The stronger downregulation of in vitro and in vivo innate antiviral responses by a very virulent strain of infectious bursal disease virus (IBDV), compared to a classical strain, is mediated, in part, by the VP4 protein

IBDV is economically important to the poultry industry. Very virulent (vv) strains cause higher mortality rates than other strains for reasons that remain poorly understood. In order to provide more information on IBDV disease outcome, groups of chickens (n=18) were inoculated with the vv strain, UK661, or the classical strain, F52/70. Birds infected with UK661 had a lower survival rate (50%) compared to F52/70 (80%). There was no difference in peak viral replication in the bursa of Fabricius (BF), but the expression of chicken IFNβ, MX1 and IL-8 was significantly lower in the BF of birds infected with UK661 compared to F52/70 (p<0.05) as quantified by RTqPCR, and this trend was also observed in DT40 cells infected with UK661 or F52/70 (p<0.05). The induction of expression of type I IFN in DF-1 cells stimulated with polyI:C (measured by an IFN-β luciferase reporter assay) was significantly reduced in cells expressing ectopic VP4 from UK661 (p<0.05), but was higher in cells expressing ectopic VP4 from F52/70. Cells infected with a chimeric recombinant IBDV carrying the UK661-VP4 gene in the background of PBG98, an attenuated vaccine strain that induces high levels of innate responses (PBG98-VP4UK661) also showed a reduced level of IFNα and IL-8 compared to cells infected with a chimeric virus carrying the F52/70-VP4 gene (PBG98-VP4F52/70), and birds infected with PBG98-VP4UK661 also had a reduced expression of IFNα in the BF compared to birds infected with PBG98-VP4F52/70. Taken together, these data demonstrate that UK661 induced the expression of lower levels of anti-viral type I IFN and proinflammatory genes than the classical strain in vitro and in vivo and this was, in part, due to strain-dependent differences in the VP4 protein.

(50%) compared to F52/70 (80%). There was no difference in peak viral replication in the bursa of 23 Fabricius (BF), but the expression of chicken IFNβ, MX1 and IL-8 was significantly lower in the BF of 24 birds infected with UK661 compared to F52/70 (p<0.05) as quantified by RTqPCR, and this trend was 25 also observed in DT40 cells infected with UK661 or F52/70 (p<0.05). The induction of expression of 26 type I IFN in DF-1 cells stimulated with polyI:C (measured by an IFN-β luciferase reporter assay) was 27 significantly reduced in cells expressing ectopic VP4 from UK661 (p<0.05), but was higher in cells 28 expressing ectopic VP4 from F52/70. Cells infected with a chimeric recombinant IBDV carrying the 29 UK661-VP4 gene in the background of PBG98, an attenuated vaccine strain that induces high levels 30 of innate responses (PBG98-VP4 UK661 ) also showed a reduced level of IFNα and IL-8 compared to cells 31 infected with a chimeric virus carrying the F52/70-VP4 gene (PBG98-VP4 F52/70 ), and birds infected 32 with PBG98-VP4 UK661 also had a reduced expression of IFNα in the BF compared to birds infected 33 with PBG98-VP4 F52/70 . Taken together, these data demonstrate that UK661 induced the expression 34 of lower levels of anti-viral type I IFN and proinflammatory genes than the classical strain in vitro and 35 Moreover, as the virus has a preferred tropism for B cells, the majority of which reside in the bursa 47 of Fabricius (BF), surviving birds are often immunosuppressed, less responsive to vaccination 48 programmes, and more susceptible to secondary infections [3,4]. 49 50 Disease severity depends on numerous factors including the age and breed of the bird, and the 51 virulence of the infecting IBDV strain [5]. Since the first identification of IBDV in the 1960s, classical 52 divided into mock-infected (n=18), recombinant wild-type (wt) PBG98-infected (n=18), PBG98-131 VP4 F52/70 -infected (n=18) and PBG98-VP4 UK661 -infected (n=18) groups. At three weeks of age, birds 132 were inoculated with either PBS or a virus dose of 1.8x10 3 TCID50/bird, delivered intranasally, 50 µl 133 per nares. Clinical scores were recorded at least twice daily according to the points-based scoring 134 system (Supplementary Figure 1). Six birds from each infected group were humanely culled at 2, 4 135 and 14 days post-infection and the BF was harvested from each bird and divided into two sections, 136 one stored in RNAlater (Thermo Fisher Scientific) for RNA extraction and one snap frozen on dry ice. 137 All animal procedures conformed to the United Kingdom ASPA 1986 under  for 5 minutes, and disrupted by sonication. Samples were subject to sodium dodecyl sulfate-187 polyacrylamide gel electrophoresis (SDS-PAGE) using a Mini-PROTEAN tetra vertical electrophoresis 188 chamber (BIO-RAD) and transferred to a nitrocellulose membrane using a Trans-Blot Turbo Transfer 189 System (BIO-RAD). Membranes were stained with rabbit anti-VP4 (a gift from Jóse Castón of Centro 190 Nacional de Biotecnología (CNB-CSIC)) and mouse anti-βactin (Thermo Fisher Scientific), followed by 191 donkey anti-rabbit-680 and donkey anti-mouse-800 (LI-COR) and imaged with an Odyssey CLx  with PBS alone. Birds were assessed clinically at least twice daily, and humanely culled when 220 humane end-points were reached. At 24 and 48 hours post-infection (hpi), 6 birds per infected group 221 were humanely culled and the BF harvested for quantification of viral replication and host gene 222 expression. At 54 hpi, 3 of the remaining 6 (50%) birds inoculated with UK661 reached their humane 223 end-points and were humanely culled, compared to 1/6 (17%) of birds inoculated with F52/70 224 ( Figure 1A), consistent with UK661 being more virulent than F52/70, as expected. The remaining 225 infected and mock-inoculated birds were humanely culled at 72 hpi. There was no statistically 226 significant difference in the BF: body weight ratio (BF:BW) between groups of birds (Supplementary 227 Figure 2), a metric that is sometimes used as a surrogate of bursal pathology [20,21]. Moreover, the 228 kinetics of disease progression was similar between the two viral strains, peaking at 54 hpi ( Figure  229 1B), and there was no significant difference in the fold change in viral transcripts measured by 230 RTqPCR between the two strains at any of the time points measured ( Figure 1C). We also 231 determined the viral titres in the bursal tissue at each time point by TCID50, and although we found 232 UK661 replicated to a lower titre than F52/70 at 24 hpi (*P<0.05), there was no significant difference 233 in viral replication at later time points, which peaked at approximately 8 log10 TCID50/g of bursal 234 tissue for both strains ( Figure 1D). 235

236
The expression of type I IFN and pro-inflammatory genes was significantly reduced in BF tissue 237 harvested from birds infected with strain UK661 compared to strain F52/70 in vivo. RNA was 238 extracted from BF samples and reverse transcribed to cDNA that was used as the template in qPCR 239 assays targeting chicken type I IFN genes IFNα, IFNβ and Mx1, and pro-inflammatory cytokines IL-6, 240 IL-8 and IL-1β. These genes were selected as they are the most relevant to studying antiviral type I 241 IFN and pro-inflammatory responses, and have previously been shown to be upregulated following 242 IBDV infection [11][12][13][14][15]. The mean expression of IFNα ( 3B-D). Expression of IL-1β and IL-6 was also significantly reduced in cells infected with UK661 262 compared to F52/70 at multiple time-points (*P<0.05, **P<0.01, ***P<0.001) ( Figure 3E and F). In 263 contrast, IL-8 expression was significantly higher in cells infected with UK661 than F52/70 at 48 hpi 264 (***P<0.01) ( Figure 3G). Taken together, these data suggest that the vv strain is able to reduce the 265 expression of mRNAs for IFNα, IFNβ, Mx1 and the pro-inflammatory cytokines IL-1β and IL-6 to a 266 greater extent than the classical strain in vitro, confirming our in vivo data. IFNβ, Mx1, IL-6, IL-8, and IL-1β quantified by RTqPCR. There was no significant difference in viral 327 replication between any of the groups ( Figure 6A). However, virus replication was somewhat low (up 328 to 10 3 fold change in viral RNA per gram of BF tissue), possibly due to the cell-culture adapted nature 329 of the backbone. Consistent with our previous observations, at day 2 post-inoculation, the 330 expression of IFNα was significantly lower in the BF of birds infected with the PBG98-VP4 UK661 virus 331 compared to the PBG98-VP4 F52/70 virus (*P<0.05) ( Figure 6B). This trend was the same at day 4 post-332 inoculation although this did not reach statistical significance. Likewise, the average IFNβ expression 333 at day 4 post-inoculation was lower in the BF of birds infected with the PBG98-VP4 UK661 virus 334 compared to the PBG98-VP4 F52/70 virus, but this did not reach statistical significance ( Figure 6C). Mx1 335 expression was similar in birds infected with PBG98 and the chimeric viruses at days 2 and 4 post-336 inoculation ( Figure 6D). Comparing the pro-inflammatory response between these viruses, IL-1β 337 expression was significantly lower in the BFs of birds inoculated with either of the chimeric viruses at 338 2 days post-inoculation compared to recombinant wt PBG98 (***P<0.001) ( Figure 6E), but there was 339 no significant difference in IL-8 expression between any of the virus groups at day 2 or 4 post-340 inoculation ( Figure 6F). Taken together, these data demonstrate that the chimeric virus containing 341 the VP4 gene from the vvIBDV strain UK661 induced a lower level of type I IFNα compared to the 342 chimeric virus containing the VP4 gene from the cIBDV strain F52/70 both in vitro and in vivo. infected ex vivo [26]. Here, we extend these observations by demonstrating that UK661 is also able 353 to down-regulate type I IFN and pro-inflammatory cytokine responses compared to a classical field 354 strain, and we confirm that this occurs not only in vitro, but also in vivo (Figures 2 and 3). vvIBDV infection compared to a vaccine strain, however the vv strain replicated to significantly 361 higher titres that the vaccine strains in both studies, making comparison of gene expression changes 362 challenging [27,28]. In contrast, in our study, there was no significant difference in peak virus 363 replication between the F52/70 and UK661 strains, meaning that differences in gene expression are 364 due to something other than the amount of virus present. Unfortunately, some studies comparing 365 the innate immune response following vvIBDV infection to caIBDV strains also inoculated birds with 366 different amounts of virus, making a direct comparison of gene expression difficult [29]. To our 367 knowledge, only one previous study, by Eldaghayes et al., has compared classical and vvIBDV strains 368 in vivo [30]. Our data are consistent with this work, which also reported that a vv IBDV strain induced 369 reduced type I IFN responses compared to a classical strain. However, the authors conducted two 370 separate in vivo studies, one with each virus, and did not compare the two viruses in the same study. 371 Moreover, birds were inoculated with a different dose of each virus, making a comparison of gene 372 expression challenging. We extend these observations by directly comparing the vv and classical 373 strains in the same in vivo study, in birds inoculated with the same dose of each virus. 374 375 We also demonstrate that the differences in IFN antagonism are, in part, due to strain-dependent 376 differences in the VP4 proteins (Figures 4, 5 and 6). The VP4 protein from vvIBDV strain Lx has 377 previously been shown to act as an IFN antagonist through an interaction with the host 378 glucocorticoid-induced leucine zipper (GILZ) protein [14,22]. GILZ plays a key role in the regulation 379 of NF-κB activation by binding to the p65 subunit and preventing its translocation into the nucleus 380 and the downstream expression of cytokines [31]. The Lx VP4 protein has been shown to bind to 381 GILZ, preventing its ubiquitination and degradation, resulting in its accumulation in the cytoplasm. 382 Consequently, this VP4-GILZ interaction enhances the inhibition of p65 translocation into the 383 nucleus, leading to a reduction in the expression of pro-inflammatory cytokines and type I IFN 384 responses [14]. We extend these observations by demonstrating that there are strain dependent 385 differences in the extent to which VP4 antagonises type I IFN induction. Elucidating differences in 386 the mechanism of action of the different VP4 proteins was beyond the scope of this study, however, 387 it might be possible that while the UK661 VP4 is capable of binding GILZ in a manner similar to the Lx 388 When the chimeric viruses were inoculated into chickens, both replicated to low titres and caused 405 few clinical signs, most likely because these viruses possess the backbone of a cell culture adapted, 406 highly attenuated virus. We were therefore unable to ascertain whether strain-dependent 407 differences in VP4 affected IBDV virulence. Previous studies, using chimeric viruses with segments A 408 and B from strains of differing virulence, found that both segments contributed to virulence [9, 32-409 34]. While the mechanism is not yet understood, the effect of VP1 mutations may be related to viral 410 replication, whereas VP2 has been shown to activate apoptosis via the reduction of the anti-411 apoptotic molecule, ORAOV1 [35], and VP5 plays a key role in apoptosis by preventing it early during 412 infection and by activating it at later time points [36]. Virulence is therefore likely to be a complex 413 phenotype, however, it is possible that the VP4 sequence could contribute to this. Five of the nine 414 amino acid residues in UK661 that were different from F52/70 are also found in diverse vv IBDV 415 strains from different geographical regions, but were not found in other classical or attenuated 416 strains (31I, 170Y, 175N, 205S, and 241D) (Supplementary Figure 3A). It is therefore possible that 417 enhanced ability to antagonise type I IFN responses is feature of vvIBDV strains, and these amino 418 acids represent a VP4 "genetic signature of virulence". Moreover, given that the classical strains 419 emerged in the 1960s and vv strains subsequently emerged in the 1980s (apparently and probably 420 by segment reassortment), it is tempting to speculate that vvIBDVs evolved to have a VP4 protein 421 with an enhanced ability to inhibit NF-κB activation, promoting increased virus fitness due to 422 suppressed antiviral responses. Consistent with this hypothesis, the replication of the UK661 virus 423 was significantly enhanced in DT40 cells compared to the F52/70 virus ( Figure 3A). Interestingly, the 424 clinical scores of birds inoculated with the F52/70 virus were actually higher than birds inoculated 425 with UK661 at 48 hpi (Figure 1 B). At first this seems counter-intuitive that the vv strain would cause 426 less severe symptoms than the classical strain, however, this may be due to the increased expression 427 of type I IFN and pro-inflammatory genes in birds inoculated with the F52/70 virus, as stimulation of 428 these innate immune responses would be expected to result in clinical signs such as lethargy, 429 depression and ruffled feathers as observed to a greater extent in the F52/70-inoculated group. This 430 further underpins the complexity of defining virulence and highlights that our understanding of why 431 some birds reach humane end points, whereas others do not, remains unknown. 432

433
Our study is not without limitations: Gene expression was quantified by RTqPCR, and only a small 434 panel of genes were investigated. It would be beneficial in the future to compare gene expression by 435 RNA-Seq to gain a more comprehensive comparison of strain-dependent differences in expression. 436 Never-the less, our dataset does allow us to draw useful conclusions. Additionally, we only 437 compared the UK661 strain with the F52/70 strain and it would be interesting to compare vv, 438 classical, and vaccine strains, and possibly also diverse strains from different geographical regions, or 439 serotype 1 compared to serotype 2. However, this was beyond the scope of the current project. 440 Despite its limitations, our study does provide useful information that can be used to inform IBDV 441 surveillance efforts and improve IBDV vaccines: Identifying genetic signatures of increased IBDV 442 virulence could be used to better inform national surveillance efforts in order to calculate the 443 potential threat of an emerging strain as early as possible. Moreover, identifying genetic signatures 444 of attenuation could be used to engineer a rationally designed live vaccine candidate. For example, it 445 might be beneficial to explore the potential, as novel vaccine candidates, of chimeric viruses 446 engineered with the VP4 gene from attenuated strains in the backbone of a field strain. 447 448 Taken together, our data demonstrate that UK661 induced the expression of lower levels of anti-449 viral type I IFN responses than the classical strain in vitro and in vivo and this was, in part, due to 450 strain-dependent differences in the VP4 protein.  1: The UK661 strain was more virulent than the F52/70 strain, but both strains replicated to the same peak titre in vivo. Birds were checked twice daily by two independent observers for clinical signs and a Kaplan Meier survival curve plotted of mock-(black), F52/70-(pink) and UK661-(grey) inoculated birds that reached their humane end points (clinical score of 11) (A). Clinical signs were quantified by a scoring system and divided into mild (1-7) and moderate (8-11). Each bird was assigned a clinical score at the indicated time points post-infection (B). Six birds per group were humanely culled at 24 and 48 hours post-infection (hpi), one F52/70 and three UK661-infected birds reached their humane end-points at 54 hpi and the remaining birds were culled at 72 hpi. The bursa of Fabricius was harvested at necropsy and the log10 fold change in viral RNA copies/g tissue determined by RT-qPCR (C). The infectious titre was determined by titration onto DT40 cells in the method described by Reed & Muench. Virus titres were expressed as log10 TCID50/g of tissue (D). The horizontal lines are the mean values. Data passed a Shapiro-Wilk normality test before analysis using a two-tailed unpaired Student's t-test (*P<0.05).   Figure 3: The expression of type I IFN and pro-inflammatory genes was significantly reduced in B cells infected with strain UK661 compared to strain F52/70 in vitro. DT40 Cells were infected at an MOI of 0.1 with either the UK661 or F52/70 IBDV strains, or mock-infected with media alone and RNA was extracted from the cells at the indicated time points post-infection. RNA was reverse transcribed and amplified by qPCR using specific primer sets. The CT values were normalised to the housekeeping gene RPLPO and the log10 fold change in virus gene expression determined for the infected samples relative to the mock-infected samples in a ΔΔCT analysis and plotted (A). The log2 fold change in host-cell gene expression was also determined for the infected samples relative to the mock-infected samples in a ΔΔCT analysis and plotted (B-G). Data subsequently passed a Shapiro-Wilk normality test before being analysed by a one-way ANOVA and a Tukey's multiple comparison test (*P<0.05) (A), or a two-tailed unpaired Student's t-test (*P<0.05, **P<0.01, ***P<0.001). Data shown are representative of at least three replicate experiments, columns represent the mean values, and error bars represent the standard deviation of the mean.

Figure 4:
The VP4 protein from the UK661 strain antagonised IFNβ induction, but the VP4 protein from the F52/70 strain did not in vitro. DF-1s were transfected with the chicken IFNβ promoter Firefly luciferase reporter and a constitutively active Renilla expression plasmid and 500ng of either eGFP-UK661-VP4 or eGFP-F52/70-VP4 expression plasmids, or a control plasmid expressing eGFP alone. Twenty-four hours post-transfection, cells were re-transfected with poly I:C. At 6 hours posttransfection, cells were lysed and luciferase activity quantified. Firefly luciferase activity was normalised to Renilla expression. Data presented are the means of three independent experiments and passed a Shapiro-Wilk normality test before analysis using a two-tailed unpaired Student's t-test (*P<0.05, ** P<0.01). Error bars represent the standard error of the mean (SEM). In a parallel experiment, transfected cells were lysed and samples denatured and subject to SDS-PAGE gel electrophoresis followed by transfer to a nitrocellulose membrane and staining with anti-VP4 and anti-β-actin antibodies in a western blot.
A F B C D E A Figure 5: The ability of the UK661 VP4 protein to antagonise type I IFN responses was reduced in the context of the whole virus in vitro. DF-1 cells were infected with PBG98, PBG98-VP4 UK661 and PBG98-VP4 F52/70 viruses at an MOI of 1, before RNA was extracted at the indicated time points postinfection and reverse transcribed. Virus specific primers were used to amplify the cDNA by quantitative PCR, the CT values were normalised to the housekeeping gene RPLPO and the log10 fold change in virus gene expression was determined for the infected samples relative to the mockinfected controls in a ΔΔCT analysis and plotted. A Kruskal-Wallis test was performed with a Dunn's multiple comparison test where no significant difference was found at any time point between the three viruses (A). A panel of genes, IFNα (B), IFNβ (C), Mx1 (D), IL-1β (E), and IL-8 (F), were amplified by quantitative PCR using specific primer sets for target genes, before the CT values were normalised to the housekeeping gene RPLPO and the log2 fold change in gene expression determined for the infected samples relative to the mock-infected controls in a ΔΔCT analysis and plotted. Data are representative of at least three replicate experiments and passed a Shapiro-Wilk normality test before analysis using a two-tailed unpaired Student's t-test (*P<0.05, **P<0.01, ***P<0.001). The mean values are plotted and the error bars are the standard error of the mean (SEM). The dashed horizontal line represents the cut-off, below which genes were significantly down-regulated.
A B C D E F Figure 6: The ability of the UK661 VP4 protein to antagonise type I IFN responses was reduced in the context of the whole virus in vitro. Birds were inoculated with 1.8x10 3 TCID 50 of the PBG98, PBG98-VP4 UK661 and PBG98-VP4 F52/70 viruses, and the bursa of Fabricius was harvested at necropsy from 6 birds per group at 2, 4 and 14 days post-inoculation. RNA was extracted prior to reverse transcription to cDNA and qPCR amplification with virus-specific primers. CT values were normalised to a housekeeping gene and expressed as log 10 fold change viral RNA relative to mock-infected samples as per the ΔΔCT method. The data passed a Shapiro-Wilk normality test before being analysed using a two-way ANOVA (not significant) (A). At 2 and 4 days post-inoculation, cDNA was amplified by qPCR for a panel genes: IFNα (B), IFNβ (C), Mx1 (D), IL-1β (E), and IL-8 (F). The CT values were normalised to the housekeeping gene RPLPO and expressed relative to mock-infected samples using the ΔΔCT method. Data are representative of at least three replicate experiments and passed a Shapiro-Wilk normality test before analysis using a two-tailed unpaired Student's t-test (*P<0.05, ***P<0.001).
Horizontal lines represent the mean and error bars represent the standard error of the mean (SEM).