Differential gene expression in chicken primary B cells infected ex vivo with attenuated and very virulent strains of infectious bursal disease virus (IBDV)

Infectious bursal disease virus (IBDV) belongs to the family Birnaviridae and is economically important to the poultry industry worldwide. IBDV infects B cells in the bursa of Fabricius (BF), causing immunosuppression and morbidity in young chickens. In addition to strains that cause classical Gumboro disease, the so-called ‘very virulent’ (vv) strain, also in circulation, causes more severe disease and increased mortality. IBDV has traditionally been controlled through the use of live attenuated vaccines, with attenuation resulting from serial passage in non-lymphoid cells. However, the factors that contribute to the vv or attenuated phenotypes are poorly understood. In order to address this, we aimed to investigate host cell-IBDV interactions using a recently described chicken primary B cell model, where chicken B cells are harvested from the BF and cultured ex vivo in the presence of chicken CD40L. We demonstrated that these cells could support the replication of IBDV when infected ex vivo in the laboratory. Furthermore, we evaluated the gene expression profiles of B cells infected with an attenuated strain (D78) and a very virulent strain (UK661) by microarray. We found that key genes involved in B cell activation and signaling (TNFSF13B, CD72 and GRAP) were down-regulated following infection relative to mock, which we speculate could contribute to IBDV-mediated immunosuppression. Moreover, cells responded to infection by expressing antiviral type I IFNs and IFN-stimulated genes, but the induction was far less pronounced upon infection with UK661, which we speculate could contribute to its virulence.


RNA viruses 23
Word count: 6,112 24 Depository: The original microarray data produced in this study have been deposited in the 25 public database ArrayExpress (http://www.ebi.ac.uk/microarray-as/ae/), with the accession 26 number: E-MTAB-5947. 27 Introduction 1960s. Next generation vaccines have also been licensed, based on a recombinant 84 herpesvirus of turkey (HVT) vector, or immune complex vaccines (6). However, despite 85 these control efforts, the infection remains endemic worldwide and new strains have 86 emerged and spread, for example immune escape antigenic variants (7, 8) and a pathotypic 87 variant of very virulent (vv) phenotype (9, 10), the latter causing increased severity of 88 disease and higher mortality which can be up to 60% in some flocks, compared to 1-2% 89 following infection with classical strains (5). 90

91
The chicken B cell is pivotal to the pathogenesis of IBDV. It is therefore crucial to 92 characterise chicken B cell-virus interactions in order to improve our current understanding 93 of viral pathogenesis and identify areas that can be exploited to develop novel strategies for 94 controlling IBDV. Key questions that remain unanswered are the basis for the increased 95 pathogenicity of the vv strain and the mechanism of attenuation of cell-culture adapted 96 strains. However, until recently, it has not been possible to culture chicken primary B cells ex 97 vivo as, when they are removed from the BF, they do not survive for long (11). 98 Consequently, it has not been possible to perform a thorough analysis of the interactions of 99 chicken B cells with different strains of IBDV, and many pathogenesis studies to date have 100 been conducted in vivo, where birds are infected and bursal tissues are harvested at 101 necropsy for downstream analysis of gene expression (12)(13)(14)(15)(16)(17). 102

103
The gene encoding chicken CD40 ligand (chCD40L), a molecule responsible for B cell 104 proliferation in vivo, was identified (18) and a soluble fusion protein containing its 105 extracellular domain was subsequently shown to support the proliferation of chicken B cells 106 in culture for up to three weeks (19). In 2015, Schermuly et al. showed that chCD40L-treated 107 B cells could be infected with Marek's disease virus (11), demonstrating that the cells have 108 the potential for studying the consequences of lymphotropic virus infection. Despite these 109 successes, chicken primary B cell cultures have not yet been applied to the study of IBDV. 110 Here we report the successful culture of chicken primary B cells ex vivo in the presence of 111 soluble chCD40L and provide data demonstrating that these cells can support the replication 112 of an attenuated cell-culture adapted strain of IBDV (D78) and a very virulent strain that does 113 not replicate in non-lymphoid cells (UK661). Furthermore, we characterise the gene 114 expression profile of B cells infected with both strains of virus, identifying differences that 115 correlate with pathogenicity. 116 117

Results 118
Chicken primary B cells can be cultured in the presence of chicken CD40L. Consistent 119 with previous reports (11,19), we found that when chicken primary B cells were cultured in 120 the presence of soluble chCD40L, the number and viability of the cells was significantly 121 increased compared to when cells were cultured in the absence of chCD40L (Fig. 1). The 122 number of cells increased 4-fold from 9.02 x10 5 to 3.63x10 6 per ml over a period of 6 days 123 when chCD40L was added to the culture media, in contrast to when it was absent (p <0.05) 124 ( Fig. 1(a)). Cell viability was also significantly improved, for example from 25% at day 3 post-125 culture in the absence of chCD40L to 48% in the presence (p <0.05) ( Fig. 1(b)). goat-anti-mouse secondary antibody labelled with Alexa Fluor 488, and counterstained with 131 DAPI. Some cells had evidence of green fluorescence around the nucleus ( Fig. 2(a)), 132 consistent with the presence of IBDV in the cytoplasm of infected cells. This was evident for 133 both D78 and UK661 ( Fig. 2(a)). RNA was extracted from infected cultures at 5, 18, 24 and 134 48 hours post-infection, and subjected to reverse transcription quantitative polymerase chain 135 reaction (RTqPCR) with primers specific to a conserved region of the IBDV VP4 gene. The 136 expression of VP4 was first normalised to a house-keeping gene (TBP) and then expressed 137 as fold change in copy number relative to mock samples in a ΔΔCt analysis. The average 138 fold change in IBDV VP4 expression increased to 16,603 copies at 48 hours post-infection 139 with D78, and 38,632 copies at 48 hours post-infection with UK661. Taken together, these 140 data demonstrate that the chicken primary B cells could support the replication of cell-culture 141 adapted and vv IBDV strains. This is in contrast to primary chicken embryo fibroblasts 142 (CEFs) or the immortalised chicken fibroblast cell line, DF-1, which do not support the 143 replication of vv IBDV without prior adaptation that can lead to viral attenuation (20). 144 145 Chicken primary B cells infected with vv UK661 and attenuated D78 show a 146 differential gene expression profile. Next, we aimed to evaluate how the primary B cells 147 responded to infection with either D78 or UK661. At 5, 18, 24 and 48 hours post-infection, 148 we determined the level of expression of type I IFN (IFNβ) and the interferon stimulated 149 gene (ISG) IFIT5 by RTqPCR. The cells infected with D78 expressed significantly more IFNβ 150 and IFIT5 than cells infected with UK661 at 18, 24 and 48 hours post-infection (*p<0.05, 151 ***p<0.001, ****p <0.0001) (Fig. 3). Moreover, at these time points, the average expression 152 of IFNβ in cells infected with UK661 was actually reduced relative to mock-infected cells, 153 (****p<0001 at 18 hours post-infection) ( Fig. 3(a)). 154

155
To acquire a broader understanding of the effect of infection on the transcriptome, gene 156 expression of mock-, D78-and UK661-infected cultures was screened at 18 hours post-157 infection using the Affymetrix Chicken Genome array. This array contains comprehensive 158 coverage of 32,773 transcripts corresponding to over 28,000 chicken genes and also 159 includes probe sets for detecting IBDV transcripts, which we used to confirm the virus 160 infection. Full raw and processed microarray data have been deposited in ArrayExpress with 161 the Accession Number E-MTAB-5947. Principal-component analysis (PCA) was used to 162 visualize three-dimensional expression patterns of the RNA data sets (Fig. S1). The samples 163 for each individual treatment (mock-, D78-, and UK661-infected samples) mapped near to 164 each other in a cluster, reflecting minor variations within replicates of each treatment. 165 However, the groups were mapped separately to one another, demonstrating their 166 transcriptomic distinctiveness from one another. 167 Analysis of the array data showed that 69 genes were differentially regulated, relative to 168 mock-infected cells, following D78 infection (p<0.05, fold change cut-off: 1.5), 12 of which 169 were also those differentially regulated following infection with UK661 ( Fig. 4(a)). Of the 69 170 genes differentially regulated following D78 infection, 53 were up-regulated and 16 were 171 down-regulated. In contrast, all 12 genes differentially regulated following UK661 infection 172 were up-regulated; there were no statistically significant down-regulated genes. 173 174 A direct comparison of gene expression between D78 and UK661 infected samples ( Table  175 S4) identified 37 differentially regulated genes, 27 of which were up-regulated by D78, 176 relative to UK661 infection, and 10 of which were down-regulated. Two of the D78 versus 177 UK661 up-regulated genes (HBG2 and HSP25), and two of the down-regulated genes 178 (LOC422305 and MCOLN2) were not identified by the comparisons with mock-infected cells. 179

180
Unsupervised hierarchical clustering analysis of the significantly differentially regulated 181 genes of the study confirmed that more transcripts were up-regulated following D78-infection 182 compared to UK661-infection ( Fig. 4(b)). All the genes could be divided broadly into four 183 similarly-sized groups: The first group included the 16 genes that were transcribed at lower 184 levels in D78-infected samples compared to mock-infected samples. These genes were 185 involved in B-cell activation and signalling (TNFSF13B (which encodes BAFF), CD72, 186 GRAP), immune processes (TLR1LA, DUSP14, PLD4, MDK, PMP2, F10, GSN), and other 187 processes such as protein ubiquitination (UBE2E1) and cholesterol transport and binding 188 (TSPO2). In addition, LOC422305 and MCOLN2, found to be significantly down-regulated in 189 cells infected with D78 relative to cells infected with UK661, were also included in this group, 190 making a total of 18 genes. The LOC422305 gene encodes the mitochondrial-like ES1 191 protein (21), and the MCOLN2 gene encodes the ion channel TRPML2 (22) that has been 192 reported to enhance the replication of yellow fever and dengue viruses (23, 24). The other 193 three groups comprised the 53 genes that were transcribed to a higher level in D78-infected 194 cells relative to mock-infected cells, as well as the two genes that were significantly up-195 regulated by infection with D78 relative to infection with UK661 (HBG2 and HSP25), making 196 a total of 55 genes. These were further sub-grouped on the basis of their expression in cells 197 infected with UK661, with their transcripts being present at: a similar level (19 genes), a 198 marginally higher level (17 genes) or a moderately higher level (19 genes) than in mock-199 probability of a random intersection between a set of genes with ontology processes was 210 estimated with the "P" value of the hypergeometric intersection (See Fig. S2-S4 for a more 211 detailed analysis). The top GO processes were "defence response to virus" and "immune 212 system processes" (Fig. 4(c)). Taken together, these data show that following IBDV infection 213 cells launch a type I IFN response, characterised by the induction of ISGs, but that the 214 response is more marked following D78-infection than UK661-infection. One limitation of the 215 enrichment analysis is that it relied on a comparison with mammalian gene counterparts. reported in previous studies. Thus 35 of the genes that we identified as being differentially 301 expressed in D78-infected cells compared to mock-infected cells were previously reported 302 as differentially expressed following IBDV infection in vivo (12,(14)(15)(16)(17). Only one 303 contradiction was found: the GSN gene was found to be down-regulated in our study, yet up-304 regulated in one in vivo study (17). GSN encodes Gelsolin, which regulates actin assembly 305 and has also been associated with inhibiting apoptosis (40). The reason for this discrepancy 306 is unknown; it could be because our experiment characterised gene expression at 18 hours 307 post-infection, whereas the in vivo study was conducted at 3 and 4 days post infection. 308

309
We demonstrated that IBDV infection led to the down-regulation of key genes involved in B 310 cell activation and signaling, such as TNFSF13B, CD72 and GRAP. This is consistent with 311 previous in vivo studies that showed down-regulation of TNFSF13B, CD72 and GRAP in 312 bursal tissue following infection with IBDV strain F52/70 (16, 17), and with an in vitro study 313 that found CD72 to be down-regulated following IBDV infection of DT40 cells (38). 314 TNFSF13B encodes the B cell activating factor (BAFF), which is essential for the survival of 315 B cells. BAFF is a transmembrane protein that is readily cleaved to release a soluble factor. 316 in cells infected at the same time with the same MOI of the two strains. Moreover, we found 353 that IFNβ expression was significantly reduced in cells infected with UK661 compared to 354 mock infected cells. This suggests that UK661 is not only able to inhibit the up-regulation of 355 type I IFN, but might actually suppress its induction. Taken together, our results imply that 356 UK661 is able to inhibit the up-regulation of antiviral responses to a greater extent than D78, 357 which we speculate could contribute to its enhanced virulence. To date, two components of 358 IBDV, VP3 and VP4, have been implicated in the suppression of innate immune responses 359 to IBDV infection. The VP3 protein binds the viral double-stranded (ds) RNA genome and is 360 thought to block the interaction of MDA5 with the dsRNA, thereby inhibiting downstream 361 events that culminate in type I IFN production (47). In addition, the VP4 protein binds to the 362 cellular glucocorticoid-induced leucine zipper (GILZ) protein, which inhibits the activation of 363 nuclear factor kappa enhancing binding (NF-KB) protein and activator protein-1 (AP-1) (48). 364 In mammalian cells, NF-KB and AP-1 co-operate with interferon regulatory factor (IRF) 3 and 365 7 to stimulate type I IFN transcription. IBDV VP4 may therefore inhibit type I IFN responses 366 via binding to and enhancing the inhibitory action of GILZ (48). It is possible that differences 367 in the sequences of these viral proteins between D78 and UK661 lead to differences in the 368 antagonism of type I IFN induction.  (19)) before undergoing centrifugation at 2,000rpm for 20 minutes at 431 4C over Histopaque 1083 (Sigma-Aldrich). Bursal cells that banded at the interface of the 432 medium and histopaque were collected, washed in PBS and counted using the TC20 TM 433 Automated Cell Counter (Bio-Rad). Cells were seeded in 24 well plates at 1x10 7 cells per ml 434 and maintained at 37C in 5% CO 2 in B cell media supplemented with chCD40L. 435 436

Virus inoculation 437
Chicken primary B cells were infected with the attenuated strain D78 and the vv strain 438 UK661 at a MOI of 3, after which cells were washed with media and cultured in B cell media 439 supplemented with chCD40L. Following incubation at 37C with 5% CO 2 for the indicated 440 amount of time, a 100µl sample of each well was obtained for processing for bioimaging. 441 The remaining cells were washed in PBS, resuspended in RLT buffer and stored at -80C 442 until nucleic acid extraction. 443 444

Bioimaging 445
Cells were pelleted, washed in PBS, and fixed in 4% paraformaldehyde (Sigma-Aldrich) for 1 446 hour at room temperature. Cells were permeabilised using 0.5% Triton X-100 (Sigma-447 Aldrich) for 30 minutes at room temperature, blocked with 4% bovine serum albumin (Sigma-448 Aldrich) for 30 minutes at room temperature on a rotating platform and then labelled with a 449 primary mouse monoclonal antibody against the IBDV VP2 protein (clone JF7-PD5) and 450 incubated for 1 hour at room temperature. After washing in PBS, the cells were incubated for 451 1 hour with a goat-anti-mouse secondary monoclonal antibody conjugated to Alexa Fluor® 452 488 (Thermo Fisher Scientific). Cells were counterstained with DAPI (Sigma-Aldrich) and 453 adhered to glass cover slips (TAAB, Aldermaston, UK) that had been coated with CellTak 454 (Fisher Scientific) by centrifugation. Cover slips were mounted onto glass slides and stained 455 cells were viewed with a Leica SP5 confocal microscope. 456 457

Nucleic acid extraction 458
Total RNA was extracted from mock-infected and IBDV-infected B cells using an RNeasy kit 459 (Qiagen) according to the manufacturer's instructions. On-column DNA digestion was 460 performed using RNase-free DNase (Qiagen) to remove contaminating genomic DNA. RNA 461 samples were quantified using a Nanodrop Spectrophotometer (Thermo Scientific) and 462 checked for quality using a 2100 Bioanalyzer (Agilent Technologies). All RNA samples had 463 an RNA integrity number (RIN) ≥ 9.6. RNA samples were halved and processed for either 464 microarrays or RTqPCR. 465 466

Quantification of virus replication by RTqPCR 467
Reverse transcription of RNA samples was performed using SuperScript TM III Reverse 468 Transcriptase (Thermo Fisher Scientific). The cDNA was synthesised according to the 469 manufacturer's instructions. Forward and reverse primers and a Taqman probe (Table S1) 470 targeting a conserved region of the IBDV VP4 sequence (Sigma-Aldrich) were used to 471 amplify the template. Briefly, the template, primers and probe were added to a Taqman 472

Universal PCR Master Mix (Applied Biosystems) and reactions were performed on 7500 473
Fast Real-Time PCR system (Life Technologies) using the following cycling conditions: 95C 474 for 10 minutes; 40 cycles of 95C for 15 seconds, 60C for 1 minute; 95C for 15 seconds; 475 60C for 1 minute; 95C for 30 seconds; 60C for 15 seconds. All target gene expression 476 levels were normalised to the housekeeping gene TBP and compared with the mock controls 477 using the comparative C T method (also referred to as the 2 -ΔΔCT method) (51). process networks. Statistical significance was measured by the number of genes that map 530 onto a given pathway and was calculated on the basis of p-value, based on hypergeometric 531 distribution (a built-in feature of MetaCore). Enrichment analysis for the mock-infected Vs 532 D78, mock-infected Vs UK661 and D78 Vs UK661 datasets is presented in Fig. S2-S4. 533 534

Microarray validation 535
RTqPCR was performed on RNA samples using a two-step procedure. RNA was first 536 reverse-transcribed into cDNA using the QuantiTect Reverse Transcription Kit (Qiagen) 537 according to the manufacturer's instructions. qPCR was then conducted on the cDNA in a 538 384-well plate with a ABI-7900HT Fast qPCR system (Applied Biosystems). Mesa Green 539 qPCR MasterMix (Eurogentec) was added to the cDNA (5μl for every 2μl of cDNA