Rapid detection and subtyping of European swine influenza viruses in porcine clinical samples by haemagglutinin‐ and neuraminidase‐specific tetra‐ and triplex real‐time RT‐PCRs

Background A diversifying pool of mammalian‐adapted influenza A viruses (IAV) with largely unknown zoonotic potential is maintained in domestic swine populations worldwide. The most recent human influenza pandemic in 2009 was caused by a virus with genes originating from IAV isolated from swine. Swine influenza viruses (SIV) are widespread in European domestic pig populations and evolve dynamically. Knowledge regarding occurrence, spread and evolution of potentially zoonotic SIV in Europe is poorly understood. Objectives Efficient SIV surveillance programmes depend on sensitive and specific diagnostic methods which allow for cost‐effective large‐scale analysis. Methods New SIV haemagglutinin (HA) and neuraminidase (NA) subtype‐ and lineage‐specific multiplex real‐time RT‐PCRs (RT‐qPCR) have been developed and validated with reference virus isolates and clinical samples. Results A diagnostic algorithm is proposed for the combined detection in clinical samples and subtyping of SIV strains currently circulating in Europe that is based on a generic, M‐gene‐specific influenza A virus RT‐qPCR. In a second step, positive samples are examined by tetraplex HA‐ and triplex NA‐specific RT‐qPCRs to differentiate the porcine subtypes H1, H3, N1 and N2. Within the HA subtype H1, lineages “av” (European avian‐derived), “hu” (European human‐derived) and “pdm” (human pandemic A/H1N1, 2009) are distinguished by RT‐qPCRs, and within the NA subtype N1, lineage “pdm” is differentiated. An RT‐PCR amplicon Sanger sequencing method of small fragments of the HA and NA genes is also proposed to safeguard against failure of multiplex RT‐qPCR subtyping. Conclusions These new multiplex RT‐qPCR assays provide adequate tools for sustained SIV monitoring programmes in Europe.


| INTRODUCTION
Influenza A virus (IAV) infections have a significant clinical impact on domestic swine populations. This is mainly related to acute respiratory disease which is frequently exacerbated by synergistic viral and bacterial co-infections. Economic losses for the pork producing industry due to swine influenza virus (SIV) infections can be substantial. [1][2][3] Swine are susceptible to infection from various IAV subtypes of both mammalian and avian origin. This is due to the expression of cellular glycan moieties (alpha 2-6 or 2-3-linked terminal sialic acids, respectively) in the porcine upper respiratory tract that act as recep- (so-called reverse zoonotic transmission). 11 However, direct infection of swine with IAV of purely avian origin has also occurred. 12 In swine, some of these viruses swiftly established porcine-adapted, independently circulating lineages which reassorted amongst each other. As a result of these processes, a variety of sublineages and genotypes of SIV is now prevalent in swine populations worldwide, but with some geographic restrictions,. 13,14 Some of these viruses may have zoonotic potential and can be retransmitted into the human population. [15][16][17] Until 2009, the IAV subtypes H1N1, H1N2 and H3N2 circulated in domestic pigs at variable rates in Europe. The current enzootic Eurasian SIV H1N1 lineage is of purely avian origin and was introduced to pigs from unknown bird populations in the late 1970s. These viruses are referred to as "avian-like" (av) H1N1 and are genetically and antigenically distinct from the "classical" (human 1918-derived) H1N1 SIVs which continue to circulate in North America and some parts of Asia, but not in Europe. 13,18 Reassortment of viruses of the Eurasian H1N1av lineage with human seasonal H3N2 viruses gave rise to the porcine H3N2 lineage in which HA and NA are of human origin, while the other six genome segments are of avian descent. 14 In 1994, an H1N2 reassortant (H1huN2) comprising the HA gene from a human seasonal H1N1 virus, the NA gene from H3N2 human-like SIV and internal genes from H1N1 avian-like SIV was first identified in the United Kingdom and subsequently detected in many further European countries. 19 In 2003, a new H1N2 SIV reassortant virus was found in Danish pigs. 20 This H1N2 virus comprised an avian-like HA gene and an NA from contemporary circulating H3N2 SIV and have since been found in several other European countries. 18 Apart from these dominant enzootic strains, several reassortants between these viruses have been described in Europe, but no sustained spread of such viruses has been reported as yet. 21 Following the emergence of the human pandemic H1N1 2009 influenza A virus (H1N1pdm), a fourth IAV lineage entered swine populations globally. Pigs were found to be highly susceptible to this virus, 22 and spread within swine populations, independent of human infections, is ongoing. An increasing amount of reassortants between porcine-adapted enzootic IAV and the H1N1pdm virus have been described. Some of these reassortant lineages, such as the H1pdmN2 lineage reported from Germany, have been in regional circulation for several years now 23 and several retained the so-called TRIG cassette of internal gene segments of the H1N1pdm virus and therefore may carry substantial zoonotic and even pandemic potential. 16,21 Despite coordinated European research efforts in surveillance of porcine IAV, such as the ESNIP consortium, 18 data on the occurrence, spread and evolution of potentially zoonotic SIV in Europe remain fragmentary. This not only impedes the early detection of emerging SIV reassortants with increased zoonotic potential but may also interfere with the development and updating of efficacious influenza virus vaccines for use in pigs. 2,3 A prerequisite for efficient SIV surveillance programmes is sensitive and specific diagnostic methods which allow for high-throughput analysis. Classically, SIV diagnosis is based on virus isolation in MDCK cells and serological subtype characterization by haemagglutination inhibition (HI) assays. However, this has recently been superseded by IAV-generic real-time RT-PCRs (RT-qPCR). [24][25][26][27] Molecular subtype characterization has, however, depended on timeconsuming sequencing of fragments of the HA and NA genes or on subtype-specific, conventional RT-PCRs. [28][29][30] Here we present two multiplex RT-qPCRs to facilitate the detection and differentiation of four and three lineages of HA and of NA gene segments, respectively, of SIV subtypes which currently cocirculate in European swine populations. In combination with a generic IAV-specific duplex RT-qPCR (enhanced by an internal control), we propose a diagnostic algorithm for a rapid, sensitive and lineagespecific large-scale monitoring of SIV infections in clinical samples from swine in Europe.

| Field samples, reference viruses
Lung tissue samples or nasal swabs derived from pigs with respiratory disease were obtained from swine holdings in several European countries in the frame of a passive monitoring programme. Samples were processed and analysed for SIV RNA by an M-gene-specific RT-qPCR as detailed previously. 31 Viral RNA of SIV-positive samples was subtyped by molecular means using conventional pan-HA and pan-NA RT-PCRs and amplicon Sanger sequencing. 32,33 In addition, virus isolation in MDCK cell cultures was attempted and isolates were subtyped by either full-length nucleotide sequence analysis of the HA and NA genome segments as described elsewhere 34 or amplicon sequencing of short HA and NA fragments. influenza A viruses of other host species were retrieved from the reference collections at FLI (Friedrich-Loeffler-Institut, Greifswald-Insel Riems, Germany) or RKI (Robert-Koch-Institut, Berlin, Germany) and used for RT-qPCR validation purposes. Similarly, other porcine respiratory pathogens of viral and bacterial nature were used for assessing analytical specificity (Table S1).

| Design of primers and probes
Primers and probes for subtype and/or lineage-specific detection of HA and NA for use in multiplexed RT-qPCRs were either selected from previously published assays (H1pdm: 35

| One-step RT-qPCR
The AgPath-ID TM One-Step RT-PCR kit (Ambion, Foster City, CA, USA) was used throughout. Thermocycling conditions on a Bio-Rad CFX96 real-time PCR detection system were optimized by adapting annealing time and temperature. These cycling conditions were found to be optimal for the generic M-specific RT-qPCR 38 : The fluorescence data were collected during the annealing step of 55°C and 58°C, respectively, in all runs.

| Sequencing
The subtype and lineage of SIV isolates and PCR-positive field samples were determined by Sanger sequencing of RT-PCR-amplified fragments of the HA and NA gene segments using the pan-HA and pan-NA conventional RT-PCRs described by Gall et al. 32,33 (2008, 2009) and further modified in this study. These PCRs generate an HA fragment of approximately 180 bp spanning the HA cleavage site and a 630-bp fragment of the 5′ region of the NA gene segment. For selected SIV isolates, full-length HA and NA sequences were obtained using RT-PCRs described by Starick et al. 34 (2011). Specific amplicons were purified from 1.5% agarose gels using a QIAquick gel extraction kit (Qiagen, Hilden, Germany) and Sanger-sequenced using the RT-PCR primers. The sequences were analysed on an ABI 310 sequencer, curated using the Chromas Lite ® software (http://www.technelysium. com.au/Chromas250Setup.exe) and assembled using the CAP3 programme (http://doua.prabi.fr/software/cap3).

| Molecular sequence analyses
The IRD or GISAID databases were screened with BLASTN2 to identify closely related sequences of the pan-HA and pan-NA sequences.

| Assembly and analytical performance of subtype-specific uniplex RT-qPCRs for the detection and differentiation of HA and NA lineages of European SIV
Significant sequence variation exists within and between the HA and NA subtypes and lineages of SIV from Europe. Sequence stretches conserved within one lineage, yet distinctively different from other lineages were found to be rare even if only the 3′ termini of potential primers and/or the central region of probes were considered.
Extensive in silico analysis showed that degeneration of primers at up to four positions or the use of more than two primers per PCR was inevitable to span the whole width of (published) sequence variation within a single lineage (Table 1). Primers selected by this strategy for the human-derived H1huN2 lineage gave rise to false (cross-subtype-) priming and resulted in non-specific amplification (data not shown).
T A B L E 1 Attributes of primers and probes employed in tetra-and triplex RT-qPCRs (A) or in multiplex RT-PCR (B) for the detection of HA and NA lineages, respectively, of swine influenza viruses currently circulating in Europe In an attempt to design more specific primers, dual priming oligonucleotides (DPO) were developed. Dual priming oligonucleotides primers are composed of a longer 5′ sequence stretch which serves as an anchor to firmly position the primer at the target site, and degenerate nucleotides can be included into this anchor region. The shorter 3′ region of the DPO primer of usually less than 12 nucleotides enforces highly specific hybridization within a fully matching target sequence.
The two regions are separated by a stretch of five inosine residues. 30 (Table 2B).
In addition, another RT-qPCR specifically detecting the N1 of the human pandemic/2009 H1N1 lineage (N1pdm) was developed. Thus, N1pdm-positive viruses gave positive results with both the N1all and the N1pdm RT-qPCRs (Table 2B).
None of the primer and probe sets gave false-positive results when tested with IAV subtypes H1 and H3 of avian origin ( Table 2).
Non-specific reactivity of these PCRs with other porcine viral or bacterial respiratory pathogens as listed in Table S1 was excluded.

| Assembly and analytical performance of tetraplex (4plex) and triplex (3plex) RT-qPCRs for the detection of haemagglutinin and neuraminidase sequences of European SIV
The HA-and NA-specific probes were labelled with different col-

| Detection of SIV double infections by HAspecific tetraplex and NA-specific triplex RT-qPCRs
Infection of the same cell by more than one IAV geno-or subtype is the prerequisite of reassortment. The possibility to detect influ-    (Table 5A). Full-length sequences of genome segment 4 encoding the HA protein were generated from these three isolates and the sequences compared to the oligonucleotide sequence of the primers and probe used in the subtyping RT-qPCRs. This analysis revealed a considerable number of mismatches (Fig. S1a). Also, the products of the tetraplex RT-qPCR of these isolates were examined by gel electrophoresis (Fig.   S1b), but no specific amplicons were detected. In two cases (AR1123/15 and 2196/15, Table 5A), the multiplex RT-qPCRs detected the presence of a mixture of H1av and H1hu subtypes in the same isolate.

| Diagnostic performance of the HA-specific tetraplex-and NA-specific triplex RT-qPCRs
Haemagglutinin and NA subtypes could be assigned to all of the clinical samples when tested by the multiplex RT-qPCRs (Table 5C/ (Table 5, "questionable") or failed to produce any amplicons at all (Table 5, "failed"). The NA subtyping was affected at a much lower rate (one isolate and six clinical samples failed). These results prompted us to establish updated primers of the pan-HA PCR that provided a better match with European SIV and yielded enhanced amplification sensitivity. Modified primers are listed in Table 1B. Using the updated pan-HA primer set in a multiplex RT-PCR (termed "pan-HA-SIV"), it was possible to sequence amplicons for all 9 isolates which had failed before, and for 16 of the 23 failed clinical samples (Table 5).
Finally, a harmonized diagnosis was made by combining the results of the multiplex RT-qPCR and the RT-PCR/amplicon sequencing: in all cases for which results for both methods were available, fully matching subtyping results were obtained for both HA and NA (Table 5A-D). For seven HA samples for which also the updated primer set failed to produce an amplicon and for one failed NA sample, multiplex RT-qPCRs were able to assign a subtype.

| DISCUSSION
Despite the fact that zoonotic and even pandemic potential in SIV has been noticed repeatedly, no sustained government-administrated   Real-time PCR is an established method in most routine diagnostic laboratories. The use of multiplex assays allows time-efficient and cost-effective examination of several parameters in the same tube. 40 The ultimate goal of this study was to distinguish HA and NA lineages of SIV circulating in Europe and was achieved by the establishment of two multiplex RT-qPCRs. Selection of primers and probes proved to be difficult as sequence stretches that are broadly conserved within one lineage (inclusivity) but distinct to all other ones (exclusivity) are scarce, especially in the HA genome segment. Therefore, primers and probes had to be constructed with a number of degenerate positions.

Isolate identification RT-qPCR (Ct-values) RT-PCR/sequencing
In addition, mixtures of up to four primers for the same target had to be used. Finally, for the distinction of the human-derived HA H1 lineage of SIV (H1hu), a double-priming strategy had to be implement-  (Table 6) which likely prevented amplification (Figs S1 and S2).
A constant control and timely adaptation of oligonucleotides in the multiplex RT-qPCRs is essential.
While the isolates and clinical samples used here for validation purposes only represent a small fraction of the ongoing monitoring project, the comparatively large fraction of HA/NA reassortants is T A B L E 6 Mismatches of primers and probes for the three SIV isolates which remained negative in the tetraplex HA RT-qPCR (Table 5A) notable (Table 5, bold-face samples). Increasing diversity of genotypes and HA/NA reassortant patterns, especially within subtype H1, has previously been noticed. 18,21,31,41 No reassortants were detected here for the H3 subtype.
Based on the results of this study, we propose an algorithm, depicted in Figure 1 The newly developed multiplex RT-qPCRs provide basic tools for a sustained monitoring programme of swine influenza in Europe. In addition, virus isolation on selected samples is required to allow further antigenic, in-depth genetic and biological characterizations of circulating virus strains.