Rainbow trout (Oncorhynchus mykiss) (30 ± 5 g) and grass carp (Ctenopharyngodon idella) (200 ± 20 g) were purchased from Dujiang Dam Rainbow Trout Farm (Chengdu, China) and Xiantao Hatchery (Xiantao, China) respectively. They were maintained and acclimated to the laboratory conditions for at least two weeks before experiments. All animal experiments in this study were approved by the Committee on the Ethics of Animal Experiments at Huazhong Agricultural University.

Human embryonic kidney 293T (HEK293T) cells were cultured in 5% CO₂ at 37°C. Dulbecco’s modified Eagle’s medium (DMEM) (HyClone) was supplemented with 10% fetal bovine serum (FBS) (Gibco), 100 g/mL penicillin (Sigma-Aldrich), and 100 g/mL streptomycin (Sigma-Aldrich).

The chromosome-level genome of grass carp was assembled by our laboratory and deposited in the NCBI BioProjects with the accession number PRJNA745929 (28). The published fish LEAPs, especially zebrafish LEAPs, were used to search against the grass carp genome using the Basic Local Alignment Search Tool (BLAST). Interestingly, in addition to LEAP-2A and LEAP-2B, LEAP-2C was found in grass carp. The grass carp LEAP-2C was cloned and submitted to the GenBank database (https://www.ncbi.nlm.nih.gov/genbank/) under the accession number OQ026323. Similarly, the rainbow trout LEAP-2C was cloned and submitted to the GenBank database under the accession number GQ870279.1.

The signal peptide of the deduced amino acid sequences was predicted using the SignalP 5.0 Server (https://services.healthtech.dtu.dk/service.php?SignalP-5.0). The LEAPs gene organizations, including exon, intron and UTR were determined by aligning the cDNA sequences with the gene sequences. The protein sequence identity was calculated using the BioEdit software (version 7.0.9). Multiple sequence alignment was conducted with the ClustalX program (version 3.0), and phylogenetic tree was constructed based on the alignments using the neighbour-joining method with 1000 bootstrap times using the MEGA program (version 4.1). All the sequences used for multiple sequence alignment and phylogenetic analysis were listed in Table 1 .

Four healthy rainbow trout and grass carp were anesthetized with MS222 (1:10000), then the blood was removed from the body by cardiac perfusion using phosphate-buffered saline (PBS; pH 7.4; Gibco). The head kidney, spleen, gut, gill, skin, liver, heart, and muscle were collected, and the total RNA was extracted using the TRIzol Reagent (Takara). The cDNA was synthesized using the PrimeScript™ RT Reagent Kit contains gDNA Eraser (Takara). The mRNA expression levels of LEAPs were detected by quantitative real-time PCR (qPCR) using the CFX Connect™ Real-Time System (Bio-Rad). The primers used are listed in Table 2 . The reaction mixture (20 μl) contained 1 μl cDNA, 10 μl SsoAdvanced™ SYBR Green Supermix (Bio-Rad), 1 μl forward primer (10 μM each) and 1 μl reverse primer (10 μM each). The amplification program was as follows: 95°C for 5 min, 45 cycles of amplification (95°C for 5 s and 60°C for 30 s), and then 65°C for 5 s. The tissue expression levels of LEAPs were determined using 2^−ΔCt method with β-actin as the internal reference.

To detect the immune responses of LEAPs during infection, rainbow trout and grass carp were intraperitoneally injected with 100 µL Aeromonas salmonicida BG1 (29) suspension culture (1×10⁷ CFU/mL) and 200 µL Aeromonas hydrophila XS91-4-1 (30) suspension culture (8×10⁶ CFU/mL) respectively, while the control fish were intraperitoneally injected with PBS instead after anesthetized with MS222 (1:10000). At 12 h, 1 d, 3 d, 5 d, and 7 d post-injection, the liver and gut of four individuals were sampled from each group. After the RNA extraction and cDNA synthesis, the expression levels of LEAPs in infected and control fish were determined by qPCR. The expression changes of LEAPs after infection were calculated using the 2^−ΔΔCt method, with β-actin as the internal reference.

The mature peptides of rainbow trout and grass carp LEAPs were synthesized by GL Biochem Ltd. The purity was confirmed to be higher than 95% by HPLC and MALDI-TOF mass spectroscopy. All peptides were stored at -80°C for later use.

Six Gram-negative bacterial strains (Escherichia coli ATCC25922, A. hydrophila XS91-4-1, A. salmonicida BG1, A. sobria CR79-1-1, Edwardsiella ictaluri HSN-1, and Vibrio fluvialis WY91-24-3) and two Gram-positive bacterial strains (Micrococcus luteus ATCC10240 and Streptococcus agalactiae ATCC13813) were used in the antibacterial activity assay. Radial-diffusion assay (RDA) was conducted as previously described (31, 32). Briefly, 5 mL underlay agarose gel containing 0.03% (wt/vol) TSB, 1% (wt/vol) low electroendosmosis (EEO) agarose (Aladdin), 0.02% (vol/vol) Tween 20 (Amresco), and 4×10⁶ CFU bacteria was added into each sterile petri dish (90 mm diameter). After the solidification of the agarose, 4 mm diameter wells were punched in the agar using a sterile steel borer. Then 6 μL peptide (125 μM) was added to each well, and sterile water was added as control. Agar plates were incubated for 3 h at the optimum bacterial growth temperature (28°C for A. hydrophila, A. salmonicida, A. sobria and E. ictaluri or 37°C for E. coli, V. fluvialis, M. luteus and S. agalactiae) to diffuse the peptides. Each plate was added 5 mL sterilized overlay agarose gel (containing 6% TSB and 1% low EEO agarose) and incubated for 24 h at the optimum bacterial growth temperature. The diameter of the clearance zone around each well was measured.

The membrane permeability of bacteria after the treatment of peptides was assessed by flow cytometry as previously described (33) with minor modifications. In brief, V. fluvialis WY91-24-3 were cultured to mid-logarithmic phase, then washed twice and resuspended to 10⁶ CFU/ml with 10 mM NaPB (pH 7.4). Thereafter, 40 μl of peptides diluted in NaPB was added to 160 μl of V. fluvialis to give a final concentration of 8 μM. LL37 and NaPB was used as the positive and negative control, respectively. After being incubated at 37°C for 1 h, propidium iodide (PI; Sigma-Aldrich) was added to the bacterial suspension at a final concentration of 9 μM. The PI positive bacterial cells were detected using the flow cytometer FACSVerse (BD Biosciences) at 3,000 events and the data were analyzed using the FlowJo software v10 (Tree Star).

The Fpn of rainbow trout (GenBank: XM_021579209.2) was amplified using the cDNAs reverse-transcribed from liver RNA using the primers listed in Table 2 . The PCR product was digested with restriction enzymes (Kpn 1and Xma 1), followed by ligation into pEGFP-N1 to construct pEGFP-Fpn. The plasmid pEGFP-Fpn was extracted from the Trans5α cells using an E.Z.N.A. Plasmid Maxi Kit (Omega). The extracted plasmid was transfected into the HEK293T cells using the TransIntro EL transfection reagents (TransGen Biotech) according to the manufacturer’s instructions. At 24 h after transfection, cycloheximide (Aladdin) was added to the HEK293T cells to give a final concentration of 75 μg/mL. After incubation for 3 h, the synthetic rainbow trout LEAPs were added to give a final concentration of 1 μM and incubated for 24 h. The changes in the position of Fpn-EGFP in the cells were investigated using a live cell station (Leica AF6000).

The statistic p value was calculated using the SPSS Statistics (version 19, IBM) by one-way ANOVA with a Dunnett post hoc test. A p value < 0.05 was considered statistically significant while a p value < 0.01 was considered highly significant.

Homologous sequence alignment showed that all LEAPs consist of a signal peptide, a propeptide and a mature peptide. The sequences of the mature peptides are more conserved than those of the signal peptides and propeptides. The mature peptide of LEAP-1 and LEAP-2 possessed eight and four conserved cysteines, respectively. Interestingly, from fish to mammals, the signature of the cleavage site (RXXR) between the propeptide and the mature peptide of LEAPs are conserved in evolution ( Figure 1 ).

LEAP-1 and LEAP-2 have the same gene structure, which is composed of 3 exons and 2 introns. The 5’ and 3’ UTR of rainbow trout and grass carp LEAP-1 are longer than those of LEAP-1s from other species analyzed. The two introns of grass carp LEAP-2C are longer than those of LEAP-2Cs from other species analyzed ( Figure 2A ).

In the phylogenetic tree constructed using the amino acid sequences of the LEAP-1 and LEAP-2 precursors ( Figure 2B ), the LEAP-1 sequences branched apart from the LEAP-2 sequences. Teleost LEAPs were distinguished from those of other vertebrates. LEAP-2 of grass carp, rainbow trout and zebrafish contain three members, named LEAP-2A, LEAP-2B and LEAP-2C.

The mRNA expression levels of the LEAPs were evaluated in several healthy rainbow trout ( Figure 3A ) and grass carp ( Figure 3B ) tissues. Rainbow trout (rt) and grass carp (gc) LEAPs were differentially expressed in various tissues and organs, with an overall predominance of LEAP-1. In general, the LEAPs were highly expressed in the liver and lowly expressed in the gill, heart, and muscle. LEAP-2A was significantly expressed in the gut and skin of rainbow trout and grass carp, and the expression of LEAP-2C in the gut of rainbow trout was higher than that in the liver.

Expression of the total transcripts of rtLEAP-1 and rtLEAP-2 in the liver ( Figure 4A ) and gut ( Figure 4B ) of rainbow trout induced by A. salmonicida was studied. The expression level of rtLEAP-1 increased significantly in the liver from 12 h to 1 d, and in the gut from 12 h to 3 d after infection. The expression level of rtLEAP-2A increased significantly in the gut at 12 h after infection. The expression level of rtLEAP-2B increased significantly in the liver from 1 d to 5 d, and in the gut from 3 d to 7 d after infection. However, the expression level of rtLEAP-2C decreased significantly in the gut from 3 d to 7 d after infection.

Expression of the total transcripts of gcLEAP-1 and gcLEAP-2 in the liver ( Figure 4C ) and gut ( Figure 4D ) of grass carp induced by A. hydrophila was studied. The expression levels of gcLEAP-1 and gcLEAP-2s increased from 12 h to 3 d after infection, and then gradually returned to normal level, with gcLEAP-2A increased most significantly in the liver and gcLEAP-1 increased most significantly in the gut.

To investigate the antimicrobial properties of rainbow trout and grass carp LEAPs, we synthesized the mature peptides of rainbow trout and grass carp LEAPs and tested their antibacterial activities against a variety of bacterial strains. The results showed that all of the rainbow trout and grass carp LEAPs were antibacterial to Gram-negative (E. coli, A. hydrophila, A. salmonicida, A. sobria, E. ictaluri, and V. fluvialis) and Gram-positive bacteria (M. luteus and S. agalactiae) ( Figures 5A, B ). For the same bacterium, the antibacterial activities of different LEAPs are different. Interestingly, rainbow trout and grass carp LEAP-2C showed significant inhibitory effects on A. hydrophila, a common and harmful pathogen in fish farming.

To clarify the antibacterial mechanism of LEAPs, we detected the integrity of cell membrane by bacterial membrane permeability assay. The results showed that rainbow trout and grass carp LEAPs could significantly increase the permeability of the cell membrane of V. fluvialis within 1 h, and rtLEAP-1 had the most significant effect, which was similar to human AMP LL-37 ( Figures 6A–K ), suggesting that LEAPs kill bacteria through membrane permeabilizing action.

Through the amino acid sequence alignment of the LEAP-1 mature peptides in rainbow trout, grass carp, zebrafish, European sea bass, Atlantic salmon, and turbot, the Q-S/I-H-L/I-S/A-L motif was found in the N-terminus of the mature peptides ( Figure 7A ). To compare the ability of LEAPs to internalize Fpn, rtLEAPs were added to the cells expressing Fpn-EGFP, and the changes in the position of Fpn-EGFP in the cells were investigated using a live cell station. The results showed that only rtLEAP-1 resulted in the loss of cell surface Fpn-EGFP and the presence of intracellular Fpn-EGFP ( Figure 7B ), indicating that only rtLEAP-1 can cause the internalization of Fpn, while rtLEAP-2A, rtLEAP-2B and rtLEAP-2C cannot.