Histophilus somni Stimulates Expression of Antiviral Proteins and Inhibits BRSV Replication in Bovine Respiratory Epithelial Cells

Our previous studies showed that bovine respiratory syncytial virus (BRSV) followed by Histophilus somni causes more severe bovine respiratory disease and a more permeable alveolar barrier in vitro than either agent alone. However, microarray analysis revealed the treatment of bovine alveolar type 2 (BAT2) epithelial cells with H. somni concentrated culture supernatant (CCS) stimulated up-regulation of four antiviral protein genes as compared with BRSV infection or dual treatment. This suggested that inhibition of viral infection, rather than synergy, may occur if the bacterial infection occurred before the viral infection. Viperin (or radical S-adenosyl methionine domain containing 2—RSAD2) and ISG15 (IFN-stimulated gene 15—ubiquitin-like modifier) were most up-regulated. CCS dose and time course for up-regulation of viperin protein levels were determined in treated bovine turbinate (BT) upper respiratory cells and BAT2 lower respiratory cells by Western blotting. Treatment of BAT2 cells with H. somni culture supernatant before BRSV infection dramatically reduced viral replication as determined by qRT PCR, supporting the hypothesis that the bacterial infection may inhibit viral infection. Studies of the role of the two known H. somni cytotoxins showed that viperin protein expression was induced by endotoxin (lipooligosaccharide) but not by IbpA, which mediates alveolar permeability and H. somni invasion. A naturally occurring IbpA negative asymptomatic carrier strain of H. somni (129Pt) does not cause BAT2 cell retraction or permeability of alveolar cell monolayers, so lacks virulence in vitro. To investigate initial steps of pathogenesis, we showed that strain 129Pt attached to BT cells and induced a strong viperin response in vitro. Thus colonization of the bovine upper respiratory tract with an asymptomatic carrier strain lacking virulence may decrease viral infection and the subsequent enhancement of bacterial respiratory infection in vivo.


Introduction
Human or bovine respiratory disease often has a multifactorial etiology. Interactions between viruses and bacteria are known to cause more severe disease than either one alone even though both may colonize the upper respiratory tract without causing disease [1][2][3]. Several studies have revealed mechanisms of viral predisposition to bacterial pneumonia [1,3]. We have investigated viral bacterial interactions in bovine respiratory disease (BRD) which is caused by a variety of pathogens including the following bacteria: Mannheimia haemolytica, Pasteurella multocida, Mycoplasma bovis and Histophilus somni and viruses: bovine respiratory syncytial virus (BRSV), bovine viral diarrhea virus (BVDV), bovine parainfluenza 3 (BPIV-3) and bovine herpes virus 1 (BHV-1). Our team has reproduced experimental pneumonia in calves with either H. somni [4][5][6] or BRSV alone [7][8][9][10]. Studies of interactions of BRSV and H. somni showed that aerosol infection of calves with BRSV 6 days before intrabronchial inoculation of H. somni, resulted in more severe pneumonia of longer duration than either pathogen alone [11]. This was associated with higher increases in specific IgE anti-H. somni levels in the dual infection than in either single infection [11]. Since BRSV and H. somni infect both upper respiratory and lower respiratory tract cells in vivo and in vitro, we have used both bovine turbinate cells (BT) and bovine alveolar type 2 (BAT2) epithelial cells for further investigation of pathogenic mechanisms. To study bacterial viral synergy, BAT2 cells were infected with BRSV 60 h before treatment with H. somni concentrated culture supernatant (CCS) for 4 h. This resulted in increased retraction of the BAT2 cells and microarray analysis showed increased BAT2 cell expression of matrix metalloproteinase (MMP)1 and MMP3 over either treatment alone [12]. The dual treatment of BAT2 cells increased passage of H. somni across the alveolar cell monolayer and increased digestion of collagen IV, a major component of the alveolar basement membrane. Thus, dual infection facilitated invasion by the bacteria [12]. We found that the IbpA was the major factor in CCS which caused retraction of BAT2 cells [13]. IbpA consists of a surface fibrillar network that is released into culture supernatant from all strains isolated from disease and most strains from asymptomatic carriers [14,15]. However, the ibpA gene was missing in four serum sensitive strains from asymptomatic preputial carriers (1P, 129Pt, 130Pf and 133P) [16]. The complete genome sequence of one of these IbpA negative asymptomatic carrier stains (129Pt) has been reported [17]. IbpA from disease isolates of H. somni has two direct repeats (DR1 and DR2), each with a cytotoxic fic motif which adenylylates Rho GTPases interfering with the cytoskeleton [18]. The motivation for the current study came from the observation that treatment of BAT2 cells with H. somni CCS as described above increased mRNA expression of four antiviral proteins over that of either BRSV or dual treated cells. Viperin (virus-inhibitory protein, endoplasmic reticulum associated, IFN-inducible) or RSAD2 (radical S-adenosyl methionine domain containing 2) and ISG15 (IFN-stimulated gene 15-ubiquitin-like modifier) were the most up-regulated antiviral genes. Therefore we hypothesized that release of H. somni factors on the surface of respiratory epithelial cells before viral infection may inhibit subsequent viral infection, the opposite of synergy. To test this hypothesis, we investigated up-regulation of viperin protein in BAT2 cells and BT cells as well as the role of H. somni toxins on increasing expression of antiviral proteins. To address antiviral function, we examined the effect of CCS treatment of BAT2 cells on BRSV replication and mechanisms of up-regulation of antiviral genes. Lastly, adherence to BT cells was investigated because H. somni does colonize the bovine upper respiratory tract in vivo [19], does form biofilms in vitro as well as in vivo and adherence is a step in biofilm formation [20,21]. Sustained adherence to the epithelial surface would allow continuous release of H. somni secreted products which stimulate increased expression of antiviral proteins.

Materials and Methods Bacteria
Pathogenic H. somni strain 2336 and asymptomatic carrier H. somni strain 129Pt, which have been previously described [4,16,17], were grown on Difco BHI agar (BD Diagnostics, Sparks, MD) plates with 5% bovine blood in Alsever's solution (Lampire Biological Laboratories, Pipersville, PA) in candle jars at 37°C. Strain 2336.A1, with the ibpA gene deleted [22], was grown on BHI blood agar supplemented with Kanamycin (145 μg/ml). Growth from 18 h plates was suspended in BHI broth containing 0.1% Tris base and 0.01% thiamine monophosphate (BHITT). For studies of interaction of live bacteria with BT or BAT2 cells, the cfus of bacteria were estimated spectrophotometrically, since 75% light Transmission at 610 nm equals approximately 2 x10 8 cfu/ml. Concentrated culture supernatant (CCS) was prepared by incubating H. somni for 7 h in BHITT, with an estimated starting concentration of 5 x 10 7 cfu/ml. After 7 h of incubation, (approximately late log phase) the culture supernatant was concentrated 40 times with Amicon Ultra-15 Centrifugal Filter Units (EMD Millipore, Billerica, MA) and filtered through 0.22 um sterile syringe filters (EMD Millipore, Billerica, MA) as previously described [12].

Bovine Respiratory Syncytial Virus preparation
A virulent clinical isolate of BRSV (strain CA-1) was propagated in primary bovine turbinate (BT) cells as previously described (12). A 1ml aliquot was used for measuring the plaque forming units (PFU) /ml and the rest stored at -80°C in aliquots of 1ml in BT cell media containing 10% horse serum and transferred to liquid nitrogen after 2-4 hours.

BAT2 and BT cell culture
Primary bovine alveolar type 2 (BAT2) cells were isolated from newborn calf lung and kindly provided by Riccardo Rosenbusch (Iowa State University) as previously described [13] and approved by the Institutional Animal Care and Use Committee at Iowa State University. They were grown in DMEM/Defined Keratinocyte-SFM (1:1, Life Technologies, Grand Island, NY) supplemented with 2% gamma-irradiated FBS, 5 mM L-glutamine (Life Technologies, Grand Island, NY), 0.02% lactalbumin dehydrogenase (Life Technologies, Grand Island, NY) for up to 13 passages. Bovine turbinate (BT) cells (ATCC, Manassas, VA) passages 8 and 9 from a bovine turbinate cell line, were grown in DMEM containing 10% horse serum (Hyclone, Ther-moFisher Scientific, Waltham, MA). Penicillin (100U/mL) and streptomycin (100 ug/mL) (Life Technologies) were added to media of both cell types, unless cells were to be treated with live H. somni. Both cell types were grown at 37°C with 5% CO 2 in flasks or plates coated with 0.1% porcine gelatin (Sigma, St. Louis, MO). When appropriate, the Trypan Blue exclusion assay was used to measure cell viability. The cell suspension was mixed with an equal volume of 0.4% trypan blue solution (ThermoFisher Scientific, Waltham, MA). Viable cells, which did not take up the dye, were counted using a hemocytometer under 20x magnification. For assays other than the microarray studies, 12 well plates (Costar, Corning, Tewksbury, MA) were seeded at 3 x10 4 cells per well and grown to confluence before treatment with H. somni or its secreted products as described in the results section. At the indicated time points, the medium was removed and cells were immediately lysed with 100ul/well protein lysis buffer (150mM NaCl, 1% Triton X-100, 1mM EDTA, 50mM pH7.5 Tris) with protease inhibitors (cOmplete protease Inhibitor Cocktail Tablets, Roche, Indianapolis, IN) and lysates were frozen at -20°C for later Western Blot analysis.

Microarray analysis
BAT2 cells at 50% confluence were treated with BAT2 media alone or with BRSV at 0.5 MOI for 60 h at which time cells were almost confluent. This was followed by 4 h incubation with H. somni strain 2336 CCS at 20X concentration (final concentration in cell culture media) or media alone. RNA was extracted with the RNeasy minikit (Qiagen, Maryland) and gene expression was profiled using Affymetrix GeneChip bovine genome arrays, which were processed in the UC Davis School of Medicine Microarray Core Facility. The experiment was done 3 times and the data analyzed using dChip (DNA-Chip Analyszer) and DAVID (The Database for Annotation, Visualization and Integrated Discovery).

Lipooligosaccharide (LOS) preparation and quantitation
Strain 2336 LOS (also called endotoxin) was prepared by hot phenol extraction as previously reported [23,24]. The amount of endotoxin in CCS was determined with a Pierce LAL Chromogenic Endotoxin Quantitation Kit (Thermo Scientific, Rockford, IL) according to the manufacturer's instructions. 1 ug/ml LOS was considered to be equal to 20,000 EU/ml as reported by others [25]. This LOS preparation was used in assays of viperin expression by BT cells to investigate the role of this H. somni toxin in stimulation of antiviral proteins as described in the Results section.

Bacterial adherence assay
Adherence of H. somni strains to BT cells was done because attachment to epithelial cells is sometimes a step in mucosal colonization and in biofilm formation. Continuous release of secreted factors from H. somni at the surface of the upper respiratory tract may increase the antiviral response. Since BT cells are from the upper respiratory tract and are susceptible to BRSV and H. somni infection, those cells were used in these experiments. Adherence was determined by ELISA in 96-well plates coated with 0.01% gelatin, and seeded with 1 x 10 4 BT cells per well, as previously described, with modifications [26]. Coated wells without BT cells served as controls. Cells were grown to confluency and fixed with 0.05% L-glutaraldehyde (Sigma, St. Louis, MO) for 1 h. Previous studies showed that adherence to fixed cells was the same as to fresh cells but with less variability [26]. Any remaining L-glutaraldehyde was neutralized with 100mM glycine, and then the plates were washed and stored for up to one week at 4°C. Midlogarithmic phase H. somni growth in BHITT, was diluted to approximately 2 x 10 8 bacteria/ ml by spectrophotometry and 100 ul of this bacterial suspension was added to each well of BT cells at 300 MOI. After 1.5 h incubation at 37°C, wells were washed 4 times with PBS, and fixed with 5% formalin (Sigma, St. Louis, MO) in PBS. ELISA plates were blocked with 3% gelatin in PBS with 0.02% sodium azide at 37°C overnight, incubated with convalescent phase bovine serum (calves E5+E7, 1:1) at 1:1000 dilution, followed by peroxidase labeled goat anti bovine IgG (KPL, Gaithersburg, MA) and then developed with TMB sure blue (KPL, Gaithersburg, MA). The plates were read on a dual wavelength Vmax kinetic microplate reader (Molecular Devices Corp., Menlo Park, CA) at 450/650nm. Morphology of H. somni adherence was determined by treating BT cells with live bacteria at 100 MOI in 12 well plates as above but with cells grown on coverslips in the wells. Conditions were the same as for ELISA evaluation of adherence. Coverslips were washed in the wells, fixed with 4% paraformaldehyde, stained with 0.5% crystal violet for 5 m, washed and air dried.

BRSV proliferation in BAT2 cells
Confluent BAT2 cells in 12 well plates coated with 0.1% gelatin were treated with 1X CCS final concentration in BAT2 media in appropriate wells. Control wells were treated with BAT2 media alone. Six hours post initial CCS treatment, media or CCS treated wells were infected with BRSV at a concentration of 5.0 MOI. Additional wells with and without CCS treatment were not treated with BRSV. Cells in appropriate wells were retreated with 1X CCS at 24-h intervals for all time points longer than 24-h. Cells and supernatant were harvested at 24 h, 36 h, 48 h, and 72 h.
For PCR, BAT2 cells were lysed in triplicate by standard freezing and thawing, through the addition of 2-4ml fresh cold media and placing the plate into -80°C for 5 min, then removing the plate and placing into a 56°C water bath until thawed. The freeze/thaw supernatants were then combined with the BAT2 cell culture supernatants. Viral supernatants were then processed to extract the total RNA using the E.Z.N.A Viral RNA kit (Omega Bio-Tek), according the manufacturer's directions. Extracted RNA was stored at -80°C until use. Viral cDNA was synthesized by using SuperScript III First Strand synthesis system (Invitrogen, CA), according to the manufacturer's directions. The cDNA thermocycling program consisted of 10 min at 25°C, followed by 50 min at 50°C, and a termination cycle of 85°C for 5 min. Synthesized cDNA was stored at -20°C until use. The qRT-PCR was performed on a 384-well plate, in a 20 ul reaction volume. The 20 ul reaction mixture contained 10ul Sybr green qPCR master mix, 2 ul of BRSV-NP-F forward primer (GCAATGCTGCAGGACTAGGTATAAT), 2 ul BRSV-NP-R reverse primer (ACACTGTAATTGATGACCCCATTC), 2ul Nuclease free water, and 4ul of fresh cDNA product. The RT-PCR thermocycling program consisted of 48°C for 30 min, 95°C for 5 min, followed by 40 cycles of 95°C for 1 min and 55°C for 1 min. Fluorescence was measured following each cycle and displayed graphically (AB Applied Biosystems ViiA-7 detection software, version 1.1). ViiA-7 software was used to determine a cycle threshold (Ct) value, which identifies the first cycle at which the fluorescence is detected above the baseline for each sample or standard. The calculated Ct values for both GAP213DH and BRSV were analyzed against a standard curve. The standard curve was created using live virus, which was independently titered by TCID50. By using the TCID50 calculations for PFU and titers as the base line, this eliminated the dead virus from the calculation, as the TCID50 provides live viral concentrations. The cDNA prepared from the live virus was diluted 6 times using 10-fold dilutions for a final dilution of 1:1,000,000. From the serial dilution of virus cDNA the software determined the Ct value for each dilution. This Ct value was then converted into a log copy number. The log copy number for the neat viral sample was converted into a BRSV viral copy number using the calculation of live viral particles from the TCID50, this created the high point for the standard curve and this value was then diluted ten fold for 6 dilutions. By using the TCID50 value only live virus is measured. The Ct value for each sample for the housekeeping gene, GAPDH, and BRSV was then converted to a copy number using the standard curve. The copy number for BRSV was then divided by the copy number for GAPDH, to get the final value for viral shedding.

Statistical analysis
All graphs and statistical analyses by unpaired t tests were made using GraphPad Prism, version 6. P values of <0.05 ( Ã ) and P < 0.01 ( ÃÃ ) were considered significant. Data presented are the means and standard error of the mean (SEM).

Induction of antiviral genes by BRSV, H. somni or dual treatment of BAT2 cells
We previously showed that treatment of BAT2 cells with BRSV, H. somni CCS (20X concentration) or BRSV plus CCS caused up-regulation of mmp1 and mmp3 genes [12], with the dual treatment causing significantly more up-regulation than BRSV or H. somni CCS alone. Therefore, we investigated other changes in gene expression by microarray analysis of BAT2 cells after exposure to BRSV, CCS or dual treatment. We found that CCS alone induces up-regulation of mRNA transcripts for antiviral genes IFIH1 (interferon induced with helicase C domain 1), ISG15, MX1 (Myxovirus resistance 1), and RSAD2, also called viperin (Table 1). Interestingly, infection with BRSV or the dual treatment of BRSV followed by CCS up-regulated antiviral gene expression much less than CCS alone. These results were of special interest because H. somni may inhibit viral infection, unlike the viral bacterial synergy which is seen when the viral infection precedes the bacterial infection. Thus, we further investigated H. somni up-regulation of antiviral genes with a view to defining the role of increased antiviral proteins in inhibition of virally induced disease. ISG15 and viperin gene expression were up-regulated the most by treatment of BAT2 cells with H. somni CCS (Fig 1). We focused on viperin for studies of up-regulation of antiviral protein expression by H. somni and for studies of mechanism of up-regulation.

Up-regulation of viperin protein expression in cells treated with H. somni or its secreted products
The kinetics of viperin induction by CCS was determined for both bovine upper respiratory and lower respiratory cells (bovine turbinate and alveolar epithelial cells) because H. somni can be carried in the upper respiratory tract but is found primarily in the alveolus during experimental pneumonia [4] and because BRSV infects both cell types. Although functional studies and microarray studies with BAT2 cells were done by stimulating cells with CCS concentrated 20 times [12,13], dose response studies showed that this CCS could be diluted at least 1000 times before less viperin protein expression was detected in BT cells or BAT2 cells by Western blotting (Fig 2A and 2B). Therefore, subsequent studies used 1X CCS (equivalent to undiluted culture supernatant, which should be physiologic). Preliminary studies using the trypan blue exclusion method, showed that this concentration did not kill or inhibit proliferation of BT cells. The time course of viperin protein expression was then studied by stimulating cells with 1x concentrated CCS (equivalent to undiluted culture supernatant). Western blots of cell lysates showed that viperin protein expression was detectable in BT cells for at least 12 h (Fig 2C) and in BAT2 cells for at least 48 h after addition of CCS ( Fig 2D). Thus, H. somni CCS up-regulates viperin protein expression in a dose dependent and time dependent manner in both BT and BAT2 cells.

Effect of antiviral proteins on BRSV replication
Up-regulation of antiviral proteins in respiratory epithelial cells should result in decreased viral load in cells treated with H. somni or its culture supernatant, if the protein is functional. We tested this hypothesis in by treating BAT2 cells with H. somni CCS before infecting with BRSV since viperin protein expression was increased for longest in that cell type. BRSV was detected by qRT PCR of BAT2 control cells by 36 hours but only at very low levels of BRSV were detected in cells treated with H. somni culture supernatant throughout the 72 h study (Fig 3). So pretreatment of BAT2 cells with H. somni secreted products greatly depressed BRSV replication.  Table 1. Note that four antiviral protein genes (IFIH1, ISG15, MX1 and RSAD2 or viperin) were greatly up-regulated in CCS treated cells as compared with BRSV treated cells. Significant differences were analyzed by two tailed unpaired t test (*, P < 0.05; **, P < 0.01, NS, not significant).   BT cells were used in these experiments since inhibition of BRSV or other respiratory virus in the upper respiratory tract could result in fewer virus particles reaching the lung to produce pneumonia. Western blotting against convalescent phase serum showed that CCS contained the high molecular weight "ladder" of immunoglobulin binding proteins (IgBPs) as is characteristic of the IbpA cytotoxin as well as immunodominant 78, 60 and 40 K antigens which have not been shown to be cytotoxic (Fig 4A). Antibodies to IbpA DR2 revealed many IbpA bands from above 250 K to below 60 K, perhaps at least partly due to the multiple start codons in the IbpA gene as demonstrated earlier [27,28]. The amount of endotoxin or LOS (the second cytotoxin) was determined by LAL assay in 3 different CCS preparations (which had been used for cytotoxicity assays in the past). Average levels of LOS in CCS were shown to be about 5.8 ug/ ml (Fig 4B). To determine whether one of the two toxins (IbpA cytotoxin or LOS endotoxin) were involved in viperin induction, we compared BT cells challenged with live H. somni 2336 and the homologous strain with ibpA gene deleted (2336.A1) [22] at 10 MOI.Both the wild type (2336) and the knockout (2336.A1) induced viperin protein expression in a similar dose dependent manner, showing that the IbpA gene and protein were not necessary for H. somni up-regulation of viperin expression (Fig 5A and 5B). To determine whether H. somni LOS stimulated viperin, BT cells were treated with LOS extracted from H. somni strain 2336. Treatment of BT cells with serial dilutions of purified LOS from strain 2336 showed that as little as 8 ng/ml LOS induced a detectable amount of viperin as determined by Western blotting ( Fig  5C). This is comparable to the amount of endotoxin in the highest dilution of CCS (10 −3 ) which induced viperin expression (Fig 2), or 5.8 ng/ml as calculated from the LAL assay ( Fig  4B). Whether other factors in CCS, such as H. somni OMPs, might contribute to up-regulation of viperin expression was not determined. The ability of a naturally occurring ibpA negative strain 129Pt [17] to stimulate viperin expression in bovine upper respiratory epithelial cells was then determined since this strain was isolated from the epithelium of an asymptomatic carrier bull [29]. An ibpA negative strain was chosen because it lacks a critical virulence factor, so would not cause disease. Live H. somni 129Pt at a low MOI (2.5) induced viperin expression, and CCS from strain 129Pt at a 1:10,000 dilution induced as strong a viperin protein expression in BT cells as strain 2336 (Fig 6A and  6B compared with Figs 2A and 4A). These were similar results to those with the virulent strain 2336. Since colonization of the bovine upper respiratory tract with viperin inducing, non-virulent H. somni may induce antiviral protein expression and attachment is one step in colonization by biofilm formation, we examined attachment of strains 129Pt and 2336 to BT cells in vitro. Both microscopy (Fig 7A, 7B and 7C) and ELISA based assays (Fig 7D) showed that H. somni strain 129Pt adheres to BT cells approximately as well as the pathogenic strain 2336.

Discussion
This study shows that H. somni live bacteria and culture supernatant stimulate an antiviral response in the bovine upper and lower respiratory tract epithelial cells. This has implications  for host resistance since bacterial/viral interactions are critical to the etiology of this polymicrobial disease, and respiratory epithelial cells are a first line of defense. All or some of the four upregulated antiviral proteins in this study (IFIH1, ISG15, MX1 and RSAD/viperin) also were shown to be up-regulated in epithelial cells by cigarette smoke, or several viruses [30][31][32][33][34]. In our studies, H. somni up-regulated the antiviral response strongly, but BRSV infection alone up-regulated it only a little, if any. Surprisingly, BRSV, followed by H. somni CCS, down-regulated these antiviral genes as compared with CCS alone. This suggests that antiviral responses may be up-regulated if the bacterial infection occurs first but down regulated if the BRSV infection occurs first. Although many viruses and some bacterial products have been shown to induce an antiviral response, there has been little investigation of the effect of bacterial stimulation of antiviral proteins on viral infection.
Induction of bacterial inhibition by bacterial respiratory pathogens has been demonstrated previously. We showed that some isolates of bovine nasal flora inhibit in vitro growth of bovine bacterial respiratory pathogens, others enhance pathogen growth while still others have no effect [35]. Other investigators found that human bacterial pathogens, such as Streptococcus pneumoniae and Haemophilus influenzae, compete with each other in vitro and in vivo [2]. Viral predisposition to bacterial infection has been reported by our team and by others [2,11,12]. However, bacterial influence on viral infection is less well investigated. Bosch et al [1] cite Fig 7. Adherence of H. somni strain 2336 and 129Pt to BT cells. Live bacteria were co-incubated with BT cells at 300 MOI in 96-well plates for 2 h, free bacteria were washed off and adherence determined by ELISA with H. somni specific antibody. The morphology of adherence is shown in Giemsa stained BT cells on coverslips treated with 100 MOI H. somni 2336, 129Pt or no bacteria. Arrows indicate adhered bacteria. Although there was variation from field to field, adherence of both strains was similar. Significant differences were analyzed by two-tailed unpaired t test (**, P < 0.01, NS, not significant). several studies that suggest a preceding bacterial infection may increase susceptibility to a subsequent viral infection. Our studies indicate that H. somni infection of respiratory epithelial cells may have the opposite effect on BRSV infection. With a qRT-PCR assay based on the dose response and time course of viperin protein expression, we showed that the antiviral response induced by secreted factors of H. somni was associated with decreased production of BRSV by BAT2 cells. Many studies have shown that antiviral responses initiated by many DNA and RNA viruses do inhibit viral production [36][37][38][39]. For example, replication of human RSV virus in respiratory tract cells is inhibited by viperin in vitro and in vivo [34]. Large numbers of interferon stimulated genes were screened recently for antiviral activity against a panel of 14 viruses [40]. Infectivity of RSV was inhibited by several ISGs. The roles of ISG15 and RSAD2 in vivo and in vitro have been reviewed [41]. Most of the in vivo and in vitro ISG studies involved single genes investigated by ectopic expression or gene deletion. Although we did not do combinatorial studies of gene networks, our studies did look at several antiviral proteins expressed together in response to bacterial stimulation. The fact that this increased expression of several antiviral genes occurred when target respiratory epithelial cells from the natural host were treated with shed factors from bacteria involved in the disease complex suggests that the results may be relevant to disease prevention. Others have shown that infection with Salmonella typhimurium, Listeria monocytogenes, Lactobacillus acidophilus or Streptococcus agalactiae resulted in up-regulation of viperin expression by macrophages or dendritic cells but the effect on viral replication was not tested and epithelial cells responses were not examined [42][43][44]. Little information is available on induction of antiviral proteins by bacteria with subsequent inhibition of viral replication in vitro or in vivo. Therefore, we investigated the mechanisms of H. somni induction of antiviral responses, showing that H. somni LOS but not IbpA increased viperin protein expression. Both are toxic for bovine respiratory tract epithelial and endothelial cells [13,15,45]. We expected that IbpA may be involved in the antiviral response, since Bordetella pertussis FHA stimulates human peripheral blood mononuclear cells to up-regulate the ISG15 antiviral protein gene [46] and H. somni IbpA and FHA are quite homologous in the N terminal region. This was not the case. The C terminal portion of H. somni IbpA, with the two fic toxic domains, has no homology to FHA and was also not associated with the antiviral response. On the other hand, very low amounts of H. somni endotoxin (LOS) stimulated strong viperin responses in a dose dependent manner. This is consistent with other studies showing that endotoxin or lipopolysaccharide is a strong inducer of antiviral proteins, including IFIH1, ISG15, MX1 and viperin [47,48] but this has not been extended to studies of in vitro or in vivo inhibition of viral replication.
Since we showed that H. somni strain, 2336.A1, lacking the critical IbpA fic cytotoxic virulence factor, could stimulate a strong antiviral response, we determined whether a naturally occurring ibpA negative isolate from an asymptomatic carrier would have a similar effect. Future attempts to introduce a naturally occurring ibpA negative strain may be more acceptable than a genetically modified strain and an ibpA negative strain would lack the ability to cause invasive disease. We previously showed that four naturally occurring asymptomatic strains which lack the ibpA gene [16], were serum sensitive, did not cause cytotoxicity and were likely non-pathogenic members of the mucosal microbiome. We chose to study strain 129Pt, from among these four, because it has been well characterized and its genome fully sequenced [17]. It does not cause retraction of BAT2 cells, unlike IbpA positive strains [13,15]. Both CCS and live bacteria of strain 129Pt stimulated strong viperin protein production in BT cells. We determined the ability of strain 129Pt to adhere to BT cells, as an example of bovine upper respiratory tract cells because adherence is the first step in biofilm formation. Attachment and biofilm formation in the upper respiratory tract with ibpA negative strains should result in release of LOS over time in vivo, and up-regulation of antiviral proteins to inhibit BRSV and perhaps other respiratory viral infections. This may occur naturally in cattle carrying H. somni in the upper respiratory tract and may partially explain individual animal differences in susceptibility to respiratory infection. Although biofilm formation by H. somni strain 129Pt in the upper respiratory tract of cattle has not been tested experimentally, the in vitro results provide hope that it may be possible to introduce a bovine asymptomatic carrier isolate of H. somni (like strain 129Pt) to the upper respiratory microbiome of calves by intranasal inoculation to stimulate antiviral responses and decrease susceptibility to respiratory disease. Lactobacilli are often used as probiotics in the gut, and Lactobacillus acidophilus induces strong antiviral protein gene up-regulation [43]. Maybe a similar approach could be used prophylactically in the upper respiratory tract.