Host factors drive the within-host competition between highly and low pathogenic H5N8 avian influenza viruses

Highly pathogenic avian influenza viruses (HPAIV) emerge from low pathogenic avian influenza viruses (LPAIV) through the introduction of basic amino acids at the hemagglutinin (HA) cleavage site. Following viral evolution, the newly formed HPAIV likely represents a minority variant within the index host, predominantly infected with the LPAIV precursor. Using reverse-genetics engineered H5N8 viruses differing solely at the HA cleavage, we tested the hypothesis that the interaction between the minority HPAIV and the majority LPAIV could modulate the risk of HPAIV emergence and that the nature of the interaction could depend on the host species. In chickens, we observed that the H5N8LP increased H5N8HP replication and pathogenesis. By contrast, the H5N8LP antagonized H5N8HP replication and pathogenesis in ducks. Ducks mounted a more potent antiviral innate immune response than chickens against the H5N8LP, which correlated with H5N8HP inhibition. These data provide experimental evidence that HPAIV may be more likely to emerge in chickens than in ducks and underscore the importance of within-host viral variants interactions in viral evolution.


77
Characterization of H5N8HP and H5N8LP viruses in cell culture 78 We used reverse-genetics to generate the wild-type H5N8HP from a HPAIV 79 isolated during the 2016 H5N8 epizootics in France (Guinat et al., 2018).  using primers specific for H5N8HP or H5N8LP HA. In chicken eggs, we 119 detected equivalent levels of viral RNA following inoculation with 10 2 EID50 120 H5N8HP alone or with H5N8LP alone regardless of the inoculum dose ( Fig.   121 2B). Upon co-infection with H5N8HP and H5N8LP in chicken eggs, the level 122 of H5N8HP and H5N8LP-specific viral RNA did not differ from mono-123 infections regardless of the quantity of H5N8LP (Fig. 2B). Thus, the 124 replication of H5N8HP is not affected by H5N8LP, regardless of the amount of 125 H5N8LP co-inoculated in chicken eggs. In duck eggs, we detected similar 126 levels of viral RNA following inoculation with 10 2 EID50 H5N8HP alone or 127 with 10 2 EID50 H5N8LP alone (Fig. 2C)   Based on these results, we investigated the interaction between a HPAIV 154 and a LPAIV in chickens. Four-week old chickens were assigned one of six 155 groups: L7, H3, H4, L7H3, L7H4 or NI (Fig. 3A). L7 animals were infected 156 with 10 7 EID50 H5N8LP. H3 animals were infected with 10 3 EID50 H5N8HP.

157
H4 animals were infected with 10 4 EID50 H5N8HP. L7H3 animals were 158 infected with a mixture of 10 7 EID50 H5N8LP and 10 3 EID50 H5N8HP. L7H4 159 animals were infected with a mixture of 10 7 EID50 H5N8LP and 10 4 EID50 160 H5N8HP. Finally, non-infected animals from group NI were administered 161 vehicle only. Neither mortality nor clinical signs were observed in chickens 162 from group L7 and group H3 (Fig. 3B). Chickens in the H4 group reached 45% mortality (Fig. 3B). When chickens were infected with a mixture of 164 H5N8LP and H5N8HP, we observed an increase in mortality compared to 165 H5N8HP mono-infected chickens: mortality reached 18% in the L7H3 group, 166 compared to 0% in the H3 group; and 72% in the L7H4 group, compared to 167 45% in the H4 group (Fig. 3B). Although the increase in mortality did not 168 reach statistical significance, these results indicate that H5N8LP and H5N8HP 169 synergized, resulting in increased pathogenesis in co-infected chickens 170 compared to H5N8HP mono-infected chickens. 171 We measured viral shedding from oropharyngeal and cloacal swabs by   190 We inoculated groups of 4-week old Pekin ducks (Anas platyrhynchos  post-infection (dpi), with a hundred-fold difference at 1 dpi (p<0.001) and 210 a ten-fold difference at 2 and 3 dpi (p<0.01) (Fig. 7C). Cloacal H5N8HP 211 shedding was also decreased in group L7H4, but the difference only 212 reached statistical difference at day 3, with a ten-fold reduction (p<0.05) 213 (Fig. 7D). There was no difference in H5N8LP shedding between groups 214 L7H4 and L7 (Fig. 7E).

215
Next, we analyzed viral RNA levels in the lungs and brain. H5N8LP was not 216 detected in the lungs or brain, indicating that H5N8LP replicated mostly in 217 the upper respiratory tract of ducks. In contrast to chickens, we detected 218 H5N8HP nucleic viral RNA from the lungs and brain of most animals at 1 and 219 3 dpi (Fig. 7F&G). H5N8HP viral RNA load was decreased in the lungs and 220 brain of H5N8LP and H5N8HP co-infected birds (group L7H4), compared to 221 H5N8HP mono-infected birds (group H4). High levels of H5N8HP viral RNA 222 were also detected in the lungs and brain of moribund ducks, which 223 developed predominantly neurological signs.

224
Extensive viral antigen staining was detected by immunohistochemistry in 225 the brain of moribund animals, while more modest viral antigen staining 226 was observed in the lungs, with no difference between groups L7H4 and H4 227 (Fig. 8A&B). We then evaluated the expression of host immune response 228 markers in the lungs and brain using RT-qPCR. Mx mRNA expression was 229 significantly increased in the lungs of infected ducks belonging to the L7H4 230 and H4 groups at 1 and 3 dpi, as well as in moribund animals (Fig. 9A). 231 Compared to NI duck, we also observed an upregulation of Mx mRNA 232 expression in the lungs of H5N8LP mono-infected ducks (L7 group) at 1 and 233 3 dpi, although the difference did not reach statistical significance. IFN-γ 234 mRNAs levels were significantly higher in the lungs of H5N8HP mono-235 infected ducks (group H4) at 1 dpi compared to NI, L7H4 and L7 ducks (Fig.   236 9B). Similar results were observed in the brain (data not shown). To To model the intra-host competition between a newly formed HPAIV and 260 its parental LPAIV, we inoculated chickens and ducks with an H5N8HP virus 261 and an H5N8LP virus differing solely at the level of the HA cleavage site.

262
When inoculated alone, H5N8HP replication and pathogenesis were 263 equivalent in chickens and ducks. However, when chickens and ducks were 264 co-inoculated with this pair of viruses, we found that the H5N8HP had a 265 stronger selective advantage over the H5N8LP in chickens than in ducks.
266 Surprisingly, we observed that the H5N8LP increased H5N8HP replication 267 and pathogenesis in chickens. By contrast, we observed that the H5N8LP