Rapid Glucocorticoid-Mediated Endocannabinoid Release and Opposing Regulation of Glutamate and γ-Aminobutyric Acid Inputs to Hypothalamic Magnocellular Neurons

Glucocorticoids secreted in response to stress activation of the hypothalamic-pituitary-adrenal axis feed back onto the brain to rapidly suppress neuroendocrine activation, including oxytocin and vasopressin secretion. Here we show using whole-cell patch clamp recordings that glucocorticoids elicit a rapid, opposing action on synaptic glutamate and γ-aminobutyric acid (GABA) release onto magnocellular neurons of the hypothalamic supraoptic nucleus and paraventricular nucleus, suppressing glutamate release and facilitating GABA release by activating a putative membrane receptor. The glucocorticoid effect on both glutamate and GABA release was blocked by inhibiting postsynaptic G protein activity, suggesting a dependence on postsynaptic G protein signaling and the involvement of a retrograde messenger. Biochemical analysis of hypothalamic slices treated with dexamethasone revealed a glucocorticoid-induced rapid increase in the levels of the endocannabinoids anandamide (AEA) and 2-arachidonoylglycerol (2-AG). The glucocorticoid suppression of glutamate release was blocked by the type I cannabinoid receptor cannabinoid receptor antagonist, AM251, and was mimicked and occluded by AEA and 2-AG, suggesting it was mediated by retrograde endocannabinoid release. The glucocorticoid facilitation of GABA release was also blocked by AM251 but was not mimicked by AEA, 2-AG, or a synthetic cannabinoid, WIN 55,212–2, nor was it blocked by vanilloid or ionotropic glutamate receptor antagonists, suggesting that it was mediated by a retrograde messenger acting at an AM251-sensitive, noncannabinoid/nonvanilloid receptor at presynaptic GABA terminals. The combined, opposing actions of glucocorticoids mediate a rapid inhibition of the magnocellular neuroendocrine cells, which in turn should mediate rapid feedback inhibition of the secretion of oxytocin and vasopressin by glucocorticoids during stress activation of the hypothalamic-pituitary-adrenal axis.

GLUCOCORTICOIDS EXERT RAPID feedback effects on neuroendocrine function that are incompatible with the classical transcriptional mechanisms of activated steroid receptors. Rapid glucocorticoid actions are best characterized by the negative feedback effects of stress-activated glucocorticoids on the hypothalamic-pituitary-adrenal axis (1–3), but fast inhibitory glucocorticoid actions have also been reported in many other hypothalamic neuroendocrine systems (4, 5). Recent work from our laboratory revealed that glucocorticoids elicit a rapid, nongenomic suppression of glutamate release onto parvocellular neuroendocrine cells of the hypothalamic paraventricular nucleus (PVN) by stimulating the retrograde release of an endocannabinoid and subsequent activation of presynaptic type I cannabinoid receptor (CB1) cannabinoid receptors (6). The glucocorticoid-induced endocannabinoid suppression of glutamate synaptic inputs was found in several different peptidergic phenotypes among the neurosecretory parvocellular neurons, including but not limited to the CRH cells, which suggests that this may represent a general mechanism responsible for the rapid glucocorticoid feedback inhibition of neuroendocrine function.

Glucocorticoids also have been reported to have rapid inhibitory effects on neurohypophysial oxytocin and vasopressin secretion from hypothalamic magnocellular neurons. Corticosterone inhibits osmotically stimulated vasopressin secretion from hypothalamic explants (4), and the inhibitory effect appears to be mediated by a membrane corticosteroid receptor (7) and linked to the osmotic activation of glutamatergic synaptic inputs (8). Here we studied the rapid actions of glucocorticoids on magnocellular neuroendocrine cells of the hypothalamic supraoptic nucleus (SON) and PVN to determine whether the fast inhibitory glucocorticoid effect on neurohypophyseal oxytocin and vasopressin secretion is mediated by steroid actions similar to those seen in PVN parvocellular neuroendocrine cells. Similar to our previous findings in parvocellular neurons, we found that glucocorticoids activate a putative membrane G protein-coupled receptor in magnocellular neurons that leads to the suppression of glutamate synaptic inputs via release of a retrograde endocannabinoid messenger. In contrast, we present evidence for the opposite effect of glucocorticoids on synaptic γ-aminobutyric acid (GABA) release in magnocellular neurons, a rapid facilitation, which was not seen in parvocellular neurons (6). The facilitatory effect of glucocorticoids on GABA release was mediated by an unknown retrograde messenger.

Materials and Methods

Slice preparation

Male Sprague Dawley rats (3–5 wk old, Charles River, Wilmington, MA) were used in these experiments according to a protocol approved by the Tulane University Institutional Animal Care and Use Committee. Hypothalamic slices containing the SON and PVN were prepared as described previously (6). Rats were deeply anesthetized with pentobarbital sodium (50 mg/kg body weight) and decapitated. The brain was quickly removed from the cranial cavity after cutting the optic nerves and immersed in a cooled (1–2 C) artificial cerebral spinal fluid (aCSF) bubbled with 100% O₂. The composition of the aCSF was (in millimoles): 140 NaCl, 3 KCl, 1.3 MgSO₄, 1.4 NaH₂PO₄, 2.4 CaCl₂, 11 glucose, and 5 HEPES; pH was adjusted to 7.2–7.3 with NaOH. The hypothalamus was blocked and glued to the chuck of a vibrating microtome (World Precision Instruments, Sarasota, FL). Two coronal hypothalamic slices (350 μm) containing the PVN and/or SON were sectioned and bisected along the midline (i.e. at the third ventricle) and the hemislices were maintained submerged in a holding chamber in oxygenated aCSF at room temperature, in which they were allowed to equilibrate for 1.5–2 h before being transferred to the recording chamber.

Electrophysiological methods

Patch pipettes were pulled from borosilicate glass (1.65 mm outer diameter, 1.2 mm inner diameter; KG33; Garner Glass, Claremont, CA) with a Flaming/Brown P-97 micropipette puller (Sutter Instruments, Novato, CA) to a resistance of 3–4 mΩ. Pipette solutions contained either (in millimoles) 120 K-gluconate, 10 KCl, 1 NaCl, 1 MgCl₂, 1 CaCl₂, 10 EGTA, 2 Mg-ATP, 0.3 Na-GTP, and 10 HEPES or 110 d-gluconic acid, 110 CsOH, 10 CsCl, 1 MgCl₂, 1 CaCl₂, 10 EGTA, 2 Mg-ATP, 0.3 Na-GTP, and 10 HEPES. pH was adjusted to 7.3 with KOH or CsOH. The osmolarity of the solution was adjusted to 300 mOsmol with 20 mmd-sorbitol.

Hemislices were transferred one at a time from the holding chamber to a submerged recording chamber, in which they were secured to the floor of the chamber with platinum wire and perfused with aCSF at a rate of 2 ml/min. Most recordings were conducted at a temperature of 32–34 C, whereas some early recordings of miniature excitatory postsynaptic currents (mEPSCs) were conducted at room temperature, as indicated in the text. The slices were allowed to equilibrate in the recording chamber for at least 15 min before starting recordings. Magnocellular neurons were then visualized via a cooled CCD camera using infrared illumination and differential interference contrast optics (IR-DIC). Upon obtaining the whole-cell recording configuration, series resistance and whole cell capacitance were adjusted and were continually monitored throughout experiments. Cells with unstable input resistance or series resistance were discarded. Only one cell was recorded from each hemislice to avoid sampling bias. All recordings were performed in voltage clamp mode using an Axopatch 1-D or Multiclamp 700 amplifier (Axon Instruments, Foster City, CA) and were monitored continuously on a digital storage oscilloscope (Hitachi, Tokyo, Japan). Data were low pass filtered at 2 kHz, converted to digital video format at 22 kHz using a Neuro-Corder (NeuroData Instruments, New York, NY) and stored on videotape for off-line analysis. Selected data were subsequently digitized at 4 kHz and recorded on a personal computer using the Digidata 1200 interface and pCLAMP 7.0 or 9.0 software (Axon Instruments).

Whole-cell patch-clamp recordings were performed in putative magnocellular neurons in the SON and PVN, which were identified on the basis of their location, their relatively large size under IR-DIC visualization and their electrical properties (9, 10). External solutions contained tetrodotoxin (TTX, 1 μm) to block spike-mediated transmitter release, and magnocellular neurons were held at a holding potential of −60 mV to record mEPSCs or 0 mV (with the cesium-containing patch solution) to record miniature inhibitory postsynaptic currents (mIPSCs). To focus on the fast, nongenomic actions of glucocorticoids, 3-min epochs of mEPSCs or mIPSCs were collected at 7–10 min of bath application of glucocorticoid and compared with 3-min control epochs acquired just before the glucocorticoid application. The traces obtained were analyzed for mean frequency, amplitude, and decay time (defined as the time from peak to 63% decay of the mEPSC/mIPSC) using the Minianalysis 5.0 program (Synaptosoft Inc., Decatur, GA).

Drug application

Water-soluble forms of the steroids dexamethasone (DEX; 0.01–1000 μm), corticosterone (1 μm), and cholesterol (5 μm) (Sigma-Aldrich, St. Louis, MO) were dissolved directly in aCSF to their final concentrations and applied in the perfusion bath. The DEX-BSA conjugate (Steraloids Inc., Newport, RI) was dissolved in the aCSF with 25% β-cyclodextrin (Sigma-Aldrich) as a carrier to increase its solubility. The concentration of DEX-BSA (10 μm) was selected to obtain an effective concentration of dexamethasone of about 1 μm because the BSA conjugation reduces the effectiveness of receptor binding by a factor of about 10. TTX (1 μm, Sigma-Aldrich), 6,7-dinitroquinoxaline-2,3-dione (DNQX, 20 μm),d-(-)-2-amino-5-phosphonopentanoic acid (AP-5, 50 μm), LY341495 (100 μm), bicuculline methiodide (30 μm), and CGP 54626 (3 μm) (Tocris Bioscience, Ellisville, MO) were dissolved in sterile water and stored in stock solutions at −20 C. The physiologically inactive steroid, isopregnanalone (5 μm), the intracellular types I and II corticosteroid receptor antagonists, respectively, spironolactone (10 μm) and mifepristone (RU486, 10 μm) (Sigma-Aldrich), and the cannabinoid reagents WIN 55,212–2 (1 μm) and AM 251 (1 μm) (Tocris) were stored as 100-mm stock solutions in dimethylsulfoxide at −20 C and were dissolved to their final concentrations in aCSF before bath application. The endogenous cannabinoids, anandamide (AEA, 0.1–1 μm) and 2-arachidonoylglycerol (2-AG, 0.5–2 μm) (Tocris), were prepared and applied under low-light conditions due to their photosensitivity. The dimethylsulfoxide, β-cyclodextrin, and the standard plain emulsion (Tocris) used to dissolve AEA were tested and had no effect on synaptic currents at the concentrations used. The nonhydrolyzable guanylyl nucleotide GDP-β-S (500 μm; Sigma-Aldrich) and, in some cases, DEX (1 μm) were included in the patch solution for intracellular application. To prevent drugs applied intracellularly from leaking from the patch pipette into the extracellular space before seal formation, the tips of the patch pipettes were filled with regular patch solution and the pipettes were then backfilled with the drug solution.

Single-cell RT-PCR

The single-cell RT-PCR procedure was similar to that described previously (6). The cell cytoplasm was aspirated into the patch pipette under visual control, taking care to avoid aspiration of the nucleus, and transferred into 5 μl of lysis buffer containing 2.9 μl diethyl pyrocarbonate-treated water, 0.7 μl BSA, 0.7 μl Oligo dT (0.5 μg/μl), and 0.7 μl RNasin (40 U/μl), which was maintained on ice prior and subsequent to receiving harvested cytoplasm. The cell lysates were stored at −80 C or used immediately for reverse transcription (RT). For the RT, the lysate-buffer mixture was heated to 70 C for 10 min and then incubated on ice for 1 min, after which 15 μl of final RT master mix, including 8.0 μl diethyl pyrocarbonate-treated water, 2 μl 10× first-strand buffer, 2 μl MgCl₂ (25 mm), 2 μl dithiothreitol (0.1 m), and 1 μl mixed deoxynucleotide triphosphates (10 mm), were added. Single-stranded cDNA was reverse transcribed from the cellular mRNA by adding SuperScript II reverse transcriptase (0.7 μl, 200 U/μl) and incubating at 42 C for 50 min, after which the reaction was terminated by heating to 70 C for 15 min. The RNA strand in the RNA-DNA hybrid was removed by adding 0.5 μl RNase H (2 U/μl) for 20 min at 37 C. All reagents were obtained from Life Technologies (Madison, WI).

PCR amplification of the reverse-transcribed mRNAs was performed in a thermal cycler (MJ Research, Watertown, MA) using a fraction of the single-cell cDNA as a template. Reaction mixtures contained (in 30 μl final volume): 2.2 mm MgCl₂, 0.5 mm of each of the deoxynucleotide triphosphates, 1.5 μm of primers, 2.5 U Taq DNA polymerase, 2.9 μl of 10× buffer, and 4 μl of the cDNA template made from the single-cell RT reaction. The thermal cycler was set at 94 C for 45 sec, 60 C for 45 sec, and 72 C for 1.1 min for 50 cycles. Primers for oxytocin, vasopressin, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were taken from Glasgow et al. (11) and synthesized by Integrated DNA Technologies (Coralville, IA). The primer pairs used were: oxytocin, 5′-GAC GGT GGA TCT CGG ACT GAA-3′ and 5′-CGC CCC TAA AGG TAT CAT CAC AAA-3′ (462 bp); vasopressin, 5′-CCT CAC CTC TGC CTG CTA CTT-3′ and 5′-GGG GGC GAT GGC TCA GTA GAC-3′ (440 bp); and GAPDH, 5′-GGA CAT TGT TGC CAT CAA CGA C-3′ and 5′-ATG AGC CCT TCC ACG ATG CCA AAG-3′ (441 bp). Each of the primers was used in every cell analyzed. Final PCR products were visualized by staining with ethidium bromide after separation by electrophoresis in 1.5–2% agarose gels. The forward and reverse GAPDH primers were designed to span an intron to detect genomic DNA in each cellular template. In addition to the single-cell GAPDH RT-PCR, whole hypothalamic slices were run in parallel with the single-cell reactions as positive controls for the primers. Negative controls for contamination from extraneous and genomic DNA were also run for every batch of neurons by running pure water and omitting the RT step from the reaction, respectively. None of the cells tested showed a genomic DNA PCR product.

Quantitative biochemical analysis of endocannabinoids

Hypothalamic slices containing the SON were prepared identically to those described above for whole-cell recordings, except that they were trimmed to restrict biochemical analysis to the SON and surrounding tissue. Thus, slices were bisected along the midline (i.e. at the third ventricle) and the hemislices were trimmed dorsally 2 mm dorsal to ventral surface and laterally 1 mm lateral to the optic chiasm. Two hemislices were subjected to drug treatment and their contralateral pairs were used as controls in each rat. After equilibration in an oxygenated holding chamber for 1.5–2 h, hemislices were secured to the floor of a perfusion chamber with platinum wire and perfused with oxygenated, heated aCSF (32–34 C). To measure DEX-induced endocannabinoid production, hemislices were perfused with DEX (1 μm) for 10 min. Control hemislices were perfused with aCSF for 10 min. Immediately after DEX or control treatment, slices were collected and homogenized in 1 ml of ice-cold methanol in preparation for liquid chromatography-tandem mass spectrometry (LC-MS-MS) analysis.

LC-MS-MS analysis of the endocannabinoids AEA and 2-AG was performed on chloroform methanol (2:1) lipid extracts, which were loaded with deuterated standard mixture (AEA-d8, 2-AG-d8) and purified by SPE extraction on C18 columns (Varian, Walnut Creek, CA). Samples were eluted with 10 ml of 1% methanol in ethyl acetate and concentrated on an N2 stream evaporator before LC-MS-MS analysis. Samples were loaded on a Biobasic-AX column (100 mm × 2.1 mm, 5 μm particle size; Thermo-Hipersyl-Keystone, Bellefonte, PA). The column was run with a 45-min gradient protocol starting with solvent solution A [40:60:0.01 methanol/water/acetic acid (pH 4.5)] at a flow rate of 300 μl/ min, reached 100% of solvent B (99.99:0.01 methanol/acetic acid) in 30 min, and run isocratic for 5 min, after which the system returned to 100% solvent solution A in 10 min. LC effluents were diverted to an electrospray-ionization probe on a TSQ Quantum triple quadrupole mass spectrometer (Thermo-Finnigan, San Jose, CA) running on negative ion detection mode. Electrospray voltage was 3 kV; sheath gas was argon at 1.5 mTorr. The instrument runs on full-scan mode to detect MS2-spectra and selected reaction mode for quantitative analysis to detect parent/daughter ion pairs simultaneously. The selected parent/daughter ion pairs were 346.3/259.3, 377.2/285.2, 353.2/266.3, and 385.4/310.2 m/z for AEA, 2-AG, AEA-d8, and 2-AG-d8, respectively.

Data analysis

All data are expressed as means ± se. Statistical comparisons of electrophysiological data were performed using the Student’s paired t test for within-group cell comparisons, the Student’s unpaired t test for between-group comparisons. For comparison of LC-MS-MS data, two slices containing SON from each rat were bisected down the midline, and hemislices from one side of the hypothalamus served as the control and the other as the experimental sample; n’s reflect numbers of animals tested. These were compared statistically using the Student’s paired t test. P <0.05 was considered significant for all comparisons.

Results

To characterize the rapid membrane effects of glucocorticoids on magnocellular neurons, we performed whole-cell patch-clamp recordings in neurons in the SON and PVN in acutely prepared hypothalamic slices. Magnocellular neurons were identified visually by their relatively large somatic size and position in the SON and posterolateral PVN under IR-DIC and electrophysiologically by their prominent transient outward rectification mediated by a large A-type potassium current (9, 10, 12). A total of 204 putative magnocellular neurons were recorded in acute hypothalamic slices from 119 rats. Miniature EPSCs were recorded as inward synaptic currents in the presence of TTX (1 μm) at a holding potential of −60 mV and were blocked by the ionotropic glutamate receptor antagonists AP-5 (50 μm) and DNQX (20 μm) (n = 5) but not by the GABA_A receptor antagonist bicuculline methiodide (30 μm) (n = 6). mIPSCs were recorded with cesium-containing electrodes as outward synaptic currents in TTX (1 μm) at a holding potential of 0 mV and were blocked by the GABA_A receptor antagonist bicuculline methiodide (30 μm) (n = 4) but not by AP-5 (50 μm) and DNQX (20 μm) (n = 3). Data were collected only after a 10-min baseline recording period was established during which a stable amplitude and frequency of synaptic currents was observed.

Rapid glucocorticoid inhibition of glutamate release

The glucocorticoids DEX and corticosterone applied in the bath perfusion suppressed glutamatergic synaptic activity by more than 10% in 18 of 21 PVN magnocellular neurons tested and 14 of 16 SON magnocellular neurons tested. Whereas most cells were recorded at a temperature of 32–34 C, a few neurons at an early stage of the study were recorded at room temperature. There was no significant difference in the response to glucocorticoids observed under the two recording conditions; therefore, the data were pooled for further analysis. Bath application of the synthetic glucocorticoid agonist DEX (1 μm) for 10 min caused a 37.6 ± 6.5 and 36.5 ± 6.2% decrease in the frequency of mEPSCs in PVN and SON magnocellular neurons, respectively (PVN: from 3.4 ± 0.7 to 1.9 ± 0.4 Hz; P < 0.01; n = 21; SON: from 2.1 ± 0.4 to 1.1 ± 0.2 Hz; P < 0.01; n = 16) but had no effect on mEPSC amplitudes (PVN: 26.6 ± 2.0 vs. 26.0 ± 2.4 pA; P = 0.55; SON: 25.6 ± 1.6 vs. 27.1 ± 2.0 pA; P = 0.29) or decay times (PVN: 3.0 ± 0.2 vs. 3.1 ± 0.2 msec, P = 0.27; SON: 2.9 ± 0.2 vs. 3.0 ± 0.2 msec, P = 0.44) (Fig. 1, A–D). The results obtained from PVN and SON magnocellular neurons were not significantly different (mEPSC frequency, P = 0.91; amplitude, P = 0.12; and decay time, P = 0.25), so the data from the two groups were pooled for all further analyses. The inhibitory effect of DEX on mEPSC frequency was dose dependent, with a threshold concentration between 10 and 100 nm and a saturating concentration of 1–10 μm (EC₅₀ = 474 nm) (Fig. 1E). The endogenous glucocorticoid, corticosterone (1 μm), had an effect of approximately the same magnitude as DEX on mEPSCs, causing a 27.7 ± 4.9% decrease in mEPSC frequency (from 1.0 ± 0.3 to 0.7 ± 0.2 Hz; P < 0.05; n = 5) (Fig. 1F) without affecting either mEPSC amplitude (45.4 ± 11.1 vs. 44.1 ± 10.5 pA; P = 0.18) or decay time (3.1 ± 0.9 vs. 3.0 ± 0.8 msec; P = 0.18). The effect of the glucocorticoids was steroid specific because the steroid precursor, cholesterol (5 μm), and the physiologically inactive steroid, isopregnanolone (5 μm), had no effect on mEPSC frequency (Fig. 1F). The glucocorticoid effect did not reverse within 60 min of washout of the steroid. These data indicated that glucocorticoids exert a fast inhibitory effect on glutamate release onto hypothalamic magnocellular neurons.

Fig. 1.

Glucocorticoid-induced suppression of glutamate release onto magnocellular neurons. A, Bath application of DEX (1 μm) elicited a reduction in the frequency of mEPSCs. B, Cumulative frequency plots showed a significant reduction in mEPSC frequency, with no change in mEPSC amplitude. C, Time course of DEX effect on mEPSC frequency from the same cell (mean ± se, 30-sec bins). D, Average change in mean mEPSC frequency, amplitude, and decay time induced by 1 μm DEX. E, Dose dependence of the DEX-induced decrease in mEPSC frequency. Percent reduction in mEPSC frequency is plotted against increasing concentrations of DEX with a computer-fitted curve. F, Corticosterone (1 μm) caused a decrease in mEPSC frequency, whereas the steroid precursor, cholesterol (5 μm), and the physiologically inactive steroid, isopregnanolone (5 μm), had no effect on mEPSC frequency. Numbers in parentheses represent numbers of cells analyzed in each condition in this and the following figures. *, P < 0.05; **, P < 0.01.

Rapid glucocorticoid facilitation of GABA release

Glucocorticoids increased GABAergic synaptic activity by more than 10% in six of six PVN magnocellular neurons and eight of nine SON magnocellular neurons tested. Bath application of DEX (1 μm) for 10 min caused a 36.3 ± 5.9 and 31.1 ± 6.6% increase in the frequency of GABA-mediated mIPSCs in PVN and SON magnocellular neurons, respectively (PVN: from 1.0 ± 0.3 to 1.3 ± 0.4 Hz; P < 0.01; n = 6; SON: from 1.6 ± 0.3 to 2.0 ± 0.4 Hz; P < 0.01; n = 9) but had no effect on mIPSC amplitudes (PVN: 39.2 ± 2.1 vs. 40.5 ± 2.0 pA; P = 0.50; SON: 45.3 ± 4.1 vs. 46.1 ± 3.7 pA; P = 0.50) or decay times (PVN: 6.9 ± 1.8 vs. 6.9 ± 1.7 msec; P = 0.81; SON: 6.7 ± 2.8 vs. 6.5 ± 2.1 msec; P = 0.31) (Fig. 2, A–D). The results obtained from PVN and SON magnocellular neurons were not significantly different (mIPSC frequency, P = 0.59; amplitude, P = 0.73; decay time, P = 0.33), so the data from each group were pooled for further analyses. The excitatory effect of dexamethasone on mIPSC frequency showed an inverted U-shaped dose-response curve (Fig. 2E), such that DEX caused an increase in mIPSC frequency at concentrations between 1 and10 μm but had no effect at both low (0.01–0.1 μm) and high (100–1000 μm) concentrations. Corticosterone (1 μm) had an effect of approximately the same magnitude as DEX and caused a similar 31.9 ± 6.1% increase in mIPSC frequency (from 0.9 ± 0.4 to 1.1 ± 0.4 Hz; P < 0.01; n = 5) without affecting either the mIPSC amplitude (47.2 ± 8.8 vs. 50.2 ± 6.1 pA; P = 0.26) or decay time (7.4 ± 0.4 vs. 7.5 ± 0.5 msec; P = 0.39) (Fig. 2D). The glucocorticoid effect did not reverse within 60 min of washout of the steroid. These data indicated that glucocorticoids have a fast facilitatory effect on GABA release onto magnocellular neurons.

Fig. 2.

Glucocorticoid-induced facilitation of GABA release onto magnocellular neurons. A, Bath application of DEX (1 μm) elicited an increase in the frequency of mIPSCs. B, Cumulative frequency plots of mIPSCs from the same cell showed a significant shift toward shorter interevent intervals or higher mIPSC frequencies (left), with no change in the mIPSC amplitude distribution (right). C, Time course of the DEX effect on mIPSC frequency from the same cell (mean ± se, 30-sec bins). D, DEX and corticosterone (CORT) induced mean changes in the average mIPSC frequency but not amplitude or decay time. E, Inverted U-shaped dose-response curve of DEX effect on mIPSC frequency. *, P < 0.05; **, P < 0.01.

Glucocorticoid actions are mediated by a membrane receptor

The rapid onset of the glucocorticoid effect (∼3 min) on both mEPSCs and mIPSCs suggested a nongenomic mechanism of steroid action. We conducted, therefore, a series of experiments to determine whether the glucocorticoid effects on glutamate and GABA release were mediated by activation of the classical intracellular corticosteroid receptors (Fig. 3, A and B). Bath application of the membrane-impermeant conjugate DEX-BSA (10 μm) retained the steroid effects on glutamate and GABA release, decreasing the mEPSC frequency by 34% (from 1.9 ± 0.7 to 1.3 ± 0.5 Hz; P < 0.05; n = 4) and increasing the mIPSC frequency by 25% (from 1.3 ± 4 to 1.6 ± 0.5 Hz; P < 0.05; n = 5). DEX (1 μm) applied intracellularly by including it in the patch pipette had no effect on the frequency of mEPSCs (101.5 ± 6.5% of baseline, n = 5). Antagonists of the types I and II corticosteroid receptors did not block the DEX effect. Thus, in the presence of the type II receptor antagonist, mifepristone (RU486, 10 μm), DEX (1 μm) caused a 47% decrease in the frequency of mEPSCs (from 1.8 ± 0.2 to 1.0 ± 0.4 Hz; P < 0.01; n = 7) and a 29% increase in mIPSCs (from 1.7 ± 0.5 to 2.1 ± 0.6 Hz; P < 0.05; n = 5). In the presence of the type I receptor antagonist, spironolactone (10 μm), DEX (1 μm) caused a 24.2% decrease in the frequency of mEPSCs (from 1.2 ± 0.4 to 0.8 ± 0.2 Hz; P < 0.01; n = 7). The effect of spironolactone on mIPSCs was not tested. These data indicated, therefore, that the glucocorticoid actions to suppress glutamate release and facilitate GABA release were not mediated by activation of the classical intracellular corticosteroid receptors and suggested mechanisms involving a membrane-associated receptor.

Fig. 3.

Rapid glucocorticoid effects are mediated by activation of a nonclassical corticosteroid receptor. A, Bath application of membrane-impermeant DEX-BSA (10 μm) maintained the inhibitory effect, whereas intracellular application of DEX (1 μm) had no effect on the frequency of mEPSCs. The classical glucocorticoid and mineralocorticoid receptor antagonists RU 38486 (10 μm) and spironolactone (10 μm), respectively, failed to block the inhibitory effect of DEX on mEPSC frequency. B, The DEX-induced increase in mIPSC frequency was maintained with DEX-BSA, and RU 38486 (10 μm) failed to block the DEX effect. C, The DEX effect on mEPSCs was not blocked by either the mGluR antagonist LY 34149 (100 μm) or the GABA_B receptor antagonist CGP 54626 (3 μm). *, P < 0.05; **, P < 0.01.

Because the activation of presynaptic group III metabotropic glutamate receptors (mGluRs) and GABA_B receptors leads to a suppression of glutamate release similar to that seen here with glucocorticoids (13, 14), we tested whether the glucocorticoid actions are dependent on the activation of presynaptic mGluRs or GABA_B receptors. Blockade of mGluRs, with a subtype nonselective concentration of LY 341495 (100 μm) and GABA_B receptors, with CGP 54626 (3 μm), failed to block the DEX-induced suppression of glutamate release onto magnocellular neurons. Thus, DEX reduced the frequency of mEPSCs by 33.5% in the presence of LY 34149 (n = 5) and 28.7% in the presence of CGP 54626 (n = 6) (Fig. 3C).

We found previously that the rapid glucocorticoid effects on glutamate release onto PVN parvocellular neurons are mediated by a postsynaptic G protein-dependent signaling mechanism (6), so we tested for a similar glucocorticoid signaling mechanism in magnocellular neurons. Intracellular application in magnocellular neurons of the G protein antagonist, GDP-β-S (0.5 mm), via the patch pipette blocked completely both the DEX-induced decrease in mEPSC frequency (96.5 ± 6.8% of baseline; P = 0.16; n = 5) and increase in mIPSC frequency (99.0 ± 10.3% of baseline; P = 0.38; n = 5) (Fig. 4). Because the GDP-β-S effect was restricted to the postsynaptic cell, this indicated that the glucocorticoid effect was G protein dependent and that its site of action was postsynaptic on the magnocellular neuronal membrane. This suggested, therefore, that the glucocorticoid effects on both glutamate and GABA release are mediated by the actions of one or more retrograde messengers.

Fig. 4.

Postsynaptic G protein dependence of rapid glucocorticoid effects. Blockade of postsynaptic G protein activity with intracellular GDP-β-S application (0.5 mm) prevented the DEX-induced reduction in mEPSC frequency and increase in mIPSC frequency. *, P < 0.05; **, P < 0.01.

Glucocorticoid-induced endocannabinoid release

Trimmed hemislices containing the SON (see Materials and Methods) were assayed for changes in endocannabinoid content in response to glucocorticoid using LC-MS-MS analysis. Both the endocannabinoids, AEA and 2-AG, were found to be increased in rapid response to glucocorticoids applied to hypothalamic slices. Bath application of DEX (1 μm) for 10 min caused an average increase in AEA levels of 290.7 ± 120.5% (from 7.0 ± 1.7 to 19.1 ± 3.6 pmol/mg protein; P < 0.05; n = 5) and an average increase in 2-AG levels of 271.8 ± 77.4% (from 2044.8 ± 375.4 to 6828.5 ± 1411.2 pmol/mg protein; P < 0.05; n = 5) (Fig. 5).

Fig. 5.

Glucocorticoid-induced increase in endocannabinoid levels measured with LC-MS-MS. A 10-min application of DEX (1 μm) caused an increase in the levels of the endocannabinoids AEA and 2-AG in trimmed hypothalamic slices. *, P < 0.05.

Glucocorticoid suppression of excitatory input via retrograde endocannabinoid release

Our previous study demonstrated that the glucocorticoid-induced suppression of glutamate release onto PVN parvocellular neuroendocrine cells is mediated by retrograde endocannabinoid release (6). We therefore tested in the current study whether the rapid inhibitory effect of glucocorticoids on mEPSCs in magnocellular neurons in the SON and PVN is mediated by a retrograde endocannabinoid messenger. Bath application of the type I cannabinoid receptor (CB1) antagonist, AM 251 (1 μm), alone did not have a significant effect on mEPSCs (frequency 88.2 ± 6.0% of baseline; P = 0.15; n = 4) but prevented the DEX-induced reduction in mEPSC frequency (96.9 ± 10.4% of AM 251 value, or 86.1 ± 8.6% of baseline; P = 0.22) (Fig. 6, A, B, and F). The high-affinity synthetic cannabinoid agonist, WIN 55,212–2, and the endocannabinoids, AEA and 2-AG, mimicked the inhibitory effect of glucocorticoids on glutamate release. Thus, WIN 55,212–2 (1 μm) reduced the mEPSC frequency by 35.8 ± 3.2% (P < 0.01; n = 8) without affecting either mEPSC amplitude (99.4 ± 3.5%) or decay time (107.1 ± 4.5%) (Fig. 5, C–F); AEA (1 μm) reduced the mEPSC frequency by 37.5 ± 4.8% (P < 0.01; n = 4) without affecting either mEPSC amplitude (102.2 ± 4.8%) or decay time (95.2 ± 7.6%), and 2-AG (1 μm) reduced mEPSC frequency by 39.6 ± 3.8% (P < 0.01; n = 6) without affecting either mEPSC amplitude (99.7 ± 10.5%) or decay time (104.5 ± 6.1%) (Fig. 6F). Therefore, the rapid decrease in mEPSC frequency caused by glucocorticoids was blocked by a selective CB1 cannabinoid receptor antagonist and mimicked by three different CB1 receptor agonists. This suggests that the glucocorticoid-induced increase in endocannabinoid levels observed in our LC-MS-MS analysis is responsible for the glucocorticoid effect on glutamate release and, therefore, that the retrograde messenger(s) that mediates this effect is an endocannabinoid. The mechanism of rapid glucocorticoid suppression of glutamate release in magnocellular neurons is similar, therefore, to that reported previously in parvocellular neurons (6).

Fig. 6.

The glucocorticoid-induced suppression of glutamate release is mediated by a retrograde endocannabinoid messenger. A, The CB1 receptor antagonist AM251 (1 μm) blocked the 1 μm DEX-induced decrease of mEPSC frequency. B, Cumulative frequency plots of data from the same cell showed no significant change in mEPSC interevent interval and amplitude distributions. C, The synthetic CB1 receptor agonist, WIN55,212–2, mimicked the glucocorticoid effect, causing a selective reduction in mEPSC frequency. D, Cumulative frequency plots of data from the same cell showed a significant reduction in mEPSC interevent intervals, with no change in the mEPSC amplitude distribution. E, Average changes in mean mEPSC frequency, amplitude, and decay time in response to WIN 55,212–2. F, Average changes in mean mEPSC frequency in the presence of DEX (1 μm), WIN55,212–2 (1 μm), AEA, 2-AG, AM251 (1 μm), and AM 251 + DEX. *, P < 0.05; **, P < 0.01.

Glucocorticoid facilitation of inhibitory input via an unknown retrograde messenger

We also tested for a role of the retrograde release of endocannabinoids in the presynaptic facilitation of mIPSCs by glucocorticoids. Bath application of the CB1 antagonist AM251 (1 μm) had no significant effect by itself on mIPSC frequency (−9.9 ± 3.2% of baseline; P = 0.08; n = 9) but completely blocked the DEX-induced increase in mIPSC frequency (+4.7 ± 2.9% vs. AM251 value, or +7.0 ± 4.3% vs. baseline; P = 0.19) (Fig. 7, A, B, and F). However, the cannabinoid receptor agonist, WIN 55,212–2 (1 μm, n = 7), had no consistent effect on mIPSC frequency (+4.3 ± 10.1%), and the endocannabinoids 2-AG and AEA had the opposite effect on mIPSC frequency when compared with DEX. Thus, 2-AG caused a dose-dependent reduction in mIPSC frequency of 10–23% at the doses tested (0.5 μm: from 1.1 ± 0.2 to 1.0 ± 0.1 Hz, P < 0.05, n = 6; 1 μm: from 2.5 ± 0.5 to 2.1 ± 0.5 Hz, P < 0.05, n = 9; 2 μm: from 1.8 ± 0.6 to 1.4 ± 0.6 Hz, P < 0.05, n = 3) (Fig. 7, C–E); AEA also caused a dose-dependent reduction in mIPSC frequency of 14–23% (0.1 μm: from 1.6 ± 0.5 to 1.6 ± 0.5 Hz, P = 0.25, n = 3; 0.5 μm: from 1.3 ± 0.2 to 1.1 ± 0.2 Hz, P < 0.05, n = 5; 1 μm: from 1.7 ± 0.6 to 1.3 ± 0.5 Hz, P < 0.05, n = 4) (Fig. 7E). We also tested for a possible combined presynaptic effect of glucocorticoids and endocannabinoids on GABA release by applying DEX (1 μm) and 2-AG (1 μm) together in the perfusion bath while blocking the postsynaptic glucocorticoid-induced release of endogenous cannabinoids with intracellular infusion of GDP-β-S (0.5 mm) via the patch pipette. The 2-AG effect on GABA release in the presence of DEX was similar to the 2-AG effect alone, causing a decrease in mIPSC frequency of 19.8 ± 3.9% (from 1.6 ± 0.6 to 1.3 ± 0.5 Hz; P < 0.05; n = 4).

Fig. 7.

The glucocorticoid-induced facilitation of GABA release is mediated by an unknown retrograde messenger. A, The CB1 antagonist AM251 (1 μm) blocked the DEX-induced increase in mIPSC frequency. B, Cumulative frequency plots of data from the same cell showed no significant change in mIPSC frequency or amplitude in response to DEX in the presence of AM251. C, Bath application of the endocannabinoid 2-AG (1 μm) had the opposite effect on GABA release, compared with that of DEX, causing a decrease in mIPSC frequency. D, Cumulative frequency plots of the data from the same cell showed a significant increase in mIPSC interevent interval or decrease in frequency, with no change in mIPSC amplitude distribution in response to 2-AG. E, DEX (1 μm) caused an average increase in mIPSC frequency, whereas the endocannabinoids AEA and 2-AG elicited the opposite effect, a dose-dependent decrease in mIPSC frequency. F, Average effects of the CB1 antagonist AM 251 (1 μm) and the vanilloid receptor antagonist capsazepine (10 μm) on the DEX-induced increase in mIPSC frequency. G, Average changes in mIPSC frequency in response to DEX (1 μm) in the presence of ionotropic glutamate receptor antagonists (DNQX/AP5 + DEX) and ionotropic and metabotropic glutamate receptor antagonists (+ LY341495 + DEX). *, P < 0.05; **, P < 0.01.

Anandamide serves as an endogenous agonist at presynaptic vanilloid receptors, in which it often leads to facilitation of neurotransmitter release (15). We tested for the possibility that glucocorticoid facilitation of GABA release might be due to retrograde endocannabinoid actions at presynaptic vanilloid receptors with the VR1 vanilloid receptor antagonist, capsazepine. Preapplication of capsazepine (10 μm) in the bath failed to block the glucocorticoid effect on GABA release, DEX causing a 20.0 ± 3.9% increase in mIPSC frequency in capsazepine (from 1.0 ± 0.2 to 1.2 ± 0.3 Hz; P < 0.01; n = 6) (Fig. 7F). Therefore, the glucocorticoid facilitation of GABA release was not mediated by endocannabinoid actions at presynaptic vanilloid receptors.

We also determined whether the facilitatory effect of glucocorticoids on GABA release onto magnocellular neurons was caused by the suppressive effect of glucocorticoid-induced endocannabinoids on glutamate release from upstream axo-axonic glutamate synapses rather than by a direct facilitatory effect of a retrograde messenger at presynaptic GABA axon terminals. We did this by testing for a role of ionotropic and metabotropic glutamate receptors in the glucocorticoid facilitation of GABA release. Blockade of ionotropic glutamate receptors with DNQX (20 μm) and AP-5 (50 μm) did not prevent the DEX-induced enhancement of GABA release onto magnocellular neurons. Thus, mIPSC frequency was increased by 20.1% (from 0.8 ± 0.1 to 1.0 ± 0.1 Hz; P < 0.05; n = 4) in the presence of DNQX and AP-5 (Fig. 7G). Similarly, blockade of both ionotropic and metabotropic glutamate receptors did not block the glucocorticoid effect on GABA release. DEX caused a 23.4% increase in the frequency of mIPSCs (from 0.7 ± 0.2 to 0.9 ± 0.3 Hz; P < 0.01; n = 5) in the presence of DNQX (20 μm), AP-5 (50 μm), and the mGluR antagonist LY341495 (100 μm) (Fig. 7G). Therefore, the glucocorticoid-mediated increase in mIPSC frequency was blocked by the CB1 receptor antagonist AM251 but was not mimicked by CB1 receptor agonists or blocked by either a vanilloid receptor or a battery of pre- and postsynaptic glutamate receptor antagonists, suggesting that the glucocorticoid-induced retrograde messenger facilitates GABA release by acting at an AM251-sensitive, noncannabinoid receptor on presynaptic GABA axon terminals.

Both oxytocin and vasopressin neurons respond to glucocorticoids

In several of the magnocellular neurons in which the glucocorticoid suppression of mEPSCs was recorded, the expression of oxytocin and vasopressin mRNA was tested after experiments using single-cell RT-PCR. Of 30 magnocellular neurons showing glucocorticoid-mediated inhibition of glutamate release and tested for the expression of both oxytocin and vasopressin, 11 expressed oxytocin mRNA and 10 expressed vasopressin mRNA (Fig. 8); five of these cells coexpressed both oxytocin and vasopressin mRNA. The effect of DEX on mEPSC frequency did not differ between oxytocin- and vasopressin-expressing neurons. The mEPSC frequency decreased by 33.9 ± 3.5% in the neurons that expressed exclusively oxytocin mRNA (n = 6) and 30.9 ± 4.0% in the neurons that expressed exclusively vasopressin mRNA (n = 5). These findings indicate that the glucocorticoid-induced suppression of glutamate synaptic inputs occurs in both oxytocinergic and vasopressinergic magnocellular neurons, which suggests a generalized rapid inhibitory feedback role of glucocorticoids in the regulation of magnocellular peptide hormone secretion.

Fig. 8.

Single-cell RT-PCR analysis of peptide expression in SON magnocellular neurons. Expression of oxytocin (OT) or vasopressin (VP) mRNA in four individual cells subjected to RT-PCR analysis after recordings; each cell responded to DEX with a significant decrease in mEPSCs. Cell identification numbers are shown above each lane. Both OT- and VP-expressing magnocellular neurons responded to glucocorticoids. Left lane, DNA ladder.

Discussion

We reported in a previous study (6) that glucocorticoids cause a rapid suppression of synaptic glutamate inputs to parvocellular neuroendocrine cells of the PVN by activating a membrane glucocorticoid receptor and stimulating the release of a retrograde endocannabinoid messenger. Here we show that glucocorticoids also cause a rapid suppression of glutamate synaptic inputs to oxytocinergic and vasopressinergic magnocellular neurons of the SON and PVN by activating a putative membrane glucocorticoid receptor that stimulates a G protein-dependent synthesis and retrograde release of an endocannabinoid messenger. It was recently shown that magnocellular neuroendocrine cells of the SON respond to stimulation with an activity-dependent release of endocannabinoids (16, 17), indicating that magnocellular neurons synthesize and release endocannabinoids under certain conditions. Our study corroborates this finding with the demonstration of the production of the endocannabinoids AEA and 2-AG in the SON in response to dexamethasone. Several studies have shown that the endocannabinoids AEA and 2-AG differ in their synthetic pathways (18) and may be produced under distinct physiological conditions or in distinct brain regions (19, 20). Here we show that glucocorticoids lead to the release of both AEA and 2-AG from the SON to a comparable degree and that both suppress glutamate release onto magnocellular neurons. It remains to be seen whether the glucocorticoid-induced suppression of glutamate release onto SON magnocellular neurons is due to the retrograde actions of AEA, 2-AG, or both endocannabinoids.

We showed that glucocorticoids facilitate GABA release in the SON and PVN, also by activating a postsynaptic membrane receptor on magnocellular neurons and stimulating the release of a retrograde messenger. Paradoxically, the glucocorticoid facilitation of GABA release was blocked by a selective CB1 receptor antagonist but was not mimicked by cannabinoid agonists. On the contrary, bath application of the endocannabinoids AEA and 2-AG suppressed GABA release, causing, therefore, an effect opposite to that of glucocorticoids. The observation that endocannabinoids suppress GABA release onto hypothalamic magnocellular neurons is consistent with reports of endocannabinoid negative regulation of GABA release in diverse brain areas, including the hippocampus and cerebellum (21–25). Endocannabinoids are endogenous ligands for the capsaicin-sensitive vanilloid receptors (26, 27), and the endocannabinoid activation of these receptors often leads to a facilitation of transmitter release (28). However, the glucocorticoid facilitation of GABA release in the SON was not blocked by blocking VR1 receptors, indicating that it was not mediated by retrograde endocannabinoid effects at presynaptic vanilloid receptors. Nor was the glucocorticoid effect on GABA release blocked by antagonists of ionotropic and metabotropic glutamate receptors, indicating that it was not due to endocannabinoid actions at intervening axo-axonic glutamate synapses. It appears, therefore, that the glucocorticoid facilitation of GABA release is mediated by a retrograde messenger that acts at presynaptic AM251-sensitive, non-CB1 receptors on GABA axon terminals.

It would appear unlikely, at first glance, that this retrograde messenger is an endogenous cannabinoid because neither AEA nor 2-AG applied exogenously caused an increase in mIPSC frequency and because the actions of retrograde endocannabinoids at most central GABA synapses studied to date have been uniformly reported as suppressing GABA release (17). Nevertheless, it is possible that the effect was mediated by a different endogenous cannabinoid released by glucocorticoids that acts at an as-yet-unknown, AM251-sensitive, CB receptor. Since the discoveries of AEA (29) and 2-AG (30, 31), several other lipid messengers with cannabinoid-like properties have been described, and many of these have been shown to have significantly lower affinity for CB1 receptors (32). Studies in CB1/2 receptor knockout mice have provided evidence for non CB1/2 receptors, or CB(x) receptors, in the hippocampus (33) and hypothalamus (34). However, in contrast to our observations here, the putative CB(x) receptors in the hippocampus are found on presynaptic glutamate, and not GABA, terminals and their activation is not blocked by AM251, and the CB(x) receptors in the hypothalamus are activated by AEA (34).

Another possible explanation for our findings is that the endocannabinoid released from magnocellular neurons by glucocorticoids, rather than acting at presynaptic GABA terminals directly, activates CB1 receptors on upstream axo-axonic GABA synapses onto presynaptic GABA terminals, reducing GABA release and effectively disinhibiting the GABA release onto the magnocellular neurons. Still another possibility is that endocannabinoids interact with other neurotransmitters to modulate GABA release because there is a large body of evidence for endocannabinoid interactions with several other neurotransmitter systems (35–37). Finally, the glucocorticoid effect on GABA release may be mediated by another, noncannabinoid retrograde messenger. A possible candidate messenger is nitric oxide, which acts as a retrograde messenger and has been shown to cause an increase in GABA release onto magnocellular neurons in the PVN (38, 39). Further studies are required to elucidate the mechanism responsible for the glucocorticoid facilitation of GABA release observed in this study.

By decreasing glutamate release and increasing GABA release onto magnocellular neurons, glucocorticoids would be expected to cause a marked reduction in the excitability of these cells in vivo. Stress and stress levels of corticosteroids have been shown to inhibit vasopressin release (40) as well as oxytocin release and the milk ejection reflex in lactating animals and delay parturition in parturient animals (41, 42). Exogenous cannabinoids have also been shown to inhibit vasopressin and oxytocin release. For example, marijuana and Δ9-tetrahydrocannabinol cause increased diuresis in humans and rats, which is thought to be mediated by central inhibitory cannabinoid actions on vasopressin release (18, 43, 44). In lactating rats, Δ9-tetrahydrocannabinol blocks suckling-induced milk ejection mediated by the secretion of oxytocin (45). The glucocorticoid-induced endocannabinoid synthesis and release described here may underlie and link these inhibitory effects of glucocorticoids and cannabinoids on neurohypophysial hormone secretion.

This work was supported by National Institutes of Health Grants MH066958 (to J.G.T.), NS23002 (to N.G.B.), and P20RR016816 (to N.G.B. and J.G.T.).

First Published Online June 30, 2005

Abbreviations

• aCSF,

  Artificial cerebral spinal fluid;

• AEA,

  anandamide;

• 2-AG,

  2-arachidonoylglycerol;

• AP-5,

  d-(-)-2-amino-5-phosphonopentanoic acid;

• CB1,

  type I cannabinoid receptor;

• DEX,

  dexamethasone;

• DNQX,

  6,7-dinitroquinoxaline-2,3-dione;

• GABA,

  γ-aminobutyric acid;

• GAPDH,

  glyceraldehyde-3-phosphate dehydrogenase;

• IR-DIC,

  infrared illumination and differential interference contrast optics;

• LC-MS-MS,

  liquid chromatography-tandem mass spectrometry;

• mEPSC,

  miniature excitatory postsynaptic current;

• mGluR,

  metabotropic glutamate receptor;

• mIPSC,

  miniature inhibitory postsynaptic current;

• PVN,

  paraventricular nucleus;

• RT,

  reverse transcription;

• SON,

  supraoptic nucleus;

• TTX,

  tetrodotoxin.