Extracellular signal-regulated kinase phosphorylation in forebrain neurones contributes to osmoregulatory mechanisms

Abstract

Vasopressin secretion from the magnocellular neurosecretory cells (MNCs) is crucial for body fluid homeostasis. Osmotic regulation of MNC activity involves the concerted modulation of intrinsic mechanosensitive ion channels, taurine release from local astrocytes as well as excitatory inputs derived from osmosensitive forebrain regions. Extracellular signal-regulated protein kinases (ERK) are mitogen-activated protein kinases that transduce extracellular stimuli into intracellular post-translational and transcriptional responses, leading to changes in intrinsic neuronal properties and synaptic function. Here, we investigated whether ERK activation (i.e. phosphorylation) plays a role in the functioning of forebrain osmoregulatory networks. We found that within 10 min after intraperitoneal injections of hypertonic saline (3 m, 6 m) in rats, many phosphoERK-immunopositive neurones were observed in osmosensitive forebrain regions, including the MNC containing supraoptic nuclei. The intensity of ERK labelling was dose-dependent. Reciprocally, slow intragastric infusions of water that lower osmolality reduced basal ERK phosphorylation. In the supraoptic nucleus, ERK phosphorylation predominated in vasopressin neurones vs. oxytocin neurones and was absent from astrocytes. Western blot experiments confirmed that phosphoERK expression in the supraoptic nucleus was dose dependent. Intracerebroventricular administration of the ERK phosphorylation inhibitor U 0126 before a hyperosmotic challenge reduced the number of both phosphoERK-immunopositive neurones and Fos expressing neurones in osmosensitive forebrain regions. Blockade of ERK phosphorylation also reduced hypertonically induced depolarization and an increase in firing of the supraoptic MNCs recorded in vitro. It finally reduced hypertonically induced vasopressin release in the bloodstream. Altogether, these findings identify ERK phosphorylation as a new element contributing to the osmoregulatory mechanisms of vasopressin release.

Introduction

Systemic osmoregulation is a vital process whereby fluctuations in plasma osmolality, detected by osmoreceptors, induce appropriate changes in salt and water intake and excretion (Bourque, 2008), to avoid irreversible organ damage and lethal neurological lesions. The osmotic control of vasopressin (VP) secretion plays a key role in body fluid homeostasis because this peptide stimulates water reabsorption by the kidney (Dunn et al. 1973). VP, as well as oxytocin (OT), a natriuretic hormone in the rat (Verbalis et al. 1991), is synthesized in magnocellular neurosecretory neurones (MNCs) located in the supraoptic and paraventricular nuclei of the hypothalamus (Swaab et al. 1975). The release of both peptides depends on plasma osmolality (Dunn et al. 1973; Verbalis et al. 1986) and is primarily determined by the firing rate of the hypothalamic MNCs (Bourque, 1998). The regulation of MNC firing rate during osmotic changes involves modulation of intrinsic mechanosensitive ion channels (Oliet & Bourque, 1993), as well as extrinsic factors (Voisin & Bourque, 2002). Since the early findings (Mason, 1980), breakthroughs in our understanding of these mechanisms have been made. First, MNCs are themselves osmoreceptors as they display the intrinsic ability to transduce osmotic perturbations into firing changes (Bourque, 1989). Osmotransduction in MNCs results from the modulation of non-selective cation channels mediated by osmotically evoked changes in cell volume (Oliet & Bourque, 1993). TRPV-channel subunits, together with the actin cytoskeleton, represent strong candidate components of the transduction channel (Sharif Naeini et al. 2006; Zhang et al. 2007). However, the involvement of intracellular pathways in the gating of mechanosensitive channels has not been explored. Second, local astrocytes can confer osmosensitivity to MNCs through the release of taurine (Hussy et al. 1997; Deleuze et al. 1998). Third, the osmotic modulation of MNC activity in situ depends in large part on the afferent signals derived from peripheral and central osmoreceptors, and notably glutamatergic afferents (Richard & Bourque, 1995; for review see Bourque, 2008).

ERK are mitogen-activated protein kinases that transduce extracellular stimuli into intracellular post-translational and transcriptional responses (Lewis et al. 1998). ERK 1/2 are expressed in hypothalamic cells and physiological stimuli induce rapid ERK activation (i.e. phosphorylation) in MNCs (Nadjar et al. 2005; Blume et al. 2008, 2009). Once activated, ERK may set in motion both post-translational and transcriptional changes in various systems, which modify intrinsic neuronal and morphological properties as well as synaptic function (Ji et al. 1999; Thomas & Huganir, 2004; Cohen-Matsliah et al. 2007). ERK are thus potentially excellent candidates for a close partnership with mechanosensitive channels, local astrocytes and/or glutamatergic inputs for osmotic modulation of MNC activity. In the present work, we investigated ERK osmosensitivity in forebrain neurones. Our data suggest that ERK is activated during osmotic challenge, thereby contributing to the osmoregulation of VP release.

Methods

Ethical approval

The experiments followed the ethical guidelines of the International Association for the Study of Pain and the European Community Council directive of 22 September 2010 (2010/63/EU). They conformed to the principles of UK regulations, as described in Drummond (2009). The project was approved by Bordeaux Ethical Committee (CEEA50) under no. 50120163-A.

Animals

Virgin adult female (150–250 g) and male (100–225 g) Wistar rats were obtained from Janvier and maintained in a controlled environment (lights on 08.00–20.00 h, 22°C) with food and water freely available. They were housed three to four per cage. All efforts were made to minimize the number of animals used. Females were not cycled. To reduce stress effects, rats were habituated daily for 7 days before each experiment.

Chemicals, drugs and antibodies

Unless specifically stated, all chemicals were obtained from Sigma-Aldrich, Saint Quentin-Fallavier, France. U 0126 [1,4-diamino-2,3-dicyano-1,4-bis(o-aminophenylmercapto) butadiene], a selective MEK1/2 (ERK kinase) inhibitor, was obtained from Tocris Bioscience (London, UK). It was dissolved in 50% DMSO and 50% artificial cerebrospinal fluid (ACSF) for in vivo injections. For in vitro experiments (electrophysiological recordings) it was dissolved in pure DMSO at 30 mm stock solution to obtain a 30 μm final concentration with DMSO <0.1%. Rabbit polyclonal anti-phosphoERK and anti-total ERK primary antibodies were obtained from Cell Signaling Technology (Massachussetts, USA) (#9101 and #9102 respectively), rabbit polyclonal anti-Fos primary antibody from Santa Cruz, Texas, USA (sc-52), mouse monoclonal anti-OT primary antibody from Millipore, Massachussetts, USA (MAB 5296, USA) and mouse monoclonal anti-glial fibrillary astrocytic protein (anti-GFAP) primary antibody from Sigma (clone G-A-5). Mouse monoclonal anti-VP primary antibody was a kind gift of V. Geenen. Secondary antibodies were goat anti-rabbit IgG conjugated to Alexa 488, goat anti-mouse IgG conjugated to Alexa 546 (both from Molecular Probes, Oregon, USA), and biotinylated goat anti-rabbit IgG (Vector, California, USA). For the immunoperoxidase experiments, the third layer was a complex of avidin–biotin and the peroxidase was developed by the nickel DAB (ABC Kit and Ni-DAB Kit; Vector). Horseradish peroxidase-conjugated goat antirabbit IgG antibody (Millipore, Paris, France) was used for Western blot experiments.

Experimental protocols

All experimental protocols were performed after rats were anaesthetized, in different cohorts of animals. Each rat was only used once. The following experimental protocols were applied: plasma osmolality measurements after various osmotic stimuli (n = 62); immunohistochemical study of forebrain ERK phosphorylation after hyperosmotic stimuli (labelling distribution, time course and relation to the strength of the hyperosmotic stimulus) (n = 49); immunoblot analysis of ERK phosphorylation in the supraoptic nucleus after hyperosmotic and hypo-osmotic stimuli (n = 20); immunohistochemical study of forebrain ERK phosphorylation after hypo-osmotic stimuli (n = 12); dual labelling immunohistochemical studies of supraoptic ERK phosphorylation after various osmotic stimuli in relation with OT neurones, VP neurones and astrocytes (n = 16); studies of the effect of ERK inhibition upon forebrain ERK phosphorylation (n = 19), Fos expression (n = 13) and upon VP blood release (n = 29) after hyperosmotic stimuli; studies of the effect of ERK inhibition upon hypertonically induced depolarization and increase in firing of identified supraoptic MNCs recorded in vitro from hypothalamic acute slices (n = 6).

Anaesthesia

All experiments (except for electrophysiology) were performed in rats deeply anaesthetized with intraperitoneal (i.p.) sodium pentobarbital (50 mg kg⁻¹; Ceva Santé Animale, Paris, France). The depth of anaesthesia was judged adequate by the lack of motor reaction, hair raising and changes in breathing frequency to the manual pinch of the distal hindpaw. If the depth of the anaesthesia was judged inadequate, additional doses of 10 mg kg⁻¹ were administered until an adequate anaesthesia was achieved. In experiments in which intragastric infusion of fluid was performed, the amount of sodium pentobarbital was reduced and animals received 100 μg atropine i.p. to prevent the effects of the vagus reflex triggered by intragastric infusion. The same dose of atropine is used as premedication for rat anaesthesia during surgery (Dallel et al. 1998). In experiments in which animals were placed in a stereotaxic frame, anaesthetic cream (Emla 5%; AstraZeneca, Rueil-Malmaison, France) was applied into the auditory canal and local anaesthesia using xylocaine (2%; AstraZeneca) was pre-emptively performed at all points of surgery.

Osmotic stimuli

Hyperosmotic stimuli were applied in sodium pentobarbital-anaesthetized rats, using i.p. injections of 3 m or 6 m saline (1 ml/300 g body weight), while control animals received isotonic (0.15 m) saline. Hypo-osmotic stimuli were applied in sodium pentobarbital-anaesthetized rats that have also received atropine, using slow (1 ml min⁻¹) intragastric infusion of 10 ml distilled water administered through a polythene tubing inserted in the oesophagus using a feeding tube and connected to a 10 ml syringe placed in an infusion pump (Harvard Apparatus, Holliston, MA, USA), while control animals received 10 ml isotonic (0.15 m) saline. The infusion lasted for 10 min and the animals were killed 10 min after the end of the infusion.

Plasma osmolality measurements

Sodium pentobarbital-anaesthetized rats were decapitated at different times (0, 5, 10, 30 or 120 min) after they received osmotic stimuli. Blood samples were collected directly in heparinized tubes and kept on ice. Blood cells and plasma were separated using centrifugation at 3000 g for 20 min and the supernatant was collected. Plasma osmolality was measured for each sample with a freezing point osmometer (3320 Microosmometer, Advanced Instrument; Radiometer Analytical, Villeurbanne, France, France).

Intracerebroventricular injections

Sodium pentobarbital-anaesthetized rats were placed in a stereotaxic frame where rectal temperature was maintained at 37°C by a thermostatically controlled electric blanket. Through a craniotomy, the tip of a 10 μl Hamilton was lowered into the third ventricle stereotaxically (anteroposterior −2.3 mm from Bregma, mediolateral 0 mm, deep 8 mm, according to the atlas of Paxinos & Watson, 1997). To test the effect of ERK inhibition upon osmotically induced forebrain ERK phosphorylation or VP release, U 0126 (10 nmol in 1 μl) was injected into the third ventricle 10 min before rats received an i.p. injection of 3 m saline and the animals were killed 10 min later (see below). An additional higher dose (50 nmol) was also tested. To test the effect of ERK inhibition upon forebrain Fos expression, U 0126 (10 nmol in 1 μl) was injected into the third ventricle every 20 min, five times. Ten minutes after the first administration of U 0126, rats received an i.p. injection of 3 m saline and the animals were killed 90 min later (see below). In all the above cases, control animals received vehicle (50% DMSO–50% ACSF) into the third ventricle instead of U 0126. The 10 nmol dose of U 0126 injected into the third ventricle is one order of magnitude larger than the dose used by some authors to block ERK activation using intracerebroventricular injections (Han & Holtzman, 2000; Kuroki et al. 2001) and similar to the one used by other authors using intrathecal injections (Chen et al. 2008).

Immunohistochemistry protocols

Rats were killed at appropriate time by an overdose of sodium pentobarbital, received an intracardiac injection of heparin (5000 U in 0.5 ml; Sanofi Aventis, Paris, France) and were perfused transcardially with phosphate-buffered saline (PBS, 0.2 m, pH 7.3) containing 4% paraformaldehyde and 0.08% picric acid for 30 min. The brain was then removed and post-fixed overnight at 4°C in the same fixative before being transferred into PBS containing 30% sucrose at 4°C until use. Coronal sections (30 μm) were cut on a freezing microtome and collected in 0.05 m Tris-buffered saline (TBS) before being processed. In all cases, sections were rinsed in TBS several times, between and after each incubation and finally transferred on to gelatinized slides before being coverslipped using either DPX medium (Fluka; Sigma-Aldrich, Saint Quentin Fallavier, France) for immunoperoxidase experiments, or Vectashield (AbCys, Paris, France) for immunofluorescence experiments. In the latter case, sections were kept at 4°C in the dark until image acquisition.

Specificity controls consisted of the omission of the primary antibody on one hand and incubation of sections in inappropriate secondary antibodies on the other. In all these control experiments, no specific staining was evident. Further evidence for the specificity of antibodies raised against OT was concluded from the fact that there was no reaction of the antibody raised against OT in the suprachiasmatic nucleus of the hypothalamus, which contains VP neurones only. Further evidence for the specificity of antibodies raised against VP was concluded from the fact that there was no reaction of the antibody raised against VP in the anterior commissural nucleus of the hypothalamus, which contains OT neurones only.

For immunoperoxidase experiments, all antibodies were diluted in TBS 0.05 m containing 0.25% bovine serum albumin, 0.5% Triton X-100 and 2% normal goat serum. Free-floating sections were incubated for 5 min in TBS solution containing 10% methanol and 3% H₂O₂ to quench endogenous peroxidases, then placed in TBS 0.05 m containing 0.25% bovine serum albumin, 0.5% Triton X-100 and 2% normal goat serum for 15 min, before incubation at room temperature in solutions containing rabbit primary antibodies directed against either phosphoERK (1:2000) or Fos (1:30,000) overnight at room temperature. The next day, sections were placed in a secondary biotinylated goat antirabbit antibody (1:400, 2 h, room temperature). Immunoreactivity for phosphoERK or Fos was visualized using nickel-DAB (ABC Kit and Ni-DAB Kit; Vector).

For immunofluorescence experiments, free-floating sections were incubated for 1 h in 0.1 m PBS containing 0.5% casein then incubated for 3 days at 4°C in different primary antibodies diluted in 0.05 m TBS containing 0.25% bovine serum albumin, 0.5% Triton X-100: rabbit polyclonal anti-phosphoERK antibody (1:2,000) and mouse monoclonal anti-VP antibody (1:30,000), rabbit polyclonal anti-phosphoERK antibody (1:2000) and mouse monoclonal anti-OT antibody (1:10,000), rabbit polyclonal anti-phosphoERK antibody (1:2000) and mouse monoclonal anti-GFAP (1:2000). After 3 days, sections were placed overnight at 4°C in a secondary antibody solution containing goat antimouse IgG conjugated to Alexa 546 (1:2000) and goat antirabbit IgG conjugated to Alexa 488 (1:2000).

Immunohistochemistry data acquisition

For immunoperoxidase experiments (phosphoERK and Fos labelling), images were acquired in a bright field on a Zeiss Axiophot 1 microscope (Carl Zeiss S.A.S, Le Pecq, France) via a 10× objective (Zeiss Fluar, NA 0.5) with a video camera DFC300X, Rueil-Malmaison, France (Leica Microsystems). The mean number of immunopositive neurones per area unit (10⁴ μm²) was counted in each structure of interest, on two to three slices/rat and on both sides for bilateral structures, using Image J software. As slices were separated by 120 μm, neurones could not be counted twice. In the paraventricular nucleus, counts were made in the lateral magnocellular part of the nucleus, where VP neurones predominate and in the medial magnocellular part, where OT neurones predominate.

For immunofluorescence experiments, images were acquired with a Leica DMI 6000 spinning disk confocal microscope equipped with a confocal scanning unit CSU22 (Yokogawa, Tokyo, Japan), using three different objectives: 40× (HCX PL Apo CS, NA 1.25), 63× (HCX PL Apo, NA 1.40) and 100× (HCX PL Apo, NA 1.40) and a Quantem camera (Roper Scientific, Evry, France) driven by Metamorph software. Each acquisition was made on the Z axis, with an optical section of 0.2 μm for 40× objective and 0.1 μm for 63× and 100×, thanks to a piezoelectric motor P721.LLQ (Roper Scientific). Laser diodes used were 473 nm (for Alexa 488) and 532 nm (for Alexa 546). A Z projection (30 images, average intensity) of each stack of acquired images was obtained with Image J software, before adjusting brightness and contrast. Counts of phosphoERK-, OT-and VP-immunopositive neurones and counts of double-labelled neurones were made on both sides at three different levels (rostral, median and caudal) of the supraoptic nucleus. Co-labelling of phosphoERK and GFAP was quantified by measuring the percentage of area where phosphoERK and GFAP colocalized.

Western blot

Ten minutes after hyper-or hypo-osmotic stimulus, sodium pentobarbital-anaesthetized rats were killed by decapitation. The brain was removed from the skull and immediately flash frozen in cold isopentane. Coronal 300 μm brain slices were cut on a cryostat at −13°C to avoid protein degradation. Supraoptic nuclei were punched on each slice (one punch on each side of each slice, three slices per rat) and kept at −80°C until use. They were homogenized at 4°C with a sonicator in 50 μl of lysis buffer containing (in mm): Tris-HCl, 50; MgCl₂, 5; dithiothreitol, 1; sodium orthovanadate, 2; 1% Triton X-100 and protease inhibitor cocktail (Sigma, France, Saint Quentin Fallavier; broad-spectrum inhibitor cocktail for serine, aspartate proteases, aminopeptidase, pepstatin A, E-64, bestatin, leupeptin and aprotinin). Homogenates were then incubated 30 min on ice and clarified by centrifugation (11,000 g, 20 min, 4°C). Supernatants were collected and the total amount of protein was determined with Dc protein assay (Bio-Rad, Marnes la Coquette, France) using a spectrophotometer (Beckman, Coulter, Villepinte, France DU 640B).

Each sample was then mixed with 2× loading buffer (Tris-HCl, pH 6.8, 62.5 mm; SDS, 10%; β-mercaptoethanol, 2.5%; bromophenol blue, 0.025%), warmed at 95°C and subjected to a 10% polyacrylamide SDS-PAGE electrophoresis (50 μg protein loaded/lane) for hypertonic stimulation and to a 12% polyacrylamide SDS-PAGE electrophoresis (20 μg protein loaded/lane) for hypotonic stimulation, before being electroblotted on to a PVDF membrane (Immobilon-Psq; Millipore). The membranes were then incubated for 2 h in blocking solution containing 5% dried milk in TBS-T (Tris-buffered saline: 10 mm; Tris-HCl, pH 8.0, 500 mm; NaCl and 0.1% Tween-20). Membranes were then incubated overnight at 4°C with primary anti-phosphoERK (1:1000) antibody or primary antitotal ERK (1:1000) antibody, diluted in TBS-T and 5% BSA. The next day, membranes were washed extensively three times with TBS-T and 5% dried milk before being incubated for 2 h at room temperature with horseradish peroxidase-conjugated goat antirabbit IgG antibody (1:2000 in TBS-T and 5% dried milk). After three washes with TBS-T and three with TBS only, immunoreactivity was then determined by enhanced chemiluminescence (GE Healthcare, Aulnay sous Bois, France). Films were scanned and processed using densitometric analysis with Image J software. With the exposure time needed here to see both phosphoERK 1 and 2, only optical density changes for 44 kDa phosphoERK were well visible. The ratio between phosphoERK (44 kDa) and ERK (44 kDa) optical densities was thus calculated for each experiment. Controls for antiphosphoERK antibody specificity were performed by dot blot using purified phosphoERK and total ERK proteins (no. 9103; Cell Signaling Technology).

Hypothalamic slices preparation

Experiments were performed on acute hypothalamic slices obtained from 1–2-month-old Wistar male rats. Rats were anaesthetized with 5% isoflurane and decapitated. The brain was quickly removed from the skull and placed in ice-cold ACSF saturated with 95% O₂ + 5% CO₂. The composition of ACSF was as follows (in mm): NaCl, 123; KCl, 2.5; Na₂HPO₄, 1; NaHCO₃, 26.2; MgCl₂, 1.3; CaCl₂, 2.5; and glucose, 10 (pH 7.4; 295–300 mosmol kg⁻¹). Coronal slices (300 μm) were cut with a vibratome (Leica VT 1000S) from a block of tissue containing the hypothalamus. Supraoptic nucleus-containing slices were then hemisected along the midline and allowed to recover for at least 1 h at 33°C in a submerged chamber containing ACSF before recordings. After 30 min of recovery at room temperature, a hemislice was then transferred and submerged in the recording chamber, in which it was continuously perfused (1–2 ml min⁻¹) with ACSF at room temperature.

Patch clamp recordings

Magnocellular neurones of the supraoptic nucleus were visually identified using infrared differential interference contrast (DIC; Olympus, Rungis, France XC-ST70CE) microscopy (Olympus BX 51 WI). The patch-clamp recording pipettes (3–5 MΩ) were filled with a biocytin (0.1 %) solution containing (in mm): 120 potassium gluconate, 1.3 MgCl₂, 0.1 CaCl₂, 10 Hepes, 1 EGTA, 61.5 mannitol, pH adjusted to 7.2 with KOH 1 m; 285 mosmol kg⁻¹. Membrane voltage was recorded at room temperature in the current clamp mode (with injected current to maintain the neurone 5 mV under action potential threshold) of the whole cell configuration of the patch clamp technique, using a Multiclamp 700 B amplifier (Molecular Devices, Sunnyvale, USA). Signals were filtered at 2 kHz and digitized at 5 kHz via a DigiData 1440A interface (Molecular Devices). Series resistance (6–15 MΩ) was monitored online, and cells were excluded from data analysis if more than a 20% change occurred during the course of the experiment. Data were collected and analysed online using pClamp 10 software (Molecular Devices). The recording protocol was as follows: baseline recording in isotonic conditions for at least 5 min followed by changing the medium for a medium made hypertonic (+30 or +60 mosmol kg⁻¹) by the addition of mannitol (8 min). After washout in isotonic condition and return to baseline level, the slice was perfused for at least 5 min with isotonic ACSF containing U 0126 (30 μm) before a similar hypertonic stimulus containing U 0126 (30 μm) was applied. For each recorded neurone, the analysed parameters were the amplitude of the hypertonically induced membrane depolarization and the hypertonically induced increase in firing frequency in the presence or absence of U 0126.

Cell identification

After recording, magnocellular neurones filled with biocytin (0.1%) were identified further by immunocytochemistry as previously described (Bonfardin et al. 2012). At the end of the recording period, slices were fixed by immersion in 4% paraformaldehyde and 0.15% picric acid (2 h, room temperature and during 2 or 3 days at 4°C). The biocytin was visualized after incubation (overnight at 4°C) with streptavidin-conjugated to 7-amino-4-methyl-coumarin-3-acetic acid (AbCys) diluted 1:200. Sections containing biocytin-positive neurones were then incubated for 7 days (at 4°C) in a mixture of a mouse monoclonal antibody raised against VP (diluted 1:20,000; gift from V. Geenen, University of Liege, Liege, Belgium) and a rabbit polyclonal serum raised against OT (diluted 1:4000; gift from T. Higuchi, University of Fukui, Fukui, Japan). Immunoreactivities were then revealed by incubation in a mixture with goat antimouse IgGs conjugated to fluorescein isothiocyanate (AbCys) (diluted 1:1600) and goat antirabbit IgGs conjugated to Texas Red (AbCys) (diluted to 1:500, overnight at 4°C). All antibodies were diluted in a solution of TBS containing 0.25% BSA and 2% Triton X-100. Controls with omission of primary antibodies yielded no specific immunolabelling. Slices were mounted in Vectashield (AbCys) and examined with a confocal microscope (Leica DMR TCS SP2 AOBS on an upright stand, using objective HCX Plan Apo CS 63×, numerical aperture 1.40). The lasers used were argon (488 nm), green helium–neon (543 nm), and a diode laser (405 nm). Images were acquired using Leica TCS software with a sequential mode to avoid interference between each channel. All confocal images are projections of 20 consecutive optical sections (0.5 μm). Contrast/brightness enhancement was done in parallel on obtained stacks using Adobe Photoshop 8.0 (Adobe Systems).

Plasma vasopressin measurements

Two protocols were used in the sodium pentobarbital-anaesthetized rats. In the first protocol, the effects of i.p. injection of 3 m hypertonic saline on plasma VP concentrations were compared to the effects of i.p. injection of isotonic saline. In the second protocol, the effects of intracerebroventricular injection of U 0126 (10 nmol) on plasma VP concentrations after i.p. injection of 3 m hypertonic saline were compared to the effects of intracerebroventricular injection of vehicle. All rats were decapitated 10 min after they received osmotic stimuli. Blood samples were collected directly in heparinized tubes and kept on ice. Blood cells and plasma were separated using centrifugation at 3000 g for 20 min and the supernatant was collected. Plasma VP was measured for each sample using a VP enzyme-linked immunosorbent assay kit as per manufacturer's instructions (Cayman Chemical Company, Ann Arbor, MI, USA).

Statistical analysis

Data are reported as means ± s.e.m. They were compared statistically using one-or two-way ANOVA followed by Bonferroni's post hoc test for all experiments except electrophysiological recordings where data were analysed using two-tailed paired Student's t test and VP measurements where data were analysed using two-tailed unpaired Student's t test. Significance was assessed at P < 0.05.

Results

Plasma osmolality in anaesthetized rats after various osmotic stimuli

We first checked that hypertonically induced rises in osmolality matched the changes in plasma osmotic pressure after i.p. injection of hypertonic saline reported in a previous work (Dunn et al. 1973). An analysis of variance of plasma osmolality with time (5, 30 and 120 min) and osmotic stimulus treatment (3 m hypertonic saline or isotonic saline) as factors revealed significant effects of treatment (P < 0.001) and time (P < 0.001) and significant interaction between treatment and time (P < 0.001). Further analysis showed that time was only effective in the 3 m hypertonic saline-treated animals. i.p. injection of 3 m hypertonic saline increased plasma osmolality from 295 ± 1 mosmol kg⁻¹ to 333 ± 2 mosmol kg⁻¹ within 5 min, afterwards plasma osmolality decreased over time, but was still elevated at 307 ± 1 mosmol kg⁻¹ after 2 h (Fig. 1). Plasma osmolality remained close to 295 ± 1 mosmol kg⁻¹ over time after isotonic saline injections (Fig. 1). At the different times, plasma osmolality was always larger in 3 m hypertonic saline-treated animals than in isotonic saline-treated animals (P < 0.001). An analysis of variance of plasma osmolality at 10 min showed that i.p. injection of 6 m hypertonic saline caused larger osmolality increases than 3 m saline injections (P < 0.001, Fig. 1). In both cases, plasma osmolality reached larger values than after isotonic saline (P < 0.001). Slow intragastric infusion of water for 10 min, but not isotonic saline, reduced plasma osmolality from 296 ± 1 mosmol kg⁻¹ to 284 ± 1 mosmol kg⁻¹ 20 min later (P < 0.001, Fig. 1).

Figure 1.

Plasma osmolality values before (basal) and after i.p. injection of isotonic (0.15 m), 3 or 6 m saline or slow intragastric infusion of water or isotonic saline are shown at different time points. Each symbol gives the mean (±s.e.m.) value obtained from five rats. See text for description of statistical differences.

Hyperosmotic stimuli induce extracellular signal-regulated protein kinase activation in the forebrain: topography, time course and relation to the strength of the osmotic stimulus

It is known that ERK can be activated rapidly and reversibly within hypothalamic neurones, for instance in response to systemic injections of interleukin-1β (Nadjar et al. 2005). In control animals that received i.p. injection of isotonic saline, we found no changes in ERK phosphorylation over time. However, we found that i.p. injection of 3 m hypertonic saline resulted 10 min after in strong ERK phosphorylation in forebrain osmosensitive regions such as the median preoptic nucleus (+139%), subfornical organ (+157%), supraoptic (+129%) and paraventricular (+144%) hypothalamic nuclei, the organum vasculosum lamina terminalis (OVLT; +113%, not shown), but not (+2%) in the thalamic anterodorsal nuclei that are not involved in osmoregulation (Figs 2 and 3). Immunostaining was localized within dendrites, cytoplasm and nuclei (Fig. 2). An analysis of variance of ERK phosphorylation with time (5, 10, 30 and 120 min) and osmotic stimulus treatment as factors revealed significant effects of treatment (P < 0.001) and time (P < 0.001) and significant interaction between treatment and time (P < 0.001) in all the osmosensitive forebrain regions analysed. Further analysis showed that time was always effective in the 3 m hypertonic saline-treated animals in all osmosensitive forebrain regions. The numbers of phosphoERK-immunopositive cell bodies, counted in each osmosensitive forebrain region, reached maximal values after 10 min and then progressively decreased (Fig. 3). ERK phosphorylation increased 5 min after the beginning of the application in all osmosensitive forebrain regions as compared to isotonic saline-treated animals (P < 0.001), except for the supraoptic nucleus where the increase started later (Fig. 3). The numbers of immunopositive cells in the 3 m hypertonic saline-treated animals were no longer different from the isotonic saline-treated animals 2 h after the stimulus application. There was also a small but statistically significant effect of time in the isotonic saline-treated animals in all osmosensitive forebrain regions, but the OVLT. In these animals, differences were found between the early (5 and 10 min) when compared to late (30 and 120 min) times (P < 0.01). No changes in ERK phosphorylation over time were found in the anterodorsal nucleus of the thalamus of all animals (Fig. 3).

Figure 2.

Bright field microphotographs (10×) show phosphoERK immunostaining in different forebrain regions 10 min after i.p. injection of 0.15 m, 3 m or 6 m saline. ADT, anterodorsal nucleus of the thalamus; ERK, extracellular signal-regulated protein kinases; MnPO, median preoptic nucleus; PVN, paraventricular nucleus; SFO, subfornical organ; SON, supraoptic nucleus. Scale bars, 100 μm.

Figure 3.

Histograms show mean (±s.e.m.) numbers of phosphoERK-immunopositive cells per area unit (10⁴ μm²) in different forebrain regions at different times after isotonic (0.15 m, white bars) or hypertonic (3 m, grey bars) saline i.p. injection. ADT, anterodorsal nucleus of the thalamus; ERK, extracellular signal-regulated protein kinases; MnPO, median preoptic nucleus; OVLT, organum vasculosum lamina terminalis; PVN, paraventricular nucleus; SFO, subfornical organ; SON, supraoptic nucleus. Numbers in bars indicate the number of animals per group. *P < 0.05, ***P < 0.001.

As ERK phosphorylation peaked 10 min after hyperosmotic stimulus application, we studied the relation between the intensity of the osmotic stimuli and the numbers of immunopositive cell bodies at this specific time point. In all osmosensitive forebrain regions, but not in the thalamic anterodorsal nuclei, the numbers of phosphoERK-immunopositive cell bodies significantly increased in proportion to the intensity of the hyperosmotic stimulus (Fig. 4), suggesting that ERK activation in forebrain osmosensitive structures relates to the intensity of osmotic stimuli. As at the osmotic set point phosphoERK-immunopositive profiles were already present in these structures, we next investigated the effects of hypo-osmotic stimuli on ERK phosphorylation.

Figure 4.

Histograms show mean (±s.e.m.) numbers of phosphoERK-immunopositive cells per area unit (10⁴ μm²) in different forebrain regions 10 min after 0.15 m (white bars), 3 m (grey bars) or 6 m (black bars) saline i.p. injection. ADT, anterodorsal nucleus of the thalamus; ERK, extracellular signal-regulated protein kinases; MnPO, median preoptic nucleus; OVLT, organum vasculosum lamina terminalis; PVN, paraventricular nucleus; SFO, subfornical organ; SON, supraoptic nucleus. Numbers in bars indicate the number of animals per group. ***P < 0.001.

Hypo-osmotic stimuli decrease extracellular signal-regulated protein kinase phosphorylation in the forebrain

In all osmosensitive forebrain regions, the numbers of phosphoERK-immunopositive cell bodies significantly decreased 20 min after a slow intragastric infusion of water for 10 min, compared to an infusion of isotonic saline (Fig. 5): in the OVLT (−49%), median preoptic nucleus (−53%), subfornical organ (−44 %), supraoptic (−51 %) and paraventricular (−40 %) hypothalamic nuclei.

Figure 5.

A, bright field microphotographs (10×) show phosphoERK immunostaining in different forebrain regions 20 min after intragastric infusion of 10 ml isotonic saline or water was started. Scale bars, 100 μm. B, histograms show mean (± s.e.m.) numbers of phosphoERK-immunopositive cells per area unit (10⁴ μm²) in different forebrain regions 20 min after intragastric infusion of 10 ml isotonic saline or water was started. Numbers in bars indicate the number of animals per group. ***P < 0.001. ERK, extracellular signal-regulated protein kinases; MnPO, median preoptic nucleus; OVLT, organum vasculosum lamina terminalis; PVN, paraventricular nucleus; SFO, subfornical organ; SON, supraoptic nucleus.

Immunoblot analysis confirms extracellular signal-regulated protein kinase phosphorylation increases and decreases in the supraoptic nucleus after hyperosmotic stimuli and hypo-osmotic stimuli, respectively

To analyse further the relation between the intensity of osmotic stimuli and ERK activation, we measured the amount of phosphoERK in the supraoptic nucleus after osmotic stimuli of different strength, using Western blot. Analysis of the 44 kDa phosphoERK bands in the supraoptic nucleus revealed a 30 ± 6% relative increase in optical density after 3 m osmotic stimulus, relative to the level measured after i.p. injection of isotonic saline, and a 61 ± 6% relative increase after 6 m osmotic stimulus (Fig. 6). In addition, there was a 26 ± 2% relative decrease in optical density after hypo-osmotic stimulus, relative to the level measured after intragastric infusion of isotonic saline (Fig. 6).

Figure 6.

A, on the left and right sides, examples of Western blot of supraoptic nucleus phosphoERK (top panel) and total ERK (bottom panel) after different hyper-or hypo-osmotic stimuli. The two middle bands show the positive and negative controls for antibody specificity. B, histograms show mean (±s.e.m.) values of the normalized ratio 44 kDa phosphoERK OD/44 kDa ERK OD. Numbers in bars indicate the number of animals per group. *P < 0.05, **P < 0.01; ***P < 0.001. ERK, extracellular signal-regulated protein kinases; tERK, total ERK; OD, optical density.

Extracellular signal-regulated protein kinase is differentially activated within oxytocin neurones, vasopressin neurones and astrocytes in the supraoptic nucleus

As local astrocytes can confer osmosensitivity to MNCs and both VP and OT are released in response to osmotic challenge, we studied ERK activation in selected cell populations of the supraoptic nucleus using dual immunohistochemical labelling techniques. The numbers of VP and OT neurones that were counted in the analysed areas gave proportions equal to 55% and 45% respectively, which is in line with the ones reported in the literature (Swaab et al. 1975). An analysis of variance of ERK phosphorylation with the cell phenotype and osmotic stimulus treatment as factors revealed significant effects of treatment (P < 0.001) and cell phenotype (P < 0.001) and significant interaction between treatment and cell phenotype (P < 0.001). The treatment effect was present in each cell population (P < 0.001) and the numbers of phosphoERK-immunopositive cell bodies increased in proportion with the intensity of the hyperosmotic stimuli in both OT and VP neurones (Fig. 7). PhosphoERK was expressed twice more in VP neurones than in OT neurones 10 min after i.p. injections of isotonic saline (Fig. 7, P < 0.001) and the overexpression of phosphoERK in VP neurones compared to OT neurones prevailed (P < 0.001) after i.p. injections of 3 m or 6 m hypertonic saline (Fig. 7). In addition to the fact that the ventral glial lamina was almost devoid of phosphoERK immunolabelling, dual immunofluorescent labelling experiments for phosphoERK and GFAP, a marker of supraoptic astrocytes (Bonfanti et al. 1993), showed no evidence of colocalization in any tested osmotic conditions (hypo-osmotic, iso-osmotic, 3 m or 6 m hyperosmotic stimulus) (Fig. 8).

Figure 7.

A, confocal images (40×, Z projections of 30 images, average intensity) show phosphoERK (left panel), OT (top middle panel) and VP (bottom middle panel) immunostaining in the supraoptic nucleus 10 min after i.p. injection of 3 m saline. Panels on the right show merged images. Scale bars, 20 μm. Insets show high-power views (100×) of selected areas (white dot lined squares). Inset scale bars, 10 μm. B, histograms show mean (±s.e.m.) numbers of double-labelled cells per area unit (10⁴ μm²) in the supraoptic nucleus 10 min after 0.15 m, 3 m or 6 m saline i.p. injection. Numbers in bars indicate the number of animals per group. For the sake of clarity, only differences between cell types at different stimulus intensities are shown. ***P < 0.001. OT, oxytocin; VP, vasopressin.

Figure 8.

Confocal images (40×, Z projections of 30 images, average intensity) show phosphoERK (left panel) and GFAP (middle panel) immunostaining in the supraoptic nucleus after various osmotic stimuli. Panels on the right show merged images. Scale bars, 20 μm. GFAP, glial fibrillary acidic protein; pERK, phosphoERK.

Extracellular signal-regulated protein kinase inhibition impairs forebrain extracellular signal-regulated protein kinase phosphorylation and Fos expression after hyperosmotic stimuli

To assess the role of ERK phosphorylation in the osmotic control of VP release, we used intracerebroventricular injection of U 0126, a selective MEK1/2 (ERK kinase) inhibitor. In preliminary experiments, we found that the range of efficacy of a single injection of U 0126 (10 nmol in 1 μl vehicle) into the third ventricle was 10–30 min (not shown). We thus tested the effect of a single injection of U 0126 10 min before i.p. injection of 3 m hypertonic saline upon ERK phosphorylation measured by immunohistochemical detection. In all osmosensitive forebrain regions, the numbers of phosphoERK-immunopositive cell bodies significantly decreased after intracerebroventricular injection of U 0126 (10 nmol in 1 μl vehicle), compared to intracerebroventricular injection of vehicle alone (Fig. 9) in the OVLT (−67%), median preoptic nucleus (−68%), subfornical organ (−54%), supraoptic (−42%) and paraventricular (−45%) hypothalamic nuclei. We also tested a higher dose of U 0126 (50 nmol) and found it had no larger effects than the 10 nmol dose of U 0126 (10 nmol) on ERK phosphorylation (Fig. 9).

Figure 9.

A, bright field microphotographs (10×) show phosphoERK immunostaining induced by i.p. injection of 3 m saline in presence (10 nmol, right panel) or in absence (left panel) of the selective ERK phosphorylation inhibitor U 0126. Scale bars, 100 μm. B, histograms show mean (± s.e.m.) numbers of phosphoERK-immunopositive cells per area unit (10⁴ μm²) in different forebrain regions 10 min after osmotic stimuli in presence (10 nmol, grey bars; 50 nmol, black bars) or in absence (white bars) of U 0126. The numbers of animals per group were four to eight. ***P < 0.001. ERK, extracellular signal-regulated protein kinases; MnPO, median preoptic nucleus; OVLT, organum vasculosum lamina terminalis; pERK, phosphoERK; PVN, paraventricular nucleus; SFO, subfornical organ; SON, supraoptic nucleus.

Fos, the product of the immediate early gene c-fos, has been widely used as a marker of neuronal activation, in particular to map forebrain osmosensitive neurones (Sharp et al. 1991; Hoffman et al. 1993; Fujihara et al. 2009). We thus also tested the effect of U 0126 upon Fos expression measured by immunohistochemical detection. Accordingly, we found that i.p. injection of 3 m hypertonic saline, as compared to i.p. injection of isotonic saline, induced a strong and significant Fos expression in forebrain osmosensitive structures. In addition, osmotically induced Fos expression in these structures significantly decreased after intracerebroventricular injection of U 0126 (10 nmol in 1 μl vehicle), compared to intracerebroventricular injection of vehicle alone (Fig. 10) in the OVLT (−53%), median preoptic nucleus (−23%), subfornical organ (−53%), supraoptic (−29%) and paraventricular (−48%) hypothalamic nuclei.

A, bright field microphotographs (10×) show Fos immunolabelling induced by i.p. injection of isotonic saline (left panel) or 3 m saline in presence (right panel) or in absence (middle panel) of the selective ERK phosphorylation inhibitor U 0126. Scale bars, 100 μm. B, histograms show mean (±s.e.m.) numbers of Fos immunopositive cells per area unit (10⁴ μm²) in different forebrain regions 90 min after i.p. injection of isotonic saline (white bars) or osmotic stimuli in presence (black bars) or in absence (grey bars) of U 0126. Numbers in bars indicate the number of animals per group. ***P < 0.001. ERK, extracellular signal-regulated protein kinases; MnPO, median preoptic nucleus; OVLT, organum vasculosum lamina terminalis; PVN, paraventricular nucleus; SFO, subfornical organ; SON, supraoptic nucleus.

Extracellular signal-regulated protein kinase inhibition decreases hypertonically induced electrical activation of supraoptic magnocellular neurones

To test functionally the involvement of ERK phosphorylation in the osmotic control of VP release, we performed electrophysiological recordings from supraoptic MNCs in acute hypothalamic slices. As previously shown (Mason, 1980), application of hypertonic solutions to the slice preparations induced a depolarization of MNC membrane potential by 8.1 ± 1.0 mV after +60 mosmol kg⁻¹ and 7.2 ± 1.3 mV after +30 mosmol kg⁻¹. In both conditions, application of U 0126 (30 μm) significantly reduced the osmotically induced depolarizations to 5.1 ± 0.6 mV (P < 0.05) and 3.5 ± 1.6 mV (P < 0.01) respectively (Fig. 11A and B). Hypertonically induced depolarizations were also accompanied by increases in the firing rate of MNCs that were reduced in the presence of U 0126 (30 μm) down to 26.8 ± 5.2 % (+60 mosmol kg⁻¹, P < 0.05) and 12.5 ± 6.0% (+30 mosmol kg⁻¹, P = 0.07) of the initial osmotically evoked increases in action potential firing (Fig. 11A and C).

Figure 11.

A, the trace shows whole cell voltage responses from a magnocellular neurosecretory cell recorded from an acute hypothalamic slice following application of hypertonic stimuli (+60 mosmol kg⁻¹) in the absence or presence of U 0126 (30 μm). B, scatter plots showing the depolarization evoked in each of cells in response to a +60 mosmol kg⁻¹ stimulus (top) or a +30 mosmol kg⁻¹ stimulus (bottom) in the absence or presence of U 0126 (30 μm). C, scatter plots showing the evoked increase in the number of spikes in each of cells in response to a +60 mosmol kg⁻¹ stimulus (top) or a +30 mosmol kg⁻¹ stimulus (bottom) in the absence or presence of U 0126 (30 μm). B and C, continuous lines join data obtained from individual cells and dotted lines join the mean response values before and after U 0126 (±s.e.m.). *P < 0.05, **P < 0.01.

Three of the recorded neurones were found to be immunopositive for VP while the others were immunopositive for OT (Fig. 12A). The osmotically induced membrane depolarizations and increases in firing rate and their reductions in the presence of U 0126 (30 μm) were similar in OT and VP cells, although they could not be compared statistically due to the small numbers of cells in each category (Fig. 12B).

Figure 12.

A, traces show whole cell voltage responses from identified VP and OT neurones recorded from acute hypothalamic slices following application of hypertonic stimuli (+60 mosmol kg⁻¹). The recorded cells (yellow arrows) displayed fluorescence for biocytin (blue) and immunoreactivity for VP (green, left) or for OT (red, right). Scale bars, 20 μm. B, bar histograms showing the mean ± s.d. depolarizations (left) and changes in firing frequency (right) evoked in VP and OT neurones in response to a +60 mosmol kg⁻¹ stimulus in absence or in presence of U 0126 (30 μm). OT, oxytocin; VP, vasopressin.

Extracellular signal-regulated protein kinase inhibition reduces hypertonically induced plasma vasopressin increases in anaesthetized rats

We finally assessed the role of ERK activation upon VP release in vivo. First, we found that i.p. injection of 3 m hypertonic saline, as compared to i.p. injection of isotonic saline, induced a strong and significant (P < 0.001) increase in VP concentrations in the circulation. VP levels increased from 2.1 ± 0.5 pg ml⁻¹ (n = 6) to 19.4 ± 5.2 pg ml⁻¹ (n = 7) 10 min after the challenge (Fig. 13). Secondly, we found that a single intracerebroventricular injection of U 0126 (10 nmol in 1 μl) 10 min before i.p. injection of 3 m hypertonic saline significantly (P < 0.05) reduced the VP blood concentration measured 10 min after the osmotic challenge. VP levels decreased from 20.9 ± 4.8 pg ml⁻¹ in control animals that received intracerebroventricular injection of vehicle (n = 7) to 10.2 ± 2.3 pg ml⁻¹ in animals that received intracerebroventricular injection of U 0126 (n = 9) (Fig. 13).

Figure 13.

Histograms show mean (±s.e.m.) plasma VP concentrations (i) left, after i.p. injection of 3 m hypertonic saline (n = 7), as compared to i.p. injection of Iso (n = 6), and (ii) right, after i.p. injection of 3 m hypertonic saline following ICV injection of vehicle (n = 7) or U 0126 (n = 9). *P < 0.05, ***P < 0.001. ICV, intracerebroventricular; Iso, isotonic saline; VP, vasopressin.

Discussion

The present work provides the first evidence that ERK is selectively activated (i.e. phosphorylated) in osmosensitive forebrain regions in response to osmotic challenges. In these regions, ERK activation was proportional to the strength of the osmotic stimulus and basal ERK phosphorylation was reduced by hypo-osmolality. Furthermore, ERK activation was found to be more predominant in VP than in OT neurones in the supraoptic nucleus and was not present in supraoptic astrocytes, even in conditions of hypo-osmolality. Finally, effective reduction of ERK phosphorylation with the selective inhibitor U 0126 significantly decreased osmotically induced: (i) forebrain expression of the activity dependent gene product Fos; (ii) depolarization and increase in firing of supraoptic MNCs recorded in vitro; and (iii) VP release in the bloodstream. Altogether, these findings show that ERK phosphorylation is involved in the integration of osmotic stimuli into the activity changes in osmoregulatory hypothalamic neurones leading to VP release.

A number of in vitro and in vivo studies have shown that neurones in the supraoptic and paraventricular nuclei, subfornical organ, median preoptic nucleus and OVLT structures can change their firing pattern after an osmotic challenge (Mason, 1980; Nakashima et al. 1985; Sibbald et al. 1988; Bourque, 1989; Vivas et al. 1990; Qiu et al. 2004), which defines them as osmoresponsive (Bourque, 2008). Selective ERK activation in these structures in response to hyperosmotic challenge that effectively changed osmolality suggests that phosphoERK could be involved in the detection and integration of osmotic stimuli that governs VP release. Accordingly, ERK was not activated in the thalamic anterodorsal nuclei that are not osmoresponsive. However, ERK activation could also result from osmotically evoked activity changes in osmosensitive neurones, without being involved in the process of osmosensitivity per se. For instance, as TRPV contributes to osmotransduction (Sharif Naeini et al. 2006), ERK could be activated by Ca²⁺ entry after TRPV channel activation, as in cortical neurones (Shirakawa et al. 2008).

The time course of ERK phosphorylation increase followed and then paralleled the hypertonically induced raise in osmolality we measured and that matched the changes in plasma osmotic pressure after i.p. injection of hypertonic saline reported in previous work (Dunn et al. 1973). ERK activation was detectable at early time points, reached a maximum after 10 min, was still present after 30 min and reversed back close to control 2 h after the i.p. injection. In vivo experiments using a protocol for administration of osmotic stimuli similar to ours have shown that the hypertonically induced increase in osmotic pressure is also accompanied by a parallel increase of the firing rate of MNCs (Brimble & Dyball, 1977). This suggests a tight temporal coupling between the transduction and integration of osmotic stimuli and ERK phosphorylation. Here the fast ERK activation is in line with the well-described rapid ERK phosphorylation previously reported as, for instance, in spinal dorsal horn neurones after nociceptive stimulation (Ji et al. 1999). The delayed response observed in ERK activation in supraoptic neurones might be explained by the facts that the OVLT and subfornical organ are circumventricular organs, devoid of blood–brain barrier, while the median preoptic and the paraventricular nuclei are closer to the third ventricle than the supraoptic nucleus: changes in extracellular fluid osmolality could influence these structures faster than the supraoptic nucleus.

We also found a positive relationship between the amplitude of changes in osmolality and the number of phosphoERK-expressing neurones in forebrain osmosensitive structures, as well as the amount of phosphoERK in the supraoptic nucleus. Accordingly, hyperosmolality increased, while hypo-osmolality decreased ERK phosphorylation. Previous in vivo electrophysiological recordings uncovered a linear relationship between osmolality and the firing rate of MNCs (Brimble & Dyball, 1977; Leng et al. 2001). Although this suggests a tight coupling between osmotic transduction and ERK phosphorylation again, it does not indicate whether ERK is activated downstream osmotransduction or whether it is involved in the process.

We found a larger expression of phosphoERK in VP vs. OT neurones in all tested osmolality conditions. Yet previous in vivo electrophysiological recordings showed that OT neurones as well as VP neurones respond in the same way to intravenous infusion of hypertonic saline by increasing their overall firing rate of action potentials (Leng et al. 2001). It is possible to speculate that the quantitative difference in ERK activation seen here might be related to the different firing patterns OT (continuous) and VP (phasic) neurones adopt in response to osmotic challenge (Poulain & Wakerley, 1982). Ca²⁺-binding proteins have been shown to determine firing patterns in MNCs (Li et al. 1995) and it is possible that the buffering of Ca²⁺ also influences ERK activation. That could explain why ERK phosphorylation predominated in VP neurones. The differences might also be related to downstream consequences. For instance, the difference in ERK activation could contribute to the difference in the coupling between excitation and release in OT and VP neurones (Bicknell, 1988).

The inhibitory effects of U 0126 upon osmotically induced forebrain expression of the activity dependent gene product Fos, electrical activity of MNCs and VP release in the bloodstream provide evidence that ERK activation in forebrain neurones contributes to the osmoregulatory mechanisms of VP release. First, U 0126, a selective inhibitor of the ERK upstream kinase, MEK 1/2, (Duncia et al. 1998; Favata et al. 1998) has been shown to efficiently block ERK phosphorylation in vivo at doses similar or one order smaller than the ones we used, with no detectable side effect, in different structures such as the dorsal root ganglia (Chen et al. 2008), neonatal brain (Han & Holtzman, 2000) or hippocampus (Kuroki et al. 2001). Accordingly we found that U 0126 significantly reduced osmotically induced forebrain ERK phosphorylation. Secondly, we show here for the first time that osmotically induced forebrain expression of Fos is dependent upon ERK activation, as U 0126 reduced osmotically induced forebrain Fos expression. Fos, the product of the immediate early gene c-fos, has been widely used as a marker of neuronal activation, in particular to map forebrain osmosensitive neurones (Sharp et al. 1991; Hoffman et al. 1993; Fujihara et al. 2009). Our findings suggest that c-fos is one of the nuclear targets activated by the ERK 1/2 pathway after osmotic stimuli. Previous work has shown that Fos expression in VP cells in response to water deprivation was greater than that in OT cells in the supraoptic and paraventricular nuclei (Fénelon et al. 1994), which is consistent with our current findings. It must be noted however, that in the osmosensitive structures, ERK phosphorylation is not systematically associated with Fos expression. For example, administration of Sar¹,Ile⁴,Ile⁸-AngII, an analog of angiotensin II, activates ERK without any expression of Fos (Daniels et al. 2005). Once expressed after osmotic stimulation, Fos may affect VP gene transcription by binding to the functional AP1 element of the VP gene promoter (Yoshida et al. 2006) and thus, contribute to osmoregulation. Thirdly, U 0126 reduced osmotically induced depolarization and increase in firing of supraoptic MNCs recorded in vitro. Finally, U 0126 also reduced osmotically induced VP release in the bloodstream. Altogether these findings indicate that ERK activation is a necessary functional element in the neuronal circuit integrating osmotic stimuli to produce activity changes in MNCs. This does not preclude other functions for ERK activation resulting from activity changes in osmosensitive neurones.

ERK activation could play a role in the functioning of forebrain osmoregulatory networks through several mechanisms. First, ERK activation could be directly involved in the process of osmotic transduction. Studies performed on acutely isolated cells have indicated that neurones in the OVLT, the subfornical organ as well as MNCs in the supraoptic nucleus can operate as intrinsic osmoreceptors, i.e. they display the intrinsic ability to transduce osmotic perturbations into changes in the rate or pattern of action potential discharge (Oliet & Bourque, 1993; Anderson et al. 2000; Ciura & Bourque, 2006). It has been suggested that the mechanosensitive channels that allow osmosensory transduction might be gated indirectly, through the action of a mechanosensitive enzyme (Bourque, 2008). In such a scenario, increases and decreases in channel activity during hypertonicity and hypotonicity would require rapid and bidirectional changes in the basal activity of the enzyme, a requirement that is compatible with the findings reported here and with the observation that ERK can be osmotically activated in a large variety of eukaryotic cells, from yeast to mammals (De Nadal et al. 2002). Secondly, ERK activation could change the gain of the osmotransduction process, for instance by stimulating actin polymerization (Prager-Khoutorsky & Bourque, 2010). Furthermore, ERK activation could also be triggered by calcium influx through TRPV channels and then provide a positive feedback to increase the gain of these channels.

It must be noted that the reproductive status influences VP release in the rat (Forsling & Peysner, 1988). ERK phosphorylation could be influenced by the oestrus cycle, resulting in changes in VP release, but we did not investigate this question. However, our electrophysiological findings in male rats suggest that ERK activation is a fundamental mechanism contributing to the osmoregulatory mechanisms of VP release even if it is modulated by the cycle.

Finally, whatever the conditions of osmolality, phosphoERK was expressed in neurones only, at least in the supraoptic nucleus where astrocytes have been shown to confer osmosensitivity through the release of taurine. This suggests that ERK activation may not be a necessary element for osmosensitivity in supraoptic astrocytes. It must be noted that others have found increased Fos expression in supraoptic astrocytes following hyperosmotic stimuli (Xiong et al. 2011), which suggests that in these cells, ERK phosphorylation and Fos expression may not be associated systematically.

To conclude, the present work shows that ERK phosphorylation in osmosensitive forebrain regions relates to osmolality changes and contributes to the osmoregulatory mechanisms of VP release.

Key points

• The mechanisms of osmotically induced vasopressin secretion from the hypothalamic magnocellular neurosecretory cells, which is crucial for body fluid homeostasis, are not yet fully understood.

• Extracellular signal-regulated protein kinases (ERK) are mitogen-activated protein kinases that transduce extracellular stimuli into intracellular post-translational and transcriptional responses and might be involved in the regulation of vasopressin release in response to changes in osmolality.

• We found that ERK was dose-dependently activated (phosphorylated) in the rat osmosensitive forebrain regions, including magnocellular neurosecretory cells, by increases in osmolality induced by hypertonic solutions.

• Inhibition of ERK phosphorylation reduced hypertonically induced activation of osmosensitive forebrain neurones and vasopressin release.

• Our results identify ERK activation as a new element contributing to the osmoregulatory mechanisms of vasopressin release.

Glossary

ERK

extracellular signal-regulated protein kinases

MNC

magnocellular neurosecretory cell

OT

oxytocin

OVLT

organum vasculosum lamina terminalis

VP

vasopressin

Additional information

Competing interests

Authors do not have any conflict of interest.

Author contributions

All experiments were performed at Inserm, U862, Neurocentre Magendie, F-33077 Bordeaux, France. Conception and design of the experiments: J.D., D.L.V. Collection, analysis and interpretation of data: J.D., V.R.R.D., V.S.F., P.F., A.A., M.E., J.M.I., S.H.R.O., D.L.V. Drafting the article or revising it critically for important intellectual content: all authors. All authors have approved the final manuscript version to be published.

Funding

This work was supported by Conseil Régional d’Aquitaine (CRA), Inserm, Université de Bordeaux, and studentship from the CRA and Inserm to JD.

Translational perspective

The impact and contribution of changes in hydration status on clinical conditions such as diabetes, heart failure, dehydration and sepsis are critical, especially in very young and old people. More generally, alterations in osmoregulatory mechanisms might play a part in the relation that links high salt intake to hypertension. There is a need therefore to better understand how the brain monitors and regulates body hydration. One key actor of this process is vasopressin, the anti-diuretic hormone, which is secreted from the hypothalamic magnocellular neurosecretory cells into the bloodstream. Here we investigated a possible mechanism involved in the osmoregularory control of vasopressin secretion. We looked at the role of Extracellular signal-Regulated protein Kinases (ERK) activation in the forebrain neurones that are involved in vasopressin neurone activation leading to subsequent vasopressin secretion. ERK are mitogen-activated protein kinases that transduce extracellular stimuli into intracellular post-translational and transcriptional responses and as such, might be involved in the response of osmoregulatory neurones to changes in osmolality. We found that ERK was dose-dependently activated (phosphorylated) in the rat osmosensitive forebrain regions, including magnocellular neurosecretory cells, by increases in osmolality induced by hypertonic solutions. Furthermore, pharmacological inhibition of ERK phosphorylation reduced hypertonically induced activation of osmosensitive forebrain neurones and vasopressin release. Altogether our results identify ERK activation as an important actor in the osmoregulatory mechanisms of vasopressin release. They thus provide a new insight in our understanding of how the brain monitors and regulates body hydration.