The Effects of Nisin-Producing Lactococcus lactis Strain Used as Probiotic on Gilthead Sea Bream (Sparus aurata) Growth, Gut Microbiota, and Transcriptional Response

The present research tested the effects of dietary nisin-producing Lactococcus lactis on growth performance, feed utilization, intestinal morphology, transcriptional response, and microbiota in gilthead sea bream (Sparus aurata). A feeding trial was conducted with fish weighting 70–90 g. Fish were tagged with passive, integrated transponders and distributed in nine 500 L tanks with 40 fish each. Fish were fed for 12 weeks with either a control (diet A) or experimental diets (diets B and C) in triplicate (3 tanks/diet). Extruded pellets of diets B and C were supplemented with a low (2 × 109 CFU/kg) and a high (5 × 109 CFU/kg) dose of probiotic, respectively. No significant differences were found between groups for the feed conversion ratio or specific growth rates. However, the final body weight of fish fed diet C was significantly higher than the control group with intermediate values for fish fed diet B. Histological analysis conducted using a semi-quantitative scoring system showed that probiotic did not alter the morphology of the intestine and did not trigger inflammation. With regard to the transcriptomic response, a customized PCR array layout was designed to simultaneously profile a panel of 44 selected genes. Significant differences in the expression of key genes involved in innate and acquired immunity were detected between fish fed probiotic and control diets. To analyze the microbiota associated to the feeds and the gut autochthonous microbial communities, we used the Illumina MiSeq platform for sequencing the 16S rRNA gene and a metagenomics pipeline based on VSEARCH and RDP databases. The analysis of gut microbiota revealed a lack of colonization of the probiotic in the host’s intestinal mucosa. However, probiotic did modulate the fish gut microbiota, confirming that colonization is not always necessary to induce host modification. In fact, diets B and C were enriched with Actinomycetales, as compared to diet A, which instead showed a higher percentage of Pseudomonas, Sphyngomonas, and Lactobacillus genera. These results were confirmed by the clear separation of gut bacterial community of fish fed with the probiotic from the bacterial community of control fish group in the beta-diversity and PLS-DA (supervised partial least-squares discriminant analysis) analyses.

The present research tested the effects of dietary nisin-producing Lactococcus lactis on growth performance, feed utilization, intestinal morphology, transcriptional response, and microbiota in gilthead sea bream (Sparus aurata). A feeding trial was conducted with fish weighting 70-90 g. Fish were tagged with passive, integrated transponders and distributed in nine 500 L tanks with 40 fish each. Fish were fed for 12 weeks with either a control (diet A) or experimental diets (diets B and C) in triplicate (3 tanks/diet). Extruded pellets of diets B and C were supplemented with a low (2 × 10 9 CFU/kg) and a high (5 × 10 9 CFU/kg) dose of probiotic, respectively. No significant differences were found between groups for the feed conversion ratio or specific growth rates. However, the final body weight of fish fed diet C was significantly higher than the control group with intermediate values for fish fed diet B. Histological analysis conducted using a semi-quantitative scoring system showed that probiotic did not alter the morphology of the intestine and did not trigger inflammation. With regard to the transcriptomic response, a customized PCR array layout was designed to simultaneously profile a panel of 44 selected genes. Significant differences in the expression of key genes involved in innate and acquired immunity were detected between fish fed probiotic and control diets. To analyze the microbiota associated to the feeds and the gut autochthonous microbial communities, we used the Illumina MiSeq platform for sequencing the 16S rRNA gene and a metagenomics pipeline based on VSEARCH and RDP databases. The analysis of gut microbiota revealed a lack of colonization of the probiotic in the host's intestinal mucosa. However, probiotic did modulate the fish gut microbiota, confirming that colonization is not always necessary to induce host modification. In fact, diets B and C were enriched with Actinomycetales, as compared to diet A, which instead showed a

INTRODUCTION
The definition of "Probiotics" has changed many times during this century. However, according to (Food and Agriculture Organization of the United [FAO] and World Health Organisation [WHO], 2001) probiotics are "live microorganisms that confer a health benefit on the host when administered in adequate amounts." The most commonly used probiotics are bacteria belonging to Lactobacillus, Bifidobacterium, Bacillus, and Enterococcus genera (European Medicines Agency [EMA], and European Food Safety Authority [EFSA], 2017; EFSA FEEDAP [EFSA Panel on Additives and Products or Substances used in Animal Feed] et al., 2018), but some fungal genera have also been reported as novel probiotics.
In the last 25-30 years, the use of probiotics in animal production has increased (Chaucheyras-Durand and Durand, 2010;Ezema, 2013). Indeed, several publications have reported numerous beneficial effects associated with the supplementation of live yeast or bacteria (mostly Lactobacillus) in the diet of terrestrial animals, including amelioration of resistance to pathogens, improvement in growth parameters (in swine and poultry), increase in productivity and quality of eggs in laying hens, and enhancement of milk production in cattle (Gallazzi et al., 2008;Shabani et al., 2012;Puphan et al., 2015;Uyeno et al., 2015;De Cesare et al., 2017;Wang et al., 2017;Dowarah et al., 2018;Forte et al., 2018).
In aquaculture, a great number of bacterial species are currently used as probiotics (for a review, please see Newaj-Fyzul et al., 2014). These microorganisms can be administered as multi-species (multi-strain) or single-species (single-strain) (Food and Agriculture Organization of the United [FAO], 2016) and provided either as a suspension in water, or added to the feed. However, use in feed is considered the best option; therefore, this approach is employed most frequently (Nayak, 2010;Jahangiri and Esteban, 2018). In the European Union (EU), probiotic strains, must obtain a market authorization by the EFSA (European Safety Food Authority) 1 , which grants a QPS (Qualified Presumption of Safety) status. The QPS is based on reasonable evidence. No microorganism belonging to a QPS status group needs to undergo a full safety assessment, but microorganisms that pose a safety concern to humans, animals, or environment are not considered suitable for QPS status and must undergo a full safety assessment. The QPS assessment requires: (1) the identity of the strain to be conclusively established, and (2) absence of resistance to antibiotics (for bacteria) or antimycotics (for yeasts) used in 1 https://www.efsa.europa.eu/en human and veterinary medicine (EFSA Panel on Biological Hazards (BIOHAZ) et al., 2020).
The increase in the use of probiotics in aquaculture is mostly related to the need to decrease or even avoid the use of antibiotics, increasing at the same time the sustainability of the aquaculture industry. The negative effects of antibiotics overuse include the accumulation of residue in the aquatic environment, particularly in the marine sediments where antibiotics can persist for months, favoring the selection of multi-antibiotic-resistant bacterial strains. Indeed, there is an increasing risk that antibiotic-resistant bacteria, initially derived from food-producing animals, could render the latest generation of antibiotics virtually ineffective for humans (Cabello, 2006;World Health Organisation [WHO] et al., 2006). Another negative outcome of antibiotics being used as growth promoters in cultured fish is the reduction of biodiversity and quantity of indigenous gut microbiota, which can impair fish immune responses (Borch et al., 2015).
For these reasons, the use of antibiotics as growth promoters in animal production has been fully banned in the EU since 2006 (Casewell et al., 2003;European Parliament  , 2017) and many research efforts have been undertaken to replace them with probiotics for animal health management (Ezema, 2013).
Several studies have demonstrated that probiotics can reduce pathogenic bacteria due to direct competition-colonizing dynamics, through which microorganisms can partition spatial niche habitats in the intestinal mucosa (Balcázar et al., 2007b;Sugimura et al., 2011). Probiotics can also produce inhibitory molecules, such as bacteriocins, siderophores, enzymes, and hydrogen peroxide, or inhibit pathogenic bacteria by decreasing the intestinal pH through the release of organic acids (Ringø, 2008;Zhou X. et al., 2010;Ustyugova et al., 2012;Perez et al., 2014;Dahiya et al., 2020).
In addition, probiotics enhance the host immune system by generating systemic and/or local responses (Balcázar et al., 2006b;Salinas et al., 2008) that include activation of various antioxidant pathways and an increase in several innate immune parameters, such as phagocytosis, lysozyme levels, respiratory burst peroxidase and antiprotease activity, cytokine production, and white blood cell count (Nayak, 2010;Lazado and Caipang, 2014;Newaj-Fyzul et al., 2014;Simó-Mirabet et al., 2017).
In cultured fish, probiotics improve fish growth and feed conversion rates, too, due to an increase in feed digestibility and absorption of nutrients (Dimitroglou et al., 2011;Martínez Cruz et al., 2012). These effects stem from the capacity of probiotics to secrete enzymes, such as proteases, amylases, and lipases that hydrolyze molecules, which the fish intestine cannot otherwise digest (Balcázar et al., 2006b;Abd El-Rhman et al., 2009). Furthermore, the use of probiotics can restore the eubiotic state of the intestinal microbiota after antibiotic treatment or a pathogenic insult or can help maintain gut microbiota homeostasis, even in larval stages, when vaccination is difficult (Abdelhamid et al., 2009;Borch et al., 2015).
According to the above findings, the aim of the present research was to evaluate the effects of the lactic acid bacteria Lactococcus lactis subsp. lactis SL242, used as feed additive, on growth performance, feed utilization, intestinal morphology, transcriptional response, and microbiota in gilthead sea bream (Sparus aurata).
The probiotic strain L. lactis subsp. lactis SL242 was selected due to important characteristics of Lc. lactis in general and SL242 in particular. Lc. lactis are mesophilic lactic acid bacteria that are present in the intestinal microbiota of fish (Tarnecki et al., 2017;Ringø et al., 2020) and can adapt to the water temperature of many reared fish species. Lactococci are proteolytic bacteria (Samaržija et al., 2001) that are potentially useful for improving the digestion of proteins contained in fish feed. The proteolytic system of lactococci includes a cell wall-associated proteinase and an extracellular peptidase (Samaržija et al., 2001). Furthermore, SL242 produces the antibiotic nisin A (Malvisi et al., 2016), which can inhibit or kill vegetative cells and bacterial spores (European Safety Food Authority [EFSA], 2005). Due to its antibacterial activity, nisin is of great interest in aquaculture. Nisin-susceptible bacterial species are found among Bacillus, Clostridium, Listeria, Staphylococcus, Streptococcus, and Vibrio genera (European Safety Food Authority [EFSA], 2005;Malvisi et al., 2016;Hamid et al., 2020), including known aquatic pathogens, such as V. parahaemolyticus, and V. alginotlyticus (Hamid et al., 2020). Lc. lactis probiotics have also shown inhibitory action against Yersinia rukeri and Aeromonas salmonicida, which can affect fish growth (Balcázar et al., 2007a(Balcázar et al., , 2006b. Furthermore, Lc. lactis probiotic has been effective against Aeromonas hydrophila in Oreochromis niloticus (Zhou X. et al., 2010).

Ethics Statement
Procedures for fish manipulation and tissue collection were carried out according to the Spanish (Royal Decree RD53/2013) and the current EU legislation (2010/63/EU) for handling of experimental fish. All procedures were approved by the Ethics and Animal Welfare Committees of Institute of Aquaculture Torre de la Sal (IATS-CSIC, Castellón, Spain) (Permit number 824/2019) and "Generalitat Valenciana" (permit number 2019/VSC/PEA/0197).

Animals
On June 2019, juveniles of gilthead sea bream were purchased from a Mediterranean hatchery (Piscimar, Burriana, Spain) and adapted for more than 2 months to the indoor experimental facilities of IATS-CSIC, under natural photoperiod and temperature conditions (40 • 5 N; 0 • 10 E). Seawater was pumped ashore (open system); oxygen content of water effluents was always above 85% saturation, and unionized ammonia remained below 0.02 mg/L. During the acclimation and experimental period, water temperature increased from 20-22 • C in June to 28 • C in August, decreasing thereafter from 24-25 • C in mid-September to 13-16 • C in December.

Diets
Extruded pellets of a control (diet A) and two experimental diets (diets B and C) were manufactured by VRM Srl Naturalleva (Verona, Italy), mimicking commercial fish feed formulations with traditional vegetable proteins and oils as the main replacers of fishmeal and fish oil, respectively ( Table 1). The mash of each diet was extruded using a single-screw extruder (X-165, Wenger United States). To ensure product stability, the probiotic was homogenized with the dietary oil and included by vacuum coating (La Meccanica vacuum coater, Italy) during the postextrusion process. During the vacuum process, only dry basal extruded pellets of diets B and C were supplemented with 2.5 and 6.2 g/100 kg of L. lactis subsp. lactis SL242, corresponding to a probiotic dosage of 2 × 10 9 CFU/kg (low dose) and 5 × 10 9 CFU/Kg (high dose), respectively. Sacco S.r.l [Cadorago (Co), Italy] provided the probiotic strain.
The two doses were chosen on the basis of our experience and literature data (Villamil et al., 2002;Adel et al., 2017) in order to verify the most effective one. They are also in line with dosages that could be used commercially in a cost-effective manner.
The final feeds were stored in a refrigerated room (6-7 • C) for the entire duration of the feeding trial. A preliminary stability study of SL242 in the feed supplemented with probiotic was conducted for 12 weeks (the duration of the experiment), at 6 • C. At the end of this period, the average loss of viability determined by plate count resulted about 50%, consistent with our expectations. Although further improvement may be warranted for a commercial probiotic product, at this stage of the process, the observed stability is considered acceptable.

Feeding Trial
In September 2019, fish weighing 70-90 g were randomly distributed in nine 500 L tanks to establish triplicate groups of 40 fish each (initial rearing density, 6.6-6.7 kg/m 3 ). All fish were tagged with PIT (passive integrated transponders) (ID-100A 1.25 Nano Transponder, Trovan) in the dorsal skeletal muscle. Fish were individually weighed and measured at initial, intermediate, Non-starch polysaccharides 18.5 Diet B and C were formulated with the addition of probiotic (5 × 10 6 CFU/g feed). a Vitamin premix (IU or mg/kg diet): DL-α tocopherol acetate 60 IU; sodium menadione bisulfate 5 mg; retinyl acetate 15,000 IU; DL-cholecalciferol 3,000 IU; thiamine 15 mg; riboflavin 30 mg; pyridoxine 15 mg; vitamin B 12 0.05 mg; nicotinic acid 175 mg; folic acid 500 mg; inositol 1,000 mg; biotin 2.5 mg; calcium pantothenate 50 mg. b Mineral premix (g or mg/kg of diet) bi-calcium phosphate 500 g, calcium carbonate 215 g, sodium salt 40 g, potassium chloride 90 g, magnesium chloride 124 g, magnesium carbonate 124 g, iron sulfate 20 g, zinc sulfate 4 g, copper sulfate 3 g, potassium iodide 4 mg, cobalt sulfate 20 mg, manganese sulfate 3 g, sodium fluoride 1 g. and final sampling points (every 4 weeks), by using a FR-200 Fish Reader W (Trovan, Madrid, Spain) for data capture and preprocessing. The trial lasted 12 weeks (October 2019-December 2019). Fish were hand-fed once daily (12 a.m.), 5-6 days per week to visual satiety with either control or experimental diets for the entire duration of the trial. Feed intake and mortalities (<1%) were recorded daily and normal fish behavior was assessed routinely by camera monitoring.

Sample Collection
At the end of the feeding trial, four fish per replicate (12 fish/diet) were anesthetized with 0.1 g/L of tricaine-methasulfonate (MS-222, Sigma-Aldrich) and then sacrificed by severing the spinal cord. The intestine (excluding the pyloric ceca) of each fish was dissected out, weighed, and measured aseptically to calculate the intestine weight index (IWI) and intestine length index (ILI). Then, anterior (AI) and posterior (PI) intestine tissue portions (∼0.4 cm) were put either into RNAlater, or in 10% neutral buffered formalin for subsequent molecular (AI) and histological (AI, PI) analyses. The remaining part of AI was opened and washed with sterile Hank's balanced salt solution before collecting the autochthonous intestinal bacteria by scraping intestinal mucosa with the blunt end of a clean scalpel. Then, mucus samples were transferred to a sterile Eppendorf tube and stored in ice until subsequent (within 2 h) DNA extraction for microbiota analysis.
To characterize feed-associated bacterial communities, two samples of 200 mg each from each feed were taken at the end of the trial and used for bacterial DNA extraction and sequencing.

Histological Analysis
Fixed samples of AI and PI were dehydrated in ethanol solutions with gradually increasing concentrations and then, embedded in paraffin. Sections of 5 µm were obtained with a microtome (Leica RM2245) and stained with hematoxylin and eosin (H&E), following standard histological protocols. The sections were examined under a stereomicroscope Eurotek Tecno NB50T (Orma Srl, Milan, Italy) and photographed with a digital camera Eurotek CMOS MDH5 (Orma Srl, Milan, Italy). Based on previous studies (Knudsen et al., 2007;Uran et al., 2008;Urán et al., 2009;Khojasteh, 2012), the semi-quantitative scoring system focused on five different gut morphological parameters (mucosal folds, connective tissue, lamina propria of simple folds, and supranuclear vacuoles). Histological alterations of each morphological parameter were classified using a score value ranging from one (normal condition) to five (severe alteration). The final values, obtained by the sum of score values for each parameter, were then used to classify the severity of the morphological damage by using a class-based scoring system: Class I (values ≤ 10)-normal tissue structure with slight histological alterations; Class II (values 11-15)-moderate histological alterations; and Class III (values > 15)-severe histological alterations of the organ.

Gene Expression Analysis
Total RNA from AI was extracted using a MagMax-96 total RNA isolation kit (Life Technologies, Carlsbad, CA, United States). The RNA yield was higher than 3.5 µg with absorbance measures (A260/280) of 1.9-2.1. cDNA was synthesized with the High-Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA, United States), using random decamers and 500 ng of total RNA in a final volume of 100 µL. Reverse transcription (RT) reactions were incubated 10 min at 25 • C and 2 h at 37 • C. Negative control reactions were run without the enzyme. As reported previously (Estensoro et al., 2016), a customized PCR array layout was designed to simultaneously profile a panel of 44 selected genes, including markers of epithelial integrity (11), nutrient transport (4), mucins (3), cytokines (9), immunoglobulins (2), cell markers and chemokines (7), and pattern recognition receptors (8) ( Table 2). qPCR reactions were performed using an iCycler IQ Real-Time Detection System (Bio-Rad, Hercules, CA, United States). Diluted RT reactions (×6) were used for qPCR assays in a 25 µL volume in combination with a SYBR Green Master Mix (Bio-Rad, Hercules, CA, United States) and specific primers at a final concentration of 0.9 µM (Supplementary Table 1). The program used for PCR amplification included an initial denaturation step at 95 • C for 3 min, followed by 40 cycles of denaturation for 15 s at 95 • C and annealing/extension for 60 s at 60 • C.
All the pipetting operations were executed by means of an EpMotion 5070 Liquid Handling Robot (Eppendorf, Hamburg, Germany) to improve data reproducibility. The efficiency of PCRs (>92%) was checked, and the specificity of reactions was verified by analyzing the melting curves (ramping rates of 0.5 • C/10 s over a temperature range of 55-95 • C), and linearity of serial dilutions of RT reactions (r 2 > 0.98). Fluorescence data acquired during the extension phase were normalized by the delta-delta CT method (Livak and Schmittgen, 2001), using beta-actin as housekeeping gene due to its stability in different experimental conditions (average CT between experimental groups varied less than 0.2).

Bacterial DNA Extraction
The bacterial DNA was extracted from feeds (2 samples/feed) and from intestinal samples (7-10 fish/dietary group). Intestinal mucus samples (200 µl) were treated with 250 µg/ml of lysozyme (Sigma) for 15 min at 37 • C. Then, DNA was extracted using the High Pure PCR Template Preparation Kit (Roche) following the manufacturer's instructions. DNA concentration, quality, and purity were measured using a NanoDrop 2000c (Thermo Fisher Scientific) and agarose gel electrophoresis (1% w/v in Tris-EDTA buffer). Samples were stored at −20 • C until sequencing. The same procedure was used to extract DNA from the control and experimental feeds (previously ground to a fine powder) to evaluate the concentration of the probiotic supplement.

Illumina MiSeq Sequencing and Bioinformatic Analysis
The V3-V4 region of the 16S rRNA gene (reference nucleotide interval 341-805 nt) was sequenced using the Illumina MiSeq system (2 × 300 paired-end run) at the Genomics Unit from the Madrid Science Park Foundation (FPCM, Spain). The details on the PCR and sequencing of amplicons have been described elsewhere (Piazzon et al., 2019). Raw sequence data were uploaded to the NCBI (National Center for Biotechnology Information) and Sequence Read Archive (SRA) under NCBI BIOPROJECT ID: PRJNA679278; NCBI BIOSAMPLE ID: SAMN16828235-61; and SRA ACCESSION: SRR13081673-99. Raw forward and reverse reads were quality filtered using FastQC 2 , and pre-processed using Prinseq (Rahlwes et al., 2019). Terminal N bases were trimmed at both ends and sequences with >5% of total N bases were discarded. Reads that were <150 bp long with a Phred quality score <28 in both of the sequence ends and with a Phred average quality score <26 were excluded. Then, forward and reverse reads were merged using fastq-join (Aronesty, 2013). Bacterial taxonomy was assigned using the Ribosomal Database Project (RDP) release 11 as a reference database (Cole et al., 2014). Reads were aligned with a custom-made pipeline using VSEARCH and BLAST (Altschul et al., 1990;Rognes et al., 2016). Alignment was performed establishing high stringency filters (≥90% sequence identity, ≥90% query coverage). Taxonomic assignment results were filtered and data were summarized in an Operational Taxonomic Units (OTUs) table. Sample depths were normalized by total sum scaling and then made proportional to the total sequencing depth, following previously described recommendations (McKnight et al., 2019). Species richness estimates and alpha diversity indexes were calculated using the R package Phyloseq (Mcmurdie and Holmes, 2013). Rarefaction curves were obtained by plotting 2 http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ the number of observed taxonomic assignations in an OTU table against the number of sequences in each sample using the R package phyloseq.

Inferred Metagenome and Pathway Analysis
Piphillin was used to normalize the amplicon data by 16S rRNA gene copy number and to infer the metagenomics content (Iwai et al., 2016). This analysis was performed with the OTUs significantly driving the separation by probiotic in the PLS-DA analysis (described in the section "Statistics"). For the analysis, a sequence identity cut-off of 97% was implemented, and the inferred metagenomics functions were assigned using the Kyoto Encyclopedia of Genes and Genomes database (KEGG, Oct 2018 Release). Raw KEGG pathway output from Piphillin was analyzed with the R Bioconductor package DESeq2 using default parameters, after flooring fractional counts to the nearest integer (Love et al., 2014;Bledsoe et al., 2016;Piazzon et al., 2019). Comparisons were also performed between different diets to evaluate possible pathway differences across diets.

Statistics
Data on growth and gene expression were analyzed by oneway ANOVA using SigmaPlot v14 (Systat Software Inc., San Jose, CA, United States). Normality of the data was verified by Shapiro-Wilk test, and Dunn's post hoc test was used for multiple comparisons between groups. For analysis of qualitative histological data, we conducted the non-parametric Kruskall-Wallis test, followed by Dunn's test for the multiple comparisons. GraphPad Prism8 (GraphPad Software, Inc., La Jolla, CA, United States) was used for both analyses. Microbiota species richness, alpha diversity indexes, and phylum abundance between experimental groups were determined by Kruskal-Wallis test followed by Dunn's post hoc test. Beta diversity was tested with permutational multivariate analysis of variance (PERMANOVA), using the non-parametric method adonis from the R package Vegan with 10,000 random permutations. To further study microbiota differences between dietary groups, supervised partial least-squares discriminant analysis (PLS-DA) and hierarchical clustering of samples were sequentially applied using EZinfo v3.0 (Umetrics, Umea, Sweden) and hclust function (gplots R package), respectively. Hotelling's T 2 statistic was calculated by employing the multivariate software package, whereby points above the 95% confidence limit for T 2 were considered as outliers and discarded. Values of normalized counts of OTUs present in 3 or more samples were included in the analyses, and the significant contribution to the group separation was determined by the minimum variable importance in the projection (VIP) values (Wold et al., 2001;Li et al., 2012), which renders an accurate clustering using the average linkage method and Euclidean distance feasible. The quality of the PLS-DA model was evaluated by the parameters R2Y (cum) and Q2 (cum), which indicate the fit and prediction ability, respectively. To assess whether the supervised model was being overfitted, a validation test consisting on 600 random permutations was performed using SIMCA-P + (v11.0, Umetrics).

Growth Performance
Data on growth performance, feed intake, and feed conversion ratio (FCR) are reported in Table 3. All fish grew efficiently during the first 30 days of the trial (FCR = 1.27-1.28), reaching an overall FCR of 1.55-1.60 at the end of trial. The decrease in the length of the day and temperature from October to December should be noted.
No statistically significant differences were found between groups for the condition factor and specific growth rates (SGR), although the highest SGR tended to be achieved in fish fed diet C (high dose of probiotic). Indeed, the final body weight of these animals was higher than in the control group (diet A) (P < 0.05) with intermediate values for fish fed diet B (low dose of probiotic). Thus, total weight gain varied from 97% in fish fed diet A to 106% in fish fed diet C.

Histological and Biometric Scoring
Histological analysis of gilthead sea bream intestine was performed according to the aforementioned morphological criteria. The intestinal scoring data are reported in Table 4. The AI (Figures 1A-C) and PI (Figures 1D-F) portions were not affected by probiotic administration. Although the mucosal folds of the PI were significantly different (P < 0.05) between groups fed diets A and B, the total scores, calculated for each group, fall within an evaluation of Class I. In particular, the simple and complex folds appeared thin and regularly branched, lamina propria and connective tissue appeared normally proportioned and supranuclear vacuoles were numerous and well-distributed. Regarding the index of intestine length (ILI) ( Table 4), diet B showed a significantly lower ILI than the control group (diet A) (P < 0.05), but no differences were observed between the other groups. No differences in the intestine weight index (IWI) were observed between groups.

Gene Expression Profiling
All genes included in the PCR-array were found at detectable levels with the highest expression level for markers of nutrient transport (alpi, fabp1, and fabp2), epithelial integrity (cx32.2), mucus production (muc2, muc13) and pattern recognition receptors (fcl) (Supplementary Table 2). Regarding the probiotic effect, statistically significant changes were found in the expression patterns of 5 out of 44 genes (P < 0.05) (Figure 2). In particular, expression of interleukin 10 (il10), interleukin (il12), and toll-like receptor 2 (tlr2) was upregulated in fish fed diet C (high probiotic dose) with intermediate values (not statistically different from the control group) in fish fed diet B (low probiotic dose). In contrast, the highest values of toll-like receptor 5 (tlr5) and galectin-8 (lgals8) were seen in fish fed diet B, whereas intermediate values were found in fish fed diet C. The probiotic treatment altered other markers (desmoplakin, dsp; interleukin 34, il34; C-C chemokine receptor 3, ccr3; and macrophage mannose receptor 1, mrc1) to a lesser extent, with an overall enhancement of gene expression that was especially evident in fish fed diet C (P < 0.1). Data are reported as mean ± SEM, different superscript letters indicate significant differences (P < 0.05) between diet groups in the same sub-column. 1 Weight gain, WG = (100 × body weigh increase)/initial body weight. 2 Specific growth rate, SGR = 100 × (ln final body weight-ln initial body weight)/days. 3 Condition factor, CF = 100 × (body weight/standard length). 4 Feed conversion ratio, FCR = dry feed intake/wet weight gain [total feed supplied (g DM, dry matter)/WG (g)].

Characterization of Feed-Associated Bacterial Communities
At the end of the trial, the normalized counts of L. lactis subsp. lactis resulted 8-11 in diet A (<0.0001% total bacterial counts); 30,204 in diet B (2.5% total counts); and 61,828 (5.4% total counts) in diet C (Figure 3). By excluding Cyanobacteria/Chloroplast (>90% total counts), Firmicutes and Proteobacteria proved to be the most highly represented bacterial phyla in the three feeds, whereas the rest of the bacterial population consisted of Bacteriodetes and Fusobacteria phyla (Supplementary Figure 1A). However, the percentage of Firmicutes varied considerably between feeds, with higher values in feed B (4.2%) and C (7.8%) than in the control feed, in which Firmicutes represented only 2% of the total counts. Thus, by recalculating the relative bacterial abundances after excluding Cyanobacteria/Chloroplast, the percentage of Firmicutes rose from 34% in the control diet A to 70% in diet B and 79% in diet C (Supplementary Figure 1B). Then, by specifically analyzing the relative abundance of the probiotic L. lactis subsp. lactis in comparison to the most representative genera within the phylum Firmicutes, the percentage of L. lactis subsp. lactis was close to 0% in the control diet, whereas in B and C diets, it was significantly higher, reaching values of 64 and 71%, respectively (Supplementary Figure 1C).

Alpha Diversity and Gut Microbiota Composition
Illumina sequencing of AI-adherent bacteria yielded 3,677,860 high-quality and merged reads, with an average value of 136,217 reads per sample (Supplementary Table 3). When annotated, the  reads were assigned to 1,313 OTUs at 97% identity threshold. Rarefaction analysis showed curves that approximated saturation (horizontal asymptote); thus, a good coverage of the bacterial community was achieved and the number of sequences for analysis was considered appropriate (Supplementary Figure 2). Indeed, up to 85% of the OTUs were classified at the level of species and more than 90% at the level of genus (94.1%), family (96%), order (97%), class (97.2%), and phylum (99%). As shown in Table 5, the richness estimator (ACE) indicated a higher OTU richness in fish fed diet B than in fish fed diet A or diet C. At the same time, alpha diversity estimators (Shannon and Simpson) disclosed a reduced evenness in fish fed diet C, which indicates that abundant OTUs predominated over the others in this group of fish. Changes in bacterial composition were also found at the phylum level (Figure 4). Proteobacteria was the most abundant phylum in the three groups, ranging from 55.9% in fish fed diet C to 55.7% in fish fed diet A, and 50.1% in the diet B fed group. The second-most abundant phylum was Firmicutes, representing the 26.6% of the OTU counts in fish fed diet A, decreasing progressively with the probiotic supplementation in fish fed diet B (26.2%) and diet C (5.6%). The same trend was shown by the phylum Bacteroidetes, ranging from 2.7% in fish fed diet A to 1.3% in fish fed diet B and 0.1% in fish fed diet C. The phylum Actinobacteria increased from 9.3% in fish fed diet A to 16.3% in fish fed diet B but decreased to its minimum level in group C (3.2%). Finally, Spirochetes appeared in a significant proportion (32%) only in fish fed with diet C, being practically absent in the other groups (<3%).

Beta Diversity, Discriminant Analysis, and Inferred Pathways
No significant differences in beta diversity were found when experimental groups were computed independently (PERMANOVA, P = 0.34, F = 1.031, R 2 = 0.04). In contrast, when B and C groups were computed together, beta diversity became statistically significant (PERMANOVA, P = 0.032, F = 1.8789, R 2 = 0.099). Taking this analysis further, a PLS-DA model was constructed with a 99% to the total variance explained (Figure 5). During the statistical processing to construct the model, two fish from the Diet A group and one fish from the Diet C group appeared as outliers and were excluded from the model. This approach displayed a clear separation of control fish and fish fed probiotic diets (B + C group) along component 1 (84.52%) with a higher individual variability within fish fed diet B than in those fed diet C. This PLS-DA model was successfully validated with a permutation test (pCV ANOVA = 0.015) discarding the possibility of over-fitting of the supervised model (Supplementary Figure 3).
Differences between control fish and the probiotic-fed merged groups were driven by 81 OTUs (VIP > 1), mainly belonging to the phyla Proteobacteria, Spirochetes, and Firmicutes. A detailed list of the VIPs can be found in Supplementary  Table 3. The inferred metagenomic analysis using DESeq2 disclosed nine differentially abundant pathways across groups ( Figure 6). Pathways related to protein digestion and absorption, as well as renin secretion were over-represented in the probiotic fed fish groups, whereas the control group showed a relative preponderance of pathways related to shigellosis, proteasome and autophagy.

DISCUSSION
In aquaculture the use of probiotics is significantly increasing and a growing number of studies are demonstrating their positive effects in the most economically important fish species (Merrifield et al., 2010;Varela et al., 2010;Mahdhi, 2012;Ridha and Azad, 2012;Chauhan and Singh, 2019).
As mentioned previously, one of the most interesting effects of probiotics is the increase in the animals' growth performance (Sun et al., 2012;Nguyen et al., 2017;Won et al., 2020). In the present study, gilthead sea bream fed diets C and B, supplemented with high and low doses of L. lactis subsp. lactis, respectively, reached a higher final biomass than control fish fed with diet A, and differences in biomass gain were statistically significant between groups C and A. Although differences between fish groups arose at the December sampling, most of the weight gain was attained during September-October, as this period still corresponds to the active fish feeding behavior at IATS-CSIC latitude. This result highlights, albeit slightly, the beneficial action of the probiotic, suggesting a more efficient digestion and utilization of nutrients in gilthead sea bream fed probiotics. Indeed, although no significant differences were detected in FCR and SGR between dietary groups, the lowest FCR (1.60 ± 0.03) and the highest SGR (0.85 ± 0.02) were registered in fish fed diet C. Similar results were obtained in gilthead sea bream by Suzer et al. (2008) and Varela et al. (2010), using Lactobacillus spp. and Shewanella putrefaciens Pdp11, respectively. Positive results in fish growth performance, using L. lactis as probiotic, were also obtained in other cultured fish species, such as common carp, European sea bass, tilapia, and olive flounder (Balcázar et al., 2006a;Carnevali et al., 2006;Heo et al., 2013;Xia et al., 2018;Feng et al., 2019).
Histological analysis was conducted using a semi-quantitative scoring system. The parameters taken into account for the AI and PI morphological evaluation were related to the mucosal folds that represent the intestinal absorptive surface area, and to the associated connective tissue (Dimitroglou et al., 2011;Khojasteh, 2012;Puphan et al., 2015). Our results confirmed that probiotic did not alter the morphology of the gut and did not trigger intestinal inflammation. Indeed, no structural modifications were detected in fish fed with diets supplemented with probiotic (diets B and C), in comparison to the control group fed diet A. In line with our results, other studies have shown that probiotics improve gut morphology, leading to an increase in intestinal absorption capacity (Batista et al., 2016;Won et al., 2020). In contrast, Cerezuela et al. (2012; reported several negative effects related to the administration of probiotics in gilthead sea bream. In particular, those authors showed that both Tetraselmis chuii and Bacillus subtilis induced intestinal inflammation with numerous signs of edema in the FIGURE 4 | Stacked bar chart representing the relative abundance (%) of bacterial phyla in fish fed control (A) and experimental (B and C) diets. The Kruskal-Wallis test (Dunn's post hoc test, P < 0.05) showed significant differences between groups for the phyla Firmicutes, Actinobacteria and Spirochetes. The differences are indicated by different letters in parenthesis that correspond to pairwise comparisons within each phylum between groups. mucosal folds. Therefore, more in-depth histological analyses are needed to better understand the effects of different probiotic strains on the adsorptive surface area in fish intestine and, in particular, on the villi length and density.
Numerous studies that have investigated the effects of probiotics on the piscine immune system have reported an enhanced immune response, thus improving survival rates and resistance to a pathogenic attack (Nayak, 2010;Lazado and Caipang, 2014). Different probiotic strains stimulate the immune system in fish, but the effect appears to be species-specific. L. lactis supplementation increased the concentration of several pro-and anti-inflammatory cytokines (Tnfα, Il1β, Il6, Il12, Il10 and Tgfβ) in common carp serum (Feng et al., 2019) and upregulated the expression of tnf α, ifnγ, hsp70, and il1β genes in the intestine of tilapia (Xia et al., 2018;Won et al., 2020). Conversely, L. lactis did not induce any differences in the abundance of cytokines and pattern recognition receptors (PRRs) transcripts in intestine or head kidney of trout (Pérez-Sánchez et al., 2011). In gilthead sea bream, the anti-inflammatory action of a Bacillus-based probiotic induced decreased expression of lgals8 and cd4 transcripts in anterior intestine, lower amounts of circulating IgM and cortisol, a lower respiratory burst activity of blood leukocytes, and lower numbers of eosinophilic granulocytes (in particular, mast cells) in the intestinal submucosa . Herein, significant differences in the expression of key genes involved in innate and acquired immunity (interleukins and PRRs) were detected between fish fed probiotic and control diets. Among the mechanisms induced by probiotics, it has been postulated that the activation of immunity derives from the interaction of the host with the probiotic microbial associated molecular patterns (MAMPs) (Yang et al., 2014). The direct effect of MAMPs was recently demonstrated by feeding grouper (Epinephelus coioides) with MAMPs isolated from the probiotic Bacillus pumilus SE5. Indeed, an activation of intestinal immunity via up-regulation of TLR signaling pathways was observed (Yang et al., 2019). Thus, the observed activation of the immune system in the present study is likely taking place by direct induction of gilthead sea bream PRRs by components on the cell wall of the probiotic, such as peptidoglycan or lipoteichoic acid, which are in fact TLR2 agonists (Dammermann et al., 2013).
The density, composition and function of intestinal microbiota of fish, including gilthead sea bream, are shaped by numerous factors, such as diet, sex, developmental stage, and rearing conditions (Piazzon et al., , 2019Rimoldi et al., 2020), as well as multiple endogenous host-microbe interactions, such as the host's genetic background (Piazzon et al., 2020), and possible intestinal disorders or intestinal diseases (Bakke-Mckellep et al., 2007;Green et al., 2013). Furthermore, microbiota vary taxonomically and functionally in different sections of the GIT of fish (Kokou et al., 2020). There is also a distinction between the allochthonous, i.e., freeliving, transient microbiota associated with the digesta (feces), and autochthonous communities that colonize the mucosal surface of the digestive tract and make up the core community (Merrifield et al., 2010;Ringø et al., 2016;Nguyen et al., 2017;Egerton et al., 2018).  Taxonomically, gut bacteria are classified according to phyla, classes, orders, families, genera, and species. The "core" intestinal microbiota, which can often persist in spite of changing factors is constituted by Proteobacteria, Firmicutes, and Actinobacteria phyla in both freshwater and marine fish species (Silva et al., 2011;Kormas et al., 2014;Ghanbari et al., 2015;Piazzon et al., 2019). These taxa are largely considered important players in nutritional provisioning, immune defense, and metabolic homeostasis (Estruch et al., 2015;Givens et al., 2015;Rimoldi et al., 2019;Terova et al., 2019).
Accordingly, in the present experiment, gilthead sea bream were fed with three different feeds and at the end of the experiment, the microbiota of these feeds was analyzed. Data revealed that Firmicutes and Proteobacteria were the bacterial phyla represented most, followed in descending order by Bacteriodetes and Fusobacteria. Then, by analyzing specifically the relative abundance of the probiotic L. lactis subsp. lactis compared to the most representative genera of Firmicutes phylum, we found that the percentage of L. lactis subsp. lactis was close to 0% in diet A (control), whereas in diets B and C, it was definitely high, reaching values of 64 and 71%, respectively. This results is in agreement with the supplementation of a low and a high dose of probiotic to diets B and C, respectively.
With regard to the gut microbiota, gilthead sea bream fed diet C showed a significant increase in bacteria belonging to the Spirochetes phylum, which were practically absent in the gut of fish fed diets B and A (<3%). In the same fish group, a decrease in Actinobacteria, Bacteroidetes, and Firmicutes phyla was recorded. The Firmicutes phylum is composed of more than 200 different genera, such as Lactobacillus, Bacillus, Enterococcus, Ruminococcus, and Clostridium. Lactic acid bacteria (LAB) include, among others, Streptococcus sp., Lactobacillus sp., Leuconostoc sp. and Carnobacterium sp., which are considered as beneficial microorganisms that contribute to an healthy status of the fish intestine (Kim et al., 2012;Terova et al., 2019). It is known that commensal Firmicutes and Bacteroidetes are the major producers of short chain fatty acids, such as butyrate, acetate, and propionate that are the end products of fiber fermentations.
While is difficult to assess from genomic data alone the physiological effect on the host of the microbiota changes we found, it is worth noting that Firmicutes/Bacteroidetes ratio in the gut has been directly related to lean body mass in both human and animals (Magne et al., 2020). Indeed, the ratio of Firmicutes vs. Bacteroidetes was increased in obese individuals as compared to lean ones. Actually, gilthead sea bream fed with diets containing probiotic showed a higher Firmicutes/Bacteroidetes ratio than control fish and this could be correlated to their better growth performances. Likewise in mice, the reduced amount of Bacteroidetes was a direct consequence of probiotic supplementation (Grazul et al., 2016). In addition, gilthead sea bream fed diet C, showing the best FCR and SGR values, had the highest percentage of Spirochetes. In swine, the Spirochaetaceae bacterial family was shown to correlate positively with the host weight (Unno et al., 2015). The gut microbiome of the feeding group C was also characterized by a Proteobacteria/Firmicutes ratio five times higher than in the other groups. This result is not surprising because Lc. lactis subsp. lactis SL242 produces the antibiotic nisin, displaying strong activity against Grampositive bacteria (Li et al., 2018), and a vast majority of Firmicutes are Gram-positive.
The analysis of gut-adherent (autochthonous) microbiota did not reveal significant differences between fish groups in relation to L. lactis, suggesting a lack of colonization of the probiotic in the host's intestinal mucosa. This was not a surprising result since it is known that probiotics generally do not colonize the digestive tract i.e., they do not become established permanently or for a long-term (weeks, months, or years) in the intestinal tract (Marco, 2019). Thus, the ingested bacteria can be beneficial while they are in the gut, but they do not have a lasting effect and continued probiotic consumption is needed for sustained impact. Thus, instead of colonizing, the new bacteria may temporarily complement resident microbial communities, forming part of a transient (allochthonous) microbiome in fish without displacing the native gut microbiota, but instead altering digestive tract function by producing active metabolites that modulate the activity of the gut microbiota, or by stimulating the intestinal epithelium directly (Marco, 2019). Hence, in the present trial, although the probiotic did not colonize the host's intestinal mucosa, it did modulate the fish gut microbiota, confirming that colonization is not always necessary to induce host modification. Indeed, diets B and C were enriched with Actinomycetales, as compared to diet A, which instead showed a higher percentage of Pseudomonas, Sphyngomonas, and Lactobacillus genera. These results were confirmed by the clear separation of bacterial community of fish fed with the probiotic from the bacterial community of control fish group (diet A) in the beta-diversity and PLS-DA analyses. Furthermore, the KEGG pathway analysis underlined such differences, highlighting several pathways potentially affected by the diet. Particularly interesting were those related to protein absorption and digestion.
In the present study, the analysis of gut microbial communities revealed significant differences between fish groups in term of species richness and diversity. Among alpha diversity indices, fish fed with diet B showed the highest level of richness estimator ACE and biodiversity, in comparison to the other two fish groups. In contrast, dietary group C, although achieving the best growth performances, showed the lowest gut bacterial diversity.
A reduction in bacterial diversity is usually considered an adverse outcome, since this could lead to less competition for opportunistic or invading pathogens due to a functionally unbalanced ecosystem (Cerezuela et al., 2013;Li et al., 2014;Rimoldi et al., 2020). However, while an increase in intestinal microbial biodiversity following prebiotics (dietary compounds that induce the growth or activity of gut microbiota) administration has been frequently described, the data currently available on the effects of probiotics in fish are more controversial. For instance, in line with our results, the species richness and diversity indexes decreased in gilthead sea bream in response to dietary administration of the probiotic Bacillus subtilis, either alone or in combination with prebiotics or microalgae (Cerezuela et al., 2012(Cerezuela et al., , 2013. In contrast, in line with what we found in fish fed diet B, lactic acid bacteria supplementation was associated with an increase in bacterial diversity in the intestinal mucus of Atlantic salmon (Gupta et al., 2019). In addition, probiotics, such as lactic acid bacteria, are known to produce several antimicrobial compounds capable of suppressing the growth of other microorganisms, which can alter the gut microbiota in terms of both composition and biodiversity (Collado et al., 2007).

CONCLUSION
According to analysis of gut-adherent (autochthonous) microbiota, the probiotic L. lactis subsp. lactis did not colonize in the host's intestinal mucosa. However, the probiotic did modulate the fish gut microbiota, confirming that colonization is not always necessary to induce host modification. Indeed, gut microbiota of fish fed diets B (low dose of probiotic) and C (high dose) were clearly separated from the bacterial community of control fish in the beta-diversity and PLS-DA analyses. Furthermore, the KEGG pathway analysis underlined such differences, highlighting several pathways potentially affected by the diet. Particularly interesting were those related to protein absorption and digestion.
With regard to fish growth performance, there were no significant differences between groups for the FCR and SGR. The only difference was the final body weight of fish fed diet C (high dose of probiotics) that resulted higher than the control group.
Dietary probiotic administration did not alter the morphology of the intestine and did not trigger inflammation.
Researches such as these highlight the interaction between fish diet and their microbiota and suggest that manipulating diet to tune the gut microbiome may be a promising intervention, together with well-designed probiotics.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCBI (accession: SAMN16828235-SAMN16828261 and PRJNA679278).

ETHICS STATEMENT
The animal study was reviewed and approved by the Ethics and Animal Welfare Committees of Institute of Aquaculture Torre de la Sal (IATS-CSIC, Castellón, Spain) (Permit number 824/2019) and "Generalitat Valenciana" (Permit number 2019/VSC/PEA/0197).

AUTHOR CONTRIBUTIONS
FM, FN-C, and MP: experimental investigation, methodology, data curation, formal analysis, and writing-review and editing. SR: methodology, formal analysis, and writing-review and editing. JC-G: methodology, data curation, formal analysis, and writing-review and editing. AG: conceptualization and writing-review and editing. IM: writing-review and editing. FB: experimental investigation and methodology. JP-S: conceptualization, experimental investigation, data curation, and writing-review and editing. GT: conceptualization, data curation, and writing-review and editing. All authors read and approved the final manuscript.

FUNDING
This work was supported by the EU Horizon 2020 Research Innovation Program under the TNA program (project AE130011) at IATS-CSIC research infrastructure within AQUAEXCEL2020 project (652831). This work was also co-funded by the EU Horizon 2020 AquaIMPACT project (Genomic and nutritional innovations for genetically superior farmed fish to improve efficiency in European aquaculture; 818367). This output reflects only the author's view and the European Union cannot be held responsible for any use that may be made of the information contained herein. MP was funded by a Ramón y Cajal Postdoctoral Research Fellowship [RYC2018-024049-I/AEI/10.13039/501100011033 cofunded by the European Social Fund (ESF) & ACOND/2020 Generalitat Valenciana].

ACKNOWLEDGMENTS
Special thanks to Micaela Antonini at the Aquaculture Unit of the University of Insubria for her assistance with the histological platform. FM is a Ph.D. student of the "Dottorato in Scienze della Vita e Biotecnologie" at the "Università degli Studi dell'Insubria, " Varese, Italy.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmars. 2021.659519/full#supplementary-material Supplementary Table 1 | Primers for qPCR gene expression. Colors in letters of column Gene show different gene functions included in the array as follows: red: Epithelial integrity; dark blue: Nutrient transport; green: Mucus production; Orange: Cytokines; purple: Immunoglobulins; black: Cell markers and chemokines; light blue: Pattern recognition receptors.
Supplementary Table 2 | Relative gene expression of anterior intestine in fish fed control (A) and experimental (B,C) diets. Data are the mean ± SEM of 9-12 fish.
All data values were referred to the expression level of hes1-b of control fish with an arbitrary assigned value of 1. 1 P-value results from ANOVA. Bold font indicates statistically significant differences between groups (P < 0.05).

Supplementary Table 3 | Sequencing and taxonomy assignment results.
Supplementary Table 4 | OTUs driving the separation between fish fed probiotic diets (B + C) and fish fed control diet A in the PLS-DA model (VIP > 1).