Mammalian IAV sequences are CpG-depleted relative to avian IAV

Viruses evolve to replicate successfully in their host species. Previous work showed that human IAV exhibited a depletion of CpG dinucleotide content relative to avian IAV sequences since the 1918 IAV pandemic [14,19,35]. Here, surveying influenza sequences in FluDB [36] collected between 1918 and 2020, we also observed a depletion of CpG dinucleotides per kilobase (CpG content) in human and swine but not avian IAV sequences over time since 1918 (S1A Fig). By comparing all human, swine, and avian IAV segment sequences en masse, we observed a lower CpG content in human and swine IAV relative to avian IAV sequences (Fig 1A). We observed similar results for ρ_CG (rho [ρ] accounts for total C and G content [13,14,17]) but less so for other dinucleotides or %GC content (S2A–S2D Fig). Notably, the largely human-specific influenza B virus [37] (IBV) and influenza C virus [38] (ICV) are CpG-depleted relative to all IAV sequences, speculatively consistent with long-term near-exclusive circulation in mammalian/human populations (Figs 1A and S1–S2). We observed that human IAV segments 1–3 encoding the heterotrimeric viral polymerase (i.e., PB2, PB1, and PA, respectively) are more CpG-depleted relative to avian IAV than segments 4–6 encoding virion proteins (i.e., HA, NP, and NA, respectively), typically subject to adaptive immune pressure [39] (S1B Fig). We found that when comparing major HA and NA subtypes, CpG content is depleted in subtypes seasonally circulating in humans (H1N1 and H3N2) relative to avian sequences of the same subtype (S1C Fig). Conversely, CpG content in human isolates of common avian-to-human zoonotic HA and NA subtypes (e.g., H5N1 and H7N9) is similar to their avian counterparts. The natural history of human IAV includes multiple introductions [40] resulting from avian zoonoses and/or reassortment of avian IAV segments with human viruses in 1918, 1957, 1968, and 2009. We compared CpG content of avian IAV to human IAV during the time periods surrounding the replacement of H1N1 with H2N2 in 1957, H2N2 with H3N2 in 1968, reintroduction of H1N1 in 1977, and pandemic “swine flu” H1N1 in 2009 (S1D Fig, bottom). We observed that CpG content is lower in human relative to avian IAV sequences during the longer time periods following pandemics in 1977 and 2009 (S1D Fig, top). This may reflect periodic reintroduction of CpG-enriched avian sequences that deplete over longer time frames.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Chicken and duck ZAP do not potently restrict replication of PR8 IAV or CpG-enriched PR8 IAV.

(A) IAV sequences exhibit depletion of CpG content (CpG per kb = # CpG/segment length in kilobases [kb]). Box and whisker plots showing CpG content of all IAV sequences from avian (orange), human (blue), and swine (green) hosts as well as IBV (pink) and ICV (purple). Thick black bars represent median CpG content, boxes define the 25–75th percentile, and whiskers represent range excluding outliers. Horizontal dashed gray lines highlight median avian IAV CpG content as a reference. The data underlying this panel can be found in https://doi.org/10.5281/zenodo.14919910. (B) Schematic of human ZAP-L, human ZAP-S, and chicken ZAP (ggaZAP) protein domain architecture including four CCCH zinc fingers and inactive poly-ADP ribose polymerase (iPARP). Amino acid alignment of ZAP zinc finger 2 (ZF2; human residues 88–110) from multiple species. Conserved CCCH residues are highlighted with green vertical bars. (C) CpG content of PR8 (shades of gray by segment length), wild-type NP (blue), and CpG-enriched NP (orange) calculated as # CpG per 100 nucleotides (nt) over a sliding window (top). Schematic of PR8 and PR8_CG viruses and displaying CpG content for each virus and total CpG number encoded by each NP segment (bottom). (D) Titers of PR8 and PR8_CG at 48 hours post-infection (hpi) in chicken DF-1 cells stably expressing mNeonGreen (mNG, gray), mNG-tagged chicken ZAP (mNG-ggaZAP, yellow), or untagged ggaZAP (orange) as well as control transduced cells (Vector, black). Bar represents mean plaque-forming units per mL (PFU/mL) with error bars representing standard deviation of the mean. Multiplicity of infection (MOI) = 0.05 (pfu per cell). Individual data points are displayed as black circle outlines. (E) Fluorescence images of DF-1 cells expressing mNG as well as fusion proteins of mNG-ggaZAP and mutants indicated (green, top). Nuclei counterstained with DAPI (blue) and shown in merge image (bottom). Scale bars represent 10 microns. (F) Titers of PR8 and PR8_CG over 48 hpi in chicken DF-1 cells stably expressing mNG (black circles), mNG-ggaZAP (orange circles), mNG-hZF2 (green circles), mNG-C710S (orange/gray squares), and mNG-hZF2.C710S (green/gray diamonds) mutants shown in (E). MOI = 0.05. Individual data points are displayed as 50% transparent jittered shapes behind mean data point with errors bars representing standard deviation of the mean. (G) Titers of PR8 and PR8_CG at 48 hpi in parental untransduced DF-1 cells (DF-1, black), pooled CRISPR/Cas9 knockout chicken DF-1 cells expressing a control sgRNA targeting Escherichia coli LacZ (LacZ, gray), or sgRNAs targeting endogenous chicken ZC3HAV1 (ggaZAP1-3 KO, shades of orange). Bar represents mean PFU/mL with error bars representing standard deviation of the mean. MOI = 0.05 (pfu per cell). Individual data points are displayed as black circle outlines. (H) Titers of PR8 and PR8_CG at 48 hpi in parental untransduced CCL-141 cells (CCL-141, black), pooled CRISPR/Cas9 knockout duck CCL-141 cells expressing a control sgRNA targeting E. coli LacZ (LacZ, gray), or sgRNAs targeting endogenous duck ZC3HAV1 (aplZAP1-3 KO, shades of yellow). Bar represents mean PFU/mL with error bars representing standard deviation of the mean. MOI = 0.05 (pfu per cell). Individual data points are displayed as black circle outlines. The data underlying the graphs in this figure can be found in S2 Table.

https://doi.org/10.1371/journal.pbio.3003471.g001

Chicken and duck ZAP may not restrict replication of CpG-enriched IAV

We hypothesized that avian species are defective in cytoplasmic CpG-sensing by ZAP and reasoned that this deficiency allows CpG-maintenance of IAV sequences in avian host species. Human ZAP-L and chicken ZAP are homologous proteins encoding four zinc fingers, an inactive poly-ADP-ribose polymerase-like domain [41], and carboxy-terminal prenylation motif [23,29] (Fig 1B, top). Previous structural and functional analysis of human ZAP identified zinc finger 2 (ZF2, residues 88–110) as responsible for directly recognizing and binding to CpG dinucleotides in the context of short (3–20 nucleotide) and long (full-length 9 kilobase HIV-1 genome) RNA molecules [20,42,43]. Using two structural prediction algorithms, we generated models of the four zinc fingers of chicken ZAP (ggaZAP, residues 1–220) by both comparative modeling based on human ZAP (RobettaCM [44], PDB: 6UEJ [42]) and ab initio folding by RoseTTAFold [45] (S3A and S3B Fig). We also compared these structures to the model of chicken ZAP predicted by AlphaFold2 [46] (S3A and S3B Fig). As expected, we observed that both ab initio predicted structures (RoseTTAFold and AlphaFold2) exhibited lower structural similarity (as measured by root mean squared deviation, RMSD) to human ZAP relative to those generated with comparative modeling (S3B Fig). In addition, we observed structural differences between human ZAP and models of duck ZAP as well as quail ZAP (S3A and S3B Fig). Alignment of the amino acid sequences of ZAP homologs revealed multiple amino acid differences between ZF2 of avian and mammalian species (Fig 1B, bottom). Therefore, we hypothesized that chicken ZAP is unable to restrict CpG-enriched viruses and tested this hypothesis using the common IAV laboratory strain A/Puerto Rico/8/1934//H1N1 (PR8, blue dashed box) and a CpG-enriched version of PR8 [47] (PR8_CG, orange dashed box; Fig 1C).

In chicken DF-1 cells stably expressing chicken ZAP (ggaZAP) as well as mNeonGreen-tagged chicken ZAP (mNG-ggaZAP), we found no difference in virus replication over 48 hours for PR8 or PR8_CG relative to vector-transduced (Vector) and mNeonGreen-transduced (mNG) cells (Fig 1D). PR8_CG exhibits consistently lower replication relative to wild-type PR8 as previously shown [47], but allows comparisons for viral replication across different cellular contexts. We confirmed expression of transduced proteins by western blot for mNG as well as blasticidin S deaminase (BSD) [48,49] expressed on the bicistronic mRNA and used for drug selection (S3C Fig). The chicken ZC3HAV1 gene encodes a single ZAP isoform of 713 amino acids with a carboxy-terminal prenylation motif (amino acids: ₇₁₀CIVC₇₁₃) that is homologous to the long isoform of human ZAP (ZAP-L). The short isoform of human ZAP (ZAP-S) is generated by alternative splicing and lacks the carboxy-terminal prenylation motif and the inactive poly-ADP-ribose polymerase-like domain. To test if either prenylation of chicken ZAP or if the divergent ZF2 regulate restriction of CpG-enriched PR8 IAV, we generated mutants of chicken ZAP lacking the critical cysteine residue (C710S) for membrane-targeting prenylation, a chimera encoding the human ZF2 (hZF2), or a combination (hZF2.C710S) of these two mutants all of which expressed and localized as expected (Figs 1E and S3D). However, no form of chicken ZAP restricted PR8 or PR8_CG replication (Fig 1F). We confirmed protein expression of transduced genes by western blot for mNG as well as BSD (S3D Fig). Finally, we found that pooled CRISPR/Cas9-mediated knockout of endogenous ZAP in chicken DF-1 and duck CCL-141 cells had minimal effects on PR8 or PR8_CG replication (Figs 1G–1H and S4). Similarly, we observed that knockdown of endogenous ZAP by stable shRNA in chicken DF-1 cells or transient siRNA in duck CCL-141 cells had minimal effects on PR8 or PR8_CG replication (S3E–S3G Fig). qPCR showed endogenous chicken ZAP is expressed in DF-1 cells, induced by interferon, and can be suppressed by shRNA (S3E Fig). Attempts at co-expression of human ZAP-S and KHNYN as exogenous transgenes in chicken DF-1 cells were complicated by apparent ZAP-S-related inhibition of KHNYN protein expression or stability (S3H Fig). Given that human ZAP can use chicken TRIM25 as a cofactor [50], we reasoned that chicken DF-1 cells are potentially deficient in ZAP-dependent antiviral activity targeting this CpG-enriched IAV and could serve as a gain-of-restriction system to test related host factors for the ability to restrict CpG-enriched viruses.

Human ZAP-S and KHNYN can independently restrict CpG-enriched PR8 IAV

Restriction of CpG RNA has been attributed to human ZAP along with its cofactors KHNYN and/or TRIM25. However, we found that avian species may lack a bona fide KHNYN orthologue. There are no annotated avian KHNYN orthologues in ensembl, NCBI, or UCSC Genome Browser and a ~400,000 bp syntenic region surrounding human KHNYN is missing in avian species, including the KHNYN paralogue—NYNRIN (S5A Fig). Human KHNYN and NYNRIN are encoded in tandem on chromosome 14 and NYNRIN is the result of a KHNYN gene duplication and fusion to a retroviral integrase domain [34] (S5B Fig). KHNYN and NYNRIN are paralogous to N4BP1 [51], a gene conserved in birds, reptiles, and fish (S5B and S5C Fig). We estimated a maximum likelihood phylogeny of human KHNYN, NYNRIN, and N4BP1 orthologues nucleotide sequences from ensembl (S5C Fig). We confirmed that N4BP1 is conserved in Gnathostomats (i.e., jawed vertebrates including fish, birds, reptiles, and mammals) and is closely related to a fish-specific N4BP1 paralogue (sometimes annotated as khnyn or khnyn-like, e.g., Danio rerio khnyn, ENSDARG00000092488). However, KHNYN found in mammalian species (i.e., monotremes, marsupials, and placentals) and NYNRIN found in therian mammals (i.e., marsupials and placentals) branched together and are more like each other than to the N4BP1 genes (S5C Fig). We noted that NCBI and ensembl contain reptile sequences annotated as KHNYN (e.g., Alligator mississippiensis KHNYN XM_014607035.3 or Anolis carolinensis khnyn ENSACAG00000009105) that may represent ancestral KHNYN orthologues based on nucleotide sequence (S5C Fig; ancient KHNYN orthologues), exon organization, and syntenic neighborhood (S5D Fig). Thus, we reasoned that avian species may be defective and/or deficient in two restriction factor genes: ZAP as well as KHNYN. As such, we generated chicken DF-1 cells that stably expressed human ZAP-L, ZAP-S, KHNYN, or TRIM25. We included a mutant of KHNYN that is expected to be enzymatically inactive within its endonuclease domain [21] due to substitution of three critical aspartic acid residues (DDD443/524/525AAA, dKHNYN). We found that ZAP-S as well as KHNYN modestly restricted PR8 relative to vector transduced cells (Fig 2A, left). KHNYN has not been shown previously to act independently of ZAP. Surprisingly, we found that human ZAP-S and KHNYN (but not ZAP-L, dKHNYN, or TRIM25) independently restricted PR8_CG relative to vector (Fig 2A, right).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Human ZAP-S and KHNYN can independently restrict CpG-enriched PR8 IAV.

(A) Titers of PR8 and PR8_CG over 48 hpi in chicken DF-1 cells stably expressing human ZAP-L (orange circles), ZAP-S (green circles), KHNYN (blue circles), dKHNYN (gray circles), or TRIM25 (pink circles) relative to Vector (black circles). MOI = 0.05. *** indicates p ≤ 0.0001 at 48 hpi by two-way ANOVA. (B) Titers of PR8 and PR8_CG over 48 hpi in chicken DF-1 cells stably expressing RNA-binding mutants of ZAP-S (green circles) including Y108A (green squares, dashed line), and chimeric encoding chicken ZF2 (gZF2; gray diamonds, dashed line) relative to Vector (black circles). MOI = 0.05. ** indicates p = 0.0093 at 48 hpi by two-way ANOVA. (C) Titers of PR8 and PR8_CG over 48 hpi in chicken DF-1 cells stably expressing C-terminal mutants of human ZAP-L (orange circles) and ZAP-S (green circles) including ZAP-L C899S (oranges squares, dashed line) and ZAP-S +CVIS motif (CVIS; green diamonds, dashed line). MOI = 0.05. * indicates p = 0.0108 at 48 hpi by two-way ANOVA. (D) Titers of PR8 and PR8_CG over 48 hpi in chicken DF-1 cells stably expressing mutants of KHNYN (blue circles) including catalytically inactive dKHNYN (gray circles), ₉₅AQ/DE₉₆ mutant (AQDE; blue squares, dashed line), and combination mutant of dKHNYN with AQDE (dAQDE; gray diamond, dashed line). MOI = 0.05. * indicates p = 0.0158 at 48 hpi by two-way ANOVA. (E–H) Western blots of chicken DF-1 cells stably expressing human ZAP, KHNYN, or TRIM25 and mutants used in (A–D) detecting ZAP, KHNYN, and/or TRIM25 as well as tubulin (TUBA) or histone 3 (H3) loading controls. Individual data points are displayed as 50% transparent jittered shapes behind mean data point with errors bars representing standard deviation of the mean. The data underlying the graphs in this figure can be found in S2 Table. Original annotated immunoblot images can be found in S1 Raw Images.

https://doi.org/10.1371/journal.pbio.3003471.g002

CpG-targeted restriction requires intact RNA-binding domain of human ZAP and extended di-KH domain of KHNYN

To confirm the role of ZAP-S and KHNYN in restricting PR8_CG, we generated mutants of each, stably expressed them in DF-1 cells, and performed multicycle IAV replication assays. To test the CpG specificity of ZAP-S in restricting PR8_CG, we generated a Y108A mutant previously shown to reduce CpG specificity [42] as well as a chimeric human ZAP-S encoding the ZF2 of chicken ZAP (gZF2). Only WT human ZAP-S remained capable of restricting PR8_CG, an ability that was lost by the Y108A and gZF2 mutants (Fig 2B). To test if the prenylation motif (amino acids: CVIS) and corresponding membrane localization of ZAP regulated restriction activity against PR8_CG, we generated a mutant of ZAP-S that encoded a carboxy-terminal CVIS motif (+CVIS) as well as a C899S mutant of ZAP-L [23]. While the C899S mutant of human ZAP-L did not gain the ability to restrict PR8_CG, human ZAP-S +CVIS lost the ability to restrict PR8_CG as expected (Fig 2C). Next, to delineate the manner of KHNYN restriction of PR8_CG, we mutated its extended di-KH domain [52]. KH-domain containing proteins encode a GXXG amino acid motif at the top of a central alpha-helix and mutations in the GXXG motif can reduce RNA-interactions in another KH-domain containing protein, HNRNPK [53]. We generated models of this extended di-KH domain (amino acids 1−213) of KHNYN by ab initio folding (RoseTTAFold) (S5E Fig) that showed a central alpha-helix with the ₉₄GAQG₉₇ motif at its apex adjacent to a three-helix bundle (S5E Fig). This model appeared structurally similar to a recent crystal structure of KHNYN extended di-KH domain [52]. We found that only WT human KHNYN was capable of restricting PR8_CG over 48 h of replication in chicken DF-1 cells, but not the catalytically inactive dKHNYN, a AQ95/96DE mutant (AQDE), or a catalytically inactive AQ95/96DE (dAQDE) (Fig 2D). We confirmed expression of transduced human proteins by western blot (Fig 2E–2H). The Vector transduced cells used in Fig 1D showing chicken ZAP had minimal, if any, effects on PR8 or PR8_CG are the same Vector transduced cells used throughout Fig 2A–2D showing human ZAP-S and KHNYN restricted PR8_CG allowing for relative comparisons in ability for these proteins to restrict these IAV. Nonetheless, we wanted to compare human ZAP-S restriction of PR8 or PR8_CG to chicken ZAP directly in multicycle IAV replication assays. Using chicken DF-1 cells stably expressing yellow fluorescent protein (YFP) tagged proteins, we found that human ZAP-S and KHNYN restricted PR8_CG while human ZAP-L, chicken ZAP, and dKHNYN failed to negatively affect PR8 or PR8_CG (S6A and S6B Fig). Human ZAP-S and KHNYN also modestly restricted an avian IAV strain (OH175) with a modest but naturally CpG-enriched PA segment (S6A–S6C Fig). Finally, we observed that human ZAP-S, KHNYN, and dKHNYN may independently interact with PR8_CG segment 5 RNA by RNA immunoprecipitation and quantitative PCR (RIP-qPCR) in infected DF-1 cells stably expressing these YFP-tagged proteins (S6D Fig). We also observed similar expression of viral proteins (NP) in PR8 and PR8_CG infected cells. Together, these results indicated that intact human ZAP-S as well as KHNYN can independently restrict PR8_CG.

Chicken and IAV mRNA are largely unaffected by human ZAP or KHNYN

To understand how these mammalian restriction factors might affect IAV, we also examined transcriptomic changes during IAV infection. We performed RNA-seq on uninfected as well as PR8 or PR8_CG infected chicken DF-1 cells stably expressing human ZAP-L, ZAP-S, KHNYN, or dKHNYN to evaluate if these human proteins modulated the chicken transcriptome and viral mRNA expression relative to vector transduced cells. We found that the transcriptomes of uninfected cells were largely unperturbed by whatever human factor was overexpressed (S7A Fig) suggesting antiviral activity may not be caused by changing the underlying gene expression profile. Next, we hypothesized that by interacting with and depleting CpG-enriched viral RNAs during multicycle IAV replication experiments, ZAP-S and KHNYN might also negatively bias CpG-enriched segment accumulation during replication. We evaluated relative levels of IAV mRNA segments in PR8 versus PR8_CG infected cells in the presence of these different human factors and found similar levels of viral mRNA within segments (S7B Fig). Infection with either PR8 or PR8_CG resulted in similar gene expression patterns for host cell factors between the two viruses, suggesting that the mechanism of restriction is not through differential induction of host antiviral responses (S7C Fig). Together, these results suggested initial mechanistic insights into how ZAP and KHNYN might or might not restrict IAV during zoonotic infections in that the restriction may not substantially deplete viral mRNAs.

Endogenous human ZAP and KHNYN restrict CpG-enriched viruses

We next tested the ability of endogenous human ZAP and/or KHNYN to restrict CpG-enriched IAV in human lung cells. Therefore, we generated multiple human lung adenocarcinoma A549 clonal cell lines depleted for ZAP, KHNYN, or both by CRISPR/Cas9 genome editing as well as a vector control (Vector) and a negative control guide targeting Escherichia coli LacZ beta-galactosidase (LacZ) clonal cell lines (Fig 3A–3B). We tested PR8 and PR8_CG by multicycle replication assay in these cell lines. We found that single knockout of only ZAP or only KHNYN failed to increase PR8 or PR8_CG replication relative to LacZ or parental A549 cells (Fig 3C). We reasoned that this was consistent with our results by overexpression in DF-1 cells such that either ZAP or KHNYN are capable of independently restricting PR8_CG. Therefore, we tested multicycle replication of PR8 and PR8_CG in double knockout (ZAP and KHNYN depleted, ZK1–2) cells revealing that genetic loss of both ZAP and KHNYN increased replication of PR8_CG relative to parental A549 or Vector cell lines (Fig 3D) supporting the hypothesis that ZAP and KHNYN can independently restrict of CpG-enriched IAV. We tested two single-cell clones per gene targeted for knockout to assess potential clonal effects and we hypothesize that the decreased replication in ZAP KO 2 is a clonal effect. The same guide (ZAP guide 2) was used to make the double knockouts that allowed for increased replication across two independent clones. We also found replication of PR8 or PR8_CG was unaffected by knockout of endogenous human TRIM25 in A549 cells (Fig 3E). Altogether, these results support the hypothesis that ZAP and KHNYN may act independently against CpG-enriched PR8 IAV.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Endogenous human ZAP and KHNYN can independently restrict CpG-enriched PR8 IAV.

(A) Schematic depicting exon organization of human ZAP (green), KHNYN (blue), and TRIM25 (pink) genes and sites targeted by CRISPR single guide RNAs (sgRNAs; carets). Colored boxes indicate coding exons and white boxes indicate noncoding portions of exons. (B) Western blots of CRISPR/Cas9 engineered knockout (KO) human A549 cell lines including parental A549 cells, control clone transduced with pLentiCRISPR1000 (Vector), control clone targeting Escherichia coli LacZ (LacZ), two ZAP KO clones (ZAP1 and ZAP2), two KHNYN KO clones (K1 and K2), two ZAP and KHNYN dual KO clones (ZK1 and ZK2), and two TRIM25 KO clones (T1 and T2). Cells were treated with vehicle (PBS) or 1,000 units per mL (U/mL) universal interferon-alpha (IFNα) for 24 hours. Histone H3 detected as loading control. (C) Titers of PR8 and PR8_CG at 48 hpi in human A549 KO cells including parental (black), Vector and LacZ (shades of gray), ZAP KO clones (shades of green), and KHNYN KO clones (shades of blue). MOI = 0.05. (D) Titers of PR8 and PR8_CG at 48 hpi in human A549 KO cells including parental (black), Vector (gray), and ZAP+KHNYN KO clones (green and blue striped). MOI = 0.05. *** indicates p ≤ 0.0001 and ** indicates p = 0.0045 by one-way ANOVA. (E) Titers of PR8 and PR8_CG at 48 hpi in human A549 KO cells including parental (black), LacZ (gray), and TRIM25 KO clones (shades of pink). MOI = 0.05. Bars represent mean PFU/mL with error bars representing standard deviation of the mean. Individual data points are displayed as black circle outlines. The data underlying the graphs in this figure can be found in S2 Table. Original annotated immunoblot images can be found in S1 Raw Images.

https://doi.org/10.1371/journal.pbio.3003471.g003

ROSV is CpG-enriched and susceptible to mammalian CpG-dependent restriction

Avian IAV exhibits an enrichment of CpG content relative to human IAV (Figs 1A, S1, and S2). We next asked if other avian viruses with different levels of human or mammalian zoonotic interactions might exhibit higher levels of CpG enrichment. We noticed that ROSV (a tractable retrovirus model in chicken) exhibits a higher CpG content relative to mammalian retroviruses, including human immunodeficiency virus 1 (HIV-1) or murine leukemia virus (MLV). We observed that ROSV is CpG-enriched along its entire length relative to mammalian retroviruses including HIV-1 or MLV (Fig 4A). To determine whether ROSV was subject to CpG-dependent restriction, we performed retroviral assembly and single-cycle infectivity assays (schematic in Fig 4B) by co-transfecting chicken DF-1 cells with a full-length ROSV reporter virus expression plasmid and a VSV-G expression plasmid along with increasing amounts of control, human ZAP-L, human ZAP-S, KHNYN, or dKHNYN plasmid. We found that human ZAP-L inhibited ROSV infectivity, but ZAP-S did not (Fig 4C). Using mutants of human ZAP-L and ZAP-S, we confirmed that for ROSV, ZAP restriction required an intact ZF2 and prenylation motif (Fig 4C). Specifically, the C899S prenylation mutant of ZAP-L failed to restrict ROSV while ZAP-S fused to the four amino acid ZAP-L prenylation motif (+CVIS) gained the ability to restrict ROSV. Next, we tested human ZAP-L, chicken ZAP, and duck ZAP as well as prenylation mutants of each for restriction of ROSV using fusions to mNeongreen. We found that human ZAP-L restricted ROSV but neither of the avian ZAP proteins or their prenylation mutants were much less restrictive (Fig 4D). Finally, we tested mammalian orthologues of ZAP from dog, pig, and platypus and found that the prenylated forms of placental mammal ZAP (pig and dog) restricted ROSV, while platypus ZAP or a prenylation mutant (C653S) were less restrictive though they expressed less readily (S8A Fig). We confirmed expected subcellular localization of mammalian and avian ZAP isoforms and mutants in chicken DF-1 cells (S9A Fig) and observed conservation of CaaX prenylation amino acid motif of representative ZAP orthologues (S9B Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Prenylated human ZAP and mammalian KHNYN can independently restrict naturally CpG-enriched ROSV.

(A) CpG content of retroviruses (ROSV in black, ROSV-mCherry reporter virus in pink dashed line, MLV in orange, and HIV-1 in blue) calculated as # CpG per 100 nucleotides over a sliding window (graph). Schematic of retrovirus open reading frame (ORF) organization (top). (B) Schematic of retroviral assembly and single-cycle infectivity assay including plasmid transfection into cells, cell lysate and supernatant harvest at 48 hours post-transfection (hpt), storage, and western blotting as well as DT40 suspension cell infection and flow cytometry as a measure of infectivity. (C) Retroviral assembly and single-cycle infectivity results of human ZAP-L, ZAP-S, and mutants effects on ROSV showing relative percent ROSV infected cells (% mCherry-positive; % mCh+) as determined by flow cytometry. DF-1 cells transfected with increasing plasmid amounts of indicated host proteins (100, 200, or 400 ng), 1,500 ng ROSV, and 200 ng VSV-G. Infectivity is relative to backbone pcDNA3.2 transfections (Mock). Below, western blots detecting ROSV cleaved Gag^p19 in supernatant virions as well as full-length ROSV Gag^Pr76, ZAP proteins, and ACTIN loading controls from cellular lysates. (D) Retroviral assembly and single-cycle infectivity results of human ZAP-L, chicken (gga) ZAP, duck (apl) ZAP, and prenylation mutant on ROSV infectivity. Below, western blots detecting ROSV cleaved Gag^p19 in supernatant virions as well as full-length ROSV Gag^Pr76, mNG-tagged ZAP proteins, and ACTIN loading controls from cellular lysates. (E) Retroviral assembly and single-cycle infectivity results of human KHNYN, mutants, and mammalian KHNYN homologs (from pig, dog, and platypus) on ROSV infectivity. Below, western blots detecting ROSV cleaved Gag^p19 in supernatant virions as well as full-length ROSV Gag^Pr76, human KHNYN proteins, and ACTIN loading controls from cellular lysates. Note mammalian KHNYN orthologues were not detected by human KHNYN antibodies. (F) Single-cycle infectivity (%mCh+ or %NP+) in A549 stable cells expressing YFP-alone (black) and YFP-tagged platypus KHNYN (blue) effects on single-cycle IAV including laboratory PR8, pandemic Cal04, and avian-origin OH175. *** indicates p = 0.000832, * indicates p = 0.025875, and ** indicates p = 0.002263 by unpaired t test. (G) IAV Minigenome reporter assay comparing mNG-tagged mammalian KHNYN (shades of blue: human, pig, dog, and platypus) relative to mNG alone (black) effects on IAV vRNA production (% mCh+ cells) using PB2, PB1, PA, and NP expression plasmids and a negative-sense mCherry vRNA reporter plasmid (encoding 34 CpG dinucleotides within the proxy vRNA). The bar represents mean relative infectivity with error bars representing standard deviation of the mean. Individual data points are displayed as black circle outlines. The data underlying the graphs in this figure can be found in S2 Table. Original annotated immunoblot images can be found in S1 Raw Images.

https://doi.org/10.1371/journal.pbio.3003471.g004

Mammalian KHNYN restricts ROSV independent of ZAP

Next, we tested human KHNYN, mutant KHNYN, and mammalian KHNYN orthologues from pig, dog, and platypus for restriction of ROSV. Pigs and dogs are domesticated mammals frequently in contact with agricultural poultry, a common source of avian-to-mammal zoonoses [4]. We selected platypus KHNYN to test because it is the only mammalian species to encode KHNYN but lack the NYNRIN gene [34] (S3C Fig). We found that pig KHNYN, but not dog KHNYN, similarly restricted ROSV as human KHNYN while platypus KHNYN potently restricted ROSV (Fig 4E). We also tested these proteins (with mNeongreen-tagged at their amino terminus) relative to catalytically inactivated mutants for ROSV restriction activity (S8B Fig). Curiously, mutation of the aspartic acid residues in the NYN-endonuclease domains inhibited the restrictive ability for pig KHNYN (DD508/508AA) but not platypus KHNYN (DDD497/498AAA) (S8B Fig). We tested multiple truncated forms of platypus KHNYN and found that truncation mutants which fully deleted the endonuclease domain lost restriction activity (S8C Fig). We observed that the NYN domains of platypus KHNYN, including the aspartic acid catalytic residues, are relatively well conserved by amino acid alignment and computational structural modeling (S10A and S10B Fig). We generated structural models of platypus KHNYN and noted a substantial structural difference in the loop above the apex of the putative GXXG motif in the KH domain in platypus KHNYN encoding a notably shorter loop relative to human, pig, and dog KHNYN (S5E Fig). We confirmed appropriate subcellular localization of these KHNYN orthologues in chicken DF-1 cells (S10C Fig). Curiously, we found that platypus KHNYN also restricted HIV-1, a CpG-enriched HIV-1, and MLV, independent of glycoprotein provided for pseudo-typing (S8D Fig). We detected cellular expression of retroviral (ROSV Gag^Pr76 and Gag^p19; HIV-1 Gag^Pr55 and Gag^p24; MLV Gag^mCh) and antiviral proteins for retroviral assembly and single-cycle infectivity assays as well as loading controls and purified virions from culture supernatants. We noted that platypus KHNYN could potently inhibit virion assembly and release of ROSV, HIV-1, and MLV with typically minimal effects on intracellular Gag expression or loading controls (notably in Figs 4E and S8D). We also found similar results for mammalian KHNYN restriction of ROSV in the absence of chicken ZAP using DF-1 cells with ggaZC3HAV1 knocked out by CRISPR/Cas9 (S8E and S4 Figs). Finally, we tested if platypus KHNYN could restrict IAV replication in human A549 cell lines expressing YFP-tagged platypus KHNYN. Relative to YFP-alone expressing cells, platypus KHNYN modestly inhibited single-cycle PR8, pandemic Cal04, and avian-origin OH175 (Fig 4F). Platypus KHNYN also restricted PR8_CG and to a lesser extent OH175 in multicycle replication assays (S6A Fig). We also observed that mammalian KHNYN negatively affects viral polymerase activity using an IAV minigenome reporter assay (Fig 4G). Together, these results support the hypothesis that mammalian ZAP and KHNYN can restrict CpG-enriched avian viruses, including strains of IAV as well as ROSV. Furthermore, we discovered an orthologue of KHNYN in platypus that may potently restrict multiple viruses from distinct virus families independent of CpG content.

Cell lines

Mammalian cells, chicken DF-1 cells, and duck CCL-141 cells were grown in DMEM supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Chicken DT40 cells were grown in DMEM supplemented with 10% FBS, 10% tryptose phosphate buffer, 1% penicillin/streptomycin, and 50 µM beta-mercaptomethanol. Cells were maintained at 37 °C, 5% CO₂, and 50% humidity. HEK293T (male human epithelial kidney, ATCC, CRL-3216), A549 (male human lung carcinoma, ATCC, CCL-185), PK-15 (male porcine kidney, ATCC, CCL-33), MDCK (female canine kidney, ATCC, CCL-34), DF-1 (unspecified sex chicken embryonic fibroblast, ATCC, CRL-12203), DT40 (unspecified sex chicken lymphoblast, ATCC, CRL-2111), and CCL-141 (unspecified sex duck embryonic fibroblast, ATCC) cells were obtained from ATCC and used without extensive passaging. Sup-T11 (male human T lymphoblast clone derived from Sup-T1, ATCC, and CRL-1942) cells have been described previously [101].

Plasmids

A CpG-enriched IAV segment 5 (encoding nucleoprotein, NP) derived from Gaunt and colleagues [47], ordered from IDT as a gBlock, and cloned into a pDZ bidirectional influenza vRNA/mRNA expression plasmid [102] via In-Fusion (Takara Bio). Expression plasmids for human ZAP-L, ZAP-S, KHNYN, and TRIM25 as well as pig ZAP, chicken ZAP, and pig KHNYN were generated using conventional molecular biology techniques. Specifically, RNA was extracted from human A549 cells, chicken DF-1, or pig PK-15 cells using Qiagen RNeasy Mini Kit, cDNA synthesized using QIAGEN Quantitect cDNA Synthesis Kit, and genes of interest were amplified by PCR from first-strand cDNA, restriction enzyme digested, and ligated into bespoke MLV-based retroviral vectors also harboring blasticidin S deaminase (BSD) or hygromycin resistance genes downstream of an IRES for bicistronic expression [100,103,104] or bespoke pcDNA3-based expression vectors. Dog and platypus KHNYN as well as dog and platypus ZAP were ordered from IDT as gBlocks. Duck ZAP was synthesized by Twist Biosciences followed by subcloning into additional plasmid backbones. Mutants, chimeras, and fluorescently tagged versions of all plasmids were generated by site-directed mutagenesis and/or overlapping PCR and subcloned into expression vectors using Phusion DNA polymerase and T4 DNA ligase (NEB). RCAS-GFP plasmid [105] was a gift of Connie Cepko (Addgene #13878). ROSV-mCherry was generated by replacing EGFP in RCAS-GFP with mCherry followed by site-directed mutagenesis to inactivate the Env glycoprotein open reading reframe. HIV-mCherry has been described previously [87]. HIV_CG-mCh was cloned by inserting a gBlock (IDT) encoding env86-561CpG [21] (from NCBI: MN685350.1 [21]) into an HIV-1 NL4-3 Env- Vpr- Nef- mCherry+ (HIV-mCh) reporter virus expression plasmid. MLV-Gag-HA-mCh was a kind gift of Nathan Sherer [106]. Codon-optimized HIV-1 NL4-3 Env (SynEnv) was synthesized as a gBlock by IDT prior to subcloning into pcDNA3.2. pCAG-RABV-G (Addgene #36398) expressing Rabies virus glycoprotein (RABV-G) was a kind gift of Connie Cepko [107]. CRISPR/Cas9 editing of human A549 cells was performed by lentiviral transduction using pLentiCRISPR1000 [108]. Attempts at genetic editing in avian cells with pLentiCRISPR1000 were unsuccessful. CRISPR/Cas9 editing of avian cells was performed by sequential retroviral transduction of Cas9-GFP expression vector followed by single guide RNA (sgRNA) expression vector—both bespoke plasmids generated for this study. Briefly, guide RNAs were designed from the literature, Synthego design tool, or RGEN Cas-Designer [109], complementary oligos with Esp3I restriction enzyme compatible ends were annealed, and subsequently ligated into pLentiCRISPR1000 or pMLV.miRFP670.iU6sgRNA via Golden-Gate ligation with T4 DNA ligase. IAV bidirectional rescue plasmids (pDZ) have been described previously [110]. Lentiviral constructs for shRNA knockdown of chicken ZAP were cloned by annealing complementary oligos with AgeI and EcoRI restriction enzyme compatible ends and subsequently ligated into pLKO.2 via Golden-Gate ligation with T4 DNA ligase. shRNAs were designed targeting chicken ZC3HAV1 using the Broad Institute’s Genetic Perturbation Platform [111]. DNA constructs were sequence confirmed by whole plasmid sequencing performed by Plasmidsaurus using Oxford Nanopore Technology or Sanger sequencing performed by Eurofins Scientific or GeneWiz from Azenta. shRNA, siRNA, and sgRNA sequences are provided in S1 Table.

Retroviral transductions and stable cell generation

Simple retroviruses for transduction were generated by transfecting HEK293T cells with 1 µg vector plasmid [100,104], 1 µg pMD-gagpol [112], and 200 ng pMD-VSVG [113] using Transit-LT1 (Mirus Bio), changing media after 24 hours, and harvesting culture supernatant at 48 hours post-transfection followed by 0.45 µm syringe filtration. Lentiviruses for transduction were generated by transfecting HEK293T with 1.25 µg pLentiCRISPR1000 or pLKO.2_shRNA plasmid, 750ng psPAX2 (Addgene plasmid #12260; encoding Gag/GagPol, Rev, Tat), and 200 ng pMD-VSVG, changing media after 24 hours, and harvesting culture supernatant at 48 hours post-transfection followed by 0.45 µm syringe filtration. Transducing virus preps were aliquoted and stored at −20 °C. Stable cells were generated as previously described [87,114]. Briefly, ~2,500 target cells were seeded into a 96-well flat bottom plate, allowed to adhere overnight, and 20–200 µL of transducing viral supernatant with up to 10 µg/mL polybrene added to each well. After 2–3 days in culture, transduced cells are washed with PBS, and fresh media containing antibiotic (2 µg/mL puromycin, 2 µg/mL blasticidin, or 200 µg/mL hygromycin; GoldBio, Inc) added. Cells were grown and expanded in the ongoing presence of antibiotic. In the case of CRISPR/Cas9 lentiviral transduction, A549 cells were single cell cloned by serial dilution (re-plating from pooled transducing culture as above into 96-well flat bottom plates at 25, 5, 1, and 0.2 cells per well or by gradient dilution down and across 96-well plates). After 2–4 days, single cell-derived colonies were marked and followed regularly, expanded, and then screened initially by western blot or PCR. Multiple independent clones were screened for each knockout. Genetic knockout in A549 cells was confirmed by western blot or genomic DNA extraction, amplification of Cas9 target site, blunt cloning into pJET1.2 (Thermo Scientific), and Sanger sequencing multiple transformed clones. Double knockout cells were transduced simultaneously (A549.ZK1 and A549.ZK2) with ZAP-targeting (blasticidin-resistant) and KHNYN-targeting (puromycin-resistant) CRISPR/Cas9 lentiviruses, selected with both antibiotics after 48 hours, single cell cloned, screened by western blot or PCR, and confirmed by Sanger sequencing. Avian cells were transduced with simple retrovirus Cas9-GFP (puromycin-resistant). Avian cell knockouts were maintained as pools stably expressing Cas9 and sgRNAs (blasticidin-resistant). Genetic knockout in avian cells was determined by genomic DNA extraction, PCR amplification of Cas9 target site, Sanger sequencing of pooled amplicons, and computational deconvolution of overlapping chromatograms by TIDE analysis [115].

Influenza viruses

Wild-type Influenza A/Puerto Rico/8/1934(H1N1) (PR8) was rescued by plasmid-based transfection into HEK293T cells using Lipofectamine 3000 (ThermoFisher Scientific) and amplified in 10-day old embryonated chicken eggs. PR8_CG virus was made by plasmid-based transfection with the WT segment 5 plasmid replaced with CpG-enriched segment 5 and amplified in embryonated chicken eggs. Single-cycle PR8 and Cal04 (A/California/04/2009(H1N1)) that lack virus-encoded HA and instead express mCherry from the HA segment were rescued by plasmid-based transfection into HEK293T cells and amplified in MDCK cells stably expressing PR8 HA [116]. Chimeric OH175 Influenza A/Green-Winged Teal/Ohio/175/1986(H2N1) viruses [67] were rescued by plasmid-based transfection into HEK293T cells using six internal segment plasmids from OH175 as well as the HA and NA segment plasmids from PR8 for biosafety reasons and subsequently amplified in MDCK cells. PR8, PR8_CG, and OH175 transfections were performed in combination with pCAGGS expression plasmids as needed. OH175 plasmids (also known as S009) were a kind gift of Andrew Mehle. Viral titers were determined by plaque assay in MDCK cells as previously described [117].

Influenza multicycle replication assays

For multicycle IAV infections, 500,000 A549 or DF-1 cells were plated in 6-well plates (or 100,000 CCL-141 in 12-well) and allowed to grow overnight for up to 24 hours. Cells were infected at an MOI of 0.05 infecting a confluent culture of cells, as previously described [117]. For each cell condition (e.g., DF-1 Vector, ZAP, and KHNYN) and for each virus (e.g., PR8 or PR8_CG) there were three independently infected well replicates (with the exception of multicycle replication assays in S6A Fig using two independently infected well replicates). Briefly, cells were washed with PBS prior to addition of 1mL infection media (1× PBS, 2.5% bovine serum albumin, 1% calcium/magnesium). Viral stocks were diluted such that 40 µL of inoculum could be added to each well yielding an MOI of 0.05 plaque-forming units per cell (pfu/cell). After the addition of viral inoculum, cells were incubated at 37 °C for 1 hour. Next, cells were washed with PBS prior to addition of 2.5 mL viral growth medium (1x DMEM, 2.5% HEPES buffer, 2.5% bovine serum fraction V [7.5%], 1% penicillin/streptomycin) supplemented with TPCK-resistant trypsin (1 µg/mL for A549 as well as 0.625 µg/mL for DF-1 and CCL-141). Culture supernatants (at least two aliquots of 200 µL each) were collected immediately after addition to wells (0 hours post-infection) and at 24- and 48-hours post-infection (hpi) and optionally at 12-hours post-infection. Supernatants were stored at −80 °C until viral titer was determined by plaque assay in MDCK cells as previously described [117]. Multicycle replication assays were performed in triplicate with key results being replicated in independent assays (e.g., ZAP-S relative to Vector in Fig 2A, 2B, and 2C). Plaque assays were performed and quantified single-blinded. For transient knockdown by siRNA, 100,000 CCL-141 cells were plate in 12-well plates and transfected with mock, 5 nM scramble control, and 1 nM or 10 nM pools of 10 duplex siRNAs targeting Anas platyrhynchus ZC3HAV1 (aplZAP) using RNAiMAX (ThermoFisher). siRNAs were designed using IDT dsiRNA design tool and sequences are provided in S1 Table.

Single-cycle retrovirus assembly and infectivity assay

For ROSV, 500,000 DF-1 cells were plated into 6-well plates in 2 mL media, cultured overnight, and transfected with 1.6 µg RCAS-mCherry, 200 ng pMD-VSVG, 100/200/400 ng gene of interest plasmid, and topped up to 2.2 µg total plasmid DNA mass with pBlueScript using TransIT-LT1 (Mirus Bio). At 24 hours post-transfection, we exchanged media. At 48 hours post-transfection, virus-containing supernatants were harvested and filtered using 3 mL syringes and 0.45 µm syringe PVDF filters. Transfections were performed as dose-titrations of gene of interest plasmid amounts rather than triplicates of a single plasmid amount. Culture supernatants were used to infect chicken DT40 suspension cells in triplicate and the remaining volume was stored at −80 °C. We performed flow cytometry to detect infected (i.e., % mCherry-positive) DT40 cells at 48 hours post-infection using a Becton Dickinson FACS Canto II with a high-throughput plate-based sample aspirator. For MLV, HEK293T cells were plated and transfected with 500 ng pMD-gagpol, 1,100 ng pMLV-Gag-HA-mCherry, 200 ng pMD-VSVG, 100/200/400 ng gene of interest plasmid, and topped up to 2.2 µg total plasmid DNA mass with pBlueScript. For HIV-1, HEK293T cells were plated and transfected with 1,600 ng pHIV-mCherry, 200 ng glycoprotein-expression plasmid (pMD2-VSVG, pCAG-RABV-G, or pcDNA.HIV.SynEnv), 100/200/400 ng gene of interest plasmid, and topped up to 2.2 µg total plasmid DNA mass with pBlueScript. MLV single-cycle infectivity was detected by infecting HEK239T cells and HIV-1 infectivity was detected using SupT11 cells (for VSV-G and HIV Env pseudo-typed virus) or HEK293T cells (for RABV-G pseudo-typed virus).

Western blotting and antibodies

A549 cells were treated with 1000 U/mL of universal interferon-alpha (PBL Assay Sciences, #11200-2) for 24 hours prior to pelleting cells and lysing in 2.5× reducing sample buffer (RSB; 125 mM Tris-HCl, 20% glycerol, 7.5% SDS, 5% β-mercaptoethanol, 250 mM DTT, 0.05% Orange G, pH 6.8), and boiled at 98 °C for 15 min prior to storage at −80 °C. Transfected DF-1 and HEK293T cells for single-cycle retrovirus assembly assays were harvested at 48 hours post-transfections, pelleted by centrifugation, lysed in 2.5× reducing sample buffer (RSB), and boiled at 98 °C for 15 min prior to storage at −80 °C. Virus-containing supernatants (1 mL) from retrovirus assays were underlaid with 200 µL of 20% sucrose (diluted in 1× PBS), centrifugated at ≥21,500g for two hours at 4 °C, liquid aspirated, and pellet resuspended in 2.5× RSB. Western blots were performed conventionally using commercial precast polyacrylamide gels (26-well Bio-Rad, 4%–20% Criterion TGX, #5671095) for electrophoresis (60 volts for 60 min followed by 120 volts for 60 min) and transferred to 0.2 µm nitrocellulose membrane using Invitrogen Power Blotter XL (ThermoFisher Scientific; ≤25 volts). Membranes were washed in PBS and PBST (PBS with 0.1% Tween-20) followed by blocking (2% dry milk in PBST). With constant rocking at all steps, membranes were incubated with primary antibodies overnight at 4 °C, followed by three washes for 5 min each with PBST, incubation with secondary antibodies in 2% milk in PBST for at least 1 hour at room temperature, followed by three washes for 5 min each in PBST, and one final wash in PBS. Primary antibodies used at indicated dilutions are provided in S1 Table. Secondary antibodies used include goat anti-mouse IgG IRDye 680LT and goat anti-rabbit IgG IRDye 800CW (both used at 1:10,000 dilution in 2% milk in PBST). Membranes were imaged using a LiCor Odyssey Fc infrared fluorescent imager for 2 min or 10 min per channel.

GFP-Trap immunoprecipitation

Chicken DF-1 cells including those stably expressing YFP-tagged ZAP and KHNYN orthologues were infected with PR8 and PR8_CG at MOI = 5 as described above in 10 cm dishes and harvested at 24 hours post-infection. A portion of each sample set aside for 10% input RNA (with RNase Inhibitor) and 10% input protein (resuspended in RSB and boiled), then stored at −80 °C. Remaining cells were washed twice with cold 1× PBS (containing RNase inhibitor) and resuspended in radioimmunoprecipitation assay buffer (RIPA: 10 mM Tris-HCl, 150 mM NaCl, 0.5 mM EDTA, 0.1% SDS, 1% Triton-X, 1% sodium deoxycholate, 2.5 mM MgCl₂, 1 mM phenylmethylsulfonyl fluoride, 1 unit/mL DNaseI, and 1 unit/mL RNase Inhibitor (Roche, #3335399001). Samples were needle-lysed with a 26G needle, slowly inverted continuously at 4 °C for 30 min, and centrifugated at ≥21,500g for 10 min at 4 °C. GFP-Trap agarose beads (ProteinTech, #gta) were equilibrated in dilution buffer (10 mM Tris-HCl, 150 mM NaCl, 0.5 mM EDTA) with 50 µL bead slurry per sample into 500 µL dilution buffer per sample. Beads were centrifugated at 2,500g for 5 min at 4 °C and washed with dilution buffer twice more. Remaining RIPA supernatant of each sample was added to 500 µL of equilibrated beads in dilution buffer and slowly inverted continuously at 4 °C overnight. Beads were centrifugated at 2,500g for 5 min at 4 °C and resuspended in wash buffer (10 mM Tris-HCl, 150 mM NaCl, 0.5 mM EDTA, 0.05% IGEPAL) twice and then split into two equivalent volume samples. One subjected to RNA extraction by Qiagen RNeasy Kit. The other centrifugated once more, supernatant aspirated, bead pellet resuspended in RSB, boiled at 98 °C for 15 min, and stored at −80 °C.

Fluorescence microscopy

Fixed-cell images were obtained on a BioTek Cytation 5 (Agilent) using a 4× (NA = 0.13, #1220519), 20× (NA = 0.40, #1220517), or 40× (NA = 0.6, #1220544) Plan Fluorite objective and LED-filter cube sets for DAPI (excitation = 377 ± 50 nm; emission = 447 ± 60 nm; #1225007 and #1225100), GFP for mNeonGreen (excitation = 469 ± 35; emission = 525 ± 39; #1225001 and #1225101), and TRITC for mCherry and AlexaFluor568 (excitation = 556 ± 29; emission = 600 ± 37; #1225012 and #1225125). We quantified single-cycle IAV infections by imaging and used the BioTek Gen5 software to calculate percent infectivity using scPR8 (MOI = 5) and scCal04 (MOI = 5) encoded mCherry as a readout for infectivity (# TRITC+ cells/ # DAPI-stained nuclei). To quantify OH175 nonreplicating infections, we performed immunofluorescence on infected cells (MOI = 0.5 without TPCK trypsin to block virus transmission) fixed with 4% paraformaldehyde at 24 hours post-infection. Briefly, we subjected cells to permeabilization (PBS with 0.25% Triton-X-100) for 10 min, blocking (PBS with 3% bovine serum albumin), primary antibody staining with mouse anti-NP (BEI, NR-43899, 1:500) in blocking buffer (3% bovine serum albumin) for 1 hour, secondary antibody staining with goat anti-rabbit AlexaFlour-568 (Invitrogen, #A-11031, 1:500), and nuclear counterstaining with DAPI (Sigma, D9542, 100 ng/mL).

Minigenome reporter assay

Chicken DF-1 cells were seeded into 96-well glass-bottom plates, allowed to adhere overnight, and then transfected with equal amounts of plasmids encoding IAV heterotrimeric polymerase (pCAGGS.PB2, pCAGGS.PB1, and pCAGGS.PA), wild-type IAV NP (pCAGGS.NP), and a vRNA/negative-sense IAV-codon-optimized mCherry expression plasmid (pPol.mCherry) [118] as well as plasmids encoding mNeongreen or mNeongreen-tagged mammalian KHNYN orthologues with TransIT-LT1. At 24 hours, cells were washed with 1× PBS, fixed with 4% paraformaldehyde, counterstained with DAPI, and imaged for mNeongreen and mCherry expression as a % mCherry-positive as described above.

Structural models of ZAP and KHNYN

For structural models of ZAP, amino acid sequences for the four zinc finger domains of different species and mutants of ZAP were used as input for RobettaCM using the human ZAP crystal structure (PDB: 6UEJ) for comparative modeling. We noted that in most cases, ab initio folding by AlphaFold or RoseTTAFold may not capture metal ion-coordinated finger structures. Sequences used are provided in S1 Table. Models generated and 6UEJ are shown in S1 Data. For structural models of KHNYN, amino acid sequences for the KH domain or NYN domain of KHNYN were used as input for RoseTTAFold using default settings. Sequences used are shown in S1 Table. Both RoseTTAFold and RobettaCM are publicly available at robetta.bakerlab.org. All models generated are provided in S2 and S3 Data. Models were visualized in Chimera (v1.13.1, build 41965, UCSC) and aligned using MatchMaker with a Needleman-Wunsch alignment algorithm and BLOSUM-62 matrix.

Phylogenetic analyses

Coding sequences (CDS) of human KHNYN, NYNRIN, N4BP1, and their orthologues were downloaded from ensembl. CDS of ZAP homologs were downloaded from ensembl. CDS were aligned using ClustalOmega in Seaview (V5.0.4). Phylogenetic trees were generated in Seaview using PhyML with 100 bootstraps using CDS. Synteny schematics were generated manually in Adobe Illustrator using data from UCSC ENCODE Genome Browser, NCBI, and ensembl. Alignments and phylogenetic trees are available in S4 and S5 Data.

Virus bioinformatic analysis

Full-length influenza virus sequences were downloaded from the Influenza Research Database for each animal host species (human A, avian A, swine A, human B, and human C), including the nucleotide sequence, host, collection date, and subtype. Dinucleotide values were counted by string count. Data were plotted using ggplot2. Data are available at a zenodo record (https://doi.org/10.5281/zenodo.14919910) and R code is available in S1 File. Sliding window CpG data were analyzed using an in-house custom script (cpg_counts.R) available in S1 File.

RNA-seq

RNA was extracted from mock and infected samples at 24 hours post-infection (MOI = 0.05) using Qiagen RNeasy Kit. Samples were submitted for polyA-selected RNA (mRNA) sequencing using an Illumina NovaSeq 6000 with 150 bp paired end reads at the University of Illinois Urbana-Champaign Roy J. Carver Biotechnology Center. Reads were trimmed of adapter sequences and low-quality reads eliminated using Trimmomatic (V0.32). Trimmed reads were mapped to a hybrid reference genome (Gallus gallus GRCg7b, PR8, and PR8_CG) using STAR (V2.7.11b) and quantified per feature using HTSeq (V2.0). Differential expression analysis was performed using DESeq2 (v1.34.0). Data were processed and analyzed in R-Studio version 1.3.1073 using R language version 3.6.3. DESeq2 R code is available in S2 File. RNA-seq data are available at the NCBI Sequence Read Archive (SRA) under accession code PRJNA1208514.

Quantitative reverse-transcription PCR

To quantify endogenous chicken ZC3HAV1 and TUBA4A mRNA transcripts in shRNA-expressing stable cells following chicken interferon-alpha treatment (1000 U/mL; Bio-Rad) for 24 hours, 500,000 cells were plated in 6-well plates and allowed to adhere overnight in culture media followed by addition of chicken interferon or vehicle. After 24 hours, cells were harvested, RNA extracted using Qiagen RNeasy kit, and cDNA synthesized used Qiagen Quantitect Reverse-Transcription Kit using supplied primer mix (includes random hexamers and oligo-dT primers) with genomic DNA wipeout according to manufacturer’s instructions. To quantify IAV segment 5 RNA in mock and infected chicken DF-1 cells (input) and GFP-Trap immunoprecipitated (IP) samples, RNA extraction and cDNA synthesis were performed as above. Quantitative PCR was performed in technical triplicate on a LightCycler 480 II (Roche Life Science) or QuantStudio5 (Applied Biosystems) using Ssofast Eva Green Supermix (Bio-Rad). Oligonucleotide primer sequences are provided in S1 Table.

Statistical analyses

Plaque assay results from IAV infections were plotted in GraphPad Prism 10 and one-way ANOVA was used at 48 hours post-infection or two-way ANOVA for multiple time points to determine statistical significance each with Dunnett’s test for correcting for multiple comparisons. Single-cycle ROSV and IAV infectivity results were plotted in GraphPad Prism 10. Unpaired t tests were used to calculate statistical significance for IAV single-cycle infectivity assays with Holm-Šidák’s test correcting for multiple comparisons. Mean and standard deviation of the mean were calculated in GraphPad Prism 10. qPCR calculations were performed in Microsoft Excel. Statistical analyses and data values underlying graphs are available in S2 Table or in a data repository at https://doi.org/10.5281/zenodo.14919910.