Lack of virological and serological evidence for continued circulation of highly pathogenic avian influenza H5N8 virus in wild birds in the Netherlands, 14 November 2014 to 31 January 2016

In 2014, H5N8 clade 2.3.4.4 highly pathogenic avian influenza (HPAI) viruses of the A/Goose/Guangdong/1/1996 lineage emerged in poultry and wild birds in Asia, Europe and North America. Here, wild birds were extensively investigated in the Netherlands for HPAI H5N8 virus (real-time polymerase chain reaction targeting the matrix and H5 gene) and antibody detection (haemagglutination inhibition and virus neutralisation assays) before, during and after the first virus detection in Europe in late 2014. Between 21 February 2015 and 31 January 2016, 7,337 bird samples were tested for the virus. One HPAI H5N8 virus-infected Eurasian wigeon (Anas penelope) sampled on 25 February 2015 was detected. Serological assays were performed on 1,443 samples, including 149 collected between 2007 and 2013, 945 between 14 November 2014 and 13 May 2015, and 349 between 1 September and 31 December 2015. Antibodies specific for HPAI H5 clade 2.3.4.4 were absent in wild bird sera obtained before 2014 and present in sera collected during and after the HPAI H5N8 emergence in Europe, with antibody incidence declining after the 2014/15 winter. Our results indicate that the HPAI H5N8 virus has not continued to circulate extensively in wild bird populations since the 2014/15 winter and that independent maintenance of the virus in these populations appears unlikely.


Introduction
Wild birds are the natural hosts of low pathogenic avian influenza (LPAI) viruses, which generally do not cause clinical signs of disease in these host species [1]. So far, virus subtypes H1 to H16 and N1 to N9 have been detected in wild birds, of which viruses of subtypes H5 and H7 have shown the ability to evolve to highly pathogenic avian influenza (HPAI) viruses in poultry, causing severe disease with high mortality in such animals. These HPAI viruses were historically mainly detected in rapidly contained sporadic outbreaks in poultry, until H5N1 viruses of the A/Goose/Guangdong/1/1996 (GsGd) lineage emerged in Asia in 1997. Subsequently, these viruses have continuously circulated in poultry with frequent detections in wild birds [2] and with significant expansion in global range.
HPAI H5N8 viruses of the GsGd lineage of clade 2.3.4.4 emerged in poultry and wild birds on multiple continents in 2014. The ancestral influenza H5N8 virus to the strains causing outbreaks from 2014 onwards was first detected in China in 2010 in a captive-held mallard (Anas platyrhynchos) [3]. In early 2014, HPAI H5N8 GsGd virus of clade 2.3.4.4 occurred for the first time in poultry in South Korea, soon after causing outbreaks also in Japan [4]. From late 2014 onwards, this virus spread to other areas of the world including Europe, North America, Russia and Taiwan [5][6][7][8]. The HPAI H5N8 virus detections in Europe were limited to sporadic cases in wild birds and a relatively small  [10][11][12] and ducks [11], the widespread detection and rapid global spread of HPAI H5 clade 2.3.4.4 viruses pose a potential threat to domestic and wild animals and should be studied further.
The major challenges in understanding the epidemiology of emerging influenza viruses in wild birds are the large numbers of potential host species and the usually short period of viral shedding, combined with the difficulty of catching and sampling representative numbers per species. For instance, mallards that were experimentally infected with HPAI H5N8 virus shed infectious virus in tracheal swabs for only up to 5 days post infection [11]. These impediments result in a low probability of detecting newly emerging avian influenza viruses in wild birds through active virological surveillance and result in a delay of implementation of effective control measures. Nevertheless, to date HPAI H5N8 virus has been detected in 30 wild bird species. In addition to the host species previously described [13,14], HPAI H5N8 viruses have been detected in wild bird species belonging to the orders Anseriformes in Asia (Aythya spp.) and North America (Branta spp.) [6]. In Europe, HPAI H5N8 viruses have been detected in bird species of the orders Anseriformes (Anas spp. and Cygnus spp.) and Charadriiformes (Larus spp.) [5,6,14].
To estimate the likelihood of the involvement of live wild birds in local and long distance movement of HPAI H5 viruses, information on recent exposure of wild bird populations to HPAI H5N8 viruses using serology, in addition to virology, would add substantial power to surveillance programmes. Studies with ferret sera have shown serological tests to have substantial discriminative power between antibodies directed to HPAI H5 viruses of different clades and LPAI H5 viruses using haemagglutination inhibition (HI) assays [12,15].
Although less is known about serology in wild birds, a study on wild birds sampled in Europe and Mongolia showed that antigenic differences between the haemagglutinin (HA) of classical Eurasian LPAI H5 viruses and GsGd lineage HPAI H5 viruses can be used to define bird populations in which HPAI viruses have previously been circulating [16]. With regard to HPAI H5N8 viruses specifically, a 2014 South Korean serology study showed evidence of a rise of H5 virus antibodies occurring in long distance migratory duck species after the onset of the HPAI H5N8 virus emergence in South Korea [4].
In this study, in response to the emergence of HPAI H5N8 virus in Europe, we present data on wild bird surveillance activities in the Netherlands, including results of virological and serological assays.

Ethical statement
The capture of free-living birds was approved by the Dutch Ministry of Economic Affairs based on the Flora and Fauna Act (permit number FF/75A/2009/067 and FF/75A/2014/054). Handling and sampling of freeliving birds was approved by the Animal Experiment Committee of the Erasmus Medical Centre (permit number 122-11-31). Free-living birds were released into the wild after sampling and all efforts were made to minimise animal suffering throughout the studies.

Study population
Immediately after the first detection of HPAI H5N8 virus in poultry in Europe, ongoing influenza surveillance activities in migrating and overwintering wild birds in the Netherlands were intensified (14 November 2014-13 May 2015). Hereafter, this period will be referred to as 'during the outbreak'. Surveillance activities in wild birds in the Netherlands were again intensified from the onset of the arrival of wild migrating birds a year after the initial HPAI H5N8 virus detection in Europe (1 September-31 December 2015). This period will be referred to as 'after the outbreak'. Sampled populations consisted of resident birds, partial migrants and long distance migrants. During both periods of Table 3 Details of positive control sera titres from experimentally infected ferrets, a domestic duck, and a domestic goose with one low pathogenic (LPAI) H5 and different highly pathogenic avian influenza (HPAI) H5 clades (n = 8 antisera) intensified surveillance, blood samples were obtained in addition to samples for virus detection. A matching historical set of serum samples was compiled based on similarity in species and family, hereafter referred to as 'before the outbreak' (2007-2013).

Sample collection
Wild birds were captured using duck decoys, clap nets, cannon nets, mist nets, leg-nooses, swan hooks, or manually. Birds were sampled routinely for virus detection using cloacal and/or oropharyngeal swabs as described elsewhere [14]. In addition, faecal samples were collected from a limited number of species for virus detection. Blood samples were collected for antibody detection. Blood samples were collected from the brachial or metatarsal vein and centrifuged at 3,000 rpm for 10 min in 0.8 mL gel separation tubes (MiniCollect tubes, Roche). Serum samples were stored below -20 °C until analysis.

Virus detection, isolation and characterisation
Samples for virus detection were analysed for the presence of HPAI H5(N8) virus using matrix-and H5-specific real-time polymerase chain reaction (RT-PCR) assays followed by H5 and neuraminidase sequencing as previously described [14]. Samples testing positive in matrix specific RT-PCR were inoculated in embryonated chicken eggs as described previously [17].

Antibody detection
Serum samples were first tested for the presence of H5-specific antibodies in an HI assay according to standard procedures [18]. Briefly, serum samples were incubated for 16 hours at 37 °C with Vibrio cholerae filtrate containing receptor-destroying enzyme to remove non-specific inhibitors of haemagglutination activity, followed by incubation for 1 hour at 56 °C. Twofold serial dilutions of serum samples with a start dilution of 1:20 were prepared using phosphate-buffered saline (PBS) in U-bottomed 96 well microtitre plates. Serum dilutions were incubated with four haemagglutinating units (HAU) of Madin-Darby canine kidney (MDCK) (all HPAI H5 clade viruses) or egg (A/Mallard/ Netherlands/3/1999) cultured virus for 30 min at 37 °C. A suspension of 1% turkey red blood cells (TRBC) was added to the serum-virus dilutions. After incubation for 1 hour at 4 °C, haemagglutination patterns were read. Negative controls, based on serum incubation without virus, were used to measure non-specific haemagglutination of each serum sample. Sera showing high background (i.e. high non-specific haemagglutination) were pre-treated with 10% TRBC for 1 hour at 4 °C and retested for the presence of H5-specific antibodies as described above. Serum samples from experimentally inoculated ferrets [12,15], a domestic duck, and a domestic goose were used as positive controls.  to enable this study within biosafety level 2 laboratories. HPAI H5 virus of clade 0 was excluded from the analyses due to high overall reactivity with all avian positive control sera as previously described [16] and thus of limited discriminative value.

Study population
A total of 11,355 birds were sampled for virus detection during and after the first detection of HPAI H5N8 viruses in poultry and wild birds in Europe. Of those, 5,387 birds were sampled during the outbreak and 5,968 after the outbreak. This report describes the results   Table 2). The majority of these samples were collected from birds wintering in Dutch wetlands.

Virus detection, isolation and characterisation
In addition to the two previously reported HPAI H5N8 virus-infected Eurasian wigeons detected in the Netherlands in November 2014 [14],  (Table 1).

Influenza A H5 virus clade-specific antibody detection
As shown previously, ferret antisera raised against prototype strains representing LPAI and HPAI H5 viruses of various clades showed almost exclusive reactivity with homologous viruses in HI assays [12] (Table  3). Importantly, a ferret antiserum raised against the clade 2. no cross-reactivity was seen with other prototype H5 strains (Table 3). These data indicate that the antigenic differences between clade 2.3.4.4 HA and HA of LPAI and HPAI viruses belonging to other clades were sufficiently large to allow serological discrimination by HI assay.

Table 6b
Titres of confirmatory haemagglutination inhibition and virus neutralisation assays for sera positive for highly pathogenic avian influenza H5 clade 2.3.4.4-specific antibodies in the initial screening, the Netherlands, 14 November 2014-31 December 2015 (n = 53 serum samples) p = 0.76 before vs during the outbreak; p = 0.39 during vs after the outbreak), while HPAI H5 clade 2.3.4.4-specific antibodies were detected exclusively in sera from five bird species, obtained during and after the HPAI H5N8 virus emergence in Europe (  (Table 6).

Discussion
In this report surveillance data for HPAI H5N8 in birds in the Netherlands are presented. In addition to bird samples previously investigated for the virus from 14 November 2014 to 20 February 2015, a new set of 7,337 samples obtained between 21 February 2015 and 31 January 2016 is analysed. One faecal sample obtained from a Eurasian wigeon (Anas penelope) on 25 February 2015 tested positive for the HPAI H5N8 virus, adding to the previous finding of the virus in two Eurasian wigeons in the country in late 2014 [14]. Virological surveillance moreover suggests that only very limited numbers of wild bird species were identified as potential hosts in Europe. Importantly, to the best of our knowledge, there are no reports of additional findings of HPAI H5N8 viruses in wild birds and poultry in Europe, since the last detection of the virus in February 2015 in the Netherlands.
Given the difficulty of detecting newly emerging HPAI virus strains in wild birds however, the application of a more sensitive and cost-effective method to detect potential host species is warranted. For this purpose, we performed serological assays specifically aimed to detect antibodies specific to HPAI H5 clade 2.3.4.4 viruses in a substantial number of sera obtained before, during, and after HPAI H5N8 emergence in the Netherlands. Three potential HPAI H5N8 host species were identified by HI assays and confirmed by VN assays; Eurasian wigeons, mute swans and common coots. Considering the results of virological studies performed worldwide since the onset of the HPAI H5N8 virus emergence in early 2014, the detection of HPAI H5 clade 2.3.4.4-specific antibodies in these species is not surprising. HPAI H5N8 virus was isolated from Eurasian wigeons in Russia [8] and the Netherlands [14], from mute swans in Sweden [6], and from a common coot in South Korea [21].
The serological results reported here were not entirely consistent between HI and VN assays and between HI assays performed in two different laboratories. Although low HI titres (e.g. in Eurasian wigeons) were reproducible within a laboratory with the same HI assay and a VN assay, they were not detected by HI assay in a second laboratory, potentially due to differences in the methods used and hence differences in sensitivity and specificity. High antibody titres in mute swan sera were reproduced by HI assay in a second laboratory and by VN assay, but low antibody titres in mute swans were not always reproduced. While it is thus clear that individual HI titres in avian sera obtained from a single test cannot be used reliably for diagnosis, use of serum panels from cohorts of birds, use of multiple tests to cross-validate results, a panel of relevant viruses and use of collections of control antisera may still enable the use of serological tests in support of HPAI H5 surveillance studies.
Previously, HI assays were shown to be discriminative enough to detect antibodies in serum samples collected from free-living wild birds in Europe and Asia to be directed to either HPAI or LPAI H5 viruses. However, widely varying results were obtained as far as HPAI H5 clade-specific antibodies were concerned [16]. In this study, most birds that tested positive for HPAI H5 clade 2.3.4.4-specific serum antibodies showed relative low HI titres. This is in accordance with findings based on experimental HPAI H5N8 virus infections of ferrets [10][11][12], possibly indicating low immunogenicity upon infection. In addition, there is limited knowledge about the longevity of avian antibodies after naturally occurring infection with avian influenza viruses. Antibodies specific to LPAI viruses were detected up to several months after experimental or natural infection [22][23][24], whereas little is known about the duration of detection of antibodies specific to HPAI viruses with a reported maximum of detection of 28 days after experimental infection in domestic ducks [25]. To date, there is no knowledge on the effect of a prior exposure to an unrelated subtype or on the phenomena of antigenic sin in avian species. Hypothetically, low immunogenicity in combination with decreasing titres in time could be an explanation for the low incidence and relative low titres of antibodies detected in wild bird sera in this study.
In conclusion, our results provide evidence that clinically unaffected long distance migratory and local wild birds sampled in the Netherlands during the H5N8 outbreak late 2014 and early 2015, and again late 2015, have been exposed to HPAI H5N8 or closely related HPAI H5 clade 2.3.4.4 viruses and seroconverted upon exposure. Since HPAI H5N8 virus has not been detected in Europe since early 2015 and because HPAI H5 clade 2.3.4.4-specific antibody incidence decreased in time, we conclude that the virus has not circulated extensively at the breeding grounds in summer and upon the return of the birds to their wintering areas in the 2015/16 winter. As a consequence, the newly emerging HPAI H5N8 clade 2.3.4.4 virus subtype appears to have already disappeared from European wild birds indicating that sustained transmission and independent maintenance may be less likely. This is an important consideration in the ongoing evolution and ecology of these viruses in wild birds and the potential risks they pose for introduction to poultry and the pathways through which they might spread. Finally we recommend that serological tools be further optimised, harmonised, and validated for avian influenza surveillance studies in wild birds.