Mutation of influenza A virus PA-X decreases pathogenicity in chicken embryos and can increase the yield of reassortant candidate vaccine viruses

Abstract

The PA-X protein of influenza A virus has roles in host cell shut-off and viral pathogenesis. While most strains are predicted to encode PA-X, strain-dependent variations in activity have been noted. We found that PA-X protein from A/PR/8/34 (PR8) strain had significantly lower repressive activity against cellular gene expression compared with PA-Xs from the avian strains A/turkey/England/50-92/91 (H5N1) (T/E) and A/chicken/Rostock/34 (H7N1). Loss of normal PA-X expression, either by mutation of the frameshift site or by truncating the X-ORF, had little effect on the infectious virus titre of PR8 or PR8 7:1 reassortants with T/E segment 3 grown in embryonated hens’ eggs. However, in both virus backgrounds, mutation of PA-X led to decreased embryo mortality and lower overall pathology; effects that were more pronounced in the PR8 strain than the T/E reassortant, despite the low shut-off activity of the PR8 PA-X. Purified PA-X mutant virus particles displayed an increased ratio of HA to NP and M1 compared to their WT counterparts, suggesting altered virion composition. When the PA-X gene was mutated in the background of poorly growing PR8 6:2 vaccine reassortant analogues containing the HA and NA segments from H1N1 2009 pandemic viruses or an avian H7N3 strain, HA yield increased up to 2-fold. This suggests that the PR8 PA-X protein may harbour a function unrelated to host cell shut-off and that disruption of the PA-X gene has the potential to improve the HA yield of vaccine viruses.

IMPORTANCE
Influenza A virus is a widespread pathogen that affects both man and a variety of animal species, causing regular epidemics and sporadic pandemics with major public health and economic consequences. A better understanding of virus biology is therefore important. The primary control measure is vaccination, which for humans, mostly relies on antigens produced in eggs from PR8-based viruses bearing the glycoprotein genes of interest. However, not all reassortants replicate well enough to supply sufficient virus antigen for demand. The significance of our research lies in identifying that mutation of the PA-X gene in the PR8 strain of virus can improve antigen yield, potentially by decreasing the pathogenicity of the virus in embryonated eggs.

Full text

1
Mutation of influenza A virus PA-X decreases pathogenicity in chicken embryos and can 1
increase the yield of reassortant candidate vaccine viruses 2
3
Saira Hussaina*, Matthew L. Turnbulla†, Helen M. Wisea‡, Brett W. Jaggerb,c§, Philippa M. 4
Bearda,d, Kristina Kovacikovaa

, Jeffery K. Taubenbergerc, Lonneke Verveldea, Othmar G 5
Engelhardte & Paul Digarda* 6
7
aThe Roslin Institute & Royal (Dick) School of Veterinary Studies, University of Edinburgh, 8
Edinburgh, UK 9
bDepartment of Pathology, University of Cambridge, Cambridge, UK 10
cNational Institutes of Health, Maryland, USA 11
dThe Pirbright Institute, Pirbright, Surrey, UK 12
eNational Institute for Biological Standards and Control, South Mimms, Hertfordshire, UK 13
14
Running Head: Influenza A virus PA-X and pathogenicity in hens’ eggs 15
16
#Address correspondence to Paul Digard, paul.digard@roslin.ed.ac.uk 17
Present addresses: *Saira Hussain, The Francis Crick Institute, London, United Kingdom; 18
†Matthew L Turnbull, Glasgow Centre for Virus Research, Glasgow, United Kingdom; 19
‡Helen M. Wise, Herriot-Watt University, Edinburgh, United Kingdom; §Brett W. Jagger, 20
Department of Medicine, Washington University in St. Louis, St. Louis, USA;

Kristina 21
Kovacikova, Leiden University Medical Centre, Netherlands. 22
23
Abstract word count: 242 24 Main text word count: 5308 25
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

2
Abstract 26
27
The PA-X protein of influenza A virus has roles in host cell shut-off and viral pathogenesis. 28
While most strains are predicted to encode PA-X, strain-dependent variations in activity have 29
been noted. We found that PA-X protein from A/PR/8/34 (PR8) strain had significantly lower 30
repressive activity against cellular gene expression compared with PA-Xs from the avian 31
strains A/turkey/England/50-92/91 (H5N1) (T/E) and A/chicken/Rostock/34 (H7N1). Loss of 32
normal PA-X expression, either by mutation of the frameshift site or by truncating the X-33
ORF, had little effect on the infectious virus titre of PR8 or PR8 7:1 reassortants with T/E 34
segment 3 grown in embryonated hens’ eggs. However, in both virus backgrounds, mutation 35
of PA-X led to decreased embryo mortality and lower overall pathology; effects that were 36
more pronounced in the PR8 strain than the T/E reassortant, despite the low shut-off activity 37
of the PR8 PA-X. Purified PA-X mutant virus particles displayed an increased ratio of HA to 38
NP and M1 compared to their WT counterparts, suggesting altered virion composition. When 39
the PA-X gene was mutated in the background of poorly growing PR8 6:2 vaccine 40
reassortant analogues containing the HA and NA segments from H1N1 2009 pandemic 41
viruses or an avian H7N3 strain, HA yield increased up to 2-fold. This suggests that the PR8 42
PA-X protein may harbour a function unrelated to host cell shut-off and that disruption of the 43
PA-X gene has the potential to improve the HA yield of vaccine viruses. 44
45
IMPORTANCE Influenza A virus is a widespread pathogen that affects both man and a 46
variety of animal species, causing regular epidemics and sporadic pandemics with major 47
public health and economic consequences. A better understanding of virus biology is 48
therefore important. The primary control measure is vaccination, which for humans, mostly 49
relies on antigens produced in eggs from PR8-based viruses bearing the glycoprotein genes of 50
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

3
interest. However, not all reassortants replicate well enough to supply sufficient virus antigen 51
for demand. The significance of our research lies in identifying that mutation of the PA-X 52
gene in the PR8 strain of virus can improve antigen yield, potentially by decreasing the 53
pathogenicity of the virus in embryonated eggs. 54
55
Introduction 56
57
Influenza epidemics occur most years as the viruses undergo antigenic drift. Influenza 58
A viruses (IAV) and influenza B viruses cause seasonal human influenza but IAV poses an 59
additional risk of zoonotic infection, with the potential of a host switch and the generation of 60
pandemic influenza. The 1918 ‘Spanish flu’ pandemic was by far the worst, resulting in 40-61
100 million deaths worldwide (1), while the 2009 swine flu pandemic caused an estimated 62
200,000 deaths worldwide (2). 63
IAV contains eight genomic segments encoding for at least ten proteins. Six genomic 64
segments (segments 1, 2, 3, 5, 7 and 8) encode the eight core “internal” proteins PB2, PB1, 65
PA, NP, M1, NS1 and NS2, as well as the ion channel M2. These segments can also encode a 66
variety of accessory proteins known to influence pathogenesis and virulence (reviewed in (3, 67
4)). Segments 4 and 6 encode for the two surface glycoproteins haemagglutinin (HA) and 68
neuraminidase (NA) respectively (5, 6) and virus strains are divided into subtypes according 69
to the antigenicity of these proteins. 70
Vaccination is the primary public health measure to reduce the impact of influenza 71
epidemics and pandemics, principally using inactivated viruses chosen to antigenically match 72
the currently circulating virus strains or newly emerging viruses of pandemic concern. 73
However, before efficient vaccine production can commence, high-yielding candidate 74
vaccine viruses (CVVs) need to be prepared. Seasonal CVVs are widely produced by 75
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

4
classical reassortment. This process involves co-infecting embryonated hens’ eggs with the 76
vaccine virus along with a high yielding “donor” virus adapted to growth in eggs (most 77
commonly the A/Puerto Rico/8/34 strain, or “PR8”). The highest yielding viruses that contain 78
the glycoproteins of the vaccine virus are then selected. Recombinant influenza viruses are 79
also made by reverse genetics (RG) (7-9), which relies on the transfection of cells with 80
plasmids engineered to express both viral genomic RNA and proteins from each of the eight 81
segments and hence initiate virus production; the resultant virus is subsequently amplified in 82
eggs. When making RG CVVs, typically the six segments encoding core proteins (backbone) 83
are derived from the donor strain whereas the two segments encoding the antigens are 84
derived from the vaccine virus. Classical reassortment has the advantage that it allows for the 85
fittest natural variant to be selected but it can be time consuming. In the case of a pandemic, 86
large quantities of vaccine must be made available quickly. Moreover, RG is the only viable 87
method for production of CVVs for potentially pandemic highly pathogenic avian influenza 88
viruses, since it allows for removal of genetic determinants of high pathogenicity in the virus 89
genome, as vaccines are manufactured in biosafety level 2 laboratories. A limited number of 90
donor strains for IAV vaccine manufacture currently exist. Although PR8 is widely used, 91
reassortant viruses based on it do not always grow sufficiently well for efficient vaccine 92
manufacture. In the case of the 2009 H1N1 pandemic (pdm09), vaccine viruses grew poorly 93
in eggs compared with those for previous seasonal H1N1 isolates (10), resulting in 94
manufacturers struggling to meet demand. Thus, there is a clear need for new reagents and 95
methods for IAV production, particularly for pandemic response. 96
In recent years, several approaches have been employed to improve antigen yield of 97
candidate vaccine viruses made by reverse genetics. These have involved empirical testing 98
and selection of PR8 variants (11, 12), as well as targeted approaches such as making 99
chimeric genes containing promoter and packaging signal regions of PR8 while encoding the 100
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

5
ectodomain of the CVV glycoprotein genes (13-21), or introducing a wild-type (WT) virus-101
derived segment 2 (21-29). Our approach was to manipulate expression of an accessory 102
protein virulence factor, PA-X (30). Segment 3, encoding PA as the primary gene product, 103
also expresses PA-X by low-level ribosomal shifting into a +1 open reading frame (ORF) 104
termed the X ORF (Fig 1A) (30). PA-X is a 29 kDa protein that contains the N-terminal 105
endonuclease domain of PA, and in most isolates, a 61 amino acid C-terminus from the X 106
ORF (30-32). It has roles in shutting off host cell protein synthesis and, at the whole animal 107
level, modulating the immune response (30, 33). Loss of PA-X expression has been shown to 108
be associated with increased virulence in mice for 1918 H1N1, H5N1 and also pdm09 and 109
classical swine influenza H1N1 strains, as well as in chickens and ducks infected with a 110
highly pathogenic H5N1 virus (30, 34-40). However, in other circumstances, such as avian 111
H9N2 viruses (40) or, in some cases, A(H1N1)pdm09 viruses (37, 41), mutation of PA-X 112
resulted in reduced pathogenicity in mice. Similarly, a swine influenza H1N2 virus (42) 113
lacking PA-X showed reduced pathogenicity in pigs. Moreover, PA-X activity in repressing 114
cellular gene expression is strain dependent (33, 34, 40, 43), with laboratory-adapted viruses 115
such as A/WSN/33 showing lower levels of activity (33). Here, we show that although the 116
PR8 PA-X polypeptide has low shut-off activity, removing its expression decreases the 117
pathogenicity of the virus in the chick embryo model. Moreover, we found that, for certain 118
poor growing CVV mimics, ablating PA-X expression improved HA yield from embryonated 119
eggs up to 2-fold. In no case did loss of PA-X appear to be detrimental to the growth of 120
CVVs, making it a potential candidate mutation for incorporation into the PR8 CVV donor 121
backbone. 122
123
Results 124
125
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

6
The PR8 virus strain PA-X has relatively low shut-off activity. 126
Previous work has noted variation in apparent activity of PA-X proteins from 127
different strains of virus, with the laboratory adapted-strain WSN showing lower activity than 128
many other strains (33). Re-examination of evidence concerning a postulated proteolytic 129
activity of PA (43) suggested that lower PA-X activity might also be a feature of the PR8 130
strain. To test this, the ability of PR8 segment 3 gene products to inhibit cellular gene 131
expression was compared to that of two avian virus-derived PA segments (from 132
A/chicken/Rostock/34 [H7N1; FPV] and A/turkey/England/50-92/91 [H5N1; T/E]. Avian 133
QT-35 (Japanese quail fibrosarcoma) cells were transfected with a consistent amount of a 134
plasmid encoding luciferase under the control of a constitutive RNA polymerase II promoter 135
(pRL) and increasing amounts of the IAV cDNAs (in pHW2000-based RG plasmids), or as a 136
negative control, the maximum amount of the empty pHW2000 vector. Luciferase expression 137
was measured 48 h later and expressed as a % of the amount obtained from pRL-only 138
transfections. Transfection of a 4-fold excess of empty pHW2000 vector over the luciferase 139
reporter plasmid had no significant effect on luciferase expression, whereas co-transfection of 140
the same amount of either the FPV or T/E segments suppressed activity to around 10% of the 141
control (Fig 1B). Titration of the FPV and T/E plasmids gave a clear dose-response 142
relationship, giving estimated EC50 values of 24 ± 1.1 ng and 32 ± 1.1 ng respectively. In 143
contrast, the maximum amount of the PR8 plasmid only inhibited luciferase expression by 144
around 30% and an EC50 value could not be calculated, indicating a lower ability to repress 145
cellular gene expression. Similarly low inhibitory activity of the PR8 segment 3 was seen in a 146
variety of other mammalian cell lines (data not shown), suggesting it was an intrinsic feature 147
of the viral gene, rather than a host- or cell type-specific outcome. 148
Several studies have shown the X-ORF to be important in host cell shut-off function 149
and virulence of PA-X (37, 44-46). To further explore the influence of X-ORF sequences on 150
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

7
virus strain-specific host cell shut-off, mutations were constructed in segment 3 in which PA-151
X expression was either hindered (via mutation of the frameshift site [FS]) or altered by the 152
insertion of premature termination codons (PTCs 1-4; silent in the PA ORF) such that C-153
terminally truncated forms of PA-X would be expressed (Figure 1A). QT-35 cells were co-154
transfected with the pRL plasmid and a fixed amount of WT, FS or PTC plasmids and 155
luciferase expression measured 48 h later. As before, the WT FPV and T/E PA-Xs reduced 156
luciferase activity by approximately 5-10 fold, while WT PR8 PA-X had no significant effect 157
(Figure 1C). Introducing the FS mutation into both PR8 and T/E segment 3 significantly 158
increased luciferase activity relative to the WT construct. Truncation of the PR8 PA-X to 225 159
AA or less (PTC mutations 1-3) significantly improved shut-off activity, although not to the 160
levels seen with the WT avian virus polypeptides, while the PTC4 truncation had no effect. In 161
contrast, none of the PTC mutations significantly affected activity of the T/E PA-X, although 162
there was a trend towards increased activity from the PTC2, 3 and 4 truncations 163
Low activity could be due to decreased expression and/or decreased activity of PA-X. 164
To examine this, expression of the low activity PR8 and high activity FPV PA-X constructs 165
were compared by in vitro translation reactions in rabbit reticulocyte lysate. Translation of 166
segment 3 from both PR8 and FPV produced both full length PA and similar quantities of a 167
minor polypeptide species of the expected size for PA-X whose abundance decreased after 168
addition of the FS mutation or whose electrophoretic mobility was altered in stepwise fashion 169
after C-terminal truncation with the PTC1-4 mutations (Figure 1D). This suggested that 170
differences in shut-off potential were not linked to intrinsic differences in PA-X protein 171
synthesis. To confirm the identity of the PR8 in vitro translated polypeptides, 172
immunoprecipitation of IVT products with sera raised either against the N-terminal domain 173
of PA, or an X-ORF derived polypeptide or pre-immune sera (30) were performed (Figure 174
1E). WT PA-X was clearly visible in samples immunoprecipitated with anti-PA-X and anti-175
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

8
PA-N but not the pre-immune serum, where it co-migrated with the product from the 176
delC598 plasmid, a construct in which cytosine 598 of segment 3 (the nucleotide skipped 177
during the PA-X frameshifting event (47)) had been deleted to put the X-ORF into the same 178
reading frame as the N-terminal PA domain (Figure 1E, lanes 2 and 7). In contrast, only 179
background amounts of protein were precipitated from the FS IVT (lane 3). Faster migrating 180
polypeptide products from the PTC3 and 4 plasmids showed similar reactivities to WT PA-X 181
(lanes 5 and 6) whereas the product of the PTC1 plasmid was only precipitated by anti-PA-N 182
(lane 4), as expected because of the loss of the epitope used to raise the PA-X antiserum 183
(Figure 1A). Overall therefore, the PR8 PA-X polypeptide possessed lower shut-off activity 184
than two avian virus PA-X polypeptides despite comparable expression in vitro, and its 185
activity could be modulated by mutation of the X-ORF. 186
187
Loss of PA-X expression results in significantly less pathogenicity in chick embryos 188
without affecting virus replication 189
190
In order to further characterise the role of PA-X as a virulence determinant, we tested 191
the panel of high and low activity mutants in the chicken embryo pathogenicity model. 192
Embryonated hens’ eggs were infected with PR8-based viruses containing either PR8 or T/E 193
WT or mutant segment 3s and embryo viability was monitored at 2 days post infection (p.i.) 194
by candling. Both WT PR8 and the WT 7:1 reassortant with the T/E segment 3 viruses had 195
killed over 50% of the embryos by this point (Figures 2A and B). Truncation of PA-X by the 196
PTC mutations led to small improvements in embryo survival, although none of the 197
differences were statistically significant. However, embryo lethality was significantly 198
reduced to below 20% following infection with the PR8 FS virus compared with PR8 WT 199
virus. A similar reduction in lethality was seen for the T/E FS virus, although the difference 200
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

9
was not statistically significant. This reduction in embryo pathogenicity following ablation of 201
PA-X expression suggested potential utility as a targeted mutation in the PR8 backbone used 202
to make CVVs. Accordingly, to characterise the effects of mutating PR8 PA-X over the 203
period used for vaccine manufacture, embryo survival was monitored daily for 72 h. Eggs 204
infected with WT PR8 showed 45% embryo survival at 2 days p.i. and all were dead by day 3 205
(Figure 2C). However, the PR8 FS infected eggs showed a statistically significant 206
improvement in survival compared to WT, with 80% and 30% survival at days 2 and 3, 207
respectively. Embryos infected with PR8 expressing the C-terminally truncated PTC1 form 208
of PA-X showed an intermediate survival phenotype with 60% and 20% survival at days 2 209
and 3, respectively. 210
To further assess the effects of mutating PA-X, the chicken embryos were examined 211
for gross pathology. WT PR8 infection resulted in smaller, more fragile embryos with diffuse 212
reddening, interpreted as haemorrhages (Figure 2D). In comparison, the PA-X null FS 213
mutant-infected embryos remained intact, were visibly larger and less red. To quantitate these 214
observations, embryos were scored blind for gross pathology. Taking uninfected embryos as 215
a baseline, it was clear that WT PR8 virus as well as the PA-X truncation mutants induced 216
severe changes to the embryos (Figure 2E). In contrast, the PA-X null FS mutant caused 217
significantly less pathology. The WT 7:1 T/E reassortant virus gave less overt pathology than 218
WT PR8 but again, reducing PA-X expression through the FS mutation further reduced 219
damage to the embryos (Figure 2E). Similar trends in pathology were also seen with 7:1 PR8 220
reassortant viruses containing either WT or FS mutant versions of FPV segment 3 (data not 221
shown). 222
Examination of haematoxylin and eosin (H&E) stained sections through the embryos 223
revealed pathology in numerous organs including the brain, liver and kidney (Figure 3). In 224
the brain of embryos infected with WT virus there was marked rarefaction of the neuropil, 225
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

10
few neurons were identifiable, and there was accumulation of red blood cells (Figure 3C). In 226
the liver of embryos infected with WT virus the hepatic cords were disorganised, and the 227
hepatocytes were often separated by large pools of red blood cells (Figure 3D). In the kidney 228
of embryos infected with WT virus, tubules were often lined by degenerate epithelial cells 229
(characterised by loss of cellular detail). In all cases the pathology noted in WT virus-infected 230
embryos was also present in the FS virus-infected embryos but at a reduced severity. Thus 231
overall, disruption of PA-X expression in PR8 resulted in significantly less pathogenicity in 232
chick embryos. 233
Reduced pathogenicity in vivo following loss of PA-X expression could be due to a 234
replication deficiency of the virus, although the viruses replicated equivalently in mammalian 235
MDCK cells (data not shown). To test if replication did differ in ovo, infectious virus titres 236
were obtained (by plaque titration on MDCK cells) from the allantoic fluid of embryonated 237
hens’ eggs infected with the panels of PR8 and T/E viruses at 2 days p.i.. However, there 238
were no significant differences in titres between either PR8 or T/E WT and PA-X mutant 239
viruses (Figures 4A, B). Since the reduced pathogenicity phenotype in ovo on loss of PA-X 240
expression was more pronounced for viruses with PR8 segment 3 than the T/E gene, embryos 241
from PR8 WT and segment 3 mutant-infected eggs were harvested at 2 days p.i., washed, 242
macerated and virus titres from the homogenates determined. Titres from embryos infected 243
with the PR8 FS and PTC4 viruses were slightly (less than 2-fold) reduced compared to 244
embryos infected with PR8 WT virus (Figure 4C), but overall there were no significant 245
differences in titres between the viruses. To see if there were differences in virus localisation 246
in tissues between PR8 WT and FS viruses, immunohistochemistry was performed on chick 247
embryo sections to detect viral NP as a marker of infected cells. NP positive cells were seen 248
in blood vessels throughout the head and body of both PR8 WT and FS-infected embryos; 249
liver, heart and kidney are shown as representatives (Figure 4D), indicating that the 250
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

11
circulatory system had been infected. However, there were no clear differences in virus 251
localisation between embryos infected with WT and FS viruses. 252
Overall therefore, the loss of PA-X expression reduced IAV pathogenicity in chick 253
embryos, as assessed by mortality curves and both gross and histopathological examination 254
of embryo bodies. This reduced pathogenicity did not appear to correlate with reduced 255
replication or altered distribution of the virus in ovo. 256
257
Ablating PA-X expression alters virion composition 258
Other viruses encode host-control proteins with mRNA endonuclease activity, 259
including the SOX protein of murine gammaherpesvirus MHV68 whose expression has been 260
shown to also modulate virion composition (48). Also, egg-grown IAV titre and HA yield do 261
not always exactly match, with certain problematic candidate vaccine viruses (CVVs) 262
containing lower amounts of HA per virion (16, 49, 50). Accordingly, we compared the 263
relative quantities of virion structural proteins between PA-X expressing and PA-X null 264
viruses. Two pairs of viruses were tested: either an entirely PR8-based virus, or a 7:1 265
reassortant of PR8 with FPV segment 3, both with or without the FS mutation. Viruses were 266
grown in eggs as before and purified from allantoic fluid by density gradient 267
ultracentrifugation before polypeptides were separated by SDS-PAGE and visualised by 268
staining with Coomassie blue. To ensure that overall differences in protein loading did not 269
bias the results, 3-fold dilutions of the samples were analysed. From the gels, the major virion 270
components of both WT and FS virus preparations could be distinguished: NP, the two 271
cleaved forms of haemagglutinin, HA1 and HA2, the matrix protein, M1 and in lower 272
abundance, the polymerase proteins (Figures 5A, B, lanes 4-9). In contrast, only trace 273
polypeptides were present in similarly purified samples from uninfected allantoic fluid (lanes 274
1-3). Densitometry was used to assess the relative viral protein contents of the viruses. The 275
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

12
two most heavily loaded lanes (where band intensities were sufficient for accurate 276
measurement) were quantified and average HA1:NP and HA2:M1 ratios calculated. When 277
the data from three independent experiments were examined in aggregate by scatter plot 278
(Figures 5C and D), a statistically significant increase in the average quantity of HA1 relative 279
to NP was evident for both PR8 and the FPV reassortant FS viruses of ~1.4-fold and ~1.6-280
fold respectively compared to WT (Figures 5 C, D). The ratio of HA2:M1 was also 281
significantly increased in the PR8 FS virus (~1.2- fold greater for WT) and a similar but non-282
significant increase was seen for the FPV virus pair. These data are consistent with the 283
hypothesis that PA-X expression modulates virion composition. 284
285
Ablating PA-X expression increases HA yield of CVVs bearing pdm2009 286
glycoproteins 287
The reduced pathogenicity and corresponding longer embryo survival time induced 288
by the PR8 FS mutant in ovo coupled with evident modulation of virion composition in 289
favour of HA content suggested a strategy to increase overall antigen yields for PR8-based 290
CVVs. Therefore, the effect of incorporating the PA-X FS mutation into CVV mimics 291
containing glycoproteins of different IAV subtypes was examined. Reasoning that a benefit 292
might be most apparent for a poor-yielding strain, 6:2 CVV mimics containing the 293
glycoprotein genes from the A(H1N1)pdm09 vaccine strain, A/California/07/2009 (Cal7) 294
with the six internal genes from PR8, with or without the FS mutation in segment 3, were 295
generated. Growth of these viruses in embryonated hens’ eggs was then assessed by 296
inoculating eggs with either 100, 1,000 or 10,000 PFU per egg (modelling the empirical 297
approach used in vaccine manufacture to find the optimal inoculation dose) and measuring 298
HA titre at 3 days p.i.. Both viruses grew best at an inoculation dose of 100 PFU/egg, but 299
final yield was both relatively low (as expected, ~ 64 HAU/50 μl) and insensitive to input 300
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

13
dose, with average titres varying less than 2-fold across the 100-fold range of inocula (Figure 301
6A). However, at each dose, the 6:2 FS virus gave a higher titre (on average, 1.6-fold) than 302
the parental 6:2 reassortant. In order to assess HA yield between the WT and FS viruses on a 303
larger scale, comparable to that used by WHO Essential Regulatory Laboratories (ERLs) 304
such as the National Institute for Biological Standards and Control, UK, 20 eggs per virus 305
were infected at a single inoculation dose. In this experiment, the average HA titre of the FS 306
virus was over 3 times higher than the WT 6:2 virus (Figure 6B). To further determine the 307
consistency of these results, HA titre yields were assessed from two independently rescued 308
reverse genetics stocks of the Cal7 6:2 CVV mimics with or without the PR8 PA-X gene as 309
well as another 6:2 CVV mimic bearing the glycoproteins from the A/England/195/2009 310
(Eng195) A(H1N1)pdm09 strain. HA yield from different stocks was assessed in independent 311
repeats of both small- (5 eggs for each of three different inoculation doses, taking data from 312
the dose that gave maximum yield) and large-scale (20 eggs per single dose of virus) 313
experiments. Examination of the average HA titres showed considerable variation between 314
independent experiments (Figure 6C). However, when plotted as paired data points, it was 315
obvious that in every experiment, the FS viruses gave a higher yield than the parental 6:2 316
reassortant and on average, there were 2.7- and 3.8-fold higher HA titres with the segment 3 317
FS mutation for Cal7 and Eng195 respectively (Table 1). 318
To directly assess HA protein yield, viruses were partially purified by 319
ultracentrifugation of pooled allantoic fluid through 30% sucrose cushions. Protein content 320
was analysed by SDS-PAGE and Coomassie staining, either before or after treatment with N-321
glycosidase F (PNGaseF) to remove glycosylation from HA and NA. Both virus preparations 322
gave polypeptide profiles that were clearly different from uninfected allantoic fluid processed 323
in parallel, with obvious NP and M1 staining, as well other polypeptide species of less certain 324
origin (Figure 6D). Overall protein recovery was higher in the FS virus than the WT 325
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

14
reassortant virus (compare lanes 3 and 4 with 5 and 6), but the poor yields of these viruses 326
made unambiguous identification of the HA polypeptide difficult. However, PNGaseF 327
treatment led to the appearance of a defined protein band migrating at around 40 kDa that 328
probably represented de-glycosylated HA1, and this was present in appreciably higher 329
quantities in the 6:2 FS preparation (compare lanes 4 and 6). Therefore, equivalent amounts 330
of glycosylated or de-glycosylated samples from the Cal7 WT and FS reassortants were 331
analysed by SDS-PAGE and western blotting using anti-pdm09 HA sera. The western blot 332
gave a clear readout for HA1 content, confirmed the mobility shift upon de-glycosylation and 333
showed increased amounts of HA1 in the 6:2 FS samples (Figure 6D lower panel). 334
Quantitative measurements of the de-glycosylated samples showed that the 6:2 FS virus gave 335
1.9-fold greater HA1 yield than the WT reassortant. To test the reproducibility of this finding, 336
HA1 yield was assessed by densitometry of de-glycosylated HA1 following SDS-PAGE and 337
western blot for partially purified virus from 9 independent experiments with the Cal7 and 338
Eng195 reassortants. When examined as paired observations, it was evident that in 8 of the 9 339
experiments, the FS viruses gave greater HA yields than the parental virus, with only one 340
experiment producing a lower amount (Figure 6E). In one large-scale experiment, the HA1 341
yield of 6:2 FS was approximately 20-fold higher compared to its 6:2 counterpart. However, 342
in all other experiments, the 6:2 FS virus gave between 1.5 and 3-fold increases in HA1 yield 343
when compared with the 6:2 virus. When the outlier was discounted (as possibly resulting 344
from an artefactually low recovery for the WT sample), average HA1 yield from the other 8 345
experiments showed 1.9- and 2.4-fold improvements with the segment 3 FS mutation for 346
Cal7 and Eng195 respectively (Table 1). 347
The HA yield of CVVs with pdm09 glycoproteins has been shown to be improved by 348
engineering chimeric HA genes which contain signal peptide and transmembrane 349
domain/cytoplasmic tail sequences from PR8 HA and the antigenic region of the HA gene 350
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

15
from Cal7 (19, 20). To test if these gains were additive with those seen with the FS mutation, 351
we introduced the NIBRG-119 construct, which is a segment 4 with the ectodomain coding 352
region of Cal7 HA and all other sequences (3
′
- and 5
′
-noncoding regions, signal peptide, 353
transmembrane domain, and cytoplasmic tail) from PR8 (19) into 6:2 CVV mimics with the 354
WT A(H1N1)pdm09 NA gene and a PR8 backbone with or without the PA-X mutation. 355
Viruses bearing the NIBRG-119 HA did not agglutinate chicken red blood cells (data not 356
shown) so HA yield from eggs was assessed by SDS-PAGE and western blot of partially 357
purified virus. Chimeric HA viruses containing the FS backbone showed an average HA 358
yield improvement of 1.54-fold compared to the WT backbone counterpart, across 359
independent small- and large-scale experiments (Table 1). Thus, the FS mutation is 360
compatible with other rational strategies for increasing egg-grown reverse genetics vaccines. 361
Following on from this, several pairs of CVV mimics were made with glycoproteins 362
from different IAV strains with either WT or FS mutant PR8 segment 3. These included 363
viruses with glycoproteins of potentially pandemic strains such as highly pathogenic avian 364
virus A/turkey/Turkey/1/2005 (H5N1), as well as low pathogenic avian strains 365
A/mallard/Netherlands/12/2000 (H7N3), A/chicken/Pakistan/UDL-01/2008 (H9N2) and 366
A/mallard/Netherlands/10/99 (H1N1), as well as the human H3N2 strain, A/Hong Kong/1/68, 367
and an early seasonal H3N2 isolate, A/Udorn/307/72 (Table 1). HA yield in eggs was 368
assessed from both the small-scale and large-scale experimental conditions described earlier, 369
by measuring HA titre and HA1 yield from partially purified virus particles. In general, the 370
two techniques were in agreement (Table 1). Ablating PA-X expression moderately improved 371
HA1 yields of some of the CVVs tested: 1.5-fold for the avian H7N3 strain, 372
A/mallard/Netherlands/12/2000 and 1.3-fold for the human H3N2 A/Udorn/307/72 strain. 373
Other CVVs showed lesser or effectively no increases. However, in no case, did ablation of 374
PA-X appear to be detrimental to the growth of CVVs. 375
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

16
376
Discussion 377
Here we show that ablating expression of PA-X resulted in reduced pathogenicity in 378
the chicken embryo model despite the PR8 PA-X protein having relatively low host cell shut-379
off activity compared to PA-X from other IAV strains. Although loss of PA-X expression had 380
no effect on infectious titres in eggs, subtle differences in virion composition were observed, 381
and more importantly, the HA yield from poor growing 6:2 reassortant vaccine analogues 382
containing the HA and NA segments from A(H1N1) pdm09 strains was significantly 383
improved. 384
The majority of studies examining the effect of loss of PA-X expression on IAV 385
pathogenicity have used mice as the experimental system. As discussed above, in most cases, 386
the outcome has been increased virulence (30, 34-40), but several studies have found the 387
opposite effect, with PA-X deficiency reducing pathogenicity in mice (37, 41, 42). In adult 388
bird challenge systems using chickens and ducks infected with a highly pathogenic H5N1 389
virus, abrogating PA-X expression caused increased virulence (35). In our infection model of 390
embryonated hens’ eggs, loss of PA-X expression markedly reduced the pathogenicity in 391
chick embryos. Thus like PB1-F2, another trans-frame encoded IAV accessory protein (51), 392
the impact of PA-X expression on viral pathogenicity seems to vary according to both host 393
and virus strain, but not in a fashion that can simply be correlated with mammalian or avian 394
settings. 395
In previous studies, changes in virulence phenotypes following loss of PA-X 396
expression have been associated with its host cell shut-off function. In the virus strains used, 397
whether from high pathogenicity or low pathogenicity IAV strains, the PA-X polypeptides 398
were shown to significantly affect host cell gene expression. Here, despite PR8 PA-X failing 399
to repress cellular gene expression, a strong phenotypic effect was seen in chicken embryos 400
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

17
following loss of PA-X expression. Furthermore, these effects on pathogenicity were more 401
pronounced in an otherwise WT PR8 virus than in a 7:1 reassortant with segment 3 from the 402
highly pathogenic H5N1 avian influenza T/E strain which encodes a PA-X with strong host 403
cell shut-off activity. This lack of correlation between repression of cellular gene expression 404
in avian cells and phenotypic effects in chicken embryos suggests that the PR8 PA-X protein 405
may harbour a function unrelated to host cell shut-off. The PR8 PA-X protein has been 406
proposed to inhibit stress granule formation, but via a mechanism linked to its endonuclease 407
activity and therefore presumably reflecting shut-off activity (52). Alternatively, it could be 408
that the PR8 PA-X polypeptide only exhibits repressive function in specific cell types, such 409
as those of the chorioallantoic membrane (the primary site of virus replication in eggs) or the 410
chick embryo itself. However, since we found low shut-off activity from it in a variety of 411
cells from different species and conversely, no great cell specificity of high activity PA-X 412
polypeptides (data not shown), we do not favour this hypothesis. 413
Several studies have found that sequences in the X-ORF make positive contributions 414
to the shut-off activity of PA-X (30, 37, 39, 45, 46). In contrast, here we found that for both 415
PR8 and T/E strains of the polypeptide, removal of X-ORF sequences actually increased 416
shut-off activity compared to the WT polypeptide. The effect was relatively modest and in 417
the case of PR8, did not confer equivalent activity to the full-length avian virus PA-X 418
polypeptides (Figure 1C). A similar outcome of greater inhibition from a truncated PA-X 419
polypeptide was seen with a triple reassortant swine influenza virus (42), suggesting that the 420
X-ORF can harbour negative as well as positive regulatory polymorphisms. 421
In some but not all studies, effects of PA-X mutations on viral pathogenicity have 422
been associated with differences in virus replication in vivo. While Jagger et al., (30) did not 423
attribute the increased virulence in mice upon loss of 1918 H1N1 PA-X to virus replication, 424
Gao and colleagues found that increased virulence in mice on loss of H5N1 PA-X was 425
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

18
associated with increased titres of
Δ
PA-X viruses in the lungs, brains and blood of infected 426
mice (34, 39). Similarly, Hu et al. found that increased virulence in chicken, ducks and mice 427
of a
∆
PA-X H5N1 virus was associated with increased virus titres in the host (35). Given the 428
postulated role of PA-X-mediated repression of cellular gene expression in controlling host 429
responses to infection, it is reasonable to hypothesise that these differing outcomes reflect the 430
variable interplay between host and virus that is well known to tip in favour of one or other 431
depending on exact circumstance (53). Our present study, where loss of a PA-X with little 432
apparent ability to modulate host gene expression had no significant effect on virus titres in 433
allantoic fluid or the chick embryos themselves, but nevertheless reduced pathogenicity, do 434
not support this hypothesis. However, differences in progeny virion composition in the form 435
of altered ratios of HA to NP and M1 between WT and FS viruses were seen. This may 436
differentially affect their ability to infect specific cell types, as the amount of virus receptor 437
varies between different tissue types and is a known determinant of tissue tropism of 438
influenza viruses (reviewed in (54, 55)). 439
Our findings have direct implications for HA yield of vaccine viruses in eggs. 440
Ablating PA-X expression did not affect yield from eggs of high growth viruses such as PR8 441
or 6:2 reassortant CVV mimics containing glycoproteins of human H3N2 strains, or 442
potentially pandemic low pathogenicity avian H9N2 or H1N1 viruses. However, mutation of 443
the PR8 PA-X gene in the background of a CVV analogue containing the HA and NA 444
segments from poor growing strains, such as A(H1N1)pdm09 viruses or a potentially 445
pandemic avian H7N3 isolate, increased HA yield by around 2-fold. The mechanism of 446
improved yield of certain virus subtypes but not others on loss of PA-X expression is unclear. 447
Others have found that mutating the FS site of PR8 PA-X has subtle effects on viral protein 448
expression in vitro, including lower levels of M1 (45), perhaps explaining the changes in HA 449
to M1 ratio we see. Beneficial outcomes to HA yield may only be apparent in low-yielding 450
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

19
strains where perhaps viral rather than cellular factors are limiting. Alternatively, changes in 451
virion composition between WT and FS viruses could result in subtype/strain-specific effects 452
depending on the balance between HA and NA activities (56). Whatever the mechanism, in 453
no case was loss of PA-X expression detrimental to yield of CVVs, when assessing HA yield 454
of a wide range of different influenza A subtypes/strains. This approach of modifying the 455
PR8 donor backbone therefore potentially supplies a ‘universal’ approach that can be applied 456
to all CVVs that is additive with, but without the need for, generation and validation of, 457
subtype/strain-specific constructs, as is required for strategies based on altering the 458
glycoprotein genes. This could be beneficial to improve antigen yield in a pandemic setting 459
where manufacturers are required to produce large amounts of vaccine quickly. 460
461
Materials and methods 462
463
Cell lines and plasmids 464
Human embryonic kidney (293T) cells, canine kidney Madin-Darby canine kidney epithelial 465
cells (MDCK) and MDCK-SIAT1 (stably transfected with the cDNA of human 2,6-466
sialtransferase; (57)) cells were obtained from the Crick Worldwide Influenza Centre, The 467
Francis Crick Institute, London. QT-35 (Japanese quail fibrosarcoma; (58)) cells were 468
obtained from Dr Laurence Tiley, University of Cambridge. Cells were cultured in DMEM 469
(Sigma) containing 10% (v/v) FBS, 100 U/mL penicillin/streptomycin and 100 U/mL 470
GlutaMAX with 1 mg/ml Geneticin as a selection marker for the SIAT cells. Infection was 471
carried out in serum-free DMEM containing 100 U/mL penicillin/streptomycin, 100 U/mL 472
GlutaMAX and 0.14% (w/v) BSA. Cells were incubated at 37ºC, 5% CO2. Reverse genetics 473
plasmids were kindly provided by Professor Ron Fouchier (A/Puerto Rico/8/34; (59)), 474
Professor Wendy Barclay (A/England/195/2009 (60) and A/turkey/England/50-92/91 (61)), 475
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

20
Dr John McCauley (A/California/07/2009; (62)), Dr Laurence Tiley 476
(A/mallard/Netherlands/10/1999 (63)), Professor Robert Lamb (A/Udorn/307/72 (64)), 477
Professor Earl Brown (A/Hong Kong/1/68 (65)) and Professor Munir Iqbal 478
(A/chicken/Pakistan/UDL-01/2008 (66)). RG plasmids for A/mallard/Netherlands/12/2000 479
(NIBRG-60) and A/turkey/Turkey/1/2005 (NIBRG-23; with the multi-basic cleavage site 480
removed (67)) were made by amplifying HA and NA genes by PCR from cDNA clones 481
available within NIBSC and cloning into pHW2000 vector using BsmB1 restriction sites. A 482
plasmid containing the Renilla luciferase gene behind the simian virus 40 early promoter 483
(pRL) was supplied by Promega Ltd. 484
485
Antibodies and sera 486
Primary antibodies used were: rabbit polyclonal antibody anti-HA for swine H1 (Ab91641, 487
AbCam), rabbit polyclonal anti-HA for H7N7 A/chicken/MD/MINHMA/2004 (IT-003-008, 488
Immune Tech Ltd), mouse monoclonal anti-HA for H5N1 (8D2, Ab82455, AbCam), 489
laboratory-made rabbit polyclonal anti-NP (2915) (68), anti-PA residues 16-213 (expressed 490
as a fusion protein with ß-galactosidase (69), anti-puromycin mouse monoclonal antibody 491
(Millipore MABE343), rabbit anti-PR8 PA-X peptide (residues 211-225) antibody (30) and 492
anti-tubulin-
α
rat monoclonal antibody (Serotec MCA77G). Secondary antibodies used were: 493
for immunofluorescence, Alexa fluor donkey anti-rabbit IgG 488 or 594 conjugates 494
(Invitrogen), for immunohistochemistry, goat anti-mouse horseradish peroxidase (Biorad 495
172-1011) and goat anti-rabbit horseradish peroxidase (Biorad 172-1019), for western blot, 496
donkey anti-rabbit IgG Dylight800 or Alexa fluor 680-conjugated donkey anti-mouse IgG 497
(Licor Biosciences). 498
499
Site–directed mutagenesis 500
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

21
The QuikChange® Lightning site-directed mutagenesis kit (Stratagene) was used according 501
to the manufacturer’s instructions. Primers used for site-directed mutagenesis of the segment 502
3 gene were designed using the primer design tool from Agilent technologies. The strategies 503
used to disrupt the frameshift site (FS) as well as generating C-terminally truncated versions 504
of PA-X via PTCs were as described (30); the cited study used the PTC1 construct. 505
506
507
Protein analyses 508
Coupled in vitro transcription–translation reactions were carried out in rabbit reticulocyte 509
lysate supplemented with 35S-methionine using the Promega TNT system according to the 510
manufacturer’s instructions. SDS–PAGE followed by autoradiography was performed 511
according to standard procedures. Immunoprecipitations were performed as previously 512
described (70). Transfection-based reporter assays to assess host cell shut-off by PA-X 513
(described previously (30)) were performed by co-transfecting QT-35 cells with a reporter 514
plasmid containing the Renilla luciferase gene along with pHW2000 plasmids expressing the 515
appropriate segment 3 genes with or without the desired PA-X mutations. 48 h post-516
transfection, cells were lysed and luciferase activity measured on a Promega GloMax 96-well 517
Microplate luminometer using the Promega Renilla Luciferase system. 518
519
Reverse genetics rescue of viruses 520
All viruses used in this study were made by reverse genetics. 293T cells were transfected 521
with eight pHW2000 plasmids each encoding one of the influenza segments using 522
Lipofectamine™ 2000 (Invitrogen). Cells were incubated for 6 hours post-transfection before 523
medium was replaced with DMEM serum-free virus growth medium. At 2 days post-524
transfection, 0.5 µg/ml TPCK trypsin (Sigma) was added to cells. Cell culture supernatants 525
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

22
were harvested at 3 days post-transfection. 293T cell culture supernatants were clarified and 526
used to infect 10-11 day-old embryonated hens’ eggs. At 3 days p.i., eggs were chilled over-527
night and virus stocks were partially sequenced to confirm identity. 528
529
RNA extraction, RT-PCR and sequence analysis. 530
Viral RNA extractions were performed using the QIAamp viral RNA mini kit with on-531
column DNase digestion (QIAGEN). Reverse transcription used the influenza A Uni12 532
primer (AGCAAAAGCAGG) using a Verso cDNA kit (Thermo Scientific). PCR reactions 533
were performed using Pfu Ultra II fusion 145 HS polymerase (Stratagene) or Taq Polymerase 534
(Invitrogen) according to the manufacturers’ protocols. PCR products were purified for 535
sequencing by Illustra GFX PCR DNA and Gel Band Purification kit (GE Healthcare). 536
Primers and purified DNA were sent to GATC biotech for Sanger sequencing (Lightrun 537
method). Sequences were analysed using the DNAstar software. 538
539
Virus titration 540
Plaque assays, TCID50 assays and haemagglutination assays were performed according to 541
standard methods (71). MDCK or MDCK-SIAT cells were used for infectious virus titration, 542
and infectious foci were visualised by either toluidine blue or immunostaining for influenza 543
NP and visualising using a tetra-methyl benzidine (TMB) substrate. 544
545
Virus purification and analysis 546
Allantoic fluid was clarified by centrifugation twice at 6,500 x g for 10 mins. Virus was then 547
partially purified by ultracentrifugation at 128,000 x g for 1.5 hours at 4ºC through a 30% 548
sucrose cushion. For further purification, virus pellets were resuspended in PBS, loaded onto 549
15-60% sucrose/PBS density gradients and centrifuged at 210,000 x g for 40 mins at 4ºC. 550
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

23
Virus bands were extracted from gradients and virus was pelleted by ultracentrifugation at 551
128,000 x g for 1.5 hours at 4ºC. Pellets were resuspended in PBS and aliquots treated with 552
N-glycosidase F (New England Biolabs), according to the manufacturer’s protocol. Virus 553
pellets were lysed in Laemmli sample buffer and separated by SDS-PAGE on 10% or 12% 554
polyacrylamide gels under reducing conditions. Protein bands were visualised by Coomassie 555
blue staining (ImperialTM protein stain, Thermo Scientific) or detected by immunostaining in 556
western blot. Coomassie stained gels were scanned and bands quantified using ImageJ 557
software. Western blots were scanned on a Li-Cor Odyssey Infrared Imaging system v1.2 558
after staining with the appropriate antibodies and bands were quantified using ImageStudio 559
Lite software (Odyssey). 560
561
Chick embryo pathogenesis model 562
Ten-day old embryonated hens’ eggs were inoculated via the allantoic cavity route with 1000 563
PFU in 100 μl per egg or mock (serum-free medium only) infected. Embryo viability was 564
subsequently determined by examination of veins lining the shell (which collapse on death) 565
and embryo movement (for a few minutes). At 2 - 3 days p.i. (depending on experiment), 566
embryos were killed by chilling, washed several times in PBS and then scored blind for overt 567
pathology by two observers in each experiment. Scores were 0 = normal, 1 = intact but with 568
dispersed haemorrhages, 2 = small, fragile embryo with dispersed haemorrhages. For 569
histology, embryos were decapitated, washed several times in PBS, imaged and fixed for 570
several days in 4% formalin in PBS. Two embryos per virus condition were sectioned 571
longitudinally and mounted onto paraffin wax. Tissue sections were cut and mounted onto 572
slides and stained with haematoxylin and eosin (H&E) by the Easter Bush Pathology Service. 573
Further sections were examined by immunohistofluorescence performed for influenza NP 574
(62). Sections were deparaffinised and rehydrated and heat-induced antigen retrieval was 575
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

24
performed using sodium citrate buffer (10 mM sodium citrate, 0.05% Tween20, pH 6.0). 576
Sections were stained with anti-NP antibody followed by an Alexa fluor-conjugated 577
secondary antibody. Pre-immune bleed serum was also used to confirm specificity of staining 578
by anti-NP antibody. Sections were mounted using ProLong Gold anti-fade reagent 579
containing DAPI (Invitrogen). Stained tissue sections were scanned using a Nanozoomer XR 580
(Hamamatsu) using brightfield or fluorescence settings. Images were analysed using the NDP 581
view 2.3 software (Hamamatsu). 582
583
Graphs and statistical analyses 584
All graphs were plotted and statistical analyses (Mantel-Cox test, t-tests and Dunnett’s and 585
Tukey’s tests as part of one-way Anova) performed using Graphpad Prism software. 586
587
Acknowledgements 588
We thank Dr. Francesco Gubinelli, Dr. Carolyn Nicolson and Dr. Ruth Harvey at the 589
Influenza Resource Centre, National Institute for Biological Standards and Control, U. K for 590
their support during experiments performed in their lab, and staff at the Easter Bush 591
Pathology service for pathology support, Bob Fleming and Dr José Pereira for imaging 592
assistance, and Dr. Liliane Chung and. Dr. Marlynne Quigg-Nicol for technical advice. 593
594
Funding information 595
This work was funded in part with Federal funds from the Office of the Assistant 596
Secretary for Preparedness and Response, Biomedical Advanced Research and Development 597
Authority, under Contract No. HHSO100201300005C (to OGE and PD), by a grant from UK 598
Department of Health’s Policy Research Programme (NIBSC Regulatory Science Research 599
Unit), Grant Number 044/0069 (to OGE) and the Intramural Research Program of the 600
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

25
National Institute of Allergy and Infectious Diseases (DIR, NIAID) (to J.K.T.), as well as 601
Institute Strategic Programme Grants (BB/J01446X/1 and BB/P013740/1) from the 602
Biotechnology and Biological Sciences Research Council (BBSRC) to PD, PB, LV and 603
HMW. BWJ, PD, and JKT are also thankful for the support of the NIH-Oxford-Cambridge 604
Research Scholars program. The views expressed in the publication are those of the author(s) 605
and not necessarily those of the NHS, the NIHR, the Department of Health, ‘arms’ length 606
bodies or other government departments. 607
608
References 609
610
1. Johnson NPAS, Mueller J. 2002. Updating the Accounts: Global Mortality of the 611 1918-1920 "Spanish" Influenza Pandemic. Bulletin of the History of Medicine 612 76:105-115. 613 2. Dawood FS, Iuliano AD, Reed C, Meltzer MI, Shay DK, Cheng P-Y, 614 Bandaranayake D, Breiman RF, Brooks WA, Buchy P, Feikin DR, Fowler KB, 615 Gordon A, Hien NT, Horby P, Huang QS, Katz MA, Krishnan A, Lal R, 616 Montgomery JM, Mølbak K, Pebody R, Presanis AM, Razuri H, Steens A, 617 Tinoco YO, Wallinga J, Yu H, Vong S, Bresee J, Widdowson M-A. 2012. 618 Estimated global mortality associated with the first 12 months of 2009 pandemic 619 influenza A H1N1 virus circulation: a modelling study. The Lancet Infectious 620 Diseases 12:687-695. 621 3. Khaperskyy DA, McCormick C. 2015. Timing Is Everything: Coordinated Control 622 of Host Shutoff by Influenza A Virus NS1 and PA-X Proteins. J Virol 89:6528-6531. 623 4. Vasin AV, Temkina OA, Egorov VV, Klotchenko SA, Plotnikova MA, Kiselev 624 OI. 2014. Molecular mechanisms enhancing the proteome of influenza A viruses: an 625 overview of recently discovered proteins. Virus Res 185:53-63. 626 5. Palese P, Schulman JL. 1976. Mapping of the influenza virus genome: identification 627 of the hemagglutinin and the neuraminidase genes. Proc Natl Acad Sci U S A 628 73:2142-2146. 629 6. Ritchey MB, Palese P, Schulman JL. 1976. Mapping of the influenza virus genome. 630 III. Identification of genes coding for nucleoprotein, membrane protein, and 631 nonstructural protein. J Virol 20:307-313. 632 7. Neumann G, Watanabe T, Ito H, Watanabe S, Goto H, Gao P, Hughes M, Perez 633 DR, Donis R, Hoffmann E, Hobom G, Kawaoka Y. 1999. Generation of influenza 634 A viruses entirely from cloned cDNAs. Proc Natl Acad Sci U S A 96:9345-9350. 635 8. Hoffmann E, Neumann G, Kawaoka Y, Hobom G, Webster RG. 2000. A DNA
636 transfection system for generation of influenza A virus from eight plasmids. Proc Natl 637 Acad Sci U S A 97:6108-6113. 638 9. Fodor E, Devenish L, Engelhardt OG, Palese P, Brownlee GG, Garcia-Sastre A. 639 1999. Rescue of influenza A virus from recombinant DNA. J Virol 73:9679-9682. 640
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

26
10. Robertson JS, Nicolson C, Harvey R, Johnson R, Major D, Guilfoyle K, Roseby 641 S, Newman R, Collin R, Wallis C, Engelhardt OG, Wood JM, Le J, Manojkumar 642 R, Pokorny BA, Silverman J, Devis R, Bucher D, Verity E, Agius C, Camuglia S, 643 Ong C, Rockman S, Curtis A, Schoofs P, Zoueva O, Xie H, Li X, Lin Z, Ye Z, 644 Chen LM, O'Neill E, Balish A, Lipatov AS, Guo Z, Isakova I, Davis CT, 645 Rivailler P, Gustin KM, Belser JA, Maines TR, Tumpey TM, Xu X, Katz JM, 646 Klimov A, Cox NJ, Donis RO. 2011. The development of vaccine viruses against 647 pandemic A(H1N1) influenza. Vaccine 29:1836-1843. 648 11. Johnson A, Chen LM, Winne E, Santana W, Metcalfe MG, Mateu-Petit G, 649 Ridenour C, Hossain MJ, Villanueva J, Zaki SR, Williams TL, Cox NJ, Barr JR, 650 Donis RO. 2015. Identification of Influenza A/PR/8/34 Donor Viruses Imparting 651 High Hemagglutinin Yields to Candidate Vaccine Viruses in Eggs. PLoS One 652 10:e0128982. 653 12. Ping J, Lopes TJ, Nidom CA, Ghedin E, Macken CA, Fitch A, Imai M, Maher 654 EA, Neumann G, Kawaoka Y. 2015. Development of high-yield influenza A virus 655 vaccine viruses. Nat Commun 6:8148. 656 13. Barman S, Krylov PS, Turner JC, Franks J, Webster RG, Husain M, Webby RJ. 657 2017. Manipulation of neuraminidase packaging signals and hemagglutinin residues 658 improves the growth of A/Anhui/1/2013 (H7N9) influenza vaccine virus yield in 659 eggs. Vaccine 35:1424-1430. 660 14. Adamo JE, Liu T, Schmeisser F, Ye Z. 2009. Optimizing viral protein yield of 661 influenza virus strain A/Vietnam/1203/2004 by modification of the neuraminidase 662 gene. J Virol 83:4023-4029. 663 15. Pan W, Dong Z, Meng W, Zhang W, Li T, Li C, Zhang B, Chen L. 2012. 664 Improvement of influenza vaccine strain A/Vietnam/1194/2004 (H5N1) growth with 665 the neuraminidase packaging sequence from A/Puerto Rico/8/34. Hum Vaccin 666 Immunother 8:252-259. 667 16. Jing X, Phy K, Li X, Ye Z. 2012. Increased hemagglutinin content in a reassortant 668 2009 pandemic H1N1 influenza virus with chimeric neuraminidase containing donor 669 A/Puerto Rico/8/34 virus transmembrane and stalk domains. Vaccine 30:4144-4152. 670 17. Harvey R, Nicolson C, Johnson RE, Guilfoyle KA, Major DL, Robertson JS, 671 Engelhardt OG. 2010. Improved haemagglutinin antigen content in H5N1 candidate 672 vaccine viruses with chimeric haemagglutinin molecules. Vaccine 28:8008-8014. 673 18. Harvey R, Johnson RE, MacLellan-Gibson K, Robertson JS, Engelhardt OG. 674 2014. A promoter mutation in the haemagglutinin segment of influenza A virus 675 generates an effective candidate live attenuated vaccine. Influenza Other Respir 676 Viruses 8:605-612. 677 19. Harvey R, Guilfoyle KA, Roseby S, Robertson JS, Engelhardt OG. 2011. 678 Improved antigen yield in pandemic H1N1 (2009) candidate vaccine viruses with 679 chimeric hemagglutinin molecules. J Virol 85:6086-6090. 680 20. Medina J, Boukhebza H, De Saint Jean A, Sodoyer R, Legastelois I, Moste C. 681 2015. Optimization of influenza A vaccine virus by reverse genetic using chimeric 682 HA and NA genes with an extended PR8 backbone. Vaccine 33:4221-4227. 683 21. Plant EP, Ye Z. 2015. Chimeric neuraminidase and mutant PB1 gene constellation 684 improves growth and yield of H5N1 vaccine candidate virus. J Gen Virol 96:752-755. 685 22. Plant EP, Liu TM, Xie H, Ye Z. 2012. Mutations to A/Puerto Rico/8/34 PB1 gene 686 improves seasonal reassortant influenza A virus growth kinetics. Vaccine 31:207-212. 687 23. Cobbin JC, Verity EE, Gilbertson BP, Rockman SP, Brown LE. 2013. The source 688 of the PB1 gene in influenza vaccine reassortants selectively alters the hemagglutinin 689 content of the resulting seed virus. J Virol 87:5577-5585. 690
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

27
24. Cobbin JC, Ong C, Verity E, Gilbertson BP, Rockman SP, Brown LE. 2014. 691 Influenza virus PB1 and neuraminidase gene segments can cosegregate during 692 vaccine reassortment driven by interactions in the PB1 coding region. J Virol 693 88:8971-8980. 694 25. Wanitchang A, Kramyu J, Jongkaewwattana A. 2010. Enhancement of reverse 695 genetics-derived swine-origin H1N1 influenza virus seed vaccine growth by inclusion 696 of indigenous polymerase PB1 protein. Virus Res 147:145-148. 697 26. Gomila RC, Suphaphiphat P, Judge C, Spencer T, Ferrari A, Wen Y, Palladino 698 G, Dormitzer PR, Mason PW. 2013. Improving influenza virus backbones by 699 including terminal regions of MDCK-adapted strains on hemagglutinin and 700 neuraminidase gene segments. Vaccine 31:4736-4743. 701 27. Giria M, Santos L, Louro J, Rebelo de Andrade H. 2016. Reverse genetics vaccine 702 seeds for influenza: Proof of concept in the source of PB1 as a determinant factor in 703 virus growth and antigen yield. Virology 496:21-27. 704 28. Mostafa A, Kanrai P, Ziebuhr J, Pleschka S. 2016. The PB1 segment of an 705 influenza A virus H1N1 2009pdm isolate enhances the replication efficiency of 706 specific influenza vaccine strains in cell culture and embryonated eggs. J Gen Virol 707 97:620-631. 708 29. Gilbertson B, Zheng T, Gerber M, Printz-Schweigert A, Ong C, Marquet R, Isel 709 C, Rockman S, Brown L. 2016. Influenza NA and PB1 Gene Segments Interact 710 during the Formation of Viral Progeny: Localization of the Binding Region within the 711 PB1 Gene. Viruses 8:238. 712 30. Jagger BW, Wise HM, Kash JC, Walters KA, Wills NM, Xiao YL, Dunfee RL, 713 Schwartzman LM, Ozinsky A, Bell GL, Dalton RM, Lo A, Efstathiou S, Atkins 714 JF, Firth AE, Taubenberger JK, Digard P. 2012. An overlapping protein-coding 715 region in influenza A virus segment 3 modulates the host response. Science 337:199-716 204. 717 31. Shi M, Jagger BW, Wise HM, Digard P, Holmes EC, Taubenberger JK. 2012. 718 Evolutionary conservation of the PA-X open reading frame in segment 3 of influenza 719 A virus. J Virol 86:12411-12413. 720 32. Yewdell JW, Ince WL. 2012. Virology. Frameshifting to PA-X influenza. Science 721 337:164-165. 722 33. Desmet EA, Bussey KA, Stone R, Takimoto T. 2013. Identification of the N-723 terminal domain of the influenza virus PA responsible for the suppression of host 724 protein synthesis. J Virol 87:3108-3118. 725 34. Gao H, Sun Y, Hu J, Qi L, Wang J, Xiong X, Wang Y, He Q, Lin Y, Kong W, 726 Seng LG, Sun H, Pu J, Chang KC, Liu X, Liu J. 2015. The contribution of PA-X to 727 the virulence of pandemic 2009 H1N1 and highly pathogenic H5N1 avian influenza 728 viruses. Sci Rep 5:8262. 729 35. Hu J, Mo Y, Wang X, Gu M, Hu Z, Zhong L, Wu Q, Hao X, Hu S, Liu W, Liu H, 730 Liu X, Liu X. 2015. PA-X decreases the pathogenicity of highly pathogenic H5N1 731 influenza A virus in avian species by inhibiting virus replication and host response. J 732 Virol 89:4126-4142. 733 36. Hayashi T, MacDonald LA, Takimoto T. 2015. Influenza A Virus Protein PA-X 734 Contributes to Viral Growth and Suppression of the Host Antiviral and Immune 735 Responses. J Virol 89:6442-6452. 736 37. Lee J, Yu H, Li Y, Ma J, Lang Y, Duff M, Henningson J, Liu Q, Li Y, Nagy A, 737 Bawa B, Li Z, Tong G, Richt JA, Ma W. 2017. Impacts of different expressions of 738 PA-X protein on 2009 pandemic H1N1 virus replication, pathogenicity and host 739 immune responses. Virology 504:25-35. 740
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

28
38. Hu J, Mo Y, Gao Z, Wang X, Gu M, Liang Y, Cheng X, Hu S, Liu W, Liu H, 741 Chen S, Liu X, Peng D, Liu X. 2016. PA-X-associated early alleviation of the acute 742 lung injury contributes to the attenuation of a highly pathogenic H5N1 avian 743 influenza virus in mice. Med Microbiol Immunol 205:381-395. 744 39. Gao H, Sun H, Hu J, Qi L, Wang J, Xiong X, Wang Y, He Q, Lin Y, Kong W, 745 Seng LG, Pu J, Chang KC, Liu X, Liu J, Sun Y. 2015. Twenty amino acids at the 746 C-terminus of PA-X are associated with increased influenza A virus replication and 747 pathogenicity. J Gen Virol 96:2036-2049. 748 40. Gao H, Xu G, Sun Y, Qi L, Wang J, Kong W, Sun H, Pu J, Chang KC, Liu J. 749 2015. PA-X is a virulence factor in avian H9N2 influenza virus. J Gen Virol 96:2587-750 2594. 751 41. Nogales A, Rodriguez L, DeDiego ML, Topham DJ, Martinez-Sobrido L. 2017. 752 Interplay of PA-X and NS1 Proteins in Replication and Pathogenesis of a 753 Temperature-Sensitive 2009 Pandemic H1N1 Influenza A Virus. J Virol 91:e00720-754 00717. 755 42. Xu G, Zhang X, Liu Q, Bing G, Hu Z, Sun H, Xiong X, Jiang M, He Q, Wang Y, 756 Pu J, Guo X, Yang H, Liu J, Sun Y. 2017. PA-X protein contributes to virulence of 757 triple-reassortant H1N2 influenza virus by suppressing early immune responses in 758 swine. Virology 508:45-53. 759 43. Naffakh N, Massin P, van der Werf S. 2001. The transcription/replication activity 760 of the polymerase of influenza A viruses is not correlated with the level of proteolysis 761 induced by the PA subunit. Virology 285:244-252. 762 44. Hayashi T, Chaimayo C, McGuinness J, Takimoto T, Abdel-Wahab M. 2016. 763 Critical Role of the PA-X C-Terminal Domain of Influenza A Virus in Its Subcellular 764 Localization and Shutoff Activity. Journal of Virology 90:7131-7141. 765 45. Khaperskyy DA, Schmaling S, Larkins-Ford J, McCormick C, Gaglia MM. 766 2016. Selective Degradation of Host RNA Polymerase II Transcripts by Influenza A 767 Virus PA-X Host Shutoff Protein. PLoS Pathog 12:e1005427. 768 46. Oishi K, Yamayoshi S, Kawaoka Y. 2015. Mapping of a Region of the PA-X 769 Protein of Influenza A Virus That Is Important for Its Shutoff Activity. J Virol 770 89:8661-8665. 771 47. Firth AE, Jagger BW, Wise HM, Nelson CC, Parsawar K, Wills NM, Napthine 772 S, Taubenberger JK, Digard P, Atkins JF. 2012. Ribosomal frameshifting used in 773 influenza A virus expression occurs within the sequence UCC_UUU_CGU and is in 774 the +1 direction. Open Biol 2:120109. 775 48. Abernathy E, Clyde K, Yeasmin R, Krug LT, Burlingame A, Coscoy L, 776 Glaunsinger B. 2014. Gammaherpesviral gene expression and virion composition are 777 broadly controlled by accelerated mRNA degradation. PLoS Pathog 10:e1003882. 778 49. Abt M, de Jonge J, Laue M, Wolff T. 2011. Improvement of H5N1 influenza 779 vaccine viruses: influence of internal gene segments of avian and human origin on 780 production and hemagglutinin content. Vaccine 29:5153-5162. 781 50. Harvey R, Wheeler JX, Wallis CL, Robertson JS, Engelhardt OG. 2008. 782 Quantitation of haemagglutinin in H5N1 influenza viruses reveals low 783 haemagglutinin content of vaccine virus NIBRG-14 (H5N1). Vaccine 26:6550-6554. 784 51. Kamal RP, Alymova IV, York IA. 2017. Evolution and Virulence of Influenza A 785 Virus Protein PB1-F2. Int J Mol Sci 19:96. 786 52. Khaperskyy DA, Emara MM, Johnston BP, Anderson P, Hatchette TF, 787 McCormick C. 2014. Influenza a virus host shutoff disables antiviral stress-induced 788 translation arrest. PLoS Pathog 10:e1004217. 789
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

29
53. Newton AH, Cardani A, Braciale TJ. 2016. The host immune response in 790 respiratory virus infection: balancing virus clearance and immunopathology. Semin 791 Immunopathol 38:471-482. 792 54. Klenk HD, Garten W, Matrosovich M. 2011. Molecular mechanisms of 793 interspecies transmission and pathogenicity of influenza viruses: Lessons from the 794 2009 pandemic. Bioessays 33:180-188. 795 55. Baigent SJ, McCauley JW. 2003. Influenza type A in humans, mammals and birds: 796 determinants of virus virulence, host-range and interspecies transmission. Bioessays 797 25:657-671. 798 56. Benton DJ, Martin SR, Wharton SA, McCauley JW. 2015. Biophysical 799 measurement of the balance of influenza a hemagglutinin and neuraminidase 800 activities. J Biol Chem 290:6516-6521. 801 57. Matrosovich M, Matrosovich T, Carr J, Roberts NA, Klenk HD. 2003. 802 Overexpression of the alpha-2,6-sialyltransferase in MDCK cells increases influenza 803 virus sensitivity to neuraminidase inhibitors. J Virol 77:8418-8425. 804 58. Moscovici C, Moscovici MG, Jimenez H, Lai MM, Hayman MJ, Vogt PK. 1977. 805 Continuous tissue culture cell lines derived from chemically induced tumors of 806 Japanese quail. Cell 11:95-103. 807 59. de Wit E, Spronken MI, Bestebroer TM, Rimmelzwaan GF, Osterhaus AD, 808 Fouchier RA. 2004. Efficient generation and growth of influenza virus A/PR/8/34 809 from eight cDNA fragments. Virus Res 103:155-161. 810 60. Brookes DW, Miah S, Lackenby A, Hartgroves L, Barclay WS. 2011. Pandemic 811 H1N1 2009 influenza virus with the H275Y oseltamivir resistance neuraminidase 812 mutation shows a small compromise in enzyme activity and viral fitness. J 813 Antimicrob Chemother 66:466-470. 814 61. Howard W, Hayman A, Lackenby A, Whiteley A, Londt B, Banks J, McCauley 815 J, Barclay W. 2007. Development of a reverse genetics system enabling the rescue of 816 recombinant avian influenza virus A/Turkey/England/50-92/91 (H5N1). Avian Dis 817 51:393-395. 818 62. Turnbull ML, Wise HM, Nicol MQ, Smith N, Dunfee RL, Beard PM, Jagger 819 BW, Ligertwood Y, Hardisty GR, Xiao H, Benton DJ, Coburn AM, Paulo JA, 820 Gygi SP, McCauley JW, Taubenberger JK, Lycett SJ, Weekes MP, Dutia BM, 821 Digard P. 2016. Role of the B Allele of Influenza A Virus Segment 8 in Setting 822 Mammalian Host Range and Pathogenicity. J Virol 90:9263-9284. 823 63. Bourret V, Croville G, Mariette J, Klopp C, Bouchez O, Tiley L, Guérin J-L. 824 2013. Whole-genome, deep pyrosequencing analysis of a duck influenza A virus 825 evolution in swine cells. Infection, Genetics and Evolution 18:31-41. 826 64. Chen BJ, Leser GP, Jackson D, Lamb RA. 2008. The influenza virus M2 protein 827 cytoplasmic tail interacts with the M1 protein and influences virus assembly at the site 828 of virus budding. J Virol 82:10059-10070. 829 65. Ping J, Dankar SK, Forbes NE, Keleta L, Zhou Y, Tyler S, Brown EG. 2010. PB2 830 and Hemagglutinin Mutations Are Major Determinants of Host Range and Virulence 831 in Mouse-Adapted Influenza A Virus. Journal of Virology 84:10606-10618. 832 66. Long JS, Giotis ES, Moncorge O, Frise R, Mistry B, James J, Morisson M, Iqbal 833 M, Vignal A, Skinner MA, Barclay WS. 2016. Species difference in ANP32A 834 underlies influenza A virus polymerase host restriction. Nature 529:101-104. 835 67. Robertson JS, Engelhardt OG. 2010. Developing vaccines to combat pandemic 836 influenza. Viruses 2:532-546. 837 68. Noton SL, Medcalf E, Fisher D, Mullin AE, Elton D, Digard P. 2007. 838 Identification of the domains of the influenza A virus M1 matrix protein required for 839
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

30
NP binding, oligomerization and incorporation into virions. J Gen Virol 88:2280-840 2290. 841 69. Blok V, Cianci C, Tibbles KW, Inglis SC, Krystal M, Digard P. 1996. Inhibition 842 of the influenza virus RNA-dependent RNA polymerase by antisera directed against 843 the carboxy-terminal region of the PB2 subunit. J Gen Virol 77:1025-1033. 844 70. Poole E, Elton D, Medcalf L, Digard P. 2004. Functional domains of the influenza 845 A virus PB2 protein: identification of NP- and PB1-binding sites. Virology 321:120-846 133. 847 71. Klimov A, Balish A, Veguilla V, Sun H, Schiffer J, Lu X, Katz JM, Hancock K. 848 2012. Influenza virus titration, antigenic characterization, and serological methods for 849 antibody detection. Methods Mol Biol 865:25-51. 850 851
852
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

31
Figures 853
854 FIGURE 1. Virus strain dependent variation in PA-X-mediated host cell shut-off 855 activity. A) Schematic showing mutations in segment 3: at the frameshift (FS) site to generate a PA-X null 856 virus, or in the X-ORF so that segment 3 expresses C-terminally truncated versions of PA-X (PTCs 1-4, size of 857 products indicated), or removing cytosine 598 (delC598) to place the X ORF in frame with PA such that only 858 PA-X is expressed. B, C) PA-X-mediated inhibition of cellular RNA polymerase II-driven gene expression in 859 QT-35 cells. B) Cells were co-transfected with 100 ng of pRL plasmid constitutively expressing Renilla 860 luciferase and a dilution series of the indicated segment 3 pHW2000 plasmids or with a fixed amount of the 861 empty pHW2000 vector (VOC). Luciferase activity was measured 48 h later and plotted as % of a pRL only 862 sample. Dose-inhibition curves were fitted using GraphPad Prism software. Data are mean ± SD of two 863 independent experiments each performed in triplicate. C) Cells were co-transfected with 100 ng of pRL plasmid 864 and 400 ng of effector pHW2000 plasmids expressing segment 3 products. Luciferase activity was measured 48 865 h later and plotted as the % of a pHW2000 vector only control. Data are the mean ± SD from 2 independent 866 experiments performed in duplicate. Dashed lines indicate groups of statistical tests (against the left hand bar in 867 each case; * p < 0.05, ** p < 0.01, *** p < 0.001) as assessed by Dunnett's test. D, E) In vitro translation of PA-868 X from PR8 segment 3 constructs. Aliquots of rabbit reticulocyte lysate supplemented with 35S-methionine were 869 programmed with the indicated plasmids and radiolabelled polypeptides visualised by SDS-PAGE and 870 autoradiography before (D) or after (E) immunoprecipitation with the indicated antisera. Arrowheads in (D) 871 indicate full length PA-X while molecular mass (kDa) markers are shown on the left. 872
873
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

32
874 FIGURE 2. Effect of PA-X mutations in a chick embryo pathogenicity model. Groups of 875 5-6 embryonated hens’ eggs were infected with 1000 PFU of the indicated viruses and A, B) embryo viability 876 was determined by candling at 2 days p.i. Data are plotted as mean ± SEM % embryo lethality from 3-4 877 independent experiments. Horizontal bars indicate statistical significance (* p < 0.05) as assessed by Dunnett’s 878 test. C) Infected eggs were monitored daily for embryo viability and survival was plotted versus time. Data are 879 from 3 independent experiments with 5 - 10 eggs per experiment. Statistical significance between WT and FS 880 viruses (** p < 0.01) was assessed by log-rank (Mantel-Cox) test. D-F) From the experiments described in A) 881 and B), embryos were imaged D) and E, F) scored blind by two observers as 0 = normal, 1= intact but bloody, 2 882 = small, damaged and with severe haemorrhages. Data are the average ± SEM pathology scores from 3-4 883 independent experiments. Horizontal bar indicates statistical significance (*** p < 0.001) as assessed by 884 Dunnett’s test. 885
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

33
886 FIGURE 3. Histopathology of chick embryos following infection with PR8 viruses. 887 Embryonated hens’ eggs were infected with segment 3 WT or mutant viruses or mock infected. 2 days p.i. 888 embryos were fixed, sectioned and stained with H&E before imaging with a Nanozoomer XR (Hamamatsu) 889 using brightfield settings; representative pictures are shown: A) head, scale bar = 5 mm, and B) body, scale bar 890 = 2.5 mm, and C) cerebral hemisphere, D) liver and E) kidney, scale bars for low and high magnification 891 images = 1mm and 100 μm or 500 μm, respectively. C = cerebral hemisphere, H = heart, L = liver, PCT = 892 proximal convoluted tubule, CD = collecting duct, HP=hepatocytes, BV= blood vessel). 893 894
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

34
895 896 FIGURE 4. Effects of mutating PA-X expression on virus replication in chick embryos. 897 Groups of 5-6 embryonated hens’ eggs were infected with the indicated viruses and at 2 days p.i., virus titres 898 determined by plaque assay from (A, B) allantoic fluid or (C) washed and macerated chick embryos. Graphs 899 represent the mean ± SEM from 3 (A, B) or 2-4 independent experiments (C). Titres of mutant viruses were not 900 significantly different compared to WT virus (Dunnett’s test). D) Embryos were fixed at 2 days p.i., sectioned 901 and stained for IAV NP and DNA before imaging using a Nanozoomer XR (Hamamatsu) on fluorescence 902 settings. Representative images of liver, heart and kidney are shown. Scale bars = 100 μm. NP = red, DAPI = 903 blue. 904 905
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

35
906 FIGURE 5. Virion composition of WT or FS mutant viruses. Embryonated hens’ eggs were 907 infected with WT or segment 3 mutant viruses or mock infected. At 2 days p.i., virus was purified from allantoic 908 fluid by sucrose density gradient ultracentrifugation and 3-fold serial dilutions A, B) analysed by SDS-PAGE on 909 10% polyacrylamide gels and staining with Coomassie blue. C, D) For PR8 and FPV, respectively, the ratios of 910 NP:HA1 and M1:HA2 were determined by densitometry of SDS-PAGE gels. Scatter plots with the mean and 911 SEM of 6 measurements from 3 independent experiments using 2 independently rescued virus stocks are shown. 912 Horizontal bars indicate statistical significance (* p < 0.05) as assessed by paired t-test. 913 914
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

36
915 916 FIGURE 6. Effect of the PA-X FS mutation on HA yield of A(H1N1) pdm09 CVV 917 mimics. Embryonated hens’ eggs were infected as indicated and A-C) HA titres in allantoic fluid measured at 918 3 days post-infection for (A) groups of 5 or (B) 20 eggs per condition. (A, B) Scatter plots of titres from 919 individual eggs with mean and SEM are shown. (*** p < 0.001) assessed by unpaired t-test. C) Average HA 920 titres from groups of eggs inoculated at the infection dose which gave maximum yield are shown as paired 921 observations. Statistical significance (* p < 0.05, n=9) assessed by paired t-test. D, E) Allantoic fluid was 922 clarified and virus pelleted by ultracentrifugation through 30% sucrose pads. Equal volumes of resuspended 923 virus pellets were separated by SDS-PAGE on a 12% polyacrylamide gel and visualized by (D) staining with 924 Coomassie blue (upper panel) or western blot for HA1 (lower panel) with (+) or without (-) prior treatment with 925 PNGase F. Molecular mass (kDa) markers and specific polypeptides are labelled. E) De-glycosylated HA1 yield 926 was quantified by densitometry of the western blots. Data points represent 8 independent experiments using 3 927 independently rescued RG virus stocks shown as paired observations. (** p < 0.01, n=8) as assessed by paired t-928 test. Circles represent Cal7 and squares represent Eng195 CVV mimics.929
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;

37
TABLE 1. Effects of the
∆
PA-X FS mutation on HA yield of CVVs grown in eggs. 930
aValues are mean ± SD. 931
bRelative means the ratio of the average HA titres (of eggs incubated at 35°C) or HA1 yield of FS (Δ PA-X) 932 viruses to their WT counterparts 933 cAn outlier from one experiment was ignored when taking the average. 934 935 936
Lineage Subtype
Strain No. of
indep endent
rescues
HA titre
6:2
virusa
HA titre
6:2FS
virus
Relativeb
HA titre Relativeb
HA1 yield No. of
expts.
Small
scale Large
scale
Human
pdm2009 H1N1 A/California/07/2009 2 106 ±
86.6 249 ±
109 2.65 ± 2.16
1.9 ± 1.07c 7 4 3
Human
pdm2009 H1N1 A/California/07/2009 chimeric
HA
(NIBRG-119) 1 - - - 1.54 ± 0.43 2 1 1
Human
pdm2009 H1N1 A/England/195/2009 1 5.71 ±
0.71 20.1 ±
9.92 3.79 ± 2.21
2.4 ± 0.37 2 1 1
Human
pdm1968 H3N2 A/Udorn/307/72 2
2200 ±
929 2100 ±
720 1.26 ± 0.47
1.35± 0.36 5 3 2
Human
pdm1968 H3N2 A/Hong Kong/1/68 1 801 ±
117 843 ±
140 1.05 ± 0.06
1.22 ± 0.39 3 2 1
Avian H7N3
A/mallard/Netherlands/12/2000
(NIBRG-60) 2 33.5 ±
29.1 45.0 ±
36.9 1.34 ±
0.18 1.55 ± 0.14 5 3 2
Avian H5N1
A/turkey/Turkey/1/2005/1/2005
(NIBRG-23) 2 47.1 ±
27.3 48.8 ±
23.2 1.13 ± 0.23
1.10 ± 0.30 5 3 2
Avian H1N1 A/mallard/Netherlands/10/99 2 123 ± 59
128 ± 37
1.22 ± 0.37
1.13 ± 0.36 5 4 1
Avian H9N2 A/chicken/Pakistan/UDL-
01/2008 2 302 ±
364 312 ±
264 0.92 ± 0.30
1.01 ± 0.18 4 2 2
All rights reserved. No reuse allowed without permission.
(which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint. http://dx.doi.org/10.1101/409375doi: bioRxiv preprint first posted online Sep. 6, 2018;