Microbiota composition and intestinal integrity remain unaltered after the inclusion of hydrolysed Nannochloropsis gaditana in Sparus aurata diet

The use of lysed microalgae in the diet of carnivorous fish can increase the bioavailability of proteins and bioactive compounds, such as unsaturated fatty acids or vitamins in the digestive tract. These are essential molecules for the proper physiological development of fish in aquaculture. However, some antinutritional components and other undesirable molecules can be released from an excess of microalgae supplied, compromising the integrity of the intestine. The inclusion of small amounts of hydrolized microalgae in the fish diet can be a good strategy to avoid negative effects, improving the availability of beneficial compounds. Nannochloropsis gaditana is an interesting microalgae as it contains nutraceuticals. Previous studies reported beneficial effects after its inclusion in the diet of Sparus aurata, a widely cultured species in Europe and in all Mediterranean countries. However, administration of raw microalgae can produce intestinal inflammation, increased intestinal permeability, bacterial translocation and disturbance of digestion and absorption processes. The aim of this study was to evaluate changes in the intestinal microbiota and barrier stability of S. aurata fed with low inclusion (5%) hydrolysed N. gaditana. Intestinal microbiota was analyzed using Illumina MiSeq technology and libraries were constructed using variable regions V3–V4 of 16S rDNA molecules. Analysis were based in the identification, quantification and comparison of sequences. The predictive intestinal microbial functionality was analyzed with PICRUSt software. The results determined that the intestinal microbiota bacterial composition and the predictive intestinal microbiota functionality did not change statistically after the inclusion of N. gaditana on the diet. The study of gene expression showed that genes involved in intestinal permeability and integrity were not altered in fish treated with the experimental diet. The potential functionality and bacterial taxonomic composition of the intestinal microbiota, and the expression of integrity and permeability genes in the intestine of the carnivorous fish S. aurata were not affected by the inclusion of hydrolysed 5% N. gaditana microalgae.

www.nature.com/scientificreports/ level, its average values increase in FH-A compared to Control-A, but not significantly. Clostridiaceae, Gemellaceae, Carnobacteriaceae and Tissierellaceae were not present in the treatment FH-A, but they are present in Control A. Fusobacteriaceae was no present in Control-A but it appears in FH-A (Fig. 3A). In the posterior section, Moraxellaceae, Vibrionaceae, Actinomycetales, and Bacilli_unclassified increased, but Enterobacteriaceae and Pseudomonadaceae decreased in FH-P. Carnobacteriaceae family and Clostridiales order disappeared in FH-P treatment. Statistically significant differences were not detected in any case and data have been represented to show the variation between treatments (Fig. 3B). Some genera were detected between diets As an example, in the anterior section some genera like Clostridiales or Anaerococcus are observable in Control-A but not in FH-A (Fig. 4A). In the posterior section, Enhydrobacter,  Table 1. Ingredient composition of experimental diets used in the feeding trial. FH diet contains 5% hydrolysed raw algal biomass. a Norsildemel (Bergen, Norway). b Lifebioencapsulation (Almería, Spain). c (81% crude protein, 8.8% crude lipid) Sopropeche (France). d (65% crude protein, 8% crude lipid) DSM (France). e Sigma-Aldrich (Madrid, Spain).

Fishmeal LT a 200 200
N. gaditana meal -50 Squid meal b 20 20 Fish protein hydrolysate c 10 10 Krill meal b 20 20 Gluten Choline chloride e 5 5 Vitamin and mineral premix 20 20 Guar gum b 10 10 Alginate b 10 10 Crude protein (%) 47 www.nature.com/scientificreports/ Actinomycetales or Bacilli_unclassifiedincreased their average values in the FH-P treatment, but Vibrio and Sphingomonas decreased in this group. (Fig. 4B). Principal coordinates analysis (PCoA) scores are plotted based on the relative abundance of OTUs of intestinal microbiota from analyzed specimens. Each point represents a single sample, and the distance between points represents how compositionally different the samples are from one another. The points did not show a clear differences in the microbial community composition between control and FH diet (Fig. 5).
Predictive study about the intestinal microbiota functionality. The functional study executed by PICRUSt software presented seven principal levels (metabolism, genetic information processing, environmental information processing, cellular processes, organismal systems, human diseases and drug development), but only three of them related to intestinal functions have been independently studied (metabolism, cellular processes and environmental information processing). Results obtained show that these functions were not affected by FH diet.
The three functions analyzed (metabolism, cellular processes and environmental information processing) did not change between treatments in the KEGG 1 category, although the processes related to environmental information processing and cellular processes were increased lightly in FH-A (Fig. 6A) and FH-P (Fig. 6B) treatments, but without statistical differences.
In KEGG 2 category, functions had similar values between treatments, except in the carbohydrates and amino acid metabolism that increased in FH-A (Fig. 6A) and FH-P (Fig. 6B), but statistical differences were not found. Statistical differences were not found between intestinal sections in any studied KEGG 3 category.
Gene expression. In general, the FH group data showed less dispersion in the gene expression data than the control individuals. In addition, a trend was observed on down-regulation of the genes involved in the intestinal permeability, especially in the posterior section, and in the pro-inflammatory response in both intestine sections (p > 0.05) (Fig. 7A-M). Non-significant slightly up-regulation was also observed in the genes pept1 and muc2 related to nutrition and intestine protection, respectively, in the anterior section in fish receiving the FH diet (p > 0.05) (Fig. 7D,N). However, no significant differences were observed in the relative expression of the genes analyzed in the intestinal samples of fish fed with control or FH diets (Fig. 7).

Discussion
The present study highlights that the carnivorous fish S. aurata fed with 5% of hydrolyzed microalga N. gaditana did not modify its intestinal microbiota after 12 weeks. This could result in an advance in the formulation of fish diets as microalgae are important sources of essential compounds and beneficial nutriceuticals.
In the present study, the tendency of more homogeneous data (less values of variance) and medium values of diversity index between the anterior and posterior sections have been observed when the fish are fed with the FH diet. Reference 41 concluded that changes in diversity are due to an adaptation of the microbiota to digest and assimilate the ingredients added, such as cell walls, polysaccharides and lipids, which affect or modify colonization by minority groups. Avoiding loss of diversity is considered as a positive aspect that protects intestinal mucose and gives protection against diseases 42 . Likewise, authors have established that microbial diversity is considered a biomarker for fish health and an image of a good metabolic capacity 43 and recently 44 , affirm that S. aurata have a plastic microbiota which effectively adapts to the metabolic challenges induced by dietary changes. The same authors have proposed studies focused on how these microbial changes correlate with health, growth, and disease resilience.
Previous studies carried out by Ref. 26 about the effect of N. gaditana on seabream's intestinal microbiota have not shown alterations in intestinal morphology and function with the same percentage of inclusion as in the present study, but the duration of the trial was limited to 1 month and N. gaditana's entire cells were used. The phylum detected in this work, Actinobacteria, Firmicutes and Proteobacteria are part of the normal microbiota in S. aurata [45][46][47] . Some groups such as Micrococcaceae and Ralstonia present antimicrobial activity, biosynthesis of bioactive compounds and can produce beneficial secondary metabolites for the host 48 , and they have been detected in the intestinal microbiota of the FH group. Micrococcaceae family contain enzyme producing bacteria capable of producing amylases, cellulases, proteases, lipases, phytases, tannases, xylanase and chitinase 49 . These enzymes can contribute to the digestion and assimilation of algal products. According to Ref. 50 members of Actinobacteria group are butyrate-producing bacteria. Butyrate is a short-chain fatty acid (SCFA) with important and demonstrated beneficial effects, also in fish such as S. aurata 51 . In this study Actinobacteria have been detected but the fine representation of all the members of this group may be minor to 1% and they are incluyed in ETC group (0.1% Streptococcus, or 0.6% Actinomyces, among others).
In the posterior section, FH group showed a low increase in the presence of Actinobacteria and a decrease in Cyanobacteria, but not statistically significant. Cyanobacteria in general produces lipopolysaccharides that are inflammatory agents and gastrointestinal irritants 52 and causes damage in the host by facilitating the colonization of other dangerous bacteria 53,54 .
Clostridiaceae is a family that appeared in FH-P group. Those group is related to dysbiosis in human 55 but in other animals such as pigs, Clostridiaceae has been related to improvements in feed efficiency and growth performance, being alsoconsidered beneficial to fish 44 . FH-A group showed Bacilli and Actinomycetales. Those groups are present when the diet is enriched in vegetals 56 .
Some genera weredetected in the present study were significantly different between samples. In the anterior and posterior sections it was observable genera like Clostridiales in Control-A but was not present in FH-A. In the posterior section, Enhydrobacter appearedin the treatment FH-P. Enhydrobacter is present in the microbiota of fish such as Dicentrarchus labrax and Salmo trutta 57 . It is known that this genus utilizes certain amino acids such as l-arginine, l-serine and l-alanine 58 , while some species present cellulase activity 57,59 . This fact may be of interest as cellulases hydrolyze the cellular wall of N. gaditana, allowing the fish to absorb the intracellular compounds of this algae. Thus, the intestinal microbiota of S. aurata is able to adapt to dietary changes, as proposed before 44 . www.nature.com/scientificreports/ By a taxonomical description of the bacterial groups present in the intestinal microbiota the information about their function at the gut is unknown 60 . The disponibility of a predictive software allows us to relate the sequences obtained with the OTUs and the predominant functions going on at the intestine. In this study, no differences were found in any of the seven principal functional pathways. The three functional pathways analyzed, metabolism, cellular processes and environmental information processing represented 63.12 ± 0.83% of total functional enrichment per sample. Additionally, KEGG 2 and 3 levels were analyzed but none of them were statistically different between treatments with the exception of a slight increasing trend detected in the www.nature.com/scientificreports/ were represented in the same quantity. The hydrolyzed cells do not affect negatively the bacterial functions, studied by predictive form.
In other studies with S. aurata juveniles, it was observed that low dietary supplementation with Arthrospira hydrolysates (2 and 4%) had no effects on the intestinal mucosa 38 . Similar results were found by Saenz et al. (personal communication/unpublished results) in S. aurata within the present study. In this sense, the results reported by a visual analysis of images obtained by transmission electron microscopy (TEM) and scanning electron microscopy (SEM) have not revealed signs of damage or irritation in the intestinal microvilli. In this regard, N. gaditana in the diet of S. aurata does not alter neither the intestinal microbiota composition nor its functions, significantly. Our in vivo results support previous information described in vitro by Ref. 32 , where N. gaditana added to S. aurata intestinal extract lead to an improvement in the digestion and assimilation of nutrients 61 .
The inclusion of 5% N. gaditana in the diet has not produced any significant alteration in the expression of genes involved in the enterocytes cell-to-cell adhesion, the cytoskeleton and the epidermal cell-basement membrane adhesion (cdh1, cdh17, cldn12, cldn15, vim, itgb6, ocln, tub, and zo1) and, therefore, it is very likely that the integrity and permeability of the analyzed fish remain unaltered. This information is further supported by the absence of changes in the expression of markers commonly used to measure pro-inflammatory processes (tnf-α and cox2), which have been associated with increased permeability and intestinal disruption in several studies 35,45,62,63 . The absence of statistically significant differences in our study could also be due to a high dispersion between individuals. However, a tendency on a down-regulation of the tight-junction genes analysed was observed in the FH group. Experimental diets that have shown to impair fish growth have also been related to a reduction in the expression of genes involved in maintaining tight junctions 45 whereas their upregulation has been associated with a healthier intestinal barrier 7 . Contradictorily, a gene up-regulation of tight junctions in response to various experimental diets has also been related to nutritional, osmoregulatory and physiological alterations, thus indicating an undesirable state in gilthead seabream 51,64 . Therefore, further analysis of immune system-related genes, as well as histological studies are necessary to confirm the actual status of the gut integrity and permeability.
The peptide transporter 1 (PepT1) is a low-affinity/high-capacity nutrient transporter that mediates the uptake of dipeptides and tripeptides from diet, which are essential for fish to sustain development, growth and metabolism 65 . In our study, no statistically significant difference was observed in pept1 gene expression between the groups analyzed. However, both Control and FH groups showed higher expression levels of pept1 in the anterior intestine section compared to the posterior section. Additionally, a slight up-regulation of pept1 was observed in the FH group compared to the control group in the anterior intestine section. Higher pept1 expression in the anterior section is consistent with studies indicating that the proximal intestine is the area of highest expression of this protein in fish, specifically at the absorptive epithelial cells along the mucosal folds 66,67 . The trend indicating a slight up-regulation of pept1 in the FH group may suggest improved nutrient absorption. In higher vertebrates, more weight gain and increased expression of pept1 in the intestine were found in response to a high-quality soybean protein diet 68 . Reference 62 found that a diet containing 15% of vegetable proteins (green pea protein concentrate) induced significantly lower levels of pept1 transcripts and were associated with lower growth in gilthead seabream 67 . Gilthead seabream fed on a strict vegetable protein diet also showed lower expression levels of pept1, and has been related to a lower digestibility and small peptide transport 69 . Taking pept1 gene expression as a marker of the dietary protein quality and absorption efficiency, our results suggest that the substitution of 5% of N. gaditana in the diet is probably not affecting the nutrient intake in gilthead seabream, or that the high dispersion between individuals may be affecting the significance of the results and, therefore, hiding possible beneficial effects.
Mucins contribute to protect the intestine epithelium against a broad spectrum of damages 70 . Enhanced gut mucin production has been related to reduced bacterial translocation in fish fed with a diet containing yeastderived oligosaccharides 63 . In this study, the inclusion of N. gaditana did not affect the mucins related genes (muc2 and imuc) between the Control and FH groups.
In summary, the inclusion of 5% N. gaditana microalgae subjected to enzymatic hydrolysis in the commercial diet does not alter the composition and functionality of the intestinal microbiota, nor the expression of integrity and permeability genes in the intestine of the carnivorous fish S. aurata.

Feed preparation. The N. gaditana biomass was produced in closed tubular photobioreactors in La
Estación Experimental de las Palmerillas (Fundación Cajamar, Almería, Spain) as reported by Ref. 71 . Raw microalgae paste at provided were collected and immediately used for enzymatic hydrolysis according to a previous described protocol 38 . Briefly, N. gaditana sludge containing up to 150 g L −1 of raw microalgae biomass was hydrolyzed using commercial enzymes with cellulase activity (Viscozyme ® ) under controlled conditions (pH 5.0 and 50 °C under continuous stirring) for 4 h, providing 2% (w/w) enzyme. The experimental aquafeed was elaborated at the CEIA3-Universidad de Almería facilities (Servicio de Piensos Experimentales, http:// www. ual. es/ stecn icos_ spe) including 5% of the hydrolyzed algae (FH diet). The microalgae concentrations employed were based on previous studies 39 . The solids were mixed in a kneader and, subsequently, the resulting mixture was extruded to obtain granules with the desired diameter (2-3 mm). Feed was stored at −20 °C until use. Control feed composition is showed in Table 1.
Experiment design and sampling. The  www.nature.com/scientificreports/ (mean ± SEM) were randomly distributed in 6 tanks of 80 L capacity (N = 15 fish per tank). The experiment was divided into two different fed conditions, one receiving a commercial diet (control group) and the other receiving a diet supplemented with 5% of fresh hydrolyzed N. gaditana paste (FH group) for 86 days. At the end of the growth assay, the animals (fasted for 12 h) were netted and deeply anesthetized with 2-phenoxiethanol (1 mL/L, Sigma-Aldrich 77699), and intestinal samples from seven fish per treatment (three fish from one tank, and two fish from the other two tanks) were extracted. Then, each tract was divided into two major sections, anterior and posterior sections, and kept at −80 °C until use. Remaining fish from each tank were employed in a parallel study (unpublished results).

DNA extraction and sequencing by Illumina Miseq technology. DNA of both intestinal sections
were extracted according to the protocol described by Ref. 72  A workflow based on the MOTHUR program (version 1.39.5) was used to remove Illumina adapter sequences and demultiplexing. The reads were filtered excluding reads < 80 bp or > 2000 bp long. Besides, the singleton sequences and the chimeras were discarded by UCHIME version 4.2 (https:// drive5. com/ uchime). Non-specific PCR amplicons were eliminated. The remaining representative, non-chimeric sequences were then subjected to taxonomic assignment against the Greengenes 16S database (May 2013), with 97% 16S similarity as the cutoff and clustered into Operational Taxonomic Units (OTUs).
After generating the taxonomic profile of microbiome samples, a comparison of taxa present in the samples was carried out. The data size was normalized to the minimum number of reads obtained in all the samples. All statistical analyses were performed using statistical software R and a web tool MicrobiomeAnalyst 74 . To determine the level of sequencing depth, rarefaction curves were obtained by plotting the number of observed OTUs against the number of sequences and Good's coverage coefficient calculated. Alpha diversity was estimated based on Shannon, Chao1, and Simpson indexes to assay taxonomic and phylogenetic structure diversity, respectively. The results are presented at phylum, family and genus taxonomic levels. The group with relative abundance less than 1% have been considered "ETC" according to Ref. 75 where'as taxa constituting ≥ 1% of the total number of cells' and 'rare phylotypes': as taxa constituting ≤ 0.1% of the total number of cells' . This limit is defined based on the levels that can be detected with PCR-dependent techniques 76 . We have used 1% because this is the limit of detection in molecular techniques such as DGGE, which we have used in the past to study intestinal microbiota 73,[77][78][79] . Non-parametric Kruskal-Wallis statistical test was performed. The differences were considered statistically significant assuming p < 0.05. Finally, a multivariate analysis of OTU data was performed via Principal Coordinate Analysis (PCoA) of OTU profiles using Bray Curtis metric to represent differences between microbiota of each group. PCoA are plotted against each other to summarize the microbial community compositional differences between samples.
Functional profiling of microbial communities. Predictive microbiota functions were made using PICRUSt (version 1.1.3), a tool designed to infer metagenomics information from 16S rRNA amplicon sequencing data 80 . The resulting metagenomics data were entered into the Greengenes database (version 13.5), and the metagenome prediction of bacterial communities was conducted using the calculated dataset after normalizing for DNAr16S copy number. Nearest Sequenced Taxon Index (NSTI) scores for evaluating the accuracy of predicted metagenomes were categorized with the Kyoto encyclopedia of genes and genomes (KEGG) pathways database 81 . Bacterial functional profiles until KEGG modules level 3 were compared, and the STAMP (Statistical Analysis of Metagenomics Profiles) was used to analyze the differential abundance of modules by intestinal sections and diet using ANOVA multiple-comparison test with post hoc Tukey-Kramer (p < 0.05).

RNA isolation and quantitative PCR (qPCR).
In order to assess the gene expression throughout the intestinal tract, RNA was isolated from each intestine section, following TRIsure™ (Bioline, England) manufacturer instructions. RNA quantity was determined in Qubit (Thermo Scientific) and reverse transcription was performed using First Strand cDNA Synthesis Kit (Thermo Scientific) with 500 ng of total RNA. One microliter of each cDNA was employed as the template in the qPCRs to analyze each gene transcription. Specific primers were used for the quantification of the relative gene expression genes involved in intestinal permeability and www.nature.com/scientificreports/ integrity, such as cadherin 1 (cdh1), cadherin 17 (cdh17), claudin 12 (cldn12), claudin 15 (cldn15), vimentin (vim), integrin 6-β (itgb6), ocludin (ocln), tubulin (tub) and zona-occludens 1 (zo1); pro-inflamamtory reactions and mucins production, such as tumoral necrosis factor α (tnf-α), cycloogygenase 2 (cox2), intestine mucin (imuc) and mucin 2 (muc2); and nutrient absorption, such as the peptide transporter 1 (pept1) ( Table 2, Supplementary Information). For normalization, the samples were analyzed in parallel with two reference genes, elongation factor 1α (ef1α) and ribosomal glyceraldehyde 3-phosphate dehydrogenase (gadph). The qPCR reactions were carried out in a C1000 Touch™ thermal cycler (BioRad, Spain) with the CFX96™ optical module (BioRad, Spain) for fluorescence measurements. Amplification reactions were performed in triplicate in 96-well plates in a final volume of 10 μL. The mixture contained 5 μL of GoTaq ® qPCR Master Mix (PROMEGA), 0.5 μL of forward and reverse primers (10 μM), 1 μL of cDNA and 3 μL of nuclease-free water. For the permeability and integrity genes, an initial activation of Taq polymerase at 95 °C for 3 min was used, followed by 40 cycles of 95 °C for 15 s and 60 °C for 60 s. For the immune and mucins related genes (tnf-α, cox2, imuc, muc2 and pept1), the PCR was performed using 95 °C for 10 min, and 40 cycles of 95 °C for 10 s and 60 °C for 20 s. Finally, in the case of the reference genes (ef1α and gadph) the reactions were incubated 90 °C for 5 min followed by 40 cycles of 95 °C for 10 s, 60 °C for 10 s and 72 °C for 15 s. Threshold amplification values (Cq) greater than 40 were considered negative. Relative expression of mRNA was calculated using the method 2 (−ΔΔCq)82 , normalizing with a geometric average of the two reference genes and in relation to the fish of each control group. The primer sequences used to analyze gene expression in S. aurata were obtained from previous studies 40,64,69,83 .
Statistical analysis was performed using XLSTAT v2014.5.03 software (Addinsoft, New York, NY, USA). Results are shown as means of fold change (2 (−ΔΔCq) ) ± standard deviation (SD). The normality and homogeneity of the data were previously evaluated using the Shapiro-Wilk and Levene tests, respectively. The existence of statistically significant differences in the RT-qPCR values between the control and FH groups was determined by One-Way Analysis of Variance (ANOVA), applying the Tukey post-hoc test. When normality of the data could not be assumed, non-normal data were logarithmic transformed, and the non-parametric Kruskal-Wallis test was performed. The differences were considered statistically significant assuming p < 0.05. Ethics approval. Fish were kept and handled following the guidelines for experimental procedures in animal research from the Ethics and Animal Welfare Committee of the University of Cadiz, according to the Spanish (RD53/2013) and European Union (2012/707/EU) legislation. The Ethical Committee from the Autonomous Andalusian Government approved the experiments (Junta de Andalucía reference number 01/04/2019/047). Declaration ARRIVAL 2.0 protocol. This study is reported in accordance with ARRIVE guidelines (https:// arriv eguid elines. org).

Consent for publication.
All the authors read and agree to the content of this paper and its publication.

Data availability
Raw read sequences of the 16S rRNA gene from S. aurata gut microbiota in this study are publicly available in the NCBI SRA depository within BioProject PRJNA700500, with BioSample accession numbers SAMN17721432-SAMN17721459.