Characterization of a natural heterodimer between MLV genomic RNA and the SD' retroelement generated by alternative splicing

Abstract

Murine leukemia virus (MLV) specifically packages both genomic RNA (FL RNA) and a subgenomic RNA, which we call SD'. SD' RNA results from alternative splicing of FL RNA. It is reverse-transcribed, and its DNA copy, integrated into the host genome, constitutes a splice donor-associated retroelement. FL and SD' RNAs share a common 5'-UTR that includes the packaging/dimerization signal (Psi). To investigate whether the mechanism of copackaging of these two RNAs involves RNA heterodimerization, we examined the spontaneous dimerization capacity of the two RNAs as large synthetic RNAs transcribed in vitro. We showed that SD' RNA not only formed homodimers with similar efficiency as the FL RNA, but that FL and SD' RNAs also formed FL/SD' heterodimers via Psi sequences. Comparison of the thermostabilities determined for these different dimeric species and competition experiments with Psi RNA fragments indicate the recruitment of similar dimer-linkage interactions within the Psi region. To validate these results, the dimeric state of the SD' RNA was analyzed in MLV particles. RNA capture assays performed with the FL RNA as bait revealed that SD', and not the host packageable U6 or 7SL RNAs, was associated with the FL RNA in virions. Heterodimerization of SD' RNA with FL RNA may argue for the recent concept of a nuclear dimerization at or near the site of transcription and raises the new hypothesis of RNA dimerization during splicing. Furthermore, FL/SD' heterodimerization may have leukemogenic consequences by influencing the pool of genomic dimers that will undergo recombinogenic template switching by reverse transcriptase.

FIGURE 1.

Schematic representation of the MLV RNAs used in this study. Native FL and SD′ RNAs (in boldface) analyzed in virions and synthetic transcripts used in vitro are depicted as large and small boxes, respectively. RNA sizes and features positions are in relation to the numbering from the cap site (+1) of the MLV RNA genome. Non-coding region elements in 5′ and 3′, R U5 and U3 R, respectively, are boxed such as the gag, pol, and env coding regions. The Psi encapsidation signal encompassing stem–loops A, B, C, and D is depicted as a triangle and its deletion by a cross. Donor and acceptor splicing sites, SD′ and SA, are indicated, as well as the SD′/SA junction.

FIGURE 2.

Conditions of spontaneous dimerization in vitro of FL (7.7 kb), SD′ (3.8 kb), and 5′-Psi (878 nt) RNAs. Comparison between the ability to dimerize with (heat+D1) or without (D1) unfolding step. All experiments were performed with 1 pmol of RNA subjected to dimerization conditions as described in Materials and Methods, analyzed by native agarose gel electrophoresis, and visualized by ethidium bromide staining.

FIGURE 3.

Influence of the Psi sequences on dimerization. (A) Dimerization ability of 5′-Psi (878 nt) and 5′-ΔPsi (640 nt) RNAs under standard D1 conditions. (B) FL RNA specifically dimerizes through the Psi sequences. 0.5 pmol of FL RNA (7.7 kb) was incubated alone (lanes 1,8) or with increasing amounts (0.5–5 pmol) of 5′-Psi RNA (lanes 2–7) or 5′-ΔPsi RNA (lanes 9–14) under D1 conditions as described in Materials and Methods. Autoradiographies obtained from similar experiments conducted with radiolabeled wt or mutant Psi RNAs are given (lanes 7 and 14, respectively). (C) SD′ RNA specifically dimerizes through the Psi sequence. The same experiments as in B were achieved with SD′ RNA (3.8 kb). Heterodimerizations were performed with either the 5′-Psi RNA (lanes 1–6) or 5′-ΔPsi RNA (lanes 7–12).

FIGURE 4.

Analysis of heterodimerization between FL and SD′ RNAs. (A) FL and SD′ RNAs were cotranscribed by mixing the SD′ template cut either with BamHI, BstEII, or ClaI to, respectively, yield 1.8-kb, 2.6-kb, and 3.8-kb SD′ RNA (lanes 3,6,9, respectively) with the FL plasmid template. The cotranscribed RNA mixes were submitted to D1 conditions. Respective FL/SD′ heterodimers are indicated by a star and FL monomer and homodimer by open and black arrowheads, respectively. In parallel, each SD′ RNA was separately submitted to D1 dimerization conditions to form (open circles) SD′/SD′ homodimers (lanes 2,5,8). (Black circles) The different SD′ monomers were obtained by denaturation for 10 min at 85°C (lanes 1,4,7). (B) FL and SD′ RNAs form heterodimers through the Psi sequences. The same amount of FL/SD′ heterodimer (indicated by a star), previously formed by mixing SD′ and FL DNA templates, was incubated with increasing amounts of 5′-Psi (lanes 2–5) or 5′-ΔPsi (lanes 6–9) RNAs for competition experiments.

FIGURE 5.

Thermal stability of the dimers. A representative experiment is shown with the corresponding T _m plots. Temperature-induced dissociation of RNA dimers were done as described in Materials and Methods with an RNA concentration of 0.1 μM. The percentage of dimer was estimated for each RNA species at different temperatures, and curves were fitted to calculate the melting temperature, T _m. Results of at least three independent experiments gave the following T _m values: 61.3° ± 1.5°C for 5′/5′; 64.5° ± 0.5°C for FL/FL; 62.0° ± 0.4°C for SD′/SD′, and 64.3° ± 0.8°C for FL/SD′.

FIGURE 6.

Ex vivo analysis of natural RNAs. (A) RNA levels in cell and virion samples of a representative experiment. Host and viral RNAs were quantitated by RT-QPCR in both (C) transfected cells and (V) virus. Values correspond to copy numbers measured in 50 ng of cellular RNA samples and in 1/1600 of total released virions. (B) Relative efficiencies of RNA encapsidation in MLV particles. Encapsidation efficiencies were determined [(V/C) × 100] and normalized to the FL level. Results represent the mean±standard deviations of at least three independent experiments.

FIGURE 7.

RNA capture assays. Abundances of SD′ and FL RNAs were quantified by RT-QPCR, and RNA copy numbers (cps) in input were given in boxes. To monitor the specificity of the capture, repartitions of FL and SD′ RNAs were determined in the different fractions collected during the RCA procedure (see Materials and Methods). (A) Specificity of the biotinylated oligonucleotide allowing the FL RNA capture. Control RNA capture experiments were conducted with the synthetic FL (7.7 kb) and SD′ (3.8 kb) RNAs transcribed separately in vitro, mixed, and kept at 4°C prior to the RCA to prevent dimerization. (B) Heterodimerization of SD′ RNA in MLV particles by the RNA capture assay. The RNA copy numbers measured in the input correspond to 30 mL of virus-containing supernatant. Average frequencies of SD′/FL and FL/FL associations were 63.4% ± 18% and 77.0% ± 17%, respectively, and were obtained from triplicate experiments. (C) The SD′ capture depends on specific FL RNA association in virions. For this control, an additional step was added before the standard RCA procedure used in B. RNA samples were heated for 10 min at 70°C and chilled on ice prior to annealing to the biotinylated oligonucleotide. (D) Heterodimerization of the host RNAs. The 7SL and U6 RNAs were quantified in the same input (number of copies in boxes) and RCA fractions as in C for viral RNA analysis. Triplicate experiments gave an average proportion of unretained RNA of 97.5% ± 0.9% and 98% ± 0.7% for 7SL and U6 RNAs, respectively.