The influence of CpG and UpA dinucleotide frequencies on RNA virus replication and characterization of the innate cellular pathways underlying virus attenuation and enhanced replication

Abstract

Most RNA viruses infecting mammals and other vertebrates show profound suppression of CpG and UpA dinucleotide frequencies. To investigate this functionally, mutants of the picornavirus, echovirus 7 (E7), were constructed with altered CpG and UpA compositions in two 1.1-1.3 Kbase regions. Those with increased frequencies of CpG and UpA showed impaired replication kinetics and higher RNA/infectivity ratios compared with wild-type virus. Remarkably, mutants with CpGs and UpAs removed showed enhanced replication, larger plaques and rapidly outcompeted wild-type virus on co-infections. Luciferase-expressing E7 sub-genomic replicons with CpGs and UpAs removed from the reporter gene showed 100-fold greater luminescence. E7 and mutants were equivalently sensitive to exogenously added interferon-β, showed no evidence for differential recognition by ADAR1 or pattern recognition receptors RIG-I, MDA5 or PKR. However, kinase inhibitors roscovitine and C16 partially or entirely reversed the attenuated phenotype of high CpG and UpA mutants, potentially through inhibition of currently uncharacterized pattern recognition receptors that respond to RNA composition. Generating viruses with enhanced replication kinetics has applications in vaccine production and reporter gene construction. More fundamentally, the findings introduce a new evolutionary paradigm where dinucleotide composition of viral genomes is subjected to selection pressures independently of coding capacity and profoundly influences host-pathogen interactions.

Figure 1.

Genome organization of E7 and positions of mutated insert regions. Insert positions are compared with genome diagram and a plot of sequence variability within species B enteroviruses at synonymous sites (blue line) and folding energies indicative of RNA secondary structure (red and pink lines). Variability at synonymous sites (left y-axis) was computed at each codon position in alignments, plotted with a window size of 41 codons. MFED values (right y-axis) for sense and antisense RNA sequences were calculated for 200 base fragments, incrementing by 48 bases; values plotted represent mean values of five consecutive fragments. Nucleotide positions were calculated relative to the pT7:E7 cDNA sequence.

Figure 2.

RNA to infectivity ratios of WT and viruses with modified CpG/UpA frequencies. WT and mutant viruses were recovered from RD cells and TCID₅₀ titres determined (note log scale). The number of viral genome copies was determined through qRT-PCR and compared with the infectivity titre. Results are the mean and standard error from three separate extractions. RNA/infectivity ratios were additionally calculated for C|C and U|U mutants in the presence of C16.

Figure 3.

Replication kinetics and titres at 24 h of WT and modified viruses infected at a low MOI. RD cells were infected with E7 WT, permuted (P|P), CpG- and UpA-high mutants (C|C and U|U; A and C) or CpG- and UpA-low mutants (c|c, u|u, cu|cu; B and D) viruses at an MOI of 0.01. Infectious titre of supernatants was quantified at indicated time points by TCID₅₀ (A and B) and mean titres and SEMs at 24 h. Results are the mean of three biological replicates.

Figure 4.

Plaque morphology of E7 WT and double region mutant viruses. RD cell monolayers in 10-cm plates were infected with a similar infectious titre of virus and incubated for 96 h at 37°C. (A) Plaque appearance. (B) Plaques sizes of WT and mutant E7 viruses calculated from 25 plaques for each virus using ImageJ software (mean values and SEMs relative to WT control shown as bar heights and error bars).

Figure 5.

Synchronized infection with equal viral genome copies. RD cells were synchronously infected with 1000 genome copies of WT, R1/R2 CpG-high or R1/R2 UpA-high virus, as calculated using qRT-PCR. Cells were trypsinized and washed 1 or 4 h post infection and the intracellular viral load determined by qRT-PCR. Results are the mean and standard error of three biological replicates.

Figure 6.

Analysis of luciferase expression driven by E7 replicons with reduced CpG/UpA frequencies. Replicons were generated with reduced CpG/UpA frequencies, based on the backbone pRiboE7luc replicon, in which the structural genes of E7 are replaced by an insect luciferase gene. In the cu|-|W replicon the luciferase gene itself was modified to minimize both CpG and UpA frequency; in the cu|-|c and cu|-|u replicons Region 2 was additionally modified to further reduce either CpG or UpA frequency. RNA was generated from replicons and transfected into RD cells. Luminescence was measured relative to the mock-transfected control. Results are the mean and standard error of three biological replicates.

Figure 7.

Fitness determination by competition assays between WT and modified viruses. RD cells were infected with an equal MOI of WT and modified virus, and the supernatant serially passaged through cells. RNA was isolated and the composition of each virus determined through selective restriction digests. Images show the virus composition in the starting inoculum and in three biological replicates following passage of CpG- and UpA-low mutants and of U|W.

Figure 8.

Pairwise fitness comparison between E7 WT and mutant with varying degrees of CpG and/or UpA under-representation. RD cells were infected with an equal MOI of two viruses represented in columns and rows of the matrix and the supernatant serially passaged. The composition of each virus was determined through restriction endonuclease cleavage (see Figure 7) and outcome displayed by colour shading. The key refers to population representations of viruses listed in columns (for example, all but one variants were out-competed by the cu|cu mutant). The fitness ranking deduced from these results is shown underneath the figure.

Figure 9.

Effect of exogenous IFN-β on viral replication. RD cells were treated with 5, 50 and 500 U/ml human IFN-β, or mock treated, then infected with WT E7, C|C, U|U and cu|cu mutants at an MOI of 1. RNA was isolated after 8 h and viral load determined by qRT-PCR. Results are the mean and standard error of two biological replicates.

Figure 10.

Analysis of gene expression in cells infected with WT E7 and mutants. RD cells were infected with E7 at an MOI of 10 and changes in expression at 8 h of a range of innate and adaptive immune response genes determined using a PCR Array. Responses were compared with those induced in cells infected by high CpG- and UpA mutants at equal copy number to WT or equal MOI. Parallel infections with Sendai virus (SV) and transfection of poly-I:C were performed. Non-specific induction of genes was controlled for by parallel assays of cells exposed to u/v inactivated supernatant corresponding in amount to that added in the equal copy number (C1) and equal MOI (C2) assays.

Figure 11.

Replication of WT and dinucleotide-modified viruses in an IRF3-blocked cell line. A549 cells expressing the viral protein N^pro and non-expressing control cells were infected with WT, C|C (R1/R2 CpG-high) and U|U (R1/R2 UpA-high) at an MOI of 0.01. RNA was extracted from cell culture supernatant at 24 h and viral load determined by qRT-PCR. Results are the mean and standard error of three biological replicates.

Figure 12.

Effect of inhibiting PKR, RIG-I and MDA-5 expression on E7 replication. (A) Comparison of virus replication in control A549 cells at 8 h compared with that in cells expressing shRNAs targeting PKR, RIG-I and MDA-5 (reduction in mRNA levels achieved is shown in Supplementary Figure S3). (B) RD cells were pre-treated for 24 h with an siRNA directed against PKR and infected with WT virus or CpG-high, UpA-high or CPg/UpA-low mutants at an MOI of 0.03. RNA was isolated after 22 h and intracellular viral loads determined by qRT-PCR. Results are shown relative to the replication rate in cells treated with a control (validated non-target) siRNA.

Figure 13.

Effect of the kinase inhibitors, C16, 2-AP and Roscovitine on E7 replication. Cells were pre-treated for 45 min with 2 μM C16, 5 mM 2-AP or 40 μM Roscovitine and infected with WT virus or CpG-high, UpA-high or CpG/UpA-low mutants at an MOI of 0.1. RNA was isolated at 24 h, and intracellular viral loads were determined by qRT-PCR. Results are shown relative to the replication rate in cells untreated cells (note log scale).

Figure 14.

Influence of the PKR inhibitor E7 cytopathology and infectivity. (A) Cytopathology of WT, C|C and U|U mutants in RD cells infected at MOIs of 0.1 and 0.01 in 2 μM C16 or untreated control. Monolayers of untreated cells remain intact on infection with C|C and U|U mutants while those treated with C16 display a complete CPE even at low MOI. (B) Infectivity determination of pre-titred virus stock supernatant (3000 PFU/ml) of WT virus or CpG-high, UpA-high or CPg/UpA-low mutants in the presence of C16. The enhanced infectivity of C|C and U|U mutants restored the RNA/infectivity ratios to those of WT virus (Figure 2).