Twenty amino acids at the C-terminus of PA-X are associated with increased influenza A virus replication and pathogenicity

The PA-X protein, arising from ribosomal frameshift during PA translation, was recently discovered in influenza A virus (IAV). The C-terminal domain ‘X’ of PA-X proteins in IAVs can be classified as full-length (61 aa) or truncated (41 aa). In the main, avian influenza viruses express full-length PA-X proteins, whilst 2009 pandemic H1N1 (pH1N1) influenza viruses harbour truncated PA proteins. The truncated form lacks aa 232–252 of the full-length PA-X protein. The significance of PA-X length in virus function remains unclear. To address this issue, we constructed a set of contemporary influenza viruses (pH1N1, avian H5N1 and H9N2) with full and truncated PA-X by reverse genetics to compare their replication and host pathogenicity. All full-length PA-X viruses in human A549 cells conferred 10- to 100-fold increase in viral replication and 5–8 % increase in apoptosis relative to corresponding truncated PA-X viruses. Full-length PA-X viruses were more virulent and caused more severe inflammatory responses in mice. Furthermore, aa 233–252 at the C terminus of PA-X strongly suppressed co-transfected gene expression by ∼50 %, suggesting that these terminal 20 aa could play a role in enhancing viral replication and contribute to virulence.


INTRODUCTION
Influenza A virus (IAV) causes considerable economic loss to the global livestock industry and poses a significant threat to public health. IAV is an enveloped negative-strand RNA virus with eight viral RNA segments encoding PB2, PB1, PA, HA, NP, NA, M1, M2, NS1 and NS2. In recent years, several additional proteins were discovered, such as PB1-F2, N40, PA-N155 and PA-N182 (Chen et al., 2001;Muramoto et al., 2013;Vasin et al., 2014;Wise et al., 2009). Previous studies showed that NS1 and PB1-F2 proteins of different influenza viruses have varying lengths that differentially affect viral properties such as pathogenicity (Hai et al., 2010;Wang et al., 2005;Zell et al., 2007). The virulence of H1N1 and H3N2 viruses, with truncated NS1 proteins at the C terminus, in mice and pigs was attenuated, unlike full-length NS1 virus infections (Kochs et al., 2009;Soló rzano et al., 2005). H9N2 viruses with different forms of NS1 showed varying replication rates and transmission ability, and induced different levels of inflammatory cytokines in chickens (Kong et al., 2015). Extension of PB1-F2 from a truncated 11 aa to full-length in pandemic 2009 H1N1 virus (pH1N1) moderately enhanced viral replication and cytokine response in pigs (Pena et al., 2012).
PA-X in the 1918 pandemic H1N1 virus reduces viral pathogenicity and modulates host responses in mice; PA-X also plays an important role in host gene expression shutoff characterized by inhibition of host protein synthesis (Desmet et al., 2013;Jagger et al., 2012). Comprehensive genetic analysis showed that the X ORFs of diverse IAVs can be divided into two groups according to protein length (Shi et al., 2012). Avian, equine and human seasonal H3N2 and H1N1 influenza viruses express a full-length PA-X protein encoded by 252 aa with 61 aa of the X ORF. In contrast, some IAVs, including human pH1N1, canine and certain swine influenza viruses possess a TGG (Trp) to TAG (stop codon) mutation at codon 42 in the X ORF. This mutation confers a truncated PA-X protein encoded by 232 aa with 41 aa of the X domain. The effects of PA-X length on the biological characteristics of IAVs remain unclear.
In the present study, we used pH1N1, a highly pathogenic avian influenza virus of the H5N1 subtype, and a low pathogenicity avian H9N2 influenza virus as prevalent exemplars of truncated human and full-length avian PA-X viruses to examine the contribution of the C-terminal end, from aa 233 to 252, of PA-X in these influenza viruses to viral replication and pathogenicity in mammalian cells and mice. We found that viruses with full-length PA-X replicated more efficiently and were more pathological in A549 cells and mice than the corresponding viruses with truncated PA-X. In contrast, pH1N1 virus had G698A, which is a TGG (W)-to-TAG (stop codon) mutation at codon 42 in the X ORF, producing a truncated PA-X protein with 41 aa in the X domain. Amino acids 233-252 in the C terminus of all PA-X mutant viruses used in the study are shown. pH1N1-61X (RG) expressing full-length PA-X was generated by introducing a A698G mutation which converted the stop codon to W (tryptophan). H5N1-41X (RG) and H9N2-41X (RG) expressing truncated PA-X were generated by introducing a G698A mutation which was a W-to-stop codon mutation at codon 42 in the PA-X ORF of H5N1-61X (WT) and H9N2-61X (WT) viruses. A698G and G698A mutations did not affect the PA ORF.

RESULTS
Generation of pH1N1, H5N1 and H9N2 viruses encoding full-length PA-X and truncated PA-X Sets of influenza viruses based on human truncated PA-X A/Beijing/16/2009 (BJ/09, pH1N1), and avian full-length PA-X A/tree sparrow/Jiangsu/1/2008 (JS08, H5N1) and A/chicken/Hebei/LC/2008 (HB/08, H9N2), viruses were generated using reverse genetics. Avian H5N1 and H9N2 influenza viruses each express a full-length 252 aa PA-X protein with 61 aa in the X domain, whilst pH1N1 virus has a G698A stop codon mutation at codon 42 in the X ORF ( Fig. 1). This stop codon produces a truncated 232 aa PA-X protein with 41 aa in the X domain. To assess the function of the C terminus (aa 233-252) of PA-X, pH1N1-61X (RG) was generated by the introduction of an A698G mutation in the PA-X ORF of pH1N1-41X (WT). Truncated PA-X viruses H5N1-41X (RG) and H9N2-41X (RG) were generated by the introduction of a G698A mutation in the PA-X ORF of H5N1-61X (WT) and H9N2-61X (WT) viruses. A698G and G698A mutations did not affect the PA ORF ( Fig. 1).
Several viral proteins (NA, M1, NS1 and PB1-F2) have been shown to promote apoptosis, associated with cellular and organ damage (Chanturiya et al., 2004;Chen et al., 2001;Morris et al., 1999;Zhirnov et al., 2002a, b  inoculated with each of the six viruses. Clinical signs, mortality and weight loss were monitored over 14 days. Three infected mice were humanely killed at 3, 5 and 7 days p.i., and lungs were collected for virus titration. For pH1N1 mutants, survival curves at 10 5 TCID 50 clearly showed that pH1N1-61X (RG) had a higher pathogenicity index than pH1N1-41X (WT) (Fig. 3a). pH1N1-61X (RG) conferred 100 % mortality by 8 days p.i.; pH1N1-41X (WT) caused no death. The loss of body weight due to pH1N1-61X (RG) was greater than that attributed to pH1N1-41X (WT) pH1N1 (10 5 TCID 50 ) pH1N1 ( Fig. 3. Full-length PA-X viruses were more virulent than the corresponding truncated PA-X viruses in mice. Data show the survival (%) of mice infected with 10 5 TCID 50 pH1N1 (a), 10 2 or 10 3 TCID 50 H5N1 (c, e) and 10 6 TCID 50 H9N2 PA-X mutant viruses (g). The body weight of mice inoculated with pH1N1 (b), H5N1 (d, f) and H9N2 (h) PA-X mutant viruses is presented as percentage of the weight on the day of inoculation (day 0). Any mouse that lost .30 % of its body weight was euthanized. Data represent the mean¡SD of the body weight of three mice per group (n53).
H9N2-61X (WT) H9N2-41X (RG) Fig. 4. Viruses with full-length PA-X caused more severe lung lesions than the corresponding viruses with truncated PA-X. Histopathological changes in mice infected with PA-X mutant viruses. A portion of lung from mice infected with 10 5 TCID 50 pH1N1, 10 3 TCID 50 H5N1 or 10 6 TCID 50 H9N2 mutant viruses at 5 days p.i. was fixed in 10 % phosphate-buffered formalin and processed for paraffin embedding. Each 5 mm section was stained with haematoxylin and eosin, and examined for histopathological changes. Bars, 100 mm.
Viruses with full-length PA-X cause more severe inflammatory responses in mice than those with truncated PA-X  Role of the C terminus of PA-X in influenza viruses pro-inflammatory cytokine and chemokine expression (Bermejo-Martin et al., 2010;Hagau et al., 2010;Lam et al., 2010;Perrone et al., 2008). To assess the effect of the aa 233-252 portion in the C terminus of PA-X on host inflammatory response, protein levels of seven cytokines and chemokines in the lungs of infected mice were determined at 3 and 5 days p.i.
These results showed that infection with pH1N1, H5N1 and H9N2 viruses housing full-length PA-X elicited a more severe inflammatory response than infection with the corresponding truncated PA-X viruses.
Amino acids 233 -252 at the C terminus of PA-X strongly suppress co-transfected gene expression Influenza virus infection can result in host gene expression shutoff, whereby the virus controls cellular transcription/ translation machinery for the preferential production of viral proteins (Katze et al., 1986a, b). Inhibition of host protein synthesis can hinder the cellular antiviral response, contributing to efficient virus production and spread. The PA-X protein is known to play a major role in the suppression of host protein synthesis (Desmet et al., 2013;Jagger et al., 2012). In order to determine the role of the aa 233-252 portion at the C terminus of PA-X in protein synthesis inhibition, the suppression activity of PA plasmids with full-length PA-X was compared with that of truncated PA-X of pH1N1, H5N1 and H9N2 viruses by co-transfection with pEGFP expression plasmid in human embryonic kidney (293T) cells over 24 h. PA and GFP expression was determined by fluorescence intensity and Western blotting (Fig. 7a, b). There was significantly less EGFP expression in the presence of PA plasmids with full-length PA-X than in the presence of the corresponding PA plasmids with truncated PA-X of pH1N1, H5N1 and H9N2 viruses (Pv0.05 or Pv0.01). The levels of PA expression were similar between PA plasmids with fulllength and truncated PA-X in the three virus strains. The differences of EGFP expression did not correlate with the levels of PA expression.
As PA plasmids with full-length PA-X plasmids rather than the corresponding truncated PA-X plasmids were more effective in suppressing co-transfected EGFP gene expression, it was possible that the aa 233-252 portion in the C terminus of PA-X was responsible for the suppression of EGFP expression. To address this, we constructed a plasmid that expressed only the C-terminal 20 aa portion of full-length PA-X. Co-transfection of this construct with pEGFP showed a clear reduction in EGFP production of *50 % (Fig. 8a, b). Therefore, the C-terminal 20 aa of full-length PA-X were important in co-transfected gene suppression.

DISCUSSION
In the present study, influenza viruses with full-length PA-X showed higher replication output and pathogenicity than those with truncated PA-X in human lung cells and mice for pH1N1, H5N1 and H9N2 viruses. Moreover, we found that aa 233-252 at the C terminus of PA-X could strongly suppress co-transfected gene expression, which might be responsible for enhanced viral replication and virulence.
Full-length and truncated forms of PA-X in pH1N1, H5N1 and H9N2 viruses showed no significant difference in virus replication in MDCK cells. Jagger et al. (2012) reported similar findings in MDCK cells, which showed no difference in single-or multi-cycle growth kinetics of the 1918 H1N1 PA-X-deficient and WT PA-X viruses (Jagger et al., 2012). In contrast, we found that pH1N1, H5N1 and H9N2 viruses with full-length PA-X showed higher replication levels in A549 cells than those with truncated PA-X. Challenge experiments showed that viruses with full-length PA-X exhibited greater pathogenicity (mortality and lung pathology) than those with truncated PA-X for pH1N1, H5N1 and H9N2 viruses. In addition, viruses with full-length PA-X showed higher replication levels in murine lungs. These results indicate that increased replication and pathogenicity in mice are features of viruses with full-length PA-X.
Severe influenza virus infection can go through a direct pathogenic pathway by inducing apoptosis, leading to cell death and loss of critical function, or more likely through an indirect pathogenic pathway of inducing excessive  cytokine/chemokine production from the infected cells (Korteweg & Gu, 2008). Jagger et al. (2012) found that loss of PA-X expression led to changes in the kinetics of the global host gene response, which notably showed increases in inflammatory, apoptotic and T-lymphocyte signalling pathways during infection with the 1918 pandemic virus. In contrast, we showed that the C-terminal 20 aa of the full-length PA-X promoted inflammatory response in mice and apoptosis in A549 cells with pH1N1, H5N1 and H9N2 virus infection. Cytokine and chemokine responses, including those of IL-1b, IFN-c, TNF-a, MIP-1a, IL-6, MIP-2, MCP-1, KC and IL-1a, are found to associate with the recruitment of macrophages and neutrophils to infected lungs, which result in acute lung inflammation (Perrone et al., 2008). In the present study, virtually all the chemokine and cytokine levels of full-length PA-X viruses examined were significantly higher than those of the corresponding truncated PA-X viruses. Therefore, full-length PA-X viruses elicited more severe pro-inflammation and pathogenicity than the corresponding truncated PA-X viruses.
Apoptosis plays an important role in the pathogenesis of many infectious diseases, including those caused by viruses (Ludwig et al., 1999;Young et al., 1997). Influenza viruses have been reported to induce apoptosis in numerous cell types, both in vivo and in vitro (Mori et al., 1995;Yang et al., 2011). Several viral proteins have been shown to promote or inhibit apoptosis, such as NS1 and PB1-F2 (Chanturiya et al., 2004;Chen et al., 2001;Zhirnov et al., 2002a). The C-terminal portion of the PB1-F2 ORF contains a mitochondrial targeting sequence (Chen et al., 2001). Expression of full-length PB1-F2 has been associated with mitochondrial targeting and apoptosis, and it has been suggested that mitochondrial disruption with subsequent cell death could contribute to virulence (Chen et al., 2001;Gibbs et al., 2003;Zamarin et al., 2005). It has also been demonstrated that NS1 proteins of different subtypes of influenza viruses promote apoptosis-inducing ability (Brydon et al., 2005;Lam et al., 2011). Here, we found that viruses with full-length PA-X induced increases in apoptosis in A549 cells for pH1N1, H5N1 and H9N2 viruses. The efficient replication of viruses with full-length PA-X in lung cells, and the induction of inflammation and apoptosis, could cause severe lung damage and reduce lung function.
Influenza virus infection can induce host cell shutoff, which is characterized by a rapid decline of host protein synthesis (Katze et al., 1986a, b). Inhibition of cellular gene expression limits the induction of an antiviral response and diverts ribosomes towards translation of viral mRNAs. NS1 can inhibit type I IFN synthesis, block nuclear export of mRNA and inhibit mRNA splicing (Fortes et al., 1994;García-Sastre et al., 1998;Talon et al., 2000). PA has been shown to suppress host protein synthesis. Desmet et al. (2013) showed that the N-terminal domain of PA, which includes the endonuclease active site, is sufficient to suppress protein expression and the suppressive effect of the N-terminal domain of PA was mainly due to PA-X. Their further results showed that the C-terminal X domain of PA-X also played a major role in the suppression of protein expression. In our study, PA plasmids with full-length PA-X were more effective in suppressing co-transfected gene expression than those with truncated PA-X. Moreover, co-transfection of a plasmid expressing only the C-terminal 20 aa of fulllength PA-X with pEGFP showed reduced EGFP expression. Our data suggest that the aa 233-252 portion at the C terminus of PA-X is involved in the suppression of host protein synthesis, and contributes to enhanced viral replication and pathogenicity in full-length PA-X viruses.
Variations in PA-X protein length in IAVs have been observed in nature. Avian influenza viruses are believed to be progenitors of all IAVs and full-length PA-X is conserved in avian influenza viruses (Shi et al., 2012). Adaptation in the form of truncation of PA-X could have occurred when avian influenza viruses were introduced into certain new hosts. For example, PA-X truncation occurred after avian H3N2 viruses were introduced into canines (Shi et al., 2012). Generation of recombinant viruses by reverse genetics. All eight gene segments were amplified by reverse transcription (RT)-PCR from BJ/09, JS/08 and HB/08 viruses, and cloned into the dualpromoter plasmid pHW2000. The mutations were introduced into the PA gene using a Site-directed QuikChange Mutagenesis kit (Agilent) according to the manufacturer's instructions. PCR primer sequences are available upon request. pH1N1 with full-length PA-X (61 aa), pH1N1-61X (RG), had a stop (UAG)-to-tryptophan (UGG) codon substitution at position 42 in the X ORF (stop42W) without changing the PA ORF (Fig. 1b). H5N1 and H9N2 viruses with truncated PA-X (41 aa), H5N1-41X (RG) and H9N2-41X (RG), had a tryptophan (UGG)-to-stop (UAG) codon substitution at position 42 in the X ORF (W42stop) without altering the PA ORF. Rescued viruses were detected using haemagglutination assays. The viruses were purified by sucrose density gradient centrifugation. Viral RNA was extracted and analysed by RT-PCR, and each viral segment was sequenced to confirm sequence identity.
Viral titration and replication kinetics. TCID 50 was determined in MDCK cells using 10-fold serially diluted virus inoculated at 37 uC and cultured for 72 h. The TCID 50 values were calculated using the method first described by Reed & Muench (1938 Mouse infection. Fifteen mice (6-week-old female BALB/c; Vital River Laboratory) per group were anaesthetized with Zoletil (tiletaminezolazepam; Virbac; 20 mg g 21 ) and inoculated intranasally with 50 ml 10 5 TCID 50 pH1N1, 10 2 or 10 3 TCID 50 H5N1 or 10 6 TCID 50 H9N2, diluted in PBS. All mice were monitored daily for 14 days; mice that lost 30 % of their original body weight were humanely euthanized. Three mice were euthanized 3, 5 and 7 days p.i. for determination of lung virus titres, histology and cytokine levels. Lungs, brains and blood were collected and homogenized in cold PBS. Virus titres were determined by TCID 50 . MLD 50 values were determined by intranasally inoculating groups of three mice with 10-fold dilutions of virus and were calculated by the method of Reed & Muench (1938). Animal research was approved by the Beijing Association for Science and Technology, and complied with the Beijing laboratory animal welfare and ethical guidelines of the Beijing Administration Committee of Laboratory Animals.
Histopathology. At 5 days p.i., a portion of lung from each euthanized mouse was fixed in 10 % phosphate-buffered formalin and processed for paraffin embedding. Each 5 mm section was stained with haematoxylin and eosin and examined for histopathological changes. Images were captured with a Zeiss Axioplan 2IE epifluorescence microscope.
Quantification of cytokine and chemokine protein levels in mouse lungs. Levels of cytokines and chemokines, including IL-1b, IL-6, KC, MCP-1, MIP-1a, TNF-a, and IFN-c, in the lung were determined using a cytometric bead array method (BD Cytometric Bead Array Mouse Inflammation kit; BD Bioscience). Briefly, 50 ml mouse inflammation capture bead suspension and 50 ml detection reagent were added to an equal amount of sample and incubated in the dark for 2 h at room temperature. Samples were later washed with 1 ml wash buffer and then centrifuged at 200 g at room temperature for 5 min. Supernatants were discarded and 300 ml wash buffer was added. Samples were analysed on a BD FACSArray bioanalyser (BD Bioscience). Data were analysed using BD Cytometric Bead Array software (BD Bioscience). Chemokine and cytokine levels were recorded as pg ml 21 in the homogenate.
Cell death assays. Virus infection assays were conducted in six-well plates. Cells were seeded at a density of 1|10 6 cells per well overnight in infection media (cell growth media with 1 % BSA was used in place of FCS). Cells were then infected with virus at m.o.i. 1.0 for 12 h. Cells from the supernatant and monolayers were then harvested, washed and stained with allophycocyanin-labelled annexin and PI (Becton Dickinson) for 20 min. After a final wash step, cells were resuspended in 100 ml FACS wash buffer (PBS containing 3 % BSA and 0.01 % sodium azide), and analysed using a FACSCalibur (BD Biosciences) and Flow Jo software (version 7.6.1). Cell death (apoptosis and necrosis) was defined as annexin-V + and PI + , whilst apoptotic cells were annexin-V + only. Viable cells were considered neither annexin-V + nor PI + .
Western blotting. Total cell protein lysates were extracted from transfected 293T cells and infected MDCK cells with CA630 lysis buffer (150 mM NaCl, 1 % CA630 detergent, 50 mM Tris base, pH 8.0). Cellular proteins were separated by 12 % SDS-PAGE and transferred to a PVDF membrane (Amersham Biosciences). Each PVDF membrane was blocked with 0.1 % Tween 20 and 5 % non-fat dry milk in Tris-buffered saline and subsequently incubated with a primary antibody. Primary antibodies were specific to IAV PA (1 : 3000; GeneTex) and GFP (1 : 1000; Abcam). Secondary antibodies included HRP-conjugated anti-mouse antibody or HRP-conjugated anti-rabbit antibody (1 : 10 000; Jackson ImmunoResearch), as appropriate. HRP presence was detected using a Western Lightning Chemiluminescence kit (Amersham Pharmacia) in accordance with the manufacturer's protocol. Protein bands were quantified by densitometry using Image Studio Digits (LI-COR Biosciences).
Statistical analysis. The statistical analyses were performed using Prism version 5.00 (GraphPad). Comparison between two forms of treatment involved the two-tailed Student's t-test; multiple comparisons were carried out using two-way ANOVA considering time and virus as factors. Differences were considered statistically significant at Pv0.05.