Species difference in ANP32A underlies influenza A virus polymerase host restriction

Abstract

Influenza pandemics occur unpredictably when zoonotic influenza viruses with novel antigenicity acquire the ability to transmit amongst humans. Host range breaches are limited by incompatibilities between avian virus components and the human host. Barriers include receptor preference, virion stability and poor activity of the avian virus RNA-dependent RNA polymerase in human cells. Mutants of the heterotrimeric viral polymerase components, particularly PB2 protein, are selected during mammalian adaptation, but their mode of action is unknown. We show that a species-specific difference in host protein ANP32A accounts for the suboptimal function of avian virus polymerase in mammalian cells. Avian ANP32A possesses an additional 33 amino acids between the leucine-rich repeats and carboxy-terminal low-complexity acidic region domains. In mammalian cells, avian ANP32A rescued the suboptimal function of avian virus polymerase to levels similar to mammalian-adapted polymerase. Deletion of the avian-specific sequence from chicken ANP32A abrogated this activity, whereas its insertion into human ANP32A, or closely related ANP32B, supported avian virus polymerase function. Substitutions, such as PB2(E627K), were rapidly selected upon infection of humans with avian H5N1 or H7N9 influenza viruses, adapting the viral polymerase for the shorter mammalian ANP32A. Thus ANP32A represents an essential host partner co-opted to support influenza virus replication and is a candidate host target for novel antivirals.

Extended Data Figure 1.. Analysis of mRNAs by PCA mapping reveals diversity of the radiation hybrid clones and their genetic instability during cell passage

Each sphere represents a microarray sample. The percentage values in the axes parentheses designate proportion of overall variance as described by each PC. PC1 principal component 1 (X-axis); PC2 principal component 2 (Y-axis); PC3 principal component 3 (Z-axis). PC1 describes the predominant amount of variance (15.6%). Selection of negative clones (red), parent Wg3H cells (blue) and positive clones: 377 (purple), 386 (orange) and 365 (green) and 476 (cyan) arrays are distinguished by colour, and passage numbers 1 and 12 are distinguished by the size of spheres. Negative arrays are dispersed, while parent cells are accumulated further to the right of PC1 and upwards of PC2. Positive clones show distinct variability in their location while passaging reduced their separation from parent cells. This analysis accompanies Figure 1.

Extended Data Figure 2.. Confirmation of chANP32A and chANP32B expression in RH clones by qRT-PCR

RNA was extracted from the RH clones after testing for influenza polymerase activity and analysed by microarray for chicken transcripts. The same RNA was used to validate identification of ANP32A by confirming the level of expression of ANP32A (and ANP32B as control) in the parent Wg3h cells, positive clones, passaged positive clones and a selection of negative clones. a, Copy numbers of chANP32A mRNA were calculated by qRT-PCR against a standard curve generated with chANP32A cDNA using primers specific for chANP32A. b, Copy number of ANP32B mRNA were measured by qRT-PCR against a standard curve generated with chANP32B cDNA using primers specific for chANP32B. This analysis accompanies Figure 1.

Extended Data Figure 3.. Knockdown of chANP32A in positive RH clone 476 diminished the ability to support avian influenza polymerase activity

a, Positive RH clone 476 cells were transfected with 100nM of siRNA targeting NP, chANP32A or no target (Allstars). After 48hrs cells were transfected with mouse-polI-firefly minigenome reporter, avian influenza polymerase (H5N1 50-92) with either PB2 627E or 627K, Renilla control and either empty plasmid or codon optimised chANP32A (codon optimization according to algorithm by Geneart with manual editing). Luciferase activity was measured after a further 24hrs. (Errors are displayed as SEM; n=3 biological replicates). b, Knockdown of chANP32A was confirmed by qRT-PCR of RNA extracted from siRNA treated cells, calculated using a standard curve generated with chANP32A cDNA, using primers specific for chANP32A (Errors are displayed as SEM; n=3 biological replicates). This analysis accompanies Figure 1.

Extended Data Figure 4.. Expression of chANP32A in human cells permits influenza polymerase activity of several avian influenza polymerases and an avianised human influenza polymerase and increases avian virus replication

293T cells were transfected with empty vector, chANP32A or huANP32A. a&b. 20hrs later, cells were transfected with pHOM1-firefly minigenome reporter, and the polymerase set from low pathogenicity avH1N1 (Bav) or H9N2 (UDL), highly pathogenic H5N1 (50-92), H5N1 (Ty05), or huH3N2 (Vic) viruses, with either PB2 627E (a) or 627K (b) and Renilla expression control. After a further 24hrs luciferase activity was measured. (Data are mean PB2 627E or K polymerase activity normalised to Renilla; error plotted as SEM of the ratio; n=3 biological replicates; pattern of results consistent in at least three independent experiments). This analysis accompanies Figure 1. c. 20hrs after transfection with ch or huANP32A or empty vector, cells were infected with avian-like influenza virus (H5N1Ty05:PR8 6:2 recombinant virus) (MOI 0.1) bearing PB2 627E (black bars) or PB2 627K (grey bars). Infected cells were incubated at 37°C and cell supernatant titrated for infectious virus at 24hrs post infection on MDCK cells by plaque assay.(Data displayed as log₁₀ plaque forming units/ml; One-way ANOVA, comparisons to empty vector, ns= not significant, *p<0.05 ***p<0.001, n=3 biological replicates, pattern of results consistent in at least three independent experiments). This analysis accompanies Figure 2.

Extended Data Figure 5.. chANP32A does not alter expression or nuclear accumulation of avian PB2 protein in human cells

293T cells were transfected with pHOM1-firefly minigenome, avian influenza polymerase and NP of H5N1 50-92 (PB2 627E) together with empty vector, chANP32A or chANP32AΔ33 or cells were untransfected (Mock). Cell monolayers were harvested after 24hrs and lysed in 0.1% NP40 lysate buffer and total fractions taken before centrifugation to generate a nuclear pellet and cytoplasmic fraction. Nuclear pellets were resuspended in 1% NP40 buffer. a. Protein levels of vinculin (cytoplasmic marker) and lamin B (nuclear marker) and of PB2 in total, nuclear or cytoplasmic fractions were analysed by immunoblotting. b. Total lysates were immunoblotted for vinculin, PB2 and FLAG peptide. c. Immunoblots were quantified using Image Studio Lite V5.2. The ratio of nuclear to cytoplasmic PB2 was calculated by dividing the ratio of PB2 to vinculin by the ratio of PB2 to lamin B from the cytoplasmic and nuclear fractions respectively. Data are the mean ratios from three independent experiments (excepting chANP32Δ33 for which only 2 data points were available), error bars are SEM. Data are not statistically significantly different by One-way ANOVA. This analysis accompanies Figure 2.

Extended Data Figure 6.. Quantification of knockdown of chANP32A in chicken cells

DF-1 cells were transduced with VSV-G lentiviral vectors that delivered a transgene expressing shRNA directed against chANP32A or a negative sequence and the puromycin gene. Puromycin selected cells were transfected with siRNA (100nM) (underlined). RNA was extracted from untreated shRNA cells and siRNA-treated shRNA cells. Knockdown of chANP32A was quantified by qRT-PCR of the extracted RNA, calculated using a standard curve generated with chANP32A cDNA, using primers specific for chANP32A. Fold decrease of RNA copies is displayed as compared to Negative shRNA DF-1 or ALLstars treated chANP32A shRNA DF-1 cells. (Error displayed as SEM; n=3 biological replicates). This analysis accompanies experiments in Figure 3a-c.

Extended Data Figure 7.. siRNA knockdown demonstrates that human adapted influenza polymerase activity is dependent on huANP32A and huANP32B in human cells

a, 293T cells were transfected with siRNA (100nM) against NP, huANP32A, huANP32B or both huANP32A and huANP32B (50nM each). After 48hrs, cells were transfected with pHOM1-firefly minigenome, human-adapted avian influenza polymerase (H5N1 50-92 PB2 627K), and Renilla expression control. Luciferase activity was measured after a further 24hrs. (Data are firefly activity normalised to Renilla, plotted as % of Allstars; error as SEM; One-way ANOVA comparisons to Allstars, ****p<0.0001; n=3 biological replicates). b, Knockdown of gene targets was verified by immunoblotting using antibody against vinculin, huANP32A and huANP32B. This analysis accompanies Figure 3 e and f.

Extended Data Figure 8.. Alignment of ANP32A proteins reveals significant homology except for an extra 33aa sequence in birds that is absent in mammals and ostrich and lacking from ANP32B family members

The protein sequences of ANP32A for chicken, duck, zebra finch, turkey, ostrich, human, mouse and pig together with sequences of ANP32B for chicken and human were aligned using Geneious R6 software. chANP32A is set as the reference sequence, and colours represent similarity of amino acid identity (Black=100%, dark grey=80-100%, light grey=60-80%, white=<60%). Gaps are annotated by dashes. Residue numbers correspond to chANP32A. The 33aa sequence found in avian species is situated between residues 176-208.This analysis accompanies Figure 4.

Extended Data Figure 9.. Expression of ANP32A and B proteins reduced human-adapted influenza polymerase activity in human cells

293T cells were transfected with FLAG-tagged ANP32 constructs and after 20hrs transfected with pHOM1-firefly minigenome reporter, human-adapted influenza polymerase (H5N1 50-92 with PB2 627K, together with Renilla expression control. Cells were assayed for luciferase activity 24hrs later. (Data are PB2 627K polymerase activity normalised to Renilla; One-way ANOVA, all constructs were significantly reduced compared to empty vector (p<0.0001); error plotted as SEM of the ratio; n=3 biological replicates; pattern of results consistent in at least three independent experiments).These data relate to Figure 4.

Figure 1. Identification of a positive avian cellular factor that permits avian influenza virus polymerase activity in mammalian cells by screening radiation hybrid clones

a, RH clones transfected with mouse-polI-firefly minigenome reporter, avian influenza virus H5N1 50-92 polymerase (either PB2 627E or PB2 627K) and Renilla expression control. Passage 1 clones black bars, Passage 12 clones grey bars. (Data as a ratio of PB2 627E/K polymerase activity (firefly normalised to Renilla); n=3 biological replicates; one-way ANOVA, comparisons to Wg3h, ****p<0.0001; error as SEM of the ratio). b, VENN diagram of microarray data analysis of P1 positive hybrid clones vs Parent Wg3h cells. c, VENN diagram of P12 positive hybrid clones vs parent and 365P1 (positive RH clone) vs 365P12 (reverted RH clone) (total gene numbers in brackets); Two-way ANOVA (variables: clone and passage number) adjusted by Benjamini–Hochberg multiple-testing correction (false discovery rate (FDR) of P<0.05). Statistically significant genes identified and those with fold-change values <±1.5 removed. d, Cloned chicken genes from chromosome 10 or chicken homologues of genes previously implicated with PB2 host range (underlined) expressed in 293T cells for 20hrs before transfection with pHOM1-firefly minigenome reporter, avian virus 50-92 polymerase (either PB2 627E (Black bars) or PB2 627K (grey bars)) and Renilla expression control. Luciferase activity was assayed after a further 24hrs. (Data are firefly activity normalised to Renilla; error as SEM; n=3 biological replicates; pattern of results consistent in at least three independent experiments).

Figure 2. Expression of chANP32A in human cells rescues transcription and replication of avian influenza virus

a-c, ch or huANP32A, ANP32B or empty vector were expressed for 20 hrs before transfection of avian virus 50-92 polymerase (either PB2 627E (black bars) or PB2 627K (grey bars)), and incubated at 37°C for 24hrs before qRT-PCR for luciferase gene of the viral minigenome reporter, vRNA (a), mRNA (b) and cRNA (c). d-f, 293T cells expressing chANP32A, huANP32A or empty vector for 20hrs before infection by avian influenza virus (H9N2 A/chicken/UDL-01/2008) virus (MOI 1.0), incubated at 37°C for 24hrs before qRT-PCR for viral segment 1; vRNA (d), mRNA (e) and cRNA (f). (Data expressed as fold change to empty vector, normalised to 18sRNA, calculated by ΔΔCT; error plotted as SEM; One-way ANOVA, comparisons to empty, ns= not significant, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001; n=3 biological replicates, pattern of results consistent in at least three independent experiments). g, 293T cells expressing ch or huANP32A or empty vector for 20hrs before infection (MOI 0.1) with avian-like influenza virus (recombinant PR8 virus with H5N1 Ty05 polymerase, M and NS gene segments and PR8 HA and NA genes) bearing PB2 627E (black bars) or PB2 627K (grey bars). Cells incubated at 33°C and cell supernatant titrated for infectious virus 24hrs post infection on MDCK cells by plaque assay. (Data displayed as log₁₀ plaque forming units/ml; One-way ANOVA, comparisons to empty vector, ns= not significant, **p<0.01 ****p<0.0001, n=3 biological replicates, pattern of results consistent in at least three independent experiments).

Figure 3. Knockdown of ANP32 reveals avian influenza polymerase dependence on chANP32A and dependence on huANP32A and B in human cells by human adapted influenza polymerase

a, DF-1 cells transduced with VSV-G lentiviral vectors delivering transgenes expressing puromycin and shRNA targeting chANP32A or Negative. Puromycin selected cells transfected with pCOM1-firefly minigenome reporter, avian 50-92 polymerase (627E), and Renilla expression control . b, siRNA (100nM) applied to DF-1 chANP32A shRNA cells. After 48hrs, cells transfected with avian 50-92 polymerase (627E), minigenome reporter and Renilla expression control. Luciferase activity measured 20hrs later. Knockdown in chicken DF-1 cells verified by immunoblotting using antibody against vinculin and chANP32A. c, DF-1 cells depleted of chANP32A by siRNA infected with avian-like influenza virus (PR8 virus bearing H5N1 Ty05 polymerase genes with PB2 627E, MOI 0.01). 24hrs later cell supernatants titrated for infectious virus by plaque assay on MDCK cells. d, 293T cells transduced with lentiviral vectors delivering transgenes expressing puromycin and shRNA targeting huANP32A, huANP32B, both huANP32A and B, or Negative. Puromycin selected cells transfected with, pHOM1-firefly minigenome reporter, human-adapted avian 50-92 polymerase (627K), and Renilla expression control.Luciferase activity measured after 20hrs. Knockdown in 293T cells verified immunoblotting using antibody against vinculin, huANP32A and huANP32B. (a,b&d, data are firefly activity normalised to Renilla, plotted as % of Negative or Allstars; error as SEM; One-way ANOVA comparisons to ALLStars or Negative, ns= not significant, **p<0.01, ***p<0.001, ****p<0.0001; n=3 biological replicates). e, Puromycin selected A549 cells expressing shRNA against huANP32 A and/or B were infected with human (H3N2) virus A/England/691/2010 (MOI 0.1). After 24hrs, cell supernatants titrated by plaque assay on MDCK cells. (c&e, Data are % pfu relative to ALLStars or Negative; One-way ANOVA, comparisons to ALLStars or Negative, ***p<0.001, ****p<0.0001, n=3 biological replicates, pattern of results consistent in at least three independent experiments).

Figure 4. Activity of avian influenza virus polymerase is dependent on a unique amino acid sequence present on avian ANP32A proteins

a, chANP32A protein schematic with 33aa insertion relative to human homologue. b, huANP32A schematic representative of ANP32A from mammals, ostrich and of ANP32B. c, 293T cells transfected with FLAG-tagged ANP32 and after 20hrs transfected with, pHOM1-firefly minigenome reporter, avian 50-92 polymerase (627E) and Renilla expression control . Luciferase activity measured 24hrs later. (Data are PB2 627E polymerase activity normalised to Renilla; One-way ANOVA, comparisons to Empty vector, ns= not significant, ****p<0.0001; error plotted as SEM of the ratio; n=3 biological replicates; pattern of results consistent in at least three independent experiments). d. Immunoblot analysis of FLAG-tagged ANP32A constructs using antibody against FLAG peptide and vinculin.