Co-infection of chickens with H9N2 and H7N9 avian influenza viruses leads to emergence of reassortant H9N9 virus with increased fitness for poultry and enhanced zoonotic potential

An H7N9 low pathogenicity avian influenza virus (LPAIV) emerged through genetic reassortment between H9N2 and other LPAIVs circulating in birds in China. This virus causes inapparent clinical disease in chickens, but zoonotic transmission results in severe and fatal disease in humans. We evaluated the consequences of reassortment between the H7N9 and the contemporary H9N2 viruses of G1 lineage that are enzootic in poultry across the Indian sub-continent and the Middle East. Co-infection of chickens with these viruses resulted in emergence of novel reassortant H9N9 viruses carrying genes derived from both H9N2 and H7N9 viruses. These reassortant H9N9 viruses showed significantly increased replication fitness, enhanced pathogenicity in chicken embryos and the potential to transmit via contact among ferrets. Our study highlights that the co-circulation of H7N9 and H9N2 viruses could represent a threat for the generation of novel reassortant viruses with greater virulence in poultry and an increased zoonotic potential. Graphical Abstract In Brief H9N2 viruses have a high propensity to reassort with other avian influenza viruses. We found that co-infection of chickens with H9N2 and H7N9 led to the emergence of reassortant viruses including the H9N9 subtype. Some reassortant H9N9 viruses exhibited increased replication fitness, increased pathogenicity in the chicken embryo, greater avidity for human and avian cell receptors, lower pH fusion and contact-transmission to ferrets. This study demonstrated the ability of viruses that already exist in nature to exchange genetic material, highlighting the potential emergence of viruses from these subtypes with increased zoonotic potential. There are nine H9 influenza A subtypes carrying different neuraminidase (NA) genes, including H9N9 viruses, while they are not common they do exist in nature as wildtypes (CDC). Highlights Co-infection of chickens with H7N9 and H9N2 led to emergence of reassortant H9N9 viruses Reassortant H9N9 viruses had an increased replication rate in avian and human cells Reassortant H9N9 viruses had a lower pH fusion and significantly higher receptor binding to α 2,3 sialoglycans Reassortant H9N9 replicated in ferrets at similar levels compared to H7N9 and transmitted via direct contact Ferrets exposed to reassortant H9N9 by aerosol contact were also found to be seropositive Experimental simulation of events that may occur naturally with circulating viruses has demonstrated the risk of emergence of viruses with increased zoonotic potential.

analysed. An H9N9 reassortant was appropriately selected for infection using a ferret model 126 to further ascertain any potential zoonotic characteristics. 127 12 Genotype 122 also displayed increased replication kinetics relative to UDL/08 in both CK 272 and MDCK cells, albeit without statistical significance (P>0.05) (Figures 5f and 6f). 273 Five reassortant viruses also showed increased embryo lethality (lower embryo lethal dose 274 (ELD 50 ) compared to the progenitor viruses (Table 1) Following initial co-infection of the D0 chickens, AIV RNA shedding at relatively high titres at 285 4 dpi (3 dpc) from the oropharyngeal cavity of the R1 chickens represented virus(es) that 286 were sufficiently fit and had successfully transmitted within this host. To elucidate the exact 287 constellation of genes within any viable reassortant virus(es), plaque purification of the 288 oropharyngeal swab samples from all the nine R1 contact chickens was carried out in MDCK 289 cells. Discrete plaques between a range of 16-28 were picked and RNA was extracted to 290 fully characterise their genotype by segment specific RT-qPCRs. The genotype frequencies 291 in each sample (expressed as percentage (%)) were calculated by taking the ratio of number 292 of times a particular genotype appeared by total plaques isolated for a particular sample 293 (Table 2). The genotypes identified by plaque purification reflected the overall genotyping as 294 identified by RT-qPCR of swab samples ( Figure 2). However, not all genotypes identified by 295 RT-qPCR of swab samples could be isolated by plaque purification. The un-reassorted 296 H9N2 UDL/08 was detected in four out of nine chickens between 100% to 13.6% genotype 297 frequencies. In addition, a total of eight novel genotypes including single, double and triple 298 13 segment reassortants were detected. Among all the viruses (  Figure 8A) showed strong binding for 6SLN receptors which was comparable to that of 326 Anhui/13 ( Figure 8C). By contrast, the parental UDL/08 displayed only marginal and 327 undetectable binding to 6SLN receptor analogues ( Figure 8B). The reassortant H9N9 328 viruses also bound strongly to 3SLN compared to the parental UDL/08 H9N2 virus, which 329 showed no binding ( Figure 8A and B) Figure 9A and 354 B). However, ferrets sharing the same airspace but separated physically (indirect) via a 355 dividing mesh in adjacent cages did not show detectable viral shedding in either groups 356 (R1 In ). 357 All directly infected (D0) and direct contact (R1 DC ) ferrets seroconverted when tested by HI 358 assay against homologous viruses ( Figure 9C and D). None of the ferrets indirectly exposed 359 to the Anhui/13 H7N9 virus infected group (R1 In ) seroconverted to H7N9 virus, but all ferrets 360 indirectly exposed to the reassortant H9N9 virus infected ferrets (R1 In ) seroconverted to 361 H9N9 virus ( Figure 9C and D). 362 Three ferrets in each of the D0 groups were culled at 4 dpi in order to provide a range of 363 organs, mainly from the respiratory tract, for post mortem examination. Both the Anhui/13 364 and reassortant H9N9 viral RNA were detected at high levels (>5log 10 REU of EID 50 ) in the 365 nasal turbinates of the D0 ferrets, with the former also detected in the olfactory lobe ( Figure  366 9E and F). Viral nucleoprotein was detected in the respiratory and olfactory epithelium of the 367 nasal turbinates in all three infected ferrets (Supplementary Figure S4). In addition, 368 histological lesions were identified in the respiratory epithelium of all ferrets for both viruses, 369 and to a lesser extent in the olfactory epithelium for both viruses (Supplementary Figure S4).   Figure S5). Neither of the H7N9 or H9N9 viruses was 377 detected in the brain or liver, although Anhui/13 was detected at a low level (>2log10 REU of 378 EID 50 ) in the kidney of one infected ferret ( Figure 9E and F), while H9N9 was detected in the 379 ileum of one ferret at the limit of detection. 380 As regards to clinical changes, the D0 ferrets directly-infected with H9N9 reassortant 381 experienced a negligible increase in body temperature and a modest reduction in weight 382 (around 2-3%) at 2 dpi (Supplementary Figure S6 C and D), compared to Anhui/13 D0 383 infected ferrets which developed fever from 1 to 3 dpi and exhibited weight loss from 1 till 384 7dpi, which decreased to ~10% of their starting weight in some ferrets (Supplementary 385 Figure S6 A and B). The increase in body temperature correlated with peak viral shedding of 386 Anhui/13 from the D0 ferrets ( Figure 9A). 387 The ferrets placed in contact with the directly infected Anhui/13 and reassortant H9N9 388 infected ferrets did not develop a significant increase in body temperature or a weight loss. and this observation was reflected in the successful generation and transmission of these 487 H9N9 genotypes in chickens in our in vivo experiment. The H9N2 and genotype 122 H9N9 488 viruses were found to have an optimal pH fusion of 5.4, which was slightly lower compared 489 to that of Anhui/13 H7N9 having an optimal pH fusion 5.6 as seen previously (Chang et al., 490 2020 Figure S4)   μ g/ml TPCK trypsin for 15 min and then exposed 694 to PBS buffers with pH values ranging from 5.2 -6.0 (at 0.1 pH-unit increments) for 5 min. 695 The PBS buffer was then replaced with DMEM containing 10% FCS. The cells were further 696 incubated for 3 h at 37 °C to allow for syncytium formation before being fixed with ice-cold 697 28 (-20℃) methanol and acetone (1:1) mixture for 12 mins. and stained with 20% Giemsa stain 698 (Sigma-Aldrich) for 1 h at room temperature. The pH at which syncytium formation was 699 judged to be greater than 50% corresponded to the pH of viral membrane fusion. 700

Experimental design: Ferret infection and transmission 701
Thirty male ferrets (Mustela putorius furo) were sourced from Highgate Farms, UK at a 702 maximum age of 3 months and weighing between 750-1000g. The ferrets were confirmed as 703 serologically negative to IAV by ID Screen® Influenza A Nucleoprotein Indirect ELISA (ID 704 Vet). Ferrets were also confirmed negative for IAV ongoing infection (shedding) by testing 705 RNA extracted from nasal washes by the M-gene RT-qPCR (Nagy et al., 2010), as above. 706 All ferrets were microchipped (bio-thermal chip) to monitor identification number and the 707 body temperature. Two groups of ferrets (n=3 per group; the D0 ferrets) were housed in 708 cages in separate containment rooms, and directly-infected via the i.n. route with 1 X 10 7 709 EID 50 of Anhui/13 (H7N9) or H9N9 (genotype 122) (Supplementary Figure S7). At 1 dpi, 710 direct-contact ferrets (n=3, i.e. the R1 DC ferrets) were introduced for co-housing in the same 711 cage with the D0 ferrets in each room. Simultaneously, indirect-contact (R1 In ) ferrets (n=3) 712 were housed in a cage adjacent to that which housed the D0 and R1 DC ferrets in each room 713 (Supplementary Figure S7). Both cages in each room were separated by a double mesh 714 which prevented direct contact between the ferrets but allowed potential IAV aerosol 715 transmission. Each room also contained a third cage which housed three ferrets directly-716 infected with the two IAVs, and these were culled at 4 dpi for post mortem (PM) analysis, 717 these six being referred to as the D0 PM ferrets. Tissues from the D0 pm ferrets were taken into 718 1ml of PBS (10% w/v) and RNA extracted for testing for influenza virus RNA using M-gene 719 RT-qPCR. All directly-infected and contact-exposed ferrets were nasal washed with 1 ml of 720 PBS (0.5 ml/nare) and similarly tested for IAV RNA using M-gene RT-qPCR until 12 dpi. The 721 remaining ferrets were culled and cardiac bled at 12 dpi, with seroconversion to IAV 722 assessed by the haemagglutination inhibition (HI) test using the homologous antigens, as 723 described. 724

Serology 725
To remove the non-specific inhibitors for HI, chicken sera were inactivated at 56 °C for 30 726 min, while ferret sera were incubated with 4 volumes of receptor destroying enzyme (APHA 727 Scientific, Weybridge) for 1 hour at 37°C before being inactivated at 56 °C for 30 min as 728 previously described (WHO, 2002). Seroconversion to the subtype-specific HA antigens was 729  Tables   Table 1:

Genotypes of the potential reassortant viruses in swabs samples which emerged after co-infection and transmission to R1 chickens.
The oropharyngeal swab samples from the contact chickens (R1) were processed and potential reassortant viruses were identified by RT-qPCR rescued by reverse genetics and compared for their 50% egg lethal dose (ELD 50 ) in 10-day old SPF embryonated eggs. Table 2: Genotypes of the reassortant viruses isolated after plaque purification from the oropharyngeal swab samples from the co-infected contact chickens at 3 dpc.
The genotype frequencies in each sample were calculated by taking the ratio of number of times a particular genotype appeared in each sample by total plaques isolated for a particular sample and expressed as percentage.  The Ct values were compared against an Anhui1/13 or UDL/08 RNA standards to determine relative equivalency units (REU of EID50). The REU values obtained from the segment-specific RT-qPCRs were converted to illustrate the percentage frequency of the origins of each gene, shown by the relative lengths of the horizontal red and blue bars. The vertical dotted lines within gene segment column represent 10%, 50% and 90% frequencies of each gene. Annotation on the left denotes the in vivo infected chickens: H9N2 and H7N9 correspond to the single-infected control groups; PM corresponds to the chickens which were pre-planned for cull and post mortem examination at 4 dpi (pathogenesis experiment); while D0 and R1 respectively indicate the direct-and contact-infected chickens following co-infection with both progenitor viruses; the final two digits of the individual chicken identifiers are discernible by the small font size at the left-end of each row. On the right, the AIV REU of EID50 for each chicken's swab are shown by dark grey horizontal lines, with the broken vertical line indicating the REU positive cut-off. The failure to detect the origins of a given viral genetic segment (shown by grey horizontal bars) among several cloacal swabs tended to occur in those with low viral shedding values.  The RNP gene complexes derived from Anhui/13 and UDL/08 viruses were reconstituted by transfection of (A) chicken DF-1 and (B) human HEK-293T cells, along with four mixed RNP combinations from the H7N9 and / or H9N2 viruses. The cells were incubated at 37℃ (HEK-293T) or 39℃ (DF-1). At 24 hours post-transfection, cells were lysed and luciferase activities measured. Co-transfection of plasmids without PB1 served as a negative control for RNP activity. Percent polymerase activity was relative to that measured in the corresponding control H9N2 or H7N9 transfections. *** indicates P value <0.0001. Replication kinetics of each reassortant virus compared to parental Anhui/13 H7N9 and UDL/08 H9N2 viruses is shown in panels (a) to (n). The genotype of each reassortant virus is shown as a combination of black and white coloured boxes with black indicating Anhui/13 H7N9 origin and white indicating UDL/08 H9N2 origin gene segments. ** denotes P value <0.005 and **** denotes P value <0.0001 compared to UDL08 H9N2.   Anhui/13 H7N9 and 6 genes from UDL/08 H9N2] to α-2,3-linked (3′SLN 6-sulphated) (black), α-2,3-linked (3′SLN) (red), or α-2,6-linked (6′SLN) (green) sialylglycan receptors was determined by biolayer interferometry. Similar receptor binding profiles were determined for (B) UDL/08 H9N2 and (C) 2+6 reassortant H7N9 [2 genes (HA and NA) from Anhui/13 H7N9) plus 6 genes from PR8]. *Since biolayer interferometry involved testing of infectious virus, due to biosafety reasons the receptor binding of H7N9 was carried out using the 2+6 reassortant of H7N9 which included internal genes from PR8.

Figure 9. Detection of influenza A virus RNA and seroconversion in ferrets after intranasal inoculation with H7N9 and reassortant H9N9 viruses.
Two separately housed groups of ferrets (D0, n=6 per group) were infected directly via the intranasal route with (A) Anhui/13 and (B) the reassortant H9N9 (genotype # 122). At 1 dpi, a group of ferrets (n=3) were placed in direct contact (R1DC) with each group of D0 ferrets, while another indirect contact group (R1In) of ferrets (n=3) were placed in an adjacent cage. Nasal washes were collected on alternate days from all ferrets which remained in the study until 12 dpi to determine viral shedding (REUs of EID50) by the M-gene RT-qPCR. Along the horizontal axis, dpi and dpc corresponds to time-points for the D0 and all the contact ferrets respectively. (C, D) Seroconversion in D0, R1DC and R1In ferrets at 14 dpi / 13 dpc. (E, F). Three ferrets from each D0PM group were euthanized for postmortem on 4 dpi to assess virus dissemination in internal organs by M-gene RT-qPCR. The broken horizontal line corresponds to the positive cut-off value for the M-gene RT-qPCR (A, B, E and F) and the HI (C and D) tests.

Immunostaining procedure -cells
Briefly, the viruses were serially diluted (two-fold) in DMEM containing 1x penicillin streptomycin to infect the Vero cells in a 96-well-plate for 1h. The inoculum was then removed and washed once with PBS before addition of DMEM medium with 5% FCS and 1x penicillin streptomycin for 16 h. The medium was removed, and cells were fixed with ice-cold methanol and acetone (1:1) mixture for 12 minutes at room temperature. The cells were washed once with PBS and blocked with blocking buffer (0.1% PBS tween containing 5% BSA) for 1hr at room temperature. The cells were incubated with anti-nucleoprotein (NP) mouse monoclonal antibody HB-65 [H16-L10-4R5 (ATCC® HB-65™)] [diluted 1:200 in dilution buffer (0.1% PBS-tween containing 0.5% BSA)] for 1 hour at room temperature. The cells were washed 4 times with wash buffer (0.1% PBS-tween) followed by incubation with horseradish peroxidase-labelled rabbit anti-mouse immunoglobulins (Dako, Denmark) (diluted 1:200 in dilution buffer). The cells were washed four times with wash buffer and developed using liquid DAB and substrate chromogen system (Dako, USA) as mentioned by the manufacturer.

Quantification of viruses using solid-phase indirect ELISA
The purified test viruses along with a reference virus [X-31 (reassortant virus carrying HA and NA from A/Aichi/2/68 and internal genes from H1N1/PR8] were diluted in carbonate bicarbonate buffer (pH 9.6) and coated on 96 well ELISA plates (Nunc MaxiSorp™) for overnight. The coated wells were permeabilized with 0.2% triton-X for 30 min at room temperature and then washed 4 times with wash buffer before incubating with blocking buffer for 1 hr. Each well was incubated with anti-NP mouse monoclonal antibody (HB65) (1:3000 dilution in dilution buffer) for 1hr. The plate was washed 4 times with wash buffer and then incubated with horseradish peroxidase labelled anti-mouse secondary antibody (Dako) (1:2000 dilution in dilution buffer) for 1 hr followed by addition of TMB substrate reagent set (BD OptEIA™). The reaction was stopped by addition of 1N H2SO4 and absorbance was measured at 450 nm. The concentration of the purified viruses was calculated by comparison of the estimated NP content of the reference virus X-31 as described elsewhere (Nicholson et al., 1998) (Lin et al., 2012) and expressed as picomolar (pM).  Figure S1. Segment specific RT-qPCR standard curves. RNA was extracted from both H7N9 Anhui/13 and H9N2 UDL/08 to achieve 1x106 EID50 / reaction. A 10-fold dilution standard curve was generated for each RNA and used in the gene specific RT-qPCR assays reactions. Ct values from each assay were plotted against EID50. The primer and probes designed to specifically detect H7N9 or H9N2 gene segment had comparable efficiency. were significantly different. '**** and *' denote significance value of P<0.0001 and P < 0.05, respectively when compared to H7N9 virus. 'oooo and o' denote significance value of P<0.0001 and P < 0.05, respectively when compared to H9N2 virus. Three ferrets were directly infected (D0) with H7N9 Anhui/13 virus or reassortant H9N9 virus. Contact ferrets (R1-D) were placed in contact with the directly infected ferrets in the same cage. A group of three ferrets were placed in an adjacent cage separated from the directly infected cage via a double mesh. The cages were maintained with a directional airflow from left to right (as shown by arrows above). A group of three directly infected ferrets (D0-PM) were culled on 4dpi to observe the virus dissemination in internal organs.