The candidate probiotics, B. subtilis WB60 was isolated from the intestines of healthy Japanese eel (Anguilla japonica) and was identified by cluster analysis via 16S rDNA sequencing. The B. subtilis WB60 was isolated according to Lee et al. (28) and incubated at 30°C for 72 h in Luria-Bertani broth (LB broth; Sigma-Aldrich, St. Louis, USA), after which the optical density (OD600) was measured at 600 nm using spectrophotometry. The B. subtilis SJ10 was isolated from jeotgal, a traditional Korean fermented dish made from salt-preserved seafood such as squid, pollock roe, and shrimp, according to Hasan et al. (29) and it was incubated from a single colony on lysogeny broth (LB, USB Corporation, USA) agar, and was subsequently cultured in 10 mL of LB broth for 16 h at 37°C in a shaking incubator. Furthermore, E. faecium SH30 was isolated from the intestine of healthy Nile tilapia (Oreochromis niloticus), and the bacteria were grown in MRS (deMan, Rogosa, and Sharpe) broth at 36°C for 48 h according to Wang et al. (30). All probiotics were washed in sterile saline and the concentration of the final suspension was calculated to be 1 × 10⁸ CFU/g for WB60 and SJ10, and 1 × 10⁷ CFU/g for SH30 in the diets (28–30).

Juvenile olive flounder were obtained from a private farm (JUNGANG Fisheries, Chungcheongnam-do, Taean-gun, Republic of Korea). Prior to the start of the feeding trial, the apparent health status of the fish was checked visually, and the fish were starved for 24 h. All the fish were then fed a commercial diet for two weeks prior to the start of the feeding trial to acclimatize to the laboratory conditions. On average, 12.1 ± 0.09 g (mean ± SD) of fish were weighed, divided into triplicate groups of 25 fish corresponding to the dietary treatment, and randomly distributed into 18 indoor fiberglass tanks (40 L each) receiving a constant flow (1.2 L/min) of filtered seawater. During the experiment, supplemental aeration was provided in each tank to maintain adequate dissolved oxygen. The temperature was maintained at 19.0 ± 1.0°C throughout the experiment by electric heaters in a concrete reservoir. Fish were fed twice a day (09:00 and 19:00) for 8 weeks at a rate of 2.5 ~ 5% body weight per day. Dead fish were immediately removed and weighed, after which the amount of feed provided to the remaining fish was adjusted. The uneaten feed was siphoned 1 h after feeding. The inside of the tanks was scrubbed once per week to minimize algal and fungal growth.

The basal or control diet (CON) formulation is shown in Table 1. Anchovy fish meal (68.75% CP, crude protein) and soybean meal (47.04% CP) were used as the main protein sources, while fish oil was used as the main lipid source. The feed additives (probiotics) used in this experiment were Pro-A: B. subtills WB60 (1 × 10⁸ CFU/g) based on Lee et al. (28), Pro-B: B. subtills SJ10 (1 × 10⁸ CFU/g) based on Hasan et al. (29), Pro-C: E. faecium SH30 (1 × 10⁷ CFU/g) based on Wang et al. (30), as well as OTC: oxytetracycline, 4 g/kg diet and AMO: amoxicillin, 4 g/kg diet which were based on Won et al. (31). The procedures for feed manufacturing and preparation were performed as previously described by Lee et al. (28). According to the feed formulation table, all fine powdered ingredients were mixed thoroughly with an electric mixer (HYVM-1214, Hanyoung Food Machinery, Republic of Korea). Then, a stiff dough was formed by adding fish oil and the desired amount of water (~10%). The dough was passed through a pellet machine (SFD-GT, Shinsung, Republic of Korea) with a 0.2 cm die. The prepared diets were air-dried in a drying room for 48 h, broken into smaller pieces and stored at -20°C. According to the proximate composition analysis, shown in Table 2, all the diets were iso-nitrogenous and iso-lipidic.

At the end of the feeding trial, fish were starved for 24 h prior to sample and data collection. The fish were subsequently counted and weighed to calculate the final weight (FW), weight gain (WG), specific growth rate (SGR), feed efficiency (FE) protein efficiency ratio (PER) and survival (SR). Four fish from each tank were selected at random, weighed individually, and dissected to obtain liver and visceral metrics for calculation of hepatosomatic index (HSI) and visceral somatic index (VSI); thereafter, the same intestinal samples were used for histological observation and enzyme activity. Three additional fish per tank were captured at random and anesthetized with ethylene glycol phenyl ether (200 mg/L for 5–10 min). After this, blood was drawn from the caudal vein, which was subsequently centrifuged at 5000 × g for 10 min to obtain the serum. Serum samples were then stored at −70°C for the analysis of non-specific immune responses, such as lysozyme, and myeloperoxidase (MPO) activities, in addition to biochemical parameters, including aspartate aminotransferase (AST), alanine aminotransferase (ALT), glucose and total protein (TP) levels. The serum levels of AST, ALT, glucose, and total protein were determined by a chemical analyzer (Fuji DRI-CHEM 3500i, Fuji Photo Film Ltd., Tokyo, Japan) following the manufacturer’s instructions.

Three additional fish from each tank were collected for whole-body proximate composition analysis. Proximate composition analyses of both whole fish and experimental diets were performed by the standard methods of AOAC (32). Whole fish and diet samples were dried at 105°C to a constant weight to determine their moisture content. The ash content was determined by incinerating the samples at 550°C. The protein concentration was determined by using the Kjeldahl method (N × 6.25) after acid digestion. Crude lipids were measured by Soxhlet extraction using Soxhlet system 1,046 (Tacator AB, Hoganas, Sweden) after the samples were freeze-dried for 20 h.

Olive flounder serum lysozyme activity was analyzed as follows: 0.1 mL of test serum was added to 2 mL of a suspension of Micrococcus lysodeikticus (0.2 mg/mL) in 0.05 M sodium phosphate buffer (pH 6.2). The reactions were carried out at 20°C, and the absorbance was measured at 530 nm. Measurements were taken between 0.5 min and 4.5 min on a spectrophotometer. One lysozyme activity unit was defined as the amount of enzyme that produced a decrease in absorbance corresponding to 0.001/min. Myeloperoxidase activity was measured according to the method described by Quade and Roth (33). Briefly, 20 μL of serum was diluted with Hank’s balanced salt solution (HBSS) without Ca²⁺ or Mg²⁺ (Sigma- Aldrich) in 96-well plates. Then, 35 μL of 3, 3′, 5, 5′ tetramethylbenzidine hydrochloride (TMB, 20 mM) (Sigma-Aldrich) and H₂O₂ (5 mM) were added. The color change reaction was stopped after 2 min by adding 35 μL of 4 M sulfuric acid. Finally, the optical density was read at 450 nm in a microplate reader.

Tissue fragments from intestine of fish were obtained and immediately stored at −80°C in TRIzol reagent (Thermo Fisher Scientific) for RNA extraction. Total RNA was extracted from 0.5 g of olive flounder tissue using TRIzol Reagent (Thermo Fisher Scientific, San Jose, CA, USA). Afterwards, the RNA was quantified and the purity was assessed spectrophotometrically. The RNA was then treated with DNase I (Cosmogenetech, Seoul, Republic of Korea) to remove genomic DNA contamination. Complementary DNA (cDNA) was synthesized using M-MuLV reverse transcriptase (Cosmogenetech). The expressions of three selected immune-related genes were analyzed by real-time quantitative polymerase chain reaction (RT-qPCR), which was performed with a Bio-Rad CFX96 (Bio-Rad, Hercules, CA, USA) using SYBR Green PCR Core Reagents (Cosmogenetech). The relative expression levels of the target gene transcripts such as flounder growth hormone (FGH), interleukin-1beta (IL-1β) and interleukin-10 (IL-10) were measured with β-actin as an internal control were using CFX Manager software version 2.0 (Bio-Rad) (Table 3). In all the cases, each PCR was performed with triplicate samples.

The anterior intestinal tissues from the fish (n = 3) were dissected and fixed in 10% neutral buffered formalin, dehydrated in a graded ethanol series and embedded in paraffin. The tissue blocks were sectioned by a microtome machine (HistoCore AUTOCUT, Leica Biosystems, Germany) each of 4 μm thick and stained with hematoxylin and eosin (H&E). At least 6 tissue sections from each sample were examined for intestinal villi length (small fingerlike projections protruded from the intestinal wall) under a light microscope (AX70 Olympus, Tokyo, Japan) equipped with a digital camera (DIXI Optics, Daejeon, Republic of Korea), and an image analysis software (Image J 1.32j, National Institute of Health, USA). Data are presented as means ± SE.

After sampling, seven fish from each tank were redistributed into 18 tanks in a non-recirculating system without water renewal to perform the 15 days of challenge test. The pathogenic bacterium, Edwardsiella tarda (E. tarda) FSW910410 was obtained from the Department of Biotechnology, Pukyong National University, Busan, Rep. Korea. The bacteria were originally sourced from diseased olive flounder and cultured on tryptic soy agar (TSA, Sigma) plates (24 h at 27°C). All the fish were subjected to intraperitoneal injection of 50 μL of active E. tarda (3 × 10⁸ CFU/mL) solution. The water temperature was maintained at 19 ± 1.0°C (mean ± SD) during 15 days of challenge test and fish mortalities were recorded daily from each tank. Dead fish were necropsied and kidney samples were taken and streaked on Salmonella-Shigella agar (SS agar, Difco). The presence of black pigments confirmed E. tarda infection.

The results are presented as means ± SE (number of replicates as indicated). All the data were analyzed by one-way ANOVA (Statistix 3.1; Analytical Software, St. Paul, MN, USA) to test the effects of the dietary treatments. Prior to the one-way ANOVA, normality and homogeneity of variances were checked. When a significant treatment effect was observed, an LSD post hoc test was used to compare the means. Treatment effects were considered significant at the p < 0.05 level.

Table 4 shows the growth performance and feed utilization of juvenile olive flounder fed different experimental diets for 8 weeks. At the end of the feeding trial, WG and SGR of fish fed the Pro-A diet were significantly higher greater those of fish fed the CON diet (p > 0.05). The FE of fish fed the Pro-B diet was significantly greater than that of fish fed the CON, OTC, or AMO diets (p > 0.05). The PER of fish fed the Pro-A diet was significantly greater than that of fish fed the CON and AMO diets (p > 0.05). Moreover, probiotics supplemented diets were not significantly different from fish fed the Pro-A, Pro-B and Pro-C diets (p > 0.05) in terms of growth performance. Furthermore, there were no significant differences (p > 0.05) in terms of SR, CF, HSI or VSI among fish fed the experimental diets.

There were no significant differences (p > 0.05) in terms of crude protein, lipid, moisture or ash content among any of the group of fish fed the experimental diets for 8 weeks (Table 5).

The blood parameters of the juvenile olive flounder fed the experimental diets are shown in Figure 1. The serum ALT level in the fish fed the Pro-A diet was significantly lower than those of fish fed the Pro-A, Pro-C and AMO diets (p < 0.05). There were no significant differences in AST, glucose, or total protein content among any of the group of fish fed the experimental diets for 8 weeks (p > 0.05).

Figure 2 shows the non-specific immune responses of juvenile olive flounder fed different experimental diets for 8 weeks. The serum lysozyme activities of fish fed the Pro-A, Pro-B and Pro-C diets were significantly greater than that of fish fed CON,OTC and AMO diets (p > 0.05). Myeloperoxidase activity of fish fed Pro-A diet was significantly higher than those from the CON and AMO diets (p > 0.05). However, there were no significant differences in MPO activities among fish fed the CON, Pro-B, Pro-C, OTC and AMO diets.

The gene expression profiles of the immunological parameters in the intestine of olive flounder fed diets supplemented with probiotics are presented in Figure 3. The mRNA expression levels of the FGH in fish fed the Pro-A and Pro-B diets were significantly greater than those in fish fed the CON, Pro-C, OTC and AMO diets (P < 0.05). However, there were no significant differences in FGH concentrations in fish fed the CON and OTC diets (p > 0.05). The IL-1β expressions in fish fed the Pro-A, Pro-B and Pro-C diets were significantly greater than those in fish fed the CON and AMO diets (P < 0.05). The IL-10 expressions in fish fed the Pro-A and Pro-C diets were significantly greater than those in fish fed the CON, OTC and AMO diets (P < 0.05). However, there were no significant differences in IL-10 mRNA expressions in fish fed the CON and AMO diets (p > 0.05).

Histological analysis of the anterior intestine of olive flounder fed different experimental diets for 8 weeks is shown in Figure 4. The intestinal villi length in fish fed the Pro-A, Pro-B and Pro-C diets had significantly greater villi lengths than the fish fed the CON, OTC and AMO diets (Figure 4A). The fish fed the Pro-A, Pro-B, Pro-C, OTC and AMO diets clearly exhibited better intestinal histomorphology with more massive villi in comparison to the CON diet (Figure 4B). In addition, the images which corresponds to the CON group, shows certain irregular shape and improper arrangement of villi in fish fed the CON, OTC and AMO diets compared to those of the Pro-A, Pro-B and Pro-C groups.

The percent cumulative survival of juvenile olive flounder challenged with E. tarda for 15 days is shown in Figure 5. During the challenge test, the first mortalities occurred on the second day. At the end of the 15 day challenge, the percent cumulative survivals of fish fed the Pro-A, Pro-B, and Pro-C diets were significantly greater than that of fish fed the CON diet (p < 0.05). However, there were no significant differences in cumulative survival rates in fish fed the Pro-A, Pro-B, Pro-C, OTC, and AMO diets (p > 0.05).