A rare variant in ANP32B impairs influenza virus replication in human cells

Viruses require host factors to support their replication, and genetic variation in such factors can affect susceptibility to infectious disease. Influenza virus replication in human cells relies on ANP32 proteins, which are involved in assembly of replication-competent dimeric influenza virus polymerase (FluPol) complexes. Here, we investigate naturally occurring single nucleotide variants (SNV) in the human Anp32A and Anp32B genes. We note that variant rs182096718 in Anp32B is found at a higher frequency than other variants in either gene. This SNV results in a D130A substitution in ANP32B, which is less able to support FluPol activity than wildtype ANP32B and binds FluPol with lower affinity. Interestingly, ANP32B-D130A exerts a dominant negative effect over wildtype ANP32B and interferes with the functionally redundant paralogue ANP32A. FluPol activity and virus replication are attenuated in CRISPR-edited cells expressing wildtype ANP32A and mutant ANP32B-D130A. We propose a model in which the D130A mutation impairs FluPol dimer formation, thus resulting in compromised replication. We suggest that both homozygous and heterozygous carriers of rs182096718 may have some genetic protection against influenza viruses.


INTRODUCTION 26
All viruses rely on host factors to support their replication. Genetic variation in such proteins 27 can affect susceptibility to infectious disease (; Ciancanelli, Abel, Zhang  Almost all the host genetic variation relevant to influenza that has been described is 60 associated with exacerbating rather than alleviating influenza virus infection, a bias 61 explained by the higher visibility of severe cases, and the opportunity to perform whole 62 5,032 in the gnomAD database is a heterozygous carrier of rs772530468 (ANP32A-S158A) 143 ( Figure 1B). 144  There was, however, an interesting exception. SNV rs182096718, encoding ANP32B-152 D130A, was relatively common in the Hispanic / Latino cohort of the gnomAD database, 153

Most variants in
where in a total pool of 35,420 individuals, 1,209 heterozygous carriers were identified, as 154 well as 25 homozygous carriers. The D130A substitution in ANP32B is thus present in 155 3.41% of the Latino cohort. The global MAF of this variant is 0.0044 which is much higher 156 than any other SNV in either Anp32A or Anp32B. Natural variants of ANP32 proteins affect support of FluPol activity 167 We have previously described a method to measure the capacity of mutated ANP32 proteins 168 to act as pro-viral factors for influenza polymerase activity. This is achieved by exogenous 169 expression of the cloned mutants in human eHAP cells lacking ANP32A and ANP32B (dKO), 170 in which influenza polymerase is unable to function in absence of complementing ANP32 171  (Figure 2). 175 Compared with wildtype ANP32A (WTA), ANP32A-D130N did not support Eng/195 176 polymerase activity at all, while R132Q and S158A substitutions had significantly reduced 177 capacity to support FluPol activity (Figure 2A). An artificial mutant with the phosphomimic 178 S158E, also had reduced capacity to support Eng/195 polymerase. In contrast, ANP32A-179 S158T supported FluPol to an extent similar to wildtype ANP32A. Compared with wildtype 180 ANP32B (WTB), ANP32B-D130A had a large deleterious effect on the support for Eng/195 181 polymerase activity, as did the leucine to valine substitution at position 128 (ANP32B-L128V) 182 ( Figure 2B). ANP32B-E133Q was also significantly less able to support Eng/195 FluPol 183 activity. Substitutions of the leucines at positions 138 and 142 to histidine (L138H) and 184 phenylalanine (L142F), respectively, did not compromise the ability of the mutant ANP32B 185 proteins to support Eng/195 polymerase activity. 186 In general the effects of natural variation in ANP32 proteins was similar for Vic/75 187 polymerase. ANP32A-D130N did not support Vic/75 polymerase activity, compared with 188 WTA ( Figure 2C), but R132Q and S158A substitutions had a smaller effect on Vic/75 FluPol 189 activity than on Eng/195 FluPol activity. The S158T substitution and the phosphomimic 190 S158E were as capable of supporting Vic/75 FluPol activity as WTA. As seen with Eng/195 191 polymerase, the ANP32B-L128V substitution was unable to rescue Vic/75 polymerase 192 activity, but the D130A mutation had less detrimental effect and resulted in only a <2-fold 193 reduction in Vic/75 polymerase activity ( Figure 2D). This is in contrast to Eng   interaction of murine ANP32A with FluPol was reduced 2.5-fold compared with mouse 225 ANP32B, which does have pro-influenza virus activity, and 3.2-fold compared with wildtype 226 human ANP32A ( Figure 3B). 227 The naturally occurring human ANP32 variants ANP32A-D130N and ANP32B-D130A also 228 showed reduced binding to FluPol ( Figures 3C and D). The luciferase signal indicating 229 binding of ANP32A-D130N to FluPol was reduced almost 3-fold, relative to wildtype ANP32A 230 and the signal indicating FluPol interaction of ANP32B-D130A was reduced >2-fold 231 compared with wild type ANP32B. These differences in binding affinity were not due to 232 differential expression of the PB1-luc1 and ANP32-luc2 constructs ( Figures 3E and F). (homozygous mutant; red bars) or a 1:1 ratio of both (heterozygous mutant; purple bars). We 259 found that adding increasing amounts of wildtype ANP32B, essentially mimicking the 260 homozygous wildtype genotype (blue bars), had no effect on FluPol activity compared with 261 FluPol supported by ANP32A alone (grey bar). This is presumably due to the presence of 262 wildtype ANP32A in the BKO cells, since ANP32A and ANP32B serve redundant roles in 263 supporting FluPol. In contrast, polymerase activity decreased significantly when ANP32B-264 D130A rather than wildtype ANP32B was co-transfected, and this effect was more dramatic 265 the more variant that was expressed (red bars). This suggests paralogue interference of 266 ANP32B-D130A over wildtype ANP32A, which would otherwise support polymerase activity 267 effectively (grey bar). Co-transfecting increasing amounts of wildtype and mutant ANP32B in 268 a 1:1 ratio, recapitulating the heterozygous mutant genotype (purple bars), also resulted in a 269 significant drop in polymerase activity. This suggests that ANP32B-D130A also exerts a 270 dominant-negative effect over wildtype ANP32B. These observations were unrelated to 271 expression levels of the FLAG-tagged ANP32 constructs (accompanying Western blots). 272 Statistical significance of the differences in Eng/195 polymerase activity between conditions 273 is shown in the accompanying Table. 274 We next investigated whether similar dominant-negative effects were exerted by the rare 275

IAV polymerase activity and replication are attenuated in ANP32B-D130A mutant cells 313
To further probe the potential significance of the relatively common ANP32B-D130A variant, 314 we generated a cell line with the wildtype Anp32A / mutant Anp32B (D130A) genotype by 315 CRISPR/Cas9 genome editing, using human codon-optimised SpCas9 and a single-316 stranded DNA (ssODN) homology-directed repair template. We also generated a cell line 317 lacking the entire pro-viral 128-130 loop (Δ128-130) ( Figure 5A). We selected an 318 unsuccessfully edited clone to serve as a negative control, and carried out Western blotting 319 analysis to ensure endogenous wildtype ANP32A and wildtype or mutant ANP32B were 320 expressed in control and mutant cell lines ( Figure 5B). Using minigenome reporter assays, 321 we found that reconstituted IAV polymerases from Eng/195 ( Figure 5C) or Vic/75 ( Figure 5D

DISCUSSION 367
Here, we used publicly available databases to perform a biased screen for naturally  press). A drawback of using a low-ploidy cell line is that heterozygous genotypes cannot 439 easily be generated. Thus far we have been unable to trace carriers of the variants but future 440 work using diploid respiratory epithelial cells that reflect the heterozygous genotype could 441 indicate whether true heterozygotes, like the reconstituted cells we used here, also display 442 reduced influenza susceptibility. 443 In conclusion, we have provided the first example of a single nucleotide variant in the coding 444 region of a human gene that may offer carriers some protection against influenza virus. This 445 work has the potential to inform future intervention. 446

Plasmids and cloning 456
Human FLAG-tagged pCAGGS-ANP32A and ANP32B expression plasmids have been  transfection, cells were lysed in 50 µl passive lysis buffer (Promega) for 30 minutes at room 518 temperature with gentle shaking. Bioluminescence generated by firefly and Renilla 519 luciferases was measured using the dual-luciferase system (Promega) on a FLUOstar 520 Omega plate reader (BMG Labtech). 521 Split luciferase complementation assay 522 pCAGGS expression plasmids encoding H3N2 Vic/75 PB1-luc1, PB2, PA, and the indicated 523 ANP32-luc2 construct were transfected into ~100,000 293T cells at a ratio of 1:1:1:1 (15 ng 524 per well). Control conditions contained pCAGGS-luc1 and untagged PB1, or pCAGGS-luc2 525 and untagged ANP32A, respectively, with all other components remaining constant. Empty 526 pCAGGS plasmid was used to ensure total transfected DNA was equal across conditions. 527 Twenty-four hours after transfection, cells were lysed in 50 µl Renilla lysis buffer (Promega) 528 for 1 h at room temperature with gentle shaking (Gaussia and Renilla luciferase share the 529 same substrate). Bioluminescence generated by Gaussia luciferase was measured using the 530 Renilla luciferase kit (Promega) on a FLUOstar Omega plate reader (BMG Labtech). 531 Normalized luminescence ratios (NLR) were calculated by dividing the signal from the 532 potential interacting partners by the sum of the two controls, as described (Mistry et al., 533 2020). 534