Shrimp weighing an average of roughly 5 g were obtained from Guangzhou, Guangdong Province, China. The shrimp were cultured for 3 days in aerated seawater (30‰ salinity, 25°C) and fed a commercial food (HAID Group) three times daily prior to the experiment. V. parahaemolyticus (isolated from a diseased shrimp by our lab) were cultured in Luria broth (LB) medium overnight at a temperature of 37°C (4). The cultured V. parahaemolyticus were quantified through counting the microbial colony-forming units (CFUs) per milliliter on LB agar plates. The final injection concentration of V. parahaemolyticus was controlled to approximately 1 × 105 CFU/50 μl (15).

The open reading frame (ORF) of LvSTING (Genbank accession KY490589.1) was cloned into pAc5.1-HA vectors (16) to generate pAc-LvSTING-HA. The ORF of LvIKKβ (Genbank accession JN180642) was constructed into pAc5.1-FLAG vector (17) for expressing FLAG-tagged LvIKKβ protein. The ORF of LvRelish (Genbank accession EF432734) was constructed into pAc5.1-V5 vector (Invitrogen, Cat No. V4110-20, USA) for expressing V5-tagged LvRelish protein. GFP sequence was constructed into pAc5.1-HA vector to express HA-tagged GFP. The promoter sequences of L. vannamei anti-LPS-factor 1 (LvALF1, Genbank accession EW713395), Crustin 1 (LvCRU1, Genbank accession AF430071.1), Lysozyme 1 (LvLYZ1, Genbank accession JN039375.1) and (LvPEN4, Genbank accession DQ206402) were cloned into pGL3-basic vector (Promega, Cat No. E1751, USA). Primer sequences were listed in Table 1 .

Co-immunoprecipitation (Co-IP) assays were performed to investigate the interaction of endogenous proteins ectopic expressed proteins in cells or in shrimp hemocytes. In Drosophila S2 cells, the plasmids pAc-LvRelish-V5 or pAc-LvIKKβ-FLAG were co-transfected with pAc-LvSTING-HA. The pAc-LvRelish-V5 or pAc-IKKβ-FLAG was also co-transfected with pAc-GFP-HA as controls. Forty-eight hours after transfection, cells were harvested and lysed with IP lysis buffer (Pierce, Cat No. 87788, USA) with a Halt Protease Inhibitor Cocktail (Merck, Cat No. 524628, Germany). The 90% of the cell lysis were incubated with agarose affinity gel of anti-HA (Merck, Cat No. A2095, Germany) or anti-V5 Agarose Affinity Gel antibody produced in mouse (Merck, A7345-1ML, Germany). The remaining 10% of cell lysis were used as input controls. All samples were subjected to SDS-PAGE assays. The primary antibodies used in western blotting included rabbit anti-FLAG antibody (Merck, Cat No. F7425, Germany), rabbit anti-V5 antibody (Merck, Cat No. AB3792, Germany) and rabbit anti-HA antibody (Merck, Cat No. H6908, Germany). Anti-rabbit IgG HRP-conjugate (Promega, Cat No. W4011, USA) was used as the secondary antibody.

In order to detect endogenous LvSTING–LvRelish interaction in vivo, hemocytes were lysed in IP lysis buffer (Pierce, Cat No. 87788, USA) with Halt Protease Inhibitor Cocktail (Thermo, Cat No. 87786, USA), and then incubated with protein G agarose beads (CST, Cat No. 37478S, USA) coated with anti-LvRelish antibody (Prepared by Genecreate, China) or a normal rabbit IgG antibody (CST, Cat No. 7074S, USA) for 3 hours at 4 °C, and finally were detected by western blotting with anti-LvSTING antibody (Prepared by Genecreate, China). Five percent of the cell lysis was loaded as the input control.

Drosophila S2 cells were transfected with reporter gene plasmids, pRL-TK renilla luciferase plasmid (as an internal control), and expression plasmid (pAc-LvRelish-V5, pAc-IKKβ-FLAG and pAc-LvSTING-HA) or empty pAc5.1/V5-His A plasmid (as a control) using the FuGENE Transfection Reagent (Promega, Cat. No. e2311, USA) following the manufacturer’s protocol. At 48 h after transfection, the firefly and renilla luciferase activity was measured following the manufacturer’s protocol. Three replicates were performed for each experiment.

DsRNAs specifically targeted to the LvSTING, LvIKKβ or LvRelish, were synthesized through in vitro transcription via T7 RiboMAX Express RNAi System kit (Promega, Cat. no. P1700, USA). DsRNA-GFP (dsGFP) targeting GFP (Genbank accession DQ389577) was used as a control. Primer sequences were listed in Table 1 .

Sample gaining, total RNA extraction and quantitative PCR (qPCR) assays were conducted as previously described (18). The expression levels of target genes were determined using the Livak (2-ΔΔCT) method following normalization to L. vannamei EF-1a (GU136229). Primer sequences were listed in Table 1 . Three replicates were performed for each experiment.

To investigate whether LvRelish played a protective role against V. parahaemolyticus, healthy shrimp were separated into two groups and injected with dsGFP or dsRNA-LvRelish (dsLvRelish). Each shrimp was injected with dsRNA (2 μg/g shrimp). Shrimp were then injected with V. parahaemolyticus or PBS after 48 h and maintained in culture flasks for approximately a week following infection. Surviving shrimp numbers were recorded every 4 h.

Another experiment was conducted to monitor the abundance of V. parahaemolyticus in LvRelish-knockdown shrimp. Quantitative PCR (qPCR) assays were conducted as previously described [10]. At 12 h after V. parahaemolyticus infection, gill tissue samples were collected from each group to extract DNA. The Marine Animals DNA Kit (TIANGEN, Cat. No. DP324, China) was used to extract gill DNA. The number of bacteria in gill tissue samples was quantified through qPCR using the V. parahaemolyticus 16S rRNA gene (rDNA, GenBank No. EU660325) with specific primers ( Table 1 ) (19). In brief, serial dilutions (108, 107, 106, 105, 104, and 103 copies) of plasmids containing V. parahaemolyticus 16S rRNA gene fragments were used to construct the standard curve. The genome copies of V. parahaemolyticus in 1 g gill tissue samples were then calculated. Hemocyte RNA was extracted to determine the expression of LvRelish for RNAi efficiency, and to detect the expression of shrimp NF-κB-mediated effector genes (LvALF1, LvLYZ1, LvCRU1, and LvPEN4). Three replicates were performed for each experiment.

The expression patterns of LvSTING in the hemocytes from V. parahaemolyticus-challenged shrimp were investigated. For immune stimulations assay, the treated groups were injected with V. parahaemolyticus solution (1 × 105 CFU), and the control group was injected with PBS solution. Hemocytes of challenged shrimp were sampled at 0, 3, 6, 12, 24, 36, 48 h post injection (hpi), and each sample was collected and pooled from 5 shrimp. Primer sequences were listed in Table 1 .

Forty-eight hours post dsGFP, dsRNA-LvSTING (dsLvSTING) or dsRNA-LvIKKβ (dsLvIKKβ) injection, shrimp were injected with 50 µl PBS or a suspension of approximately 1 × 105 CFU of V. parahaemolyticus. At six hours after V. parahaemolyticus infection, shrimp hemocytes were obtained through centrifugation (1000 g for 5 min) at 25 °C and seeded onto the slides. The hemocytes were fixed with 4% paraformaldehyde, and then were permeabilized with methanol at -20 °C. The slides were blocked using 3% BSA for 1 h at 25 °C and then incubated overnight (at 4°C for approximately 8 h) in a mixture of rabbit anti-LvRelish antibody (Genecreate, China) and mouse anti-β-actin antibody (Merck, Cat. No. A2228, Germany). The hemocytes were then washed with PBS and incubated with the fluorescent antibody (CST, Cat. No. 4412S/8890S, USA) for 1 h at 25 °C in the dark. The hemocytes were washed with PBS and stained with Hoechst 33258 (Yeasen, Cat. No. 40729ES10, China) for 10 min at 25 °C before being washed another six times. The fluorescence was visualized using a confocal laser scanning microscope (Leica, TCS-SP8, Germany). WCIF ImageJ software was used to analyze the colocalization of LvRelish and Hochest-stained nuclei in hemocytes.

Shrimp hemocytes were harvested at 6 hours post V. parahaemolyticus infection and lysed with IP lysis buffer (Pierce, Cat No. 87788, USA) with a Halt Protease Inhibitor Cocktail (Merck, Cat No. 524628, Germany). All samples were subjected to SDS-PAGE assays. The primary antibodies used in western blotting included rabbit anti-pLvRelish antibody (Genecreate, China) and mouse anti-β-actin antibody (Merck, Cat. No. A2228, Germany). Anti-rabbit IgG (H+L) HRP-conjugate (Promega, Cat No. W4011, USA) and Anti-mouse IgG (H+L) HRP-conjugate (Promega, Cat No. W4021, USA) were used as the secondary antibody.

To investigate whether LvSTING can activate LvRelish, healthy shrimp were separated into two groups and injected with dsGFP or dsLvSTING. Each shrimp was injected with dsRNA (2 μg/g shrimp). 48 hours post dsRNA injection, shrimp were infected with V. parahaemolyticus or PBS. And 12 hours post V. parahaemolyticus infection, hemocytes were harvested for qPCR, western blotting and immunofluorescence assay, and gill tissue samples were collected for V. parahaemolyticus numbers. Three replicates were performed for each experiment.

To explore whether LvIKKβ participates in LvRelish activation, healthy shrimp were separated into two groups and injected with dsGFP or dsLvRelish. Each shrimp was injected with dsRNA (2 μg/g shrimp). 48 hours post dsRNA injection, shrimp were infected with V. parahaemolyticus or PBS. And 12 hours post V. parahaemolyticus infection, hemocytes were harvested for qPCR, western blotting and immunofluorescence assay. Gill tissue samples were collected for counting V. parahaemolyticus numbers. Three replicates were performed for each experiment.

To prove that the LvSTING–LvIKKβ–LvRelish–AMPs axis plays a protective role against V. parahaemolyticus, healthy shrimp were separated into four groups and injected with dsGFP (4 μg/g shrimp), dsLvSTING (2 μg/g shrimp) plus dsGFP (2 μg/g shrimp), dsLvIKKβ (2 μg/g shrimp) plus dsGFP (2 μg/g shrimp), and dsLvSTING (2 μg/g shrimp) plus dsLvIKKβ (2 μg/g shrimp). Shrimp were then injected with V. parahaemolyticus or PBS after 48 hours and maintained in culture flasks for approximately a week following infection. Surviving shrimp numbers were recorded every 4 h. And 12 hours post V. parahaemolyticus infection, hemocytes were harvested for qPCR, and gill tissue samples were collected for counting V. parahaemolyticus numbers. Three replicates were performed for each experiment.

All the data are presented as mean ± SD. Student’s t test is used to calculate the comparisons between groups of numerical data. For survival rates, data are subjected to statistical analysis using GraphPad Prism software to generate the Kaplan ± Meier plot (log-rank χ2 test). The following p values are considered to be statistically significant: *p < 0.05, **p < 0.01 and ***p < 0.001.

Shrimp NF-κB pathway is crucial for AMPs expression, and LvRelish is a key transcription factor of NF-κB pathway. In this study, RNAi was performed to investigate the relationship between LvRelish and AMPs. We designed and synthesized dsRNA-LvRelish (dsLvRelish) targeting LvRelish expression, and checked the silencing efficiency of LvRelish at 48 h post dsRNA injection. The injection of dsLvRelish resulted in a significant decrease in LvRelish expression levels down-regulating to ~0.11-fold of the control group ( Figure 1A ), which was sufficient for the following experiments. Accordingly, the expression levels of LvALF1, LvCRU1, LvLYZ1 and LvPEN4 were remarkably down-regulated to ~0.47-fold, ~0.45-fold, ~0.36-fold and ~0.57-fold compared to those of dsGFP group at 48 h post dsRNA injection ( Figure 1A ).

Considering the relationship between LvRelish and AMPs, we were curious about the role played by LvRelish in the host defense against bacterial infection. As shown in Figure 1B , the expression of LvRelish in dsLvRelish treated group was down-regulated to ~0.30-fold comparing with the dsGFP group, which meant dsLvRelish was competent for LvRelish knockdown during V. parahaemolyticus infection. And the detection of AMPs during V. parahaemolyticus infection showed that the expression of AMPs in dsLvRelish injected shrimp were lower than those of the control group ( Figure 1B ). These results suggested that LvRelish could regulate AMPs expressions in both uninfected shrimp ( Figure 1A ) and V. parahaemolyticus-infected shrimp ( Figure 1B ).

During V. parahaemolyticus infection, the survival rate of dsLvRelish group was much lower than that of dsGFP group (χ2: 8.674, p = 0.0340), which indicated that LvRelish silenced shrimp were more sensitive to V. parahaemolyticus infection ( Figure 1C ). Besides, the higher numbers of V. parahaemolyticus were observed in dsLvRelish group at 12 h post V. parahaemolyticus infection ( Figure 1D ) that correlated well with the survival percent recorded in Figure 1C , and further confirmed that LvRelish played an antibacterial role in the innate immune response.

Relish is the vital transcription factor of STING-mediated pathways in silkworm and fruit fly (12, 20), but whether LvSTING participates in Relish regulation is still unclear. In this study, we found that V5 tagged LvRelish was co-immunoprecipitated with HA tagged LvSTING, but no appreciable binding was observed for HA tagged GFP protein ( Figure 2A ), which suggested that LvSTING could interact with LvRelish.

Given the vital role acted by LvRelish in inducing AMPs, the discovery of the association between LvSTING and LvRelish implied that LvSTING might induce AMPs by LvRelish. As Figure 2B shown, LvRelish-triggered promoter activities of LvALF1, LvCRU1, LvLYZ1 and LvPEN4, were promoted by the co-expression of LvSTING in S2 cells, which demonstrated that LvSTING could enhance LvRelish-mediated AMPs expression in vitro.

To prove the interaction between LvSTING and LvRelish existed in shrimp hemocyte, IP experiments were performed with LvRelish antibody in vivo. IP assays demonstrated that LvRelish could interact with LvSTING in hemocyte ( Figure 3A ). In hemocyte of V. parahaemolyticus infected shrimp, LvSTING expressions were up-regulated at 3, 6, 12, 24 hours post V. parahaemolyticus infection, indicated that LvSTING also responded to V. parahaemolyticus infection in hemocyte ( Figure 3B ). Then, we confirmed the effects of LvSTING on LvRelish activation in vivo. As Figure 3C shown, dsLvSTING could efficiently suppress LvSTING expression to ~0.19-fold of control. LvRelish has been proved to transfer from cytoplasm into nuclear after V. parahaemolyticus infection (21), and LvRelish nuclear location was inhibited by LvSTING knockdown during V. parahaemolyticus infection ( Figures 3D, E ). And the phosphorylation of LvRelish was weakened in hemocytes from LvSTING knocked down shrimp ( Figure 3F ), suggesting that LvSTING was involved in activating LvRelish. LvSTING knockdown led to down-regulated the expression of LvALF1, LvCRU1, LvLYZ1 and LvPEN4 during V. parahaemolyticus infection ( Figure 3G ), which indicated that LvSTING was involved in AMPs expression in shrimp. To sum up, LvSTING prompted LvRelish phosphorylation and nuclear location to induce AMPs expression.

IKKβ is the phosphokinase targeting Relish in Drosophila IMD pathway (22), but no direct evidence supporting that shrimp IKKβ activates Relish. In this study, dsLvIKKβ was used to inhibit the expression of LvIKKβ ( Figure 4A ), and the effect of LvIKKβ on LvRelish activation was examined. LvRelish subcellular location was observed by immunofluorescence assay. The results showed that LvRelish moved into the nuclear in response to V. parahaemolyticus infection, and dsLvIKKβ injection could suppress nuclear import of LvRelish ( Figures 4B, C ). LvRelish phosphorylation levels in hemocytes were significantly reduced by silencing LvIKKβ expression ( Figure 4D ). QPCR performed that dsLvIKKβ inhibited AMPs expression ( Figure 4E ), which is consistent with the AMPs expression changes caused by dsLvRelish or dsLvSTING. The data above indicated that LvIKKβ induced AMPs expression via activating LvRelish.

To explore the effects of LvSTING on LvIKKβ–LvRelish signaling transduction, the interaction between LvSTING and LvIKKβ was examined. The Co-IP demonstrated that FLAG-tagged LvIKKβ interacted with HA-tagged LvSTING but not HA-tagged GFP, which indicated that LvSTING could recruit LvIKKβ ( Figure 5A ). To prove the recruitment functions of LvSTING, FLAG-tagged LvIKKβ and V5-tagged LvRelish were co-expressed with or without HA-tagged LvSTING in S2 cells. As shown in Figure 5B , FLAG-tagged LvIKKβ was co-immunoprecipitated with V5-tagged LvRelish with the help of HA-tagged LvSTING, and the absence of LvSTING resulted in less LvIKKβ was co-immunoprecipitated with LvRelish. To confirm the effect of LvSTING on LvIKKβ–LvRelish–AMPs pathway, the dual luciferase assays was performed in S2 cells. The result showed that the promoter activities of shrimp AMPs (LvALF1, LvCRU1, LvLYZ1 and LvPEN4) could be positively regulated by the co-expression of LvIKKβ and LvRelish, which could be further upregulated by ectopic expression of LvSTING in a dose-dependent manner ( Figure 5C ).

As mentioned above, there was a LvSTING–LvIKKβ–LvRelish–AMPs signaling pathway in shrimp. We were curious about the roles played by the above signaling pathway in the host defense against V. parahaemolyticus infection. Double knockdown experiments were done to investigate whether LvSTING regulated the expression of AMPs through LvIKKβ. We observed that LvSTING- or LvIKKβ-knockdown reduced the expression of AMPs during V. parahaemolyticus infection. When compared to dsLvIKKβ-injected shrimp, LvSTING and LvIKKβ knockdown combined suppressed the expression of LvALF1 (~0.34-fold), LvCRU1 (~0.43-fold), LvLYZ1 (~0.53-fold) and LvPEN4 (~0.67-fold) ( Figure 6A ). Furthermore, Figure 6B showed that knockdown of LvSTING or LvIKKβ increased shrimp susceptibility to V. parahaemolyticus infection compared to the dsGFP group (χ2 = 10.32, p = 0.0013; χ2 = 5.662, p = 0.0173). In addition, when compared to LvIKKβ-silenced shrimp, LvSTING and LvIKKβ knockdown resulted in higher cumulative mortality (χ2 = 9.974, p = 0.0016), suggesting that LvSTING improved shrimp resistance to V. parahaemolyticus infection via LvIKKβ ( Figure 6B ). In agreement with the survival curves, LvSTING- or LvIKKβ-knockdown boosted V. parahaemolyticus levels in shrimp gill tissues, and V. parahaemolyticus numbers in the dsLvSTING + dsLvIKKβ group were substantially greater than those in the LvIKKβ-silenced alone group ( Figure 6C ). Taken together, these results suggested that LvSTING could trigger an antibacterial response via the LvIKKβ–LvRelish–AMPs pathway during V. parahaemolyticus infection.