Transmission and pathogenicity of novel reassortants derived from Eurasian avian-like and 2009 pandemic H1N1 influenza viruses in mice and guinea pigs

Given the present extensive co-circulation in pigs of Eurasian avian-like (EA) swine H1N1 and 2009 pandemic (pdm/09) H1N1 viruses, reassortment between them is highly plausible but largely uncharacterized. Here, experimentally co-infected pigs with a representative EA virus and a pdm/09 virus yielded 55 novel reassortant viruses that could be categorized into 17 genotypes from Gt1 to Gt17 based on segment segregation. Majority of novel reassortants were isolated from the lower respiratory tract. Most of reassortant viruses were more pathogenic and contagious than the parental EA viruses in mice and guinea pigs. The most transmissible reassortant genotypes demonstrated in guinea pigs (Gt2, Gt3, Gt7, Gt10 and Gt13) were also the most lethal in mice. Notably, nearly all these highly virulent reassortants (all except Gt13) were characterized with possession of EA H1 and full complement of pdm/09 ribonucleoprotein genes. Compositionally, we demonstrated that EA H1-222G contributed to virulence by its ability to bind avian-type sialic acid receptors, and that pdm/09 RNP conferred the most robust polymerase activity to reassortants. The present study revealed high reassortment compatibility between EA and pdm/09 viruses in pigs, which could give rise to progeny reassortant viruses with enhanced virulence and transmissibility in mice and guinea pig models.

restrictions [4][5][6] . EA H1N1 virus first emerged in Europe in 1979 and has replaced the classical swine H1N1 virus which indicates that the EA virus has competitive advantage in pigs [7][8][9] . In 2007, EA virus was first reported in Hong Kong and eventually became dominant in China 10,11 . The EA virus has been documented to cause respiratory infection in humans 12,13 . Since the global human population is expected to have little to no immunity against EA virus or virus with EA HA gene [14][15][16][17] , the EA swine lineage of viruses could be a potential genetic source for the development of future pandemic strains.
The pdm/09 virus continues to circulate in humans and has effectively become a seasonal virus. Pigs are highly susceptible to pdm/09 virus, which repeatedly infected pigs around the world and reassortment with pdm/09 virus has also occurred with many other SIV lineages apart from the EA lineage 18,19 . Previous studies have identified reassortant swine influenza virus derived from EA and pdm/09 lineages in Hong Kong, Germany, Italy and United Kingdom [20][21][22][23][24][25] . Recent surveys of SIVs from 2009-2012 in South China found marked changes in the evolutionary patterns and ecosystem of SIVs and that only the internal genes of the pdm/09 virus were present in dominant lineages in pigs 10 . Collectively, these observations suggest that reassortants between pdm/09 and EA H1N1 viruses could potentially generate reassortant viruses with novel antigenicity for humans. Although many reassortants between pdm/09-origin viruses and enzootic SIVs have been detected, no systematic study has been reported on reassortants between pdm/09 and EA viruses.
To better understand the genetic diversity and biological properties of novel reassortants between pdm/09 and EA influenza virus, we experimentally co-infected pigs with EA swine H1N1 and pdm/09 viruses, and examined the reassortant patterns of progeny viruses and their pathogenicity/transmissibility in mice and guinea pigs. We found that the two virus lineages could generate reassortant viruses with enhanced pathogenicity and transmissibility in both animal models, and that the most pathogenic and transmissible reassortants retained the EA H1 and full complement of pdm/09 RNP gene segments.

Reassortants from co-infection of pigs with parental EA/FJ04 and pdm/BJ09 H1N1 viruses.
To systematically examine reassortants obtained from co-infection of swine with EA and pdm/09 viruses, a representative EA virus A/Swine/Fujian/204/2007 (EA/FJ04) and a representative pdm/09 virus A/Beijing/7/2009 (pdm/BJ09) were selected for coinfection of six pigs inoculated by intranasal and intratracheal routes. At 3 days post-infection (dpi), all pigs exhibited clinical symptoms including coughing and pyrexia. Daily nasal washes from 1 to 5 dpi, and bronchoalveolar fluids (BALFs) from 3 euthanatized pigs each time at 3 and 5 dpi were collected. Ten plaque purified viruses were isolated and sequenced from each nasal wash and BALF sample. A total of 300 purified viruses (about 50 per pig) were isolated, of which 240 viruses were from nasal washes and 60 viruses were from BALFs. Based on the parental origin of eight gene segments, plaque-purified viruses were classified as EA/FJ04 or pdm/BJ09 genotype if they contained all eight genes from only one parental virus, and as reassortants if they contained segments of both parental viruses.

Distribution of reassortant viruses in respiratory tracts of co-infected pigs.
There is clear differential expression of avian-like sialic acid α 2,3-galactose (SAα 2,3-Gal) linked and human-like sialic acid α 2,6-galactose (SAα 2,6-Gal) linked receptors along the porcine respiratory tract, with SAα 2,3-Gal linked receptors mainly found in the LRT 26 . We found that the frequency of reassortment was low in nasal passage throughout the infection period, but progressively increased during infection in lungs (Fig. S1). Thirty-six out of 60 (60%) viruses isolated from BALFs and 19 out of 240 (8%) from nasal washes were reassortant viruses. The 19 reassortants isolated from the nasal passage comprised five genotypes of Gt7, Gt10, Gt13, Gt14, and Gt15 (Table 1) Transmissibility of EA/FJ04 and pdm/BJ09 reassortants in guinea pigs. Guinea pigs were used to determine the transmissibility of reassortant viruses. As shown in Fig. 1, both parental EA/FJ04 and pdm/BJ09 viruses readily infected guinea pigs. Pdm/BJ09 virus replicated more efficiently in the nasal passage than EA/FJ04 virus and was detected in the nasal washes of all three inoculated and three in-contact guinea pigs, which indicated efficient horizontal transmission. However, there was no transmission of EA/FJ04 virus to in-contact guinea pigs. Based on virus transmissibility, the 17 reassortant genotypes were categorized into three groups: 1. Transmissibility group I comprised 6 reassortant genotypes, Gt2, Gt3, Gt7, Gt10, Gt13 and Gt14, with high transmissibility that was similar to that of the parental pdm/BJ09 virus (Fig. 1). The six genotypes made up 67% (37/55) of all reassortant isolates. Most of the reassortants of Gt2, Gt3, Gt7, and Gt10 in this group harbored EA/FJ04 H1 gene. Notably, all viruses from this group contained PB2, PB1, PA and NP segments of the pdm/BJ09 virus. 2. Transmissibility group II composed of 8 reassortant genotypes, Gt1, Gt4, Gt6, Gt9, Gt12, Gt15, Gt16 and Gt17, that showed reduced transmissibility but still higher than the parental EA/FJ04 virus. Gt1, Gt4, Gt6, Gt12, Gt15 and Gt17 reassortants were transmitted to two out of the three in-contact guinea pigs, while Gt9 and Gt16 reassortants infected one out of three in-contact animals (Fig. 2). The 8 genotypes comprised 20% (11/55) of reassortant viruses. Interestingly, all viruses in this group except Gt1 contained PB2, PA and NP gene segments of the pdm/BJ09 virus and PB1 segment of the EA/FJ04 virus. 3. Transmissibility group III comprised 3 reassortant genotypes, Gt5, Gt8 and Gt11, which showed no horizontal transmission which was similar to that of the parental EA/FJ04 virus. These viruses were only detected in the nasal washes of all three inoculated guinea pigs (Fig. S2). The two genotypes (Gt8, Gt11) harbored PB2, PA and NP gene segments of the pdm/BJ09 virus and PB1 segment of the EA/FJ04 virus.
Seroconversion occurred in all inoculated animals and in infected in-contact animals. For statistical analyses, the area under the nasal wash curve between 2 and 6 dpi (AUC 2-6 dpi ), which measured duration of peak virus shedding, was calculated for each guinea pig, as previously described 27 . The AUC 2-6 dpi value of the EA/FJ04 virus group was significantly lower than that of the pdm/BJ09 virus group and all other reassortant genotypes (Fig. S3). These results suggested that in guinea pigs pdm/BJ09 genes enhanced replication of reassortant viruses.
Pathogenicity of EA/FJ04 and pdm/BJ09 reassortants in mice. All reassortant genotypes and the two parental EA/FJ04 and pdm/BJ09 viruses were assessed for pathogenicity in BALB/c mice. Both parental viruses readily infected mice, and LD 50 of EA/FJ04 and pdm/BJ09 was > 10 5.5 and 10 4.5 TCID 50 respectively. Pdm/ BJ09 virus replicated more efficiently in the LRT than EA/FJ04 virus on day 4 following intranasal virus inoculation at 10 5 TCID 50 . Mice that lost more than 25% of their body weight were euthanized. Pdm/BJ09 virus caused > 25% weight loss which ensured complete mortality in mice whereas EA/FJ04 virus caused only 1% weight loss with no mortality. All tested reassortants readily infected mice and replicated in the LRT ( reassortant genotypes were additionally categorized into three groups according to their relative pathogenicity in mice compared with parental viruses: 1. Pathogenicity group I comprised 6 reassortant genotypes, Gt2, Gt3, Gt5, Gt7, Gt10 and Gt13, with pathogenicity higher than either parental virus ( Table 2). With LD 50 at10 3.5 to10 3.8 TCID 50 and mean survival time (MST) of 4 to 5.3 days, they constituted 58% (32/55) of reassortant viruses. Infected mice displayed obvious clinical signs of huddling, ruffled fur, severe dyspnea by 4 dpi and complete mortality after 5 days. With this group of reassortants, even at a reduced infection dose of 10 4 TCID 50 , severe clinical signs were evident (data not shown). Hematoxylin and eosin (H&E) lung histopathology performed on three mice from each genotype at 4 dpi revealed severe bronchopneumonia and interstitial pneumonia, characterized by edema, epithelial erosion and massive infiltration of inflammatory monocytic cells (Fig. 3A,B).   (Fig. 3C,D). However, this group of reassortants at a lower inoculation dose of 10 4 TCID 50 elicited no clinical signs in mice (data not shown). 3. Pathogenicity group III comprised 3 reassortant genotypes, Gt1, Gt9 and Gt12, which showed reduced virulence similar to that of the EA/FJ04 virus. This group did not cause any death over 2 weeks of infection and constituted only 5% (3/55) of total reassortant viruses. Clinical signs from infected mice were not apparent and lung histopathological changes were mild to moderate (Fig. 3E,F). Notably, EA/FJ04 virus grew to lower virus titers in lungs at 4 dpi than all reassortants, which suggested that introduction of pdm/BJ09 genes into EA/FJ04 virus enhanced viral replication in mice ( Table 2). In summary, five out of six reassortant genotypes in the most pathogenic category (Gt2, Gt3, Gt5, Gt7 and Gt10 in pathogenicity group I) housed the H1 gene from the EA/FJ04 virus; in particular Gt7 reassortant viruses only contained H1 gene from EA/FJ04, indicating that EA H1 contributed to virulence in these reassortants. All six genotypes in this high pathogenicity group housed the 4 RNP genes of PB1, PB2, PA and NP from the pdm/ BJ09 virus, suggesting that pdm/BJ09 RNP is critical to enhanced pathogenicity. Notably, highly pathogenic Gt2, Gt3, Gt7, Gt10 and Gt13 reassortants were earlier found to be highly transmissible in guinea pigs (Fig. 1).

HA-222G was critical in the pathogenicity of EA H1 reassortant virus in mice. Our above results
suggested that the EA/FJ04 HA gene is an important virulence factor in the reassortant viruses. Receptor-binding specificity of HA is an important molecular determinant of host range, transmissibility and replication 28 . To ascertain the receptor-binding properties of the two parental viruses, we performed binding of EA/FJ04 and pdm/ BJ09 viruses to a glycan array panel of SAα 2,3-or SAα 2,6-linked sialosides (Fig. 4). In keeping with the previous reports 29 , our results showed that pdm/BJ09 virus bound to SAα 2,6 Gal linked receptor only. However, EA/FJ04 virus additionally exhibited SAα 2,3 Gal linked receptor binding.
Previous studies showed that HA D222G mutation in 2009 pdm virus resulted in additional SAα 2,3 linked receptor binding preference and enhanced virulence in mice and macaques 30,31 . EA/FJ04 virus also contains G222 in its HA protein. To determine whether residue G222 in EA HA plays a role in pathogenicity, we generated a reassortant rgEAH1/pdm virus that comprised H1 gene from EA/FJ04 virus and all other genes from pdm/ BJ09 virus, which had the same genotype as Gt7 and showed binding affinity for SAα 2,6 and SAα 2,3 Gal linked receptors. Additionally, we introduced a HA-G222E mutation into the EA H1 gene of rgEAH1/pdm to generate a further reassortant (rgEAH1-G222E/pdm). We found that HA G222E mutation introduced into rgEAH1/pdm virus abolished binding affinity for SAα 2,3 Gal linked receptor (Fig. 4). We further determined that rgEAH1/ pdm virus caused 100% mortality in mice at challenge doses of 10 4 TCID 50 and10 5 TCID 50 (Fig. S4); however, rgEAH1-G222E/pdm mutant only caused slight weight loss and no mortality at the 10 4 TCID 50 challenge doses. Therefore, our results demonstrated that not only EA H1 gene but also HA 222G from parental EA/FJ04 virus contributed to the virulence of EA H1 reassortant virus by conferring, at least in part, affinity for the avian SAα 2,3-Gal linked receptor which is mainly localized in the porcine LRT 26 . However, HA222G did not enhance the virulence of wild type EA/FJ04 in mice, suggesting that combination of HA222G with genes from pdm/BJ09 was responsible for viral virulence in mice.  (Fig. 5). RNP activities of hybrid RNP complexes, however, varied substantially. All hybrid RNPs exhibited polymerase activities that were much less than the pdm/BJ09 virus RNP and more similar to the EA/FJ04 virus RNP complex. Therefore, we suggest that the selectivity for homologous RNP complex of pdm/BJ09 virus in reassortant viruses was likely due to its relatively high polymerase activity.

RNP of pdm
To evaluate the role of pdm/RNP in the pathogenicity of reassortants, a pdmRNP/EAH1N1 virus containing RNP genes from pdm/BJ09 virus and the remaining four genes from EA/FJ04 virus was generated and tested in mice. We found that the pdm/RNP genes significantly enhanced virus pathogenicity of pdmRNP/EAH1N1 in mice compared with the parental EA/FJ04 virus (Fig. S5), which suggests that pdm/RNP contributed to the virulence of reassortant virus in mice.

Low antigenic cross-reactivity of EA SIV with human influenza vaccine strains. H1N1/
pdm09-like and H3N2 antigenic components are presently incorporated in human seasonal vaccines. To assess cross-protection of such seasonal influenza vaccines against EA viruses, we performed hemagglutination inhibition (HI) assays with a panel of different sera (Table S1). The results showed that EA H1N1 viruses (Shangdong/07/2011 and Fujian/04/2007) reacted well with homologous virus sera with HI titer of 1:1280; however, they showed no reactivity with sera against human seasonal H1N1 (A/Tianjin-jinnan/15/2009, not H1N1/pdm09-like) and pdm/09 H1N1 viruses. This suggests that neither seasonal H1N1 nor pdm/09 H1N1 vaccine can currently provide effective humoral protection against EA/FJ04. Likewise, sera against the EA H1N1 viruses showed little cross-reactivity with pdm/09 H1N1 viruses of A/California/07/2009 and A/Beijing/7/2009, implying that EA H1N1 vaccine may in turn offer little protection to pdm/09 H1N1 virus infection. Notably, low cross-reactive HI antibodies against EA/FJ04 were observed in sera collected from children vaccinated by bivalent vaccine of H3N2 and H1N1 viruses, which is in accordance to previous studies 16,17,25 . Collectively, these results showed that EA H1N1 and pdm/09 H1N1, seasonal H1N1 and H3N2 viruses are antigenically distinct.

Discussion
Eurasian avian-like swine H1N1 viruses and 2009 pandemic H1N1 viruses are prevalent in pigs in Eurasia 7,20,21 . Co-infection with two different influenza viruses in a host can mathematically produce up to 256 different  reassortant viruses. In our co-infection of pigs with EA/FJ04 and pdm/BJ09 viruses, 55 out of 300 isolated virus clones were reassortants. We found that majority of the reassortant viruses (65%, 36/55) were generated from the LRT. Seventy eight % of LRT reassortants (28/36) carried the H1 HA gene from the parental EA/FJ04 virus. The 5 highly infectious reassortants genotypes of Gt2, Gt3, Gt7, Gt10 and Gt13 in guinea pigs were also amongst the most pathogenic in mice. These 5 genotypes all harbored the pdm/BJ09 RNP gene segments (PB1, PB2, PA and NP). We further demonstrated that pdm/09-RNP and EA-H1HA contributed to enhanced virulence and transmissibility in the reassortant viruses. No progeny EA/FJ04 virus was isolated which suggests that the competing pdm/BJ09 virus was much more competent at replication than EA/FJ04 virus. The present findings revealed high reassortment compatibility between EA and pdm/09 viruses in pigs, which could generate novel progeny viruses with enhanced virulence and transmissibility in mice and guinea pigs. A H3N2 variant derived from endemic H3N2 TR virus and pdm/09-origin virus in pigs has caused more than 340 human infection cases in the USA since 2011, which emphasized the role of pdm/09-origin internal gene in the genesis of influenza reassortants with the potential to infect humans 34 .
Pigs are commonly considered as "mixing vessels" for the reassortment of influenza viruses; however, how and where reassortment occurs in the pig remains unclear. Reassortment is likely to be influenced by infectious dose, cell tropism, host species and pre-existing immunity 35 . Ma et al. intratracheally co-infected pigs with an EA H1N1 virus and North American triple reassortant H1N1 virus. They found that reassortment mainly occurred in the lung of co-infected pigs 36 . In our study, we inoculated equal virus doses intranasally and intratracheally. Eight percent (19/240) of the isolated virus clones from the nasal passage and 60% (36/60) from the LRT were reassortants, indicating that most reassortants were isolated in the LRT.
In this study, the Gt7 reassortant viruses only containing H1 gene from EA/FJ04 showed increased pathogenicity in mice relative to either parental virus, which indicated that the EA H1 gene could be responsible for enhanced virulence in mice. Amino acid at G222 of pdm/09 HA was reported to be associated with severe disease and fatality in human patients 37,38 . HA-G222 variants have dual receptor specificity for SAα 2,3-Gal linked and SAα 2,6-Gal linked receptors, which confer expanded tissue tropism leading to increased pathogenicity in mice and ferrets 31,39 . We showed that a HA-G222E mutation in the EAH1/pdm reassortant resulted in significant reduction in pathogenicity in mice and eliminated its binding to SAα 2,3-Gal linked receptors. However, the HA222G isoform in parental EA/FJ04 virus did not confer elevated virulence in mice, which suggested that additional contextual genes in the reassortant virus are necessary for pathogenicity. The ability of EA-H1 reassortants to bind avian SAα 2,3-Gal linked receptors is expected to enhance their propagation in the porcine and human LRT where avian receptors-like are present 26,40 .
The other major virulence factor identified in the reassortants from co-infection with EA/FJ04 and pdm/BJ09 viruses is the dominance of the pdm/09 RNP complex. Polymerase activity of the pdm/09 RNP complex was significantly higher than those of parental EA and reassortant viruses. Collectively, the EA-H1 gene and members of pdm/09-RNP complex are major contributors to the virulence and transmissibility of EA H1 reassortants. Since recent surveillance data indicated that the internal genes of pdm/09 are widely established in pig populations in China 10 , emergence of novel reassortant viruses comprising EA-HA and pdm/09 RNP genes could pose potential threat to public health.
EA viruses have been circulating in pig populations for over 30 years. There have been occasional human cases of infection but no persistent human transmission has been documented 13,41 . Hence, there is little specific immunity against EA H1N1 viruses in the human population. Although, cross-reactive antibodies against EA H1N1 were found in a small proportion of human population, most do not have such antibodies 16,17,25 . We found that current seasonal H1N1 and H3N2 influenza vaccines do not appear to confer cross-reactive humoral protection against EA viruses. Therefore, there is a need of a novel EA H1N1 vaccine in anticipation of its possible emergence.
In summary, our data highlight the potential public health danger of novel viruses arising from reassortment in pigs between circulating EA swine H1N1 and pdm/09 H1N1 viruses. Such reassortants for which there is little apparent cross protection from current vaccines could be much more pathogenic and transmissible than parental viruses.
Co-infection of pigs and samples collection. Four to 5 weeks old Landrace-Large White cross pigs were sourced from a high health status farm. All animals were sero-negative for circulating influenza viruses (H1N1, H3N2, H5N1 and H9N2) by Hemagglutination inhibition (HI) assay. Six pigs were inoculated intranasally and equally intratracheally with 10 6.0 TCID 50 of EA/FJ04 and pdm/BJ09 in 3 ml phosphate buffered saline (1.5mL for each challenge route). All inoculated pigs were observed twice daily for clinical signs until the end of the experiment. Nasal washes were collected from all pigs every 24 h for 5 days post infection (dpi). Three pigs were euthanized at 3 and 5 dpi. The lungs were flushed with 50 ml fresh DMEM to obtain bronchoalveolar fluid (BALF) from each euthanized pig.

Virus plaque purification and identification of the genetic component. Each BALF sample and
nasal wash from co-infected pigs was serially diluted tenfold and inoculated onto a monolayer of MDCK cells grown in six-well plates. Single plaques were randomly selected (approx. 10 plaques per sample) and plaque purification was performed three rounds in order to purify the virus. Viral RNA was extracted from purified virus by using the QIAamp Viral RNA Minikit (Qiagen), and RT-PCR was performed to amplify the segments of influenza A virus as described previously 44 . Full genomic DNA was sequenced by Tingske Biological Technology Co. Ltd (Beijing, China). To identify the reassortant patterns of progeny virus, all sequence results were analyzed by using DNA star software (Lasergene 7.1).

Receptor binding assays.
Receptor binding specificity was analyzed by a solid-phase direct binding assay of biotinylated sialylglycopolymers as previously described 29 . Sodium salts of sialylglycopolymers poly-l-glutamic acid backbones containing N-acetylneuraminic acid linked to galactose through either an α -2,3 [Neu5Acα 2,3Galβ 1,4GlcNAcβ 1-pAP] or an α -2,6 [Neu5Acα 2,6Galβ 1,4GlcNAcβ 1-pAP] were kindly provided by the Glycotech, Gaithersburg, MD Ribonucleoprotein (RNP) minigenome luciferase assay. RNP genes (PB2, PB1, PA and NP) from EA/FJ04 and pdm/BJ09 viruses were cloned into pcDNA3.1 expressing vector. Polymerase activity of RNP from different gene combination of EA/FJ04 and pdm/BJ09 viruses was assayed in 293T cells as previously described 45 . Renilla luciferase plasmid pRL-TK (Promega) was also co-transfected to normalize the transfection efficiency. Firefly luciferase expression was normalized to Renilla luciferase expression (relative activity). The relative activity of the EA/FJ04 RNP was set as 100%, and the relative activities of the other reassortant RNP were presented relative to that.
Reverse genetics. All segments of EA/FJ04 and pdm/BJ09 were cloned into dual-promoter plasmid PHW2000 as described previously 29 ; all plasmids were verified by sequencing to preclude any mutations. A pdm-RNP/EAH1N1 reassortant virus containing RNP of pdm/BJ09 and other four genes from EA/FJ04 was generated by co-transfecting 293T cells with plasmids of PHW2000-EA/FJ04-HA, NA, M, NS and PHW2000-pdm/ BJ09-PB2, PB1, PA, NP. Similarly, a reassortant virus rgEAH1/pdm with HA gene from EA/FJ04 and the other seven genes from pdm/BJ09 was generated. To generate rgEAH1-G222E/pdm virus that contains a G to E substitution at residue 222 of HA, HA-G222E was introduced into PHW2000-EA/FJ04-HA by site-directed mutagenesis kit (Qiagen). All viruses were propagated in MDCK cells and sequence verified prior to animal studies.

Mouse experiments.
To evaluate the virulence of all reassortant viruses, groups of nineteen 5-week-old female BALB/c mice were intranasally inoculated with 50 μ L of 10-fold serial dilutions of each virus in PBS. Three mice infected with 10 5 TCID 50 virus from each group were euthanized at 4 dpi and lung tissues were harvested for virus titration and hematoxylin and eosin staining. Body weights of all mice were monitored daily for 14 dpi, and mice that lost more than 25% of their body weight were euthanized on humane ground. LD 50 were calculated and expressed as the TCID 50 value corresponding to 1 LD 50 .
In order to examine the roles of EA/FJ04 H1-222G and pdm/BJ09 RNP genes in the pathogenicity of reassortant viruses, groups of six mice were intranasally inoculated with rgEAH1-G222E/pdm, rgEAH1/pdm, pdmRNP/ EAH1N1 and EA/FJ04 at two infection doses of 10 4 and 10 5 TCID 50 , weight loss and survival rate were monitored daily until the end of this study (14dpi).
Guinea pig experiment. Female SPF/VAF guinea pigs (Hartley strain, 300-350 g, Vital River Laboratory Animal Technology), serum-negative for influenza virus, were used in these studies. Animals were anesthetized by muscular injection with zoletil 100 (tiletamine-zolazepam; Virbac S.A., Garros, France) (10-15 mg/ kg). Groups of three animals were intranasally inoculated with 10 5 TCID 50 of test viruses and housed in a cage per group. At 24 h post infection, three naïve guinea pigs were placed in the same cage and co-housed in this way for a total of 14 days. On 2, 4, 6, and 8 dpi, nasal washes of the three inoculated animals and three contact animals were collected and titrated. The area under the nasal wash curve between days 2 and 6 post inoculation (AUC 2-6 dpi ) was calculated from the virus titers of nasal wash of all guinea pig using GraphPad Prism 6 software as previously described 27 . All animals were euthanized at 14 dpi and sera were collected for seroconversion tests.