Three experimental diets were designed to evaluate the effects of D. hansenii (CBS 8339) on fish performance and feed utilization and evaluate the effects of its supplementation on the skin through transcriptomic analysis. A 7% fish meal-based basal diet was formulated and considered a control diet ( Table 1 ). Subsequently, the control diet was used as a basal diet to create both experimental yeast diets, which only differed on D. hansenii content proportion (Yeast_1.1% diet with 1.1%, 17.2 x 10⁵ cfu; and Yeast_2.2% diet 2.2%, 35.7 x 10⁵ cfu) as shown in Table 1 . These percentages have been determined in previous studies (17–19, 21). Diets were isoproteic (48.4%), isolipidic (17.2%), and isoenergetic (21.7 MJ/Kg feed). Each of the diets (pellet size: 2mm) was manufactured by SPAROS Lda (Portugal) under the conditions detailed by Gisbert et al. (33) and stored at 4°C throughout the experiment to prevent feed deterioration. Diet manufacture conditions, packaging, and storage did not compromise the viability of the probiotic (18, 20, 21). D. hansenii was provided by CIBNOR (La Paz, Mexico) and cultured as described by Tovar-Ramírez et al. (18).

A total of 500 S. aurata juveniles (BW = 14.5 ± 1.2 g) were obtained from a commercial farm (PISCIMAR SL; AVRAMAR, Burriana, Spain) and transported and acclimated to IRTA-La Ràpita research facilities (La Ràpita, Spain). During acclimatization, fish were fed a commercial diet (Optibream, Skretting) and maintained in a 2,000 L tank connected to IRTAmar^® recirculating system under natural conditions and constant temperature (23.0 ± 1.0°C).

Prior to the experiment, 300 fish were slightly anesthetized (50 mg/L tricaine methanesulfonate, MS-222, Sigma-Aldrich, Madrid, Spain) and randomly distributed in four replicate tanks per diet (N = 25 fish per tank). The experiment lasted for 70 days, and diets were distributed in four meals at 08:00, 11:00, 13:00, and 16:00 h (the corresponding feed ratio was distributed during 1 h), and uneaten pellets were collected and weighed to calculate daily feed ingesta values. The trial was run under natural photoperiod (14h light/8h dark), salinity (35-36 ‰), constant temperature (23.3 ± 1.3°C), dissolved oxygen (5.7 ± 0.2 mg/L) (OXI330, Crison Instruments, Spain), and pH (8.2 ± 0.1) (pHmeter 507, Crison Instruments). Ammonia and nitrite levels (0.13 ± 0.1 mg /L and 0.18 ± 0.1 mg /L, respectively) were continuously controlled using HACH DR 900 Colorimeter (Hach Company, Spain).

Growth performance and feed utilization indicators were calculated using the following formulae:

- Weight gain rate (WGR) % = 100 x [(BWf – BWi)/BWi]; where BWf and BWi are the final and initial mean BW of fish.

- Specific growth rate (SGR) % BW/day = 100 x [(ln BWf – ln BWi)/days]

- Survival rate (SR) % = 100 x (Final number of fish/Initial number of fish).

- Feed conversion ratio (FCR) = Feed intake (g)/Increase in fish biomass (g).

At the end of the experiment, all fish were anesthetized (100 mg/L tricaine methanesulfonate, MS222) and measured individually for the calculation of key performance indicators as previously described and to sample skin mucus following the method described by Fernandez-Alacid et al. (34). According to the results obtained from the growth and skin mucus metabolites, only control and Yeast_1.1% diets were considered for the skin transcriptomics and skin mucus defensive capacity analyses.

In addition, three randomly selected fish per replicate tank (N = 12) were euthanized (300 mg/L MS222) for transcriptional analysis purposes. A ca. 1 cm² of skin section from the mid-region of the fish body (over the lateral line, right side of each fish) was dissected and the muscle tissue attached to it was discarded. Samples were incubated overnight (4°C) in RNAlater™ and stored at -80°C for further RNA extraction.

Equality of variances and normality were checked using Levene’s test. Differences among the key performance parameters and SMABs were performed using a one-way ANOVA test, followed by post-hoc Bonferroni’s test (if equal variances were assumed) or Dunnett’s test (if variances among groups were unbalanced) to test the effects of yeast supplementation levels (p < 0.05). In addition, curvilinear estimation of regression analysis was performed to a deep comprehension of KPI and SMAB differences (p < 0.05). Student’s t-test, assuming data homoscedasticity, was used to compare bacterial growth differences among the Yeast_1.1% diet and control diet (p < 0.05). Data on KPI and SMABs were expressed as mean values ± standard deviations using SPSS Statistics for Windows (v.25; IBM Corp., Armonk, NY, USA), and bacterial growth values were expressed as mean ± standard error, with GraphPad PRISM 9.

Results on fish growth performance and key performance indicators are represented and summarized in Table 2 . Briefly, fish fed the Yeast_1.1% diet performed better than the rest of the dietary groups. Values of WGR, SGR, and FCR were improved in fish fed the Yeast_1.1% diet compared to those from both control and Yeast_2.2% diets. Furthermore, the feed conversion ratio (FCR) value was lower (p < 0.05) in fish fed the Yeast_1.1% diet when compared to fish fed the reference diet, whereas fish fed the Yeast_2.2% diet showed an intermediate value. However, no significant differences in feed intake (FI) were found among the experimental groups ( Table 2 ; p > 0.05).

A detailed transcripteractome was performed from Biological processes, Molecular functions, and Cellular components to unravel the modulatory effects of the yeast-based supplemented diet on the skin mucosal tissue. The enrichment analysis revealed multiple GO annotations corresponding to binding molecular functions, mainly “Protein binding” (GO.0005515), “Enzyme binding” (GO:0019899), and “Protein-containing complex binding” (GO:0044877). In addition, these functions are associated with the presence of regulatory biological processes, which might be correlated with the defensive pluripotency attributed to the fish skin. Specifically, among the main mucosal tissues (i.e., gill, gut, and skin), the skin is considered the principal defensive barrier since it covers almost the entire body of the fish (6, 50). In this line, the analysis of the “Cellular component” GO showed that the DEGs performed their functions, especially in membranes (“Intracellular membrane”, GO:0043231; “Side of membrane”, DO:0098552) in addition to being part of vesicles and exosomes (“Cytoplasmic vesicle”, GO:0031410; “Extracellular exosome”, GO:0070062). In fish, the skin tissue is composed of active cells, with a high turnover ratio to maintain the viability and protective barrier function of this tissue (51). Moreover, the skin epidermis also contains specialized mucus-producing cells which generate and secrete the mucus layer that covers the skin. Therefore, the relation of these biological aspects could represent an improvement in the barrier function of the skin, either by modulating the secretion and/or the composition of the mucus (52) or by enhancing the physical protection capacity at cell-cell junctions or/and cell and extracellular matrix levels (“Anchoring junction”, GO:0070161). These effects are crucial to protecting the organism from environmental fluctuations and biotic stressors (5, 53). The above-mentioned structural and functional characteristics based on the transcriptomic data obtained demonstrate that the supplementation at 1.1% of D. hansenii in a low fish meal-based diet (7%) was capable of modulating and supporting skin functions in S. aurata. These results may be attributed to the contribution of bioactive compounds from D. hansenii since this yeast species is rich in β-glucans, polyamines, and mannan oligosaccharides (19, 20, 54).

Different studies in fish have shown the possibility of modulating skin functions through dietary strategies (27, 30). In our study, we have observed a modulation of some pathways related to amide metabolism (aass, abcc2, acss1, adam10, asah1, eif1, eif3k, glyat, mrpl1, mrps10, mrps11, mthfs, naca, psap, rpl14, rpl18, rpms17, rps12, and shmt1) and amino acid metabolism (shmt1, gldc, glyat, aass, ahcy, serinc2, and arhgef17). Similar findings related to the modulation of amide metabolism were reported in a parallel study in which we evaluated the effect of D. hansenii dietary supplementation at the intestinal level in the same fish species (21); the results may be attributed to the high content of polyamines in this yeast (19). Polyamines are a part of and regulate a large variety of key cellular processes, such as cell growth, differentiation, or proliferation, whose control is crucial to maintain cell homeostasis (55). In the skin, we found two upregulated genes involved in polyamine regulation, oaz1 and azin1. In response to elevated levels of polyamines, ornithine decarboxylase antizyme 1 (OAZ1) and antizyme inhibitor 1 (AZIN1) regulate the entry of polyamine compounds into cells and may inhibit ornithine decarboxylase (ODC). Thereby, these molecules initially regulate polyamine metabolism (55, 56). In addition to this pathway, the slc6a6 gene was upregulated in our study, whose function is the transport of taurine, urea, and biogenic amine (57). This modulation of amide metabolism and its related transport mechanisms are undoubtedly of great interest since D. hansenii dietary supplementation not only has effects at the intestinal level (21) but also at the skin mucosa level as herein observed, which may indicate a similar effect at cell growth, differentiation, or proliferation levels in different MALTs. Moreover, polyamines and their derived compounds participate in protein metabolism, which is highly relevant in carnivorous fish (58). In this sense, the enrichment analysis revealed several biological processes involved in the regulation of protein turnover, suggesting a modulation of the amino acid and protein metabolism in the fish skin by D. hansenii supplementation. Our study also revealed that the slc7a2 gene, an amino acid transporter member of the amino acid-polyamine-organocation (57, 59), was upregulated in the skin cells. The SLC7A2 transporter is responsible for the cellular uptake of arginine, lysine, and ornithine. An increased uptake of arginine and ornithine, which also participates in polyamine metabolism, supports the above-mentioned hypothesis of increased polyamine functions in the skin through dietary uptake. Moreover, the expression and functions of SLC7A2 have been highly related to both innate and adaptive immunity. In particular, SLC7A2 has been reported to control macrophages’ function toward a bacterial infection and under inflammation in mammal peripheral monocytes (60). The up-regulation of these DEGs related to protein metabolism and polyamine turnover might explain a more efficient use of protein sources (61), which would support the improvement of intestinal health status and nutrient digestion (21), altogether with modulation of immune-related processes.

Furthermore, several mediators of the immune response were also observed to be upregulated in the skin of S. aurata fed the Yeast_1.1% diet. For instance, the TNF Receptor Superfamily Member 14 (tnfrsf14) gene expression was highly increased. The TNFRSF14 membrane-bound receptor binds several ligands mediating both coinhibitory and costimulatory pathways. Therefore, TNFRSF14 regulates immune cell homeostasis (62–64). In accordance, among its ligands, we found that it down-regulated the B- and T-lymphocyte attenuator (btla) gene, a member of the CD28 family, and its interaction with TNFRSF14 delivers a coinhibitory signal regulating T-cell activation (65). However, due to their wide cellular distribution, their functional analysis is complex. For instance, the balance herein observed in their expressions suggests a naïve or final stage of the T-cell activation phase, and/or an immature phase of dendritic cells (66). In this line, we observed the upregulation of the cd28 gene, a costimulatory receptor that triggers T-cell activating functions. Thus, the lack of modulation of costimulatory T cell receptors (TCR), like their cognate CD80/CD86 ligands, and/or interleukin modulatory expressions, allowed us to discard a proliferative T-cell response associated with the upregulation of CD28, supporting the idea of a naïve/resting stage of T-cells modulation in the skin of S. aurata after feeding with the Yeast_1.1% diet for 70 days (67, 68). In addition, T-cell activation requires, among the above-mentioned signals, a complex formed by MCH with T-cell receptors (TCR) (69). Under this scenario, the downregulation of the human leukocyte antigen (hla-a) gene, a major histocompatibility complex class I (MHC-I) molecule present in the cell surface of all nucleated cells, along with genes related to “positive regulation of T cell activation” (btla, tfrc, and sash3; GO:0050870), coupled to the modulation of genes related to “regulation of T cell proliferation” (sox12, and btn2a2; GO:0042129), would also support the above-mentioned hypothesis about the T-cell stage found under the current experimental conditions. Regarding B-cell regulation, one of the co-stimulatory signals for B-cell activation is the CD22 molecule as it has been described in turbot (Scophthalmus maximus) and Japanese flounder (Paralichthys olivaceus) (70). In our study, fish fed the Yeast_1.1% diet presented a downregulation of the cd22 gene, which has been suggested as one of the B-cell biomarkers in Atlantic salmon (Salmo salar) as it is consistently expressed in IgM⁺ or IgT⁺ lymphoid cells together with cd79a (71). Under the current experimental conditions, the enrichment analysis of DEGs showed several regulatory processes related to B-cell proliferation and adaptative immune responses coupled with the observed upregulation of the cd38 gene. In this sense, CD38 participates in the regulation of intracellular Ca²⁺, a cell signaling system, which in turn modulates immune functions. In fish, CD38 has been shown to participate in antiviral and antibacterial B-cell responses, and its expression seems to modulate different B-cell stages (72). However, the enrichment analysis herein shows several repressed genes involved in the positive activation and proliferation of B-cells (i.e., mef2c, tfrc, and sash3; GO:0050871 and GO:0030890, respectively). The functionality and modulation of these above-mentioned lymphocyte-related genes reinforced the hypothesis of an immunomodulatory effect of D. hansenii on the skin of fish. With the downregulation of the membrane-attack complex member c7 gene, its function and expression have been observed in response to infection in fish (73), and the modulation of several viral processes and antiviral genes (i.e., trim25, pvr, f11r, among others) suggest that the supplementation of D. hansenii in low fish meal-based diets is capable of modulate and regulate immune functions in the skin of S. aurata. Similar results related to T and B cell development were reported when comparing the effect of D. hansenii on the skin (present study) and the gut in S. aurata fed the Yeast_1.1% (21) even though different DEGs were involved in this MALTs, which may be attributed to different tissue-specific response to dietary immunomodulators, being that the gut is more tolerogenic than the skin (3, 50).

Regarding the function of the skin protecting the fish, one of the most distinctive features is the production and secretion of mucus by goblet cells and club cells (74). The mucus layer is mainly composed of glycoprotein molecules called mucins, soluble proteins, several metabolites, and hormones, among other compounds (28, 34). In our study, we found that D. hansenii dietary supplementation could modulate mucin properties through mucin synthesis and secretion. For example, through the upregulation of the GalNAc sialyltransferase (st6galnac6) gene, which adds terminal sialic acids contributing to the mucin negative charges (75). This is of special relevance since sialic acid can act as ligands for receptors on viruses, acting as competitive inhibitors, blocking receptor binding, and limiting infection of epithelial cells (76, 77). Moreover, the upregulation of the myo5a gene, a class V myosin motor capable of moving along actin filaments (78), and the esyt3 gene, a synaptotagmin Ca²⁺ sensor required for vesicle fusion, might suggest an increased mucin secretion (75, 79). However, this hypothesis could not be supported by an increase in the expression of MUC18, which has been described as the most abundant mucin in S. aurata that plays an important role in adhesion and protection against pathogens (80). In our study, we found a repressed expression of muc17, a membrane-bounded and no secreted mucin, which could indicate a differential modulatory pattern for skin mucins under the present experimental conditions. However, the available information regarding this specific issue is limited, and further studies are needed to discern the complex pattern of mucins, their changes, and their implications in skin mucus homeostasis.

Regarding mucus composition, the SMABs profile analyzed did not show major changes except for a slight increase in skin mucus glucose levels. Glucose, lactate, and protein levels are markers of metabolic and stress factors in fish mucus (34). Increased glucose levels could indicate a stressful condition or a switch in metabolic status. However, the levels of cortisol, the major corticosteroid indicator of stress in fish (81), remained constant through dietary groups. Indeed, when the levels of metabolites were standardized per protein ratios, the Control and Yeast_1.1% diets did not show differences in glucose levels, but higher levels on the Yeast_2.2% were exhibited. Taking this into consideration, these results also supported that the above-mentioned hypothesis of increased mucus production, as transcriptomic data hinted, is due to the inclusion of yeast in diets. As increased mucus production is shown to be the first defensive mechanism in fish (82), this might be considered a beneficial measure against potential biotic and abiotic stressors due to its incorporated defensive components, as the in vitro skin mucus co-culture with pathogenic bacteria also indicated. Among the above-mentioned metabolites, the skin mucus is also composed of a great variety of innate immune components and defensive molecules that, working together, confer the defensive capacity to the mucus (5, 83). Hernández-Contreras et al. (31) found increased activity of several immune-related enzymatic parameters in the skin mucus (e.g., proteases and lysozyme) when fish were fed a diet supplemented with D. hansenii. In our study, some enzymatic-related genes were modulated, specifically two protease genes (prss2, and prss3), and one phosphatase inhibitor-related gene (set). The upregulation of the serine protease 2 (prss2), which participates in the formation of antimicrobial peptides (AMP) (84, 85), could suggest an increased AMPs level, potentiating the antibacterial defense. The modulation of these defense components has been shown in the grater amberjack fish (Seriola dumerili) when fed a diet supplemented with yeast-derived mannan oligosaccharides (54). This suggests that yeast components might modulate skin defense components potentiating its defensive capacity. Accordingly, the observed reduced growth capacity of V. anguillarum in the skin mucus of S. aurata fed yeast-supplemented diet is especially relevant because Vibrio strains are capable to adhere to surface cells and penetrate host epithelial and vascular tissues (86). Besides, P. anguilliseptica growth was also affected, exhibiting in the final 2 hours of co-culture a decreased growth. Pseudomonas is considered an opportunistic pathogen whose infections in S. aurata occur under environmental stress (87). The results observed herein suggest that feeding the fish with D. hansenii not only could increase mucus production but also could increase its enzymatic and antibacterial defense components, thus protecting the host from potential pathogenic bacteria (88).

In the same line, we found the upregulation of antioxidant-related DEGs (txnl1, gpx3, ipcef1, and sesn2). The antioxidant modulatory defense effects of D. hansenii have been demonstrated in different animal species including fish (46, 89, 90). However, this modulating effect is not reflected in the total antioxidant capacity of the skin mucus, whose values in the Yeast_1.1% diet remained close to those from the Control diet. Under certain situations, the skin mucus’s antioxidant capacity is promoted to protect the mucosal surface from oxidative attack, either due to environmental stress (91, 92) or skin disorders (93). Due to the lack of oxidative challenging situations, we hypothesize that the dietary yeast supplementation promoted sentinel antioxidant defensive processes rather than activating response stress by reactive oxygen species (ROS) presence, which is in agreement with other studies focused on functional feeds (30, 32). In addition, skin transcriptomics did not reflect responses linked to oxidative damage but to the downregulation of several genes involved in oxidative phosphorylation, which is the main metabolic ROS producer in cell metabolism (94, 95). This may support the hypotheses that D. hansenii supplementation might result in a better condition of the skin, including a tight regulation of its immune response, enabling a more efficient use of available energy and nutrients for growth (21).