Spillback transmission of European H1N1 avian-like swine influenza

viruses to turkeys: a strain-dependent possibility?

Highlights
  Replication of avian-like swine H1N1 viruses in turkeys is a strain-dependent feature
 Avian-like swine H1N1 viruses replicating and transmitting in turkeys possess human-
 like receptor binding specificity
 Isolates displaying opposite agglutinating capabilities share the same key amino acids at
the level of the hemagglutinin, suggesting other determinants are involved.

Abstract
In 1979, an avian influenza virus of the H1N1 subtype began to circulate in European swine herds,
rapidly replacing classical swine H1N1 viruses. Spill-back transmissions to turkeys were recorded
occasionally, but they might have been underreported due to the asymptomatic nature of the
infection and the lack of specific surveillance. In our study, we evaluated the infectivity and
transmissibility in turkeys of seven strains of H1N1 avian-like swine viruses isolated from 1979 to
2006, and compared them with their closest progenitor A/duck/Bavaria/1/77 (H1N1), to establish
whether the adaptation to pigs has gradually decreased their fitness in turkeys. Our data indicate that
the circulation of European H1N1 in pigs might have impaired the possibility of infecting turkeys.
Nevertheless, the two swine-origin strains, which showed the ability to replicate and transmit in
turkeys, possess typical swine-like genetic traits, not different from the rest of the tested isolates,
suggesting replication of avian-like swine H1N1 viruses in turkeys as a strain-dependent polygenic
feature.

Introduction
Avian influenza (AI) viruses occasionally leave their natural reservoir, wild aquatic birds, and cross
species and class barriers, infecting poultry and different families of mammals. Due to modern
industrial farming and a globally connected human population, specific subtypes of AI viruses have
become endemic in humans, chickens, swine herds and horses worldwide, while for species like
dogs, seals and minks influenza outbreaks still have a regional and cyclical pattern.
Pigs are thought to play a crucial role in the evolution of zoonotic influenza viruses, as they host
strains of different species origin, probably due to their receptor profile at the level of the upper
respiratory tract (Ito et al., 1998). For this reason, they are commonly addressed as the so-called
mixing vessel.
In Europe and North America, the ecology of swine influenza dramatically changed when avian-
like H1N1 (Brown, 2000) and triple reassortant H3N2 (Webby et al., 2000) viruses emerged and
spread throughout the respective continents, replacing the extant classical H1N1 lineage of human
origin. Both viruses carried polymerase genes of avian origin and were able to reassort with avian
and human viruses, accepting different haemagglutinin (HA) and neuraminidase (NA) genes
(Brown, 2013; Lorusso et al., 2013). Reassortment of these viruses led to the emergence of the
2009 H1N1 pandemic strain (Malik Peiris et al., 2009), a virus that has become seasonal and has
been frequently isolated in pigs and turkeys around the world (Pereda et al., 2010; Reid et al.,
2012). Turkeys, like pigs, are particularly susceptible to influenza viruses of different origin as their
respiratory epithelia express both avian and mammal-like receptors (Pillai and Lee, 2010), offering
a favorable environment for virus reassortment, and the emergence of potentially zoonotic strains
against whom the human population would be immunologically naïve.
Transmission of swine influenza to turkeys has been reported in the United States, as a result of
proximity between hog and turkey farms and insufficient biosecurity measures, leading in some
areas to the implementation of preventive vaccination against swine influenza in turkeys (Yassine et
al., 2013).
Considering the high farming density of both pigs and turkeys in Europe, and the fact that avian-
like H1N1 viruses derived from a duck isolate (A/duck/Bavaria/1/77), reports of transmission to
poultry have been surprisingly rare, and only in one case they were associated with clinical
disease (Starick et al., 2011; Wood et al., 1997). Subclinical infections and a lack of subtype-
specific surveillance in poultry might be accounted for underreporting of spillback transmission
events. Studying how the adaptation of these viruses to pigs affected their fitness in turkeys could
help us identify the molecular basis of swine influenza host-range restriction, as well as the
virulence and pathogenicity markers of mammalian strains in turkeys.

4
In this study, we investigated the replication and transmission among turkeys of seven strains of
avian-like H1N1 swine influenza, isolated between 1979 and 2006 in Europe.

Materials and methods

Viruses
Eight H1N1 viruses were selected for the experimental infections: A/duck/Bavaria/1/77 (Dk/77),
A/swine/Belgium/1/79 (Sw/79), A/swine/Belgium/1/83 (Sw/83), A/swine/Gent/196/92 (Sw/92),
A/swine/England/195852/92 (Sw/Eng/92), A/swine/Gent/74/95 (Sw/95), A/swine/Belgium/1/98
(Sw/98) and A/swine/England/453/06 (Sw/06) (kindly supplied by Kristien Van Reeth, Ghent
University, and Jens Peter Teifke, Friedrich-Loeffler-Institut). The total number of passages in
eggs since the first isolation was not available.
Stocks of viruses were produced inoculating 9-to 11 day-old embryonated specific pathogen free
(SPF) hen’s eggs via the allantoic cavity. The infectious allantoic fluid was harvested 48 hours
post inoculation, aliquoted and stored at -80°C until use.
The stocks were titrated in a standard plaque assay on MDCK cells and expressed as plaque
forming units (PFU) per ml.

Animals
Day-old turkeys (Meleagris gallopavo) were obtained from a commercial farm and tested for the
presence of antibodies against Influenza A viruses. Birds were housed in negative pressure, high
efficiency particulate air (HEPA) filtered poultry isolators in the BSL-3 containment facilities of the
Istituto Zooprofilattico Sperimentale delle Venezie (IZSVe). Prior to challenge, animals were
housed for 4 weeks. Animal experiment procedures were conducted in strict accordance with the
national and international animal care guidelines [Convention of the European Council no. 123 and
Italian Law on protection of animals used for scientific purposes (DLgs 116/92)], and approved by
the Institute’s Ethics Committee.

Experimental design
At the age of 4 weeks, eight groups of 10 birds each were formed and challenged via the oro-
nasal route with a dose of 106 PFU/0.1 ml of the selected virus. Groups were named after the
challenge viruses (e.g. group Dk/77). On day 2 post infection, four sentinels were introduced in
each challenged group. Challenged birds and sentinels were swabbed on days 2, 4, 7 and 10 post
infection (p.i.) or post contact (p.c.), respectively. Birds were monitored twice a day, to record the

5
appearance of clinical signs. Twenty-one days from either the challenge of naïve birds, or the
introduction of sentinels, blood was collected to assess the level of seroconversion. Sera with
HI titers lower than 4 log2 were further tested by means of a type A ELISA.

Virus detection, characterization and serology

Viral RNA was extracted from 50 µL of phosphate-buffered saline (PBS) containing a swab
suspension using the Ambion MagMax-96 AI-ND Viral RNA Isolation kit on the MICROLAB®
STARlet bench-top liquid handling system (Hamilton Robotics). Isolated RNA from each sample
was analyzed by Real time RT-PCR for the identification of the influenza A virus matrix (M) gene
using the published probes and primers from Spackman et al. (Spackman et al., 2002). For each
virus serial dilutions of selected positive oropharyngeal and cloacal swab samples were tested both
by means of RRT-PCR and by virus isolation in five SPF embryonated chicken eggs, to establish a
cut-off value for virus viability. Swabs with Ct values higher than the cut-off, were tested by virus
isolation to prevent reporting false negatives, while samples with Ct values lower than the cut-off
were considered positive for isolation. Mean RRT-PCR Ct values were also plotted irrespective of
their viability in eggs as an accessory measure of virus RNA shedding.
To compare the receptor binding preference of the challenge viruses, the hemagglutination test
(HA) was performed using erythrocytes from different avian and mammalian species (Ito et al.,
1997) on undiluted original viral stocks. Erythrocytes from the turkey, guinea pig and human
species are known to express mainly α 2,6 sialic acid (SA) receptors, while chicken and horse red
blood cells predominantly contain α 2,3 SA; in particular horse erythrocytes express only α 2,3
and no α 2,6 SA. The HA was performed according to the OIE guidelines (World Organisation for
Animal Health (OIE), 2015).
To evaluate post infection seroconversion, the hemagglutination inhibition test (HI) was carried
out using a homologous antigen to the challenge virus, according to the OIE procedure (World
Organisation for Animal Health (OIE), 2015).
Before carrying out the HI test, sera were treated with 10% chicken erythrocytes for 30 min to
remove non-specific agglutinins. To remove non-specific inhibitors of hemagglutination, sera were
treated with a receptor destroying enzyme from Vibrio cholerae (RDE) overnight at 37°C, prior to
heat inactivation by incubation at 56°C for 30 minutes. Type A ELISA (ID Screen ® Influenza A
Antibody Competition, IDvet, France) test was used as accessory serological test to investigate the
presence of antibodies in serum samples with HI titers lower than 4 log2, since it is a more
sensitive assay compared with the HI test.

6
Sequencing and phylogenetic analysis
Sequencing of the whole genome was carried out on allantoic fluids containing the challenge
viruses, and selected swabs collected from challenged and sentinel birds that shed viable virus on
days 7 p.i/p.c. Viral RNA was extracted from the allantoic fluid and swab suspensions using the
Nucleospin RNA II kit (Macherey- Nagel, Duren, Germany). Complete influenza A virus
genomes were amplified from the RNA using the SuperScript III one-step reverse transcription-
PCR (RT-PCR) system with PlatinumTaq high fidelity (Invitrogen, Carlsbad, CA) by following
the protocol of Zhou et al. (Zhou et al., 2009). The amplified genomes were fragmented down to
600 to 800 bp and tagged with sequencing adapters with indexes in a single step, using the
Nextera DNA sample preparation kit from Illumina. Finally, the indexed libraries were pooled in
equimolar concentrations and sequenced in multiplex for 250-bp paired ends on Illumina MiSeq.
Raw sequence reads were inspected using FASTQC to assess the quality of data coming from the
high-throughput sequencing pipelines. Fastq files were cleaned with PRINSEQ and Trim Galore
to remove adaptors and primers and exclude reads with a Phred quality score below 30. Only reads
longer than 80 bp were considered and mapped to the reference sequences using Stampy (Lunter
and Goodson, 2011). Consensus sequences were obtained from the BAM files using Samtools
version 0.1.14 (Li et al., 2009) and deposited in GISAID under accession numbers EPI662096 to
EPI662159.
Each gene segment of the analysed samples was compared with representative sequences of the H1
subtype available in GenBank and with sequences resulted from the BLAST search. Nucleotide
sequence alignments were constructed for each gene segment using the online version of MAFFT
v.7 program (http://mafft.cbrc.jp/alignment/server/index.html). To infer the evolutionary
relationships for each gene segment, we employed the maximum likelihood (ML) method
available in the PhyML program, incorporating a GTR model of nucleotide substitution with a
gamma distribution of among-site rate variation (with four rate categories, Γ4) and an SPR branch-
swapping search procedure (Guindon and Gascuel, 2003). The GTR + 4 model of nucleotide
substitution was found to be the best-fit to the data in hand using the model selection utility implemented in
MEGA 5.2. Phylogenetic trees were visualized with FigTree v.1.1.2
(http://tree.bio.ed.ac.uk/software/figtree/).

Molecular analysis
To identify mutations potentially associated with adaptation of the virus to avian or swine species,
we compared the amino acid composition of the eight H1 viruses under study, as well as European
swine H1 viruses of the avian-like lineage previously described as linked to a spill-back

7
transmission event to turkeys (A/turkey/Germany/2482/90 and A/turkey/Germany-NI/R778/09) as
terms of reference for their proven replication capabilities in turkeys. In addition, mutations
previously described as associated to adaptation of avian influenza viruses to mammals or to an
increased virulence in mammals, were analysed. Moreover, we compared the amino acid
composition of the challenge viruses with the respective strains after replication and transmission
among turkeys, in order to identify either adaptive or escape mutations.

Statistical analyses
Frequency of virus shedding was analyzed by Fisher’s exact test (P < 0.05). Swab Ct values and HI
titers were expressed as mean ± standard deviation and analyzed by unpaired Student’s t-test (P <
0.05).

Results

Clinical data
Only challenged and sentinel birds of groups Dk/77, Sw/79 and Sw/Eng/92 showed clinical
signs after the challenge. In groups Dk/77 and Sw/79, reduced activity, altered droppings and
minor respiratory distress were recorded. In group Sw/Eng/92, the infection caused mild
depression and led to laboured breathing and nasal discharge, while droppings appeared normal.
All of the remaining animals looked healthy throughout the duration of the trials.

Viral shedding
Results of virus isolation and RRT-PCR mean Ct values on tracheal and cloacal swabs are shown,
in figures 1 and 2. Cut-off values for viability of viruses that replicated and transmitted in turkeys
were determined for each type of swab, and are reported in table 1.
Challenged birds
In groups Dk/77, Sw/79 and Sw/Eng/92, at least 70, 100 and 60% of the challenged birds shed
viable virus by the tracheal route, on a given sampling day (Fig.1). Tracheal shedding peaked at
day 4 in groups Dk/77, Sw/79 and at day 7 p.i. in group Sw/Eng/92.
Cloacal shedding in groups Dk/77 and Sw/79 peaked at day 7 p.i., recording 60 and 100% positivity
among the challenged birds, respectively, while in group Sw/Eng/92, none of the challenged birds
shed viable virus. Virus RNA shedding profiles of Dk/77 and Sw/79 were similar in trends and
mean values (Fig. 2), while significant differences were recorded when comparing those groups
with the Sw/Eng/92 group.

8
In groups Sw/83, Sw/92 and Sw/98 no viable virus was detected from swabs, while in groups
Sw/95 and Sw/06 only one animal per group shed viable virus on a sampling day by either the
trachea or the cloaca.
Sentinels
In group Dk/77, Sw/79 and Sw/Eng/92, the frequencies of sentinels shedding viable virus by the
tracheal route were equal or higher than those of the challenged birds, apart on day 7 p.i., in
group Sw/79 and on day 2 and 10 p.i. in group Sw/Eng/92. In particular, on day 2 p.i., in groups
Dk/77 and Sw/79 the number of shedders was significantly higher than that among challenged
birds. Cloacal shedding of viable virus was restricted to 25% of the birds in group Dk/77 as
opposed to the 100% of sentinels in group Sw/79. In group Sw/Eng/92, no shedding of viable
virus by the cloacal route was detected among sentinels.
In the remaining groups, none of the sentinels shed viable virus during the study.

Serology
In groups Dk/77, Sw/79 and Sw/Eng/92, both challenged and sentinel birds scored HI geometric
mean titers (GMT) greater than 4 Log2 and 100% seroconversion rates (Fig. 3). The highest
GMTs were recorded in group Sw/Eng/92, for both challenged (10.2 Log2) and sentinel (8.0
Log2) birds. Of the remaining groups, challenged birds of group Sw/95 were the only other
subjects to record a HI GMT greater than 4 Log2 (4.4 Log2), while only one sentinel
seroconverted, resulting positive to both the HI and ELISA tests.
Although 30 and 50% of the birds seroconverted in groups Sw/83 and Sw/92, respectively,
GMTs were lower than 3.0 Log2.
Challenged and sentinel birds in groups Sw/98 and Sw/06 recorded negative HI results
(GMT<1.0 Log2) and only 4 subjects in group Sw/98 tested positive by the ELISA test.

Erythrocyte binding preference

When erythrocytes of turkey, guinea pig, human and chicken origin were used, HA titers of all
challenge viruses ranged from 4.0-9.0 Log2 (Tab. 2). The only viruses that could agglutinate horse
red blood cells were Dk/77 and Sw/79, achieving HA titers of 7.5 and 5.0 Log2, respectively.
The HA titres of Sw/98 (3-4 Log2) were considered unusually low given the high infectivity in
MDCK cells of the viral stock (1,7·107 PFU/0,1ml). The assay was repeated by two operators and
results were confirmed. In our experience (unpublished data) influenza viruses range widely in
HA titres with varying correspondence to the infectivity of the strain, depending on HA-NA
balance, mutations at the level of the RBS and other variables.

9
Phylogenetic analyses
The eight hemagglutinin sequences clustered within the group of Eurasian avian-like swine viruses
(Fig. 4). The analysis revealed HA nucleotide and amino acid identities of 88.3 – 94.0% and 90.2 –
97.3%, respectively between Dk/77 and the swine challenge viruses (Tab. 3). The neuraminidase
(NA) genes grouped in the same genetic lineage, showing nucleotide and amino acid similarities
of 85.7 – 90.9% and 89.3 – 94.7%, respectively with the progenitor virus Dk/77 (Tab. 3). A similar
picture emerged from the phylogenetic analyses of the remaining genes, as all of the challenge
viruses isolated from pigs fell within the monophyletic group of Eurasian avian-like swine viruses.
On the other side, the avian strain A/duck/Bavaria/77 fell within the avian group, and showed a
sister-group relationship with the swine avian-like clade for all the gene segments, except for the
NS gene, which belongs to two distinct allele: the avian virus possesses the allele B, while the
swine viruses contain the allele A.
As expected, we observed a gradually decrease of HA and NA sequence identity of the
swine viruses from Dk/77 according to their sampling dates.

Molecular analysis
As highlighted by the phylogenetic analyses, the challenge viruses, although belonging to the same
genetic lineage, had low nucleotide and amino acid similarities. Hence, the first objective of these
analyses was to verify if the three challenge viruses that had efficiently replicated in turkeys,
shared unique amino acid mutations compared to the non-replicating phenotypes. Comparisons
among viruses of all genes did not result in any specific signature that was unique of Dk/77, Sw/79
and Sw/Eng/92.
For this reason, we focused on the more genetically and phenotypically related strains, Dk/77
and Sw/79, and in this case 32 unique molecular features were identified for the HA, NA, PA,
PB1, PB2, PB1-F2, NP, M1, NS1 and NS2 genes.
The identified mutations in the HA and NA genes do not possess any known biological function
(Supplementary table S1). Similarly, mutations in the other genes have never been associated with
adaptation or increased virulence in mammals, with few exceptions. Mutation D701N in PB2 (D
in Dk/77 and Sw/79, and N in the remaining challenge viruses and natural spill-back transmission
isolates) is known for increasing host range, polymerase activity and pathogenicity in humans and
mammals (Li et al., 2005). Moreover, the avian signatures E227 (NS1) and S70 (NS2) were found
in Dk/77 and Sw/79, while all of the other viruses carry a G227 and a G70 (human signature) in
NS1 and NS2, respectively (Chen and Shih, 2009).

10
Our second objective was to analyze the amino acid sequences to identify adaptive mutations
associated with increased virulence, transmission, replication and binding affinity of avian
influenza viruses in pigs and other mammalian species (Tab. 4). In the HA, all of the swine strains
irrespective of their phenotype in turkeys, shared at positions 155, 159 and 190 the classical swine
features, previously described as markers of avian-to-pig adaptation (T155V, T159N and E190D)
(Imai and Kawaoka, 2012; Matrosovich et al., 2000). The only exceptions were Sw/79, as it carried
a T155I mutation instead of the classical swine T155V, and Sw/92 that carried a T159D mutation as
opposed to the classical swine T159N. Previous studies demonstrated that Asp190 and Gly225
confer mixed α2-6/α2-3 specificity in the H1 subtype (Stevens et al., 2006; Zhang et al., 2013).
Notably, all the swine strains, with the only exception of Sw/98, possessed both residues, while
Dk/77 contained the glutamic acid (Glu) at position 190, which is sufficient to revert the HA
receptor preference to that of classical avian strains.
At the level of the PB1 gene, the L473V mutation is known to increase the polymerase activity of
influenza viruses in mammalian cells and mice (Xu et al., 2012). This mutation was present in all
viruses, including the ones involved in natural back-transmission events.
The R99K mutation in the NP gene, previously associated with airborne transmission of a
chimeric H5N1 virus in ferrets (Herfst et al., 2012), was identified only in some of the swine
challenge strains (Sw/83, Sw/92, Sw/98 and Sw/06).
The mutation P42S in the NS1 responsible for the increased pathogenicity of H5N1 viruses in mice
(Jiao et al., 2008) was identified in all of the swine strains and not in the Dk/77 virus.
In Dk/77 and Sw/79, the C-terminal residue of the NS1 protein carried the classical avian
feature ESEV(227-230), while all of the remaining strains had the motifs GSEV or GPEV of
unknown biological value.
Molecular analyses were conducted on further 15 samples collected at day 7 p.i/p.c. from
challenged birds and sentinels infected with Sw/Eng/92 (4 samples), Dk/77 (4 samples) and Sw/79
(7 samples), as these are the only viruses that consistently replicated in and transmitted among
turkeys. The complete genome was generated for 9 samples, while for the remaining six samples
only the sequences of 4-8 genes were obtained. Each analysed virus shows from 0 to 6 amino acid
mutations, compared to the challenge strain, distributed along the entire genome, with the HA
protein displaying the highest number of polymorphisms (Supplementary tables S2-S4).
Specifically, we identified a total of 12 amino acid changes in the HA protein of viruses collected
from both challenged birds and sentinels of group Dk/77 (8 mutations), Sw/79 (3 mutations) and
Sw/Eng/92 (3 mutations). Four of these substitutions (L53I, S112P, N189D, N197D) were
identified in more than one isolate belonging to different birds in the same group, while no mutation

11
on the HA was shared among isolates of different groups. Two samples in groups Dk/77 and
Sw/Eng/92 selected for mutations in the neuraminidase protein at the level of antigenic
(S364N) (Jiang et al., 2015) and enzymatic sites (R430Q) (Qi et al., 2012), respectively.
Of note, three samples from challenged and sentinel birds in group Sw/Eng/92 acquired the
mutation K99R in the NP protein, recovering the amino acid typical of avian strains (Herfst et al.
Science 2012).
Six samples collected from sentinels and challenged birds in groups Dk/77 and Sw/79 shared the
same mutation at position 231 in the M1 protein (D231N), a determinant of virion morphology in
equine influenza viruses (Elton et al., 2013).
Interestingly, a virus detected in a swab in group Dk/77, carried the mutation R79Q in PB1-F2
associated in mice with non-inflammatory and anti-bacterial effect (Alymova et al., 2011). None of
the other mutations identified in the internal proteins have been previously described as associated
with a change in either host preference, transmissibility, pathogenicity, or tissue tropism.

Discussion
Of the eight strains only the Dk/77, Sw/79 and Sw/Eng/92 replicated in all of the challenged birds
and transmitted to the sentinels. Replication of these strains differed in terms of pathogenicity, as
both Dk/77 and Sw/79 elicited minor clinical signs, such as reduced activity and altered
droppings, while challenge with Sw/Eng/92 caused overt respiratory distress and nasal discharge,
but in this case droppings appeared normal.
In keeping with these observations, birds infected with Dk/77 and Sw/79 shared similar patterns of
viral shedding, as both challenged and sentinel birds shed viable virus via both the tracheal and
cloacal routes. On the other hand, challenged and sentinel birds of the Sw/Eng/92 group shed viable
virus exclusively by the tracheal route. Previous studies (Ali et al., 2013) tested the replication
efficiency of recent North American SIVs in juvenile and layer turkey hens, describing subclinical
infections associated with consistent shedding of virus by both the tracheal and cloacal routes,
irrespective of the subtype and lineage of virus. In contrast, our data portrait a variable phenotype
of avian-like swine viruses, with extremely different outcomes in terms of clinical signs and viral
shedding depending on the tested isolate. These experimental evidences could in part explain why
outbreaks of SIV in turkeys are more frequent in the United States compared to Europe.
Interestingly, the erythrocyte binding preference of Dk/77 and Sw/79 differed substantially with
that of Sw/Eng/92 and the rest of the challenge strains. The progenitor virus and the early swine
isolate were in fact the only strains to agglutinate horse red blood cells, which are known to express
almost exlusively α 2,3 SA (Pawar et al., 2012), suggesting that the rest of the challenge viruses

12
either lost or strongly decreased their binding affinity for these receptors few years after their
introduction in the European swine farming industry. Interestingly, Matrosovich et al.,
(Matrosovich et al., 2000) studied the shift in receptor binding to synthetic syalylglycopolymers
of European avian-like swine H1N1 viruses, and recorded a similar trend for strains isolated from
1979 to 1992.
To this respect, the lack of cloacal shedding recorded in the Sw/Eng/92 group, relates with the
lack of affinity of the virus for α 2,3 SA and with the evidence that the small and large intestines
of 4-week-old turkeys display exclusively α 2,3 SA receptors (Pillai and Lee, 2010). On the other
hand, we could not explain why Sw/Eng/92 elicited more severe clinical signs and a stronger
seroconversion, given that the virus replication was significantly lower than Dk/77 and Sw/79, and
it was restricted to the respiratory tract.
Interestingly, Sw/79 replicated and transmitted more than the progenitor Dk/77, indicating that
the species jump from aquatic birds to swine, possibly through a passage in poultry, brought
about adaptive mutations that are favorable not only for mammals but also for turkeys.
Molecular analyses aimed at understanding whether the strains efficiently replicating in turkeys
shared specific signatures, and if replication in turkeys could select for adaptive or escape mutants.
Features uniquely shared by Dk/77 and Sw/79 with a known biological function were the amino
acids at positions 701 and 227 of the PB2 and NS1 genes, respectively. Both viruses shared the
classical avian traits (D701; E227) while the rest of the swine strains shared the asparagine (N) at
position 701 of the PB2 and a glycine (G) at position 227 of the NS1 protein. The D701N mutation
is generally associated with the enhancement of polymerase activity and replication efficiency of
different influenza viruses in mammalian species. Nevertheless, it is not clear whether an
asparagine at position 701 could also act as a limiting factor for the replication of mammal-
adapted viruses in birds.
Sw/79 and Sw/Eng/92, replicated and transmitted in turkeys and yet their HA genes carried 2 of the
three amino acids that are classically present in both classical and avian-like Eurasian swine influenza
viruses (Matrosovich et al., 2000). Of these mutations, the amino acid change at position 190 is thought
to be decisive in the adaptation of avian viruses to both human and swine receptors, increasing their
affinity for α 2,6 SA (Matrosovich et al., 2000). Interestingly, these two viruses possessed the same
amino acid in this position but displayed opposite agglutinating capabilities and different tissue tropism,
suggesting that other amino acids at the level of the hemagglutinin play a fundamental role in
determining the binding preference of receptors in H1N1 swine viruses. Molecular analyses conducted
on samples collected 7 days from the onset of infection identified different mutations at the level the
HA and NA. Interestingly, 8 out of the 12 amino acid changes on

13
the HA were located within antigenic sites or in the receptor binding domain, suggesting that
they might have been selected as a result of adaptation to the new host, or during the immune
evasion process, as both innate and adaptive immunities are already in place after 7 days from
the onset of infection. Interestingly, mutation K99R in the NP protein was selected in different
subjects within group Sw/Eng/92, indicating a beneficial role of Arginine in this position, as
this amino acid was only present at a consensus level in Dk/77 and Sw/79, the other two
viruses that replicated in turkeys.
Mutation D231N in the M1 protein, deserve further investigation as it is the only amino acid
change that was selected in challenged and sentinel birds infected with Dk/77 and Sw/79
viruses, suggesting a convergent advantageous evolution of this gene.
Spill-back transmission of avian-like swine strains to turkeys such as Tk/90 and Tk/09 revealed
that the viruses did not revert to the classical avian molecular features (Starick et al., 2011).
These findings together with our results indicate that Eurasian avian-like swine viruses with a
human-like receptor binding specificity have the capability to infect and transmit in turkeys,
maintaining the molecular features required for replication in mammals. Due to the limited
number of tested isolates and the evidence that H1N1 pig-to-turkey transmissions occurred in
France and Germany, in 1990, 2005 and 2009 (Starick et al., 2011), we hypothesize that the
capability to infect turkeys is a polygenic strain-dependent feature, whose frequency in the swine
virus population is unknown. The undetected circulation of these viruses in turkeys could pose a
silent threat, as it offers the opportunity for strains potentially transmissible to humans, to
reassort with avian viruses of low and high pathogenicity. In this perspective, the
implementation of enhanced surveillance for the detection and characterization of avian-like
swine viruses in high-risk turkey farms could prove to be a valuable tool in monitoring the
emergence and circulation of potentially zoonotic viruses.

18
Figure 1. Tracheal and cloacal shedding of challenged and sentinel birds in each group. Primary
vertical axes represent the percentage of swabs positive for virus isolation. P.i: post infection;
p.c.: post challenge; Ch: challenged birds; Se: sentinels; TS: Tracheal swabs; CS: cloacal swabs

19
Figure 2.Mean RRT-PCR Ct values ± standard deviations of tracheal and cloacal swabs, of
challenged and sentinel birds of groups Dk/77, Sw/79 and Sw/Eng/92. Black asterisks indicate
significant differences between the groups. **:p<0.005; ****: p<0.00005; Ch: challenged birds;
Se: sentinels; TS: Tracheal swabs; CS: cloacal swabs
Figure 3. Mean HI titers ± standard deviations of challenged and sentinel birds in each group.
Black asterisks indicate significant differences between the groups. ***:p<0.0005; Ch: challenged
birds; Se: sentinels

Figure 4. ML phylogenetic tree for the HA gene segment of H1 subtype avian, human, avian-like
and classical swine influenza viruses. Challenge virus are highlighted in red, bold letters indicate
natural spill-back transmission isolates in turkeys.
99 A/swine/England/WVL11/1994_H1N1
90

95
 Swine
Avian-like

Avian

PDM

Classical
Swine

Human
Seasonal

Swine
european

0.04
Table 1. RRT-PCR cut-off values for virus viability in SPF embryonated eggs of strains that

replicated and transmitted in turkeys.

RRT-PCR (Ct values)

Challenge virus TS CS

Dk/77
28.90 33.12
Sw/79
32.35 31.10
Sw/Eng/92
31.20 30.60

23
Table 2. Hemagglutination titers (Log2) of challenge viruses using erythrocytes from different

species

Hemagglutination titers (Log2)

Challenge virus Turkey Guinea Pig Human Chicken Horse
Dk/77 9.0 9.0 7.5 7.5 7.5
Sw/79 7.0 7.0 6.5 6.5 5.0
Sw/83 6.0 4.5 4.5 5.05 0.0
Sw/Eng/92 9.0 8.5 8.5 8.0 0.0
Sw/92 8.5 8.5 9.0 7.0 0.0
Sw/95 7.0 6.0 7.0 6.75 0.0
Sw/98 3.0 n.d. 3.0 4.0 0.0
Sw/06 6.0 n.d. 5.0 6.0 0.0

24
Table 3. Nucleotide and amino acid sequence identity of HA and NA proteins of avian-like

swine viruses with the progenitor A/Dk/Bavaria/1/77.

Nucleotide (amino acid*) similarity (%)

Challenge
virus HA NA PB1 PB2 PA MA NP NS
100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Dk/77
(100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0)
94.0 90.9 99.9 99.9 99.9 100 99.9 99.7
Sw/79
(97.3) (94.7) (100) (100) (100) (100) (100) (99.7)
93.2 90.4 99.9 99.9 99.9 100 99.9 99.7
Sw/83
(95.2) (93.8) (100) (100) (100) (100) (100) (99.7)
91.1 88.6 99.9 99.9 99.9 99.9 99.9 99.7
Sw/Eng/92
(93.0) (91.5) (100) (100) (100) (100) (100) (99.7)
91.0 88.1 99.9 99.9 99.9 99.9 99.9 99.7
Sw/92
(93.2) (91.5) (100) (100) (100) (100) (100) (99.7)
90.6 87.6 99.9 99.9 99.9 99.9 99.9 99.7
Sw/95
(92.5) (89.7) (100) (100) (100) (100) (100) (99.7)
90.0 87.4 99.9 99.9 99.9 99.9 99.9 99.7
Sw/98
(91.5) (90.4) (100) (100) (100) (100) (100) (99.7)
88.3 85.7 99.9 99.9 99.9 99.9 99.9 99.7
Sw/06
(90.4) (89.3) (100) (100) (100) (100) (100) (99.7)
*similarity calculated on the longest protein codified by each segment

25