Efficient Translation of Pelargonium line pattern virus RNAs Relies on a TED-Like 3´-Translational Enhancer that Communicates with the Corresponding 5´-Region through a Long-Distance RNA-RNA Interaction

Abstract

Cap-independent translational enhancers (CITEs) have been identified at the 3´-terminal regions of distinct plant positive-strand RNA viruses belonging to families Tombusviridae and Luteoviridae. On the bases of their structural and/or functional requirements, at least six classes of CITEs have been defined whose distribution does not correlate with taxonomy. The so-called TED class has been relatively under-studied and its functionality only confirmed in the case of Satellite tobacco necrosis virus, a parasitic subviral agent. The 3´-untranslated region of the monopartite genome of Pelargonium line pattern virus (PLPV), the recommended type member of a tentative new genus (Pelarspovirus) in the family Tombusviridae, was predicted to contain a TED-like CITE. Similar CITEs can be anticipated in some other related viruses though none has been experimentally verified. Here, in the first place, we have performed a reassessment of the structure of the putative PLPV-TED through in silico predictions and in vitro SHAPE analysis with the full-length PLPV genome, which has indicated that the presumed TED element is larger than previously proposed. The extended conformation of the TED is strongly supported by the pattern of natural sequence variation, thus providing comparative structural evidence in support of the structural data obtained by in silico and in vitro approaches. Next, we have obtained experimental evidence demonstrating the in vivo activity of the PLPV-TED in the genomic (g) RNA, and also in the subgenomic (sg) RNA that the virus produces to express 3´-proximal genes. Besides other structural features, the results have highlighted the key role of long-distance kissing-loop interactions between the 3´-CITE and 5´-proximal hairpins for gRNA and sgRNA translation. Bioassays of CITE mutants have confirmed the importance of the identified 5´-3´ RNA communication for viral infectivity and, moreover, have underlined the strong evolutionary constraints that may operate on genome stretches with both regulatory and coding functions.

Fig 1. Schematic representation of the PLPV genome.

PLPV gRNA contains five ORFs (represented by grey boxes) flanked by 5´- and 3´- UTRs (striped boxes). The viral RNA dependent-RNA polymerase, p87, is the ribosomal read-through (RT) product of p27, an auxiliary replication protein. PLPV sgRNA is tri-cistronic and serves as mRNA for expression of two small movement proteins, p7 and p9.7, and of protein p37, which functions as coat protein and as the suppressor of RNA silencing.

Fig 2. Examination of the structure of the 3´-terminal region of PLPV through in silico, in vitro and comparative structural analyses.

(A) Secondary structure of the 3´-terminal region of PLPV (bottom). The represented structure embraces the 3´-terminal sequence of the p37 gene and the entire 3´ UTR (arrow), contains several hairpins (HP1 to HP4) and corresponds to the optimal one according to Mfold predictions using the entire PLPV gRNA. Nucleotides likely involved in an interaction that presumably acts as repressor of minus-strand synthesis are within grey circles. Underlined nucleotides are predicted to base-pair with a segment of the gRNA out of the 3´-terminal region. HP4, that might be a CITE of the TED class, shows a SHAPE-derived flexibility profile (autoradiographs at the top) consistent with the Mfold predictions. In autoradiographs of SHAPE analysis, the lanes have been labelled as: N, for the NMIA treated RNA, D, for the DMSO treated RNA, G, C and A, for sequencing ladders. Residues in the putative TED and flanking region with high and medium reactivity to NMIA are denoted with red and green colors, respectively, whereas those with low or no reactivity to the reagent are in black. Bulges (B) and loops (L) of HP4 have been numbered on the secondary structure representation to facilitate comparison with SHAPE profile. The apical loop and a lateral bulge of the upper part of HP4 have been labelled as AL and BG, respectively. The region of HP4 previously proposed as putative TED-like CITE (15) is indicated by a square bracket with discontinuous line at the left side of the structure. (B) SHAPE profiles across a portion of the 3´-terminal region of the PLPV wild-type (wt) gRNA and a mutated (mut) gRNA carrying nucleotide replacements (in italics) in the apical loop of HP4. Strong NMIA reactivity of the apical loop in the mutant is observed when compared with the wt molecule. This result is in agreement with the involvement of the loop in RNA-RNA interaction as previously proposed [15]. The lanes have been labelled as in panel A and the sequencing ladders have been performed on the wt template. (C) Distribution of natural sequence heterogeneity along HP4. Twenty natural PLPV variants [43] were included in the analysis. Nucleotide replacements are in blue and the number of variants with a given mutation is shown in parentheses. Co-variations and mutations resulting in conversion of canonical Watson-Crick to wobble base pairs, or vice versa, in stems are boxed. Only two minority mutations affecting residue U3723 (marked by an asterisk), could lead to slight destabilization of a stem.

Fig 3. Proposed structures of STNV-TED and of TED-like CITEs predicted in several viruses.

Virus acronyms: Satellite tobacco necrosis virus (STNV), Calibrachoa mottle virus (CbMV), Pelargonium chlorotic ring pattern virus (PCRPV), Elderberry latent virus (ELV), Pelargonium ring spot virus (PelRSV), Rosa rugosa leaf distortion virus (RrLDV) and Pelargonium line pattern virus (PLPV). CbMV is a member of genus Carmovirus and PCRPV, ELV, PelRSV, RrLDV and PLPV are recommended to be included in the tentative new genus Pelarspovirus, all within family Tombusviridae. The region corresponding to the previously proposed STNV-TED as well as those corresponding to the previously proposed TED-like CITEs of CbMV, PCRPV and PLPV [15] are shown on a grey background.

Fig 4. In vitro translation assay in WGE of wt and mutant genomic PLPV transcripts.

A scheme of PLPV gRNA constructs carrying deletions (depicted as dashed lines) in the 3´-UTR is shown. Segments at the 3´-region forming potential hairpins (HP1-HP4) are indicated by small white rectangles on the wt construct. A representative autoradiograph showing p27 levels produced in vitro from each construct (top) and a graphic representation of such levels (bottom) are shown within an inset at the left. For the graphic representation, protein expression levels from wt gRNA were set to 100% and the translation efficiencies from other templates are indicated as percentages with respect the wt. Each percentage is shown as the mean and standard error of the mean from three replicates. Other details as in Fig 1.

Fig 5. In vivo translation assay of reporter-based genomic PLPV transcripts.

(A) A parental genomic PLPV construct (gFF at the top) for assessment of translation efficiency was generated by replacing most of the p27/RdRp ORF by the Fluc reporter gene (represented by a white box). This parental construct served as template for the generation of a series of mutant constructs carrying deletions at the 3´-UTR (gFF-D1 to D4). Uncapped transcripts from each construct were inoculated into N. benthamiana protoplasts along with a control capped Rluc transcript and luciferase levels were measured 3 h post-inoculation. Translation efficiencies of constructs are given as percentages with regard the parental gFF (set to 100%) The mean and standard error of the mean from three replicates are indicated. (B) gRNA reporter constructs carrying deletions at the 5´-terminal coding region (gFF-s5R) or the 3´-terminal coding region (gFF-s3R1 to s3R3). Uncapped transcripts derived from these constructs were assayed as indicated in panel A. (C) Stability of parental construct gFF and deletion constructs gFF-s3R1, gFF-s5R and gFF-D2 in protoplasts. Uncapped transcripts were employed to inoculate N. benthamiana protoplasts and total RNA was isolated at 1 h (left) and 3 h (right) after inoculation. A mock inoculated control (-) was also included. RNA was subjected to Northern blot analysis using an RNA probe derived from Fluc gene. Two replicates of the experiment were done (one shown here) with similar results. Ethidium bromide staining of ribosomal RNAs is shown below the autoradiograph as loading control. Other details as in Fig 3.

Fig 6. Effect of mutations in the TED-CITE and/or in a hairpin within p27 ORF on translation of reporter-based genomic PLPV transcripts.

(A) Putative secondary structure of the 5´-proximal 125 nt of PLPV according Mfold predictions. A potential long-range interaction between the apical loop of a hairpin (gHP3) predicted within p27 ORF and the apical loop of the 3´-CITE is shown. Nucleotides that can putatively pair are connected by dotted lines. (B) Translation efficiency of transcripts bearing mutations that disrupted or reconstituted the potential kissing-loop interaction between apical loops of gHP3 and the CITE. All mutants were generated from parental construct gFF and assayed in N. benthamiana protoplasts as described in Fig 5 legend. Only the sequences of the loops involved in the potential base-pairing are shown. The engineered nucleotide substitutions are on a gray background and putative G:U base pairs are in italics. (C) Translation efficiency of transcripts bearing mutations in the CITE outside of its apical loop. In B and C, levels of translation as a percentage of that of gFF construct are given. Results are from three experiments with standard deviations. Other details as in Fig 5.

Fig 7. Effect of the addition of a 5´-cap on translation of reporter-based genomic PLPV transcripts.

Uncapped and capped transcripts from the wt construct gFF and from CITE mutants gFF-D2 and gFF-M2 (depicted in Figs 5 and 6, respectively) were generated and their translation efficiencies estimated through assay in N. benthamiana protoplasts. Translation efficiencies of transcripts are shown as percentages with respect to that of the uncapped gFF (set to 100%) The mean and standard error of the mean from three replicates are represented.

Fig 8. Assessment of the relevance of the 3´-TED and/or a 5´-hairpin, that can potentially establish a kissing-loop interaction, on reporter-based PLPV subgenomic translation.

A) Putative secondary structure of the 5´-proximal 33 nt of PLPV sgRNA according Mfold predictions. The start codon of p7 ORF is indicated by an arrow. A potential long-range interaction between the apical loop of a 5´-hairpin (sgHP1) and the apical loop of the TED-like CITE is shown. Nucleotides that can putatively pair are connected by dotted lines. (B) In vivo translation assay of subgenomic PLPV transcripts. A parental subgenomic construct for assessment of translation efficiency was generated by replacing almost the entire p7 ORF by the Fluc reporter gene. This parental construct, named sgFF, is shown at the top and mutants derived from it are depicted below. One of the mutants had a shorter 5´-region (sgFF-s5R) and the others carried mutations that disrupted (sgFF-M1 and M3) or preserved (sgFF-M1/M3) the potential kissing-loop interaction between the 3´-CITE and the 5´-sgHP1. For the latter mutants, only the sequences of the loops involved in the potential base-pairing are shown. The engineered nucleotide substitutions are on a gray background and putative G:U base pairs in potential kissing-loop interaction(s) are in italics. Translation efficiencies were measured in protoplasts and the mean and standard error of the mean from three replicates is shown. Other details as in Fig 5.

Fig 9. Assessment of infectivity of PLPV mutants.

Uncapped transcripts corresponding to the gRNA of wt and CITE-related mutants of PLPV were inoculated onto N. benthamiana plants. Local leaves were collected at 15 days post-inoculation and viral accumulation was determined by Northern blot analysis. Bioassayed mutants are indicated above the lanes of the autoradiograph. A mock inoculated sample (lane M) was also included. Positions of PLPV gRNA and sgRNA are indicated at the right. Ethidium bromide staining of ribosomal RNAs is shown below the autoradiograph as loading control.

Fig 10. STNV-TED and putative TED-like CITEs predicted in the 3´ UTR of several viruses (middle panel) can potentially base-pair with the 5´-region in the gRNA (upper panel) or sgRNA (lower panel) context.

Virus acronyms as is Fig 3.