DF-1 is a chicken embryonic fibroblast cell line from East Lansing strain eggs. The chIRF7 gene knockout DF-1 cell line (IRF7^−/−) was obtained through the CRISPR/Cas9 technique, as described in our previous study (36). Chicken embryo fibroblast cells (CEFs) were prepared from 10-day-old specific-pathogen-free (SPF) embryos which were obtained from Merial Vital (Beijing, China). The CEFs, DF-1 and IRF7^−/− DF-1 cells were maintained in high-glucose Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, USA) containing 10% fetal bovine serum (FBS) (Gibco) and 1% penicillin-streptomycin (Gibco). All cells were incubated in 5% CO₂ at 37°C. The A/chicken/Shanghai/010/2008 (H9N2) virus (SH010), a H9N2 subtype of AIV, was isolated from chickens in Shanghai, China, in 2008. The NDV strain NSD14 was isolated from chickens in a farm of Shandong province, China. Green fluorescent protein (GFP) tagged NDV low virulent strain LaSota named NDV-GFP, and GFP tagged vesicular stomatitis virus (VSV) VSV-GFP, were stored in our laboratory. These viruses were purified, propagated, and stored as described in our previous study (37).

Based on the chDDX1 sequence (NM_204563.1) obtained from the National Center for Biotechnology Information (NCBI), the primers chDDX1-F and chDDX1-R ( Table 1 ) located outside the chDDX1 open reading frame (ORF) were designed and used to amplify chDDX1 cDNA via RT-PCR from DF-1 cells. The PCR product was ligated into a pTOPO-Blunt vector (Aidlab, Beijing, China) for sequencing, and the positive colonies were sent to the Beijing Genomics Institute (Beijing, China) for sequencing. The amino acid sequence of chDDX1 was aligned with the other animal DDX1 proteins from humans, mice, pigs, and ducks using Clustal W and edited with ESPript 3.0 (http://http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi). Sequence homology and a phylogenetic analysis of the DDX1 amino acid sequences were conducted using DNASTAR. A phylogenetic tree was constructed based on the DDX1 from 18 different species, including mammals, birds, and fish. Different domains in the chDDX1 amino acid sequences were predicted using the simple modular architecture research tool (SMART) program (http://smart.embl-heidelberg.de/). Homology modeling for chDDX1 was conducted using the online protein-modeling server Swiss-Model (http://swissmodel.expasy.org/).

FLAG-tagged chDDX1 plasmids were constructed by inserting full-length chDDX1 into the Kpn I and BamH I sites p3 × FLAG-14-CMV of the expression vector using a ClonExpress II one-step cloning kit (Vazyme, Nanjing, China). The primers used in the PCR are listed in Table 1 . The truncated plasmids of chDDX1, including dSPRY domain, dHELICc domain, d1-100 aa (deletion of 1-100 amino acids), d1-250 aa, d1-500 aa, d251-500aa, and d501-740 aa, were constructed using a modified homologous recombination method and the primers listed in Table 1 . The chIFN-β promoter luciferase reporter plasmids (pGL-chIFN-β-Luc), containing -158 to +14 of the chicken IFN-β promotor motif, were constructed as described in our previous study (37).

The CEFs, DF-1 cells and IRF7^−/− DF-1 cells were plated in 24-well plates at 5 × 10⁵/0.5 mL and were transiently transfected with the reporter plasmid pGL-chIFN-β-Luc (120 ng/well) and the control Renilla luciferase (pRL-TK, 60 ng/well); the cells were also transfected with the indicated plasmids using Nulen PlusTrans™ Transfection Reagent (Nulen, Shanghai, China). The cells were lysed 24 hours after transfection, and luciferase activity was detected using a Dual-Luciferase Reporter Assay System kit (Promega, USA) according to the manufacturer’s instructions. Renilla luciferase activity was used for normalization. All reporter assays were repeated at least three times.

In order to detect the expression of DDX1 mRNA in different chicken tissues, tissues from the brain, trachea, lungs, and 13 other types of tissues were collected from three specific pathogen free (SPF) chickens (Merial Vital Laboratory Animal Technology Company, China) at four weeks of age. All animal experiments were conducted in accordance with the animal welfare standards approved by the Ethical Committee for Animal Experiments of Shanghai Jiao Tong University, China. After the RNA was extracted from the chicken tissues or cells using an HP Total RNA kit (Omega, USA), the RNA was reverse transcribed to cDNA using a cDNA synthesis kit (Vazyme). The viral RNA in the supernatant is extracted by a TaKaRa MiniBEST Viral RNA/DNA Extraction Kit (TaKaRa, Japan). Quantitative real-time PCR (qRT-PCR) tests were conducted using the primers listed in Table 1 and an ABI 7500 RT-PCR system. The qRT-PCR was performed according to the manufacturer’s instructions using a ChamQ™ SYBR^® qPCR Master Mix (Vazyme). The conditions and data processing of the qRT-PCR were the same as those in our previous study (38).

Specific chDDX1 gene knockdown in DF-1 cells was conducted via pGPU6-based silencing (GenePharma, Shanghai, China). A negative control sequence that is not present in human or chicken genome databases and two RNA interference (RNAi) target sequences against chDDX1 were designed ( Table 2 ). The complementary short hairpin RNA (shRNA) template oligonucleotides were synthesized based on the RNAi sequences designed against chDDX1 and the negative control. Next, the complementary oligonucleotides were annealed and inserted into the shRNA expression vector pGPU6-neo. The recombinant shRNA plasmids were called shDDX1-1, shDDX1-2, and shNC, respectively. For silencing, the DF-1 cells were plated in 12-well or 24-well plates and transfected with shRNA at 1 µg/well or 0.5 µg/well using Nulen PlusTrans™ Transfection Reagent (Nulen). Forty-eight hours after transfection, the cells were used for the subsequent experiments.

The DF-1 cells were plated in 6-well plates at 2 × 10⁶/2 mL and then transfected with a total of 4 µg empty plasmid or various expression plasmids. Thirty-six hours later, the transfected cells were washed twice with phosphate buffer saline (PBS) (Gibco) and then lysed with a cell lysis buffer (Beyotime, Shanghai, China) containing a protease cocktail (Yeasen, Shanghai, China) and phenylmethylsulfonyl fluoride (PMSF) (Yeasen). The lysate was centrifuged at 13,000 rpm for 15 minutes to obtain the supernatant, and a 5 × SDS loading buffer was added before the lysates were boiled for ten minutes. The proteins isolated from the cell lysates were separated via SDS-PAGE and analyzed using western blotting. Images were obtained using the Tanon 5200 imaging system (Tanon, Shanghai, China), as described in our previous study (35).

The DF-1 cells were plated in 6-well plates and transfected with various plasmids. The transfected cells were harvested 24 hours after transfection and lysed in the same way as before the western blot analysis. The poly(I) and agarose beads coupled with poly(C) were resuspended in a buffer containing 50 mM Tris (pH = 7) and 150 mM NaCl to a final concentration of 2 mg/mL and mixed at a ratio of 2 to 1 by volume. The mixture was incubated at 56°C for 30 minutes; the temperature was then slowly reduced to 4°C to obtain poly(I:C)-coated beads. The poly(C) and poly(I:C) coupled agarose beads were incubated with cell lysates at 4°C for six hours on a low-speed rotating instrument (15 rpm/min). After the mixture was centrifuged at 6000 rpm for one minute, the supernatant was discarded, and the beads were washed three times with TBS. A loading buffer of 60 µL 1 × SDS was added to the beads, and this mixture was boiled for ten minutes and then centrifuged under the same conditions as the supernatant for the western blot analysis.

The DF-1 cells were plated, washed twice with PBS, and infected at 0.1 multiplicity of infection (MOI) with AIV, NDV, or VSV-GFP. The RNA from the cells, which was infected with the viruses at different times, was then collected for qRT-PCR. Poly(I:C) (Invivogen, San Diego, CA) was transfected into the DF-1 cells at 0.05 µg/well (in 24-well plates), and the cells were used for the subsequent experiments 12 hours after transfection.

The data were expressed as means ± standard deviations. Significance was determined using a two-tailed independent Student’s t test (*p < 0.05, **p < 0.01).

To better understand the role of chDDX1 in innate immunity in chickens, chDDX1 was cloned and studied using bioinformatics analysis. The open reading frame of chDDX1 contains 2223 bp and encodes 741 amino acid residues. The DDX1 amino acid sequences in humans (Homo sapiens, NP_004930.1), mice (Mus musculus, NP_598801.1), pigs (Sus scrofa, ANR02618.1), and ducks (Anas platyrhynchos, XP_027311055.1) are 93.2%, 93.1%, 93.5%, and 98.6% identical to those in chickens (Gallus gallus, NP_989894.1), respectively ( Figure 1A ). The protein domains of chDDX1 were predicted using SMART. The results show that chDDX1 consists of an N-terminal domain in SPla and the RYanodine receptor (SPRY) (aa 130-246) and a C-terminal helicase superfamily c-terminal (HELICc) domain (aa 520-610) ( Figure 1B ).

A phylogenetic tree was developed based on multiple alignments of DDX1 from various species, including fish, birds, and mammals ( Figure 2A ). The resulting phylogenetic tree consists of three major branches. The DDX1 protein sequences of chicken, finches (predicted), geese (predicted) and ducks (predicted) belong to one subgroup. The DDX1 of mammals, including humans, mice, rabbits (predicted), chimpanzees (predicted), baboons (predicted), cattle, horses (predicted), goats (predicted), dogs (predicted), pigs, bats (predicted), and cats (predicted) belong to another subgroup. Fish DDX1 sequences from zebrafish and salmon (predicted) belong to a third subgroup. These results reflect the genetic relationships among these species. The amino acid sequence homologies of different animals were conducted using MegAlign; the results are shown in Figure 2B . The predicted three-dimensional structures of chDDX1 are shown in Figure 2C .

The chicken tissues tested included brain, trachea, lung, spleen, bursa, heart, liver, esophagus, kidney, pancreatic gland, stomach (glandular and muscular stomach), intestine (small and large intestine), and muscle. The results showed that chDDX1 mRNA was widely expressed in most tissues analyzed. The highest levels were observed in the pancreatic gland; higher levels were detected in muscle, brain, trachea, heart, liver, and kidney tissue; moderate levels in the muscular stomach, bursa, lungs, spleen, esophagus, and small and large intestine; and low levels in the glandular stomach ( Figure 3A ).

In mammals, DDX1 is involved in the type I IFN-mediated antiviral innate immune response. However, the role of chDDX1 in the antiviral response remains unknown. For the host, upregulating the expression of certain immune-related genes is an important strategy in infection resistance. To determine whether chDDX1 could induce an antiviral response to infection with AIV, NDV, or VSV-GFP, we conducted a preliminary analysis of the expression of chDDX1 in DF-1 cells following infection with SH010 AIV, NSD14 NDV, and VSV-GFP. The results showed that the mRNA levels of chDDX1 were significantly upregulated during the early stages of viral infection ( Figures 3B–D ).

DDX1 has been reported to activate type I IFN signaling in mouse dendritic cells (39). To investigate whether chDDX1 is also involved in the regulation of IFNs production, DF-1 cells, an immortalized chicken embryonic fibroblast cell line, were cotransfected with chDDX1 expression plasmids and with chIFN-β luciferase reporter plasmids. We found that overexpression of chDDX1 in DF-1 cells activated the IFN-β promoter, and this activation correlated positively with the dosage of the transfected chDDX1 plasmids ( Figure 4A ). Twenty-four hours after transfection, chDDX1 had the strongest activation effect on chIFN-β ( Figure 4B ). To further confirm the ability of IFN activation of chDDX1, we prepared primary chick embryo fibroblasts (CEFs). And luciferase assays were conducted with CEFs. Similarly, the overexpression of chDDX1 in CEFs activated the IFN-β promoter in a dose-dependent manner ( Figure 4C ).

To further study the effect of chDDX1 on the expression of innate immune genes, qRT-PCR was conducted after chDDX1 or empty vector transfected into DF1 cells. Consistent with the luciferase assay, the expression of chIFN-β mRNA was significantly upregulated by chDDX1 transfection ( Figure 4D ). Besides IFN-β, the mRNA levels of IFN-α, IFN-λ, and the IFN-stimulated genes (ISGs) (PKR and Mx-1) were also upregulated by chDDX1 transfection ( Figures 4D, E ). It is noted that the proinflammatory cytokines (IL-6 and IL-8) were also upregulated in chDDX1 transfected DF1 cells ( Figure 4F ).

Next, the chDDX1-overexpressing and normal DF-1 cells were infected with VSV-GFP and NDV-GFP, respectively, and fluorescence was measured with a fluorescence microscope. The fluorescence intensities of both VSV-GFP and NDV-GFP in chDDX1 overexpression cells were significantly lower than that in the control DF-1 cells at all tested time points ( Figures 4G, H ). This result suggests that chDDX1 overexpression in DF-1 cells suppresses VSV-GFP and NDV-GFP viral replication. Then, the chDDX1-overexpressing and normal DF-1 cells were further infected with an unmodified moderate virulent NDV strain NSD14, and the viral RNAs inside the cells and in the supernatants were detected by qPCR, respectively. Results showed that both the viral loads inside the cells and in the supernatants were significantly lower in chDDX1 overexpressing cells than those in the empty vector transfected groups ( Figures 4I, J ). In summary, the overexpression of chDDX1 in DF-1 cells induced the expression of IFNs, ISGs, and proinflammatory cytokines and also inhibited VSV-GFP and NDV replications.

Based on the secondary structure of chDDX1 predicted by the SMART program, a series of truncated mutants of chDDX1 were generated. The ability of these mutants to activate the IFN-β promoter was measured and compared to that of wild-type chDDX1 with dual luciferase reporter assays ( Figures 5A, B ). As shown in Figure 5B , the deletion of 100 (d1-100 aa) or 250 (d1-250 aa) amino acids (aa) at the chDDX1 N-terminal slightly increased IFN-β promoter activity. The further deletion of 500 residues in the chDDX1 N-terminal (d1-500 aa) resulted a significant decrease of promoter activity. The C-terminal deletion mutant (d501-740 aa) and the mutant deleted 251-500 amino acids (d251-500 aa) also led to a weaker ability to activate IFN-β promoter. However, deletion of the SPRY domain (dSPRY) or the HELICc domain (dHELICc) from chDDX1 did not significantly affect its ability to activate the IFN-β promoter. The expressions of the truncated chDDX1 proteins were detected using western blot, and all the protein bands were consistent with their indicated sizes ( Figure 5C ). These results indicate that the first 250 amino acids of chDDX1 is possible a negative regulatory domain for chDDX1’s IFN activation, and the 250-740 aa of chDDX1 contains pivotal domains that required for activating IFN-β signaling. To further verify the IFN activation abilities and the antiviral functions of the truncations, the DF1 cells transfected p3×FLAG, p3×FLAG-chDDX1, chDDX1-d1-100aa and chDDX1-d1-500aa were infected with NDV-GFP and fluorescence was measured. Consistently with the result of the IFN activation in the Figure 5B , the chDDX1-d1-100aa transfected cells showed a slight stronger inhibitory effects, while the chDDX1-d1-500aa cells exhibited an obvious weaker inhibitory effects on the NDV-GFP replication, when compared with the wild type chDDX1 transfected cells ( Figure 5D ).

To further investigate the role of chDDX1 in the chIFN-β signaling pathway in chicken cells, we constructed RNAi plasmids of chDDX1 (shDDX1-1, shDDX1-2, and shNC). To verify the specific knockdown target, the DF-1 cells were transfected with the shRNA plasmids and then transfected with expression plasmids 36 hours later. Twenty-four hours after that, the cells were harvested for western blot analysis. The results indicated that both shDDX1-1 and shDDX1-2 substantially reduced the expression of chDDX1. The results of the western blot and gray analyses are shown in Figure 6A . To verify whether knockdown of chDDXI increases virus replication, chDDX1 knockdown DF-1 cells were infected with VSV-GFP and NDV-GFP, and real-time fluorescence images were used to measure virus proliferation. The results showed that the virus replications of both VSV-GFP and NDV-GFP were faster in chDDX1 knockdown DF-1 cells than those in the control group ( Figures 6B, C ). Then, the chDDX1 knockdown (shDDX1-2) and normal DF-1 cells were infected with the NSD14 NDV and the viral RNAs inside the cells and in the supernatants were extracted separately for qPCR. The results showed that the virus copies both inside the cells ( Figure 6E ) and in the supernatants ( Figure 6D ) from chDDX1 knockdown cells were obviously higher than those in the control group. The chDDX1 knockdown cells transfected with luciferase reporter plasmids were then incubated with poly(I:C), AIV, or NDV, and luciferase reporter assays were performed. The results demonstrated that knocking down chDDX1 led to inhibited IFN-β promoter activation after the DF-1 cells were stimulated with poly(I:C) or infected with a virus ( Figures 6F–H ). These data showed that SH010 AIV, NSD14 NDV, and VSV-GFP induce chIFN-β via a chDDX1-dependent pathway and that chDDX1 may play an essential role in the antiviral innate immunity mediated by IFN-β in DF-1 cells.

To further explore the signal pathway through which chDDX1 activates IFN-β, DF-1 cells were transfected with a chDDX1 plasmid and an empty plasmid. Twenty-four hours after transfection, RNA was extracted for a qRT-PCR test, which showed that the transcription level of IRF7 was significantly upregulated ( Figure 7A ). After we endogenously knocked down the chDDX1, chIRF7 was significantly downregulated ( Figure 7B ). More critically, when chDDX1 was overexpressed in IRF7^-/- DF-1 cell lines, chIFN-β could not be activated ( Figure 7C ). Therefore, chIRF7 plays a critical role in the regulation of chIFN-β by chDDX1, and chDDX1 activates IFN signaling via the chIRF7 pathway.

Some members of the DExD/H-box helicases family have proven to be extremely important antiviral PRRs in mammals. It is unclear whether chDDX1 plays a similar function in innate immunity in chickens. In-vitro pulldown assays were used to explore the specific functions of chDDX1 in the IFN regulatory pathway. We found that chDDX1 bound to poly(I:C), a simulant of dsRNA, but not to poly(C), a simulant of ssRNA ( Figures 8A, B ). The in-vitro pulldown assay showed a strong and direct interaction between poly(I:C) and the chDDX1 protein, indicating that chDDX1 may act as an RNA PRR during IFN activation.