Understanding off-target growth defects introduced to influenza A virus by synonymous recoding

Abstract

CpG dinucleotides are underrepresented in the genomes of most RNA viruses. Synonymously increasing CpG content of a range of RNA virus genomes reliably causes replication defects due to the recognition of CpG motifs in RNA by cellular zinc-finger antiviral protein (ZAP). Prior to the discovery of ZAP as a CpG sensor, we described an engineered influenza A virus (IAV) enriched for CpGs in segment 5 that displays the expected replication defects. However, we report here that this CpG-high ("CpGH") mutant is not attenuated by ZAP. Instead, a pair of compensatory nucleotide changes, resulting in a stretch of eight consecutive adenosines (8A), were found to be responsible. Viral polymerase slippage occurs at this site, resulting in the production of aberrant peptides and type I interferon induction. When the nucleotides in either one of these two positions were restored to wild-type sequence, no viral attenuation was seen, despite the 86 extra CpGs encoded by this virus. Introduction of these two adenosines into wild-type virus (thereby introducing the 8A tract) resulted in viral attenuation, polymerase slippage, aberrant peptide production and type I interferon induction. That a single nucleotide change can offset the growth defects in a virus designed to have a formidable barrier to wild-type reversion highlights the importance of understanding the processes underlying viral attenuation. Poly(A) tracts are a correlate for the emergence of polybasic cleavage sites in avian IAV hemagglutinins to produce highly pathogenic strains. These results thereby uncover possible insights into the intermediary events of this important evolutionary process.

Keywords: RNA polymerase; frameshift; influenza virus; synonymous recoding.

FIGURE 1.

CpG enrichment in segment 5 of the IAV genome results in ZAP-independent attenuation. WT PR8, CDLR control, and CpGH viruses were used to infect permissive cells and counterpart ZAP-pathway knockout cells at MOI 0.01 for 48 h, and infectious virus production was measured. (A) A549-cas9 or paired ZAP−/− cell infections. (B) HEK293-cas9 or paired TRIM25−/− cell infections. (C) A549-cas9 or paired KHNYN−/− cell infections.

FIGURE 2.

CpG enrichment in segment 5 of the IAV genome does not impair viral packaging. (A) RNA was extracted from 10⁷ PFU virus stocks for WT PR8, CDLR control, and CpGH viruses, and relative copy numbers of segment 1 and segment 5 RNA were quantified by qPCR. A “4c6c” packaging mutant was used as a negative control. (B) 10¹⁰ PFU of egg-derived virus panel virions were purified, then genomic RNA was extracted, and individual segments were separated using urea-PAGE and visualized by silver staining. Numbers indicate location of each genome segment. Mock is RNA from the allantoic fluid of uninfected eggs, prepared using the same methods.

FIGURE 3.

CpG enrichment in segment 5 of the IAV genome reduces NP transcript and protein abundance in cells. (A) To assay transcription efficiency, in vitro transcription assays using T7 RNA polymerase were performed using the segment 5 plasmids, with RNA quantified by Qubit at 10, 20, 45, and 90 min. (B) To assay combined transcription and translation in a cell-free system, limiting dilutions of segment 5 plasmids were used in rabbit reticulocyte lysate coupled transcription and translation assays, and NP protein was detected by western blotting using an NP-specific antibody. (C, D) A549 cells were infected at MOI 10 for 10 h, then RNA abundance was examined by northern blotting (C) and protein abundance by western blotting for the indicated species (D).

FIGURE 4.

Tiled reversion of the segment 5 CpGH sequence to WT identifies a short region that is common to viruses with WT virus fitness. (A) To identify the recoded region of the CpGH virus contributing the attenuated phenotype, overlapping fragments A–E were reverted to WT PR8 sequence as indicated. (B) Fragmented reversion viruses (e.g., CpGH virus with fragment A reverted to WT PR8 sequence, denoted as CpGH/A.PR8) were used to infect A549-cas9 cells at MOI 0.01 for 48 h, and virus production was measured.

FIGURE 5.

Serial passage of segment 5 CpGH IAV identifies reversion mutations that fall within an 8 nucleotide stretch of adenosines. Virus panel generated from either egg (A) or MDCK (B) rescue was serially passaged 10 times at MOI 0.01 in A549 cells, with virus titered after each passage, for four biological repeats (two with starting inoculum of egg rescue and two of MDCK). At passage 10, virus was deep sequenced. (C) Read depth from deep sequencing of CpGH virus. (D) Mutations occurring exclusively in the CpGH virus were plotted. None of these mutations occurred at CpG sites. Their position relative to fragments A–E (Fig. 4) are indicated by the black bars at the head of the panel. (E) Serially passaged viruses derived from egg rescue were sequenced after 1, 2, and 3 passages; for the CpGH virus, mutation frequency at position 312 is shown. (F) Reversion mutations at positions 312 and 315 corresponded with an 8-adenosine tract introduced exclusively into the CpGH virus. (G) Variability of segment 5 nucleotide positions in nature (10,000 human sequence isolates analyzed over 5 years between 2018 and 2022).

FIGURE 6.

Single nucleotide reversions at positions 312 and 315 of the CpGH virus restored WT fitness. The CpGH virus (S5CpGH) incorporated an 8A tract in the coding sequence. This was removed by reversion to WT sequence at these nucleotide positions through mutations A312U and/or A315G. Conversely, the poly(A) stretch was introduced into WT PR8 via U312A and/or G315A mutations. (A) Titer of virus panel grown at MOI 0.01 for 48 h in A549 cells. (B) To determine whether ZAP sensing was apparent for any mutants, the virus panel was grown in A549-cas9 or ZAP−/− cells, and titers were normalized to WT PR8 titers. (C) To determine whether input RNA from incoming virions was degraded, A549-cas9 cells were infected with the virus panel at MOI 10 in the presence of cycloheximide; RNA was harvested at 24 h postinfection, electrophoresed, and northern blotted using probes for IAV segments 1 and 5 (negative orientation). Ribosomal RNA (rRNA) served as a loading control.

FIGURE 7.

Introduction of the 8A tract into IAV segment 5 resulted in viral polymerase slippage, aberrant protein production, and triggered type I IFN. (A) WT PR8, CpGH, and cognate viruses with switched nucleotides at positions 312 and 315 were used to infect A549 cells at MOI 3 for 8 h, after which RNA was extracted, amplified, and sequenced. Chromatogram traces were examined for evidence of multiple RNA species generated downstream from the 8A sequence of CpGH virus, encoded at nucleotide positions 312–319. Faded black arrows indicate direction of sequencing read. Solid black arrows indicate sites of polymerase slippage. (Top left) CpGH plasmid was transfected into HEK293T cells, and +RNA was produced from a pol.II promoter. (Top middle) WT PR8 virus infection. (Top right) CpGH virus infection. (Middle left) CpGH A312U virus (7A tract) infection. (Middle middle) CpGH A315G virus (4A tract) infection. (Middle right) CpGH A312U A 315G (4A tract) virus infection. (Bottom left) PR8 U312A virus (4A tract) infection. (Bottom middle) PR8 virus G315A (7A tract) infection. (Bottom right) PR8 U312A G315A virus (8A tract) infection. (B) CpGH plasmids and infections were deep sequenced, and percentage of sequence reads with changes in the length of the 8A tract were calculated. (C) Top panel—Nucleotide alignment of PR8, PR8 G312A U315A, and CpGH sequence surrounding the poly(A) site, with alignments to show the nucleotide sequence resulting from +1, +2, −1, and −2 polymerase slippage events. The bottom panel shows the resulting peptide species arising from these transcripts. (D) To examine for the presence of frameshifted peptide production, A549 cells were infected at MOI 3 for 8 h with either WT PR8, CpGH, or PR8 A312U A315G viruses, and NP peptide production was assessed using mass spectrometry to examine for a shift into alternative reading frames. Due to differences in m/z ratios for the peptides unique to +1 and +2 frameshifted translations, relative abundance cannot be compared across peptide species. The −1 frameshift followed by gluC digestion resulted in predicted peptides that were not of sufficient length for detection by mass spectrometry. No peptides predicted to have arisen from the −2 frameshift were detected. (E) Type I interferon competent A549 cells were infected with virus panel for 10 h at MOI 10, after which time supernatant was harvested, UV treated to inactivate infectious virus and assayed for IFN content using HEK Blue cells. HEK Blue cells were also treated with IFN standard (light gray bars). Means of three biological repeats are shown.

FIGURE 8.

IAV polymerase slippage preferentially occurs on 8U tract over 8A, offering a putative mechanism for the generation of polybasic cleavage sites. (A) Two plasmid constructs were designed to encode the EGFP ORF downstream from an 8-adenosine or 8-uracil tract, flanked by PR8 NS1 segment UTRs that are recognized by the viral polymerase and under an RNA pol.I promoter. These constructs were used in a minigenome assay (together with plasmids that together reconstitute viral polymerase) to determine whether polymerase slippage could be evidenced through reporter gene detection. Due to the incorporation of the 8A/8U tract upstream of the GFP ORF, the ORF is out of frame. Polymerase slippage on the 8A/8U resulting in a single nucleotide insertion will put the GFP ORF in frame and yield GFP signal. (B) GFP signal was read for in-frame GFP, GFP downstream from an 8A oligonucleotide, and GFP downstream from an 8U oligonucleotide, under IAV promoter and polymerase expression (annotations refer to sequence in positive polarity). Fluorescence was read every 24 h and normalized to reporter transfected with incomplete IAV polymerase components (no PA). (C) Relative fluorescence of GFPs encoded downstream from 8A and 8U sequences compared with in-frame GFPs for PR8, Cal04, Mallard, Swine, and Udorn polymerases, and CMV-promoter driven host polymerases. (D–I) To determine whether polymerase slippage occurred for other IAV polymerases, the IAV PR8, Cal04, Mallard, Swine, Udorn, and human CMV polymerases were used to amplify the transcripts containing 8A upstream of the GFP sequence, and amplicon sequencing was performed to determine the extent of polymerase slippage on the 8A sequence.

Colin P. Sharp