Dynamic and Differential Expression of Duplicated Cxcr4/Cxcl12 Genes Facilitates Antiviral Response in Hexaploid Gibel Carp

Chemokine receptor cxcr4 and its ligand cxcl12 have evolved two paralogs in the teleost lineage. In this study, we have identified four duplicated cxcr4 and cxcl12 genes from hexaploid gibel carp, Carassius gibelio, respectively. Cgcxcr4bs and Cgcxcl12as were dynamically and differentially expressed in immune-related tissues, and significantly up-regulated in head kidney and spleen after crucian carp herpesvirus (CaHV) infection. Blocking Cxcr4/Cxcl12 axis by injecting AMD3100 brought more severe bleeding symptom and lower survival rate in CaHV-infected fish. AMD3100 treatment also suppressed the up-regulation of key antiviral genes in head kidney and spleen, and resulted in more acute replication of CaHV in vivo. Consistently, the similar suppression of up-regulated expression of key antiviral genes were also observed in CAB cells treated by AMD3100 after poly(I:C) stimulation. Finally, MAPK3 and JAK/STAT were identified as the possible pathways that CgCxcr4s and CgCxcl12s participate in to promote the antiviral response in vitro.

Gibel carp (Carassius gibelio) (3n = 6x = ∼150 chromosomes) has been considered to undergo two extra polyploidy events (an early allopolyploidy followed by a recent autopolyploidy) in addition to the three rounds of whole genome duplication (WGD) shared by all teleost fishes (25)(26)(27)(28)(29). As a recurring polyploidy animal, it is an ideal system to investigate divergent expression and function of duplicated genes through WGD. Since 1980s, gibel carp has become an important aquaculture species in China (30). Along with the application of several improved varieties with excellent growth performance, such as allogynogenetic gibel carp "CAS III" (clone A + ) and "CAS V" (clone F), the current annual production capacity has increased to about 3 million tons per year in China (31)(32)(33). However, due to the highly intensive culture, outbreak of epizootic crucian carp (Carassius auratus) herpesvirus (CaHV) has led to enormous economic losses in main culture areas of Jiangsu province since 2012 (34)(35)(36). It is an urgent need to breed novel variety with CaHV resistance. In our previous studies, IFN system genes and immunoglobulin genes were found to play vital roles in defensing CaHV (32,37,38). Especially, some genes annotated as cxcl12 and cxcr4 were identified from the comparative transcriptomes between the diseased and control individuals (32), suggesting the antiviral role in response to CaHV infection. In this study, we first analyzed the diversification, evolution and biased expression pattern of cxcr4/cxcl12 genes in hexaploid gibel carp. Then, we explored the role of Cxcr4/Cxcl12 axis in the antiviral immune response by blocking CXCR4 with AMD3100 and revealed the related signaling pathways.

CaHV Infection and Sample Collection
Six-month-old individuals of gibel carp clone F with 67.68 (±2.16) g average weight were collected from the GuanQiao Experimental Station, Institute of Hydrobiology, Chinese Academy of Sciences. Ten tissues (spleen, head kidney, kidney, thymus, gill, intestine, brain, heart, muscle and liver) from three individuals were collected for quantitative reverse transcription PCR (qPCR) analysis before CaHV infection. The CaHV challenge experiments were performed as previously described (32,39). In brief, sixty individuals were randomly divided into two groups, and were intraperitoneally injected with 500 µl CaHV viral suspension (2.915 × 10 8 virus particles) or PBS per fish in the experimental group or in control group respectively. Head kidney and spleen were collected from three control (0 day) and infected individuals at 1, 3 and 5 days post injection (dpi), respectively. All samples were preserved in RNAlater (Qiagen, Dusseldorf, Germany) and stored at −20 • C for nucleic acid extraction. All experiments were performed in triplicate and the results were representative of three independent experiments.
Healthy individuals were gradually acclimatized with aerated water at 24 (±1) • C for 2 weeks before infection, and fed twice a day. After deep and overdosed anesthesia with styrylpyridine (30-50 mg/L; Aladdin, Shanghai, China), the fish were euthanized by immediately cutting off the spinal cord adjacent to the head. All procedures in this study were approved by the Institutional Animal Care and Use Committee of Institute of Hydrobiology, Chinese Academy of Sciences (protocol number 2016-018).

Sequence and Phylogenetic Analyses
The complete cDNA sequences of gibel carp cxcr4s and cxcl12s (GenBank accession numbers MT330400, MT330401, MT330402, MT330403, MT330404, MT330405, MT330406, and MT330407) were amplified by 3 ′ and 5 ′ RACE using SMARTer R RACE 5 ′ /3 ′ Kit (Clontech, San Francisco, USA) from head kidneys of diseased fish after CaHV infection. Amino acid sequences and transmembrane-domain were predicted by ORF Finder (https://www.ncbi.nlm.nih.gov/orffinder/) and SMART (http://smart.embl-heidelberg.de/), respectively. To analyze the evolution of vertebrate Cxcr4 and Cxcl12, the sequence information of 7 species were obtained from the Ensembl Genome browser (http://asia.ensembl.org/index.html) and National Center for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov/). Multiple protein sequences were aligned by Clustal W program and the phylogenetic tree was constructed by bootstrap analysis (1000 replicates) using the neighbor-joining method (NJ) in MEGA 7.0 software. The exonintron structure prediction was made by mRNA and genomic sequences alignment of cxcr4s and cxcl12s. The conserved synteny was analyzed among the chromosomal regions around cxcr4s and cxcl12s genes in human (Homo sapiens), chicken (Gallus gallus), spotted gar (Lepisosteus oculatus), zebrafish, crucian carp and gibel carp.

AMD3100 Treatment in vivo
One hundred twenty gibel carp individuals were randomly divided into four groups. The healthy and CaHV-infected individuals intraperitoneally injected with 200 µl PBS were considered as negative and positive control group (NC and PC), respectively. In the two experimental groups (E1 and E2), the healthy and CaHV-infected individuals were intraperitoneally injected with CXCR4 antagonist AMD3100 (MedChemExpress, New Jersey, USA) (100 µg in 200 µl PBS). Briefly, PBS or AMD3100 was injected per day until the fish died, and CaHV was injected only once at the 24 h after the first injection of PBS or AMD3100 in PC and E2 group, while NC and E1 group injected PBS instead. The procedure of CaHV infection was described above. After CaHV challenge, the individuals in four groups were monitored every 24 h to score mortality. Head kidney and spleen were collected from individuals at 3dpi to extract RNA for subsequent qPCR assays. The experiments were performed in triplicate.

RNA Extraction and qPCR
RNA extraction and qPCR were performed as previously described (37,38). Total RNAs from tissues were extracted using SV Total RNA isolation System (Promega, Madison, USA) according to the manufacturer's protocol. The firststrand cDNAs were synthesized in a 20 µL reaction volume following the protocol of GoScript TM Reverse Transcription System (Promega, Madison, USA). qPCR was performed on a CFX96TM Real-Time PCR System (Bio-Rad, California, USA) using an iTaq TM Universal SYBR R Green Supermix (Bio-Rad, California, USA). Primers used for qPCR analysis were designed with Oligo Calc (Oligonucleotide Properties Calculator) (http:// biotools.nubic.northwestern.edu/OligoCalc.html) and listed in Supplementary Table 1. The specificity of each pair of primers was confirmed by sequencing. All samples were analyzed in triplicate, and relative expression levels of target genes were calculated using the 2 − CT method. The optimal reference gene, eukaryotic translation elongation factor 1 alpha 1, like 1 (eef1a1l1) (M value = 0.74 < 1.5) was selected as the normalizer for qPCR (37).

AMD3100 Treatment in vitro
Crucian carp (C. auratus) blastula embryonic (CAB) cells were used to perform the Cxcr4 inhibition experiments with a final AMD3100 concentration of 10 µM. Controls were treated with same volumes of ethanol. Cells after 24 h treatment were harvested to extract RNA for subsequent qPCR assays. The experiments were also performed in triplicate and the results were representative of three independent experiments.
For the control, the expression plasmid was replaced by empty vector pCS2+. At 24 h post-transfection, the cells were treated again with poly(I:C) infection and the controls were added serum-free medium instead. After 48 h transfection, the cells were harvested and lysed according to the Dual-Luciferase Reporter Assay System (Promega, Madison, USA). Luciferase activities were measured by a Junior LB9509 luminometer (Berthold, Pforzheim, Germany) and normalized to the amounts of Renilla luciferase activities. All experiments were performed in triplicate and the results were representative of three independent experiments.

Western Blotting Assay
CAB cells were seeded in 6-well plates overnight and treated with ethanol (as a control) and AMD3100 for 24 h. Then, the cells were transfected with 1 µg/ml poly(I:C) or added equal volume serum free medium. After 24 h stimulation, the cell were collected and boiled together with SDS-PAGE protein loading buffer (Beyotime, Wuhan, China) for 8 min, and then analyzed by western blotting using crucian carp IRF3 (CaIRF3) polyclonal Ab (40), zebrafish STAT1 polyclonal Ab (HuaAn, Hangzhou, China) and zebrafish IRF7 monoclonal Ab (HuaAn, Hangzhou, China). The anti-tubulin was purchased from ABclone. To reveal the pathways, CAB cells were seeded in 6-well plates overnight and transfected with different CgCxcl12s and CgCxcr4s combinations. After 24 h transfection, the cells were collected as above described, and analyzed by western blotting using p-JAK2 monoclonal Ab (Cell Signaling Technology, Danvers, USA) and p-MAPK monoclonal Ab (Cell Signaling Technology, Danvers, USA).

Statistical Analysis
qPCR and luciferase activity assay data were shown as means ± SD of three independent experiments, and each performed in triplicate. The statistical analysis was performed by SPSS software (SPSS Inc.) using one-way ANOVA. A probability (p) < 0.05 was considered statistically significant ( * ), and p < 0.01 was considered extremely significant ( * * ).

Molecular Characterization and Evolution of Duplicated cxcr4/cxcl12 Genes
Four cxcr4 genes were identified in gibel carp (Supplementary Figure 1). Based on the identities, we supposed that gibel carp has two groups of paralogues (cxcr4a and cxcr4b) (caused by the Ts3R), and each of them contains two homoeologues (caused by the allopolyploidy), so they were named as Cgcxcr4a-A, Cgcxcr4a-B, Cgcxcr4b-A and Cgcxcr4b-B respectively. Different from three alleles (≥99.00% identity) of Cgbmp15s or Cgnanos2s (46,47), only one sequence was identified for each homoeologue, which indicates that the alleles of each Cgcxcr4 gene may be identical. The average identity between homoeologues was 89.65 ± 0.75%, while the average identity between paralogs was about 71.5 ± 0.2% (Supplementary Table 2). The SMART and TMHMM analyses revealed that all CgCxcr4s possessed 7 hydrophobic transmembrane (TM) domains, four extracellular loops (ECL), four intracellular loops (ICL), an extracellular N-terminus and an intracellular C-terminus (Supplementary Figure 3). The major differences existed in the N-terminus and the ECL3 (Supplementary Figure 3). There were also four cxcr4 genes in tetraploid crucian carp (2n = 4x = 100 chromosomes), and their identities to the corresponding genes of gibel carp ranged from 97.5 to 100% (Supplementary Table 2). Same as cxcr4s, four cxcl12 genes were also identified from gibel carp and crucian carp, respectively (Supplementary Figures 2, 3).
Consistent with the accepted species phylogeny, CgCxcr4as and CgCxcr4bs were first clustered with CaCxcr4as and CaCxcr4bs respectively, then clustered with common carp (Cyprinus carpio) and zebrafish Cxcr4a and Cxcr4b, respectively ( Figure 1A). Teleost Cxcr4a and Cxcr4b branch were grouped together and then clustered with spot gar Cxcr4, while tetrapod CXCR4s and coelacanth Cxcr4 were clustered into another branch ( Figure 1A). Similar phylogenetic tree of vertebrate Cxcl12 was shown in Figure 2A. The phylogenies of Cxcr4/Cxcl12 confirmed the assumption that both gibel carp and crucian carp are originated from a common allotetraploid ancestor (27).

Dynamic and Differential Expression Patterns of cxcr4s and cxcl12s
Cgcxcr4s were abundantly expressed in the immune-related tissues, such as spleen, kidney, head kidney and thymus ( Figure 3A). Except gill, Cgcxcr4bs showed remarkably higher expression than Cgcxcr4as in the analyzed tissues, suggesting that Cgcxcr4bs may play a vital role in immune regulation. Cxcr4 homoeologues also exhibited divergent expression patterns. Cgcxcr4b-A was expressed significantly higher than Cgcxcr4b-B in most analyzed tissues, while Cgcxcr4a-A and Cgcxcr4a-B showed well-matched expression bias in different tissues. More abundant Cgcxcr4a-A transcripts than Cgcxcr4a-B were detected in head kidney, kidney, gill and heart, while reversely differential expression patterns were observed in spleen, thymus, and intestine ( Figure 3A). Consistent with the specific chemotactic interaction of grouper Cxcr4b-Cxcl12a (48), Cgcxcl12a was the mainly expressed paralog in most analyzed immune-related tissues, as well as Cgcxcr4b (Figure 3B). Except in intestine and brain, Cgcxcl12a-B showed significantly higher expression than Cgcxcl12a-A ( Figure 3B). These results indicate that the duplicated cxcl12 genes also exhibit a differential expression pattern in gibel carp.
Dynamic expression changes of Cgcxcr4s or Cgcxcl12s in gibel carp after CaHV infection were also evaluated by qPCR (Figures 3C-E). At 1 dpi in head kidney, Cgcxcr4b-A and Cgcxcr4b-B increased up to 3.7 and 5.4 folds respectively, and then both returned to the basic expression level (Figure 3C). Although fewer Cgcxcr4a transcripts were detected in head kidney compared to those of Cgcxcr4b, they were also slightly up-regulated (1.7-2.5 folds) at 1 dpi ( Figure 3C). In spleen, Cgcxcr4b-A and Cgcxcr4b-B increased their expression and reached a peak at 3 dpi ( Figure 3D). No significant expression change of Cgcxcr4as was detected in spleen after CaHV challenge ( Figure 3D). For four Cgcxcl12 genes, only Cgcxcl12a-A was significantly up-regulated in head kidney (4.0 folds at 1 dpi) and spleen (25.5 folds at 3 dpi) after CaHV infection (Figures 3E,F).

CgCxcr4/CgCxcl12 Axis Controls Antiviral Response Against CaHV
To explore the role of CgCxcr4/CgCxcl12 axis in antiviral response to CaHV, we investigated the symptoms and the CaHV abundance among four gibel carp groups treated with/without AMD3100 that specifically blocks the binding between CXCR4 and CXCL12 (49,50). AMD3100 injection did not affect the survival of the healthy individuals (not challenged by CaHV) (Figures 4A,B). After CaHV infection, the individuals exhibited hemorrhage at the base of fins and on abdomen at 3 dip (arrows in Figure 4A). Cxcr4-blocked individuals showed more severe bleeding and lower survival rate (Figures 4A,B), indicating more susceptibility to CaHV infection than control individuals (injecting PBS). The death started at 3 dpi in Cxcr4-blocked group and all died at 5 dpi, while the first death of control group occurred at 4 dpi and about 7% survivors appeared healthy at 10 dpi ( Figure 4B). To further confirm this susceptibility induced by Cxcr4 inhibition, we detected the abundances of five CaHV genes to evaluate the CaHV replication (38). After 35 amplification cycles, no DNA fragment of CaHV was detected from the healthy individuals, while the specific bands were amplified in the CaHVinfected individuals (Figure 4C). Accordant with more severe symptoms, the Cxcr4-blocked fish had markedly stronger bands than the control individuals at 3 dpi ( Figure 4C).
We also analyzed the expression of several key antiviral genes in head kidney and spleen (37), such as interferon (ifnϕ1 and ifnϕ3), interferon regulatory factor (irf3 and irf7), host pattern recognition receptors PRRs [toll-like receptor 9 (tlr9) and retinoic acid induce gene I (rig1)] and PRRmediated IFN signal pathway genes [mediator of irf3 activation (mita) and myeloid differentiation primary-response gene 88 (myd88)]. Similar to cxcr4 and cxcl12, they all possessed A and B homoeologues with divergent sequences and with biased expression in gibel carp (Figure 5). After CaHV infection, they were all up-regulated, and the up-regulation folds were significantly suppressed when Cxcr4-Cxcl12 binding was blocked    ( Figure 5 and Supplementary Figure 4). Together with above results, we suggest that blocking Cxcr4/Cxcl12 axis should suppress gibel carp antiviral response to CaHV.

CgCxcl12/CgCxcr4 Axis Enhances Cellular Antiviral Response Through MAPK3 and JAK/STAT Pathway
The function and signal pathways of about 20 IFN system genes have been well documented in CAB cells (40,41,(51)(52)(53), so we used CAB cells to explore the antiviral response pathways related to CgCxcr4/CgCxcl12 axis. After treatment with poly(I:C) that mimics the double-strand RNA (dsRNA) virus infection, all of the analyzed IFN system genes, including ifn, ifi58 (interferon inducible protein 58), pkr (dsRNA-activated protein kinase R), irf1 (interferon regulatory factor 1), irf3, irf7, viperin, mx1 (MX dynamin-like GTPase 1), tbk1 (TANK binding kinase 1), stat1(signal transducer and activator of transcription 1), tlr3 (toll-like receptor 3) and rig1, were significantly up-regulated ( Figure 6A). Consistent with the results in vivo (Figure 5), the similar suppression effects were also observed in the CAB cells treated with AMD3100 ( Figure 6A). Furthermore, ADM3100 pre-treatment also inhibited the activities of CaIFN and CgVip promoters, as well as elevating the amounts of STAT1, IRF7 and phosphorylated IRF3 (40), induced by poly(I:C) stimulation (Figures 6B,C). Finally, both CgCxcr4 or CgCxcl12 overexpression (Supplementary Figure 5) promoted the activation of IFN promoter induced by poly(I:C) ( Figure 6D). In addition, The up-regulation folds affected by CgCxcr4bs or CgCxcl12as were bigger than their paralogues respectively, and no obviously differences observed between each homoeologous pair (CgCxcr4b-A: 33 (Figure 6D).
Considering the bias expression in immune organs, we suppose that Cxcr4b and Cxcl12a might be the major antiviral couple in gibel carp. Altogether, Cxcr4/Cxcl12 axis facilitates the cellular antiviral response stimulated by poly(I:C).
To avoid the possible bias effects of poly(I:C), we cotransfected different CgCxcr4s-CgCxcl12s combinations in CAB cells to directly detect the CaIFN promoter luciferase activities. Compared to the control, co-transfection of CgCxcl12a-A-CgCxcr4b-A combination resulted in stronger activation of CaIFN promoter (Figure 7A), and other different ligand-receptor combinations showed the similar results (Supplementary Figure 6). CXCR4/CXCL12 axis can activate PI3K/Akt, PLC/IP3, ERK1/2 pathways (14,54), and β-arrestin/JAK/STAT pathway in mammals (55,56). To reveal the molecular mechanisms by which Cxcr4/Cxcl12 axis regulates IFN expression, we co-transfected CgCxcr4b-A and CgCxcl12a-A with dominant negative mutants of MAPK1, MAPK3 or STAT1 (MAPK1-DN, MAPK3-DN or STAT1a-C+STAT1b-C) in above vitro system. After 24 h of transfection, the activity of CaIFN promoter was diminished by the co-overexpression of MAPK3-DN or STAT1a-C+STAT1b-C, while overexpression of MAPK1-DN did not obviously affect the induction of IFN expression (Figure 7B). To further verify the results, we used western blotting to detect the expression of phosphorylation JAK2 (p-JAK2) and p-MAPK3 proteins. As shown in Figure 7C, overexpression of CgCxcr4b-A and CgCxcl12a-A promoted the expression of p-JAK2 and p-MAPK3. The increased expression was significantly reduced by co-transfecting MAPK3-DN or STAT1a-C+STAT1b-C ( Figure 7D). These results indicate that CgCxcl12/CgCxcr4 axis can regulate IFN expression through MAPK3 and JAK/STAT pathways.
During rediploidization, the sub-functionalization and neofunctionalization are the major maintenance ways of gene duplicates (68,69). The dynamic expression pattern was observed in some fishes after pathogen infection, elucidating that cxcr4/cxcl12 is not only important in immune defense against bacterial and viral infection but also may restrain the immune system in some other physiological processes (17)(18)(19)(20). In gibel carp, both paralogous and homoeologous pairs of cxcr4b and cxcl12a show dynamic and differential expression patterns and antiviral responses. Our results indicate that cxcr4bs and cxcl12as should play a dominant role in immune response and confirm the specific ligand-receptor recognitions of Cxcl12b and Cxcr4a (48,70). Although derived from the common ancestor (71), the expression patterns of duplicated genes evolved divergently after separation. For example, a common carp cxcl12b (Cccxcl12b)   , ifi58, pkr, stat1, irf1, irf3, irf7, tbk1, viperin, mx1, tlr3, and rig1. (B) AMD3100 inhibited the activation of fish IFN and viperin promoters in (Continued) Frontiers in Immunology | www.frontiersin.org exhibits highest identity with Cgcxcl12b-B, and they are both highly expressed in the brain (72). But besides, Cgcxcl12b-B also shows high expression in spleen ( Figure 3B). In gibel carp, all cxcr4 and cxcl12 paralogues/homoeologues can promote IFN gene expression in CAB cells, although Cxcl12as and Cxcr4bs show a bit higher ability ( Figure 6D). It seems that the subfunctionalization and neo-functionalization of cxcr4s or cxcl12s mainly occur at the transcriptional level, and cxcr4 or cxcl12 copies may still overlap for some functions (68,73). Although CXCR4 blockade benefits to autoimmune diseases, it is deleterious in acute inflammation and ischemia, and has a negative impact on negative physiological defense (74)(75)(76). In mammals, CXCR4 triggered by CXCL12 binding activates antiinflammatory signaling pathways and inhibits inflammation (77,78). It increases the production of pro-inflammatory cytokines, including IFN-γ, TNF-α, IL-6 and so on (79,80). AMD3100 administration also results in the significant reduction of IFNγ, TNF-α and IL-6 in the acute stage of ischemic stroke or after spinal cord injury (11,12). In addition, interruption of the CXCR4/CXCL12 axis not only down-regulates the expression of IFN-γ and TNF-α, but also negatively influences TLR4 signaling and leads to the decrease of inflammatory responses in the LPS treatment mice (76,81). In zebrafish, Cxcl12/Cxcr4 axis plays an important role to antagonize wound-induced inflammatory signals, but Cxcl12/Cxcr4 signaling also leads to the inappropriate retention of neutrophils at the inflammatory sites (8,23,82). Interestingly, impairment in the Cxcl12a/Cxcr4b signaling axis does not affect the recruitment of neutrophils when cxcr4b mutants are infected by Salmonella typhimurium (8). Therefore, the role of Cxcl12/Cxcr4 signaling axis against infection may be not affected by the retention and/or migration of leukocyte population. In this study, Cxcr4 blockade not only promotes the duplication of CaHV, but also suppresses the up-regulation expression of the virus recognition receptors tlr9, tlr3 and the following downstream antiviral genes, both in gibel carp after CaHV infection and CAB cells with poly(I:C) stimulation (Figures 4-6). Moreover, overexpression of CgCxcr4s or CgCxcl12s both instigate the induction of IFN by poly(I:C) ( Figure 6D). The innate immune response triggered by virus infection increases the expression of IFN and downstream IFN-stimulated genes (ISGs) to protect the host both in mammals and teleost (53,83,84). Taken together, we suggest that Cxcr4/Cxcl12 axis may be important for the innate immune responses to resist virus invasion in teleost.
Cxcl12 was reported to enhance dimerization of Cxcr4, and then to activate the downstream pathways (85). Activating some signaling pathways such as ERK, JAK/STAT and PI3K can regulate the transcription factors, including AP1, NF-κB and NFAT, to regulate the expression of cytokines (86)(87)(88)(89). They also play some roles in IFN induction by virus infection, as their domain negative mutants markedly inhibited the induction of IFN ( Figure 7B). Following CXCl12 binding, CXCR4 activates the Janus Kinases by changing their conformation to promote the JAK/STAT signaling pathway and the downstream transcription factor NF-κB that stimulates IFN secretion (85,90,91). In addition, CXCR4 internalization also activates the β-arrestin pathway, which severs as a scaffold for ERK and enhance the ERK activation (56,92). In our study, blocking MAPK3 and STAT1 has a significant suppression effect on IFN induction by CgCxcr4b-A and CgCxcl12a-A over-expression (Figure 7), indicating MAPK and JAK/STAT maybe the major pathways that CgCxcr4s/CgCxcl12s axis participates in antiviral response. However, blocking MAPK1 does not affect the induction of IFN, and the MAPK1 protein is much less than MAPK3 in CAB cells (data not shown), which suggest that MAPK1 might play a different role from MAPK3 in teleost.
In conclusion, the current study represents the diversification and dominant expression in immune tissues of Cxcl12/Cxcr4 axis in hexaploid gibel carp. It is the first time to report its function in teleost antiviral response, which regulates IFN expression mainly through MAPK3 and JAK/STAT pathways. These findings would expand our knowledge about the complex function of Cxcl12/Cxcr4 axis in teleost, which is a link between the innate and the adaptive immune system and play a key role to improve the adaptive immunity (22,93). Understanding the regulation mechanisms of Cxcl12/Cxcr4 signaling can provide useful information for disease control with effective immune protection in gibel carp. Further study of Cxcl12/Cxcr4 signaling and its downstream factors will help us to better understand the mechanism of host against CaHV infection, and provide resistant-related candidate genes for disease-resistance breeding of gibel carp.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.

ETHICS STATEMENT
The animal study was reviewed and approved by Institutional Animal Care and Use Committee of Institute of Hydrobiology, Chinese Academy of Sciences (protocol number 2016-018).

AUTHOR CONTRIBUTIONS
W-JL and F-XG performed the experiments, analyzed the data, and drafted the manuscript. YW participated in the data analysis and revised manuscript. ZL participated in the sample collection. Y-LZ and X-JZ participated in the data analysis. LZ designed the studies, analyzed the data, and drafted the manuscript. J-FG conceived the study and revised the manuscript. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
We thank Prof. Yi-Bing Zhang (Institute of Hydrobiology, Chinese Academy of Sciences) for providing some plasmids used in this study.