Atlantic salmon (Salmo salar L.) were raised indoors at the Conservatoire National du Saumon Sauvage (CNSS, France, 45°N¹). The CNSS hatchery produces juvenile salmon, which are to be released at different developmental stages in areas of the Loire-Allier basin requiring supplementation, as part of a restoration program to enhance the native Atlantic salmon population. The fish used for the mRNA detection in the different tissues were from a group of smolts used in a previous study (Nisembaum et al., 2020). Briefly, they were reared indoors under simulated natural conditions of photoperiod and natural temperature [February: 11L_{(07:30–18:30)}/13D and 4°C; July: 16L_{(06:00–22:00)}/8D and 15°C]; they weighed ∼30 g in February, and ∼80 g in July. Details are provided in Figures 1, 2 of Nisembaum et al. (2020). The fish used for the ISH and IHC (250 g b.w.), and in vitro pharmacological studies (600 g b.w.) were from the hatchery’s broodstocks (each year, a batch of ∼8 months old fish exceeding 145 mm is retained in the hatchery for 3 years to become the future broodstocks). The 1st generation hatchery-reared progeny was obtained from wild male and female adult salmon caught in the Allier River (620 km from the Loire estuary). Rearing occurred under simulated natural photoperiod and natural temperature, and standard hatchery conditions as described elsewhere (Martin et al., 2012; Nisembaum et al., 2020). At their juvenile stage fish were distributed in four 9 m³ cylindrical tanks (depth 0.5–0.7 m) at a density that did not exceed 10 kg/m³. The tanks were supplied with running water from the Allier River, at a flow of 3 l/s until April, which was progressively increased to 7 l/s when water temperature reached values above 13°C. This ensured a concentration of dissolved oxygen higher than 7 mg/l (Martin et al., 2012). As detailed elsewhere (Nisembaum et al., 2020), all fish were fed during daytime using automatic feeders. The acclimation conditions at the CNSS were in accordance with the “Agreement N° B43 056 005” (Arrêté N° DDCSPP/CS/2016/40), and the experimentation followed the guidelines and regulations approved by the “Ethics Committee for Animal Experiment of Languedoc-Roussillon (C2EA-LR/C2EA-36)” N° A6601601, and the European Union regulations (European directive 2010/63/EU).

Sampling was performed in the morning (between 10:00 and 12:00). Animals were anesthetized with 2-phenoxy−ethanol (0.5 mg/l) and then killed by decapitation. Tissues and organs were sampled and immediately dipped in the appropriate solution: (i) RNA later, first at + 4°C for 24 h, and then at −80°C until RNA extraction, for the real time quantitative PCR (RT-qPCR) studies; (ii) ice cold fixative [freshly prepared 4% paraformaldehyde (PFA) in phosphate buffer (PB; 0.1 M, pH 7.4)] for the ISH and IHC studies; (iii) ice cold freshly prepared culture medium for the culture of pineal organs and pharmacology studies.

Total RNA extraction from adipose fin, blood, cerebellum, diencephalon, gills, heart, intestine, kidney, optic tectum, pineal organ, pituitary, retina, skin (including the lateral line), spleen, and saccus vasculosus, was performed using an automated system and kit (Maxwell^®; Promega, Charbonnières-les-Bains, France) according to the manufacturer’s protocol. Retro-transcription was performed with 1 μg of RNA for all tissues except for the saccus vasculosus (for which 0.5 μg were used), using the PrimeScript™ 1st strand cDNA synthesis kit (Takara Bio Inc., Ozyme, Saint-Quentin-en-Yvelines, France). The abundance of the mRNA corresponding to the genes studied [trpv1 (NM_001140498.1; Leong et al., 2010), trpv4, (KJ135123.1; Nisembaum et al., 2015) and the reference gene ef1α (NM_001141909.1; Leong et al., 2010)] was quantified using a Light-Cycler™ system 2.0 (Roche; Meylan, France). The reactions were performed in a 20 μl final volume, containing 10 μl of LightCycler-FastStart DNA Master SYBR-Green I™ Mix (Roche Diagnostics; Meylan, France), 0.2 μM specific primers (Eurofins, Ebersberg, Germany; Table 1), and 2 μl of 1/5 diluted cDNA. The amino-acid sequences amplified are detailed in Supplementary Figure 1. The amplification protocol was as follows: 1 cycle of enzyme activation at 95°C for 3 min, and 40 cycles consisting in 95°C for 3 s, 60°C for 30 s and 72°C for 20 s. All samples were analyzed in duplicates and the relative expression (ΔΔCT) was performed according to Livak and Schmittgen (2001), taking into account the efficiency of the PCR reactions (Pfaffl, 2001). The efficiency of the amplification for all the genes investigated was around 100%. The specificity of the amplification reactions was confirmed by the melting temperature in each sample, through a melting curve protocol at the end of the 40 cycles of amplification, and by the size of the PCR products, obtained in an agarose gel. The data are presented as the mean ± SEM of n = 4–8 samples.

Pineal glands from adult fish (250 ± 50 g b.w.), were used for these assays. After 24 h fixation in PFA (see above) the samples were washed in PB, dehydrated in graded ethanol series (70, 95, 100%), dipped 3 min in toluene and then in Paraplast^® (at 60°C); after 15 h of impregnation, they were embedded in a new bath of Paraplast^®. Eight micrometers thick sections (using a MicroM HM 340^E microtome) were layered on glass slides (coated with a 2% solution of 3−aminopropyl-triethoxy-silane). The ISH and IHC (3 glands for each procedure) were performed on sections that were successively deparaffinized in toluene, rehydrated (through descending ethanol series) and placed in PB saline (PBS).

The ISH was performed on proteinase K treated sections (5 mg/ml; for 6 min at 37°C), using digoxigenin (DIG) labeled probes. The preparation of the riboprobes and the ISH procedure were as detailed elsewhere using a commercial kit (Roche-Diagnostics DIG labeling kit) (Besseau et al., 2006; Nisembaum et al., 2015; Paulin et al., 2015). The primers sequences for the preparation of the probes are given in Table 1 and Supplementary Figure 1.

The IHC was performed on pineal organs and retinas. Pineal and retinal sections were dipped in a 3% H₂O₂ solution (in PBS) for 10 min in the dark and rinsed in PBS. The first antibody, from a commercial origin (Abcam, Cambridge), was a polyclonal rabbit anti-zebrafish antibody directed against either TRPV1 (1/100), or TRPV4 (1/500); it was applied for 16 h at + 4°C. Revelation was then performed using a commercially available kit (IHC Select^® HRP/DAB) that contained the second antibody (anti-rabbit coupled to horseradish peroxidase [HRP]) and the substrates [diaminonobenzidine (DAB) and H₂O₂], and following the protocol instructions. The first antibody was omitted in the control sections as indicated in the kit instructions.

Pineal glands of adult fish (600 ± 50 g b.w.) were cultured in 24-wells culture plates (Nunclon™ Surface; VWR International, Fontenay-sous-Bois, France) as detailed elsewhere (Bégay et al., 1992; Nisembaum et al., 2015). Each well contained one gland in 500 μL of medium (RPMI 1640 without phenol red) complemented with penicillin (100 U/ml), streptomycin (100 μg/ml), glutamine (2 mM), and fungizone (2.5 μg/ml). The culture plates were placed in a MIR-154 incubator (Sanyo; Osaka, Japan) under the photoperiod and temperature conditions the fish had been acclimated to. The media were renewed every 24 h. After 48 h, the pineal glands (n = 7–8/group) were placed (at 12:00) for 6 h in the dark, in the presence of vehicle or increasing concentrations of either capsaicin (TRPV1 agonists) or 4α-Phorbol 12,13-didecanoate (4αPDD; TRPV4 agonist); these experiments were performed in the absence or presence of 1 μM of the respective antagonists (1 μM capsazepine for TRPV1, 10 μM ruthenium red for TRPV4). More details are given in the results section and legend of the figures. Ten millimolar stock solutions were prepared in the appropriate vehicle [100% methanol (capsazepine), DMSO (4αPDD), ethanol (capsaicin) or ultrapure water (ruthenium red)]. The final solvent concentration did not exceed 0.2% and controls contained an equivalent amount of vehicle. At the end of the 6 h, the culture media were collected and frozen at −20°C until melatonin quantification. All experiments were run in duplicate, the data are presented as the mean ± SEM.

The concentration of melatonin released in the culture medium was determined by High Performance Liquid Chromatography (HPLC) using either (1) a 100 × 4.6 mm C8 reversed-phase analytic column (Waters Spherisorb; Milford, MA) with particles size of 3 μm and an Agilent fluorescence detector (1,100 series; Santa Clara, CA, United States) or (2) a 125 × 4.6 mm C18(2) reversed phase analytic column (Luna, Phenomenex; Le Pecq, France), with particles size of 5 μm and a Dionex™ ULTIMATE™ 3100 fluorescence detector (Thermo Scientifique™; Villebon-sur-Yvette, France). Two to ten microliter of each sample were directly injected in the HPLC system. The excitation and emission wavelengths were 280 and 340 nm, respectively; the column temperature was maintained at 30°C; the mobile phase consisted of 0.1 M Na₂HPO₄ containing 10% (protocol 1) or 20% (protocol 2) acetonitrile; pH was adjusted to 6.5 with orthophosphoric acid. The mobile phase flow was 1 ml/min (protocol 1) or 1.5 ml/min (protocol 2), and the retention times of melatonin standards and samples were of ∼31 min (protocol 1) and ∼7 min (protocol 2). Standard curves were prepared after diluting a stock melatonin solution (10 mM; prepared in 100% methanol) in HPLC- grade water.

The analysis included one- or two-way ANOVA followed by the Holm-Sidak or Sidak post hoc tests depending on the dataset. Individual means were compared using the Two-tailed Student’s t-test. Drawings and statistics were performed using the Prism.v6 (GraphPad™ Software Inc., San Diego, CA).

3−aminopropyltriethoxysilane, 4-α-Phorbol 12,13-didecanoate (4αPDD), capsaicin, capsazepine, EDTA, eugenol, fungizone (Amphotericin B), L-glutamine-penicillin, paraplast^®, 3−aminopropyltriethoxysilane, streptomycin solution, ruthenium red, and RPMI culture medium were from Sigma-Aldrich (Saint-Quentin Fallavier, France). Melatonin standard was from Acros Organics™ (Fisher Scientifics, Villebon- sur-Yvette, France). Acetonitrile and hydrogen peroxide solution-HPLC grade were from Fisher Scientifics (Villebon- sur-Yvette, France). RNA later was from Life technologies SAS (Saint Aubin, France). The DIG labeling kit was from Roche Diagnostics (Meylan, France). The polyclonal rabbit anti-zebrafish antibodies (anti-TRPV1, Ab68969; anti-TRPV4, Ab69094) were from Abcam (Cambridge, England). The IHC revelation kit (IHC Select^® HRP/DAB) was from Merck-Millipore (Molsheim, Alsace, France).

The amino acid sequences corresponding to S. salar TRPV1 and TRPV4, as well as their alignment with the corresponding TRPV sequences from other vertebrates, are given in Supplementary Figures 1–3. The sequences amplified by the primers used in this study (Table 1) are also highlighted (Supplementary Figure 1). The couples of primers, chosen for either the qPCR or the ISH, displayed no significant alignment.

TRPV1 and TRPV4 mRNA were ubiquitously distributed in S. salar brain and peripheral tissues (Figure 1 and Supplementary Figure 4). However, their relative abundance varied from one tissue to another (Supplementary Figure 4). TRPV1 mRNA abundance was particularly high in the kidney, spleen and blood cells at both months investigated, while TRPV4 mRNA was particularly abundant in the gills (February and July), adipose fin and heart (February only).

In some tissues, variations in abundance were found between the February and July samples (Figure 1). TRPV1: abundance was 2–10-fold higher in July compared to February in the diencephalon, pituitary, heart, intestine, kidney, gills, and blood cells. The opposite held true for the optic tectum, spleen and saccus vasculosus, while no variation was seen in the case of the pineal organ and the retina. TRPV4: significantly higher amounts were detected in February vs. July in the retina, saccus vasculosus, heart, skin (including lateral line), liver, spleen and adipose fin; conversely, amounts were higher in July in the diencephalon, cerebellum, pituitary, kidney and blood cells. Again, no change was detected in the pineal organ.

Immuno-detected TRPV1 and TRPV4 proteins of the Atlantic salmon were found in all cell layers of the retina (Figure 6), while no staining was seen in the controls (Figure 6A).

Together with the previous studies [trout O. mykiss TRPV1 and TRPV4 (Nisembaum et al., 2015); chum salmon Oncorhynchus keta TRPV4 (Lee et al., 2021); tilapia Oreochromis mossambicus TRPV4 (Watanabe et al., 2012); half-smooth tongue sole Cynoglossus semilaevis TRPV4 (Shang et al., 2020); sea bass Dicentrarchus labrax (Bossus et al., 2011); zebrafish Danio rerio (Gau et al., 2013)], this study provides evidence that TRPV1 and TRPV4 are widely distributed in nervous and non-nervous tissues of fish. The experiments conducted here in S. salar at the months of February and July suggest variations in mRNA abundance may exist from a month to another in some tissues, as indicated by studies in the rainbow trout (Nisembaum et al., 2015). These differences might result from seasonal variations related to environmental changes (photoperiod and/or temperature) and/or developmental differences (fish studied here were developing yearlings) and/or smoltification (the fish were smolts in February and had achieved smoltification in July). More experimentation is needed to elucidate this point. These observations and the possible existence of differences among species or experimental procedures, make difficult any comparison concerning the levels of abundance. For example, a previous in situ hybridization study in S. salar allowed detection of TRPV1 and TRPV4 transcripts only in the telencephalon and optic lobes (Boltana et al., 2018), which contrasts with the present data showing expression in all brain areas, including the pineal gland, retina, pituitary gland, saccus vasculosus, telencephalon, diencephalon, optic tectum and cerebellum.

The variety of tissues expressing the TRPV channels is most probably related to the fact that these channels are multimodal effectors sensitive to a large number of stimuli including temperature (Patapoutian et al., 2003; Dhaka et al., 2006; Cohen and Moiseenkova-Bell, 2014; Li et al., 2020), ionic balance (Bossus et al., 2011; Seale et al., 2012), pressure and stretching (Liedtke and Kim, 2005; Watanabe et al., 2012; Startek et al., 2019), pH, ligands and ions [Ca²⁺, Mg²⁺] (Vriens et al., 2009), H₂0₂, lipids and lipid derived metabolites (arachidonic acid, anandamide, N-arachidonoyldopamine, lipoxygenase) (Leonelli et al., 2011; Zheng, 2013; Raboune et al., 2014; Cordero-Morales and Vasquez, 2018). It is beyond the scope of this study to discuss in depth the presence and role of TRPV1 and TRPV4 channels in the different tissues of S. salar. As detailed below, the main focus was S. salar pineal and, because the pineal gland and the retina are two homologous organs, derived from the same diencephalic origin (O’Brien and Klein, 1986), we also ran some parallel experiments in the retina.

In the pineal epithelium both, TRPV1 and TRPV4 transcripts, were detected only in the photoreceptor cells. The photoreceptors were identified by (i) their position in the upper part of the pineal epithelium, bordering the lumen of the organ, in which their apical part protrudes (i.e., they are in direct contact with the CSF) and (ii) their shape and segmented organization. These results are comparable to those previously obtained in the pineal gland of O. mykiss (Nisembaum et al., 2015). In the Atlantic salmon, the IHC detection of the corresponding proteins suggested a localization in membranes rather than the cytosol, with the channels concentrating in the apical part of the photoreceptor cells, which bathes into the CSF. The position of the faint IHC labeling also seen within the epithelium suggests they might correspond to neuropil areas. These areas contain photoreceptor endings making synaptic contacts with the pineal second-order neurons. In this regard, it is interesting that an ISH study in the rat showed that TRPV1 was associated with the synaptic ribbons of the pinealocytes (Reuss et al., 2010).

In vitro studies have shown that in both, the rat and the trout pineal organs in culture, the production of melatonin was modulated by the TRPV1 agonist capsaicin, and the effects were antagonized by the TRPV1 antagonist capsazepine (Reuss et al., 2010; Nisembaum et al., 2015). Here we show that melatonin secretion is also modulated by TRPV1 analogs in the Atlantic salmon. The modulation of melatonin production by TRPV1 appears thus as an ancestral character. Of interest is the observation that capsaicin had similar dose-dependent bell-shaped effects in trout and salmon pineal glands; and, in both cases the optimal effect was obtained at the micromolar concentration of the agonist; but in S. salar the effects of the agonist were observed only in the presence of the antagonist. Another similarity between trout and salmon lies in their response to TRPV4 analogs. In both cases 4αPPD had no effect on melatonin secretion [except in the Atlantic salmon at the highest (0.1 mM) concentration used]; but, in the presence of the antagonist ruthenium red, 4αPPD induced a similar U-shaped dose-response curve in both rainbow trout and Atlantic salmon (statistically significant in the Atlantic salmon only). The complexity of the responses to the TRPV agonists and antagonists has been discussed in the rainbow trout study (Nisembaum et al., 2015). Altogether, we conclude that TRPV1 and TRPV4 contribute to modulate in vitro melatonin secretion in a similar manner in the rainbow trout and Atlantic salmon pineal organs.

The question raises to know what triggers TRPV opening and closure. We believe that temperature is one possible candidate because: (i) the in vitro pharmacological responses to the TRPV analogs in the trout and salmon pineal glands were quite similar, and it is known that at least capsaicin mimics the effect of elevated temperature in TRPV1 channel, (ii) in the trout the effects of the TRPV1 and TRPV4 antagonists depended on temperature (∼16°C for TRPV1 and ∼8°C for TRPV4) (Nisembaum et al., 2015), and (iii) the TRPV1 and TRPV4 sequences from both fish displayed high identity (97% for TRPV1 and 98% for TRPV4). This would agree with previous observation indicating that (i) the fish pineal gland and melatonin are involved in behavioral thermoregulation (Kavaliers and Ralph, 1980; Ekström and Meissl, 1997), as is also the case in lizards (other ectotherms; Ralph et al., 1979a,b; Skinner, 1991), and (ii) temperature modulates both the hormonal and nervous pineal outputs (Zachmann et al., 1992; Tabata and Meissl, 1993; Tabata et al., 1993; Thibault et al., 1993; Falcón et al., 2007). Altogether it is reasonable to believe that the TRPV1 and TRPV4 channels of the Atlantic salmon pineal gland are thermo-receptors. Other functions are, however, not excluded, in keeping with the observation that TRPV channels are multimodal channels as commented above. Indeed, the observation that most of the ISH detected TRPV channels were located in the apical part of the photoreceptors, bathing into the CSF, in the Atlantic salmon pineal gland would agree with pioneer studies suggesting the pineal organ of ectotherms is involved in the regulation of pressure or composition of the CSF (Kelly and Vandekamer, 1960). This is particularly relevant in the Atlantic salmon, a migratory species, in which the life cycle involves adaptation to waters of different salinities, and migration start is triggered by changes in both photoperiod and temperature. In line with this, we found TRPV1 and TRPV4 channels in the apex of saccus dorsalis cells, which plays a major role in the production and regulation of the CSF in trout (Jansen et al., 1976a,b). The saccus dorsalis is apparently an analog of the choroid plexus (absent in trout) and is involved in a number of functions including fluid secretion, catabolism and extrusion of organic substances (monoamines, GABA) into the ventricular system, and uptake of organic substances from the CSF. It is interesting that the pineal gland and the saccus dorsalis of ectotherms, both receive innervation from arginine vasotocin fibers originating from the preoptic area (Vandendungen et al., 1982; Ramallo et al., 2012). Arginine vasotocin is a regulator of water balance and osmotic homoeostasis.

The cellular localization of TRPV1 and TRPV4 in the retina of S. salar was performed using IHC only (preliminary investigations using ISH indicated both, ISH and IHC, provided similar results; data not shown). We felt interesting to investigate the localization of TRPV1 and TRPV4 in the salmon retina for several reasons: (i) the fish pineal organ and retina are two homologous organs, (ii) previous studies indicated that TRPV channels are present in the vertebrates’ retina (Gao et al., 2019; Toft-Bertelsen et al., 2019; Bouskila et al., 2020) but (iii) data on fish remain scarce and concern the retina of 3 species only, namely D. rerio (Zimov and Yazulla, 2004; Amato et al., 2012; Sanchez-Ramos et al., 2012), C. auratus (Zimov and Yazulla, 2004), and O. mykiss (Nisembaum et al., 2015). Major differences between the pineal organ and the retina lie in the facts that the former contains only cones and does not possess a complex network of interneurons between the photoreceptors and ganglion cells. Also, the pineal is an “open organ,” while the retina is a “closed organ,” in other words, at their apical parts, the inner and outer segments of the photoreceptors bath into the CSF in the pineal gland, while those of the retina are nested into the extensions of the retinal pigment epithelial cells; at their basal part, the pineal organ is opened to the blood circulation, being directly surrounded by vessels, while the retina bathes into the vitreous humor.

In a general manner, Atlantic salmon TRPV1- and TRPV4-like proteins distributed as described in other vertebrate species [fish: (Sanchez-Ramos et al., 2012; Nisembaum et al., 2015); mouse: (Ryskamp et al., 2011; Lakk et al., 2018); Monkeys and human: (Gau et al., 2013; Sappington et al., 2015)]. Most of the studies dealing with TRPV1 and TRPV4 in the vertebrates’ retina indicate a role in detecting variations in temperature, osmotic pressure, mechanical, volume and hydrostatic changes (linked to systemic changes in blood pressure, hydrostatic pressure from the CSF and intrinsic intraocular pressure), as well as in mediating the response to chemicals (lipids and endocannabinoids) (Alessandri-Haber et al., 2006; Ryskamp et al., 2011, 2014; Ye et al., 2012; Sappington et al., 2015; Toft-Bertelsen et al., 2019; Pang et al., 2021; Redmon et al., 2021). In the Atlantic salmon retina, TRPV1 and TRPV4 were expressed in some, but not all, photoreceptors, highlighting a heterogeneity among the photoreceptor cell types; this contrasts with the situation observed in the pineal gland of this same species, or the retina of the rainbow trout, in which all the ONL appeared to express TRPV1 and TRPV4. These differences observed from a study to another might be due to either species-specific requirements or to differences in the experimental protocols (e.g., time of day or season) and technical approaches (e.g., ICC vs. ISH), not mentioning that TRPV mRNA expression may change from eye to eye, as reported to occur for TRPV1 in mice (Sappington et al., 2015). Whether these channels contribute to controlling melatonin production by the retinal photoreceptors, as is the case for their pineal analogs, remains an open question. It is also possible that they contribute to modulate neural transmission to bipolar cells, as they have been localized associated to the photoreceptor synaptic ribbons in D. rerio and C. auratus (Zimov and Yazulla, 2004), and to mediate synaptic transmission in rod bipolar cells in mice (Shen et al., 2009). Another interesting possibility is that TRPV channels contribute to controlling the temperature driven shifts between rhodopsin and porphyropsin observed in various fish species (including salmonids) and which affects nocturnal spectral sensitivity (Cristy, 1976; Saszik and Bilotta, 1999; Flamarique, 2005).