PA-X is an avian virulence factor in H9N2 avian influenza virus

Influenza A viruses encode several accessory proteins that have host- and strain-specific effects on virulence and replication. The accessory protein PA-X is expressed due to a ribosomal frameshift during translation of the PA gene. Depending on the particular combination of virus strain and host species, PA-X has been described as either acting to reduce or increase virulence and/or virus replication. In this study, we set out to investigate the role PA-X plays in H9N2 avian influenza viruses, focusing on the natural avian host, chickens. We found that the G1 lineage A/chicken/Pakistan/UDL-01/2008 (H9N2) PA-X induced robust host shutoff in both mammalian and avian cells and increased virus replication in mammalian, but not avian cells. We further showed that PA-X affected embryonic lethality in ovo and led to more rapid viral shedding and widespread organ dissemination in vivo in chickens. Overall, we conclude PA-X may act as a virulence factor for H9N2 viruses in chickens, allowing faster replication and wider organ tropism.


INTRODUCTION
Influenza A viruses (IAV) have segmented negative-sense RNA genomes encoding 10 core proteins and a variable number of strain-specific accessory proteins [1,2]. Due to its small genome size and nuclear replication, IAV has evolved a number of ways to increase its protein coding capacity including the use of splice variants (M2, NEP and the more recently discovered M42, PB2-S1 and NS3 proteins), encoding multiple open reading frames (ORFs), both nested and overlapping on a single gene segment (e.g. PB1-F2, PB1-N40, NA43) and ribosomal frameshifts leading to multi-ORF fusion proteins (PA-X) [3][4][5][6][7][8] Influenza A virus segment three primarily encodes the polymerase acidic protein (PA). PA is an integral part of the influenza virus RNA-dependent RNA polymerase (RdRp) and contains two functional domains, an N-terminal endonuclease (endo) domain, responsible for cleaving host capped RNAs used to prime viral transcription, and a C-terminal domain, associated with the core of the RdRp [9]. The 3′-boundary of PA endo domain coding sequence contains a conserved ribosomal frameshift site where a rare arginine codon facilitates ribosomal stalling, which is followed (at low level) by ribosomal slippage into a +1 open reading frame (ORF), facilitated by an upstream UUU codon that allows tRNA realignment [10,11]. The resulting protein, PA-X, is a fusion between the first 191 amino acids of PA (the endo domain) and up to 61 amino acids translated from the X-ORF in the +1 reading frame. PA-X has been shown to mediate degradation of host cell mRNAs and disruption of host mRNA processing, leading to host cell shutoff and a dampened innate immune response [11][12][13][14][15][16][17].
H9N2 avian influenza viruses are enzootic through much of Asia, the Middle East and Africa [18]. H9N2 viruses cause massive economic damage due to their impact on poultry production systems, causing moderate morbidity and mortality, especially in the context of viral or bacterial coinfection. Additionally, H9N2 viruses pose a direct zoonotic threat to humans and are considered viruses with pandemic potential [18,19]. As well as being zoonotic threats in their own right, H9N2 viruses have contributed polymerase genes to multiple zoonotic avian influenza viruses, including epidemic H7N9 [20].

ACCESS
Several studies have investigated the role of PA-X on virulence and replication of avian influenza viruses in mammalian and avian hosts. There is little consensus between these studies and there appears to be virus strain-and host speciesspecific differences in whether PA-X expression increases or decreases replication, virulence and transmissibility of IAV [11,15,[21][22][23][24][25][26][27][28][29]. In H9N2 viruses, expression of full length PA-X has been shown to be a virulence factor in mammalian infection systems [21,23]. However, it is unclear what role PA-X has in these viruses in their natural chicken hosts.
In this study we set out to investigate the role of PA-X in a contemporary G1-lineage H9N2 virus, typical of viruses circulating in the Middle East and South Asia [18]. We found that the H9N2 virus expressed a PA-X capable of robust host shutoff which correlated with PA-X expression. Removing PA-X expression decreased viral replication in mammalian, but not avian cell culture systems, although it did reduce embryonic lethality in ovo. In vivo, in the natural chicken host, ablation of PA-X expression led to delayed shedding and restricted viral dissemination. Overall, our results suggest PA-X may be an H9N2 virulence factor in the natural avian host, allowing faster replication and increased visceral organ tropism.

Ethics statement
All animal experiments were carried out in strict accordance with the European and United Kingdom Home Office Regulations and the Animal (Scientific Procedures) Act 1986 Amendment regulation 2012, under the authority of a United Kingdom Home Office Licence (Project Licence Numbers: P68D44CF4 X and PPL3002952).
Viruses were rescued as described elsewhere [32]. Briefly, 250 ng of each segment plasmid were co-transfected into 293Ts plated in six well plates using lipofectamine 2000. At 16 h post-transfection, media was changed to reverse genetics media (DMEM +2 mM glutamine, 100 U ml −1 penicillin, 100 U ml −1 streptomycin, 0.14 % BSA, 5 µg ml −1 TPCK-treated trypsin). Following a 48 h incubation at 37 °C, 5 % CO 2 , supernatants were collected and inoculated into embryonated hens' eggs for subsequent harvest of virus infected allantoic fluid. Virus segment three sequences were confirmed by RT-PCR and Sanger sequencing.

In vitro translation and autoradiography
For in vitro translations to visualise radiolabelled proteins, the TnT Coupled Reticulocyte Lysate system (Promega; #L4610) was utilised as per the manufacturers' instructions using 200 ng of T7 promoter containing PHW2000 plasmid DNA. Reactions were incubated at 37 °C for 90 min then denatured in 50 µl protein loading buffer and analysed on a 15 % SDS-PAGE, to allow greatest and autoradiography. X-ray films were developed using a Konica SRX-101A X-ograph film processor as per the manufacturers' instructions.

Host cell shutoff assays
The β-gal shutoff reporter assays were performed as described elsewhere [11]. Briefly, 293T or DF-1 cells were co-transfected with expression plasmids for the influenza segment 3 (PA) and β-galactosidase (β-gal) reporter. Then 48 h later, cells were lysed with 100 µl of 1 x Reporter lysis buffer (Promega). β-gal expression was assayed using the β-galactosidase enzyme assay system (Promega). A Promega GloMax Multi Detection unit was used to measure absorbance at 420 nm. Western blotting was performed according to standard protocols. PA was detected using a rabbit polyclonal raised to bacteriallyexpressed PA [33], while tubulin was detected with rat (Abcam; YL1/2) or rabbit (Abcam; ab15246) monoclonal antibodies as a loading control.
For the shutoff activity assays using live virus, MDCKs were infected with virus at a high MOI of 5. At 7.5 h postinfection, cells were washed and the medium changed to 1 ml of complete DMEM containing 10 µg ml −1 of Puromycin dihydrochloride from Streptomyces alboniger for 30 min. Cells were washed and lysed in protein loading buffer and run on an SDS-PAGE and were then Western blotted, probing for puromycin with a mouse monoclonal antibody to puromycin (Millipore; MABE343). Puromycylated protein synthesis was quantified in the region of the gel between 45 kDa and 80 kDa. Protein quantification following Western blot was measured by densitometry using ImageJ analysis software.

Mini-replicon assays
The 293T or DF-1 cells were transfected with plasmids encoding PB2, PB1, PA and NP, along with a firefly luciferase vRNA-like reporter under a cell-type specific polI promoter (human for 293T, chicken for DF1). The following concentration of plasmid were used for 293Ts, PB2-80 ng, PB1-80 ng, PA-20 ng, NP-160 ng, pPol I Luc-800 ng, these amounts were doubled for DF1s. After 48 h, media was removed, and cells were lysed in 100 µl of 1 x Passive Lysis Buffer (Promega). A Promega GloMax Multi Detection unit was used to measure luciferase activity following the manufacturer's instructions.

Virus replication assays
MDCK and CK cells were infected at a low MOI of 0

Virus infection, transmission and clinical outcome in vivo
In vivo studies were performed with 3 week-old White Leghorn birds (VALO breed). Prior to the start of the experiments, birds were swabbed and bled to confirm they were naïve to the virus. All infection experiments were performed in self-contained BioFlex B50 Rigid Body Poultry isolators (Bell Isolation Systems) at negative pressure. Then 10 birds per group were directly inoculated with 10 4 p.f.u. of virus via the intranasal route. Mock infected birds were inoculated with sterile PBS as an alternative. One day post-inoculation eight naïve contact birds were introduced into each isolator to determine viral transmissibility.
Throughout the experiment, birds were swabbed in the buccal and cloacal cavities (on day 1-8, 10 and 14 post-infection). Swabs were collected into 1 ml of virus transport media (WHO standard). Swabs were soaked in media and vortexed for 10 s before centrifugation. Viral titres in swabs were determined via plaque assay.
At day 2 post-infection, three birds per group (directly infected, contact and mock infected) were euthanised via overdose of pentobarbital (at least 1 ml). A panel of tissues were collected and stored in RNA later at −80 °C until further processing for qRT-PCR of virus gene copies and cytokines. Birds were observed twice daily by members of animal services and whilst procedures were carried out for the presence of clinical signs of infection. On day 14 post-infection, all remaining birds were culled via overdose of pentobarbital or cervical dislocation.

RNA extraction and RT-PCR from chicken tissues
The 30 mg of tissue collected in RNA later was mixed with 750 µl of Trizol. Tissues were homogenised using the Retsch MM 300 Bead Mill system (20 Hz, 4 min). Then 200 µl of chloroform was added per tube, shaken vigorously and incubated for 5 min at room temperature. Samples were centrifuged (9 200 g, 30 min, 4 °C) and the top aqueous phase containing total RNA was added to a new microcentrifuge tube, subsequent RNA extraction was then carried out using the QIAGEN RNeasy mini kit following manufacturers' instructions.
Next 100 ng of RNA extracted from tissue samples was used for qRT-PCR. All qRT-PCR was completed using the Superscript III platinum One-step qRT-PCR kit (Life Technologies) following manufacturer's instructions for reaction set up. Cycling conditions were as follows: i) 5 min hold step at 50 °C, ii) a 2 min hold step at 95 °C, and 40 cycles of iii) 3 s at 95 °C and iv) 30 s annealing and extension at 60 °C. Cycle threshold (CT) values were obtained using 7500 software v2.3. Mean CT values were calculated from triplicate data. Negative controls were included within each plate to determine any unspecific amplification or contamination. Within viral M segment qRT-PCR an M segment RNA standard curve was completed alongside the samples to quantify the amount of M gene RNA within the sample from the CT value. T7 RNA polymerase-derived transcripts from UDL-01 segment seven were used for the preparation of the standard curve.
Within the cytokine qRT-PCRs, three housekeeping genes were included per sample (RPLPO-1, RPL13 and 28S rRNA) that had been previously determined to be stable in a broad range of tissues. Briefly, the geNorm algorithm (Vandesompele et al., 2002) was adopted to calculate the stability for each reference gene and the optimal reference gene number from raw Cq values of candidate reference genes using qbase +real time qPCR software version 3.0 (Biogazelle).

Statistical analysis
All statistical analysis was carried out using GraphPad Prism 6/7 software. Parametric vs non-parametric distribution of data was assessed prior to deciding on the statistical test to use. Correlation coefficients were calculated using Spearman's Rank correlation coefficient.

Generation of H9N2 viruses with altered PA-X expression
To investigate the role of PA-X in an H9N2 background we generated a panel of mutants in the background of a contemporary G1-lineage H9N2 virus UDL-01, typical of viruses still circulating in South Asia and the Middle East [34,35]. Mutants were generated (in cDNA copies of segment three cloned into a reverse genetics plasmid) using a previously validated approach, with the frameshift site mutated in such a way that ribosomal slippage should be inhibited (FS); additionally, a panel of truncated PA-X mutants with premature termination codons (PTC) spaced throughout the X-ORF, were made (Table 1, Fig. 1a) [11,25]. All mutations were synonymous in the coding sequence of PA.
To confirm ablation of PA-X expression and/or validate the truncations, in vitro transcription/translation assays in rabbit reticulocyte lysate were carried out directly from the plasmids. Comparable levels of PA expression were seen across every construct, confirming the PA-X mutations did not affect PA expression (Fig. 1b, c). As expected [11,15,25], PA-X expression was visible as a minor polypeptide species migrating above the 25 kDa marker, whose abundance was drastically reduced in the FS mutant ( Fig. 1b red asterisks,  D). Upon X-ORF truncation, a ladder effect where the size of PA-X was progressively decreased by the PTC mutations could be seen, while both PTC1 and PTC2 also showed a reduction of PA-X expression. Overall, these results indicate that a previously used strategy for altering PA-X expression is also successful in this H9N2 virus background.

A PA-X frameshift mutation abrogates host cell shutoff activity
PA-X plays a key role in IAV host cell shutoff, therefore the ability of the H9N2 PA-X variants to repress cellular gene expression was tested using transfected segment three plasmids in β-galactosidase (β-gal) reporter assays. HEK 293Ts or DF-1 cells were transfected with the β-gal plasmid along with segment three plasmids or an empty vector control, β-gal accumulation was measured by enzyme assays 48 h later and normalised to empty vector. UDL-01 WT PA-X but not the A/Puerto Rico/8/34 (PR8) PA-X showed robust repression of β-gal expression in both cell types, and introduction of the FS mutant ablated the UDL-01 activity (Fig. 2a, b), indicating that the shutoff activity of segment three is dependent on PA-X expression but varies according to IAV strain, as previously shown [11,17,25]. When shut-off activity of the UDL-01 PTC mutants was tested in 293 T cells, PTC1 and PTC2 showed a minor and non-statistically significant reduction in host shutoff activity compared to UDL-01 WT while PTC3 and PTC4 had no apparent effect. Western blot analysis of PA expression confirmed similar expression levels of the various PA polypeptides, with the exception of the UDL-01 FS mutant, which accumulated to higher levels in both 293T and DF1 cells (Fig. 2a, lower panels). The increased expression of the FS mutant may reflect removal of the auto-repressive PA-X function from the segment, as previously noted [11]. Thus UDL-01 encodes an active PA-X polypeptide, whose shutoff function does not strongly depend on the full X-ORF sequence.
To investigate whether the effect of the FS mutation on host shutoff activity of UDL-01 segment three seen with plasmidbased assays could be recapitulated with infectious virus, mutant viruses were generated by reverse genetics. MDCK cells were then infected with WT and FS UDL-01 viruses at a high MOI [5], and pulsed with puromycin for 30 min to label nascent polypeptides [36], before being lysed. Lysates were then run on SDS-PAGE Western blotted for puromycin (Fig. 2c). The region of the blot corresponding to ~80-50 kDa, above where a pair of virally induced protein species (Fig. 2c, arrowheads) were seen, was assessed by densitometry to measure host protein synthesis levels within the cell. UDL-01 WT virus reduced cellular protein synthesis by over 50 % compared to uninfected cells while the FS mutant only caused <20 % host shutoff (Fig. 2d). Furthermore, we have recently shown that UDL-01 WT virus is able to cause host cell shutoff in avian cells [37]. These data corroborate the plasmid-based methods previously used and showed that in the context of infectious virus, UDL-01 expresses a classically active PA-X protein.

PA-X expression does not affect H9N2 polymerase activity
Influenza PA-X has been suggested to modulate polymerase activity [15,22,24,27], although these studies did not all agree on whether PA-X promotes or suppresses polymerase activity; the effect may be strain-as well as cell-type dependent.
Therefore, we investigated the effect of PA-X expression on H9N2 polymerase activity in avian and mammalian cells. Cells were co-transfected with plasmids encoding the polymerase components and NP, alongside a viral RNA-like reporter encoding luciferase. In 293Ts, PB2 and PB1 from the mammalian adapted strain PR8 were used to overcome the restriction of avian IAV polymerase in these cells, whereas in avian DF-1 cells, the full polymerase from UDL-01 was used.
Exchanging WT UDL-01 segment three with the different PA-X mutants had no significant effect on polymerase activity in either mammalian or avian cells (Fig. 2e, f). Furthermore, upon Western blotting cell lysates, no differences in PA or tubulin expression were seen, further confirming that PA-X expression did not alter PA accumulation (Fig. 2g).

Viruses with abrogated PA-X expression have a minor replicative defect in mammalian but not avian cells
Although no difference in polymerase activity was seen between the different PA-X mutants, multicycle growth curves were performed in mammalian and avian systems to determine whether differences in PA-X may affect viral replication in a more biologically relevant context. PA-X removal has been shown to impact viral replication in several previous studies [15,22,23,27], though similarly to polymerase activity these studies tend to disagree about whether PA-X expression enhances or suppresses viral replication and here too, the effect may be virus strain and host-specific.
Initially, plaque size in MDCKs was assessed to determine if any gross replication defects could be seen in the mutants as plaque phenotype is a proxy for replicative fitness in influenza viruses. A modest, but statistically significant reduction in plaque diameter was observed within UDL-01 when PA-X was removed or truncated up to PTC3 (Fig. 3a, b). Average plaque diameters decreased from 1.7 mm to 1.2 mm, 1.25 mm, and 1.36 mm respectively. As UDL-01 FS had the largest impact on plaque diameter and host shutoff, replication  kinetics of this virus were examined. MDCKs were infected at a low MOI and virus titres were assessed over a time course. UDL-01 WT and FS showed very similar growth kinetics, although UDL-01 WT displayed significantly higher titres at 72 h post-infection compared to UDL-01 FS (Fig. 3c). From 48 h post-infection there was a trend for decreased viral titres with UDL-01 FS compared to UDL-01 WT, implying in this particular H9N2 strain PA-X expression may slightly enhance viral replication in MDCK cells. However, when a similar growth kinetics experiment was performed in avian primary chicken kidney (CK) cells, there was no significant difference between UDL-01 WT and FS at any time point, suggesting a host-specific effect (Fig. 3d). To assess this further, replication kinetics were assessed in 10-day-old fertilised hens' eggs. Eggs were inoculated with 100 p.f.u. of each virus and allantoic fluid was harvested periodically. In ovo, UDL-01 WT and FS viruses did not exhibit any significant differences in viral replication throughout the course of infection (Fig. 3e).
Overall, the impact of mutating PA-X on the replication of UDL-01 was variable and the small differences appeared to be host-dependent, with PA-X playing a role in replication in mammalian, but not avian systems.

Viruses with abrogated PA-X expression have lower embryonic lethality
PA-X expression has previously been shown to alter viral pathogenicity in animal models e.g. [11, 22-24, 27, 28] as well as in ovo [25]. Prior to performing an in vivo experiment we infected 10-day-old embryonated hens' eggs with serial dilutions of UDL-01 WT or FS mutants and assessed embryonic lethality over 84 h, as previously described [25]. When percentage survival was plotted against viral dilution a clear difference could be seen between UDL-01 WT and FS (Fig. 3f). The inoculum size of UDL-01 WT negatively affected embryo survival in a dose dependant manner as seen by the ascending trend line (R 2 =0.98, ρ=1 by Spearman's rank correlation coefficient), whereas UDL-01 FS survival was not dose-dependent (R 2 =0.00, ρ=0 by Spearman's rank correlation coefficient). Overall, these data were suggestive of a difference in a pathogenicity between the WT and PA-X deficient viruses with reduction of PA-X expression potentially leading to lower pathogenicity in ovo, as we have previously described [25].

Viruses lacking PA-X expression have delayed shedding and reduced visceral tropism in vivo
As the effect of PA-X expression by H9N2 viruses has yet to be assessed in their natural host, we performed an in vivo experiment to test the effect of PA-X on virus replication, transmission, tropism, pathogenicity and cytokine expression in chickens. Groups of ten 3 week-old White Leghorn (VALO breed) chickens were inoculated intranasally with 10 4 p.f.u. of either UDL-01 WT or FS virus (or sterile PBS). One day post-inoculation, eight naïve contact birds were introduced into each directly infected group to assess viral transmission. Birds were swabbed daily in both buccal and cloacal cavities to determine viral shedding. Throughout the study period birds were monitored for clinical signs, however only very minimal signs were seen in any birds (data not shown). Furthermore, in both directly infected and contact birds, no culturable virus was detected from cloacal swabs. Directly infected birds in both groups showed robust buccal shedding from days 1-6, peaking at titres of over 10 4 p.f.u. ml −1 (Fig. 4a). However, birds infected with UDL-01 FS showed a delayed buccal shedding profile compared to UDL-01 WT infected birds, shedding significantly less on days 1-2. By day 3 post-infection, shedding levels were comparable between groups and by day 6, an increased number of animals infected with UDL-01 FS shed compared to WT (2 of 7 for UDL-01 WT vs 4 of 6 for UDL FS). Within both groups, viral shedding was cleared by day 7 post-infection. To estimate the total amount of virus shed by each group, the area under the shedding curves (AUC) were calculated for both viruses; giving 81 674 for WT UDL-01 versus 241 229 for the FS mutant. This suggested that UDL-01 FS infected birds shed more virus buccally over the course of the experiment than the UDL-01 WT infected birds.
All contact birds in both groups became infected and showed buccal shedding from 1 day post-exposure to the directly infected birds, indicating robust contact transmission of both viruses (Fig. 4b). A similar buccal shedding pattern to the directly infected birds was seen in the contact birds; UDL-01 FS again showed delayed shedding kinetics, with significantly less virus shed on days 1-2 post-exposure and significantly more virus shed by day 5 post-exposure. Within both contact groups viral shedding was cleared by day 6 postexposure. When AUC values were calculated for the contact bird populations, values were comparable; UDL-01 WT AUC was 94 171 whereas the UDL-01 FS AUC was 104 655. Overall, the shedding profiles of the infected birds suggested that expression of PA-X by the WT virus led to accelerated, but not increased, buccal shedding of virus compared to a virus which lacked PA-X expression.
On day 2 post-inoculation three birds from each groupdirectly infected, contact or mock -were euthanised and a panel of tissues were taken to assess viral tropism and cytokine profiles. After RNA extraction, qRT-PCR for the viral M gene was performed to assess viral replication within different tissues. In tissues isolated from directly infected birds, viral replication was primarily observed within the upper respiratory tract in both UDL-01 WT and UDL-01 FS infected birds (Fig. 4c). There were no significant differences in M gene copy number within the nasal tissue and trachea. However, significantly more viral RNA was detected for UDL-01 WT in the lower respiratory tract in the lung. Within other visceral organs little viral RNA was detected, particularly in tissues collected from UDL-01 FS infected birds. UDL-01 WT RNA could be detected in both the liver and kidneys to significantly higher levels than UDL-01 FS. Therefore, UDL-01 WT virus showed increased viral dissemination compared to UDL-01 FS at day 2 post-inoculation. Tissues taken from the contact birds at day 2 of the experiment (i.e. day 1 post-exposure), again showed robust RNA levels in the upper respiratory tract tissues, the nasal tissue and trachea (Fig. 4d). Viral loads were not significantly different between UDL-01 WT and FS infected birds in any tissues and very low levels of RNA were detected in non-respiratory tissues. Overall, these data show that removal of PA-X from UDL-01 led to reduced viral dissemination at day 2 post-infection for directly inoculated birds.

Cytokine expression in the upper respiratory tract of directly infected birds correlated with viral titres
PA-X is known to alter host responses, at least in part by downregulating protein synthesis [11]. Within different host species, modulation of PA-X expression has been shown to alter host expression of cytokines and chemokines; for example an H9N2 AIV unable to express PA-X has been shown to have decreased expression of IL-6, IL-1β, CCL3, IFN-γ and TNF-α within a mouse model compared to a virus with PA-X expression [23]. Therefore, it was assessed whether UDL-01 WT and FS led to differential cytokine and chemokine responses within the chicken host. The upper respiratory tract tissues were chosen due to robust and comparable viral replication in directly infected animals (Fig. 4c). qRT-PCR for a range of chicken cytokines and markers of the interferon response were assessed. The host gene, RPLPO-1, was used for gene normalisation.
Nasal tissue from directly infected birds displayed little differences in immune response (Fig. 5a). Some minor differences were seen with expression of IFN-β, IFN-γ and IL-18 with UDL-01 WT infected birds generally expressing higher levels of cytokines. These differences only reached significance with the expression of IL-1β and the innate immune effector gene Mx, with UDL-01 WT infected birds expressing higher levels of these immune markers. Immune responses in the tracheas of directly infected birds showed a similar trend to those in the nasal tissue (Fig. 5b). Few differences in cytokine expression were seen between UDL-01 WT and UDL-01 FS infected animals. UDL-01 WT infected animals again tended to have increased expression of IFN-β, IFN-γ and IL-1β although this only reached significantly different levels with CXCLi2. It was worth noting that although UDL-01 WT trended towards having higher levels of cytokines in the upper respiratory tract, these tissues also had higher levels of viral RNA, therefore it is difficult to draw conclusions about whether PA-X is having a direct role on cytokine expression; the trend for reduced cytokines in FS mutants may be a result of reduced viral RNA in tissues infected with these viruses.

DISCUSSION
In this study we set out to investigate whether PA-X expression is a virulence factor for H9N2 viruses in their natural poultry host. As expected, ablating PA-X expression resulted in loss of host shutoff activity, showing that, as with other IAV strains, the shutoff activity of H9N2 viruses is partly due to PA-X expression. We found that PA-X expression resulted in slightly increased replication in mammalian MDCK cells but had no effect on titres in primary chicken cells or eggs, although virus with PA-X caused more embryonic lethality in ovo at higher input doses. We further found that in vivo, ablating PA-X expression led to a delayed shedding profile and lower visceral organ tropism. These results, the first study to investigate the impact of PA-X expression in a low pathogenicity avian influenza virus in its natural host suggest that PA-X expression aids faster virus replication and dissemination in vivo, although we saw little evidence for this acting through viral modulation of induced cytokine levels.
The finding here that expression of PA-X causes more rapid virus shedding after infection but that a virus with ablated PA-X expression potentially sheds for longer than WT virus is similar to what has been seen for H1N1-infected mice [15] and H9N2-infected mice [23]. The latter study showed that when mice were infected with a different lineage of H9N2 virus to the one used here (BJ94 versus G1) that had PA-X ablated, there was decreased virulence associated with reduced virus titres in mouse lungs. Interestingly, PA-X tells a contrasting story in high pathogenicity avian influenza viruses and the 1918 H1N1 pandemic virus. Loss of PA-X in 1918 H1N1 and HPAI H5N1 viruses caused increased virulence in mice [11,[21][22][23] and in ducks and chickens [27]. While Jagger and colleagues did not link the increased virulence in mice of 1918 H1N1 IAV lacking PA-X to effects on virus replication, Gao and colleagues showed that increased virulence in mice after loss of H5N1 PA-X was associated with increased titres of the PA-X null virus in the lungs, brain and blood of infected mice. Similarly, Hu and colleagues showed that increased virulence in chickens, ducks and mice on loss of H5N1 PA-X correlated with increased virus titres of the PA-X null virus.
The authors of the H9N2 infection study in mice (Gao et al. 2015c) proposed that differences in effects of PA-X on virulence could be due to the fact that high pathogenicity viruses induce high levels of cytokine responses but low pathogenicity viruses do not typically induce high levels of cytokines. Since a ΔPA-X mutant of a low pathogenicity virus is less effective at host cell shut off, it was more effective at eliciting an antiviral response resulting in reduced virus replication in their study.
Although no difference in pathogenicity in viruses expressing or lacking PA-X was seen in this study it should not be ruled out there may be a role. UDL-01 WT has shown to be a good model for a moderately pathogenic H9N2 virus in previous studies, causing clear clinical signs and even limited mortality [38,39]. However these studies used Rhode Island Red breed chickens, whereas here we used white Leghorn birds which appear to be more resilient to avian influenza virus infection [40,41]. Therefore, it is possible that the virus lacking PA-X is less pathogenic than UDL-01 WT, but this was not observed due to the lack of clinical signs and mortality in the UDL-01 WT infected groups in this system. It is worth noting we have previously correlated visceral tropism with pathogenicity and clinical signs in UDL-01 [38,39]; in this study UDL-01 WT did show greater visceral tropism than UDL-01 FS, perhaps suggesting an attenuated pathogenicity when PA-X expression is ablated.
Overall, this work suggests PA-X may play a role in H9N2 viruses in birds by allowing more rapid replication and dissemination throughout the host, potentially leading to higher pathogenicity. This work will be useful in future surveillance efforts allowing the assessment of newly sequenced viruses as it suggests viruses expressing a full-length PA-X are likely to have a wider tropism and higher pathogenicity than those that do not. Furthermore, this work suggests that there may be slightly different roles for PA-X in mammalian and avian hosts, potentially helping explain the mechanism by which PA-X works.