Mammalian ANP32A and ANP32B Proteins Drive Differential Polymerase Adaptations in Avian Influenza Virus

Abstract

ANP32 proteins, which act as influenza polymerase cofactors, vary between birds and mammals. In mammals, ANP32A and ANP32B have been reported to serve essential but redundant roles to support influenza polymerase activity. The well-known mammalian adaptation PB2-E627K enables influenza polymerase to use mammalian ANP32 proteins. However, some mammalian-adapted influenza viruses do not harbor this substitution. Here, we show that alternative PB2 adaptations, Q591R and D701N, also allow influenza polymerase to use mammalian ANP32 proteins, whereas other PB2 mutations, G158E, T271A, and D740N, increase polymerase activity in the presence of avian ANP32 proteins as well. Furthermore, PB2-E627K strongly favors use of mammalian ANP32B proteins, whereas D701N shows no such bias. Accordingly, PB2-E627K adaptation emerges in species with strong pro-viral ANP32B proteins, such as humans and mice, while D701N is more commonly seen in isolates from swine, dogs, and horses, where ANP32A proteins are the preferred cofactor. Using an experimental evolution approach, we show that the passage of viruses containing avian polymerases in human cells drove acquisition of PB2-E627K, but not in the absence of ANP32B. Finally, we show that the strong pro-viral support of ANP32B for PB2-E627K maps to the low-complexity acidic region (LCAR) tail of ANP32B. IMPORTANCE Influenza viruses naturally reside in wild aquatic birds. However, the high mutation rate of influenza viruses allows them to rapidly and frequently adapt to new hosts, including mammals. Viruses that succeed in these zoonotic jumps pose a pandemic threat whereby the virus adapts sufficiently to efficiently transmit human-to-human. The influenza virus polymerase is central to viral replication and restriction of polymerase activity is a major barrier to species jumps. ANP32 proteins are essential for influenza polymerase activity. In this study, we describe how avian influenza viruses can adapt in several different ways to use mammalian ANP32 proteins. We further show that differences between mammalian ANP32 proteins can select different adaptive changes and are responsible for some of the typical mutations that arise in mammalian-adapted influenza polymerases. These different adaptive mutations may determine the relative zoonotic potential of influenza viruses and thus help assess their pandemic risk.

Keywords: ANP32; ANP32A; ANP32B; avian influenza; canine influenza; equine influenza; host factors; influenza; pandemic; pandemic influenza; swine influenza; zoonotic.

FIG 1

PB2 Q591R, E627K, and D701N specifically adapt influenza virus polymerase to human ANP32 proteins. Minigenome assays performed in wild-type (WT) human engineered-haploid (eHAP) cells (A) or WT chicken DF-1 cells (B) with avian 50-92 polymerase bearing different mammalian adaptations. (C) Western blot of mutant PB2 expression. Minigenome assays performed in human eHAP dKO cells (D to F) or chicken DF-1 AKO cells (G to I) with avian 50-92 polymerase bearing different mammalian adaptations transfected in along with different chicken or human ANP32 proteins. Data are representative of triplicate repeats (n = 3) and are plotted in triplicate. Data are plotted as means and standard deviation normalized to PB2 WT. Statistical significance was determined by one-way analysis of variance (ANOVA) with multiple comparisons against WT on log-transformed data. Log-normality of data were confirmed by Shapiro-Wilk test of normality. *, 0.05 ≥ P > 0.01; **, 0.01 ≥ P > 0.001; ***, 0.001 ≥ P > 0.0001; ****, P ≤ 0.0001.

FIG 2

Different species’ ANP32 proteins display different patterns of dominance for influenza polymerase. (A) Minigenome assays performed in human eHAP dKO cells with a typical human influenza polymerase (A/England/687/2010 [pH1N1]), co-transfected with different mammalian ANP32 proteins expressed. Statistical significance was determined by multiple t tests between different species’ ANP32A and ANP32B proteins. (B) Western blot of ANP32 proteins as shown in panel A. (C and D) Minigenome assays performed in human eHAP dKO cells with different human or dog ANP32B mutants. Data are representative of triplicate repeats (n = 3) and are plotted in triplicate. Data plotted as means and standard deviation. Statistical significance was determined by one-way ANOVA with multiple comparisons, statistical tests performed between WT and mutant ANP32 proteins. *, 0.05 ≥ P > 0.01; **, 0.01 ≥ P > 0.001; ***, 0.001 ≥ P > 0.0001; ****, P ≤ 0.0001. (E) Western blot analysis of mutant ANP32 proteins as shown in panel C and D.

FIG 3

PB2 E627K shows a greater preference than Q591R or D701N for using mammalian ANP32B proteins. Minigenome assays performed in human eHAP dKO cells (A to D) or avian DF-1 AKO cells (E to H) with avian 50-92 polymerase bearing different mammalian adaptations transfected in along with different avian or mammalian ANP32A (blue bars) or ANP32B proteins (orange bars). Statistical significance was determined by one-way ANOVA with multiple comparisons, statistical tests without comparison bars indicate a comparison against empty vector and between ANP32A and ANP32B proteins from the same species. *, 0.05 ≥ P > 0.01; **, 0.01 ≥ P > 0.001; ***, 0.001 ≥ P > 0.0001; ****, P ≤ 0.0001. (I) Western blot of different ANP32 proteins used in panels A to H. (J) Minigenome assays performed in human eHAP dKO cells with chicken ANP32A, or human ANP32A or ANP32B, titrated in. Data are representative of triplicate repeats (n = 3) and are plotted in triplicate. Panels A to H replotted from same representative repeat as in Fig. 1D to I. Data plotted throughout as means and standard deviation. Statistical significance was determined by multiple t tests between human ANP32A and ANP32B, statistical tests performed between WT and mutant ANP32 proteins. *, 0.05 ≥ P > 0.01; **, 0.01 ≥ P > 0.001; ***, 0.001 ≥ P > 0.0001; ****, P ≤ 0.0001. (K) Western blot of highest concentrations used in ANP32 titration as shown in panel J.

FIG 4

Mammalian species with dominantly pro-viral ANP32B are associated with E627K. Mammalian adaptations seen in avian-origin viruses during (A) zoonotic or likely dead-end cross-species infections/mouse passage experiments (middle panel only) and (B) stable circulation and prolonged adaptation to a mammalian host. Human and swine cross-species infections (panel A, left and right) were calculated by downloading all non-H1-3 human/swine influenza virus strain PB2s from NCBI, performing an alignment, curating out any seasonal human influenza segments, and looking at the identities of positions 591, 627, and 701. Mouse adaptation studies (panel A, middle) were calculated by performing a literature review of any influenza mouse adaptation studies using avian-origin influenza virus without any prior mammalian adaptation. Only virus strains with good evidence for having stably circulated in their respective mammalian species were included in the timeline (B). pH1N1/pH2N2/pH3N2 are represented by the same line because they contained the same PB2 originally from the spillover which caused the 1918 virus, reassorted into each new pandemic virus. Swine H9N2 (*) was included because phylogenetic and molecular evidence strongly suggested that although this virus appears to potentially co-circulate in both swine and chickens, it shows some clear mammalian adaptation markers.

FIG 5

Experimental evolution of an avian influenza virus in human cells abrogated for ANP32B does not lead to the PB2 E627K adaptation. Sequencing summary of reassortant PR8 viruses containing the polymerases constellations from avian origin 50-92 (A) or Anhui (B) in human eHAP cells ablated for ANP32A (AKO) or ANP32B (BKO) or control cells (con). Each pie chart indicates 6 independently passaged populations. Depth of color indicates rough estimation of the proportion of the population: light shades indicate a mixed population of WT and indicated residues at a position, while darker shades indicate a complete change. Striped bars indicate a mix of both E627K and D701N. Gray slices indicate that no changes were detectible at positions 591, 627, or 701.

FIG 6

Differences in the low-complexity acidic region (LCAR) of human ANP32A and ANP32B are responsible for the preference of PB2 E627K viruses for ANP32B. (A) Non-aligned comparison of human ANP32A and ANP32B sequences, with sites of interest indicated in red. (B) Minigenome assays performed in human eHAP dKO cells with avian 50-92 or Bavaria polymerase with different mammalian adaptations transfected in along with different chimeric human ANP32 protein. Data are representative of triplicate repeats (n = 3) and are plotted in triplicate. Data are plotted as means and standard deviation. Statistical significance was determined by one-way ANOVA with multiple comparisons, statistical tests without comparison bars indicate a comparison against empty vector and between the different ANP32 proteins. ****, P ≤ 0.0001. (C) Western blot of chimeric human ANP32 proteins used in panel B.

FIG 7

Model of how ANP32 protein dominance in different species may influence mammalian adaptation. Summary model of the data from this study showing how avian-origin influenza viruses adapt to the dominant pro-viral ANP32 protein in a new host and gain different adaptive mutations in PB2. Size of circles in each host indicate the relative pro-viral ability of its ANP32 proteins.