Evidence of exposure of domestic pigs to Highly Pathogenic Avian Influenza H5N1 in Nigeria

Avian influenza viruses (AIV) potentially transmit to swine as shown by experiments, where further reassortment may contribute to the generation of pandemic strains. Associated risks of AIV inter-species transmission are greater in countries like Nigeria with recurrent epidemics of highly pathogenic AI (HPAI) in poultry and significant pig population. Analysis of 129 tracheal swab specimens collected from apparently healthy pigs at slaughterhouse during presence of HPAI virus H5N1 in poultry in Nigeria for influenza A by RT-qPCR yielded 43 positive samples. Twenty-two could be determined by clade specific RT-qPCR as belonging to the H5N1 clade 2.3.2.1c and confirmed by partial hemagglutinin (HA) sequence analysis. In addition, 500 swine sera were screened for antibodies against influenza A virus nucleoprotein and H5 HA using competition ELISAs and hemagglutination inhibition (HI) tests. Serologically, 222 (44.4%) and 42 (8.4%) sera were positive for influenza A virus NP and H5 antibodies, respectively. Sera reacted to H5N1 and A/H1N1pdm09 strains by HI suggesting exposure of the Nigerian domestic pig population to these viruses. We report for the first time in Nigeria, exposure of domestic pigs to H5N1 virus. This poses potential public health and pandemic risk due to interspecies transmission of avian and human influenza viruses.


Molecular characterisation and phylogenetic analyses.
Apparently healthy domestic pigs from backyard and free range husbandry system in Plateau State, north central Nigeria slaughtered at the Jos central abattoir from 2015-2016 were investigated. Forty three of 129 (33%) tracheal swab specimens collected from these pigs were positive for influenza A virus (M and NP) genes by conventional and RT-qPCR assays. In order to avoid multiple freeze-thawing steps, and to save expensive RT-qPCR material, only samples with lower Cq values were used for further characterisation. Also, not all of these positive samples were investigated in the subsequent tests because of low volume of remaining RNA. Thirty positive samples were tested in multiplex RT-qPCR assays specific for HA and NA subtypes of European porcine influenza viruses. All of them were negative in the HA test (i.e. H1pdm-, H1av-, H1hu-and H3-negative), but 26 were positive for N1. Out of the 23 samples tested by the H5N1 duplex assay, 22 were positive in the H5 reaction and two in the N1 part (Table 1). In addition, these 22 samples could be determined as belonging to the H5N1 clade 2.3.2.1c by a clade-specific RT-qPCR 13 .
Query BLAST of 5 sequences that were generated via conventional RT-PCR and Sanger sequencing of an HAII fragment (about 650 nucleotides) returned closely related viruses. This included A/Chicken/ Nigeria/15VIR399-1/2015, other poultry sequences from Nigeria and from other African countries in 2015 as well as from Asian countries (all H5N1) with 99% sequence homology. As determined with the clade-specific RT-qPCR and by sequence analysis, the swine H5N1 virus genomes detected in this study clustered together and along avian H5N1 clade 2.3.2.1c.
Unfortunately, the NA gene sequencing attempts repeatedly failed even with different primer sets including conventional panNA RT-PCR, both variants 14 and new designed and customized primers for N1 based on avian H5N1 sequences from western Africa, 2015. Attempts to isolate the virus from swabs in chicken embryonated eggs and tissues in MDCK (+) cells (cell line number 0671 of the Collection of Cell Lines in Veterinary Medicine CCLV, FLI) were also not successful.
The analysis of the nucleotide sequences of the swine viruses showed only three silent nucleotide exchanges in the HAII gene fragment compared to A/Chicken/Nigeria/15VIR399-1/2015 H5N1. These include G, A, and C replacing A, G, and T respectively in position 1152, 1177 and 1335 of the HA gene. The partial sequences of the HA gene of the five amplifiable H5N1 has been deposited in GISAID-Epiflu TM database and assigned accession number EPI262933 (A/swine/Nigeria/49/H5N1), EPI262932 (A/swine/Nigeria/55/H5N1),

Discussion
Infections of swine with AIV in nature seem to be rare, and only very few studies have described detection of HPAIV H5N1 in healthy domestic pigs in Indonesia and China 15,16 . In those studies, the virus appeared to have been transmitted from poultry. Our investigation revealed natural exposure of HPAIV H5N1 clade 2.3.2.1.c possibly from infected poultry in Nigeria to domestic pigs as we could find H5N1 RNA in 43 tracheal specimens of In an RT-qPCR 3plex assay for the detection of swine influenza NA subtypes, it was possible to detect 22 N1-positive samples. However, sequencing of the neuraminidase failed due to the compromised quality of the samples as a result of multiple freeze thawing. Therefore, the sequence identity of the N1 could not be ascertained. Further investigation through surveillance, virus isolation and full genome sequencing is required. Another limitation was the inability to sequence the internal genes due to low quality of the remaining RNAs and low volume of original material which restricted the range of investigations including virus isolation attempts. Significant exposure to influenza viruses was demonstrated serologically with 222 (44.4%) and 42 (8.4%) out of 500 sera samples being positive in the competitive ELISAs for the detection of NP and H5 antibodies, respectively. This was also confirmed by HI assay against H5 and H1 antigens for selected samples. Interestingly, the HI test also indicated cross reactive antibodies in low titres to other serotypes of porcine influenza viruses including H1avN1av, H1huN2, and H3N2 in the study population (Table 3a). However, the sera were almost exclusively strongly positive for the A/H1N1pdm09 antigen. The serological results also confirm exposure of swine to HPAIV H5N1 and H1N1pdm09 in Nigeria.
In a study published in 2001 17 on co-circulation of avian H9N2 and contemporary human H3N2 viruses, possible inter-species transmission and genetic reassortment in pig was suggested. Also in 2011, influenza A H3N2 variant virus infection in a person who attended an agriculture fair in the USA was detected for the first time and characterised as a reassortant of a SIV H3N2 with A/H1N1pdm09 18 which was further spreading among swine and humans in following years 19,20 . These reports raised concerns on the possible emergence of zoonotic influenza virus from swine as consequences of inter-species transmission 18 . Our results provide additional scientific evidence of natural exposure of HPAIV subtype H5N1 currently circulating in Nigerian poultry to pigs in the same epidemiological zone. The risk of possible gene reassortments may be due to presence of different IAV subtypes in domestic animals in the region. Furthermore, this study shows that HPAI H5N1 clade 2.3.2.1c was present in swine population in the study area as also revealed serologically through detection of H5 antibodies.
We speculate that exposure in swine is due to intensity of AIV outbreaks in the region and the breakdown in control measures. Because AIV detection in swine was not only once and not in the same group of pigs based on different sampling dates and ownership, we can also speculate that exposure was more than once within the epidemiological zone likely due to between-farms transmission similar to observations by Ssematimba et al. 21 . Mechanical carriage may be possible, but this does not exclude other routes of transmission. Specifically, tracheal swabs (lower respiratory tract) not nasal samples were collected. In addition 8.4% seroconversion to H5N1 in pigs was also observed in the sampled population.
In a recent experiment by Kaplan et al. 22 , it was observed that clade 2.3.4.4 HPAIV H5 viruses were poorly adapted for replication and transmission in pigs. In that study, inoculated pigs did not produce HI antibodies. Comparatively, our investigation showed that H5 clade 2.3.2.1c produced high antibody titre in naturally exposed pigs and corroborate observation by Kaplan et al. 22 , that other lineages of H5 and H1N1pdm09 are more reactive serologically in swine. Though as also observed by Balzli et al. 23 . AIV may produce lower respiratory tract infection in pigs based on detection of IAV by RT-qPCR in bronchoalveolar lavage fluid (BALF). However poor replication efficiency shown by high Cq values and lack of virus isolation (observed in our investigation and in the data by Kaplan et al.) suggests low adaptation in swine host 22 . Though domestic pigs showed low susceptibility to HPAI H5N1 24,25 , genetically engineered reassortants of H5N1 and H1N1pdm09 have been shown experimentally to have potential for mammalian transmissibility in guinea pigs 26 .
The virus RNA detected in our study were obtained from apparently healthy pigs; asymptomatic and therefore silent transmission of a potentially zoonotic HPAIV in a country with a large population of pigs in intensive and free range husbandry bears public health risks 27,28 . Thus, the large concentration of people and livestock in urban and semi-urban settings provides a veritable ground for emergence and spread of a zoonotic pathogen at the human-animal interface. This was emphasised in a previous study on the prediction and prevention of the next pandemic 29 and also verified by serological investigations performed among Nigerian poultry workers which demonstrated H5 reactive sera within this risk group 30 . In addition, it was also observed that the emergence of zoonotic pathogens that have dominated the pandemics of the past 100 years correlates strongly with human population density and global distribution of animals 29 in the current study suggests possibilities of enhanced factors including increased contact rates between avian and swine and widespread H5N1 virus dissemination in the environment, which could aid exposures and biological adaptation in swine host. Furthermore, a large proportion of serum samples showed antibodies against IAV most of which were reactive to A/H1N1pdm09 and therefore calls attention to the theoretical possibility of simultaneous infections with different avian and mammalian influenza A viruses in the population, the basis for reassortment of genes in pigs being a mixing vessel. This also indicates that possibly, human pandemic H1N1 infections (or porcine adapted variants thereof) play the main role for swine influenza in Nigeria. Alternatively, other swine-adapted viruses may circulate which are antigenically distinct from those of Europe or other regions.
In conclusion, the exposure to H5N1 HPAIV from avian to swine, poses yet unknown public health risk if subclinical infection in pigs is sustained and may combine with A/H1N1pdm09 also detected in the population. This might allow the generation of new potentially zoonotic viruses by gene reassortment. The potential of this IAV transmission to humans thus requires further monitoring and investigation of the swine population in Nigeria, especially in regions with high swine density and endemicity of HPAIV in poultry though the risk is up to now theoretical.

Methods
Sample collection. In the study, 129 tracheal swabs and 129 lung tissues were concurrently and randomly collected through convenience sampling from apparently healthy pigs at a slaughter house in December 2015 through February 2016 in the period of avian influenza epizootic in Nigeria (Figs 2 and 3). Different owners (pig farmer/trader) brought 3 to 10 pigs in one batch to the abattoir for contract slaughter during each slaughtering day. Each batch is followed in the slaughtering line and 3 tracheal and 3 lung samples (sometimes less) were collected from each batch/farm/trader. Sampling limitation included inability to collect specimens from every batch due to owners' resistance on account of perceived lack of compensation in the event of a positive case. This procedure was carried out on a weekly basis. These animals originated from free range and backyard pig farms in Plateau State, North central Nigeria (Fig. 3). Animals were transported from backyard and free range pig holdings in the region to the central city abattoir where only cattle, small ruminants and pigs were slaughtered in separate lines. The central city abattoir is located on approximate 400 × 200 square metre land area. The build-up facility covers up to 200 × 100 square metre and the pig slaughter section is about 80 × 15 square metre. The abattoir handles up to 70% of bred pigs in Plateau state and it serves as the major outlet for pork in the region. In the same longitudinal study 300 sera were obtained from slaughtered pigs in the facility over a period of six months from February to July 2016. Two hundred sera collected in 2013 from the same abattoir and in Enugu (southern Nigeria) were also included in the investigations for comparison. Pre-extracted nucleic acid, swabs and tissues were stored at −80 °C while sera were stored at −20 °C until they were all shipped on dry ice for analyses at the OIE/FAO and National reference laboratory for avian influenza, Friedrich-Loeffler-Institute (FLI), Isle of Riems in Germany.

Serology.
To assess levels and extent of exposure to influenza viruses in domestic pigs in the study area, 300 recently collected sera (2016) along with 200 archived sera (2013) were analysed for Influenza A, NP and H5 specific antibodies, respectively, using "ID-Screen" Competition ELISA assays obtained from IDVet according to manufacturer's instructions. A subset of ELISA positive sera were subtyped by hemagglutination inhibition (HI) assay 44  The datasets generated and/or analysed during the current study are available in the GISAID-EpiFlu TM -www. platform.gisaid.org/ with the following accession numbers: EPI262932, EPI262933, EPI262934, EPI262935, EPI262950.