Exploring the potential for an evolutionarily conserved role of the taste 1 receptor gene family in gut sensing mechanisms of fish

In this study, we investigated the transcriptional spatio-temporal dynamics of the taste 1 receptor (T1R) gene family repertoire in seabream (Sparus aurata [sa]), during larval ontogeny and in adult tissues. In early larval development, saT1R expression arises heterochronously, i.e. the extraoral taste-related perception in the gastrointestinal tract (GIT) anticipates first exogenous feeding (at 9 days post hatching [dph]), followed by the buccal/intraoral perception from 14 dph onwards, supporting the hypothesis that the early onset of the molecular machinery underlying saT1R expression in the GIT is not induced by food but rather genetically hardwired. During adulthood, we characterized the expression patterns of saT1R within specific tissues (n = 4) distributed in oropharingeal, GIT and brain regions substantiating their functional versatility as chemosensory signaling players to a variety of biological functions beyond oral taste sensation. Further, we provided for the first time direct evidences in fish for mRNA co-expression of a subset of saT1R genes (mostly saT1R3, i.e. the common subunit of the heterodimeric T1R complexes for the detection of “sweet” and “umami” substances), with the selected gut peptides ghrelin (ghr), cholecystokinin (cck), hormone peptide yy (pyy) and proglucagon (pg). Each peptide defines the enteroendocrine cells (ECCs) identity, and establishes on morphological basis, a direct link for T1R chemosensing in the regulation of fish digestive processes. Finally, we analyzed the spatial gene expression patterns of 2 taste signaling components functionally homologous to the mammalian G(i)α subunit gustducin, namely saG(i)α1 and saG(i)α2, and demonstrated their co-localization with the saT1R3 in EECs, thus validating their direct involvement in taste-like transduction mechanisms of the fish GIT. In conclusion, data provide new insights in the evolutionary conservation of gut sensing in fish suggesting a conserved role for nutrient sensors modulating entero-endocrine secretion.


Introduction
Vertebrates recognize a wide variety of food-related substances by olfactory and taste chemosensory systems to detect chemical cues mediating both appetitive and aversive behaviors to foods. In the classical view, the sense of taste is associated to gustation produced in the oral cavity epithelium, where taste qualities are perceived by specific receptors. In the case of sweet, umami and bitter, taste signaling is initiated by specialized taste G-proteincoupled receptors (GPCR) type 1 (T1R) and 2 (T2R), mainly expressed in lingual taste buds (Lindemann, 2001;Chandrashekar et al., 2006). Particularly, the T1R-mediated chemosensing associated to metabolic and hedonic signals initiates in specialized type II-taste receptor cells (TRCs-II) expressing 3 T1R gene paralogs (T1R1, T1R2 and T1R3) that function as heterodimeric complexes prototypically encoding for sweet (T1R2/T1R3) or umami (T1R1/ T1R3) taste modalities (Li et al., 2002;Nelson et al., 2002;Finger, 2005). T1R signal transduction within TRCs-II is accomplished via the heterotrimeric G-protein complex G-abg that dissociates in the 2 functional components Ga-and Gbg upon receptor/ligand binding. The best described cellular pathway implicated in mammalian taste transduction relies on the Gia subunit gustducin (Ga-gust)-dependent activation of multiple downstream effectors including phospholipase Cb2 (PLCb2), inositol triphosphate receptor 3 (IP 3 R3) and transient receptor ion channel 5 (TRPM5), ultimately leading to the elevation of intracellular calcium, taste cell membrane depolarization and afferent neuronal transmission to the gustatory cortex (reviewed by Ahmad and Dalziel, 2020).
A multitude of studies in the last decades uncovered that the role of T1R and T2R in chemosensing is not limited to canonical gustatory functions driving food choices towards ingestion or rejection, but it rather extends far beyond oral cavity sensing (reviewed by Finger and Kinnamon, 2011). Indeed, T1R and T2R expression and associated signaling pathways have been identified in extra-oral tissues of endodermic origin (i.e. digestive and respiratory apparatuses), within a large polymorphic population of isolated or clustered cells presumably involved in immune and digestive functions, and collectively recognized as the diffuse chemosensory system (DCS) (Braun et al., 2011;Uhlen et al., 2015;Hass et al., 2010;Taniguchi, 2004;Osculati, 2003, 2005). Accordingly, expression of T1R has been documented in mammalian enteroendocrine cells (EECs) of the gastrointestinal tract (GIT), along with functional evidences on their implication in the modulation of gut hormone release. Gut peptides secreted upon T1R activation in the GIT are important endocrine factors responsible for the regulation of many physiological processes including satiation and satiety, digestive (acid, bile and enzyme secretion, and gut motility) and absorptive (nutrient transporter expression and nutrient uptake) functions, epithelial cell proliferation and regeneration as well as metabolism (energy and glucose homeostasis) (Dyer et al., 2005;Depoortere, 2014;Raka et al., 2019;Jang et al., 2007;Alpers, 2010).
From an evolutionary perspective, the 3 T1R orthologs are conserved across vertebrates, including fish, whose T1R families have greatly expanded mostly due to additional T1R2 duplicates that apparently evolved to increase taste plasticity for amino acid sensing (Hashiguchi et al., 2007;Baldwin and Ko, 2020;Oike et al., 2007;Angotzi et al., 2020). Emerging evidences based on quantitative molecular studies indicate that several taste receptors and canonical components of T1R transduction signaling are also present in the fish GIT, suggesting that the T1R-mediated gut sensing mechanisms could have been conserved during evolution (Polakof and Soengas, 2013;Latorre et al., 2013;Ronnestad et al., 2016;Calo et al., 2021). On the other hand, the Ga-gust system is absent in the genome of both amphibians and teleost fishes as a result of 2 independent gene losses in their last common ancestors (Oka and Korsching, 2011;Ohmoto et al., 2011), somehow implying that other G(i)a-related proteins might be involved in the initial steps of taste signaling in these lineages. In line with this hypothesis, in a recent study where we comprehensively described the T1R gene repertoire of the carnivorous marine fish gilthead seabream (Sparus aurata [sa]), and it was also shown in vitro that heterologous expression of saT1R heterodimers co-transfected with the Gi alpha protein subunits saG(i)a1 and saG(i)a2 triggered both stimulatory and inhibitory taste transduction mechanisms upon amino acid activation (Angotzi et al., 2020). Hence, the overall emerging picture suggests a large degree of conservation of the T1R-mediated taste signaling across vertebrates, including fish.
However, despite the important progresses made to describe the functional and evolutive aspects of T1R and related taste signaling cascades in teleosts, many basic aspects of T1R biology remain largely unexplored. For instance, a putative element of the taste signaling pathway, namely Gi alpha protein-like immunoreactivity, has been localized in the GIT, in cells with an endocrine appearance, co-localizing with some peptides in the fish stomach (Latorre et al., 2013), but to our knowledge there is no direct evidence linking the presence of T1R and gut hormones in the same cell type (specifically EECs). Such evidence would be a fundamental stepping stone towards establishing the possible existence of gut sensing mechanisms operating in fish similarly to mammals. To the best of our knowledge, no study has characterized the ontogeny of the T1R gene system in early life stages of fish. Indeed, only a few published studies examined developmental aspects related to the fish gustatory system mostly focused on taste bud morphology (Hansen et al., 2002;Wang et al., 2016), cell patterning and distribution (Varatharasan et al., 2009) or development of oral taste functionality by behavioral methods (Kasumyan, 2001).
Having in mind these knowledge gaps, the objective of this study was to address aspects related to spatio-temporal gene expression patterns, and obtain anatomical information on the full set of saT1R genes in fish larvae at different stages of ontogeny and in selected tissues during adulthood. Moreover, we analyzed the specific gene expression patterns of the 2 signal-transducing components saG(i)a1 and saG(i)a2 to establish their potential colocalization with the saT1R3 gene expression, as the common subunit of T1R heterodimeric complexes, in the GIT of adult fish. Finally, we aimed to provide direct evidence for mRNAs coexpression of a subset of saT1R genes (mostly saT1R3) with selected gut peptides defining EEC-type identity such as ghrelin (ghr), cholecystokinin (cck), peptide YY (pyy) and proglucagon (pg), to establish a morphological link indicating possible roles of T1R chemosensing in gut nutrient-sensing mechanisms and in the regulation of fish digestive processes.

Animals and ethical statement
Gilthead seabream adults and newly hatched larvae were obtained in January 2018 from the fish farm Avramar (Spain), and were maintained in fiber-glass aerated tanks supplied with a continuous flow of seawater (37 g/L salinity, 16.9e17.2 C), and under a natural photoperiod at the facilities of the IATS institute (CSIC, Torre la Sal, Spain). Following the complete absorption of the yolk sac at 8 d post hatching (dph), larvae were fed on rotifers once per day from 9 to 17 dph, and gradually replaced by a mixed diet of rotifers and Artemia naupli as development progressed, until the last day of sampling (21 dph). Adult fish were fed twice daily on a standard commercial diet (Biomar, Spain), and were fasted for 24 h prior to tissue sampling. All experimental procedures were performed in compliance with the European Union guidelines for Care and Use of Laboratory Animals (2010/63/EU), and after the approval of the Welfare and Bioethical Committee of Instituto de Acuicultura de Torre de la Sal (IATS-CSIC) under the code 015/2013 and according to Royal Decree RD53/2013.

Quantification of saT1R mRNA abundance by real-time quantitative PCR (RT qPCR)
Real-time quantitative PCR analyses of saT1R1, saT1R2a, saT1R2b, saT1R2d, saT1R2e, saT1R2f and saT1R3 genes were performed using RNA pools of whole seabream larvae collected at 1, 3, 5, 7, 10 and 12 dph, i.e. spanning lifeestage transition from yolk-sac sustenance to exogenous feeding (initiated at 9 dph). For each stage analyzed, total RNA was extracted from triplicate samples, each containing approximately 15 pooled whole-body larvae. For qPCR analyses of adult tissues, 4 fish (n ¼ 4; 348 ± 53 g) were euthanized with an overdose of tricaine methane-sulfonate (MS-222; 400 mg/ L), and tissue samples ranging from 50 to 100 mg were dissected from the oropharyngeal area, including lips (L), gill filaments (G), the epithelium overlying the bony basyhyal (homologous to the tongue of tetrapods, T), and the mucous epithelium lining the inside of the oral cavity (OC). For GIT tissue sampling, stomach (St; posterior part) and intestine samples were dissected. The intestine was first equally divided into 3 major antero-posterior segments; then the middle portion of each segment was dissected for further processing and hereby defined as foregut (Fg), midgut (Mg) and hindgut (Hg). The 3 brain tissue compartments analyzed included telencephalic/hypothalamic (Forebrain, Fb), mesencephalic (midbrain, Mb) and romboencephalic (hindbrain, Hb) regions, respectively. Larvae and dissected tissues were mechanically homogenized in 1 mL TRIzol reagent (Invitrogen, St. Louis, MI, USA), and the concentration and purity of RNAs were determined by the optical density 260/280 ratio (>1.9), using a NanoDrop 2000c spectrophotometer (Thermo Scientific, United States). To eliminate potential genomic DNA, samples were treated with the TURBO DNA-free kit (Ambion, Life-Technologies, Austin, TX, USA) according to the supplier's protocol. The cDNAs were synthesized from 2 mg of DNase-treated RNAs using oligo(dT)12-18 primer and Superscript III (Invitrogen, Carlsbad, CA, USA) following the manufacturer's instructions. Target mRNAs were quantified in duplicate samples by real-time qPCR (Bio-Rad CFX96) using Sybr green PCR master mix (Invitrogen), 300 nM of forward and reverse primers, 50 ng cDNA template and nuclease-free water up to a final volume of 25 mL. All primers were designed using the free software Oli-goAnalyzer Tool (Integrated DNA technologies), to ensure similar melting temperatures, avoidance of self and hetero dimerization and a balanced G/C content. Sequences of primers used for reference and target genes are listed in Supplementary Table S1. Primer pair efficiency (E) was evaluated using a 2-fold dilution curve ranging from 100 to 6 ng cDNA pools, and was determined by formula E (%) ¼ (10 À1/slope À 1) Â 100; the primers with efficiency in the range of 95% to 105% were selected for quantitative gene expression analysis. PCR conditions were as follow: 50 C for 2 min; 95 C for 10 min; 40 cycles of 95 C for 15 s, 60 C for 1 min. Melting curve analysis to evaluate potential non-specific amplification was performed by ramping from 60 to 92 C, rising by 0.2 C every 1 s. Specificity of amplified PCR products was further confirmed by electrophoresis on a 1.2% agarose gel. Fold-change gene expressions of target transcripts were estimated using the mean normalized expression method of the Q-Gene application (Muller et al., 2002;Simon, 2003), using the stably expressed gene-elongation factor 2 as internal reference for data normalization. Relative gene expression results are expressed as the mean ± SEM. Statistical comparisons were analyzed by one-way analysis of variance (ANOVA) followed by Tukey multiple test, using GraphPad Prism 8. A P-value < 0.05 was considered to be statistically significant.

Chromogenic (CISH) and dual fluorescent mRNA in situ hybridization (FISH)
To prepare specimens for mRNA in situ hybridization on tissue sections, samples from stomach, pyloric caeca, proximal, middle and distal intestines were all dissected and immediately perfused with ice-cold PBS until the effluent was clear, fixed in PFA for 48 h at 4 C, and then dehydrated through a graded ethanol series before being embedded in paraffin wax. Sections of 7 mm were prepared on a Microm HM 355 microtome (Fisher scientific), collected on poly-L-lysine coated slides (VWR, Germany) and preserved at À80 C until the experiments were performed. Prior to in situ hybridization (ISH) procedures, slides were air-dried for 1 h and heated at 60 C for 10 min. Deparaffinization and rehydration were performed in xylene and ethanol series, 100% xylene (3 Â 5 min), 100% ethanol (2 Â 3 min), 95% ethanol (1 Â 3 min), 70% ethanol (1 Â 3min) and 50% ethanol (1 Â 3min). Tissue sections were then fixed in 4% PFA/PBS (pH 7.4) (1 Â 10 min), and permeabilized in 10 mg/mL of Proteinase K in 0.05 M Tris-HCl (pH 7.5; 1 Â 10 min) followed by washes in 1Â PBS (1 Â 5 min) prior additional fixation in 4% PFA/PBS (1 Â 10 min). Sections were rinsed in 1Â PBS (2 Â 5 min) and treated with freshly made acetylation solution containing 0.25% acetic anhydride and 0.2% HCl in 0.1 M triethanolamine buffer (pH 8; 1 Â 10 min). After rinsing again in 1 Â PBS (3 Â 5 min), tissues were rehydrated in a graded series of ethanol (95, 80 and 70%, 1 min each) and prehybridized in HB for 90 min; HB was then removed and replaced by one (for single FISH or CISH) or 2 (for dual FISH) cRNA probes. Different cRNA concentrations between 1 and 20 ng/mL were tested, and the best results were achieved using 8 ng/mL. Sections were covered with Grace Bio-Labs HybriSlip (Sigma Aldrich) and incubated for 16 h at 63 C in humidified chamber with wipes soaked in 5XSSC. Post-hybridization washes included 2XSSC (1 Â 30 min) and 50% formamide in 2XSSC (1 Â 30 min) at 63 C, followed by 10 mg/mL RNase A treatment, for 30 min at 37 C.
A.R. Angotzi, E. Leal, S. Puchol et al. Animal Nutrition 11 (2022) 293e308 For CISH procedure, slides were incubated in Immuno buffer containing 1:2,000 diluted anti-DIG/Fab fragments conjugated to AP and incubated O/N at 4 C. To develop the staining, sections were then washed in buffer 2 and incubated O/N at 4 C with NBT/BCIP chromogen substrates. Finally, sections were washed in buffer 3 (1 Â 30 min) to stop the reactions and mounted on 60% glycerol in buffer 3 until visualization.
For dual FISH detection, experiments were carried out as described by Hoang et al. (2016). Briefly, subsequent to the RNase A treatment, sections were incubated O/N at 4 C with an antibody solution containing 1:250 diluted anti-fluorescein Fab fragments conjugated with horse-radishperoxidase antibody (Sigma Aldrich). After several washes in 1Â maleate buffer (2 Â 10 min) and TNT buffer (100 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.05% Tween-20), (3 Â 5 min), green fluorescent signals were developed by tyramide stock solution 1:200 using 1Â Plus Amplification Diluent, according to the manufacturer's specifications (TSA PLUS fluorescein kit, PerkinElmer). The reaction was subsequently stopped in buffer 3 (4 Â 5 min), and the slides were incubated O/N in humidified chamber at 4 C with antibody solution containing 1:2,000 diluted anti-DIG/Fab fragments conjugated to AP, as previously described for single CISH detection. Sections were then washed in visualization buffer (100 mM Tris-HCl pH 8.2 and 0.1% Tween-20) and red fluorescent staining visualized with SIGMA FAST TR/ naphtholAS-MX/Tris buffer, prepared according to the manufacturer's procedures (Sigma, Aldrich). For blue fluorescent DNA stain, slides were incubated (1 Â 1 min) in 0.5 mM of bisBenzimide H 33342 trihydrochloride (Hoechst; Invitrogen). Finally, cover slips were mounted onto slides using Pro-Long Gold mounting media (Thermo Fisher scientific). Images were visualized with BX41 fluorescent microscope (Olympus), using 10Â and 20Â objectives, and documented with cellSens 1.18 software (Olympus), followed by brightness and contrast adjustments using Adobe Photoshop CS5 (Adobe Systems Inc., San Jose, CA), before being analyzed and mounted on multi-panel figures.

Riboprobes preparation
The DIG and fluorescein-labeled riboprobes for saG(i)a1-2, saT1R1, neurogenic differentiation 1 (nd1), ghr, cck, pyy, pg and pomcb genes were all synthesized from PCR-amplified DNA fragments using the primers listed in Table S1. All PCR-products were cloned into pGEM-T Easy vector plasmids (Promega, Madison, WI), and were Sanger-sequenced to confirm DNA specificity. Linearized gene specific plasmids were then subjected to in vitro transcription using 25 U of SP6 or T7 RNA polymerases (Promega) in the presence of DIG-labeled or Fluorescein-labeled UTPs, following the manufacturer's instructions (Roche Diagnostic). Synthesized cRNA probes were subsequently precipitated with 2.5 Â 100% ethanol/ LiCl (3M), and spectrophotometrically quantified. To make sure that the different cRNA probes used in multiple ISH do not cross-react, we aligned the cRNA sequences of interest using CLUSTALX V1.81. Probe sizes and percentage of nucleotide identity between the conserved sequence targets of the T1R's probes are presented in Table S1 and Supplementary Table 2 (Table S2), respectively.

saT1R qPCR in adult tissues
Expression profiles of the 7 saT1R genes were also examined by qPCR in several adult tissues. Their distribution patterns were visualized by plotting the relative mRNA abundance of each saT1R gene in 3 regions: oropharyngeal, GIT and brain tissues. In the oropharyngeal region ( Fig. 2A), the 7 saT1R gene transcripts were found in all tissues investigated (lips, tongue, gill filaments and oral cavity epithelium), and significantly higher expression levels were found for the following: 1) saT1R3 in gills (P < 0.05 for saT1R1/ saT1R2a/d/e comparisons), tongue (P < 0.05 and P < 0.01, for saT1R1 and saT1R2a/b/d/e comparisons, respectively) and in the oral cavity epithelium (P < 0.05, for saT1R2a/b/d/e comparisons), and 2) saT1R2b and saT1R2f genes in the gills (P < 0.01 for saT1R1/saT1R2a/ d/e comparisons). In the GIT (stomach, foregut, midgut and hindgut, Fig. 2B), lower saT1R mRNA levels were generally observed when compared to oropharyngeal tissues (note the differing y-axes scales of Fig. 2A versus Fig. 2B). The saT1R3 gene was significantly higher expressed than the remaining T1R gene-set in both midgut (P < 0.05) and hindgut (P < 0.01) segments, while no significant differences were found between saT1R1/saT1R2a/d/e/f in these 2 regions, and among all saT1R in stomach and foregut compartments. In the brain (Fig. 2C), some saT1R genes showed remarkable high levels of expression, especially in fore-and hindbrain regions. In the forebrain, saT1R2d was the highest expressed gene for all statistical comparisons (P < 0.001), followed by saT1R2e (P < 0.01 and P < 0.05, for saT1R1/saT1R2a/b/R3 and saT1R2f comparisons, respectively). In the midbrain, significantly higher expression levels were found for saT1R2b when compared to saT1R2d (P < 0.05), saT1R1/R3 (P < 0.01) and saT1R2e/f (P < 0.001), and for saT1R2a when compared to saT1R2e/f transcripts (P < 0.05). In the hindbrain, saT1R2d was expressed at roughly 20-fold higher levels than saT1R2e, and 250-or 2,000-fold higher than saT1R2f and saT1R1/ saT1R2a/b/R3, respectively (P < 0.001 for all comparisons); all remaining saT1R comparisons were not statistically significant.

WISH studies of saT1R in seabream larvae
The WISH techniques were used to localize saT1R mRNA expression in whole larvae in five selected stages, including yolksac nourishment (5 dph) and exogenously feeding (on rotifers) larvae at 4 post-feeding stages (11, 14, 17 and 21 dph). An overview of the saT1R's sites of expression in developing larvae is shown in Figs. 3 and 4, where the anatomical localization is indicated. In prefeeding larvae, the 7 saT1R exhibited overlapping expression patterns in the stomach and the foregut (Fig. 3AeD; Fig. 4AeC), while additional expression in the most posterior region of the intestine was also observed for saT1R2b (Fig. 3C). At 11 dph, saT1R patterns remained essentially conserved in the 3 developing portions of the gastrointestinal tract ( Fig. 3EeH; Fig. 4DeF), and though exogenous feeding had already started, saT1R expression was not detected in oral taste tissues (lips, tongue, oral cavity epithelium) for the entire saT1R gene set. During post-feeding stages, mRNAs became gradually detected in the oral cavity epithelium and/or tongue and pharynx, while overall maintaining a strong expression in the stomach and intestine. In particular, oral tissue saT1R2e (E), saT1R2f (F), saT1R3 (G) at 1, 3, 5, 7, 10, 12 d post hatching (dph). saT1R2b mRNAs expressed roughly 300-fold higher levels of saT1R2f and 60, 30, 10 and 5-folds of saT1R3, saT1R2e, saT1R2a and saT1R1, respectively (H). Data are represented as the fold change differences of target gene expression to the reference gene-elongation factor 2, per 100 ng of input RNA/sample. All experiments were performed in triplicates (n ¼ 3), each containing approximately 15 pooled whole-body larvae; data are expressed as means ± standard error of the mean (SEM). Different letters indicate significant differences between experimental groups. Asterisks indicate significance levels (*P < 0.05, **P < 0.01, ***P < 0.001) after one-way ANOVA followed by Tukey's multiple comparisons tests (GraphPad Prism version 8.0). expression was observed for saT1R2b at 14, 17 and 21 dph (Fig. 3K, O, S); for saT1R1, saT1R2d and saT1R3 at 14 and 17 dph (Fig. 3M, Q;  Fig. 3P, T and Fig. 4L, O, respectively) and for saT1R2e at 21 dph (Fig. 4M). A summary of gut vs oral sites of saT1R expression throughout ontogenesis is provided in Table 1. Negative controls using DIG-labeled saT1R1/R2b/R3 sense RNA probes were virtually devoid of labeling in gastrointestinal regions, although chromogenic signals were some time detected in the developing otic vesicles (Fig. 5AeC). The cRNA antisense probes synthesized from the pituitary/hypothalamic gene marker pomcb yielded, as expected, well-defined chromogenic signals in the medio-basal hypothalamus (Fig. 5DeF).

ISH studies of Neurod1, saT1R, saG(i)a1 and saG(i)a2 in the intestine of adult seabream
This set of experiments was designed to investigate the spatial pattern of gene expressions of nd1, saT1R1, saT1R2b, saT1R3 and saG(i)a1-2 in presumptive EECs of the intestine. An additional objective was to investigate the potential colocalization of gene expression in EECs for the following: 1) two of the subunits of heterodimeric complexes forming functional taste receptors (saT1R2/saT1R3), and 2) saT1R3 (the common subunit of T1R heterodimeric complexes) with either saG(i)a1 or saG(i)a2 paralogs, to potentially provide insights into the evolutionary conservation of the G(i)a-mediated-taste signal transduction.
By using the CISH methods, we first characterized the pattern of expression of nd1, a member of the basic-helix-loop-helix (bHLH) family of transcription factors. Transcripts of nd1 were found throughout the gut portions analyzed, in 3 main locations: 1) in the upper layer of the intestinal mucosa, adjacent to mucus-secreting goblet cells (Fig. 6A), 2) in the intermediate mucosa ( Fig. 6B1-B2 and 3) in the bottom layer of the mucosa (Fig. 6C). The nd1 positive (þ) cells were particularly abundant in the midgut, where proliferative outbreaks next to the lamina propia were also identified (Fig. 6D). The nd1 expression was drastically reduced and essentially restricted to few presumptive EECs in distal (hindgut) segments of the intestine (Fig. 6E).
Based on the previous qPCR analysis, we selected the highest saT1R expressed genes in the intestine (saT1R3, saT1R1 and saT1R2b) for further detailed investigation. Initial examination by CISH methods showed that they were mostly expressed in the upper mucosal lining, contiguously to goblet cells, exhibiting an expression pattern similar to that shown by a subpopulation of nd1 (þ) cells ( Fig. 6F and G). These observations were further corroborated by single-color fluorescent detections using fluorescein-TSA ( Fig. 6H; saT1R3) and Dig-FastRed ( Fig. 6I and J for saT1R3 and saT1R2b, respectively). Next, we investigated whether saT1R3 and saT1R2b gene transcripts were co-expressed in the same presumptive EECs by dual FISH methods, providing evidences that both genes could have either common or independent spatial patterns of expression. Specifically, saT1R3 (þ) cells were by far more abundant than saT1R2b (þ), and rarely co-expressed saT1R2b ( Fig. 6L; Fig. 6K-K2). saT1R2b was almost always coexpressed with saT1R3, although it was occasionally found in presumptive EECs not expressing saT1R3 (Fig. 6L). To verify the hypothesis that T1R genes have, at least partially, nutrient sensing roles in EECs, we next sought to test by dual FISH if, and to what extent, nd1 (þ) EECs also expressed saT1R3. Consistently with the CISH experiments reported above, nd1 (þ) cells were mainly found in proximal and medial intestine segments (Fig. 6M2, N2), where they occasionally co-expressed saT1R3 transcripts (Fig. 6M3, N3).
Furthermore, we conducted a series of ISH experiments to characterize the spatial expression patterns of the 2 gene paralogs saG(i)a1 and saG(i)a2 by single CISH methods. Both transcripttypes were found abundantly expressed in a widespread fashion resembling the 3 main spatial domains previously described in this study for the EEC-marker nd1 in pyloric, fore-and midgut segments ( Fig. 7AeC and Fig. 7EeG for saG(i)a1 and saG(i)a2, respectively). In the hindgut, saG(i)a1 and saG(i)a2 expressing cells were less frequent (data not shown). When dual FISH was employed for both saG(i)a1/saT1R3 and saG(i)a2/saT1R3 probe combinations, several cases of co-localization in saT1R3 (þ) EECs with both saG(i)a 1-2 transcripts were observed (Fig. 7D-D3 and Fig. 7H-H3 for saG(i)a1/saT1R3 and saG(i)a2/saT1R3 combinations, respectively). Graphical representation of RT qPCR analyses of seabream (sa) taste 1 receptor (T1R) subunits, saT1R1, saT1R2a-f and saT1R3 in adult seabream tissues of oropharyngeal (A), gastrointestinal tract (B) and brain tissues (C). Data are represented as the fold change differences of target gene expression to the reference gene-elongation factor 2, per 100 ng of input RNA/tissue. L ¼ lips; G ¼ gill filaments; T ¼ tongue; Oc ¼ oral cavity epithelium; FG ¼ foregut; MG ¼ midgut; HG ¼ hindgut; FB ¼ forebrain; MB ¼ midbrain; HB ¼ hindbrain. All experiments were performed in quadruplicates (n ¼ 4); data are expressed as means ± standard error of the mean (SEM). Different letters indicate significant differences after one-way ANOVA followed by Tukey's multiple comparisons tests (see section of results for significance levels).
Animal Nutrition 11 (2022) 293e308 3.5. ISH studies of ghr, cck, pyy and pg hormone genes and their colocalization with the saT1R3 subunit in the intestine of adult seabream In another set of ISH experiments, we aimed to provide direct evidence for mRNAs co-expression of the gut peptides ghr, cck, pyy and pg with saT1R3 gene, to support the hypothesis that gut hormone secretion upon T1R-mediated gut sensing might also occur in fish. Using cRNA probes we found that ghr was abundantly expressed within scattered cells in the gastric mucous membrane (Fig. 8A); this first experiment was intended to test the effectiveness of our protocol in a tissue where this gene is known to be highly expressed. When using intestinal tissues, ghr (þ) cells were also clearly identified in presumptive EECs of the proximal intestine by CISH (Fig. 8B). Dual FISH in this area showed several ghr (þ) cells that did not co-express saT1R3 (Fig. 8C), as well as saT1R3 (þ) cells devoid of ghr expression (Fig. 8D). However, a few cases of ghr/ saT1R3 co-localization were also identified ( Fig. 8E-E3). No evidence of ghr expression was detected in distal portions of the gut (not shown).
Next, we analyzed cck mRNA localization throughout the intestine, initially by CISH, and found a particularly high density/ abundance of transcripts, as deduced by the intense chromogenic staining. A strikingly elongated cck cell morphology was observed in proximal ( Fig. 8F and G), but not in distal (Fig. 8H) gut segments. Dual FISH experiments using cck and saT1R3 cRNA probes revealed that the 2 targets largely exhibited independent spatial domains (Fig. 8I), although some cases of co-localization were also observed (Fig. 8JeJ3). Likewise, pyy and pg-expressing EECs were also found along intestinal segments, although usually with low abundance. Particularly, flask shaped pyy (þ) cells were clearly identified in the midgut, by both CISH (Fig. 9A, C) and green-FISH (Fig. 9B-B2), and additionally in the foregut, where a few cases of co-localization with saT1R3 transcripts were observed (Fig. 9DeD2). The pg A.R. Angotzi, E. Leal, S. Puchol et al. Animal Nutrition 11 (2022) 293e308 transcripts were even more rarely detected than those of pyy. Indeed, pg (þ) cells were only sparsely noticed, with no clear concentration or pattern (Fig. 9F), although few cases of colocalization of expression with saT1R3 were also observed ( Fig. 9E-E2 and Fig. 9G-G2, in fore-and midgut respectively). Additionally, pg transcripts were identified in presumptive EECs of the hindgut, although clear ISH signals were only visible by CISH methods (Fig. 9H).

Discussion
In a recent study, we reported the complete T1R gene repertoire of gilthead seabream, consisting of eight members including saT1R1, saT1R3 and six saT1R2 (a-f), and functionally characterized the in vitro responses of a subset of heterodimers, namely saT1R1/ R3, saT1R2a/R3 and saT1R2b/R3, to L-amino acids and sweet ligands (Angotzi et al., 2020). Here, we further explored the mRNA A.R. Angotzi, E. Leal, S. Puchol et al. Animal Nutrition 11 (2022) 293e308 expression profiles of all saT1R in both larval and adult tissues. During early larval development, mRNA levels were quantified in whole-body of larval stages encompassing the transition from endogenous to first exogenous feeding (initiated at 9 dph). The entire T1R gene repertoire was expressed from 1 dph onwards, without significant variations until 10 dph. At 12 dph, as soon as first exogenous food is digested and metabolized, all saT1R transcripts, and especially saT1R2b, significantly increased their expression levels. The highest level of expression and abrupt rise of saT1R2b expression following first exogenous feeding and its earlier appearance in oropharyngeal tissues as deduced by the whole mount in situ analyses, suggests that this paralog may be playing major roles related to feeding. In support of this hypothesis, saT1R2b/R3 was the most responsive and sensitive heterodimer to L-amino acid stimulations in this species (Angotzi et al., 2020). Paradoxically, the saT1R3 gene encoding the shared subunit of receptors signaling both sugar (T1R2/T1R3) and protein (T1R1/T1R3) rich foods in mammals (Roper, 1989;Hoon et al., 1999;Adler et al., 2000;Yarmolinsky et al., 2009), and responding to a wide spectrum of L-amino acids (T1R1/T1R3 and T1R2 n /T1R3) in fish (Oike et al., 2007), was among the lowest expressed gene throughout the larval stages analyzed. It is believed that alternative dimeric arrangements among T1R subunits might also potentially occur in cell membranes, including homodimerization among the highest expressed saT1R subunits (Damak et al., 2003;Herness, 2018), albeit these types of protein combinations have mainly been reported for the T1R3 homodimer (Masubuchi et al., 2013;Kojima et al., 2014;Lee and Cohen 2015;Mafi et al., 2021). Therefore, the reason behind these unexpected differences in gene expression levels remains elusive at present. The gene expression profiling during early development was extended (up to 21 dph) through mRNA localization studies by whole-mount ISH. Surprisingly, the results revealed that at the earliest larval stages analyzed (5 and 11 dph), the entire set of saT1R transcripts were virtually confined to the developing GIT. Expression of saT1R in oropharyngeal regions was detected at 14 dph for saT1R2b, and gradually followed by the other saT1R members, except T1R2f whose expression was only observed in the GIT at 21 dph. Altogether, these findings suggest that saT1R expression during early larval stages spanning first exogenous feeding arise heterochronously, and that extraoral (gastrointestinal) taste-related perception occurs earlier than the buccal/intraoral perception. However, starting at 14 dph onwards both tissue modalities coexist. The observed "delayed" onset of saT1R expression in oropharyngeal tissues is consistent with the notion that, in most fish, functional taste buds and the capability to efficiently perceive gustatory qualities might arise later in development, although morphologically mature taste buds appear just before or at the onset of exogenous feeding in some species (Hansen et al., 2002;Kasumyan, 2001). Marine fish larvae, in particular, are considered primarily visual feeders, with olfaction and the lateral line providing additional inputs to visually oriented feeding (Rønnestad et al., 2013). For instance, the study of the development of sensory systems in sharpsnout seabream (Diplodus puntazzo) larvae showed that at the start of exogenous feeding larvae rely on olfactory ciliated and microvillate sensory cells and free cephalic neuromasts, besides vision, to detect the presence of food, while intraoral and extraoral taste buds only differentiated in the post-larval stage at 48 and 51 dph, respectively (Boglione et al., 2003). Accordingly, behavioral experiments testing agar pellets indicated that fish larvae at first feeding respond only to a limited number of taste stimuli, but as larval development progresses, the spectrum and effectiveness of amino acid perception greatly increases (Hughes, 1991(Hughes, , 1993Kasumyan and Sidorov, 2005;Kasumyan, 2001).
The biological significance of saT1R expression in lecithotrophic stages undergoing organogenesis of the gastrointestinal tract is unclear considering that larval development and nutrition at this point are fueled by the endogenous energy supplies of the yolk. The transition period from exhaustion of yolk reserves to the onset of Table 1 Summary of saT1R(s) patterns of expression as deduced by whole mount in situ hybridization analyses in gut and oral tissues during larval ontogenesis.

saT1R(s)
5 dph Gut Gut Gut Gut saT1R ¼ seabream taste 1 receptor; dph ¼ days post-hatching. A.R. Angotzi, E. Leal, S. Puchol et al. Animal Nutrition 11 (2022) 293e308 exogenous feeding is a critical step for larval survival, and has been associated with massive mortalities depending on the fish species (Yúfera et al., 2005;Palazzi et al., 2006). Successful transition to exotrophic life stages largely relies on the functional maturation of the gastrointestinal structures necessary for food digestion and absorption, alongside the development of efficient signaling pathways through which the brain and the gastrointestinal system communicate to regulate food intake and energy homeostasis.
Newly hatched larvae have a rudimentary tubular intestine, but the intestinal mucosa starts to increase its thickening and folding concomitantly with mouth opening (i.e., few days before the transition to exogenous feeding). At this point, enterocytes initiate their apical differentiation by developing a brush border microvillus membrane that becomes fully functional at approximately the third week post-hatching (Cahu and Zambonino-Infante, 1994;Moyano et al., 1996;Ribeiro et al., 1999; Zambonino-Infante et al., A.R. Angotzi, E. Leal, S. Puchol et al. Animal Nutrition 11 (2022) 293e308 2008). Despite the low degree of morphological differentiation, at mouth opening most fish species investigated so far, including marine (Zambonino-Infante and Cahu, 2001;Rønnestad et al., 2013;Yúfera and Darias, 2007) and freshwater (Lahnsteiner, 2017) species, have active cytosolic (intracellular; e.g., leucineealanine peptidases) and secretory (pancreatic, trypsins, lipases and amylases) proteolytic, lipolytic, and carbohydrate splitting enzymes in their intestines. These findings suggest that the onset of digestive enzyme expression and activity is not induced by food but rather genetically hardwired. Similarly, mRNA transcripts coding for endocrine hormones and neuropeptides of the brain-gut axis have been reported prior to the onset of exogenous feeding, including ghr, cck, neuropeptide Y, pomc, cocaineamphetamine-regulated transcript and prepro-orexin (Hoang et al., 2016;Ping et al., 2014;Kurokawa et al., 2000). In line with these observations, it is reasonable to postulate that the early (prefeeding) onset of the molecular machinery underlying saT1R expression in the gastrointestinal tract might be an anticipatory and genetically programmed mechanism, gradually yielding functional saT1R proteins to possibly exert chemosensory roles in the gut of older larvae and adult fish. Next, we sought to ascertain the distribution of saT1R mRNA transcripts in different adult fish tissues, to establish possible similitudes with mammalian vertebrates, in which T1R are widely expressed in body tissues, where they perform chemosensory functions beyond oral taste sensation (Finger and Kinnamon, 2011). Hence, gene expression was quantified by qPCR in different tissues within oropharingeal, GIT and brain regions. Our findings are in accordance with previous studies that reported several taste receptors and taste signaling components in the GIT of different fish species (Polakof and Soengas, 2013;Latorre et al., 2013;Ronnestad et al., 2016;Yuan et al., 2020;Calo et al., 2021;Kinnamon, 2012;Morais, 2017), reinforcing the hypothesis that the T1R-mediated gut sensing mechanisms could be conserved throughout vertebrate evolution. The significantly higher levels of saT1R3 expression observed in oropharyngeal and GIT tissues suggest that this gene might be locally demanded at higher transcriptional rates due to heterodimerization with saT1R1 or saT1R2 subunits, since saT1R3/ A.R. Angotzi, E. Leal, S. Puchol et al. Animal Nutrition 11 (2022) 293e308 saT1R3 homodimers do not seem to respond to L-amino acids stimulations in this species (Angotzi et al., 2020). However, the existence of T1R3 homo-oligomers cannot be dismissed since these have been described in mammals and proposed to sense calcium and magnesium taste (Nelson et al., 2001). Interestingly, there were significantly higher mRNA levels of both saT1R2b and saT1R2f than saT1R2a/T1R2d/T1R2e in gills. These gene expression patterns are in agreement with those recently described for T1R2b/e ortholog counterparts in grass carp (Ctenopharyn godonidellus) (Yuan et al., 2020), suggesting that the T1R2-mediated chemosensing functions may have been highly retained in gills throughout teleost radiation. We found also high levels of expression for the paralog subtypes saT1R2d/e and saT1R2d/e/f in forebrain and hindbrain, respectively, suggesting that these genes might have a tissuespecific chemosensory role in these brain compartments. In accordance, specialized glucose-sensing neurons mainly located in paraventricular and arcuate nuclei of the hypothalamus and in the nucleus of the solitary tract of the brainstem, are known to regulate extracellular glucose concentration through sweet taste-like signaling in murine models (Ren et al., 2009;Herrera Moro Chao, 2016;McCaughey, 2021). Similarly, the T1R nutrient-sensing functions with implications on food intake have also been described in some brain regions of rainbow trout (Oncorhynchus mykiss) (Otero-Rodiño et al., 2015;Comesaña et al., 2018a;2018b).
In the present study, we additionally designed a comprehensive set of experiments with the purpose of describing the spatial pattern of expression of saT1R1, saT1R2b, saT1R3, saG(i)a1 and saG(i)a2 genes, and to substantiate possible mRNA co-expression of the saT1R3 subunit with both saG(i)a1-2 and with selected gut hormones (ghr, cck, pyy and pg) in presumptive EECs along the intestine. In both mammalian and fish models, it has been previously shown that the bHLH transcription factor nd1 plays essential roles to direct intestinal progenitor cells to an EEC fate, and that it is selectively expressed in this GI cell population (Li et al., 2011(Li et al., , 2019Ye et al., 2019). Therefore, we first identified presumptive EECs employing nd1 as a specific EEC-marker, and found that it is A.R. Angotzi, E. Leal, S. Puchol et al. Animal Nutrition 11 (2022) 293e308 expressed in cells located in different layers of the intestinal mucosa, possibly reflecting a continuous epithelial renewal in spatially distinct compartments (Sun et al., 2018). EECs represent a small population of scattered and highly specialized gut epithelial cells that respond to luminal contents, acting as chemoreception units capable of releasing signaling factors (Raybould, 2010;Young, 2011). Their sensory properties are exerted by different nutrient and non-nutrient-sensing receptors, mainly GPCRs, implicated in the perception of glucose, amino acids, fatty acids, bile acids, phytochemicals or secondary products derived from microbial fermentation (Gribble and Reimann, 2019). We demonstrated that saT1R genes are mostly expressed in presumptive mature EECs located in the upper epithelial lining of the intestinal mucosa, and that saT1R3 (þ) cells are usually found lying contiguously to mucus secreting goblet cells, often in contact with the gastrointestinal lumen. In addition, saT1R3 (þ) EECs were also detected near the base of the lamina propria membrane, apparently without reaching the intestinal lumen. Although the interpretation of this cell patterning heterogeneity is challenging, it correlates well with the "open" or "closed" EEC-types described in higher vertebrates to sense gut contents either directly (open-type), or indirectly (closed-type) through neural or humoral pathways (Sternini, 2007;Latorre et al., 2016). The examination of overlapping expressions of saT1R2b/R3 by dual FISH revealed the existence of a greater population of saT1R3 (þ) cells whose major fraction did not express saT1R2b, while the latter was almost always co-expressed with saT1R3. Similar mutual distributions of the single T1R subunit components have been previously described in both oral (Nelson et al., 2002;Li, 2009) and gastrointestinal (Daly et al., 2013) tissues of mammals, and in oral tissues of fish (Oike et al., 2007). However, this study provides for the first time in situ morphological evidence that heterodimerization is likely an evolutionary preserved mode of taste receptors coupling in fish gut sensing. While the presence of saT1R3 (þ) cells that did not express saT1R2b (þ) is consistent with the existence of different T1R3 (þ) subpopulations selectively co-expressing one (or A.R. Angotzi, E. Leal, S. Puchol et al. Animal Nutrition 11 (2022) 293e308 more) T1R subunits (Oike et al., 2007), the functional significance of the small fraction of saT1R2b (þ) cells devoid of saT1R3 expression is yet to be explained, and again suggests that this receptor subunit could additionally function as a monomer or homodimer (Herness, 2018). Since the first report uncovering the lack of Ga-gust orthologs in fish genomes (Oka and Korsching 2011;Ohmoto et al., 2011), other Gai subunits have been proposed to mediate taste signal transduction, and studies using immunoreactive and quantitative molecular assays further documented Gai expression in the GIT of some fish species (Latorre et al., 2013;Calo et al., 2021). Here, we provided clear evidence that saG(i)a1 and saG(i)a2 genes are highly expressed in the proximal GIT, with spatial expression patterns resembling those previously described for the EEC marker nd1. Through dual FISH assays, we further demonstrated that both saG(i) a genes are expressed in EECs potentially implicated in saT1R3mediated molecular sensing, thus supporting their functional homology to Ga-gust as intracellular taste-like transducer(s) in the GIT (Bertrand, 2009;Young, 2011;Angotzi et al., 2020).
Mammalian EECs are known to produce several peptides, and have been traditionally classified according to the hormones they secrete. The best characterized EECs are the X/A-cells (in mice) or P/ D1-cells (in humans) producing GHR, L-type cells producing glucagon-like peptides 1 and 2 and peptide YY (PYY), the I-type cells producing cholecystokinin (CCK), and the K-type cells producing glucose-dependent insulinotropic polypeptide (Sj€ olund et al., 1983). Furthermore, different EEC subsets can overlap in the co-expression of multiple hormones (Habib et al., 2012;Latorre et al., 2016;Fothergill and Furness, 2018). To verify assumptions based on mammalian studies that T1R might function as EECsensory transducers (Burman and Kaji, 2021) in fish, we aimed to determine whether the saT1R3 gene was expressed in fish-like X/A-L-, or I-specialized EEC types. Using dual FISH methods, we identified some cases of co-localization between saT1R3 and ghr, cck, pyy and pg genes in different regions of the GIT, as well as independent and non-overlapping expression domains. While these spatially correlated patterns of expressions corroborate a plausible direct role for saT1R as nutrient-sensing targets regulating hormone secretion in seabream, the identification of additional saT1R3 (þ) cell-subsets devoid of endocrine peptides expression could indicate that saT1R-mediated chemosensing functions might occur in these tissues via mechanisms that are both dependent and independent of endocrine pathways. Indeed, in addition to their potential function as nutrient sensors participating in food digestion, nutrient absorption and metabolism, mammalian T1R, together with T2Rs, have been proposed to also regulate gut innate immune responses to compounds secreted by microbial pathogens (Lee and Cohen 2015;Triantafillou et al., 2018). On the other hand, putative L-, K-or I-EEC types that did not express saT1R3 might potentially be equipped with other nutrient sensors such as extracellular calcium sensing receptors, taste variants of metabotropic glutamate receptors and free fatty acid receptor 2/3, among others (Raka et al., 2019;Burman and Kaji, 2021).
Interestingly, single-cell RNAseq surveys of the murine small intestine recently identified a broader set of genes for different epithelial cell lineages (goblet, Paneth, or tuft cells), including Krüppel-like factors (KLf3-6), mucosal pentraxin 2, and epithelial cytokines (thymic stromal lymphopoietin and leukocyte common antigen) (Haber et al., 2017). In the current work, the employment of seabream orthologs to these gene markers would have contributed to the characterization of diverse epithelial cell types potentially present in the gut of fish. Unfortunately, many of these markers have not yet been annotated in the seabream genome, and therefore it was not feasible to perform such type of cellular screening. Finally, drawbacks related to practical aspects of fluorescent imaging procedures (variability among experiments and photobleaching, among others), coupled to the remarkably low number of identified saT1R3þ cells, hampered our efforts to acquire reliable quantitative estimations of gut hormone þ cells coexpressing saT1R3. Future work in fish model species that are amenable to a dual reporter transgenic approach in vivo, would further verify the hypothesis of T1R chemosensory roles in the regulation of fish digestive processes put forward in this study.

Conclusions
Altogether, these findings provide new information on the T1Rmediated chemosensing capabilities in the GIT of a carnivorous fish species, and suggest a likely evolutionarily conserved role for saT1R as nutrient-sensors modulating gut hormone secretion. Furthermore, our data support the hypothesis that the saT1R-mediated gut sensing mechanisms might occur at least partially, through the involvement of the sensory transducers saG(i)a1 and saG(i)a2, thus validating their functional homology to the mammalian G(i)a subunit gustducin as taste-like intracellular components in the GIT of fish.

Declaration of Competing Interest
We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, and there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the content of this paper.