Background
Swine influenza is an economically important disease of pigs caused by infections with influenza A viruses (IAV). Economic losses are caused by retarded growth of fattening pigs due to influenza-induced or -aggravated respiratory disease. In addition, febrile influenza virus infections in sows may cause fertility problems [
1]. Influenza viruses are an important factor in the polymicrobial respiratory syndrome of swine [
2]. In contrast to human influenza, infections in swine do not appear to be seasonal, and virus circulation, especially in larger herds, is observed year-round [
3]. Control of swine influenza is difficult and requires strict zoosanitary measures and herd vaccination programs. Licensed inactivated whole virus vaccines are available but high and continuous vaccine coverage within herds is required [
4].
In addition to being affected by disease, pigs are considered to be an integral part of the wider epidemiology of influenza, bridging the avian influenza world to mammalian influenza. The porcine respiratory epithelium is lined by cells which express the two sialic acid glycan receptor structures to which avian- or mammalian-adapted IAV bind [
5]. As such, pigs can be infected by IAV of human and of avian origin and this provides opportunities for reassortment between these viruses [
6,
7]. Following historic transspecies transmission events of IAV from human or avian sources to pigs some of these viruses have established stable circulating lineages in swine populations worldwide. These porcine lineages continue to reassort amongst each other and with other IAV of more recent human or avian origin. In Europe, this scenario has lead to current presence of at least four distinct lineages of swine influenza viruses in pig populations [
8,
9].
Since the late 1970s IAV of subtype H1N1 of purely avian origin (H1N1av) dominate the influenza epidemiology in swine in many European countries including Germany. This lineage, referred to as H1N1av, has fully adapted to swine and can be distinguished genetically and antigenically from current avian-adapted H1N1 viruses. Viruses of this lineage have sporadically been detected also in humans and turkeys in Europe due to single transmission events from infected pigs [
10,
11]. A second porcine lineage consists of viruses of subtype H3N2. The progenitor of the currently circulating porcine H3N2 strains originated from human-adapted H3N2 viruses which had caused the Hong Kong flu pandemic in 1968 [
8]. In the early 1980s the descendants of this virus reassorted with H1N1av and, apart from the hemagglutinin H3 and the neuraminidase N2 segments, all further six genome segments were replaced with those of H1N1av [
8]. In the early 1990s, a new porcine triple reassortant virus, H1N2, arose from reassortment events between human seasonal H1N1 and H3N2 and porcine H1N1av viruses. This porcine H1N2 virus carried hemagglutinin (HA) and neuraminidase (NA) of human origin and the cassette of six further segments of H1N1av [
8]. These three lineages continue to co-circulate at varying prevalences in different European countries.
In 2009, a new human pandemic H1N1 strain (H1N1pdm) emerged. This virus carried reassorted gene segments from several American and Eurasian swine influenza lineages and was rapidly introduced from the human population to pigs [
8]. Pigs proved to be highly susceptible to this virus and stable transmission chains were easily maintained [
12]. To date H1N1pdm appears to circulate independently from the human population in swine in several countries worldwide. Recently we and others found evidence for the emergence of reassortants between H1N1pdm and authentic porcine influenza virus lineages in Germany. In particular, a reassortant lineage of subtype H1pdmN2 which carried seven segments of the H1N1pdm virus and a neuraminidase of subtype 2 that was derived from different porcine or human HxN2 lineages, circulated stably [
9,
13].
Measures aiming at control of swine influenza must be based on subtype-specific virological and serological diagnosis. Real-time RT-PCR (RT-qPCR) has provided ample applications for rapid and sensitive molecular virological diagnosis [
14]. Serology in swine influenza has been found useful for retrospective epidemiological investigations e.g., [
15], estimation of disease incidence e.g., [
16], and control of vaccination success e.g., [
17]. Several commercial ELISAs for detection of generic IAV nucleocapsid protein (NP)-specific antibodies in pigs are available and recommended for use e.g., [
18]. With regard to detection of antibodies at the subtype-specific level the hemagglutination inhibition assay (HI) is still held gold standard despite several draw-backs of this method including time- and labour-consuming performance and dependence on labile and difficult-to-standardize components (viral antigen, erythrocytes). Furthermore, due to interference of antibodies against the viral neuraminidase component interpretation of HI-results are particularly difficult [
19,
20]. Antibodies to the IAV hemagglutinin protein (HA) become detectable by HI assay from the second week post infection on and HI titers correlate with protection from clinically overt disease [
21,
22].
Despite mentioned problems of HI assays very few swine influenza ELISA applications aiming at subtype differentiation at the antibody level have been reported and the current commercially available assays for subtypes H1 and H3 have not superseded HI assays, at least in Europe [
23]. The reported lack of sensitivity of these assays may be related to the American origin of the IAV isolates used which are antigenically distinct from those circulating in Europe. Low specificity of these assays may be caused by use of whole virion preparations which contain group specific antigens such as the nucleocapsid protein.
Here we show that recombinant HA1 antigen of European swine influenza viruses which was bacterially expressed and refolded in vitro can be used in indirect ELISAs for detection and differentiation of subtype-specific antibodies in porcine sera.
Discussion
Serodiagnosis of porcine influenza virus infections in Europe and elsewhere is significantly challenged by co-circulation of several subtypes, and use of multivalent vaccines further complicates this situation. While generic antibodies directed against the well conserved influenza virus NP protein can be detected by commercial blocking ELISAs, the differentiation of subtype-specific antibodies requires use of the fastidious HI assay. Subtype-specific ELISA assays suitable for high throughput investigations of porcine sera would aid in promoting more intense studies on porcine influenza seroepidemiology.
Aiming to develop such assays we have successfully expressed recombinant full-length NP and HA1 fragments in bacteria and refolded proteins
in vitro. Recombinant proteins were co-translationally mono-bitoinylated which facilitated purification and binding to solid, streptavidin-coated supports. Previous work by [
26] has shown that bacterially expressed HA proteins can be refolded to acquire native conformation. Recombinant proteins representing recent isolates of the major subtypes and lineages of porcine influenza virus currently circulating in Germany were recognized by specific porcine or mustelid immune sera in Western blotting and indirect ELISA. Dynamics of generic and subtype-specific antibody development in experimentally infected swine as measured by recombinant indirect ELISAs fully paralleled results obtained by a commercial NP blocking ELISA and homologous HI assays.
However, despite use of the less conserved HA1 section of the HA glycoprotein as diagnostic antigen residual cross reactivity between the different subtypes was noticed in both Western blot (Figure
1) and indirect ELISA (Figure
2) even when using experimental post infection sera which were reasonably discriminatory between subtypes in HI assays (Table
2). Thus, a number of conserved epitopes exists between the different subtypes in the HA1 which is not detected by the functional HI assay. Yet, a considerable number of subtype-specific epitopes must have been represented in the recombinant HA1 proteins as well since, in indirect ELISA, the homologous HA1-serum pairs always resulted by far in the highest signal intensity (Figure
2). Human pandemic H1 HA1 represented by isolate R26/11 and that of the porcine-adapted variant H1pdmN2 R2035/11 were indistinguishable by Western blot and indirect ELISA although slight antigenic differences have been reported when using an HI assay [
9].
The comparative examination of porcine field sera showed a variable agreement for the different recombinant antigens when individual sera were compared. In particular, ELISAs based on recombinant H1av and H1N2 revealed only moderate agreement when compared to HI. Fewer sera scored positive in the indirect ELISAs compared to HI assays. The majority of sera missed by the indirect ELISA showed low HI titres (3 or 4 log
2) while reasonable correlation was seen between S/P ratios and HI titres for higher-titred sera (Figure
5). The blurring of results obtained with low-titred HI-positive sera may be due to low sensitivity of the indirect ELISAs but could as well have been caused by lack of specificity of the HI assays; problems with reproducibility and standardization of HI assays are notorious, especially when testing low-titred sera [
20]. In addition, the HI assays here were carried out with antigens selected and used in routine diagnosis while the recombinant antigens were produced from more recent circulating viruses and differed slightly in HA1 amino acid sequences (not shown). This may have introduced further discrepancies as observed for individual sera. However, when results were compared on a herd basis a full match between HI and indirect ELISA results was evident (Table
5). Herds found to be seropositive by HI for a certain subtype were similarly positive in the respective HA1 indirect ELISA. Moreover, herds negative by HI for a certain subtype tested negative by the corresponding indirect ELISA. This indicates that on herd base the HI assay may be replaced, without loss of diagnostic quality, by the indirect ELISAs.
Material and methods
Virus and cell culture
Influenza A viruses were propagated in serum-free MDCK cell cultures in the presence of TPCK-trypsin as detailed elsewhere [
11]. Isolates were obtained from the virus repository maintained at the Friedrich-Loeffler-Institut. Molecular characteristics of recent porcine field isolates have previously been reported [
9]. A list of viruses used in this study for production of recombinant proteins is provided in Table
1.
Bacterial expression, in vivobiotinylation and purification of influenza virus HA1 and NP proteins
The HA1 fragments of the viral hemagglutinin open-reading frames (ORF) were cloned into the pET19b vector by a target-primed technique using Phusion polymerase amplification and
Dpn I digested amplificates [
27]. Sequences of primers are available on request. Expressed sequences stretched from the first amino acid of the mature protein to the arginin residue immediately proximal to the first glycin residue of the HA2 fusion peptide. Downstream of this arginin residue an Avi-Tag consensus sequence [
28] was inserted. The central lysin residue of the 15 amino acid Avi-Tag sequence provides an acceptor site for covalent linkage of D-biotin which is specifically catalyzed by the bacterial biotin transferase
BirA[
29]
.
HA1-pET19b expression constructs were co-transformed into Rosettagami E. coli with plasmid pBIRAcm (Avidity, Aurora, CO, U.S.A.) for overexpression of BirA. Dually transformed cells were selected using ampicillin and chloramphenicol (CM). Since CM is also required to maintain the genotype of Rosettagami E. coli cells, presence of both plasmids in selected colonies had to be confirmed by plasmid/insert-specific PCRs (primer sequences available on request). TYH medium supplemented with D-biotin at a concentration of 50 μg/ml was used for expression of co-translationally mono-biotinylatied Avi-tagged recombinant protein. The full length ORF of the nucleocapsid gene of the porcine influenza virus isolate R1738/10 was cloned and expressed similarly. However, the Avi-Tag was placed at the N-terminus of the protein. Lysates of Rosettagami cells transformed with plasmid pBIRAcm and an empty pET19b vector were used as a negative expression control.
Monobiotinylated bacterially expressed recombinant proteins were purified from inclusion bodies (IBs) by centrifugation and washing steps as previously described [
26]. Proteins sequestered in purified IBs were then subjected to solubilization in 6 M guanidin-HCl and refolding using a panel of up to 30 different primary and up to nine secondary buffer conditions in a stepwise solubilization strategy using the ProteoStat kit (Enzo, Lörrach, Germany). Protein solubilization and reactivity were screened with conformation-dependent monoclonal antibodies and specific polyclonal sera in ELISA to sort out optimal refolding conditions for each of the recombinantly expressed proteins. Here, refolding conditions were used which had been validated using avian influenza virus H5 HA1 protein and two conformation-dependent monoclonal antibodies, 3H12 and 5 F3 (see [
30], for properties of monoclonal antibodies). Final concentrations of recombinant proteins in appropriate refolding buffers were measured using a Coomassie protein assay kit (ThermoScientific, Rockford, IL, U.S.A.).
Production of subtype-specific antisera
Pigs or ferrets were experimentally infected by the oronasal route with 10
6 TCID
50 of MDCK cell culture-grown influenza viruses in 1 ml cell culture medium using a nebulizer device (Wolfe Tory Medical, Salt Lake City, Utah) as previously described [
31]. All experiments had received legal approval by an ethics commission (LALLF M-V/TSD/7221.3-2.5-004/10). Prior to infection animals were tested seronegative for influenza NP-specific antibodies in a commercial blocking ELISA (ID.Vet). Virus isolates used for infection are listed in Table
2. Blood samples used in further serological studies were obtained on day 21 post inoculation (p.i.). Reactivity of post infection sera in hemagglutination inhibition assays is detailed in Table
2.
Origin of field sera
Porcine field sera from 12 swine holdings in Germany were submitted for routine diagnostic procedures. The history of these holdings for vaccination against influenza and/or clinical episodes of influenza virus infection was not documented.
Indirect ELISA
Bacterially expressed proteins in refolding buffer were adjusted to a concentration of 5 μg/ml using TRIS-buffered saline (TBS). A total of 100 μl per well was used for binding to streptavidin-coated plates for 2 hours at room temperature or overnight at 4°C. Each plate included different recombinant antigens for each row with following strains: A:SIV/R1738/10 (H1N1),B: SIV/ R1207/11 (H1N2),C: –SIV/ R26/11 (H1N1pdm)D: SIV R2035/11 H1pdmN2, E: SIV/– R1931 /11 (H1N2), F: R76/11 (H3N2), G: R96/11 (H3N2), H: R1738/10 nucleocapsid protein. After washing wells were blocked using 5% nonfat milk-TBS containing 0.05% Tween 20 (TBST) (see Postel et al., 2011) for 2 hours at room temperature and then washed four times with TBST. Individual sera (100 μl per well, pre-diluted 1:200 in sample dilution buffer [ID.Vet, Montpellier, France]) were pipetted into columns (1A-1H) of the microtitre plate. This procedure assured that the reactivity of each serum against all antigens was measured in the same plate. Sera were incubated at room temperature for 1 hour. Wells were washed again four times with TBST before 100 μl of appropriately diluted goat-anti-swine IgG peroxidase conjugate (Dianova) was added for one hour at room temperature. Antibody was removed and after a final washing cycle with TBST, 50 μl of chromogenic TMB substrate was added. OD
450 values were measured after 10 minutes of incubation and addition of 50 μl of 1 N H
2SO
4 to each well. Results were calculated and expressed in S/P units:
Avidity measurement of sera by indirect ELISA
The indirect ELISA was performed as described above. However, after incubation of sera in the wells a washing step using urea in TBST was carried out. Different urea concentrations (0.5 and 2 M) and incubation times were evaluated, Final assays were carried out with 6 M urea for ten minutes at room temperature. Consecutive washing steps and conjugate incubation were carried out with TBST without urea. The sera were tested in parallel with and without the urea-buffer washing step and an avidity index (AVI) was calculated:
Generic nucleocapsid protein blocking ELISA (NP-bEIA)
For detection of group specific antibodies a commercial NP-bEIA was purchased (ID.Vet, Montpellier, France) and used according to recommendations of the manufacturer. Accordingly, samples were considered positive if the S/N (sample OD450/negative-control OD450 × 100) ratio was less than 45%, negative if the S/N ratio was more than 50%, and doubtful if the S/N ratio was between 45% and 50%.
Hemagglutination inhibition assay (HI)
HI assays were performed according to O.I.E. recommendations essentially as described by [
31]. Four hemagglutinating units of cell culture-grown influenza viruses were used throughout. All porcine and ferret sera were heat-inactivated for 30 minutes at 56°C and treated with receptor-destroying enzyme (neuraminidase from
Bacillus subtilis).
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
ZN, CG and TCH conceived the study and drafted the manuscript. ZN carried out the molecular and serological work, and analysed samples. SK participated in the molecular and serological work. EL carried out animal experiments and provided samples. MB conceived the study, provided funds and edited the manuscript. All authors read and approved the final manuscript.