IFN and cytokine responses in ducks to genetically similar H5N1 influenza A viruses of varying pathogenicity

Abstract

Ducks, the reservoir host, are generally permissive to influenza A virus infection without disease symptoms. This natural ecology was upset by the emergence of H5N1 strains, which can kill ducks. To better understand host–virus interactions in the reservoir host, and influenza strain-specific molecular contributions to virulence, we infected White Pekin ducks with three similar H5N1 viruses, with known differences in pathogenicity and replication rate. We quantified viral replication and innate immune gene activation by qPCR, in lung and spleen tissues, isolated on each of the first 3 days of infection. The three viruses replicated well, as measured by accumulation of matrix gene transcript, and viral load declined over time in the spleen. The ducks produced rapid, but temporally limited, IFN and cytokine responses, peaking on the first day post-infection. IFN and proinflammatory cytokine gene induction were greater in response to infection with the more lethal viruses, compared to an attenuated strain. We conclude that a well-regulated IFN response, with the ability to overcome early viral immune inhibition, without hyperinflammation, contributes to the ability of ducks to survive H5N1 influenza replication in their airways, and yet clear systemic infection and limit disease.

Introduction

Influenza A viruses are ubiquitous pathogens that periodically cause widespread epidemics in humans and livestock, resulting in significant public health and economic burdens [1–4]. Ducks and other waterfowl are the natural hosts of most influenza virus types, with a long co-evolutionary history [5, 6]. Despite the existence of host tropism restrictions waterfowl can be a reservoir of aggressive epizootic strains like A/HK/156/1997 (H5N1) and A/Shanghai/patient 1/2013 (H7N9), both of which have caused occasional human infections with high mortality [7–9]. Historically, the majority of such strains, even those bearing the hallmark of highly pathogenic influenza – a multibasic cleavage site in the haemagglutinin (HA) protein – cause little disease in the ducks themselves, allowing them to act as the ‘Trojan horses’ of influenza [10–15]. However, certain highly pathogenic avian influenza (HPAI) H5N1 viruses isolated since 2002 are pathogenic and even lethal to ducks, and produce higher titres in the trachea than in the cloaca, which shows a preference for replication in the airways rather than the intestines [16–18]. These strains, which can harm ducks, represent a change in the natural host–pathogen relationship. The precise mechanisms of pathogenesis and resistance in the natural host are not entirely clear.

Pathogenicity of influenza viruses is associated with immune dysregulation and unusual tissue tropism. Hypercytokinemia, a ‘cytokine storm’ that results from over-activation of inflammatory immune signalling, can contribute to influenza pathology in humans, mice and chickens [19–24]. In side-by-side infections of chickens and ducks with two HPAI H5N1 viruses – A/Muscovy duck/Vietnam/453/2004 and A/duck/Indramayu/BBVW/109/2006 – Burggraaf and colleagues found that the inflammation responses differed dramatically between hosts [19]. Proinflammatory cytokines, like IL-6, were potently induced in the chickens, but only modestly in ducks, and only at later times in the infection period. The authors concluded that the controlled response in ducks contributed to apparent resistance and decreased mortality. Similar trends were also observed in chicken embryonic fibroblasts and duck embryonic fibroblasts infected with H5N1 virus [25]. On the other hand, Wei and colleagues found that infection of Muscovy ducks with the HPAI H5N1 virus A/duck/Guangdong/383/2008 greatly induced transcription of IL-6, IL-8 and IFN-gamma (IFNG) in brain tissue, along with apparently unchecked viral replication [26]. Their results suggest that hyperinflammation may also contribute to neurological symptoms and mortality in ducks.

To help clarify the mechanisms of virulence and resistance we performed experimental infections of mature Pekin ducks using three similar HPAI (H5N1) influenza strains with different reported levels of pathogenicity; A/duck/Thailand/D4AT/2004 (D4AT), a very virulent avian influenza virus, and A/Vietnam/1203/2004 (VN1203), which is an avian-origin virus isolated from a fatal human case [27, 28]. In an experimental infection of juvenile Pekin and mallard ducks both of these strains produced comparable titres in tracheal and cloacal swabs, but D4AT was more lethal than VN1203 in both directly infected and contact-infected animals [16]. In 4–6-week-old mallard ducks infected with 10^6-7 EID₅₀ (50% egg infectious doses), D4AT killed all of the ducks by 6 days post-infection (p.i.), while VN1203 killed 50 % of the ducks at 5 days p.i., leaving survivors that shed the virus. D4AT was also 100 % lethal to experimentally infected Khaki Campbell ducks [28]. Interestingly, this is opposite to lethality in ferrets, where VN1203 is more lethal than D4AT [27]. The third virus strain we used is a reverse-genetics, recombinant version of VN1203 (rgVN1203) [29] that has 11 silent nucleotide substitutions and is attenuated in ducks compared to the original isolate [30]. It elicits clinical symptoms in ducks, but with a lower mortality rate than the original isolate VN1203 due to three amino acid substitutions (T51M, V56A, E87G) in the PB1-F2 gene, which decrease viral polymerase activity and replication efficiency in a minigenome assay [31]. The original isolate VN1203 killed 55 % of the ducks, compared to only 20 % for rgVN1203.

Here we examine transcription levels of IFN, IFN-stimulated genes (ISG), and proinflammatory cytokines in White Pekin ducks infected with these viruses via the ‘natural route’ – dripping into nares, eyes and trachea. We compare immune responses in lung and spleen tissue of the ducks over the course of 3 days p.i. Previously we reported that rgVN1203 infection induces a rapid induction of ISGs in duck lungs, which returns to the baseline by the third day p.i. [17], and we expected similar immune response profiles in this experiment. We anticipated that signatures of pathogenicity may be apparent for D4AT and VN1203, compared to rgVN1203, in proinflammatory cytokines and IFNs. We show that all three H5N1 strains induced rapid, transient IFN responses and moderate inflammation, which likely protected some ducks from manifestation of disease while permitting viral replication and shedding.

Results

Ducks upregulate RIG-I expression in response to H5N1 infections

We infected ducks by the natural route with two closely related H5N1 viruses that differed in known pathogenicity [16, 31], the original isolate of A/Vietnam/1203/04 (H5N1) (VN1203), or A/duck/Thailand/D4AT/04 (H5N1) (D4AT), and a recombinant version of A/Vietnam1203/2004 (H5N1) (rgVN1203). These viral strains differed in several key amino acid residues we predicted might affect innate immune responses (Table 1). Ducks were tested for infection by taking tracheal and cloacal swabs, and positive tracheal swabs were titrated further to reveal that viruses replicated to similar levels (Fig. 1a). To assess differences in viral replication in tissues, we quantified viral matrix gene copy numbers in the tissues of the animals, isolated on each of 3 days p.i. In lungs, matrix gene copy numbers remained relatively stable throughout the three-day period, and D4AT replicated most abundantly of the three viruses with the highest viral titre at 2 days p.i. (Fig. 1b). rgVN1203 had the lowest viral load in the lung, but this difference was statistically significant only for 2 days p.i. In spleen tissues, matrix copy numbers peaked at 1 day p.i., with D4AT remaining high at 2 and 3 days p.i. in some individuals (Fig. 1c). For all three viruses we see significant viral load in the lung and evidence of systemic spread to the spleen, with clearance over time.

Table 1. Known amino acid differences between the viral proteins of three influenza virus strains: recombinant A/Vietnam/1203/2004 (H5N1) (rgVN1203), original isolate A/Vietnam/1203/2004 (H5N1) (VN1203) and A/duck/Thailand/D4AT/2004 (H5N1) (D4AT).

The three amino acid differences between rgVN1203 and the original isolate are highlighted in bold.

Protein	Amino acid position	rgVN1203	VN1203	D4AT
PB2	195	D	D	N
	391	Q	Q	E
	627	K	K	E
PB1-F2	42	Y	Y	C
	51	M	T	M
	56	A	V	V
	79	R	R	Q
	87	G	E	E
PA	421	I	I	S
HA	52	K	K	T
NP	456	V	V	A
NA	5	Q	Q	K
	19	I	I	M
	29	M	M	L
	45	Q	Q	K
	46	S	S	A
	264	N	N	D
NEP	48	T	T	A
	115	A	A	T
NS1	57	A	A	S
	71	G	G	E
	73	S	S	T
	122	T	T	V
	200	N	N	S
	216–225	del.	del.	KMARTIESEV

Fig. 1.

H5N1 influenza A viruses replicated to high levels in infected ducks. (a) Tracheal virus titres in ducks infected with different H5N1 strains. Tracheal and cloacal swabs were collected at 1, 2 and 3 days p.i. and neat swab material was tested for the presence of influenza virus by egg inoculation, and positive tracheal samples were titrated further by calculating EID₅₀. H5N1 replication in tissues was determined by amplification of the influenza A matrix 1 gene. RNA was extracted from lung (b) and spleen (c) tissues of ducks at 1, 2 and 3 days p.i. with rgVN1203, VN1203 or D4AT viruses. Influenza matrix gene copy number is determined against a known copy number of an influenza matrix M1 clone. Each dot represents one duck. Significant differences in mean viral titre each day were determined by two-way ANOVA (P<0.05). *P<0.05, ns not significant.

RIG-I is a key pattern recognition receptor of the IFN response in influenza-infected cells [32]. RIG-I induces IFN, and its own expression is, in turn, induced by IFN and cytokine signalling, in birds [33, 34], fish [35] and humans [36]. To compare the relative expression of the gene encoding RIG-I (DDX58) in different H5N1 infections, we examined its transcript abundance in the lung and spleen tissues of ducks either mock-treated, or infected with rgVN1203, VN1203 or D4AT, normalizing to the housekeeping gene GAPDH. The relative abundance of duck DDX58 transcripts in lung tissue is increased at 1 day p.i., the greatest induction following infection with D4AT (Fig. 2a). By 3 days p.i., the relative abundance is decreased. In spleen tissue, DDX58 upregulation peaks at 1 day p.i. (Fig. 2b). While the response to each virus follows a slightly different time course, RIG-I is highly upregulated in both the lung and spleen, peaking at 1 day p.i.

Fig. 2.

RIG-I is upregulated early in lung and spleen tissues of H5N1 influenza A-infected ducks. RNA was extracted from lung (a) and spleen (b) tissues of ducks at 1, 2 and 3 days p.i. with rgVN1203, VN1203 or D4AT viruses. DDX58 transcription was analysed by qPCR and normalized to GAPDH. Fold-induction compared to a mock-treated animal is shown (each dot represents one duck). Significant differences between the mean gene expression levels in tissues infected with different viruses were determined by two-way ANOVA (P<0.05), *P<0.05.

Ducks upregulate IFNs early in response to H5N1 infections

To examine the contribution of different IFNs, we examined their transcription levels in lung and spleen tissues of the ducks by qPCR (Fig. 3). Genes coding IFN alpha (IFNA), beta (IFNB), gamma (IFNG) and lambda (IFNL) were upregulated on the first day p.i., compared to mock-treated animals. IFNA transcript abundance in the lungs increased at 1 day p.i. with rgVN1203, and in one VN1203-infected animal, and not at all by D4AT (Fig. 4a). In spleen tissue at 1 day p.i. IFNA was highly induced by VN1203 and especially D4AT (Fig. 3b). IFNB, conversely, was more strongly upregulated in lung tissue than in spleen, in concordance with RIG-I expression. Lung IFNB transcripts were upregulated most by D4AT on the first day, and returned to the baseline by the third day (Fig. 3c). In rgVN1203-infected lungs IFNB was not upregulated on the first day, but increased at 2 days p.i., before returning to mock level by 3 days p.i. In the spleen, IFNB transcript abundance was highest at 1 day p.i. for all viruses, but induction was modest (Fig. 3d). IFNG transcription in lung tissue was upregulated most by D4AT at 1 day p.i., and returned to the baseline at 3 days p.i. (Fig. 3e). In spleen tissue, IFNG upregulation was more modest, and returned to the baseline by 2 days p.i. (Fig. 3f). IFNL was upregulated in both tissues, with D4AT eliciting the highest response in the lung (Fig. 3g) and spleen (Fig. 3h). Thus, all IFNs are induced by infections with highly pathogenic avian influenza, with D4AT eliciting the highest IFN response.

Fig. 3.

IFNs are upregulated in lung and spleen tissues of H5N1 influenza A-infected ducks at 1 day p.i. RNA was extracted from lung and spleen tissues of ducks at 1, 2 and 3 days p.i. with rgVN1203, VN1203 or D4AT viruses. IFN transcription was analysed by qPCR and normalized to GAPDH. Fold-induction compared to mock-infected animals is shown for genes encoding IFN-α in lung (a) and spleen (b); IFN-β in lung (c) and spleen (d); IFN-γ in lung (e) and spleen (f); IFN-λ in lung (g) and spleen (h). Each dot represents one duck. Significant differences between mean expression levels in tissues infected with different viruses were determined by two-way ANOVA (P<0.05). *P<0.05, ** P<0.01, ***P<0.001, ****P<0.0001, ns not significant.

Fig. 4.

ISGs are upregulated early in lung and spleen tissues of H5N1 influenza A-infected ducks. RNA was extracted from lung and spleen tissues of ducks at 1, 2 and 3 days p.i. with rgVN1203, VN1203 or D4AT viruses. ISG transcription was analysed by qPCR and normalized to GAPDH. Fold-induction compared to mock-infected animals is shown for OASL in lung (a) and spleen (b); IFIT5 in lung (c) and spleen (d); MX1 in lung (e) and spleen (f). Each dot represents one duck. Significant differences between mean expression levels in tissues infected with different viruses were determined by two-way ANOVA (P<0.05), *P<0.05, ns not significant.

Ducks highly upregulate ISGs in response to H5N1 infections

To compare gene expression downstream of RIG-I and IFN induction we looked at the upregulation of three representative ISG transcripts in the lung and spleen tissues. OASL was most upregulated at 1 day p.i. by D4AT in the lung (Fig. 4a) and spleen (Fig. 4b). IFIT5 transcription upregulation was highest at 1 day p.i. in the lung (Fig. 4c) and spleen (Fig. 4d). MX1 transcripts were upregulated in lungs at 1 day p.i. (Fig. 4e), but were highly induced in spleen over all 3 days (Fig. 4f). All ducks showed a strong upregulation of three ISGs in the lung and spleen early in the immune response, which decreased over time, and showed individual variation in expression.

Ducks upregulate proinflammatory cytokines early in response to H5N1 infections

To assess the potential contribution of hypercytokinemia to H5N1 pathology in ducks we examined the transcription of three proinflammatory cytokine genes – IL1B, IL6 and IL-18 – and the anti-inflammatory cytokine gene IL10. IL-1β and IL-6 are key mediators of lung inflammation in severe influenza infections of humans and chickens [37, 38]. In our experimentally infected ducks there was a moderate upregulation of cytokines, but the response appears controlled and not sustained, with all transcript levels in both tissues returning to control levels by 3 days p.i. IL1B was only slightly induced in the lung (Fig. 5a) and spleen (Fig. 5b). In duck lungs IL6 was upregulated most by VN1203 (Fig. 5c). In spleen tissue, relative transcript abundance was increased similarly by all viruses on the first day p.i., and returned to control levels by 2 days p.i. in the spleen (Fig. 5d). IL18 was slightly upregulated at 2 days p.i. for rgVN1203 in the lung (Fig. 5e), but slightly induced by all viruses at 1 day p.i. in the spleen (Fig. 5f). IL10 induction was low in both tissues, peaking in the lungs at 1 day p.i. in VN1203-infected ducks (Fig. 5g), and peaking in the spleen at 1 day p.i. in response to D4AT (Fig. 5h). Thus, ducks make a proinflammatory cytokine response to H5N1 infections, with greater response to the lethal VN1203 and D4AT viruses, compared to the attenuated rgVN1203.

Fig. 5.

ILs are upregulated in lung and spleen tissues of H5N1 influenza A-infected ducks at 1 day p.i. RNA was extracted from lung and spleen tissues of ducks at 1, 2 and 3 days p.i. with rgVN1203, VN1203 or D4AT viruses. IL transcription was analysed by qPCR and normalized to GAPDH. Fold-induction compared to mock-infected animals is shown for IL1B in lung (a) and spleen (b); IL6 in lung (c) and spleen (d); IL18 in lung (e) and spleen (f) and IL10 in lung (g) and spleen (h). Each dot represents one duck. Significant differences between mean expression levels in tissues infected with different viruses were determined by two-way ANOVA (P<0.05). *P<0.05, ** P<0.01, ns not significant.

Discussion

We report that experimental infections of White Pekin ducks with three similar highly pathogenic H5N1 viruses produced rapid and robust IFN responses and inflammation, which peaked on the first day p.i. and were not sustained. Between the individual ducks there was a lot of variation, as expected. Viral load was initially high in both organs and remained stable in the lungs, but decreased over time in the spleen. Although the three viral strains are closely related, between them D4AT replicates most quickly and induces higher IFN responses, and conversely rgVN1203 replicates most slowly and elicits a lower magnitude response, as expected. Proinflammatory cytokine responses to infection with VN1203 and D4AT showed increased IL6 and IFNG gene expression, compared to the attenuated rgVN1203. Unexpectedly, the more lethal strain D4AT, elicited higher IFN responses and lower IL-6 than VN1203, suggesting dysregulated innate responses may contribute to the pathogenicity of these viruses in ducks. Alternatively, the pathogenicity of both VN1203 and D4AT in Pekin ducks is very high [16] and may be too close for comparison between them, or the kinetics of the innate responses were different.

RIG-I transcripts were upregulated in response to all three viruses starting on the first day p.i., and modest upregulation persisted through the third day. In the infected ducks, induction was highest in response to D4AT on the first day in the lungs. RIG-I upregulation suggested a productive IFN response, so we examined the transcription of type I, type II and type III IFNs in the ducks and saw that induction spiked sharply on the first day p.-i., and returned to the baseline by the third day, in both tissues. Type I IFN signalling controls inflammation and limits influenza virus-induced lung injury [39]. Consistent with RIG-I signalling leading to release of, primarily, IFN-β, we observed greater induction of IFNB transcription than IFNA. IFN-β knockout mice show decreased survival and higher influenza viral titres in the lung [40], and fibroblasts from RIG-I knockout mice fail to induce several downstream ISGs [41]. Thus, we predict that a rapid IFN response protects ducks from succumbing to disease. Expression of IFNB is delayed in lungs infected with rgVN1203, which peaks at 2 days p.i., potentially due to slower replication. Attenuation was shown to be due to amino acid differences between these strains in PB1-F2 [31], which is also known to antagonize IFN in mammals [42]. Surprisingly, the IFNB response was highest to D4AT, the virus known to cause the highest mortality in ducks. Of note, the influenza NS1 protein, the primary IFN antagonist of the virus [43], differs between the D4AT and VN1203 strains by five amino acid substitutions, and a ten amino acid C-terminal deletion in VN1203. The C-terminal ESEV motif, present in D4AT NS1, was initially thought to be important in IFN antagonism. However, VN1203 NS1 variants with truncated wild-type or ESEV termini equally inhibited an IFN-β promoter reporter in MDCK cells [44]. Our results are consistent with this, since D4AT, with the terminal ESEV, induced higher IFNB gene expression in duck lung and spleen tissues than VN1203. Alternatively, differences in internal genes between strains account for the different responses. A severe outcome in mice correlating with type I IFN and proinflammatory cytokine release in lungs, is due to increased viral replication in myeloid cells, and is determined by internal genes of avian H5N1 viruses [45]. One residue of D4AT known to contribute to slightly increased pathogenicity in ducks is PB2 627E [46].

IFNG was upregulated in the lung and spleen at 1 day p.i. with D4AT, and less for VN1203 or rgVN1203. We observed a 100-fold upregulation of IFNG transcripts at 1 day p.i. in response to D4AT. IFN-γ and other markers of a Th1 type immune response are seen in VN1203-infected chickens at 1 day p.i. [47]. IFNG transcripts were also increased in the brain and hearts of infected ducks at 3 days p.i., which has been implicated in neurological symptoms [19]. Muscovy ducks infected with H5N1 strains showed very high expression of IFNG transcripts in brain, and lung (increasing by 3 days p.i.), with increased expression correlating with virulence of the strains [26]. In our H5N1 infections of ducks, we observed that increased IFNG expression in the lung and spleen correlates with known pathogenicity of the strains.

ISG transcription was induced to very high levels at 1 day p.i., and declined gradually in all three viral infections. Upregulation of IFIT5 and of OASL was higher in the lungs than in the spleen, while MX1 induction was very moderate in the lungs and highly induced in the spleen. In humans, IFT5 binds and sequesters viral 5′-triphosphate RNA [48], and OASL enhances RIG-I activity [49, 50]. Mx1 is a key component of the antiviral response in mice [51], expressed in all mouse tissues stimulated with poly (I:C) or influenza, and is thought to be expressed by all vertebrate cell types [52, 53]. High Mx induction may explain the decrease of viral titre in spleens compared to lungs. However, the original in vitro investigation of two duck Mx alleles found that they conferred no resistance to lab strains of influenza, compared to mouse Mx [54]. However, a variety of alleles exist in wild ducks [55], and resistance to influenza is allele and strain-dependent in other species. Thus, duck Mx function remains an open question.

The cytokine response in the infected ducks appeared relatively controlled. Hypercytokinemia, or ‘cytokine storm’, can contribute to H5N1 influenza pathology [56] in humans [20–24], mice [57, 58] and chickens [19]. In VN1203- and D4AT-infected ducks, transcripts for proinflammatory cytokines IL-1β and IL-6 were upregulated only on the first day p.-i., and upregulation of IL-1β was very modest. IL6 transcript induction was highest in response to VN1203, and lowest to the attenuated rgVN1203. Our results differ from previous work showing rapid over-induction of proinflammatory cytokines in chickens, but not ducks, in response to infection with two H5N1 strains, one potentially lethal to ducks (A/Muscovy duck/Vietnam/453/2004) and one which did not kill ducks (A/Duck/Indramayu/BBVW/109/2006) [19]. Proinflammatory cytokines were not induced at 1 day p.i. in ducks infected with VT453 and IND109, instead peaking at 3 days p.i. for VT453. Viral strain, replication kinetics and host genetics may all contribute to the different responses observed. While dysregulated proinflammatory responses were expected to be a hallmark of the lethal strains, similar to proinflammatory gene induction increasing each day in lung tissue in DK383 H5N1-infected Muscovy ducks [26], this was not seen within the 3 days p.-i. examined. Thus, we believe the immune responses seen are protective, contributing to survival, at least within the timeframe examined.

Methods

Viruses

Two closely related H5N1 viruses were used in this study A/Vietnam/1203/2004 (H5N1) (VN1203); A/duck/Thailand/D4AT/2004 (H5N1) (D4AT); and a recombinant version of A/Vietnam/1203/2004 (H5N1) (rgVN1203), which was made by reverse genetics [29]. Previously, the intravenous pathogenicity indexes (IVPI) for D4AT and VN1203 in mallards were determined to be 2.87 and 2.32, respectively [16]. The IVPI for VN1203 was higher (2.85) in domestic Pekin ducks [16]. A comparison of pathogenicity index following inoculation by the natural route yielded 1.24 for wtVN1203 and 0.58 for rgVN1203 [31]. Phylogenetic trees based on HA sequence place both D4AT and VN1203 in clade 1 [59, 60] and other genes are also closely related [61]. Amino acid sequence information to generate Table 1 is derived from GenBank accession numbers: A/VietNam/1203/2004 (H5N1) – PB2 (HM006756.1); PB1 (HM006757.1); PA (HM006758.1); HA (HM006759.1); NP (HM006760.1); NA (HM006761.1); M (HM006762.1); NS (HM006763.1). D4AT with original strain name A/Dk/Thailand/71.1/2004 (H5N1) - PB2 (AY651716.1); PB1 (AY651662.1); PA (AY651608.1); HA (AY651331.1); NP (AY651496.1); NA (AY651443.1); M (AY651385.1); NS (AY651551.1). rgVN1203 sequence differences were previously reported [31].

Duck infections, swab titrations

Outbred 6-week-old White Pekin ducks purchased from Metzer Farms were infected with 10⁶ EID₅₀ of VN1203, rgVN1203 or D4AT by dripping virus into nares, eyes and trachea, or mock-treated with PBS-only. Tracheal and cloacal swabs were taken from ducks prior to sacrifice and stored until titration in swab freezing media, made as described previously [62]. Ducks were confirmed infected by egg inoculations of neat swab material, and positive tracheal swabs were titred by determining their EID₅₀ by the method of Reed and Muench [63]. Briefly, 10-fold dilutions were injected in three eggs, incubated for 2 days and allantoic fluid tested for HA activity [64]. The titre is the inverse of the dilution at which 50 % of the eggs were infected, and is calculated from the number of positive eggs at each dilution. Most cloacal swabs were negative, but a few undiluted cloacal swabs tested positive (3/8 for VN1203, 1/8 rgVN1203, and 3/8 D4AT), but not all eggs were positive, so they were not titred further. All tracheal swabs were titred in eggs. Ducks were euthanized and lung and spleen samples were harvested 1, 2 or 3 days p.i.

All virus strains caused noticeable disease signs, including cloudy eyes, loss of appetite and reduced mobility, but no neurological symptoms were observed, and no ducks succumbed to death within the 3 days. Weight differences were expected to be small and were not measured in this study. Previously, wtVN1203-infected ducks showed no weight gain, while rgVN1203 continued to gain weight, but weights were not significantly different before day 4 [31]. All animal experiments in this study were approved by the Animal Care and Use Committee of St. Jude Children’s Research Hospital and performed in compliance with relevant institutional policies, National Institutes of Health regulations, and the Animal Welfare Act.

Quantification of gene expression by qPCR amplification

Total RNA was extracted from weighed tissues by immediate homogenization in TRIzol (1 ml per 200 mg tissue) and RNA extracted according to their protocol (Invitrogen) and precipitated. Resuspended RNA was quantified by NanoDrop spectrophotometer (Thermo Scientific), and stored at −80 °C until use. An aliquot was treated with DNase I and used for the generation of cDNA. First strand cDNA synthesis was performed using OligodT and Superscript III in the presence of RNaseOUT Recombinant RNAse Inhibitor (Invitrogen).

To evaluate transcript abundance of DDX58, type I-III IFNs, ISGs: MX1, IFIT5, OASL and cytokines: IL6, IL1B, IL10, IL18 in influenza-infected tissues, we used qPCR with the 7500 Fast Real Time PCR instrument (Applied Biosystems) and the FastStart TaqMan Probe Master mix (Roche). Primers and probe sets (Integrated DNA Technologies (IDT)) were designed using an online IDT Tool and validated using GAPDH as a control (Table S1, available in the online version of this article). Quantitative PCR amplifications were done using cDNA for mock-treated or influenza-infected lung and spleen tissues. PCR conditions were 95 °C for 10 min, and 40 cycles of denaturation at 95 °C for 15 s followed by annealing-extension at 60 °C for 1 min. All samples were run in triplicate, each assay was repeated at least twice. Expression levels of the target genes were normalized to the housekeeping gene GAPDH using the relative quantitation method (ΔΔCT), and the analysis was done using the 7500 Fast system software version 1.4 (Applied Biosystems). Absolute quantification of the influenza matrix 1 (M1) gene was accomplished with qPCR, by comparing sample cDNA dilution series to a standard curve of an M1 clone in the pGEM-T Easy vector (Promega).

Statistical analyses

Analysis of variance was done using a two-way ANOVA, with a post hoc Tukey multiple comparisons test, comparing the mean value of each group to every other group. While infected samples at 1 day p.i. were always significantly different from mock-treated samples, and different from samples at other days, these differences are obvious and bars indicating significance were omitted for clarity. Statistically significant differences between infections at each timepoint are indicated where present.