Evidence for ribosomal frameshifting and a novel overlapping gene in the genomes of insect-specific flaviviruses

Abstract

Flaviviruses have a positive-sense, single-stranded RNA genome of approximately 11 kb, encoding a large polyprotein that is cleaved to produce approximately 10 mature proteins. Cell fusing agent virus, Kamiti River virus, Culex flavivirus and several recently discovered flaviviruses have no known vertebrate host and apparently infect only insects. We present compelling bioinformatic evidence for a 253-295 codon overlapping gene (designated fifo) conserved throughout these insect-specific flaviviruses and immunofluorescent detection of its product. Fifo overlaps the NS2A/NS2B coding sequence in the -1/+2 reading frame and is most likely expressed as a trans-frame fusion protein via ribosomal frameshifting at a conserved GGAUUUY slippery heptanucleotide with 3'-adjacent RNA secondary structure (which stimulates efficient frameshifting in vitro). The discovery bears striking parallels to the recently discovered ribosomal frameshifting site in the NS2A coding sequence of the Japanese encephalitis serogroup of flaviviruses and suggests that programmed ribosomal frameshifting may be more widespread in flaviviruses than currently realized.

Fig. 1

Simple phylogenetic trees based on (A) NS2A+NS2B and (B) FIFO. The amino acid neighbor-joining tree was produced with CLUSTALX (Larkin et al., 2007). Columns with alignment gaps were excluded. Numbers indicate bootstrap support (out of 1000), while the scale bar represents the number of substitutions per site.

Fig. 2

Conservation at synonymous sites in CxFV and QBV. Conservation was calculated for an input alignment comprising the polyprotein coding sequences from the CxFV sequences AB262759, AB377213, FJ663034, FJ502995, GQ165808 and EU879060, and the QBV sequence FJ644291. Panel 1 shows a map of the CxFV genome, along with the proposed overlapping gene fifo and the proposed frameshift site. Panels 2–4 show the positions of stop codons in the three forward reading frames. The + 0 frame is the polyprotein frame and is therefore devoid of stop codons. Note the conserved absence of stop codons in the − 1/+ 2 frame within the fifo ORF. Panels 5–8 depict the conservation at polyprotein-frame synonymous sites. Panels 5 and 7 depict the probability that the degree of conservation within a given window could be obtained under a null model of neutral evolution at synonymous sites, while panels 6 and 8 depict the absolute amount of conservation as represented by the ratio of the observed number of substitutions within a given window to the number expected under the null model (see Firth and Atkins, 2009, for details). Note that the large sliding window size (75 codons) is responsible for the broad smoothing of the conservation scores at either end of the fifo ORF. The range of features detected by this type of analysis depends partly on the sliding window size − 75 codons is optimal for features with a width of 75 codons. If features cover fewer codons (e.g., many non-coding functional elements), then the corresponding conservation peaks may be diluted to such an extent that they are not distinguishable from background noise.

Fig. 3

Potential stimulatory RNA secondary structures for − 1 ribosomal frameshifting in insect-specific flavivirus sequences. (A) Structure predictions for representative CxFV and CFAV sequences. (B) Sequence alignments for the 14 available insect-specific flavivirus sequences, showing the conserved presence of a slippery heptanucleotide (G GAU UUY, or G UUU UUU in NAKV; orange) and potential to form a 3′-adjacent RNA hairpin (Clade 1) or pseudoknot (Clade 2). Base pairings in stem 1 are indicated with ‘()’s or ‘{}’s and yellow or tan background, while base pairings in stem 2 of the pseudoknot are indicated with ‘[]’s and green background. Paired base substitutions that preserve base pairings are highlighted in pink, while single base substitutions that replace canonical Watson-Crick base pairings with G:U base pairings are highlighted in pale blue. A potential extension to stem 1 in CFAV is indicated by underscores, although these base pairings are expected to be disrupted once the ribosome is positioned on the slippery site.

Fig. 4

Amino acid alignment for CxFV, QBV and NAKV FIFO. Note that amino acids ‘DF’ at positions 1–2 come from the zero-frame at the shift site. FIFO is predicted to be expressed as part of a trans-frame fusion protein with the N-terminal region of NS2A (not shown). Predicted transmembrane regions are underlined.

Fig. 5

Amino acid alignment for CFAV, KRV, AEFV and CSA FIFO. Note that amino acids ‘DF’ at positions 1–2 come from the zero-frame at the shift site. FIFO is predicted to be expressed as part of a trans-frame fusion protein with the N-terminal region of NS2A (not shown). Predicted transmembrane regions are underlined. The three premature termination codons in the fifo-defective CFAV-1993 sequence are indicated with ‘⁎’s on black rectangles.

Fig. 6

Conservation at synonymous sites in CFAV, KRV, AEFV and CSA. See Fig. 2 for details. Here, the alignment comprises the polyprotein coding sequences from AY149905 (KRV), AY149904 (KRV), AB488408 (AEFV), AF411835 (CSA; chromosome-integrated sequence), M91671 (presumed fifo-defective CFAV-1993) and GQ165810 (CFAV-RioPiedras02). AY149905 and AY149904 are nearly identical. Only AY149905, AB488408, AF411835 and GQ165810 were used to calculate the synonymous site conservation statistics (panels 5–6). There is considerable variation in the 3′ UTR lengths between the different sequences (e.g., 559, 1208 and 945 nt, respectively, in CFAV M91671, KRV AY149905 and AEFV AB488408; see Gritsun and Gould, 2006, for a discussion) — hence the wiggly line in the genome diagram. The CFAV-1993 sequence (isolated from a laboratory mosquito cell line) has lost the fifo ORF, indicating that the product is non-essential in cell culture.

Fig. 7

Detection of FIFO by immunofluorescence in CxFV-infected mosquito cells. C6/36 cells were infected with CxFV-Mex07 (first row) or CxFV-Iowa07 (second row) or mock infected (third row). At 4 days p.i., cells were fixed with methanol and immunostained with rabbit polyclonal antibodies against FIFO (FIFO Ab 1 and FIFO Ab 2; panels A and B, respectively), followed by Alexa 594-conjugated donkey anti-rabbit IgG. Scale bars = 10 μm.

Fig. 8

SDS PAGEs for frameshifting assays. (A) Translation of products from pF25A constructs in the insect cell-free system. The frameshifting cassette was fused between upstream and downstream ORFs such that translation of the downstream ORF – as a fusion with the upstream ORF – occurs only if frameshifting occurs. The predicted sizes of the termination product (upstream ORF, no frameshifting) and frameshift product (fusion of both ORFs) are marked. Frameshifting efficiencies were calculated from the intensities of labelled bands after normalization for the number of methionine residues in each product. WT sequence (especially CFAV) stimulates high levels of frameshifting, while frameshifting is reduced to background levels in the frameshift (fs) site knockout mutants. Unfortunately background bands (first lane) at the same positions as both the termination and frameshift products makes precise measurement difficult for the shift site mutants, although it is clear that frameshifting is greatly diminished in the mutants. (B) Comparison of frameshifting efficiencies between the insect system (I) and rabbit reticulocyte lysate (R). Both vectors pF25A and pDluc were tested in rabbit reticulocyte lysate but only the insect-optimized pF25A vector was usable in the insect system.

Fig. 9

Potential slippery heptanucleotides and stimulatory RNA secondary structures for − 1 ribosomal frameshifting in Nounane virus, Kedougou virus, Chaoyang virus and Lammi virus. In contrast to the insect-specific flavivirus fifo frameshift site, these cases are merely candidate frameshift sites and currently have either no (Nounane, Kedougou) or only very modest (Chaoyang, Lammi) phylogenetic support. Structures were predicted with pknotsRG (Reeder et al., 2007).

Fig. 10

SDS PAGE of translation products from pDluc constructs in rabbit reticulocyte lysate. See Fig. 8 for further details. Lane 1 shows translation products for an in-frame control (IFC) – in which an extra nucleotide has been inserted next to the slippery heptanucleotide to put both ORFs into the same reading frame – to show the expected size of the frameshift fusion product. Note that the faint band corresponding to the frameshift product for the CFAV and CxFV frameshift site (fs) mutants may represent a background of low-level slippage occurring anywhere within the overlap region between the upstream and downstream ORFs and does not necessarily represent slippage just at the mutated shift site.