A functional coupling between extrasynaptic NMDA receptors and A-type K⁺ channels under astrocyte control regulates hypothalamic neurosecretory neuronal activity

Abstract

Neuronal activity is controlled by a fine-tuned balance between intrinsic properties and extrinsic synaptic inputs. Moreover, neighbouring astrocytes are now recognized to influence a wide spectrum of neuronal functions. Yet, how these three key factors act in concert to modulate and fine-tune neuronal output is not well understood. Here, we show that in rat hypothalamic magnocellular neurosecretory cells (MNCs), glutamate NMDA receptors (NMDARs) are negatively coupled to the transient, voltage-gated A-type K⁺ current (I_A). We found that activation of NMDARs by extracellular glutamate levels influenced by astrocyte glutamate transporters resulted in a significant inhibition of I_A. The NMDAR–I_A functional coupling resulted from activation of extrasynaptic NMDARs, was calcium- and protein kinase C-dependent, and involved enhanced steady-state, voltage-dependent inactivation of I_A. The NMDAR–I_A coupling diminished the latency to the first evoked spike in response to membrane depolarization and increased the total number of evoked action potentials, thus strengthening the neuronal input/output function. Finally, we found a blunted NMDA-mediated inhibition of I_A in dehydrated rats. Together, our findings support a novel signalling mechanism that involves a functional coupling between extrasynaptic NMDARs and A-type K⁺ channels, which is influenced by local astrocytes. We show this signalling complex to play an important role in modulating hypothalamic neuronal excitability, which may contribute to adaptive responses during a sustained osmotic challenge such as dehydration.

Key points

• In hypothalamic magnocellular neurosecretory cells, activation of glutamate NMDA receptors leads to inhibition of the transient voltage-gated A-type K⁺ current (I_A), in a Ca²⁺- and protein kinase C-dependent manner.

• The negative NMDAR–I_A functional coupling involves activation of extrasynaptic (e)NMDARs. The eNMDAR–I_A coupling is engaged by endogenous extracellular glutamate, whose levels are controlled by astrocyte glutamate GLT1 transporters.

• The eNMDAR–I_A coupling is enhanced during dehydration, a condition in which astrocyte GLT1 efficiency is blunted.

• The eNMDAR–I_A coupling results in increased neuronal excitability and firing activity in magnocellular neurosecretory neurons.

• Taken together these studies support the concept that the eNMDAR–I_A coupling is a powerful mechanism by which glutamate increases magnocellular neurosecretory excitability and firing activity.

Introduction

Complex interactions among intrinsic membrane properties and synaptic inputs influence the degree and pattern of neuronal activity in the central nervous system. Moreover, astrocytes are now recognized to influence a wide spectrum of neuronal functions, ranging from fine-tuning of synaptic strength to changes in overall neuronal network output (Araque & Navarrete, 2010; Giaume, 2010). In addition to releasing a variety of neuroactive substances, including ATP and d-serine (Panatier et al. 2006; Gordon et al. 2007) astrocytes affect neuronal function via the expression of selective transporters that regulate ambient neurotransmitter levels (Rothstein et al. 1996). Yet the precise mechanisms by which these three pivotal factors (i.e. intrinsic properties, synaptic mechanisms and astrocytes) act in concert to modulate neuronal output have not been explored in detail.

The hypothalamic magnocellular system constitutes an ideal network model to study this phenomenon. It is a simple, well-characterized system comprising magnocellular neurosecretory vasopressin and oxytocin cells (MNCs) in the supraoptic (SON) and paraventricular (PVN) nuclei. MNC firing activity directly correlates with neurosecretion from the posterior pituitary (Cazalis et al. 1985). The key intrinsic and synaptic properties influencing MNC activity are well characterized.

The A-type potassium current (I_A) is a dominant intrinsic, voltage-dependent subthreshold current that determines spike onset and interspike interval during MNC repetitive firing (Bourque, 1988; Luther & Tasker, 2000). Moreover, we showed that downregulation of I_A contributes to increased hypothalamic neuronal excitability during hypertension (Sonner et al. 2008, 2011).

Glutamate is the major extrinsic excitatory neurotransmitter in the hypothalamus (van den Pol et al. 1990). Acting primarily on ionotropic NMDA receptors (NMDARs), glutamate not only stimulates overall MNC activity, but also contributes to the adoption of bursting firing, which is critical to optimize the neurosecretory output of this system (Hu & Bourque, 1992; Nissen et al. 1995).

Finally, the magnocellular system possesses a unique and dynamic neuro-glial microenvironment. Under basal conditions, astroglial processes closely enwrap neurons and their synapses. This tight microenvironment enables astrocyte glutamate transporters to efficiently buffer extracellular glutamate, restricting activation of extrasynaptic NMDARs (eNMDARs) (Fleming et al. 2011) and presynaptic metabotropic receptors (Oliet et al. 2001; Boudaba et al. 2003). Physiological challenges including dehydration and lactation evoke a rapid and reversible retraction of astroglial processes (Perlmutter et al. 1984; Oliet et al. 2001) enhancing ambient glutamate levels and actions at these extra/perisynaptic sites (Oliet et al. 2001; Fleming et al. 2011). This dynamic neuro-glial remodelling is necessary for the adoption of bursting firing patterns during conditions of high hormonal demand (Theodosis & Poulain, 1993; Tasker et al. 2012). Thus, given the importance of astrocytes, NMDARs and I_A in MNCs, we explored whether these mechanisms acted in an interrelated manner to influence MNC firing activity. Our results show that in addition to inducing a direct excitatory inward current, activation of eNMDARs are negatively coupled, in a Ca²⁺- and protein kinase C (PKC)-dependent manner, to I_A. Moreover, we show that, by modulating ambient glutamate levels, astrocyte GLT1 transporters influence the degree of activation of the NMDAR–I_A coupling. Finally, we demonstrate that this functional coupling efficiently influences MNC membrane excitability and firing output.

Methods

Ethical approval

All experimental procedures were in strict compliance with NIH guidelines, and were approved by the Georgia Regents University Institutional Animal Care and Use Committee.

Animals

Male Wistar rats purchased from Harlan Laboratories (Indianapolis, IN, USA) (4–5 weeks old) were housed under standardized conditions (12 h:12 h light–dark cycle, lights on 07.00 h) with food and water available ad libitum. For dehydration, water was removed for 48 h with free access to dry food. In these rats, plasma osmolarity was significantly elevated (328.5 ± 1.8 vs. 308 ± 2.1 mosmol l⁻¹, in dehydrated and euhydrated rats, respectively, P < 0.05).

Hypothalamic slices

Rats were anaesthetized with pentobarbital (50 mg kg^–1 i.p.), quickly decapitated and brains dissected out. Coronal slices were cut (250 μm thick) utilizing a vibroslicer (Leica VT1000). An oxygenated ice-cold artificial cerebrospinal fluid (ACSF) was used during slicing (containing in mm): 119 NaCl, 2.5 KCl, 1 MgSO₄, 26 NaHCO₃, 1.25 NaH₂PO₄, 20 d-glucose, 0.4 ascorbic acid, 2.0 CaCl₂ and 2.0 pyruvic acid; pH 7.4; 290–310 mosmol l⁻¹). After sectioning, slices were placed in a holding chamber containing ACSF and kept at room temperature (22°C) until used.

Drugs

Tetrodotoxin (TTX) and (2S,3S,4R)-2-carboxy-4-isopropyl-3-pyrrolidineacetic acid (DHK, dihydrokainic acid) were purchased from Abcam Biochemicals (Cambridge, MA, USA). N-Methyl-d-aspartate (NMDA), bicuculline, MK801, 4-aminopyridine, 4-hydroxyquinoline-2-carboxylic acid (kynurenic acid), 6,7-dinitroquinoxaline-2,3-dione (DNQX) and glycine were purchased from Sigma-Aldrich (St Louis, MO, USA). 1,2-bis(o-Aminophenoxy)ethane- N,N,N′,N′-tetraacetic acid (BAPTA) was purchased from Invitrogen (Grand Island, NY, USA). dl-2-Amino-5-phosphonopentanoic acid (AP5) and chelerythrine chloride were purchased from Tocris Bioscience (Bristol, UK).

Electrophysiological recordings

Slices were placed in a submersion style recording chamber, and bathed with solutions (3.0 ml min⁻¹) that were bubbled continuously with a gas mix of 95% O₂–5% CO₂, and maintained at near physiological temperature (32°C). Thin-walled (1.5 mm o.d., 1.17 mm i.d.) borosilicate glass (G150TF-3, Warner Instruments, Sarasota, FL, USA) was used to pull patch pipettes (3–6 MΩ) on a horizontal Flaming/Brown micropipette puller (P-97, Sutter Instruments, Novato, CA, USA). Whole-cell patch-clamp recordings from SON neurons were visually made using diffraction interference contrast (DIC) videomicroscopy. Recordings were obtained with a Multiclamp 700A amplifier (Axon Instruments, Union City, CA, USA). The voltage output was digitized at 16-bit resolution, 10 kHz (Digidata1320A, Axon Instruments), and saved on a computer to be analysed offline using pCLAMP9 software (Axon Instruments). Mean series resistance (SR) was 13.4 ± 0.5 MΩ and experiments were discarded in cases in which the series resistance was unstable (changes >20%). In a control set of experiments, SR was compensated for ∼60% and a similar degree of NMDAR-mediated I_A inhibition was observed as we report without SR compensation. Moreover, the overall magnitude of the recorded currents was minimally affected by the compensation (less than 20% at the highest voltage tested). Finally, MNCs in SON/PVN are electrotonically compact, displaying very rudimentary dendritic trees (Armstrong & Smith, 1990; Stern & Armstrong, 1998; Luther & Tasker, 2000; Stern, 2001), minimizing potential space clamp problems during recordings. Given this, and the fact that the goal of our study was to report changes in I_A current ‘before and after’ a treatment, rather than reporting precise amplitude/kinetic values of these currents, all subsequent studies were performed without SR compensation. The internal solution contained (in mm): 140 potassium gluconate, 0.2 EGTA, 10 Hepes, 10 KCl, 0.9 MgCl₂, 4 MgATP, 0.3 NaGTP and 20 phosphocreatine (Na⁺); pH 7.2–7.3. All recordings were obtained in a low Mg²⁺ ACSF (10 μm MgSO₄) in order to facilitate measurements of NMDA currents (I_NMDA).

Voltage clamp

Unless otherwise stated, glycine (10 μm) and tetrodotoxin (0.5 μm) were added to the ACSF for voltage-clamp recordings. I_NMDA was activated either by focal, transient application of NMDA via a picospritzer or via bath application of NMDA. I_NMDA was also activated by inducing build-up of extracellular glutamate levels following blockade of glial GLT1 transporters with dihydrokainate (DHK, 300 μm; Fleming et al. 2011). Cells not showing an evident I_NMDA following DHK (n = 2) were not included in the study. I_NMDA was quantified as the change in the holding current measured as the average of a 2 min segment of steady-state baseline obtained before and after a 5–10 min application of NMDA or DHK. The voltage-dependent properties of activation of I_A were assessed before and after NMDA/DHK application (when a new steady-state current was achieved), using conventional protocols (depolarizing steps from −70 to +20 mV in 10 mV increments, 200 ms; Sonner et al. 2008). These protocols evoked both I_A and the delayed-rectifier current I_KDR. For bath-applied NMDA/DHK, due to the very slow nature of these experiments, we were unable to electronically or pharmacologically isolate these two components. Nonetheless, at the range of membrane potentials tested, I_A was clearly differentiated from I_KDR based on several characteristic properties including hyperpolarized activation threshold, overall larger magnitude and faster activation kinetics and rapid inactivation (see Fig. 1). During transient NMDAR activation, however, I_A was isolated electronically by digital subtraction of currents evoked from two separate protocols, in which preconditioning steps of either −80 mV and −10 mV were used, to fully evoke, and to completely inactivate I_A, respectively (Sonner et al. 2008). Since the focus of the present study was on I_A, the properties of I_KDR were not further studied. To determine the I_A half-activation potential, current amplitudes normalized to the maximum peak were plotted as a function of the command step potentials, and fitted with a Boltzmann function. To determine the voltage dependence of inactivation, I_A currents were evoked by voltage steps to +10 mV (300 ms duration) from conditioning steps between −120 and −30 mV in 10 mV increments (50 ms duration) before and after application of NMDA/DHK. To determine the half-inactivation potential, current amplitudes normalized to the maximum peak were plotted as a function of the conditioning step (Sonner et al. 2008).

Figure 1. Sustained NMDA receptor activation inhibits I_A.

A, DIC image of a patched MNC (left) showing the characteristic transient outward rectification during evoked depolarizations (right, arrow, 50 pA step^–1). B, NMDA (30 μm)-induced inward shift in I_holding evoked in the same neuron as in A. C, samples of I_A (arrow) and delayed rectifier (I_KDR, arrowhead) outward K⁺ currents activated before and during NMDA application (at time points 1 and 2 in B, respectively). D, mean (left) and before–after (right) plots summarizing the effect of NMDAR activation on I_A density evoked at 0 mV before (open symbols) and during (filled symbols) NMDAR activation (n = 12), E, mean plot of the normalized I_A density vs. command potential before (open squares) and during (filled squares) NMDA application (n = 9). F, plot of mean percentage I_A inhibition in the presence of NMDA vs. the command potential (n = 9). **P < 0.01; paired t test. Scale bar in A, 10 μm.

Current clamp

Action potentials were triggered with depolarizing current steps of varying amplitudes (Δ50 pA) or duration (Δ200 ms) while holding neurons at ∼−80 mV. Plots of number of action potentials as a function of step current amplitude or duration were constructed. The mean time delay to the first spike, and the total number of action potentials triggered were calculated and compared before and after drug applications. In the case of bath-applied NMDA or DHK, neurons were hyperpolarized with an additional DC current injection to maintain neurons at ∼−80 mV before depolarizing steps were applied again. To test for the influence of I_A on evoked firing properties, the K⁺ channel blocker 4-aminopyridine (4-AP) was used. Since 4-AP facilitates the presynaptic release of neurotransmitter (Flores-Hernandez et al. 1994), which could mask the direct effects of 4-AP on intrinsic properties, these experiments were performed in the presence of the GABA_A and AMPA receptor blockers bicuculline (20 μm) and DNQX (30 μm), respectively, a condition in which the DHK-mediated current was still present (not shown).

Statistical analysis

All values are expressed as means ± SEM. Student's paired t test was used to compare the effects of a drug treatment on I_NMDA or I_A. Between-group (e.g. MK801 vs. glycine) differences were compared using either unpaired t tests or analysis of variance repeated measures (ANOVA-RM). Where the F ratio was significant, post hoc comparisons were completed using the Bonferroni post hoc test or Dunnett's multiple comparison test when comparing values against a basal control. Pearson's correlation test was used to determine if correlations existed between two parameters. Differences were considered statistically significant at P < 0.05 and n refers to the number of cells. All statistical analyses were conducted using GraphPad Prism (GraphPad Software, San Diego, CA, USA).

Results

NMDAR activation inhibits the magnitude of I_A

Whole-cell patch clamp recordings were obtained from 204 SON MNCs (Fig. 1A). We found that following bath-applied NMDA (30 μm, Fig. 1B), I_A in MNCs was significantly inhibited (Fig. 1C and D, n = 12). Mean normalized I–V plots showed that the voltage-dependent activation properties of I_A were not affected following NMDAR activation (Fig. 1E, half-activation potential (V_½): −8.26 ± 1.97 mV and −10.94 ± 1.79 mV before and after NMDA, respectively; P = 0.2, paired t test). Moreover, once I_A was evoked, a similar degree of NMDA-mediated inhibition was observed at all voltages measured (−40 mV to +20 mV, F = 0.588, P > 0.5, repeated measures ANOVA, Fig. 1F), indicating that the NMDA-mediated inhibition of I_A was voltage independent. Importantly, we found that blockade of NMDARs per se (100 μm AP5) increased the basal magnitude of I_A (control: 134.46 ± 22.34 pA pF^–1, AP5: 176.90 ± 36.14 pA pF^–1, P = 0.05 paired t test, n = 10).

To better study the time course of the NMDA effects, we focally and transiently puffed NMDA (20 μm, 1 p.s.i., 5 s duration, n = 11) directly onto the recorded neuron, while I_A was repetitively evoked (single depolarizing pulse to 0 mV, 1 s interval) before, during and after NMDAR activation (Fig. 2A). I_A was significantly inhibited 2 s following the initiation of the NMDA stimulus, returning to control levels at 12 s (Fig. 2B, F = 7.443, P < 0.0001, repeated measures ANOVA). The mean and individual I_A density before and after NMDA application is shown in Fig. 2C. Finally, we found no correlation between the degree of I_A inhibition and I_NMDA magnitude (Fig. 2D, Pearson's correlation r²: 0.0864 P > 0.05, Pearson's correlation test). The NMDA-mediated inhibition of I_A persisted in a subset of recordings obtained with a normal Mg²⁺ ACSF (1 mm MgSO₄) (%NMDA–I_A peak inhibition = 3.7 ± 1.1, n = 5), although as expected (given that recordings were obtained at a holding potential of −70 mV), the effect was substantially smaller when compared to that observed in a low Mg²⁺ ACSF (∼20% inhibition, see Fig. 2B). Collectively, our results support the presence of a negative functional coupling between NMDARs and A-type channels in hypothalamic MNCs.

Figure 2. Transient NMDA receptor activation inhibits I_A.

A, transient inward shift in I_holding following a focal brief application of NMDA (20 μm, 5 s) to an MNC. The lower traces show I_A before (0 s), at the end (5 s) and after (15 s) NMDA application. I_A was isolated electronically by digital subtraction of currents evoked from two separate protocols, in which preconditioning steps of either −80 mV and −10 mV were used, to fully evoke, and to completely inactivate I_A, respectively. B, plot of the mean percentage NMDA-evoked inhibition of I_A amplitude as a function of time (n = 11). C, mean (left) and before–after (right) plots of I_A density before (open symbols) and 3.5 s after (filled symbols) initiation of transient NMDAR activation. D, plot of the percentage inhibition of I_A as a function of I_NMDA, depicting a lack of correlation between the two parameters (r² = 0.0864). *P < 0.05 vs. baseline I_A (Dunnett's post hoc test); ***P < 0.0001 (paired t test).

Astrocytes influence the NMDAR–I_A functional coupling via glutamate-mediated activation of extrasynaptic NMDARs

We recently showed that astrocytes regulate ambient glutamate levels and the degree of eNMDAR activation in MNCs via the activity of GLT1 glutamate transporters (Fleming et al. 2011; Potapenko et al. 2013). Thus, we explored here whether astrocytes indirectly affect I_A magnitude via this mechanism. As previously reported (Fleming et al. 2011), GLT1 blockade (300 μm dihydrokainate, DHK) induced an inward shift in I_holding (Fig. 3A, DHK-mediated current density: 3.15 ± 0.94 pA pF^–1), which we previously showed to be blocked by AP5 (Fleming et al. 2011). DHK resulted in a significant inhibition of I_A (P < 0.05, n = 13, Fig. 3B), an effect that was voltage independent (Fig. 3C, F = 2.271, repeated measures ANOVA) and did not involve a change in the voltage-dependent activation properties of I_A (Fig. 3D, V_½ activation: −10.06 ± 2.18 mV and −9.69 ± 2.20 mV before and after DHK, respectively; P = 0.85, paired t test).

Figure 3. I_A magnitude is inhibited following blockade of astrocyte GLT1 transporters.

A, DHK-induced inward shift in I_holding in an MNC. Lower traces show I_A before (1) and during (2) DHK application. B, mean (left) and before–after (right) plots of I_A density (at 0 mV) before (open symbols) and after (filled symbols) DHK (300 μm) (n = 13). C, plot of mean percentage inhibition of I_A in the presence of DHK vs. the command potential. D, mean plots of the normalized I_A density vs. the command potential before (open squares) and during (filled squares) application of DHK. *P < 0.05 (paired t test).

To confirm the contribution of eNMDARs to the NMDA–I_A coupling, experiments were repeated in conditions in which we selectively blocked either synaptic or extrasynaptic NMDARs. To selectively block sNMDARs, we used a well-characterized approach based on the combined use of the NMDA open channel blocker MK801 (20 μm) with 4-AP (2.5 mm) and bicuculline (20 μm), to chemically evoke synaptic release of glutamate (Hardingham et al. 2002; Jung et al. 2008). In a set of control experiments, we found that at this concentration, 4-AP evoked a 30.0 ± 5.6% inhibition of I_A (n = 5, P < 0.01), which within 10 min after washout recovered almost completely (90.6 ± 5.6% of baseline magnitude, n = 5). We found that blockade of sNMDARs did not affect the DHK-induced shift in I_holding and DHK-induced I_A inhibition (n = 9, P < 0.05, paired t test, Fig. 4A, C, D and E). Conversely, removal of glycine from the ACSF, a critical co-agonist of extrasynaptic (but not synaptic) NMDARs (Papouin et al. 2012) prevented the DHK-induced I_A inhibition and largely reduced the DHK-evoked I_NMDA (n = 7, P = 0.89, paired t test, Fig. 4B, C, D and E). In fact, when results were plotted as percentage changes within each individual cell, a tendency for an enhancement of I_A was observed, which was significantly different to the inhibition observed in cells treated with MK801 (Fig. 4D). The DHK-mediated inhibition of I_A was almost completely blocked by the broad-spectrum glutamate receptor blocker kynurenic acid (138.68 ± 19.43 and 138.91 ± 23.37 pA pF^–1 before and after DHK, respectively, P > 0.8, paired t test, n = 11). Taken together, these results indicate that the eNMDAR–I_A functional coupling stands as a functionally relevant signalling mechanism by which astrocytes influence membrane excitability in MNC neurons.

Figure 4. Astrocyte GLT1 transporters influence I_A via activation of extrasynaptic NMDA receptors.

A, blockade of synaptic NMDA receptors (sNMDARs) with a combined application of 4-AP (2.5 mm) and bicuculline (BIC; 20 μm), followed by MK801 (20 μm) did not affect the DHK-evoked inhibition of I_A (right panel). B, blockade of eNMDARs (0 gly ACSF) blunted the DHK-induced shift in I_holding (left panel) and the inhibition of I_A (right panel). Mean I_A density (C), mean percentage I_A change (D), and mean shift in I_holding density (E) evoked by DHK are shown for 4-AP/BIC/MK801 (n = 9) and 0 glycine (n = 7) groups. *P < 0.05, paired t test; **P < 0.01, unpaired t test; #P < 0.05, unpaired t test.

Blunted evoked NMDA–I_A coupling in dehydrated rats

We recently showed that during dehydration, a condition known to lead to astrocyte process retraction and diminished astroglial ensheathment of MNCs (Hatton et al. 1984; Tweedle & Hatton, 1984; Tasker et al. 2002), GLT1-mediated glutamate uptake was blunted, resulting in elevated ambient glutamate levels, enhanced eNMDAR activation and increased MNC firing activity (Fleming et al. 2011). Thus, to further assess the contribution of astrocytes to the regulation of the eNMDA–I_A functional coupling, we repeated a set of experiments in MNCs from 48 h-dehydrated rats, in which we assessed changes in I_A magnitude following transient activation of NMDARs. We found that while transient NMDAR activation still evoked a significant inhibition of I_A, this effect was significantly blunted when compared to control rats subjected to the same experimental protocol (i.e. data points shown in Fig. 2B; 2-way ANOVA = experimental group × I_A sweep; F_group: 5.6, P < 0.0001; sweep: 21.2, P = 0.0001; interaction: F = 1.8, P = 0.03; Fig. 5). No differences in the magnitude of the evoked I_NMDA were observed between MNCs in control and dehydrated rats.

Figure 5. Blunted evoked NMDA–I_A coupling in dehydrated rats.

Plot of the mean percentage NMDA-evoked inhibition of I_A amplitude as a function of time obtained from MNCs (n = 11) recorded from dehydrated (48 h water-deprived) rats. Representative traces of I_A recorded at the time points indicated are shown. *P < 0.05 vs. baseline I_A (Dunnett's post hoc test).

The eNMDAR–I_A functional coupling requires an increase in intracellular calcium and activation of PKC

We found that inhibition of I_A, evoked either by bath-applied DHK or by a transient NMDA puff, was largely blunted in neurons dialysed with the fast Ca²⁺ chelator BAPTA (10 mm) (Fig. 6A, B, C and D). No differences in the magnitude of the evoked I_NMDA or I_DHK were observed between MNCs dialysed with or without BAPTA (Fig. 6E). A similar block of the NMDA–I_A functional coupling was found in MNCs dialysed with the slower Ca²⁺ chelator EGTA (5 mm) (97.32 ± 16.67 and 91.35 ± 14.23 pA pF^–1, before and after NMDA, respectively, P > 0.41, paired t test, n = 10).

Figure 6. The eNMDAR-mediated inhibition of I_A is Ca²⁺ dependent.

A, representative trace showing a DHK-induced inward shift in I_holding and lack of I_A inhibition in an MNC dialysed with BAPTA (10 mm). B, a focal and brief application of NMDA (20 μm, 5 s) also failed to inhibit I_A in an MNC dialysed with BAPTA. I_A was isolated electronically by digital subtraction of currents evoked from two separate protocols, in which preconditioning steps of either −80 mV and −10 mV were used, to fully evoke, and to completely inactivate I_A, respectively. C, mean (left) and before–after (right) plots of I_A density (at 0 mV) before (open symbols) and after (filled symbols) DHK (300 μm) application in the presence of BAPTA in the patch pipette (n = 8). D, mean (left) and before–after (right) plots of I_A density (at 0 mV) before (open symbols) and after (filled symbols) NMDA (20 μm) application in the presence of BAPTA in the patch pipette (n = 6). E, mean NMDA and DHK-evoked current density with and without BAPTA in the patch pipette.

To determine if the NMDAR–[Ca²⁺]_i-mediated inhibition of I_A involved a Ca²⁺-dependent, protein kinase C (PKC)-dependent pathway, we carried out a similar experiment in which NMDARs were transiently activated in MNCs dialysed with the PKC inhibitor chelerythrine chloride (10 mm). We found that while NMDAR activation still significantly inhibited I_A magnitude (135.68 ± 25.49 and 125.74 ± 22.45 pA pF^–1, before and after NMDA, respectively, n = 9, P = 0.02 paired t test), the degree of inhibition was significantly smaller than that observed in MNCs under control conditions (see Fig. 2C): control −24.7 ± 7.0%; PKC: −6.0 ± 1.6%, n = 11 and 9, respectively, P = 0.03 unpaired t test).

Activation of eNMDAR enhances the steady-state voltage-dependent inactivation of I_A

We previously showed that an increased voltage-dependent inactivation of I_A was a main factor contributing to blunted I_A magnitude in hypothalamic neurons during hypertension (Sonner et al. 2008, 2011), a pathological condition known to also involve an increased NMDAR activity within the hypothalamus (Li & Pan, 2007; Li et al. 2008). Thus, we explored whether the eNMDAR–I_A functional coupling involved changes in I_A inactivation properties. To evoke variable amounts of I_A inactivation, neurons were depolarized using a command pulse to 10 mV, from a range of conditioning steps (−120 to −30 mV, 10 mV increments, 300 ms duration, Fig. 7A). Blockade of GLT1 transporters shifted the voltage dependence of inactivation of I_A towards a more hyperpolarized membrane potential (leftward shift, Fig. 7B), reflected also as a significant hyperpolarization of the V_½ inactivation potential (P < 0.01, paired t test, Fig. 7C). This is indicative of a higher degree of I_A inactivation at any given V_m.

Figure 7. Blockade of astrocyte GLT1 transporters increases I_A steady-state inactivation.

A, representative example of I_A following voltage-dependent inactivation protocols (inset) before and during application of DHK in an MNC. B, plot of normalized I_A amplitude vs. inactivating conditioning voltage steps from the same neuron shown in A, before and during DHK, showing a clear hyperpolarizing shift in the curve in DHK. C, mean (left) and before–after (right) plots of I_A half-inactivation potential before and during DHK-mediated activation of NMDARs (n = 6). D, mean DHK-mediated change in V_½ inactivation in the presence and absence of BAPTA in the patch pipette. ***P < 0.0001, unpaired t test; **P < 0.01, paired t test.

Finally, in MNCs dialysed with BAPTA, DHK failed to significantly shift the V_½ inactivation (−39.87 ± 6.5 mV and −35.01 ± 4.6 mV before and after DHK, respectively, P > 0.5, unpaired t test, n = 9, see Fig. 7D).

The eNMDAR–I_A coupling enhances repetitive firing discharge of MNCs

To determine the functional consequences of the GLT1–eNMDAR–I_A signalling, we measured repetitive firing activity properties of MNCs before and after GLT1 blockade. The mean input resistance of neurons in this subset of experiments was 569.2 ± 33.6 MΩ. We found that the input–output function following bath-applied DHK was significantly strengthened, resulting in an overall increased number of evoked action potentials as a function of the injected current (2-way ANOVA = drug × current injection; F_drug: 24.02, P < 0.0001; F_current: 36.55, P < 0.0001; interaction: F = 0.95, P = 0.5, n = 9, Fig. 8A and C). In a different set of recordings, we found that DHK effects were prevented in the presence of the A-type K⁺ channel blocker 4-AP (2.5 mm) (2-way ANOVA = drug × current injection; F_drug: 0.5, P = 0.5; F_current: 88.2, P < 0.0001; interaction: F = 0.5, P = 0.8, n = 6, Fig. 8D). Similarly, depolarizing steps of increasing duration (175 pA, from 200 ms to 1800 ms, 200 ms increment) evoked more spikes in the presence of DHK (2-way ANOVA = drug × current injection; F_drug: 24.0, P < 0.0001; F_duration: 36.6, P < 0.0001; interaction: F = 0.9, P = 0.5, n = 9, Fig. 8F and G), an effect prevented by 4-AP (2-way ANOVA = drug × current injection; F_drug: 0.24, P = 0.6; F_duration: 11.3, P < 0.0001; interaction: F = 0.02, P = 1.0, n = 6, Fig. 8H).

Figure 8. The eNMDAR–I_A functional coupling enhances MNC repetitive firing discharge.

A, sample of firing activity evoked in an MNC in control ACSF. B, sample traces from the same neuron (150 pA step) before and during application of DHK. Note the reduced delay to the first spike (duration between two arrows) and the increased number of evoked action potentials in DHK. C, plot of the mean number of evoked spike numbers vs. current step amplitude before and during DHK application (n = 9). D, similar plot to that in C but in the presence of 2.5 mm 4-AP (n = 6). E, bar graph summarizing the mean delay to the first evoked spike (150 pA step) in DHK in the presence and absence of 4-AP. F, sample of firing activity evoked by successively longer depolarizing pulses (200 ms increments) in an MNC in control ACSF. G, plot of the mean number of evoked spikes vs. pulse duration before and during DHK application (n = 9). H, similar plot to that in F but in the presence of 4-AP (n = 6). I, representative traces (175 pA, 400 ms step) before and during DHK application. J, bar graph summarizing the mean delay to the first evoked spike in DHK (200 ms step duration). *P < 0.05 vs. respective control (Bonferroni's post hoc test); #P < 0.05 vs. control (paired t test).

We found that the latency to the first evoked spike, a property in MNCs largely determined by the transient I_A activation (Bourque, 1988) was significantly shortened during DHK (P < 0.05, paired t test, Fig. 8E and J), an effect prevented in the presence of 4-AP (Fig. 8E). Spike threshold did not change significantly in the presence of DHK (Δ −2.2 ± 14.9 mV).

The increased evoked firing activity following NMDAR activation was also observed following a transient, focal application of NMDA (20 μm, 5 s, n = 17, Fig. 9A–C), which occurred concomitantly with a diminished delay to the first evoked spike (Fig. 9B, before NMDA: 0.13 ± 0.01 s; after NMDA: 0.09 ± 0.01 s, P < 0.01, paired t test).

Figure 9. Transient activation of NMDARs potentiates subsequently evoked firing activity.

A, sample of repetitive firing activity evoked before and after transient activation of NMDA receptors (20 μm, 3.5 s) in an MNC. I_inj: current injected. B, the evoked firing for the periods 1 and 2 and 3 from panel A are shown at an expanded time scale. Note the shorter delay to the first evoked spike during the second depolarization (2) and the partial recovery during the third depolarization (3). C, plot of mean percentage changes in the evoked number of spikes following transient NMDAR activation (n = 17). *P < 0.001 vs. basal depolarization; #P < 0.01 vs. 7.5 s (Bonferroni's post hoc test).

Discussion

We report here a novel signalling mechanism that involves interactions among three major components influencing neuronal and network activity: astrocytes, intrinsic membrane properties and synaptic mechanisms. Specifically, our studies show that astrocytes, via the activity of GLT1 transporters, influence the degree of activation of a functional coupling between eNMDARs and the voltage-gated K⁺ current I_A. Moreover, we show that this functional coupling enhances NMDAR efficacy, and increases hypothalamic neuronal activity in response to membrane depolarization.

Extrasynaptic NMDA receptor activation under astrocyte control inhibits the magnitude of I_A

Our primary finding in this work is a significant inhibition of I_A following NMDAR activation. Similar to other CNS neurons (Hardingham & Bading, 2010), we recently reported two molecularly and functionally distinct pools of NMDARs in MNCs: synaptic and extrasynaptic NMDARs (Fleming et al. 2011; Potapenko et al. 2013; Fig. 10). Synaptic NMDARs display a high affinity for glutamate, and rapidly desensitize following sustained activation. Conversely, eNMDARs display lower affinity, but do not desensitize in the presence of sustained glutamate levels. Our results support a major contribution of eNMDARs to the NMDAR–I_A coupling. Even though focal and bath-applied NMDA accessed both types of NMDARs, the rapid inactivation of sNMDARs probably minimized their participation in this phenomenon. Moreover, we recently showed that eNMDARs in hypothalamic neurons, including MNCs, are tonically activated by extracellular glutamate, whose levels are tightly controlled by astrocyte GLT1 glutamate transporters. Thus, blockade of GLT1 function resulted in a build-up of extracellular glutamate and activation of eNMDARs, leading to a persistent tonic current (tonic I_NMDA) and a concomitant increase in neuronal firing discharge (Fleming et al. 2011). Here, we found that, similar to direct NMDA application, astrocyte GLT1 blockade lead to I_A inhibition. The fact that this effect persisted when synaptic NMDARs were blocked (i.e. 4-AP/MK801 protocol) (Hardingham et al. 2002; Jung et al. 2008; Hardingham & Bading, 2010), but was blunted when eNMDARs were rendered inactive by removing the eNMDAR (but not sNMDAR) co-agonist glycine (Papouin et al. 2012), further supports a contribution of eNMDARs to the NMDA–I_A functional coupling. This is in agreement with a recent study in cultured hippocampal neurons showing that activation of extrasynaptic (but not synaptic) NMDARs resulted in the downregulation of Kv4.2 channel subunits expression (Lei et al. 2010). However, whether activation of synaptic NMDARs during robust and synchronous synaptic activity can lead to I_A inhibition was not explored in this study. The importance of this mechanism is further supported by the fact that blockade of NMDARs per se increased the basal magnitude of I_A, suggesting that persistent activation of a small proportion of eNMDARs by ambient glutamate is sufficient to continuously suppress I_A activity.

Figure 10. Schematic model showing the proposed functional coupling between eNMDARs and I_A, via Ca²⁺-dependent activation of PKC.

The model depicts the role of astrocyte GLT1 transporters in regulating the ability of ambient extracellular glutamate (Glu) to activate eNMDARs, which in a Ca²⁺- and PKC-dependent manner results in inhibition of I_A.

It could be argued that the inhibition of I_A following prolonged NMDAR activation was due to changes in overall membrane conductance during sustained NMDAR activation. Several pieces of evidences argue against this, however, including: (a) that transient NMDAR activation leads to a similar I_A inhibition observed during sustained NMDAR activation; (b) the fact that the NMDA-mediated inhibition of I_A was blocked by different treatments (i.e. BAPTA, EGTA, PKC inhibition) in which the relatively large change in basal NMDA-mediated change in conductance was still present; and (c) a lack of a significant correlation between the magnitude of the evoked I_NMDA and the associated percentage changes in I_A. Taken together, these results show that activation of eNMDARs inhibits I_A, and underscore the ability of astrocytes to influence the degree of activation and efficacy of the eNMDAR–I_A coupling.

The eNMDAR–I_A functional coupling is Ca²⁺- and PKC-dependent

We found that the eNMDAR–I_A functional coupling was blocked when cells were dialysed with a slow (EGTA) or a fast-acting (BAPTA) Ca²⁺ chelator, supporting the contention that NMDAR- I_A inhibition was dependent on a rise of intracellular Ca²⁺. The fact that similar results were obtained with both Ca²⁺ chelators suggests that the functional coupling did not involve a local Ca²⁺ microdomain. This argues against an intimate physical interaction between NMDARs and A-type channels, and suggests that brief and transient activation of sNMDARs during physiological levels of synaptic activity may not generate a sufficiently large rise in global Ca²⁺ to inhibit I_A. However, alternative approaches such as co-immunoprecipitation are needed to more conclusively determine whether eNMDARs or sNMDARs and I_A channel subunits are physically coupled in these neurons. We also found the NMDA–I_A functional coupling to be blunted in MNCs dialysed with a PKC blocker. Given previous studies showing eNMDAR-mediated Ca²⁺-dependent activation of PKC (Sun & Liu, 2007), and that Ca²⁺-dependent PKC activity influenced I_A properties, including inactivation (Ritter et al. 2012), it is then reasonable to speculate that the NMDAR–I_A coupling in SON neurons involved a Ca²⁺-dependent activation of PKC. While NMDAR activation did not affect I_A voltage-dependent activation properties, it increased I_A voltage-dependent steady-state inactivation at membrane potentials (V_m) ≥−80 mV. Thus, as we recently reported in presympathetic PVN neurons of hypertensive rats (Sonner et al. 2008), the enhanced steady-state inactivation at these V_m values could account, at least in part, for the diminished I_A availability and magnitude following NMDAR activation. It is important to note, however, that the NMDA–I_A functional coupling was still observed when steady-state inactivation was completely removed (see Fig. 6A), suggesting that a factor other than inactivation (e.g. A-type channel internalization; Kim et al. 2007), could also contribute to the NMDAR-mediated I_A inhibition.

Functional implications of the NMDAR–I_A coupling

Besides influencing interspike intervals during repetitive firing, activation of I_A in MNCs underlies a transient outward rectification that delays the onset of spiking upon membrane depolarization (Hu & Bourque, 1992; Luther & Tasker, 2000). Within this context, we found that astrocyte GLT1 blockade or transient pharmacological NMDAR activation both diminished the delay to spike onset, increasing in turn the number of evoked spikes. These effects were blocked by the A-type K⁺ channel blocker 4-AP. Pharmacological block of I_A with 4-AP would be expected per se to diminish the delay to the first evoked spike (Fisher et al. 1998), and to increase firing activity. In our studies, however, we did not observe a significant difference in these parameters when comparing independent recordings obtained from different cells in the presence or absence of 4-AP. This was probably due to cell–cell variability in resting membrane potential and cell input resistance. This limitation in our experimental approach, however, did not preclude our ability to clearly demonstrate that 4-AP efficiently occluded DHK effects on firing activity. Similar results and conclusions to ours were obtained in comparable studies that evaluated the effects of other neurotransmitters/receptors, including activation of metabotropic glutamate receptors (Schrader & Tasker, 1997) and angiotensin II (Ma et al. 2006).

Taken together, our results indicate that in addition to its effect via direct membrane depolarization, the negative functional coupling to I_A further contributes to NMDA-mediated increased membrane excitability and firing discharge in MNCs. Moreover, our studies show that the efficacy of this process is modulated by the local astrocyte microenvironment. In this sense, we recently showed that activity-dependent retraction of astrocyte processes during dehydration resulted in blunted GLT1 glutamate buffering and enhanced activation of eNMDARs (Fleming et al. 2011). Here, we show that this physiological challenge also led to an enhanced tonic activation of the NMDAR–I_A signalling pathway in MNCs, resulting in (a) a basally diminished I_A magnitude, and (b) an occluded response to an additional NMDAR activation. These results together support an increased tonic occupancy of eNMDARs and a concomitant inhibition of I_A due to build-up of extracellular glutamate secondary to blunted GLT1 efficacy during dehydration (Fleming et al. 2011).

The NMDA–I_A coupling may not only have functional consequences at the single cell level, but also at the entire population level. Similar to the hippocampus (Kim et al. 2005; Chen et al. 2006), A-type K⁺ channels in hypothalamic neurons modulate dendritic excitability, and we showed that their inhibition facilitates dendritic Ca²⁺ signalling and propagation (Sonner et al. 2011). A rise in dendritic Ca²⁺ in MNCs results in the local release of oxytocin and vasopressin (Ludwig & Leng, 2006), a phenomenon that can be locally initiated by NMDAR activation (de Kock et al. 2004; Son et al. 2013). Thus, by inhibiting I_A and enhancing Ca²⁺ signalling, NMDAR activation would further potentiate dendritic release. This effect is expected to have important functional implications in hypothalamic information processing. Dendritic peptide release is not only involved in the autoregulation and optimization of MNC firing activity (Ludwig & Leng, 1997; Gouzenes et al. 1998), but we recently showed this to be a critical mechanism mediating neurosecretory-to-presympathetic interpopulation crosstalk during physiological challenges (Son et al. 2013). Thus, it will be important in future studies to determine the impact of the bidirectional NMDA–I_A functional coupling in the generation of multimodal homeostatic hypothalamic responses.

In summary, our studies support a novel signalling mechanism involving a functional coupling between extrasynaptic NMDARs and transient A-type K⁺ channels, whose degree of activation is modulated by local astrocytes, and which plays an important role in the regulation of magnocellular neurosecretory neuronal excitability.

Glossary

4-AP

4-aminopyridine

AP5

dl-2-amino-5-phosphonopentanoic acid

DHK

dihydrokainic acid

DIC

diffraction interference contrast

DNQX

6,7-dinitroquinoxaline-2,3-dione

eNMDARs

extrasynaptic NMDARs

I_A

A-type potassium current

I_holding

holding current

I_KDR

delayed-rectifier current

I_NMDA

NMDA current

MNC

magnocellular neurosecretory cell

NMDAR

NMDA receptor

PKC

protein kinase C

sNMDAR

synaptic NMDAR

SON

supraoptic nucleus

SR

series resistance

V_m

membrane potential

Additional information

Competing interests

The authors declare no competing financial interests.

Author contributions

Both authors contributed to the conception and design of the experiments, collection, analysis and interpretation of data, and drafting the article.

Funding

This work was supported by a National Heart, Lung, and Blood Institute Grant NIH HL112225 to J.E.S.