T. ovatus (average weight 18.5 g) from Hainan Province was temporarily reared for 1 week before the experiments. Golden pompano snout (GPS) cells, kindly provided by Professor Qin, were cultured in L-15 medium [containing 10% fetal bovine serum (FBS, Gibco), 100 U/ml penicillin, and 100 U/ml streptomycin] at 26°C (32). Human embryonic kidney (293T) cells were incubated in Dulbecco’s modified Eagle’s medium with 10% FBS and incubated at 37°C (5% CO₂ incubator).

V. harveyi that was isolated by our laboratory from golden pompano was used as the pathogen and cultured in Luria–Bertani (LB) medium (containing 100 μg/ml ampicillin, Amp) at 30°C (31). The suspension was diluted to 3 × 10⁷ colony-forming units (CFU)/ml when the OD₆₀₀ value reached approximately 0.6. The fish were intraperitoneally injected with 0.1 ml of the suspension, and the same volume of phosphate-buffered saline (PBS) was injected as the control. The liver, spleen, and head kidney were collected at 6, 9, 12, and 24 h post-infection (hpi). Three separate samples were prepared as well.

Total RNA of tissues and cells was extracted using E.A.N.A. Total RNA Extraction Kit (OMEGA, USA) and digested with DNase (OMEGA, USA). cDNAs were synthesized using Eastep^® RT Master Mix Kit (Promega, USA). qRT-PCR was performed to quantify the target gene mRNA level using SYBR ExScript qRT-PCR Kit (Promega, China). Beta-2-microglobulin was used as the housekeeping gene, and data were analyzed by the 2^-ΔΔCT method (33). The primers used are listed in Supplementary Table S1 .

The full open reading frame of the TroIGFBP5b sequence was amplified with primers TroIGFBP5b-F1/TroIGFBP5b-R1 by using T. ovatus liver cDNA as the template. The TroIGFBP5b sequence was blasted at the National Center for Biotechnology Information (NCBI, http://www.ncbi.nlm.nih.gov/blast). The structural domain was predicted at SMART online (http://smart.embl-heidelberg.de/). The three-dimensional (3D) structure prediction of TroIGFBP5b was carried out on the SWISS-MODEL website, and the visualization of the predicted protein 3D structure was achieved using PyMOL software. The phylogenetic tree constructed by MEGA 7.0 employed the neighbor-joining (NJ) method.

The TroIGFBP5b sequence, except for the signal peptide (SP) domain, was amplified with primer TroIGFBP5b-F2/TroIGFBP5b-R1, using the pEASY-T-TroIGFBP5b plasmid as a template by PCR, and named TroIGFBP5b-ΔSP. Subsequently, two HBM-deleted mutants (²¹⁵RKGFFKRKQCKPSRGRKR²³² to ²¹⁵RKGFFPSRGRKR²²⁶, delete ²²⁰KRKQCK²²⁵) of TroIGFBP5b were amplified with full-length TroIGFBP5b or TroIGFBP5b-ΔSP using the primer pairs TroIGFBP5b-F1/TroIGFBP5b-R3 or TroIGFBP5b-F2/TroIGFBP5b-R3 and TroIGFBP5b-F3/TroIGFBP5b-R1 by overlap PCR assay and named TroIGFBP5b-ΔHBM and TroIGFBP5b-Δ(HBM+SP), respectively.

To construct a eukaryotic expression vector for overexpressing IGFBP5 in vivo, TroIGFBP5b and TroIGFBP5b-ΔHBM were inserted into the pCN3 vector, which expresses the human cytomegalovirus immediate-early promoter, at the EcoR V site, resulting in pTroIGFBP5b and pTroIGFBP5b-ΔHBM (34). pEGFPX-N3 was used for subcellular localization and was reformed from the pEGFP-N3 vector (35). Recombinant GFP plasmids were constructed by connecting TroIGFBP5b, TroIGFBP5b-ΔSP, TroIGFBP5b-ΔHBM, and TroIGFBP5b-Δ(HBM+SP) to pEGFPX-N3 at the SmaI site, resulting in pTroIGFBP5b-WT-N3, pTroIGFBP5b-ΔHBM-N3, pTroIGFBP5b-ΔSP-N3, and pTroIGFBP5b-Δ(HBM+SP)-N3, respectively. To obtain biologically active recombinant TroIGFBP5b-ΔSP and TroIGFBP5b-Δ(HBM+SP) proteins, pET-32a, which could express a His-tag and a thioredoxin protein (Trx), was used and linearized at the EcoRV site. All of the above-mentioned positive constructs were confirmed by colony PCR and sequencing. The plasmids used in the cell-related experiments, as well as those injected into the fish body, were endotoxin-free plasmids harvested using Plasmid Extraction Kit (TransGen, China) according to the supplier’s instructions.

The siRNA synthesis followed the instructions of RiboMAX™ Express RNAi System (Promega, USA) as described (36). Briefly, two pairs of primers, siTroIGFBP5b-P1/siTroIGFBP5b-P2 and siTroIGFBP5b-P3/siTroIGFBP5b-P4, were designed to obtain two DNA oligonucleotides after incubation at 95°C for 5 min. The templates were allowed to cool slowly to room temperature (RT). Next, these two DNA oligonucleotides were used to separately synthesize the sense strand RNA or the antisense strand RNA templates at 37°C for 2 h. Afterwards, the DNA template was removed from the separate short RNA strands by digestion with DNase, and then the two RNA strands were mixed to synthesize the siRNA. Finally, the synthesized siRNA was purified following the manufacturer’s instructions. The control siRNA (siTroIGFBP5b-C) was synthesized with siTroIGFBP5b- C-P1/P2 and siTroCCL4-C-P3/P4 as described above. The primers used in this study are listed in Supplementary Table S1 .

To evaluate the in vivo role of TroIGFBP5b, the fish were intramuscularly injected with 15 μg (0.1 ml) overexpression plasmids (pTroIGFBP5b and pCN3). The knockdown of TroIGFBP5b was achieved by an intramuscular injection of 15 μg siTroIGFBP5b or siTroIGFBP5b-C into the fish. For these experiments, 0.1 ml PBS was injected as a control. To further study the function of HBM on bacterial infection in vivo, the pTroIGFBP5b-ΔHBM plasmid was also injected into the fish as described above, and the pTroIGFBP5b, pCN3, and PBS groups were repeated to compare differences.

After the post-injection of overexpressing plasmids for 5 days and the post-injection of siRNA for 12 h, 0.1 ml of V. harveyi (2 × 10⁷ CFU/ml) suspension was injected into all groups intraperitoneally. The appropriate size of the liver, spleen, and head kidney tissue blocks was determined aseptically at 6, 9, and 12 hpi. The weighed tissue blocks were ground in 300 µl PBS, and 700 µl PBS was added. In total, 100 µl homogenate was spread evenly onto LB plates (containing 100 μg/ml Amp) to measure the bacterial loads of the different tissues. Finally, the number of colonies per gram was calculated.

GPS cells were cultivated to analyze the subcellular localization of TroIGFBP5b according to the method of Chen et al. (33). Briefly, before transfection, GPS cells were grown in six-well plates overnight until reaching 60% confluence. pTroIGFBP5b-WT-N3, pTroIGFBP5b-ΔHBM-N3, pTroIGFBP5b-ΔSP-N3, pTroIGFBP5b-Δ(HBM+SP)-N3, and pEGFPX-N3 (4 μg) were transfected into GPS cells using LipoFiter3.0 (Hanbio, Shanghai, China). At 48 h post-transfection, the cells were fixed for 15 min with 4% (v/v) paraformaldehyde at RT. 4′,6-Diamidino-2-phenylindole (1 μg/ml) was used to stain the nuclei for 20 min, and rhodamine B was used to highlight the whole cell. Finally, the cells were washed with PBS until the cleaning solution became colorless, and the cells were observed using an inverted microscope. To monitor the localization of TroIGFBP5b after stimulation, V. harveyi (1 × 10⁵ CFU/ml) or lipopolysaccharide (LPS; 100 ng/ml) was added to the transfected cells (TroIGFBP5b-WT and TroIGFBP5b-ΔHBM) for 4 h after transfection for 48 h. The ensuing steps were the same as those described above.

Recombinant TroIGFBP5b (rTroIGFBP5b), HBM mutant (rTroIGFBP5b-ΔHBM), and rTrx were purified as described in a previous report (33). After exploring the induction conditions, the optimum induction condition was determined by incubating at 20°C for 8 h after adding isopropyl β-d-1-thiogalactopyranoside (0.5 mM). After purification by a Ni Sepharose column, PBS was used to dialyze the recombinant proteins which were concentrated with PEG8000. The concentration of the purified recombinant protein was measured by the Bradford method.

T. ovatus head kidneys were collected and rinsed with PBS three times aseptically. Density gradient centrifugation was used to obtain head kidney macrophages (HKMs) from T. ovatus as previously described with Percoll (GE Healthcare, USA) (37). The isolated cells were added to a six-well plate (1 × 10⁷ cells/well) containing L-15 medium for 2 h, and then 100 μl each of 200 μg/ml rTroIGFBP5b, rTroIGFBP5b-ΔHBM, and rTrx was infused into each well; the mixture was incubated at 26°C overnight. Green fluorescent microspheres (Aladdin, China) were diluted to 1% (w/v) L-15 (containing 10% FBS) medium and then added to each well at 26°C in the dark for 2 h. Before being subjected to Guava easyCyte™ Flow Cytometer (Millipore, USA), the cells were washed and resuspended in L-15 to 1 × 10⁹ cells/ml. The experiment was performed in triplicate.

T. ovatus head kidneys were collected and rinsed aseptically with PBS three times. Density gradient centrifugation was used to obtain head kidney lymphocytes (HKLs) from T. ovatus as previously described by Percoll (GE Healthcare, USA) (36). The prepared HKLs resuspended in L-15 medium with 10% FBS were added (90 μl) to a 96-well culture plate (1 × 10⁵ cells/well). Subsequently, 10 μl of rTroIGFBP5b and rTroIGFBP5b-ΔHBM (final concentrations of 10, 50, 100, 200, and 300 μg/ml) or rTrx (300 μg/ml) was incubated for 12 h at 26°C, and PBS was added as a control. As per the instructions, Cell Counting Kit-8 (CCK8) (Hanbio, Shanghai, China) was applied to measure the proliferation of HKLs. The results were calculated as follows: (A ₄₅₀ of protein-treated cells − A ₄₅₀ of the empty well)/(A ₄₅₀ of control cells − A ₄₅₀ of the empty well). The empty well contained only medium and CCK-8 solution. The experiment was repeated in triplicate.

293T cells were transfected with pTroIGFBP5b, pTroIGFBP5b-ΔHBM, and pCN3 as described above. The immunofluorescence microscopy assay was performed as previously described (38). The primary antibody used was an anti-p65 polyclonal antibody (Bioss, Beijing, China) at 1/200 dilution.

GPS cells were transfected with pTroIGFBP5b-WT-N3, pTroIGFBP5b-ΔHBM-N3, pTroIGFBP5b-ΔSP-N3, pTroIGFBP5b-Δ(HBM+SP)-N3, or pEGFPX-N3 in 10-cm-diameter culture dishes. Nuclear and Cytoplasmic Extraction Reagent Kit (Beyotime, Beijing, China) was used to separately extract nuclear and cytoplasmic proteins. After protein separation by 15% SDS-PAGE and transfer to a PVDF membrane (Millipore, Germany), the membrane was blocked with 5% BSA for 1 h. Then, the membrane was incubated with anti-EGFP (1/2,000 dilution, Bioss, Beijing, China) at 4°C overnight. After 10 min of washing with TBST three times, the secondary antibody (HRP-conjugated goat anti-mouse IgG and 1/2000 dilution) was added and incubated for 1 h at RT. Anti-β tubulin and anti-Histone H3 (Bioss, Beijing, China) were used as the nuclear and cytoplasmic internal references, respectively. The experiment was performed in triplicate.

A DLR assay was performed to examine the activation of NF-κB. GPS and 293T cells (1 × 10⁶ cells/well) were seeded in a 24-well plate. A total of 0.2 mg of NF-κB-specific firefly luciferase reporter vector, the pGL4.32 vector (luc2P/NF-κB, Promega, USA), 0.05 mg of pRL-CMV (a control vector), and 0.2 mg of pTroIGFBP5b or pTroIGFBP5b-ΔHBM were co-transfected into cells using LipoFiter3.0 (Hanbio, Shanghai, China). After transfection for 48 h, the firefly and Renilla luciferase activities in the cell lysates were measured using Dual-Luciferase Reporter Assay Kit (Promega, USA). All experiments were conducted three times independently.

All data in this study were statistically processed using GraphPad Prism (version 8.0.2). Statistically significant differences were evaluated by the t-test, with the P-value indicated (*P < 0.05, **P < 0.01).

TroIGFBP5b is 801 bp in length and encodes 266 amino acids (a.a.) (NCBI GenBank accession number OP712620). According to SMART prediction analysis, TroIGFBP5b is composed of a signal peptide (SP) (1–20 a.a.), an insulin growth factor-binding (IB) domain (23–100 a.a.), and a thyroglobulin type I repeat (TY) domain (210–261 a.a.) ( Figure 1A ). In addition, TroIGFBP5b contains a highly conserved HBM (220–226 a.a.) in the NLS sequence (215–232 a.a.) of the C-terminal domain ( Figure 1B ). The sequence alignment analysis showed that the IB and TY domains of TroIGFBP5b were very highly conserved among vertebrates, including mammals (Homo sapiens and Mus musculus), amphibians (Xenopus tropicalis), reptiles (Chelonia mydas), avians (Gallus gallus), and teleosts (D. rerio), suggesting its conservation during species evolution. The N-domain contains 12 conserved cysteine residues, and the C-domain contains six, which contributed to the structural stability by intradomain disulfide bonds between cysteine residues ( Figure 1C ). From the 3D predicted structure, HBM is located in the cavity formed by the C- and N- termini ( Figure 1D ). Identities with mammals, avians, reptiles, and amphibians are relatively lower, ranging from 51.66% to 58.74%. According to the multiple sequence alignment analysis, identities vary in teleosts, ranging from 60.15% to 97.38%. TroIGFBP5b shows a high identity with IGFBP5b in Seriola dumerili (97.38%), Lates calcarifer (94.91%), and Larimichthys crocea (92.54%) ( Table 1 ). A phylogenetic tree was constructed using the NJ algorithm, showing that TroIGFBP5b clusters with IGFBP5b of other teleosts and resembles Lates calcarifer IGFBP5b, the closest phylogenetically, with a bootstrap value of 92 ( Figure 1E ).

The expression pattern of TroIGFBP5b in T. ovatus was detected in 11 tissues using qRT-PCR: heart, stomach, brain, liver, skin, head kidney, intestine, spleen, gills, blood, and muscle. The results showed that it was the lowest in the muscle, set as 1. In contrast to the set criterion of muscle, the expression levels of the other tissues ranking from high to low were as follows: liver (134.44-fold), spleen (33.19-fold), brain (30.43-fold), gill (27.51-fold), head kidney (17.69-fold), intestine (17.18-fold), stomach (12.67-fold), skin (10.12-fold), blood (5.51-fold), and heart (5.11-fold) ( Figure 2A ). The results for the expression of TroIGFBP5b during V. harveyi infection in the three main immune organs all displayed a significant enhancement with a similar expression pattern, which tended to decrease after an initial increase. The time points at which the peaks appeared were different. In the liver, the expression peak was at 9 hpi with a 6.86-fold change, and it was also at 9 hpi with a 4.64-fold change in the head kidney. For the spleen, the highest expression (10.93-fold) was observed at 12 hpi ( Figure 2B ).

To explore the function of TroIGFBP5b in response to bacterial infection, TroIGFBP5b was overexpressed in vivo by injection with pTroIGFBP5b, pCN3, or PBS (as a control). On the 5th day after the plasmid injection, the expression levels of TroIGFBP5b in the liver, spleen, and head kidney of fish treated with pTroIGFBP5b were significantly higher than in the control using qRT-PCR analysis, indicating that the overexpression of TroIGFBP5b was successful ( Supplementary Figure S2 ). In the liver of the pTroIGFBP5b overexpression group, the bacterial load decreased by 1.7- and 1.85-fold compared with that of the control at 9 and 12 hpi, respectively. In the spleen, it was decreased by 1.05-fold and 2.23-fold in the pTroIGFBP5b group compared with the control group at the two time points. Furthermore, the pTroIGFBP5b group had an approximately 1.46- and 1.67-fold reduction in head kidney bacterial load at 9 and 12 hpi, respectively ( Figure 3A ).

In vivo siRNA technology was used to further analyze the effect of TroIGFBP5b against pathogen infection. The qRT-PCR analysis showed that the expression levels of TroIGFBP5b in the liver, spleen, and head kidney of fish treated with pTroIGFBP5b were significantly decreased than in the control group using qRT-PCR analysis, indicating that the knockdown of TroIGFBP5b was successful after siRNA injection for 12 h ( Supplementary Figure S3 ). After being challenged by V. harveyi, the liver bacterial counts were approximately 1.90-, 1.50-, and 1.55-fold higher in the siTroIGFBP5b-injected group at 6, 9, and 12 hpi than in the control group, respectively. On the other hand, at 6, 9, and 12 hpi, the splenic bacterial load after siTroIGFBP5b injection was approximately 1.97-, 1.87-, and 1.69-fold higher than that in the control group. In the head kidney, the bacterial loads increased by 1.63-, 1.54-, and 1.94-fold compared with the control at 6, 9, and 12 hpi, respectively ( Figure 3B ).

As mentioned above, TroIGFBP5b contains an SP sequence and an HBM in its NLS domain. To explore the role of these two motifs in the subcellular localization characteristics of TroIGFBP5b, recombinant plasmids containing TroIGFBP5b wild type, HBM-deleted TroIGFBP5b, SP-deleted TroIGFBP5b, and both HBM- and SP-deleted TroIGFBP5b were constructed based on pEGFPX-N3, which were named TroIGFBP5b-WT, TroIGFBP5b-ΔHBM, TroIGFBP5b-ΔSP, and TroIGFBP5b-Δ(HBM+SP), respectively ( Figure 4A ). GPS cells were transfected using LipoFiter3.0. The inverted fluorescence microscopy results showed IGFBP5-WT and IGFBP5-ΔHBM to be mainly observed in the cytoplasm, indicating that TroIGFBP5b might be localized in the cytoplasm of GPS cells. While in the absence of the SP sequence its localization changed, the main localization was transferred from the cytoplasm to the nucleus ( Figure 4B ). The protein level displayed by the green fluorescence was consistent with those observed by fluorescence microscopy ( Figure 4C ).

Although the subcellular localization of IGFBP5-WT and IGFBP5-ΔHBM did not seem different through observation, we wondered if the HBM deficiency would affect the subcellular localization under pathogen stimulation. To explore this postulate, GPS cells transfected with TroIGFBP5b-WT and TroIGFBP5b-ΔHBM were treated with LPS or V. harveyi. According to the results, green fluorescence could be observed in the nucleus after stimulation in the TroIGFBP5b-WT transfected cells. However, the cells transfected with TroIGFBP5b-ΔHBM did not undergo such transfer, suggesting that the deletion of HBM would result in the loss of nuclear transfer reaction in response to stimulation ( Figure 5A ). We also evaluated the green fluorescence at the protein level, and the results were in agreement with the results mentioned above ( Figure 5B ).

Since TroIGFBP5b was involved in defense against the pathogen in vivo, we wonder whether TroIGFBP5b had any effect on immune cell activities and if HBM contributed to the processes. To prove this opinion, rTroIGFBP5b, rTroIGFBP5b-ΔHBM, and rTrx (control) were purified, and the lymphocytes and macrophage cells were extracted from the head kidney. According to the CCK-8 assay results, rTroIGFBP5b enhanced HKL proliferation in a dose-dependent manner, and it was shown that 200 mg/ml exhibited the best effect, compared with the cells incubated with PBS and rTrx, while rTroIGFBP5b-ΔHBM had no effects on HKL proliferation ( Figure 6A ). From the results derived by checking using a flow cytometer, it was shown that the phagocytosis rate of the rTroIGFBP5b group was extremely higher than that in the rTrx and rTroIGFBP5b-ΔHBM cells, while the phagocytic activity of those cells that were treated with rTroIGFBP5b-ΔHBM was comparable with that of the rTrx-treated group ( Figure 6B ).

A dual-luciferase reporter assay was performed to determine the role of TroIGFBP5b in the NF-κB pathway. It was found that the overexpression of TroIGFBP5b significantly activated the NF-κB luciferase reporter activity in a dose-dependent manner, while the activating effects on the NF-κB signaling did not occur in the TroIGFBP5b-ΔHBM overexpression cells in both 293T and GPS cells ( Figures 7A, B ). This indicated that HBM was important in TroIGFBP5b activation to the NF-κB pathway. The immunofluorescence staining results proved that TroIGFBP5b overexpression promoted the nuclear translocation of p65 in 293T cells. However, this p65 translocation was blocked by overexpressing TroIGFBP5b-ΔHBM in 293T cells ( Figure 7C ). However, due to the poor specificity of the p65 antibody in GPS cells, we did not achieve satisfactory results (data not shown). These results demonstrated that the HBM of TroIGFBP5b seemed to function as a key for activating the NF-κB pathway.

Since the HBM sequence could affect the nuclear transfer of TroIGFBP5b protein in response to pathogenic stimulation, we further investigated whether it could affect its antimicrobial activity in vivo. Taking the same approach as before, the fish were injected with overexpression plasmids pTroIGFBP5b-WT, pTroIGFBP5b-ΔHBM, pCN3, or PBS (control), respectively. After having been infected with V. harveyi for 9 and 12 h, the bacterial loads in the liver, spleen, and head kidney of the pTroIGFBP5b-ΔHBM group were all significantly higher than that in the pTroIGFBP5b group. In the head kidney, the bacterial load in the mutant group was not any different from that of the control group, while in the liver the bacterial load was lower than the control at 9 and 12 hpi, and in the spleen, it was shown to be lower at 12 hpi ( Figure 8A ).

Furthermore, the in vivo function mechanism after the overexpression of TroIGFBP5b and TroIGFBP5b-ΔHBM was figured out by examining the expression of immune-related genes using qRT-PCR. According to the results, TroIGFBP5b overexpression could significantly raise the expression level of the selected immune-related genes (p65, IKBα, IL-8, IL-10, IL-1β, and TNF-α). TroIGFBP5b-ΔHBM could also affect some genes compared with pCN3, such as TNF-α, IKBα, IL-8, and IL-1β in the liver and IL-8 and IL-1β in the spleen. However, compared with the TroIGFBP5b group, the upregulated effect on those genes was significantly decreased in TroIGFBP5b-ΔHBM. To sum up, the upregulated expression of immune-related genes induced by TroIGFBP5b was practically shut down in the absence of HBM ( Figures 8B–D ).