double-stranded DNA. How the RNA structure of the IRES element interacts with the NC protein and whether this will have any effect on the single, quite stable secondary structure of this RNA remain uncertain.
Studies of recombination within retroviral sequences have identified cases where recombination frequencies are determined by a combination of structural features and sequence homologies. Such work has focused upon template shifts between homologous stem–loop structures such as the proposed kissing-loop dimerization structure within the leader region of MLV (30) and the TAR structure near the 5' end of the human immunodeficiency virus RNA (31). In these cases of recombination between identical or closely related sequences it could not be concluded if the determining structural features reside within the donor or acceptor sequence or in an overall dimeric structure. However, using an human immunodeficiency virus (HIV-1) derived in vitro homologous recombination system, Moumen et al. (15) recently acquired evidence that secondary structure at the acceptor was the determining factor for template switching. This feature applied under conditions both with and without added NC protein. Interestingly, this work mapped the preferred acceptor sites to hairpin structures, but, due to the sequence identity stretches at donor and acceptor templates, exact switch positions could not be defined. In our case, however, looking at non-homologous recombination, it was possible to assign transfer events more precisely to a single or a few nucleotides. By such analysis we found a statistically significant preference for unpaired versus paired acceptor nucleotides, i.e. the loop or bulge parts of the stem–loop structures. Although it is intriguing to compare the in vitro results from Moumen et al. (15) on homologous recombination with our in vivo data on non-homologous recombination, further work will be required to assess if structural determinants are the same for the two types of events.
Further studies will have to reveal if such structural determinants also play major roles in the recombination between natural retroviral sequences and therefore are important in processes that generate genetic diversity of retroviruses.
ACKNOWLEDGEMENTS
We thank Morten Frydenberg for statistical advice and Lene Svinth J?hnke for expert technical assistance. This work was supported by Karen Elise Jensen’s Fund, the Danish Cancer Society and the Danish Research Agency.
REFERENCES
Hu,W.S. and Temin,H.M. (1990) Genetic consequences of packaging two RNA genomes in one retroviral particle: pseudodiploidy and high rate of genetic recombination. Proc. Natl Acad. Sci. USA, 87, 1556–1560.
Negroni,M. and Buc,H. (2001) Mechanisms of retroviral recombination. Annu. Rev. Genet., 35, 275–302.
Zhang,J., Tang,L.Y., Li,T., Ma,Y. and Sapp,C.M. (2000) Most retroviral recombinations occur during minus-strand DNA synthesis. J. Virol., 74, 2313–2322.
Hu,W.S. and Temin,H.M. (1990) Retroviral recombination and reverse transcription. Science, 250, 1227–1233.
Pathak,V.K. and Temin,H.M. (1990) Broad spectrum of in vivo forward mutations, hypermutations and mutational hotspots in a retroviral shuttle vector after a single replication cycle: deletions and deletions with insertions. Proc. Natl. Acad. Sci. USA, 87, 6024–6028.
An,W. and Telesnitsky,A. (2002) Effects of varying sequence similarity on the frequency of repeat deletion during reverse transcription of a human immunodeficiency virus type 1 vector. J. Virol., 76, 7897–7902.
Pfeiffer,J.K. and Telesnitsky,A. (2001) Effects of limiting homology at the site of intermolecular recombinogenic template switching during Moloney murine leukemia virus replication. J. Virol., 75, 11263–11274.
Coffin,J.M. (1979) Structure, replication and recombination of retrovirus genomes: some unifying hypotheses. J. Gen. Virol., 42, 1–26.
Hwang,C.K., Svarovskaia,E.S. and Pathak,V.K. (2001) Dynamic copy choice: steady state between murine leukemia virus polymerase and polymerase-dependent RNase H activity determines frequency of in vivo template switching. Proc. Natl Acad. Sci. USA, 98, 12209–12214.
Delviks,K.A. and Pathak,V.K. (1999) Effect of distance between homologous sequences and 3' homology on the frequency of retroviral reverse transcriptase template switching. J. Virol., 73, 7923–7932.
Pfeiffer,J.K., Georgiadis,M.M. and Telesnitsky,A. (2000) Structure-based moloney murine leukemia virus reverse transcriptase mutants with altered intracellular direct-repeat deletion frequencies. J. Virol., 74, 9629–9636.
Pfeiffer,J.K., Topping,R.S., Shin,N.H. and Telesnitsky,A. (1999) Altering the intracellular environment increases the frequency of tandem repeat deletion during Moloney murine leukemia virus reverse transcription. J. Virol., 73, 8441–8447.
Svarovskaia,E.S., Delviks,K.A., Hwang,C.P. and Pathak,V.K. (2000) Structural determinants of murine leukemia virus reverse transcriptase that affect the frequency of template switching. J. Virol., 74, 7171–7178.
Moumen,A., Polomack,L., Roques,B., Buc,H. and Negroni,M. (2001) The HIV-1 repeated sequence R as a robust hot-spot for copy-choice recombination. Nucleic Acids Res., 29, 3814–3821.
Moumen,A., Polomack,L., Unge,T., Veron,M., Buc,H. and Negroni,M. (2003) Evidence for a mechanism of recombination during reverse transcription dependent on the structure of the acceptor RNA. J. Biol. Chem., 278, 15973–15982.
Jang,S.K., Krausslich,H.G., Nicklin,M.J., Duke,G.M., Palmenberg,A.C. and Wimmer,E. (1988) A segment of the 5' nontranslated region of encephalomyocarditis virus RNA directs internal entry of ribosomes during in vitro translation. J. Virol., 62, 2636–2643.
Pilipenko,E.V., Blinov,V.M., Chernov,B.K., Dmitrieva,T.M. and Agol,V.I. (1989) Conservation of the secondary structure elements of the 5'-untranslated region of cardio- and aphthovirus RNAs. Nucleic Acids Res., 17, 5701–5711.
Jang,S.K. and Wimmer,E. (1990) Cap-independent translation of encephalomyocarditis virus RNA: structural elements of the internal ribosomal entry site and involvement of a cellular 57-kD RNA-binding protein. Genes Dev., 4, 1560–1572.
Bachrach,E., Pelegrin,M., Piechaczyk,M., Pedersen,F.S. and Duch,M. (2002) Efficient gene transfer into spleen cells of newborn mice by a replication-competent retroviral vector. Virology, 293, 328–334.
Morgan,R.A., Couture,L., Elroy-Stein,O., Ragheb,J., Moss,B. and Anderson,W.F. (1992) Retroviral vectors containing putative internal ribosome entry sites: development of a polycistronic gene transfer system and applications to human gene therapy. Nucleic Acids Res., 20, 1293–1299.
Jespersen,T., Duch,M., Carrasco,M.L., Warming,S. and Pedersen,F.S. (1999) Expression of heterologous genes from an IRES translational cassette in replication competent murine leukemia virus vectors. Gene, 239, 227–235.
Koo,H.M., Brown,A.M., Kaufman,R.J., Prorock,C.M., Ron,Y. and Dougherty,J.P. (1992) A spleen necrosis virus-based retroviral vector which expresses two genes from a dicistronic mRNA. Virology, 186, 669–675.
Lund,A.H. and Pedersen,F.S. (1999) The nucleotide sequence of the high-leukemogenic murine retrovirus SL3-3 reveals a patch of mink cell focus forming-like sequences upstream of the ecotropic envelope gene. Arch. Virol., 144, 2207–2212.
Duch,M., Carrasco,M.L., Jespersen,T., Hansen,B.D. and Pedersen,F.S. (2004) Transgene stability for three replication competent murine leukemia virus vectors. Gene, 329, 61–69.
Graham,F.L. and van der Eb,A.J. (1973) A new technique for the assay of infectivity of human adenovirus 5 DNA. Virology, 52, 456–467.
Diaz,L., Cristofaro,J.V. and DeStefano,J.J. (2000) Human immunodeficiency virus reverse transcriptase base misincorporations can promote strand transfer. Arch. Virol., 145, 1117–1131.
Kulpa,D., Topping,R. and Telesnitsky,A. (1997) Determination of the site of first strand transfer during Moloney murine leukemia virus reverse transcription and identification of strand transfer-associated reverse transcriptase errors. EMBO J., 16, 856–865.
Logg,C.R., Logg,A., Tai,C.K., Cannon,P.M. and Kasahara,N. (2001) Genomic stability of murine leukemia viruses containing insertions at the Env-3' untranslated region boundary. J. Virol., 75, 6989–6998.
Huthoff,H. and Berkhout,B. (2002) Multiple secondary structure rearrangements during HIV-1 RNA dimerization. Biochemistry, 41, 10439–10445.
Mikkelsen,J.G., Lund,A.H., Duch,M. and Pedersen,F.S. (2000) Mutations of the kissing-loop dimerization sequence influence the site specificity of murine leukemia virus recombination in vivo. J. Virol., 74, 600–610.
Kim,J.K., Palaniappan,C., Wu,W., Fay,P.J. and Bambara,R.A. (1997) Evidence for a unique mechanism of strand transfer from the transactivation response region of HIV-1. J. Biol. Chem., 272, 16769–16777.(Mogens Duch1, Maria L. Carrasco1, Thomas)