CpG dinucleotide enrichment in the influenza A virus genome as a live attenuated vaccine development strategy

Abstract

Synonymous recoding of RNA virus genomes is a promising approach for generating attenuated viruses to use as vaccines. Problematically, recoding typically hinders virus growth, but this may be rectified using CpG dinucleotide enrichment. CpGs are recognised by cellular zinc-finger antiviral protein (ZAP), and so in principle, removing ZAP sensing from a virus propagation system will reverse attenuation of a CpG-enriched virus, enabling high titre yield of a vaccine virus. We tested this using a vaccine strain of influenza A virus (IAV) engineered for increased CpG content in genome segment 1. Virus attenuation was mediated by the short isoform of ZAP, correlated with the number of CpGs added, and was enacted via turnover of viral transcripts. The CpG-enriched virus was strongly attenuated in mice, yet conveyed protection from a potentially lethal challenge dose of wildtype virus. Importantly for vaccine development, CpG-enriched viruses were genetically stable during serial passage. Unexpectedly, in both MDCK cells and embryonated hens' eggs that are used to propagate live attenuated influenza vaccines, the ZAP-sensitive virus was fully replication competent. Thus, ZAP-sensitive CpG enriched viruses that are defective in human systems can yield high titre in vaccine propagation systems, providing a realistic, economically viable platform to augment existing live attenuated vaccines.

Fig 1. Influenza A viruses with CpG enrichment in segment 1 have replication defects mediated by ZAP.

WT PR8, synonymously permuted control virus (CDLR) and CpG enriched (CpGH) viruses were used to infect various cell types at low MOI for 48 hours. Synonymous recoding was performed in segment 1. Viruses were grown in A549 and paired ZAP -/- cells (A) or in 293 and paired TRIM25 -/- cells (B). C. siRNA knockdown of ZAP or TRIM25 in NHDF primary cells was confirmed by western blot. D. After knockdown, NHDFs were infected at high MOI for 24 hours. Impact of ZAP or TRIM25 knockdown on replication of the seg1 virus panel in NHDFs was determined. E. Seg1 virus panel was grown in A549 and paired KHNYN -/- cells. F. The number of CpGs introduced into the CpGH construct was depleted by a third (80 CpGs added, giving a segment total of 131) or by two thirds (46 CpGs added, giving a segment total of 97). When 46 CpGs were added, this was done either by retaining the 46 CpGs added to the 3’ end of RNA (positive orientation), or by spacing the 46 CpGs added across the segment. This panel of CpG mutants, along with CDLR and CpGH, were titred in WT A549s and paired ZAP-/- cells as in (A). G. ZAPL/ ZAPS expression was reconstituted in A549 cells using lentivirus expression constructs, and isoform-specific expression was confirmed by western blotting. H. A549 ZAP-/- cells were reconstituted for isoform-specific ZAPL or ZAPS expression, and titres of seg1 virus panel were determined. For validation of KO phenotypes, please refer to S1 Fig. Confocal visualisation of ZAP reconstitution is presented in S2 Fig.

Fig 2. CpG enrichment does not affect packaging but may alter electrophoretic mobility.

A. Virus stocks of the seg1 virus panel were titred by plaque assay, and then the relative copy numbers of segment 1 and segment 5 were determined by qPCR. RNA:PFU ratios were then calculated. A ‘4c6c’ mutant virus [36,46] was used as a packaging mutant control. To visualise packaged viral RNAs, viral stocks were purified by ultracentrifugation and subjected to urea-PAGE, thereby separating the 8 segments of the viral genome (B). A shift in the electrophoretic mobility of segment 1 of the CpGH virus was observed (large black arrow), and in order to determine whether this was due to RNA modification, RNAs generated through in vitro transcription (i.e. a cell-free system, free from RNA modifying enzymes) were run under the same conditions (C). For ‘mix’ samples, equimolar combinations of CDLR and CpGH constructs were combined and run to demonstrate whether a mobility shift was apparent in CpGH constructs.

Fig 3. CpG enrichment of viral RNA does not alter the methylation profile.

Segment 1 modified viruses were processed through bisulphite conversion reactions and then deep sequenced to determine the frequency of methylation in unmodified versus CpG-enriched transcripts. A. As virus reconstitution was performed in embryonated hens’ eggs, chicken ribosomal RNA was sequenced from viral stocks as a positive control due to a known methylation site in this RNA. B. 5-methylcytosine signal intensity across PR8 segment 1 vRNA. C. 5-methylcytosine signal intensity across CDLR segment 1 vRNA. D. 5-methylcytosine signal intensity across CDLR segment 1 vRNA. E. 5-methylcytosine signal intensity across PCR amplicons of either vRNA (PR8 egg/ CDLR egg/ CpGH egg) or in vitro transcribed (IVT) RNA, across segment 1 positions 106–304. This region was selected due to the spike in signal observed across segment 1 vRNAs for the segment 1 virus panel (B-D, red boxes). For PR8 whole genome methylation plots please refer to S4 Fig.

Fig 4. Interaction of CpG-enriched viruses with the IFN pathway.

A. A549 cells were infected with WT PR8, CDLR or CpGH viruses at MOI 10 for 24 hours, and then type I IFN induction was quantified by HEK-Blue assay. Titrated IFN was used as a standard, and an IAV mutated in segment 8 to inhibit IFN induction (R38K41A) was used as a positive control. B. IAV WT PR8, CDLR or CpGH viruses were further mutated in segment 8 to abrogate the ZAP/TRIM25 targeting activity of NS1 encoded on this segment. Mutations introduced into segment 8 were L95S99A, reportedly defective in TRIM25 targeting [55], E96/97A, reportedly defective in ZAP targeting [56], and PKQK107-110A, reportedly defective in ZAP targeting [56]. Seg1/8 mutant viruses were used to infect A549 cells at MOI 0.01, for 48h, and viral titres were quantified. C. Western blots examining ZAP and TRIM25 expression in the same infections. D. The seg1/8 mutant virus panel was used to infect A549 cells at high MOI for 10 hours, and then IFN induction was quantified by HEK-Blue assay (A640 read-out).

Fig 5. CpG enrichment in the IAV genome results in reduced transcriptional levels during virus replication.

Transcription and translation were investigated in the context of infection using high MOI time courses. The virus panel was analysed for segment 1 (A) or segment 5 (B) RNA production. C. For assessment of protein production, PB2 (segment 1) and NP (segment 5) were assayed by western blot. D. Equal amounts of segment 1 WT (PR8), permuted control (CDLR) or CpG-enriched (CpGH) amplicons under T7 promoters were used in in vitro transcription assays to assess whether synonymous recoding impacted transcriptional efficiency in a cell free system. E. The efficiency of translation of RNAs was then assessed by titrating the synthesised RNAs in in vitro translation assays. F. A549 WT or ZAP-/- cells were infected with either WT PR8 or CpGH virus at an MOI of 15 for 8 hours, after which time RNA was harvested and probed by northern blotting using specific primers to detect segment 1 and segment 5 vRNA or positive polarity RNA. Segment 1 and segment 5 vRNAs or +RNAs were probed for on the same membrane, but due to differences in sensitivies of the probes, different exposures are presented for the two segments.

Fig 6. CpGH IAV is attenuated in mice and conveys protection from challenge with a lethal dose of WT virus.

Groups of six, 6 week-old female BALB/c mice were infected with 200 PFU of either WT PR8 virus, CDLR permuted control, CpGH virus or mock infected (n = 2) for 5 days; weight change measured daily (A). On day 5, lower left lung lobes were harvested and the presence of infectious virus was quantified (B). To determine the protective effect of CpGH virus infection, groups of 6–10 mice were infected with a low dose (20 PFU) of either PR8 (n = 6), CDLR control (n = 10), CpGH virus (n = 10) or were mock infected (n = 10) and weighed daily for 10 days until full recovery (C). At day 20 post-exposure, tail bleeds were used to collect sera for ELISA to quantify anti-IAV antibody levels (D). The next day, five mice exposed to either CDLR or CpGH virus, or mock exposed, were challenged with a potentially lethal dose of PR8 WT virus (200 PFU) and weighed daily for 5 days (E) at which time mice were culled and lower left lung lobes were collected for virus titration (F) and antibody levels in sera were quantified by ELISA (G). Sera were also tested for IAV protein binding using western blotting (S5 Fig).

Fig 7. A ZAP-sensitive attenuated CpG-enriched IAV replicates equivalently to WT virus in vaccine propagation systems.

A. PR8 WT, CDLR and CpGH viruses were grown in embryonated hens’ eggs or in MDCK cells and titred. B. Embroyonated hens’ eggs were inoculated with 100 PFU virus at developmental day 5 or day 9, and viral titre in allantoic fluid was assessed 24 hours later. C. Lethality of the virus panel in eggs was fit to nonlinear regression curves for LD50 calculations. D. The virus panel was grown in chicken epithelial cell line CLEC213 and viral titres were quantified by plaque assay. E. The virus panel was used to infect chicken enteroids at an MOI of 10 for 24 hours, and the production of infectious virus, and viral protein production (F) were assayed. For confocal imaging of infected enteroids, please refer to S8 Fig. For innate gene expression analyses from infected enteroids, please refer to S9 Fig.

Fig 8. Serial passage of CpG modified IAV does not enable deselection of CpG motifs.

Egg (A) or MDCK (B) derived virus stocks were passaged at an MOI of 0.01 ten times in A549 cells, and titred after each passage (n = 2). After ten passages, viruses were whole genome sequenced to test for reversion or epistatic compensatory mutations. (C). Total cumulative genome mutations after ten passages were calculated, with proportional representation of mutations that had not become fixed (e.g. if 10% of sequences contained a mutation at a given position, this scored as 0.1 mutations) calculated for egg virus stock passages and MDCK virus stock passages. (D). Cumulative mutations were calculated for each segment for each of the three viruses in the panel across four replicates (E). Mutation frequency at recoded positions across segment 1 for CpGH virus. The mutation frequencies of the four serially passaged replicates are overlaid. No mutations at CpG sites were observed. For sequence coverage, please refer to S10 Fig and for full genome mutational profiles to S11 Fig.