Epithelioma papulosum cyprini (EPC) cells and grass carp (Ctenopharyngodon idellus) ovary (CO) cells were cultured in medium 199 basic (Gibco) supplemented with 10% fetal bovine serum (FBS) at 28°C, and HEK293T cells were in DMEM basic (Gibco) with 10% FBS at 37°C in a humidified incubator containing 5% CO₂. Spring viraemia of carp virus (SVCV) and grass carp reovirus (GCRV) were propagated in EPC cells and CO cells, respectively, and the titer was determined with a 50% tissue culture infective dose (TCID₅₀) assay according to Reed and Much methods. Zebrafish (Danio rerio) strain AB were raised according to standard protocols, which was approved by the Animal Care and Use Committee of Institute of Hydrobiology, Chinese Academy of Sciences. For viral infection, zebrafish adults (2-month-old) were injected i.p. (intraperitoneally injection) with SVCV.

The open reading frames (ORF) of 11 zebrafish IRF family genes (IRF1: NM_205747.1; IRF2: NM_001328374; IRF3: NM_001143904; IRF4b: XM_021468630; IRF5: NM_001327817; IRF6: NM_200598; IRF7: NM_200677; IRF8: NM_001002622; IRF9: NM_205710; IRF10: NM_212879; IRF11: NM_001040352) were inserted into the Nhe I/Kpn I site of pEGFP-N3 to construct expression plasmids expressing respective IRF protein that was fused to GFP (IRF1/2/3/4b/5/6/7/8/9/10/11-wt). 11 zebrafish IRF mutants, including IRF1-mut (K78A/R82A), IRF2-mut (K77A/R81A), IRF3-mut (K76A/R80A), IRF4b-mut (K88A/R92A), IRF5-mut (K80A/R84A), IRF6-mut (K80A/R84A), IRF7-mut (K75A/R79A), IRF8-mut (K77A/R81A), IRF9-mut (K81A/R85A), IRF10-mut (K80A/R84A) and IRF11-mut (K78A/R82A), were obtained by mutation of the indicated two residues to alanine based on the wild type plasmids. Promoter-driven luciferase constructs including CaIFNpro-luc, DrIFNφ1pro-luc, DrIFNφ3pro-luc, and ISRE-luc, were described previously (31, 32, 34).

Plasmid transfection was carried out by Polyethylenimine Linear (PEI, MW25000, Aldrich, USA) according to the manufacturer’s protocol or our previous reports (31, 36, 45). Luciferase assays were performed as described previously (31, 36). Briefly, EPC cells seeded in 48-well plates overnight were transfected with various plasmids at a ratio of 1:10:10 (pRL-TK, promoter-driven luciferase reporter plasmid, expression construct). If needed, the cells were simultaneously transfected with poly(I:C). 30h later, the cells were collected for detecting luciferase activities in a Junior LB 9509 luminometer (Berthold, Pforzheim, Germany) using Dual-Luciferase Reporter Assay System (Promega). The relative expression values of luciferase activities were normalized to the amounts of Renilla luciferase activities according to the protocol. The results were the representative of at least three independent experiments, each performed in triplicate.

SVCV-injected zebrafish were sampled for different tissues, followed by extraction of total RNAs using TRIZOL Reagent (AIDLAB, China). First-strand cDNA was synthesized using random primers (AIDLAB, China). RT-qPCR was performed with Universal Blue qPCR SYBR Green Master Mmix (YEASEN, China) in a DNA Engine Chromo 4 real-time system (BioRad, USA). Gene expression values were normalized to β-actin in a same sample. The results were the representative of at least three independent experiments, each performed in triplicate. The primers used in this study were listed in Table 1.

Subcellular localization was determined in HEK293T cells by GFP reporter assays (31). Briefly, HEK293T cells plated overnight on microscopic cover glasses in six-well plates were transfected with plasmids expressing respective zebrafish IRF fused to GFP. 24 h later, the transfected cells were washed three times with PBS, fixed with 4% (v/v) paraformaldehyde at room temperature for 30 min, washed again with PBS three times, incubated with 0.2% Triton X-100 for 15 min, and finally stained with DAPI (500µg/ml) for 30 min. Cells were examined by a confocal microscope [ZEN Blue Lite confocal system. Objectives: ×40; analysis software: ZEN 2.3 (blue edition)].

Subcellular localization of the endogenous IRF3 and IRF7 protein in CO cells was determined by immunofluorescence analyses. Briefly, CO cells plated overnight on microscopic cover glasses in six-well plates transfected with zebrafish TBK1 (2μg), or infected with SVCV or GCRV for 24h. After fixed with paraformaldehyde and permeabilized with Triton X-100, the cells were blocked with 10% ADB (10% goat serum、3% BSA, 0.05% Triton X-100) at 37°C for 1 h, incubated overnight with the first antibody of IRF3 and IRF7 at 4°C and then with Alexa Fluor Plus 555 TRITC (Invitrogen) fluorescent second antibody and DAPI in the dark at 37°C for 1 h, followed by confocal microscope observation. The antibodies specific to crucian carp IRF3 and IRF7 was made by immunizing white rabbit with prokaryotic expressed DBD domains of IRF3/7 as described previously (31, 35).

DNA pull-down assays were performed as described previously (28, 34, 35). Briefly, HEK293T cells inoculated overnight in 10cm petri dishes were transfected with plasmids expressing GFP-fused IRF. 24h later, the cells were harvested in HKMG buffer with protease inhibitor (Roche) and further lysed by ultrasonication. One-tenth of cell lysates were taken as input. Appropriate amounts of cell lysates were incubated at 4°C with 15 μl M-280 streptavidin Dynabeads (Invitrogen), and biotinylated promoter DNA probes (20ng). The cell lysates should contain excessive overexpressed-IRF proteins to ensure the completed binding to DNA at a constant amount. Another 24h later, the beads were washed with HKMG buffer five times, followed by western blotting analysis of promoter-bound IRF protein with the antibody specific to GFP (Dia-an Biological Technology, Wuhan, China). The results were the representative of at least three independent experiments.

Student’s t-test is applied for statistical analysis of the data derived from luciferase assays and RT-PCR assays.

By analyzing the genome data of zebrafish, we cloned a total of 11 IRF genes (IRF1-11) from zebrafish. There are two copies of IRF4 (IRF4a and IRF4b) reported in zebrafish genome (46). Zebrafish IRF4a is essential in T lymphoid-primed progenitors (47). In the study, IRF4b was cloned for further assays. Expression comparison showed that IRF1/2/3/7/8/9/10 were significantly upregulated by SVCV infection in all five zebrafish tissues (gill, spleen, head kidney, body kidney and liver), and IRF4b/5/6/11 did not to respond to viral infection in certain tissues, such as IRF6 in liver, IRF5 and IRF11 in spleen, body kidney and liver (Figure 1A). Of 11 zebrafish genes, IRF4b displayed the weakest induction in zebrafish tissues except for liver (Figures 1A, B). Under the same conditions, two zebrafish IFN (IFNφ1 and IFNφ3) and Mx genes were markedly induced (Figure 1C). These results indicated that zebrafish IRF family genes were differentially induced by SVCV infection when IFN response was activated.

Luciferase assays revealed that overexpression of IRF1/3/5/6/7/8/9/11 significantly stimulated crucian carp IFN promoter-driven luciferase activity (CaIFNpro-luc), albeit to different degrees (Figure 2A). Instead, overexpression of IRF2/4b/10 did not show any stimulation (Figure 2A) but resulted in a decreased luciferase activity triggered by poly(I:C) transfection (Figure 2B). Zebrafish has four IFN genes, which are divided into group I and group II subfamilies, being represented by IFNφ1 and IFNφ3 (32). Crucian carp IFN and zebrafish IFNφ1 belong to group I IFN subfamily (48). Therefore, when IFNφ1pro-luc was used to replace CaIFNpro-luc, overexpression of IRF1/3/5/6/7/8/9/11 or IRF2/4b/10 yielded similar results (Figure 2C). Compared to IFNφ1pro-luc, IFNφ3pro-luc was activated only by IRF1/7/11, but not by IRF3/5/6/8 (Figure 2D), indicating that zebrafish IFNφ1 and IFNφ3 are transcriptionally regulated by different IRF family members.

Similarly, the IFN activators IRF1/3/5/6/7/8/9/11 were capable to activate ISRE-containing promoters (ISRE-luc) (Figure 2E), and IRF2/4b/10 downregulated poly(I:C)-triggered activity of ISRE-luc (Figure 2F). Notably, despite inability to stimulate fish IFN promoters (Figure 2A), overexpression of IRF4b alone yielded 10-fold luciferase activity of ISRE-luc over the control (Figures 2E, F). These results together indicate that zebrafish IRF members have the ability to regulate the expression of IFN genes and ISGs, with some (IRF1/3/5/6/7/8/9/11) functioning as positive regulators of IFN gene transcription, and others (such as IRF2/4b/10) as negative regulators. In addition, IRF3/5/6/8 differentially activates different fish IFN gene expression.

There are three ISRE/IRF-E motifs within IFNφ1 promoter and two in IFNφ3 promoter (Figure 3A), which are the binding sites of DBDs of crucian carp IRF3 and zebrafish IRF1/3/7/11 (28, 34, 35). We wondered whether this might be the case for the other zebrafish IFN proteins. To this end, DNA pull-down assays were performed by incubation of biotin-labeled promoter DNAs of IFNφ1 (-586 to +38) or IFNφ3 (-1447 to -912), with IRF proteins that were overexpressed in HEK293T cells. Western blot analysis of promoter-bound products showed that all zebrafish IRF proteins effectively bound to IFNφ1/3 promoters (Figure 3B). Of 11 zebrafish IFN proteins, IRF4b and IRF8 displayed the weakest binding to two fish promoter DNA, followed by IRF9.

Transcription factors have to translocate to nucleus first for their function; therefore, it is of great interest to investigate the subcellular localization of zebrafish IRF family proteins. To this end, the ORFs of IRF genes were fused to GFP for transfection of HEK293T cell. Confocal microscopy examination showed that, distinct from the control GFP protein that was expressed ubiquitously in the whole cells (Figure 4A), 11 zebrafish IRF proteins exhibited three patterns of subcellular localization (Figures 4B–D). The first is nuclear accumulation, including IRF1/2/9/11 (Figure 4B). The second is cytoplasmic accumulation, including IRF3/5/7 (Figure 4C). And the third contains IRF4b/6/8/10 accumulating largely in nucleus and partially in cytoplasm (Figure 4D). Quantitation of nuclear/cytoplasmic GFP intensities supported the conclusions above (Figure 4E).

Similar to previous results (28, 30) zebrafish IRF family can be divided into four subfamilies: IRF1 subfamily (IRF1/2/11), IRF3 subfamily (IRF3/7), IRF4 subfamily (IRF4/8/9/10) and IRF5 subfamily (IRF5/6) (Figure 5A). Multiple alignments showed zebrafish IRF proteins displays a high level of homology in DBDs and a weak one in C-terminal IAD) (Figure 5B). Based on human IRF1 structure (5), the DBDs should form a conserved α/β architecture, with a cluster of three α-helices (α1–α3) flanked on one side by a mixed four-stranded β-sheets (β1–β4) (Figure 5B).

We recently revealed that zebrafish IRF11 constitutively localizes to nucleus, exclusively dependent on a function-integrated tripartite NLS motif, which is composed of 3 aa stretches: NLS1 (aa75-82) and NLS2 (aa97–103) within DBD, and NLS3 (aa119–133) immediately adjacent to the C-terminal DBD (28) (Figure 5B). Notably, the regions corresponding to IRF11 NLS1, particularly two basic residues K78 and R82, are conserved across zebrafish and human IRF family, and instead, the regions corresponding to the NLS2 and NLS3 are less conserved (Figure 5B). Since K78 and R82 of zebrafish IRF11 contribute to its DNA-binding and nuclear import (28), we hypothesized that both amino acids have similar roles in other IRF family proteins. To this end, we generated IRF mutants by collective mutation of both basic residues to alanine (Figure 5C), to further characterize the DNA binding and nuclear retention features of zebrafish IRF family members.

IRF1 subfamily is composed of IRF1, IRF2 and IRF11. Compared to wild type IRF1/2/11 that exclusively localized to nucleus, collective mutation of both conserved amino acids to alanine (IRF1-mut, IRF2-mut and IRF11-mut) significantly altered their constitutive localization patterns, resulting in partially cytoplasmic localization (Figure 6A). Pull-down assays revealed that either IRF2-mut or IRF11-mut largely lost the binding affinity to DrIFNφ1 promoter DNA and slightly to DrIFNφ3 promoter DNA, but IRF1-mut appeared to have equal promoter binding affinity to the wild type IRF1 (IRF1-wt) (Figure 6B). Luciferase assays showed that, compared to the wild type IRFs, IRF1-mut and IRF11-mut failed to activate CaIFNpro-luc (Figure 6C), DrIFNφ1pro-luc and DrIFNφ3pro-luc (Figure 6D). Similarly, IRF2-mut completely lost the ability to downregulate the luciferase activity of CaIFNpro-luc triggered by poly(I:C) (Figure 6E). These results together indicated that mutation of both conserved amino acids of α3 helices in DBD domains resulted in a completely functional loss of three zebrafish IRF1 subfamily members, likely through impairing nuclear retention and DNA binding (IRF2 and IRF11), or only through impairing nuclear retention (IRF1).

IRF3 subfamily contains IRF3 and IRF7. In contrast to IRF1 subfamily members, wild type IRF3 and IRF7 (IRF3-wt and IRF7-wt) constitutively resided in cytoplasm (Figure 7A). Interestingly, combined mutation of both basic residues did not make any influence in constitutively cytoplasmic retention of IRF3 and IRF7 (Figure 7A), because both residues are essential for nuclear import but not nuclear export of zebrafish IRF11 (28). Compared to IRF3-wt and IRF7-wt that efficiently bound to both fish IFN promoters, IRF3-mut completely lost the DNA binding affinity, but no alteration was seen for IRF7-mut (Figure 7B). IRF3-wt mainly regulates the expression of group I IFN (crucian carp IFN and zebrafish IFNφ1) but not group II IFN (zebrafish IFNφ3) (Figures 7C, D). IRF7-wt has a strong stimulatory potential to zebrafish IFNφ3 and a weak one to zebrafish IFNφ1 (Figure 7D). Despite of these difference, combined mutation of both basic residues (IRF3-mut and IRF7-mut) completely abrogated their stimulatory potential to fish promoters (Figures 7C, D).

IRF5 subfamily is composed of IRF5 and IRF6. IRF5-wt was constitutively expressed in cytoplasm, and the same is true for IRF5-mut (Figure 8A, upper panel). Distinct from IRF5-wt, IRF6-wt largely accumulated in nucleus and slightly in cytoplasm; on the contrary, IRF6-mut was largely expressed in cytoplasm and partially in nucleus (Figure 8A, lower panel). These results indicated that mutation of both conserved amino acids have made a significant influence on the constitutively subcellular localization of IRF6 but not of IRF5. Regarding the promoter binding activity, IRF5-mut and IRF6-mut showed a weak or no binding to IFNφ1 promoter DNA, and a medium one to IFNφ3 promoter DNA compared to their wild types (Figure 8B). Despite that IRF5-wt and IRF6-wt effectively activated crucian carp IFN promoter, overexpression of IRF5-mut and IRF6-mut gave a basal luciferase activity, comparable to the control (Figure 8C). IRF5-wt and IRF6-wt selectively activated zebrafish IFNφ1 promoter but not zebrafish IFNφ3 promoter; similarly, IRF5-mut and IRF6-mut completely lost the ability to stimulate IFNφ1 promoters (Figure 8D).

IRF4 subfamily consists of IRF4, IRF8, IRF9 and IRF10, showing two distinct patterns of constitutively subcellular localization, only in nucleus (IRF9), and most in nucleus with a little in cytoplasm (IRF4b/8/10) (Figure 9A). Regardless of which localization pattern they possessed, mutation of both basic residues significantly changed their constitutive localization, with much more proteins residing in cytoplasm (Figure 9A). Similar to mammalian counterparts (12, 49), zebrafish IRF4b and IRF8 exhibited a weaker binding to both zebrafish promoters than IRF9 and IRF10 (Figure 9B).Mutation of both amino acids in IRF4b/8/9 (IRF4b-mut, IRF8-mut and IRF9-mut) but not in IRF10 (IRF10-mut) severely impaired their binding to zebrafish IFNφ1 promoter, but no significant change was seen to zebrafish IFNφ3 promoter except for IRF9-mut (Figure 9B). Compared to the wild types, IRF8-mut and IRF9-mut did not activate crucian carp IFN promoter anymore (Figure 9C), and IRF4b-mut and IRF10-mut failed to downregulate poly(I:C)-triggered IFN promoter activation (Figure 9D). IRF8-wt and IRF9-wt obviously activated zebrafish IFNφ1 promoter but not zebrafish IFNφ3 promoter; consistently, IRF8-mut and IRF9-mut completely lost their immune-stimulatory potential (Figure 9E).

As mentioned above, mutation of both basic residues did not change the constitutively cytoplasmic retention of zebrafish IRF3/7 (Figure 7A) and IRF5 (Figure 9A). We wondered whether mutation of both basic residues impaired the induced nuclear import of IRF3/5/7. As shown in Figure 10A, in response to 24h-poly(I:C) transfection, less than 1% overexpressed cells showed a subcellular translocation of IRF3/5/7-wt from cytoplasm to nucleus. However, similar nuclear import was not observed for IRF3/5/7-mut under the same condition (Figure 10A).

To corroborate the findings, we replicated the assays using TBK1 instead of poly(I:C), since TBK1 is the protein kinase specific to IRF3/5/7 activation (8, 31, 32). A time-course observation showed that overexpression of TBK1 facilitated the cytoplasmic to nuclear accumulation of IRF3/5/7-wt in a time-dependent manner, but not of IRF3/5/7-mut by statistic counting of cells that had undergone nuclear import (Figure 10B). When TBK1 was overexpressed together, the green fluorescence signal that was spatially homogenous in cytosol was generally condensed, and finally converged to the peripheral nuclear membrane; however, only wild types of IRF3/5/7 rather than the mutants were observed to translocate from cytoplasm to nucleus (Figure 10B). These results indicate that although mutation of both basic residues does not alter the cytoplasmic accumulation of IRF3/5/7 in resting cells, but significantly impairs the nuclear import of IRF3/5/7 following poly(I:C)- or TBK1-transfection.

The inducible nuclear import of endogenous IRF3 and IRF7 proteins were next determined in CO cells stimulated by TBK1 or poly(I:C) transfection, or by GCRV or SVCV transfection (Figure 10C). Using the polyclonal antibodies specific to fish IRF3 and IRF7, immunofluorescence microscopy showed that in resting cells (IRF3-mock and IRF7-mock), a red signal was weakly observed in cytoplasm, indicating that the constitutively expressed IRF3 and IRF7 proteins resided in cytoplasm. However, in TBK1-overexpressed cells (IRF3-TBK1, IRF7-TBK1), a strong red signal was highlighted, and some obviously located to nucleus, indicating that TBK1 induced the expression of endogenous IRF3 and IRF7 proteins, some of which in turn translocated from cytoplasm to nucleus. Similar results were seen in poly(I:C)-transfected cells (IRF3-poly(I:C), IRF7-poly(I:C)), SVCV-infected cells (IRF3-SVCV, IRF7-SVCV) and GCRV-infected cells (IRF3-GCRV, IRF7-GCRV) (Figure 10C). Pulldown assays showed that zebrafish IFNφ1 and IFNφ3 promoter DNA could bind to the endogenous IRF3 and IRF7 proteins derived from CO cells that were transfected with empty vector or with TBK1 (Figure 10D). It is noted that the low binding intensities in empty vector-transfected CO cells might be due to the low level of constitutive expressed IRF3 and IRF7, which finally resulted in unsaturated protein binding to DNA in the pulldown assays.