To investigate whether teleost Avp and Oxt are produced locally in the seabream ovary, we initially determined the changes in the concentration of the nonapeptides both in the plasma and in the ovary of females showing different stages of ovarian follicle development ( Figure 1A ) using high performance liquid chromatography (HPLC). Plasma and ovarian levels of Avp progressively increased in females with ovaries at the vitellogenic stage and undergoing meiotic maturation and hydration ( Figure 1B ). However, in females with ovaries containing fully matured and hydrated follicle-enclosed oocytes and ovulated eggs, the ovarian Avp levels markedly dropped, while the plasma levels remained elevated ( Figure 1B ). By contrast, the plasma Oxt levels in females were approximately three times higher than those of Avp and showed a cyclical pattern ( Figure 1C ). Thus, circulating Oxt concentrations were higher during vitellogenesis and the fully mature and hydrated stages compared to other stages ( Figure 1C ). However, the intraovarian changes of Oxt during the reproductive cycle reflected those observed for Avp, with a precipitous drop during the maturation and hydration stage ( Figure 1C ).

To rule out the possibility that intraovarian Avp and Oxt detected in our analysis were the result of their accumulation from circulating products, we investigated whether these peptides can also be produced within the seabream ovary, as reported for the stinging catfish (38, 40). For this, we first assessed the expression of pro-avp and pro-oxt genes in ovarian follicles by reverse transcription-polymerase chain reaction (RT-PCR) and in situ hybridization (ISH) using gene-specific oligonucleotide primers and riboprobes, respectively. The RT-PCR experiments indicated that pro-avp and pro-oxt mRNAs were expressed in the pituitary (as a positive control) and the ovary, whereas low but detectable transcripts from both genes were amplified in isolated vitellogenic ovarian follicles ( Figure 2A ). The ISH performed with histological sections of ovaries at the primary growth and vitellogenic stages showed that primordial follicle cells adjacent to previtellogenic oocytes were strongly positive for pro-avp, whereas these cells were devoid of pro-oxt mRNAs ( Figure 2B ). In vitellogenic ovarian follicles, in which follicle cells are fully differentiated into cuboidal granulosa cells and enlarged and elongated theca cells, pro-avp transcripts were detected only in the granulosa cells, while both granulosa and theca cells appear to express pro-oxt mRNAs ( Figure 2B ). In the cytoplasm of vitellogenic oocytes, some weak pro-avp and pro-oxt staining was also detected ( Figure 2B ). For both transcripts, no extrafollicular expression (i.e., interstitial tissue) was noted in the ovarian stages examined, and control sections incubated with sense probes were negative ( Figure 2B ).

The above observations were further confirmed by immunofluorescence microscopy using anti-Avp and anti-Oxt rabbit antisera, which were previously validated by immunostaining of the seabream posterior pituitary gland ( Supplementary Figure S1 ). These experiments showed that primordial follicle cells associated to primary growth oocytes were negative for Avp or Oxt immunostaining, and confirmed that granulosa cells surrounding vitellogenic oocytes, as well as those surrounding oocytes undergoing meiotic maturation and hydration, synthesize Avp ( Figure 2C ). In contrast, Oxt immunoreaction was observed in theca and granulosa cells in both follicular stages, in agreement with the ISH data ( Figure 2C ). Preadsorption of the antisera with the corresponding Avp and Oxt peptides led to a complete absence of staining in all the follicular stages analyzed, except a cross-reaction of the Avp antiserum with the vitelline envelope in vitellogenic follicles ( Figure 2C ), indicating the specificity of the signals in the follicular cells.

To assess whether Avprs and Oxtrs are also expressed in ovarian follicles we carried out RT-PCR and ISH using specific primers and riboprobes for two seabream Avpr subtypes, Avpr1aa and Avpr2aa (previously termed Avt1a2r and Avt2r, respectively), and one previously identified Oxtrb (formerly Itr) (41). These receptors were selected as representative Avprs and Oxtrs with a potential for different ligand specificities and the activation of different signal transduction pathways. The results of the RT-PCR assays, using kidney total RNA as a positive control, showed that both avpr1aa and avpr2aa transcripts could be amplified from the ovary and isolated follicles, while oxtrb expression was also noted in the ovary but not detected in follicles ( Figure 3A ). The ISH data indicated that previtellogenic oocytes express avpr2aa and oxtrb, but not avpr1aa, whereas oxtrb mRNAs were also observed in the primordial follicle cells ( Figure 3B ). During vitellogenesis, follicles start to express avpr1aa, but only in the oocyte and not in the follicle cells, whereas avpr2aa expression in oocytes is maintained and activated in the differentiated theca and granulosa cells ( Figure 3B ). At this stage, the oxtrb transcripts in oocytes seemed to be slightly reduced and were no longer detected in the follicle cells ( Figure 3B ). For the three transcripts sense probes gave negative results ( Figure 3B ).

Altogether, the above data demonstrate the local production of Avp and Oxt nonapeptides in the seabream ovarian follicle, as well as the expression of their cognate receptors in the follicle cells and/or oocyte, which would be consistent with a role of a vasopressinergic and oxytocinergic regulatory system controlling aquaporin intracellular trafficking and oocyte hydration.

Since the different Avprs and the Oxtrs are considered to trigger distinct PKA and PKC signaling cascades, and it has been shown that these kinases can regulate seabream Aqp1aa, -1ab1 and -1ab2 trafficking (12, 36), we next investigated the signaling pathways activated by seabream Avpr1aa, Avpr2aa and Oxtrb. For this, human embryonic kidney cells 293T (HEK293T) were transiently transfected with avpr1aa, avpr2aa or oxtrb full-length cDNAs, isolated from the seabream ovary based on publicly available data. Cells were transfected at the same time with β-Galactosidase (β-Gal) together with a cAMP-responsive reporter gene plasmid (pCRE-luc), or a plasmid for activation of the phospholipase C (PLC)/PKC pathway (pSRE-luc), and exposed to increasing doses of the nonapeptides ranging from 10^-12 to 10^-5 M ( Figure 4 ). The receptor-induced increase in intracellular Ca²⁺ concentration ([Ca²⁺]_i), as a second messenger in the PLC signal transduction pathway, was determined using Fluo-4 AM, a cell-permeable, fluorescent Ca²⁺ indicator.

In cells expressing the Avpr1aa or Oxtrb receptors, incubation with Avp or Oxt, respectively, resulted in a concentration-dependent increase of pSRE-induced luciferase activity and of [Ca²⁺]_i, while it did not induce measurable increases in pCRE-mediated luciferase activity ( Figures 4A–C ). However, the Avpr1aa was less affective at activating PKC or increasing the [Ca²⁺]_i than the Oxtrb, since the half-maximal effective concentrations (EC₅₀) of Avp on its cognate receptor were ~10 times higher compared with those of Oxt on the Oxtrb ( Figures 4B, C ). In contrast, in Avpr2aa expressing cells Avp induced a potent dose-dependent response in pCRE-induced luciferase activity, while PKC was not activated and the [Ca²⁺]_i was not affected ( Figures 4A–C ). Receptor cross-activation experiments showed that both Avpr1aa and Avpr2aa could be activated by Avp and Oxt, although the EC₅₀ for Oxt on these receptors was ~10 times higher with respect to that of Avp ( Figures 4D, E ). On the contrary, the Oxtrb was more specific for Oxt, since only a slight increase of the [Ca²⁺]_i in response to Avp was noted in cells expressing this receptor ( Figure 4F ).

To search for effective and selective inhibitors of seabream Avprs and Oxtrb, which could be used in further physiological experiments, we tested the effect of different non-peptide antagonists selective for mammalian AVPR1A (SR49059) (42), AVPR2 (tolvaptan, TLV) (43) and OXTR (L-371,257) (44). In these experiments, cells were transfected with Avpr1aa, Avpr2aa or Oxtrb and reporter plasmids as above, and preincubated with increasing doses (from 10^-12 to 10^-5 M) of the different receptor antagonists, and then subsequently treated with the approximate EC₅₀ of Avp or Oxt on each cognate receptor (10^-8 M). The results showed that activation of the Avpr1aa was equally inhibited by SR49059 and TLV, which showed half maximal inhibitory concentration (IC₅₀) of ~10^-9 M, although the SR49059 was slightly more effective ( Figure 4G ). Both SR49059 and TLV also inhibited the Avpr2aa, although in this case TLV was ~10 times more potent than SR49059 at blocking the receptor ( Figure 4H ). In contrast, the L-371,257 antagonist had no effect on Avpr1aa or Avpr2aa activation ( Figure 4G, H ). Finally, the Oxtrb was well blocked by L-371,257, showing an IC₅₀ of ~10^-9 M, while it was not affected by SR49059 or TLV ( Figure 4I ). These data therefore indicate that TLV and L-371,257 are, respectively, potent and selective inhibitors of seabream Avpr and Oxtr receptor subtypes.

To investigate whether Avprs and Oxtrb can regulate the intracellular trafficking of Aqp1aa, -1ab1 or -1ab2, we initially employed oocytes from X. laevis as a surrogate experimental system ( Figure 5 ). Oocytes were injected with avpr1aa, avpr2aa or oxtrb cRNAs, synthesized from human influenza hemagglutinin (HA)-tagged cDNAs, which drives the constitutive expression of the receptors in the oocyte plasma membrane ( Supplementary Figure S2 ). To test the effect of receptor activation on the trafficking of the TSA1C channels, each receptor was coinjected with cRNAs encoding Aqp1aa, Aqp1ab1 or Aqp1ab2, or water as a negative control. Oocytes were exposed to Avp or Oxt, depending on the type of receptor expressed, or to the hormone vehicle dimethyl sulfoxide (DMSO) as control, and subsequently submitted to swelling assays to determine the osmotic water permeability (P _f). Since Aqp1ab2 can only traffic to the frog oocyte plasma membrane when it is bound to the teleost-specific YwhazLb protein in the presence of the intracellular cAMP activator forskolin (FSK) (12), Aqp1ab2-injected oocytes were also coinjected with seabream YwhazLb cRNA and, in some experiments, exposed to FSK prior to nonapeptide treatment.

Oocytes expressing the Avpr1aa and Aqp1aa or Aqp1ab1 showed a 3-4-fold increase in P _f with respect to the control oocytes after a hypoosmotic challenge. However, Avp treatment reduced the P _f by 80% in oocytes expressing Aqp1aa, but not in those injected with Aqp1ab1 ( Figure 5A ). In the case of oocytes expressing Avpr1aa plus Aqp1ab2, treatment with Avp did not affect the oocyte P _f triggered by FSK exposure ( Figure 5A ). Preincubation of Avpr1aa+Aqp1aa oocytes with the PKC inhibitor bisindolylmaleimide II (Bim-II) prevented the Avp-induced inhibition of P _f ( Figure 5B ), in agreement with the activation of the PLC/PKC signaling pathway by the Avpr1aa previously observed in HK293T cells. Immunoblotting of total and plasma membrane extracts of oocytes and immunofluorescence microscopy using Aqp1aa-specific antibodies revealed that the hormone-induced reduction in the oocyte P _f was due to the internalization of the channel in response to Avpr1aa activation, which was blocked in the presence of Bim-II ( Figures 5C, D ). The same results were obtained in oocytes expressing the Oxtrb and Aqp1aa, Aqp1ab1 or Aqp1ab2 plus YwhazLb ( Figures 5I–L ), indicating that activation of Avpr1aa or Oxtrb by their cognate ligands inhibit Aqp1aa trafficking to the oocyte surface through PKC signaling, while Aqp1ab1 and -1ab2 are not regulated by these receptors.

In contrast, the P _f of oocytes expressing Avpr2aa and Aqp1ab1 or Aqp1ab2+YwhazLb (in the latter case without FSK pretreatment) was increased by ~2- and ~4-fold, respectively, after Avp exposure, whereas that of the oocytes co-expressing Avpr2aa and Aqp1aa did not change ( Figure 5E ). The Avpr2aa-mediated increase in the P _f of oocytes expressing Aqp1ab1 or Aqp1ab2+YwhazLb occurred through the Avp-induced insertion of the channels into the oocyte plasma membrane as revealed by immunoblotting and immunolocalization experiments using paralog-specific antibodies ( Figures 5F–H ). However, this trafficking was prevented by the PKA inhibitor H89 ( Figures 5F–H ). These data thus show that Avpr2aa can positively regulate Aqp1ab1 and Aqp1ab2 trafficking to the plasma membrane through the PKA signaling cascade, which is activated in both oocytes and cultured cells.

The above data are consistent with our recent study using direct pharmacological activation of PKC and PKA in X. laevis oocytes, which showed that in euacanthomorph teleosts, such as the seabream, PKC negatively regulates Aqp1aa trafficking to the plasma membrane, while PKA positively regulates this mechanism in Aqp1ab1 and Aqp1ab2 (12). This latter study also identified the kinase phosphorylation sites involved in trafficking regulation of the different channels: Thr¹⁴⁵ (loop D) in Aqp1aa, Ser¹⁵⁶ and Ser²⁵³ (intracellular loop D and C-terminus, respectively) in Aqp1ab1, and Thr²⁶² (C-terminus) in Aqp1ab2 ( Figures 6A, D, I ). However, plasma membrane trafficking of Aqp1ab1 partially depends on channel binding to the teleost-specific YwhazLa, through PKA phosphorylation of Ser²⁵³ within the YWHA binding motif, while that of Aqp1ab2, as mentioned above, strictly relies on PKA-mediated Thr²⁶² phosphorylation and the YwhazLb interaction (12) ( Figures 6A, D, I ). Therefore, in order to investigate whether Avp and Oxt control Aqp1aa, Aqp1ab1 and Aqp1ab2 trafficking through the same mechanisms, X. laevis oocytes were injected with cRNAs for avpr1aa, avpr2aa or oxtrb and wild-type TSA1C channels or mutant forms in which the PKA and PKC phosphorylation sites were replaced by non-phosphorylatable Ala residues (12), together with HA-tagged YwhazLa or -zLb. The P _f of oocytes, the trafficking of the channels to the oocyte plasma membrane, and the interaction of Aqp1ab1 and Aqp1ab2 with YwhazLa or -zLb carrier proteins, respectively, in response to the nonapeptides, were then examined by using swelling assays, immunoblotting and co-immunoprecipitation using Aqp1ab1- and Aqp1ab2-specific antibodies.

The results of these experiments showed that the reduction of oocyte permeability and channel accumulation in the plasma membrane in response to Avp or Oxt, in oocytes expressing wild-type Aqp1aa plus Avpr1aa or Oxtrb, was completely reversed when oocytes expressed the Aqp1aa-V145T mutant instead of the wild-type ( Figures 6B, C ), as observed when PKC is directly activated in Aqp1aa expressing oocytes (12). In oocytes expressing wild-type Aqp1ab1, YwhazLa-HA and Avpr2aa, Avp triggered a more robust increment in the oocyte P _f and of the relative amount of the channel in the plasma membrane ( Figures 6E, F ) than in oocytes without YwhazLa ( Figure 5E ), which was partially or completely prevented, respectively, in oocytes expressing the Aqp1ab1-S156A or -S253A mutants ( Figures 6E, F ). However, the positive effect of Avp on plasma membrane trafficking of both wild-type Aqp1ab1 and Aqp1ab1-S156A channels was equally abolished in the presence of the H89 inhibitor ( Figures 6E, G ). Co-immunoprecipitation experiments using the Aqp1ab1 antibody confirmed that Avp promoted the interaction of wild-type Aqp1ab1 and Aqp1ab1-S156A with YwhazLa, which was inhibited in both cases in the presence of H89 ( Figure 6H ). However, the Aqp1ab1-S253A mutant, which did not traffic to the oocyte surface after Avp stimulation, was not able to interact with YwhazLa regardless of H89 treatment ( Figure 6H ). These data agree with our previous study (12) and suggest that Avp-mediated targeting of Aqp1ab1 to the plasma membrane depends on PKA phosphorylation of C-terminal Ser²⁵³, which drives YwhazLa binding, and of loop D Ser¹⁵⁶ once the complex Aqp1ab1-YwhazLa is formed. In contrast, oocytes expressing Aqp1ab2 and Avpr2aa, together with YwhazLb-HA, only showed a higher P _f, together with an increased channel expression in the oocyte plasma membrane and channel binding to YwhazLb, after Avp treatment, which were all blocked in oocytes expressing the Aqp1ab2-T262A mutant ( Figures 6J–L ). These findings therefore indicate that the Avp-mediated regulation of Aqp1ab2 trafficking is specifically mediated by PKA phosphorylation of Thr²⁶² in the C-terminus, allowing YwhazLb binding and further insertion of the channel into the plasma membrane.

The previous experiments show that seabream ovarian follicles can synthesize Avp, and that the ovarian levels of the nonapeptide in vivo increase during the stage of oocyte maturation and hydration. These observations would be compatible with a role of Avp regulating the intracellular trafficking of Aqp1ab1 and Aqp1ab2 in maturing oocytes, as observed in surrogate amphibian oocytes. However, our data also suggest that the local production of Avp may be under gonadotropic regulation, since a surge of the circulating luteinizing hormone (Lh) typically occurs in seabream females during the maturation stage (45, 46). To test this hypothesis, we carried out in vitro experiments in which the production of Avp by postvitellogenic ovarian explants in response to piscine single-chain recombinant Lh (rLh) was evaluated by a custom made Avp enzyme-linked immunosorbent assay (ELISA) ( Supplementary Figure S3 ). To investigate the specificity of the rLh potential regulation, in these experiments we also tested the effect of single-chain recombinant follicle-stimulating hormone (rFsh). The results showed that rLh increased the concentration of Avp in the explants by ~4-fold with respect to the water-exposed explants (controls), while the Avp levels in the rFsh-treated explants did not differ from the controls ( Figure 7A ).

Since the progestin 17α,20β-dihydroxypregn-4-en-3-one (17,20β-P) is known to be the maturation-inducing steroid secreted by granulosa cells in response to Lh in sparids (47), we further investigated whether 17,20β-P could mediate the Lh induction of Avp synthesis in postvitellogenic ovarian explants. In addition, to examine the type of progestin receptor potentially involved in this mechanism, treatment with 17,20β-P was applied in the presence or absence of the nuclear progesterone receptor inhibitor RU486 (20), while other explants were treated with the membrane progestin receptor agonist OD 02-0 (10- ethenyl-19-norprogesterone) (48) in the absence of 17,20β-P. The results of these trials confirmed that 17,20β-P induced a 3-fold increase in Avp synthesis in ovarian explants, comparable to that observed in response to rLh, which could be completely abolished by RU486, and not reproduced by OD 02-0 treatment ( Figure 7B ). These data therefore suggest that the specific Lh induction of Avp synthesis in the granulosa cells of seabream postvitellogenic follicles is mediated by a genomic mechanism through the nuclear progestin receptor.

To investigate the physiological significance of the Avp regulation of the TSA1C channel trafficking in seabream oocytes, we determined the dynamics of the subcellular localization of endogenous Aqp1aa, Aqp1ab1, and Aqp1ab2, and their interactions with YwhazLa and -zLb proteins, after Avp treatment in vitro, using the Aqp1ab1- and Aqp1ab2-specific antisera and commercial mammalian YWHAZ antibodies specific for seabream YwhazLa/b or -zLb (12). For this, we incubated isolated seabream vitellogenic ovarian follicles with Avp, in the presence or absence of the seabream Avpr1aa and Avpr2aa antagonist TLV or the PKA inhibitor H89, while control follicles were exposed to DMSO vehicle alone ( Figure 8 ).

The experiments showed that Avp did not induce meiotic maturation or oocyte hydration, in the latter case because the osmotic drive generated by yolk hydrolysis was not activated by the nonapeptide. Immunofluorescence microscopy indicated that Aqp1aa was localized to the epithelium surrounding the ovarian follicles, and not in the theca cell as previously reported (12), and that this localization did not appear to change with Avp treatment ( Figures 8A, B ). In contrast, both Aqp1ab1 and Aqp1ab2 were more accumulated in the most cortical cytoplasmic region of the oocytes, just below the plasma membrane, while Aqp1ab2 was also localized in the granulosa and theca cells ( Figures 8A, B ). After Avp exposure, Aqp1ab1 immunoreactivity appeared more concentrated in the most distal part of the microvilli crossing the vitelline envelope, whereas the Aqp1ab2 channel became localized in a more basal region of the microvilli ( Figures 8A, B ). The differential spatial localization of Aqp1ab1 and Aqp1ab2 in the oocyte microvilli in response to Avp mimics that observed in FSK-treated postvitellogenic follicles (12). However, it was prevented in follicle-enclosed oocytes preincubated with TLV or H89 prior to Avp treatment ( Figures 8C, D ). These observations were corroborated by Aqp1ab1 and Aqp1ab2 fluorescence quantification in the oocyte plasma membrane ( Figure 8E ), as well as by radiolabeled water uptake assays, which showed that water permeability of Avp-treated follicles was higher than that of follicles treated with DMSO alone and reduced in the presence of Avp plus the TLV or H89 inhibitors ( Figure 8F ).

The potential changes in the phosphorylation state of Aqp1ab1 and Aqp1ab2 channels, and of their interactions with YwhazLa and -zLb carrier proteins, during Avp treatment were subsequently investigated by co-immunoprecipitation and immunoblotting. These trials confirmed previous observations (12, 16) indicating that in control postvitellogenic follicles, when Aqp1ab1 is held in the most cortical region of the oocyte and some inserted in the basal region of the microvilli, the channel is already phosphorylated and bound to both YwhazLa and -zLb ( Figure 8G ). However, after Avp treatment Aqp1ab1 becomes more phosphorylated, while maintaining the interaction with YwhazLa and -zLb ( Figure 8G ), which coincides with the secondary loop D Ser¹⁵⁶ phosphorylation and trafficking of the channel to the most distal region of the oocyte microvilli. Phosphorylation of Aqp1ab2 was also increased after Avp stimulation, but in this case the YwhazLb interaction was only detected in the presence of the nonapeptide ( Figure 8G ), when the channel is translocated to the basal region of the oocyte microvilli. The Avp-mediated increase in the phosphorylation of Aqp1ab1 and Aqp1ab2, as well as the binding of the latter channel to YwhazLb, was blocked by TLV and H89 ( Figure 8G ), suggesting that the differential membrane trafficking regulation of the channels is regulated by the Avpr2aa-PKA transduction pathway.

Adult gilthead seabream were obtained from the Institut de Recerca i Tecnologia Agroalimentàries (IRTA) aquaculture facilities (Tarragona, Spain) and transported to the facilities in Institut de Ciències del Mar (Barcelona, Spain). Fish were maintained under natural conditions of photoperiod and temperature and fed three times a week with dry pellets (Seabream Alterna AE, Skretting) and squid ad libitum. Samples of blood and gonads were collected throughout the year from sacrificed females as previously described (20), frozen in liquid nitrogen and stored at -80°C. The stage of ovarian development in each female was determined by standard histological analysis as described previously (59).

Adult X. laevis were purchased from the Centre de Ressources Biologiques Xénopes (University of Rennes, France) and maintained at the AQUAB facilities of the Universitat Autònoma de Barcelona (UAB, Spain). Frogs were kept in tanks with filtrated freshwater at 18°C, under a 12-h light-dark cycle, and fed two days a week with beef heart or pellets (Xenopus Sticks, AQUA Schwarz GmbH, Göttingen, Germany). Oocytes were collected by surgical laparotomy from anesthetized females.

The nomenclature used for the seabream vasopressin and oxytocin nonapeptides and corresponding cognate receptors is adopted according to the Human Genome Nomenclature Committee (HGNC) (www.genenames.org), the zebrafish information network (ZFIN) (www.zfin.org) and Ensembl (v109) (www.ensembl.org), according to that previously described (28). The specific nomenclature of the gilthead seabream Avprs and Oxtrs was established through Bayesian inference of two million MCMC generations (aamodel = mixed; burnin = 25%) of a ClustalX amino acid alignment of the seabream Avprs and Oxtrs in relation to 88 actinopterygian orthologs downloaded from Ensembl (v109) ( Supplementary Figure S5 ). The resultant tree topology matches that computed by maximum likelihood by Daza et al. (27) with the exception that the zebrafish Avpr2ba and Avpr2bb receptors are named in accordance with the Ensembl (v109) nomenclature.

Affinity-purified rabbit or chicken polyclonal antibodies specific for seabream Aqp1aa, -1ab1 and -1ab2 have been described elsewhere (12, 60). Antisera against Avp and Oxt were generously provided by Prof. Olivier Kah (INSERM-Université de Rennes 1, France). The HA tag rabbit antibody was from Thermo Fisher Scientific (Invitrogen, #PA1-985), whereas anti-YWHAZ antibodies were purchased from GeneTex (#GTX101075) and Signalway Antibody LLC (#SAB40525), which specificity for seabream YwhazLa and/or YwhazLb has been validated elsewhere (12). Synthetic Avp (CYIQNCPRG-NH₂) and Oxt (CYISNCPIG-NH₂) were obtained from PeptideSynthetics (Peptide Protein Research Ltd.). All other reagents were purchased from Merck unless indicated otherwise.

Ovarian and plasma levels of Avp and Oxt in vivo were determined using HPLC with fluorescence detection preceded by solid-phase extraction (SPE). Pieces of ovary (150-350 mg) previously frozen and stored at -80°C were sonicated in 1 ml of Milli-Q water, acidified with 3 μl of glacial acetic acid and incubated in a boiling water bath during 3.5 min. The samples were centrifuged (30 min, 4°C, 10,000 × g). Plasma samples (1.5-2 ml each) were acidified with 1 M HCl (100 μl) and centrifuged at 6000·× g for 20 min at 4 °C. In both cases, the supernatants were loaded onto previously equilibrated SPE columns (Strata-X, 30 mg/ml; Phenomenex, Torrance, USA) and eluted in 80% acetonitrile. Eluates were evaporated to dryness using TurboVap LVTM (Caliper Life Sciences, PerkinElmer, Inc.), and frozen and stored at -20°C until HPLC analysis. Before quantitative analysis, the dried samples were reconstituted with 0.1 ml of 4% acetic acid and divided into two aliquots for duplicate analysis. For plasma, samples were redissolved in 0.1 ml of acetonitrile:H₂O (2:1). For pre-column derivatization, 50 μl of sample and 50 μl of 0.05 M borate buffer (pH 8) were mixed, and then 3 μl of NBD-F (4-fluoro-7-nitro-2,1,3-benzoxadiazole; 30 mg/ml of acetonitrile) was added. The solution was heated at 40˚C for 10 min, cooled on ice, and acidified with 5 μl of 1 M HCl. Chromatographic analysis was achieved using Agilent 1200 Series Quaternary HPLC System (Agilent Technology, USA). Peptide separations were performed on a Kinetex C18 column (4.6 x 150 mm, 5 μm, Phenomenex, USA). A gradient elution system was applied for separation of derivatized peptides. The mobile phase consisted of solvent A (0.1% TFA in H₂O) and solvent B (0.1% TFA in 3:1 acetonitrile:H₂O). A linear gradient was 40-65% of eluent B in 20 min. Flow rate was set at 1 ml/min and the column temperature was set to 20˚C. Fluorescence detection was carried out at 530 nm with excitation at 470 nm.

Full-length cDNAs encoding seabream Aqp1aa, Aqp1ab1, Aqp1ab2, YwhazLa and -zLb have been previously isolated (GenBank accession numbers AY626939, AY626938, MW960021, OK572451 and OK572452, respectively). Mutated Aqp1aa, Aqp1ab1 and Aqp1ab2 constructs in PKA and PKC phosphorylation sites were previously produced (12). Full-length cDNAs encoding seabream Avpr1aa, Avpr2aa and Oxtrb were isolated by RT-PCR using intestine or rectum total RNA as previously described (18) and oligonucleotide primers designed based on publicly available sequences (GenBank accession numbers KC195974, KC960488 and KC195973, respectively; Supplementary Table S1 ). The cDNAs were subcloned into the pT7Ts (61) or pcDNA3 (Invitrogen) expression vectors, and fused with an HA epitope tag (YPYDVPDYA) in the C-terminus before the stop codon by using PCR. The final clones were Sanger sequenced to confirm that the constructs were correct.

Seabream Avp and Oxt receptor cDNA constructs in pcDNA3 were transiently expressed in HEK293T cells (ATCC # CRL-11268). The cells were grown in 24 wells plates in DMEM supplemented with penicillin/streptomycin, 2 mM glutamine, and 10% fetal bovine serum (FBS) (Life Technologies) and were incubated at 37°C in 5% CO₂. Receptor activation was measured using luciferase reporter gene assays and intracellular Ca²⁺ levels. At approximately 50-60% confluence, cells were transfected using Lipofectamine 3000 (Invitrogen) with empty pcDNA3 or Avpr1aa (5 ng), Avpr2aa (5 ng), or Oxtrb (50 ng) cDNAs, 50 ng of pCRE-luc (Agilent Technologies) or 200 ng of pSRE-luc (BD Biosciences Clontech), and 5 ng or 50 ng of β-Gal plasmid (Promega Corp.) to normalize the transfection efficiency. The transfected cells were grown for 36 h at 37°C in an air/CO₂ [95:5 (v/v)] atmosphere before starvation for 16 h by replacing the DMEM by serum-free DMEM medium. The cells were then incubated in triplicate with different concentrations of Avp or Oxt (from 10^-5 M to 10^-12 M), or sterile dimethyl sulfoxide (DMSO) vehicle (0.01% final concentration), for 3 h. In some experiments, the cells were incubated with increasing concentrations (from 10^-12 to 10^-5 M) of different AVP (SR49059, Merck #S5701; TLV, Merck #T7455) and OXT (L-371,257, Tocris #2410) receptor antagonists for 1 h prior to stimulation with 10^-8 M Avp or Oxt. After incubation, the cells were harvested in lysis buffer (25 mM Tris-phosphate, pH 7.8, 2 mM DTT, 10% glycerol, 1% triton X-100), frozen at -80°C for 16-24 h and centrifuged at 14,000 × g for 1 min. To determine the luciferase activity, 20 µl of the supernatant was mixed with 100 µl of reconstituted Luciferin (BioThema) in a 96-well plate (Nunc F96 MicroWell Black and White Polystyrene Plate, Thermo Fisher Scientific), and the luminescence measured in an Orion II microplate luminometer (Titertek-Berthold). Luciferase activity was normalized to β-Gal activity measured by colorimetric detection using an Infinite M200 microplate reader (Tecan Group Ltd.). The data was finally expressed as fold-change of luciferase activity with respect to control cells exposed to DMSO only.

To determine the intracellular Ca²⁺ levels after receptor activation we used the cell permeant Fluo-4-AM vital dye (Life technologies #F14201). For this, HEK293T cells were transfected as above with 2.5 µg of Avpr1aa, Avpr2aa or Oxtrb cDNAs in 6-well plates (Thermo Scientific). After 16 h, the cells were transferred to a black 96-well microplate and incubated 24 h at 37°C in an air/CO₂ [95:5 (v/v)] atmosphere. Fluo-4 AM was diluted in Hank’s Balanced Salt Solution (HBSS) supplemented with glucose, containing 8% NP-40 and 20% DMSO, and added to the cells at 5 µM final concentration. Cells were incubated for 1 h at 37°C in an air/CO₂ [95:5 (v/v)] atmosphere, and subsequently exposed in triplicate to the nonapeptides, in the presence of absence of inhibitors, as described above. After washing twice with fresh DMEM, fluorescence intensity was measured at excitation and emission wavelengths of 494 and 506 nm, respectively, using an Infinite M200 microplate reader (Tecan Group Ltd.). The background signal of cells incubated with HBSS without Fluo-4 AM was subtracted from each value.

For RT-PCR, total RNA was extracted from different adult tissues, including pituitary, kidney, ovary and isolated ovarian follicles, using the RNeasy Minikit (Qiagen) and DNAse I treatment, following the manufacturer’s instructions. Total RNA (5 μg) was reverse transcribed using 0.5 μg oligo(dT)₁₇ primer, 1 mM deoxynucleotide triphosphates (dNTPs), 40 IU RNAse out (Life Technologies Corp.), and 10 IU SuperScript II Reverse Transcriptase enzyme (Life Technologies Corp.) for 1.5 h at 42°C. The PCR was carried out with 1 μl of the RT reaction in a final volume of 50 μl containing PCR buffer with Mg²⁺, 0.2 mM dNTPs, 1 IU of Taq polymerase (Roche), and 0.2 μM of forward and reverse oligonucleotide primers specific for seabream pro-avp, pro-oxt, avpr1aa, avpr2aa and oxtrb ( Supplementary Table S1 ). Reactions were amplified using one cycle at 95°C for 2 min; 35 cycles at 95°C for 30 sec, 60-62°C (depending on primer Tm) for 30 sec, and 1 min at 72°C; and 7 min for final elongation at 72°C. The 18S ribosomal RNA (rps18) was used as reference gene. The PCR products were run on 1% agarose gels and photographed.

Cellular localization of pro-avp, pro-oxt, avpr1aa, avpr2aa and oxtrb expression in ovarian follicles was carried out by ISH using digoxigenin-labelled gene-specific riboprobes as described previously (62). The sense and antisense riboprobes were synthesized using T7- or SP6-RNA polymerases from a PCR-amplified template using specific oligonucleotide primers ( Supplementary Table S1 ). The pro-avp and pro-oxt probes corresponded to the C-terminal end and part of the 3´UTR, or the proximal 3´UTR, respectively. For the receptors, the probes spanned ~300 bp from the middle of the avpr2aa cDNA sequence, and ~200 bp of the avpr1aa and oxtrb cDNAs including the C-terminus and a portion of the 3´UTR.

The cRNAs corresponding to the different constructs were synthesized in vitro with T7 RNA polymerase from XbaI-digested pT7Ts vector containing the cDNAs. The isolation and microinjection of oocytes, and subsequent swelling assays, were carried out as previously described (63). Oocytes were injected with 50 nl of water alone (controls) or containing 15 ng of Avpr1aa, Avpr2aa or Oxtrb cRNAs. Oocytes expressing the receptors were also injected with wild-type or mutant Aqp1aa (0.1 ng), Aqp1ab1 (0.5 ng) or Aqp1ab2 (25 ng) plus YwhazLb (25 ng). In some experiments, oocytes expressing Avpr2aa and Aqp1ab1 constructs were also injected with YwhazLa (25 ng). The oocyte P _f was determined after 3 h incubation with 10 µM Avp or Oxt (depending on the type of receptor expressed in oocytes). The role of PKA or PKC on the oocyte P _f was tested by preincubating the oocytes with 10 µM of H89 or Bim-II for 1 h prior to the addition of the hormones. Oocytes expressing Avpr1aa or Oxtrb together with Aqp1ab2 plus YwhazLb were treated with 10 µM FSK as previously described (12). Control oocytes were treated with 0.1% DMSO.

Pieces of the ovary at the vitellogenic stage (400-500 mg) were dissected from naturally spawning seabream females and placed in Petri dishes containing 75% Leivovitz L-15 culture medium with L-glutamine and 100 µg/ml gentamicin at pH 7.5. The explants were treated with 100 ng/ml of sea bass (Dicentrarchus labrax) rFsh or rLh (64), which activate the seabream Fsh and Lh receptors, respectively (59), 0.1 µg/ml 17,20βP, in the presence or absence of 100 µM of RU486 (Merck M8046), or with 10 µM of OD 02-0 (Axon Medchem, 13258-85-0), for ~20 h at 18°C in a temperature-controlled incubator. Control groups were treated with an equivalent volume of the cell culture medium used for rLh and rFsh production, or with 0.5% ethanol. After incubation, explants were frozen and stored at -80°C until Avp extraction and determination by ELISA.

In other experiments, groups of fully-grown postvitellogenic ovarian follicles (n = 10), manually isolated from the ovary, were incubated with 10 µM Avp in the same culture medium as above, in the presence or absence of 10 µM of the receptor antagonist TLV, or with 0.1% DMSO (controls) for 6 h at 18°C. After incubation, follicles were fixed for immunofluorescence microscopy, frozen in liquid nitrogen and stored at -80°C for further protein extraction, or processed for water uptake assays using radiolabeled water as previously described (12).

To determine the Avp levels in ovarian explants incubated in vitro, a competitive ELISA was developed using the specific anti-Avp antiserum and the synthetic Avp to create the standard curve ( Supplementary Figure S3 ). For this, polystyrene ELISA microtiter plates (F96 Maxisorp Nunc Immuno Plates; Nunc) were coated with 100 µl/well of 50 ng Avp/ml carbonate buffer (50 mM sodium carbonate, pH 9.6) and incubated overnight at 4°C. For the blank and the non-specific binding, two wells were either not coated or coated with 100 µl/well of 50 ng BSA/ml carbonate buffer, respectively. Ovarian explants were homogenized on ice with a glass dounce tissue homogenizer in water containing 0.25% glacial acetic acid in a proportion of 500 µl/100 mg tissue. After incubation of the homogenate at 100°C for 5 min, the samples were centrifuged a 10,000 × g for 5 min at 4°C, and the supernatant recovered in a new tube. The standards (0.01-500 ng Avp/ml) and experimental samples were diluted in PBST (PBS plus 0.05% Tween 20) with 0.5% normal goat serum (NGS), containing the Avp antibody diluted at 1:15,000, and were pre-incubated overnight at 4°C. Displacement curves for ovarian samples were obtained by serial dilutions from 1:2 to 1:16 to confirm the parallelism with the standard curves. For the measurement of the experimental samples a 1:4 dilution was applied, and to evaluate the cross-reactivity of the assay with Oxt, the same standard curve as for Avp was created (0.01-500 ng/ml). All standards and samples were processed in duplicate. After coating, the plates were washed 3 times with 200 µl/well of PBST, and blocked with 200 µl/well of PBST buffer containing 2% NGS for 1 h at room temperature. Then, 100 µl of samples and standards preincubated with the antibody were added to each well, and plates were incubated at 4°C for another 24 h. The plates were finally washed 3 times in PBST and each well was incubated with 100 µl/well of EIA grade affinity purified goat anti-rabbit IgG (H + L) HRP conjugated (Bio-rad Laboratories, Inc.) diluted 1:3000 in PBST plus 0.5% NGS for 2 h at room temperature. After washing, enzymatic color development was carried out by the addition of 100 µl/well of tetramethylbenzidine (TMB) peroxidase EIA substrate kit (Bio-rad Laboratories, Inc.) in darkness during 30 min, and the reaction was stopped with 100 µl/well of 1 N sulfuric acid. Absorbance was read at 450 nm with a VICTOR3™ Multilabel Plate Reader (PerkinElmer, Inc.). The amount of Avp was normalized to the wet weight of the ovarian explants. The intra-assay coefficient of variation was 4.64% (calculated by assaying eight replicates of an ovarian sample at 50% of maximum binding on the same plate), whereas the inter-assay coefficient of variation was 8.35%. The sensitivity of the assay for Avp was of 10 pg/ml, while its crossreaction with Oxt was calculated to be 8.43 ± 1.72%.

Isolated ovarian follicles were processed for immunofluorescence microscopy as previously described (65, 66). Control sections were incubated with antibodies (1:400 dilution) preadsorbed with the corresponding peptides used for immunization (for aquaporin antibodies), or with the Avp or Oxt synthetic peptides. Secondary antibodies were sheep anti-rabbit IgG Cy3 (Merck #C2306) or goat anti-chicken IgY Alexa Fluor 488 (Invitrogen #A11039) at a 1:1,000 dilution. Sections were counterstained with DAPI (1:10,000), and mounted with Fluoromount™ Aqueous Mounting Medium (Merck #F4680). Immunofluorescence was observed and documented with a Zeiss Axio Imager Z1/Apotome fluorescent microscope (Carl Zeiss Corp.). Changes in Aqp1ab1 and Aqp1ab2 in the oocyte plasma membrane after Avp treatment were evaluated using the ImageJ software v.1.52o (https://imagej.nih.gov/ij/) as previously described (12).

The total and plasma membrane fractions of X. laevis oocytes (n = 10), and total membrane fractions of seabream ovarian follicles (n = 50), were isolated as described previously (12) and resuspended in Laemmli sample buffer. For immunoblotting, Laemmli-mixed protein samples were denatured at 95°C for 10 min and subjected to 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and blotted onto Immun-Blot nitrocellulose 0.2 μm membranes (Bio-Rad Laboratories, Inc.), as described previously (67). Membranes were blocked with 5% nonfat dry milk in TBST (20 mM Tris, 140 mM NaCl, 0.1% Tween, pH 8) for 1 h at room temperature, and subsequently incubated overnight at 4°C with the seabream-specific Aqp1aa, Aqp1ab1 or Aqp1ab2 antisera, or HA antibodies, diluted (1:1,000) in TBST with 5% milk. Bound antibodies were detected with horseradish peroxidase-coupled anti-rabbit (Bio-Rad Laboratories, Inc., #172-1019) or anti-chicken (Thermo Fisher Scientific, #A16054) secondary antibodies diluted 1:5000 as above, and reactive protein bands were revealed using ImmobilonTM Western chemiluminescent HRP substrate (Millipore, #WBKLS).

X. laevis oocytes and seabream follicles (n = 10) were homogenized in the lysis buffer from the Pierce™ Crosslink Magnetic IP/Co-IP Kit (Pierce Thermo Fisher Scientific, #88805) with a bouncer on ice, and centrifuged at 14,000 × g for 1 min. An aliquot (10%) of the protein extract was collected as “input” and mixed with 2 × Laemmli sample buffer, whereas the remaining extract was incubated overnight at 4°C under constant agitation with magnetic beads previously coated with anti-Aqp1ab1 or anti-Aqp1ab2 antibodies. For rabbit primary antibodies, the immunoprecipitation was carried out using the kit indicated above, and final eluted proteins in 50 µl of elution buffer were mixed with 4 × Laemmli sample buffer supplemented with protease and phosphatase inhibitors. For chicken antibodies, the immunoprecipitation was performed using Dynabeads™ M-280 Tosylactivated (Invitrogen Thermo Fisher Scientific, #14203) following the manufacturer’s instructions. The input and immunoprecipitated samples were immunoblotted as indicated above.

Comparisons between two independent groups were made by the two-tailed unpaired Student’s t-test. The statistical significance among multiple groups was analyzed by one-way ANOVA, followed by the Tukey’s multiple comparison test, or by the non-parametric Kruskal-Wallis test and further Dunn’s test for nonparametric post hoc comparisons. Percentages were square root transformed previous analyses. Statistical analyses were carried out using the GraphPad Prism v9.1.2 (226) (GraphPad Software).