Mineralocorticoid receptors are indispensable for nongenomic modulation of hippocampal glutamate transmission by corticosterone

Abstract

The adrenal hormone corticosterone transcriptionally regulates responsive genes in the rodent hippocampus through nuclear mineralocorticoid and glucocorticoid receptors. Via this genomic pathway the hormone alters properties of hippocampal cells slowly and for a prolonged period. Here we report that corticosterone also rapidly and reversibly changes hippocampal signaling. Stress levels of the hormone enhance the frequency of miniature excitatory postsynaptic potentials in CA1 pyramidal neurons and reduce paired-pulse facilitation, pointing to a hormone-dependent enhancement of glutamate-release probability. The rapid effect by corticosterone is accomplished through a nongenomic pathway involving membrane-located receptors. Unexpectedly, the rapid effect critically depends on the classical mineralocorticoid receptor, as evidenced by the effectiveness of agonists, antagonists, and brain-specific inactivation of the mineralocorticoid but not the glucocorticoid receptor gene. Rapid actions by corticosterone would allow the brain to change its function within minutes after stress-induced elevations of corticosteroid levels, in addition to responding later through gene-mediated signaling pathways.

The release of corticosterone, i.e., the prevailing endogenous adrenal corticosteroid in rats and mice, is markedly enhanced after exposure to a stressor. Hormone levels rise within minutes and return to baseline ≈60–90 min after the stress (1). Corticosterone can quickly enter the brain and bind to two subtypes of discretely localized receptors, the high-affinity mineralocorticoid receptor (MR) and the lower-affinity glucocorticoid receptor (GR) (2). Hippocampal CA1 neurons abundantly express both receptor types. Activated corticosteroid receptors are known to modify transcription of responsive genes, either through DNA binding of homodimers or through protein–protein interactions with other transcription factors (3, 4). In this way, hippocampal cell properties, like calcium influx, were found to be altered in a slow but persistent manner (5). In recent years, though, preliminary evidence has accumulated that corticosterone can also induce rapid effects in brain (6–10), which would vastly expand the time window over which adrenal hormones can affect brain function. Although in amphibian brain a specific membrane receptor was isolated mediating rapid behavioral effects (11), the underlying mechanism and presumed membrane receptor involved in rapid effects of the mammalian hormone remained unresolved. We here demonstrate that a rapid nongenomic effect by corticosterone through membrane-located receptors critically depends on the MR, which hitherto was regarded only as a nuclear receptor.

Methods

Animals. All experiments, unless stated otherwise, were performed on male C57BL/6 mice, ≈6 weeks of age and group-housed in cages with a light/dark cycle of 12 h (lights on at 0800 hours). Food and water were given ad libitum. The experiments were carried out with permission of the local Animal Committee (file no. DED 91). One mouse per day was decapitated under rest around 0930 hours, i.e., when plasma corticosterone levels are low (ref. 5; ≈2 μg/dl).

A limited number of experiments was performed in brain-specific GR (12) and MR knockout mice. Details about the generation of MR mutants are described in Supporting Text, which is published as supporting information on the PNAS web site. In these two mutant lines, immunoreactivity of the GR and MR protein, respectively, was lost in all CA1 neurons, which did express the proteins in control animals. In view of the elevated corticosterone levels in GR mutants, both the mutant mice and their control littermates (GR^loxP/loxP) were adrenalectomized, which in mice resulted in low (≈1 μg/dl) levels of circulating corticosterone, not unlike the levels in adrenally intact mice, which were decapitated in the morning under rest.

Slice Preparation and Recording. On the day of the experiment, the animal was decapitated, and the brain was removed from the skull and stored in continuously gassed medium. Transverse slices of the hippocampus were made with a tissue chopper and stored at room temperature in medium containing 124 mM NaCl, 3.5 mM KCl, 1.25 mM NaH₂PO₄, 1.5 mM MgSO₄, 2 mM CaCl, 25 mM NaHCO₃, and 10 mM glucose (osmolarity = 300 mOsm; pH 7.4). One slice at a time was fully submerged in a recording chamber and continuously perfused with the medium (32°C, 2–3 ml/s), as described in ref. 13. Bicuculline methiodide (20 μM, Sigma) was added to block GABA_A receptor-mediated signals.

Whole-cell patch clamp recordings were made with an Axopatch 200B amplifier (Axon Instruments, Union City, CA) by using borosilicate glass electrodes (1.5-mm outer diameter; Hilgenberg, Malsfeld, Germany; pulled on a Sutter micropipette puller), with an impedance of 3–4 MΩ. The intracellular pipette solution contained 120 mM Cs methane sulfonate, 17.5 mM CsCl, 10 mM Hepes, 2 mM MgATP, 0.1 mM NaGTP, 5 mM BAPTA, and 10 mM QX-314 (osmolarity = 295 mOsm; pH 7.4), adjusted with CsOH. BAPTA was obtained from Molecular Probes, the sodium channel blocker QX-314 was from Alomone (Jerusalem), and all other chemicals were from Sigma. Under visual control the recording electrode was directed toward a CA1 neuron by using positive pressure. Once a patch electrode was sealed on the cell (≈1 GΩ) the membrane patch under the electrode was ruptured and the cell was held at a holding potential of -70 mV. Recordings with an uncompensated series resistance of <2.5 times the pipette resistance were accepted for analysis. Data acquisition was performed with pclamp 8.2 and analyzed with clampfit 8.2 and strathclyde software (J. Dempster, University of Strathclyde, Glasgow, U.K.).

Stimulus Evoked of Excitatory Postsynaptic Currents (EPSCs). A bipolar stainless steel stimulus electrode was placed in the Schaffer collaterals. Biphasic stimuli were applied through a stimulus isolator driven by pclamp 8.2. Input–output curves of EPSCs evoked in CA1 neurons were made at holding potential (-70 mV) by stimuli of increasing intensities (from 16 μA to 500 μA), given once every 10 seconds. Evoked EPSCs were recorded with a sampling frequency of 10 kHz. Signals were stored and corrected off-line for leak. Input–output curves were fitted with a Boltzmann equation: R(i) = R_max/[1 + exp{(i - i_H)/I_C}], where R_max is the maximal evoked current, i_H is the half maximal stimulus intensity, and I_C is proportional to the slope. Based on this curve the half maximal stimulus intensity was determined. This intensity was used to record responses to paired pulse stimulation (interstimulus interval, 100 ms).

Miniature EPSCs (mEPSCs). mEPSCs were always recorded in the presence of 0.5 μM TTX, either added to the buffer after recording of evoked EPSCs in the same cell or present during the entire experiment (i.e., in case evoked EPSCs were not recorded). At a holding potential of -70 mV, mEPSCs were recorded for 5 min. During some recordings the non-NMDA receptor blocker CNQX or the selective α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor blocker GYKI 53655 was perfused to confirm that the mEPSCs were indeed mediated by AMPA receptors. The digitized data (stored on PC via Digidata interface) were analyzed off-line by using strathclyde software with detection threshold levels set above 5 pA. The currents were identified as mEPSCs when the rise time was faster than the decay time (13). Of all cells measured, the following mEPSC characteristics were determined: inter-mEPSC interval and frequency, rise time, peak amplitude, and τ of decay. The decay of each mEPSC was fitted with a monoexponential curve using the wholecell program of the strathclyde software. This program uses the Levenberg–Marquardt algorithm to iteratively minimize the sum of the squared differences between the theoretical curve and the data curve. As criterion for the goodness of the fit the residual standard deviation should be <0.3. Fitting with a biexponential curve instead of a monoexponential curve did not increase goodness of the fit (data not shown).

Statistics. Rapid effects on frequency, amplitude, and decay time were analyzed with a two-tailed paired Student t test. In all cases, significance was set at P < 0.05.

Results

mEPSCs were recorded from 82 CA1 pyramidal cells in acutely prepared adult mouse hippocampal slices (Fig. 1A). These currents reflect the spontaneous fusion of glutamate-containing vesicles with the presynaptic membrane (14). All currents were recorded at a holding potential of -70 mV, when NMDA receptors are presumably blocked by Mg²⁺ ions (15). The exclusive involvement of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors was further evidenced by effective reversible blockade of the mEPSCs by the non-NMDA receptor antagonist CNQX and the selective AMPA receptor antagonist GYKI 53655 (data not shown, see refs. 13, 16, and 17).

Fig. 1.

Corticosterone rapidly and reversibly enhances mEPSC frequency in hippocampal cells. (A) Typical example showing enhanced mEPSC frequency in a CA1 pyramidal cell during 100 nM corticosterone (cort) application (Middle) compared with the control situation (Top). Bottom shows the typical shape of an mEPSC. (B) Example showing that mEPSC frequency is rapidly and reversibly enhanced by corticosterone but not by vehicle (0.009% ethanol). (C) Cumulative frequency histogram of the mEPSC frequency in a CA1 hippocampal cell, showing that the mEPSC frequency is enhanced in the presence of corticosterone and returns to pretreatment levels after corticosterone has been washed out. (D) Percentual increase in mEPSC frequency caused by various concentrations of corticosterone (filled circles, based on n = 3–5 cells for each concentration), showing that 10 nM corticosterone, but not lower concentrations of corticosterone, induces a significant (P < 0.05) increase in mEPSC frequency. The membrane-impermeable BSA–corticosterone conjugate (100 nM; n = 5) also resulted in a marked increase in mEPSC frequency. In the presence of the protein synthesis inhibitor cycloheximide (100 nM; n = 5), corticosterone was still able to enhance the mEPSC frequency. Symbols represent the mean + SEM values.

Bath application of 100 nM corticosterone rapidly and reversibly enhanced the frequency of mEPSCs (see examples in Fig. 1 A–C). On average, corticosterone induced a 60% increase in the mEPSC frequency (Table 1). In the hippocampus, enhancement of mEPSC frequency was seen already at 10 nM corticosterone (Fig. 1D). The mEPSC amplitude was not affected by the hormone (Table 1). Similarly, kinetic properties like the time constant of the decay were unaffected (Table 1). The enhancement of glutamate transmission may be hippocampus-specific, because 100 nM corticosterone rapidly reduced mEPSC frequency in parvocellular neurons of the hypothalamic paraventricular nucleus (38.9 ± 15.9%, n = 3), in line with an earlier report (9).

Table 1. Effect of corticosterone on frequency, amplitude, and decay time of mEPSCs in CA1 hippocampal neurons.

Condition	mEPSC frequency, Hz	mEPSC amplitude, pA	τ of decay, s
Before corticosterone	0.85 ± 0.14	22.8 ± 2.3	17.5 ± 1.6
During corticosterone	1.33 ± 0.25*	24.5 ± 1.9	15.8 ± 0.4

All values represent the mean ± SEM, based on a total of 3,242 events from five cells. Corticosterone significantly (*, P < 0.05) enhanced the mEPSC frequency (determined between 5 and 10 min after application was started), whereas the amplitude and time constant of the decay (based on a monoexponential fit) remained unaffected. Control data were collected between 5 and 0 min before corticosterone application was started.

The rapid onset and reversibility of the effect indicate that a nongenomic pathway underlies the hormone action. In agreement, the increase in mEPSC frequency by corticosterone was of comparable magnitude when tested in the presence of a translation inhibitor, cycloheximide (Fig. 1D; P = 0.04). Earlier studies in nonneuronal tissues have shown that nongenomic steroid effects can be mediated by a receptor located in the membrane (18). This also appears to be the case for the present effect of corticosterone, because administration of a corticosterone–BSA conjugate (100 nM), which cannot pass the membrane, increased the mEPSC frequency (Fig. 1D; P = 0.03) with an efficacy that was comparable to that of corticosterone itself.

Increases in mEPSC frequency, but not amplitude or kinetics, point to a change in presynaptic properties of glutamate transmission (19). To distinguish between an increased release probability and an enhanced number of synaptic contacts we examined paired-pulse responses to stimulation of the Schaffer collateral afferents (interstimulus interval, 100 ms). We observed that corticosterone decreases the ratio of the second to the first response (before, 1.20 ± 0.03; during, 1.00 ± 0.04, n = 4; P = 0.04). This finding is compatible with the notion that the hormone increases the release probability for glutamate (20). In agreement, a recent study using microdialysis demonstrated a rapid, nongenomic increase of glutamate but not GABA release by a high dose of corticosterone (8). Such a rapid increase in glutamate release potentially enhances excitability in the CA1 region, but it should be emphasized that overall excitability depends not only on processes taking place in presynaptic terminals as presently studied, but also on other factors, e.g., the stability of synaptic contacts that are located on higher-order dendritic branches and intrinsic network properties related to feedforward or feedback inhibition.

To determine whether “classical” steroid receptors are involved in these rapid, nongenomic effects of corticosterone, we used selective receptor agonists and antagonists as well as brain-specific GR or MR knockout mice. Based on the relatively high threshold concentration, we assumed that the rapid effects were mediated by GR rather than MR. However, the highly selective GR agonist RU 28362 (21) did not increase mEPSC frequency (Fig. 2A). Also, the effects by corticosterone were not significantly blocked by RU 38486, which interferes with transactivation via the GR (21) (Fig. 2 A). Finally, in hippocampal slices prepared from brain-specific GR mutant mice (12), the mEPSC frequency was still significantly enhanced (Fig. 2B). Evidently, the GR is not involved in these rapid hormone effects. By contrast, aldosterone, which in rodents has a higher affinity for MR than GR (21), markedly enhanced mEPSC frequency, at a concentration (10 nM) that induced only threshold effects with corticosterone (Fig. 2 A). In view of this higher sensitivity to aldosterone than corticosterone and the fact that aldosterone was tested in the presence of the GR antagonist RU 38486, involvement of an MR rather than GR is indicated. In agreement, the effect of aldosterone (Fig. 2 A) in hippocampus was completely blocked by spironolactone, an MR antagonist (22). Most importantly, no effect at all of corticosterone on mEPSC frequency was observed in CA1 cells of forebrain-specific MR knockout mice as opposed to the controls (Fig. 2B).

Fig. 2.

Increased mEPSC frequency during corticosterone application is mediated by the MR and not the GR. (A) The selective GR agonist RU 28362 (100 nM) does not increase the mEPSC frequency (n = 4). The GR antagonist RU 38486 (500 nM; n = 4) does not significantly affect the percentual increase in mEPSC frequency caused by corticosterone when compared with the response evoked by corticosterone alone (shown on the far left). By contrast, the endogenous mineralocorticoid aldosterone (10 nM; n = 5), applied in the presence of the GR antagonist RU 38486, potently increased the mEPSC frequency. Additional application of the MR antagonist spironolactone (100 nM; n = 5) completely abolished the response to aldosterone. (B) Corticosterone (100 nM) significantly enhances the mEPSC frequency in CA1 pyramidal cells of brain-specific GR knockout mice (GR^NesCre)(n = 4 cells) as well as of controls (GR^loxP/loxP)(n = 4). In cells of brain-specific MR knockout mice (MR^CamKIICre) (n = 6), however, corticosterone application was ineffective, and normal effects were seen in control mice (MR^loxP/loxP)(n = 6). All data represent the percentual increase in mEPSC frequency, expressed as the ratio of the averaged frequency determined between 5 and 10 min after administration of the agonist and the averaged frequency between 5 and 0 min before administration was started (*, P < 0.05). Details on the transgenesis are provided in Supporting Text.

Discussion

The present study demonstrates a fast-onset, rapidly reversible, and nongenomic enhancement by corticosterone of glutamate transmission in the CA1 hippocampal area. With the use of a unique forebrain-specific MR knockout mouse model we demonstrated that these nongenomic effects critically depend on a gene that encodes the MR. Until now this receptor was thought to mediate only slow actions in brain, through transcriptional regulation of responsive genes. The threshold corticosterone concentration for rapid effects was presently found to be ≈10 nM, whereas earlier electrophysiological investigations showed that 10- to 20-fold lower concentrations suffice to activate the intracellular MR (23). This lower apparent affinity of the MRs involved in rapid corticosteroid effects could lend a new meaning to this receptor, of which the role in the stress response has remained somewhat difficult to understand given its very high affinity and hence considerable occupancy already with low circulating levels of the hormone.

Because corticosterone does not have to pass the membrane to accomplish these effects in hippocampus, it is likely that the receptor is localized within the membrane. Imaging and immunocytochemical studies so far have demonstrated the MR to be present primarily in the nuclear compartment (24–26), although membrane localization was reported with immunoelectron microscopy for human kidney cells (27). We propose that part of the brain MRs (the degree of which may depend on long-term averages of circulating hormone levels) can shuttle from the nucleus/cytoplasm to the cell membrane. Shuttling of steroid receptors is not unprecedented, because incorporation of a small fraction of the estrogen receptors (<10%) into the cell membrane has been demonstrated (28). This mechanism could explain the rapid effects on excitability of hippocampal (29) and hypothalamic neurons (30) seen with estrogens. Shuttling of the receptor may involve rearrangement of chaperone proteins, such as heat-shock protein 90, and 3D adaptation of the receptor. These as well as other factors involving the receptor structure could explain the apparent decrease in affinity of the receptor (31).

The requirement of at least 10 nM corticosterone to induce rapid effects on hippocampal glutamate transmission indicates that such effects will not occur with basal levels of the hormone but only when hormone levels are elevated, i.e., ≈10–60 min after stress exposure or possibly during the circadian peak (32). The MR could thus serve as a “cortico-sensor” in the brain, rapidly changing excitatory transmission while corticosteroid levels are enhanced. This mechanism would supply the brain with the means to react quickly to stress through the MR, next to other rapid stress-activated systems involving noradrenaline and corticotropin-releasing hormone. Gene-mediated stress effects by the hormone (13), which in hippocampus typically develop with a delay of 1–2 h and mostly through the GR (23), would next adapt hippocampal activity when circulating hormone levels have normalized again after stress and, thus, when rapid MR effects have subsided. Through this binary mode of action, one hormone (corticosterone) could alter hippocampal activity over a very prolonged period, during as well as after exposure to a stressful situation.