Activation of Neurotensin Receptor 1 Facilitates Neuronal Excitability and Spatial Learning and Memory in the Entorhinal Cortex: Beneficial Actions in an Alzheimer's Disease Model

NT induces persistent enhancement of neuronal excitability in the superficial layers of the EC via activation of NTS1

Autoradiographic studies have shown that NT-binding sites are detected in the EC, especially in layer II of the EC (Wolf et al., 1994, 1995; Lahti et al., 1998; Hamid et al., 2002). Accordingly, we examined the effects of NT on neuronal excitability by recording AP firing from the principal neurons in the medial EC of rats using whole-cell current-clamp configuration. The intracellular solution contained K⁺-gluconate, and the extracellular solution was supplemented with 10 μm DNQX, 50 μm dl-APV to block glutamatergic, and 10 μm bicuculline and 2 μm CGP55845 to block GABAergic transmission. In the superficial layers (layer II/III), stellate and pyramidal neurons are the two major neuronal types in layer II (Alonso and Klink, 1993; Canto et al., 2008), whereas pyramidal neurons are the principal neuronal type in layer III (Dickson et al., 1997). Application of NT (0.25 μm) for 5 min increased the AP firing frequency by 8.93 ± 1.94-fold in stellate (n = 7, p = 0.004; Fig. 1A₁,A₂) and 7.58 ± 1.71-fold in pyramidal (n = 8, p = 0.003; Fig. 1B₁,B₂) neurons of layer II. The effect of NT lasted for at least 60 min after wash in NT-free extracellular solution in our whole-cell recording configuration. We also examined the effects of NT on AP firing frequency in the pyramidal neurons of layer III and layer V. Application of NT (0.25 μm) induced a much smaller increase in AP firing frequency in the pyramidal neurons of layer III (0.66 ± 0.26-fold increase, n = 8, p = 0.04; Fig. 1C₁,C₂) but failed to significantly enhance the AP firing frequency in layer V pyramidal neurons (0.25 ± 0.13-fold increase, n = 9, p = 0.09; Fig. 1D₁,D₂). These results demonstrate that NT selectively facilitates neuronal excitability in the superficial layers (especially layer II) of the EC, consistent with previous and our immunohistological results (see below) showing that a high density of NT receptors exists in layer II of the EC (Wolf et al., 1994; Lahti et al., 1998; Hamid et al., 2002). Because stellate neurons are the principal neuronal type (∼69%) in layer II (Alonso and Klink, 1993), we recorded from the stellate neurons to further determine the underlying ionic and signaling mechanisms for the rest of the experiments. The EC₅₀ of NT was measured to be 41 nm (effective concentration: 0.01–1 μm; Fig. 1E). Because the maximal effect of NT was observed at 0.25 μm, we used NT at this concentration for the rest of the experiments unless stated otherwise.

Whereas NT has been reported to modulate the release of glutamate (Yin et al., 2008; Kadiri et al., 2011) and GABA (O'Connor et al., 1992; Rakovska et al., 1998; Petkova-Kirova et al., 2008) in some brain regions, NT-induced persistent increase in AP firing frequency is unlikely to be mediated by glutamate or GABA because the extracellular solution contained blockers for glutamatergic and GABAergic receptors. Inclusion of MCPG (1 mm), a broad-spectrum metabotropic glutamate receptor (mGluR) blocker, in the extracellular solution failed to alter NT-induced facilitation of AP firing (n = 6, p = 0.48 vs NT alone; Fig. 1F), excluding the involvement of mGluRs. NT has also been reported to modulate the releases of acetylcholine (Ach) (Lapchak et al., 1990, 1991; Rakovska et al., 1998; Petkova-Kirova et al., 2008) and dopamine (DA) (Faggin et al., 1990; Reyneke et al., 1992). Pretreatment of slices with and continuous bath application of atropine (10 μm, muscarinic receptor blocker) and mecamylamine (100 μm, nicotinic receptor blocker) failed to change NT-induced facilitation of AP firing frequency (n = 6, p = 0.44 vs NT alone; Fig. 1F), suggesting that it is unlikely that NT facilitates neuronal excitability via releasing Ach. Furthermore, application of SCH23390 (10 μm, D₁-like receptor antagonist) and sulpiride (100 μm, D₂-like receptor antagonist) in the same fashion did not alter NT-induced enhancement of AP firing frequency (n = 5, p = 0.99 vs NT alone; Fig. 1F). These results together indicate that it is unlikely that NT-induced facilitation of AP firing is mediated indirectly by modulating the release of Ach and DA. Finally, application of NT in the pure extracellular solution without any blockers for any receptors still enhanced the AP firing frequency to the same level (8.4 ± 3.1-fold increase, n = 6, p = 0.04; data not shown) supporting a physiological role for NT-induced facilitation of neuronal excitability in the EC. Bath application of the active NT fragment (NT8–13, 0.25 μm) increased the AP firing frequency to the same level as NT (n = 6, p = 0.35 vs NT alone; Fig. 1F), whereas application of the inactive fragment (NT1–8, 0.25 μm) failed to increase AP firing frequency (n = 7, p = 0.093 vs baseline; p < 0.0007 vs NT alone; Fig. 1F), suggesting that NT acts via activation of NT receptors.

We then tested the roles of NT receptors (NTS1 and NTS2) in NT-induced enhancement of neuronal excitability. Bath application of the selective NTS1 blocker, SR48693 (1 μm, for 10 min), did not significantly alter AP firing frequency (98 ± 6% of control, n = 6, p = 0.68; data not shown), suggesting that this compound by itself does not modulate AP firing. We then pretreated the slices with SR48692 (1 μm) for ∼20 min, and the same concentration of SR48692 was continuously bath-applied. In the presence of SR48692, application of NT increased AP firing frequency to a significantly smaller level (n = 6, p = 0.001 vs NT alone, two-way ANOVA; Fig. 2A), suggesting the requirement of NTS1 receptors. Because PD149163 is a small-molecule NTS1 agonist that crosses the blood–brain barrier (Feifel et al., 2004, 2010; Azmi et al., 2006), we tested the effects of PD149163 on neuronal excitability recorded from stellate neurons in the EC. Bath application of PD149163 (0.25 μm) robustly increased the AP firing frequency (n = 10, p = 0.002; Fig. 2A), further supporting the requirement of NTS1 receptors. However, pretreatment of slices with and continuous bath application of NTS2 receptor antagonist levocabastine (30 μm) was without effect on NT-induced facilitation of AP firing frequency (n = 10, p = 0.35 vs NT alone; Fig. 2A), suggesting that NTS2 is not necessary for NT-induced facilitation of neuronal excitability. We further tested the roles of NT receptors in NT-induced facilitation of neuronal excitability by using NT receptor KO mice as previously (Maeno et al., 2004). Application of NT did not significantly increase the AP firing frequency in NTS1 KO mice (n = 12 cells from 4 mice, p = 0.98; Fig. 2B), whereas application of NT still robustly increased the AP firing frequency in NTS2 KO (n = 12 cells from 4 mice, p < 0.0001; Fig. 2B) and WT (n = 10 cells from 3 mice, p < 0.0001; Fig. 2B) mice, further confirming the involvement of NTS1. Consistent with our electrophysiological data, immunoreactivity for NTS1 was mainly detected in the superficial layers (especially layer II) of the EC (Fig. 2C). Western blot showed a single protein band of ∼60 kDa in the lysates of the EC prepared from rats and WT mice, corresponding to the molecular mass of NTS1 receptors (Martorana et al., 2006), whereas no band was detected in this range of molecular mass for the lysates of the EC from NTS1 KO mice (Fig. 2D).

NTS1 receptors are coupled to G-proteins. We next tested the roles of G-proteins in NT-induced facilitation of neuronal excitability. Intracellular dialysis of G-protein inhibitor GDP-β-S (2 mm) blocked NT-induced enhancement of AP firing frequency (n = 7, p = 0.79; Fig. 2E). The major intracellular signal activated by NTS1 is PLCβ, which hydrolyzes PIP₂ to produce IP₃ to facilitate intracellular Ca²⁺ release and diacylglycerol to activate PKC. We accordingly examined the roles of this pathway. Slices were pretreated with U73122 (10 μm), a PLC inhibitor, and the same concentration of U73122 was continuously bath-applied. Under these conditions, application of NT failed to change the AP firing frequency significantly (n = 6, p = 0.12; Fig. 2E), whereas application of U73343 (10 μm), the inactive analog, in the same fashion did not significantly alter NT-induced enhancement of AP firing frequency (n = 6, p < 0.0001 vs baseline; Fig. 2E). Similarly, application of edelfosine (10 μm), another PLC inhibitor, blocked NT-induced facilitation of AP firing frequency (n = 9, p = 0.09; Fig. 2E). We then used PLCβ KO mice to further confirm our results. Among the 4 isoforms of PLCβ (PLCβ1–4), PLCβ1 is expressed in the hippocampal formation (Watanabe et al., 1998). We therefore used PLCβ1 KO mice as previously described (Deng et al., 2006; Lei et al., 2007) to examine the roles of PLC in NT-induced facilitation of AP firing. Application of NT robustly enhanced the AP firing frequency (n = 10 cells, p < 0.0001) in slices cut from 3 WT mice, whereas it increased the AP firing frequency to a significantly smaller level (n = 13 cells) in slices cut from 4 PLCβ1 KO mice (p < 0.0001 vs WT mice; Fig. 2E). Together, these results demonstrate that NT-mediated enhancement of neuronal excitability requires the function of PLCβ1. We then tested the roles of PKC and intracellular Ca²⁺ release in NT-induced facilitation of neuronal excitability. Slices were pretreated with GF109203X (1 μm) and Ro318220 (1 μm) separately, and the same concentrations of the PKC inhibitors were continuously bath-applied. Under these circumstances, application of NT failed to increase AP firing frequency significantly in the presence of GF109203X (n = 6, p = 0.2; Fig. 2F) or Ro318220 (n = 6, p = 0.83; Fig. 2F), demonstrating the involvement of PKC. Because PKC is an upstream activator of MEK/ERK in some cell types (Hu and Gereau, 2003), we also tested whether MEK/ERK is required for NT-induced increase in AP firing frequency. Slices were pretreated with the selective MEK/ERK inhibitor PD98059 (50 μm) for >2 h and the extracellular solution continued to contain the same concentration of PD98059. Under these circumstances, bath application of NT still enhanced the AP firing frequency to a comparable level (n = 5, p = 0.02; Fig. 2F), suggesting that MEK/ERK activity is not required for NT-induced facilitation of neuronal excitability. To test the involvement of IP₃ receptors, we included xestospongin C (5 μm, IP₃ receptor inhibitor) in the recording pipettes and waited for ∼20 min after the formation of whole-cell configuration. In this condition, application of NT still increased the AP firing frequency to a comparable level (n = 6, p = 0.83 vs NT alone; Fig. 2F). Similarly, depletion of intracellular Ca²⁺ stores by including thapsigargin (10 μm) in the pipettes for 20 min did not significantly alter NT-induced facilitation of AP firing frequency (n = 6, p = 0.41 vs NT alone; Fig. 2F). These results suggest that intracellular Ca²⁺ release is not required for NT-induced increases in AP firing. We then included BAPTA (10 mm) in the recording pipettes to chelate intracellular Ca²⁺. In the presence of BAPTA, bath application of NT increased the AP firing frequency to a significantly smaller level (n = 7, p = 0.007 vs NT alone; Fig. 2F). Because NT still significantly increased the AP firing frequency in the extracellular solution without Ca²⁺ (prepared by replacing extracellular Ca²⁺ with the same concentration of Mg²⁺ and adding 1 mm EGTA, n = 9, p = 0.013; Fig. 2F), one explanation for the effect of BAPTA is that a certain basal level of intracellular Ca²⁺ is required for the activation of signaling molecules, such as PLCβ and PKC.

NT-induced facilitation of AP firing is generated by membrane depolarization

We recorded RMPs in the presence of TTX (0.5 μm) to block AP firing. Application of NT generated membrane depolarization (control: −57.7 ± 1.2 mV, NT: −53.2 ± 2.0 mV, n = 6, p = 0.006; Fig. 3A) and significantly increased the input resistance assessed by injection of a negative current (−50 pA for 500 ms) before (115 ± 6 mΩ) and after (133 ± 6 mΩ) NT application (n = 6, p = 0.002; Fig. 3A, inset). We then recorded the HCs at −55 mV, a potential close to the RMPs of these neurons. Application of NT generated a net inward HC (−44.5 ± 5.4 pA, n = 16, p < 0.001; Fig. 3B), indicating that NT increases neuronal excitability by generating membrane depolarization.

Cationic channels are not involved in NT-induced membrane depolarization

We probed the ionic mechanisms whereby NT generates membrane depolarization. NT could cause depolarization by opening a cationic conductance and/or by inhibiting a background K⁺ conductance. We accordingly tested these possibilities by replacing the intracellular K⁺ with Cs⁺. Application of NT failed to induce an increase in inward HCs when intracellular K⁺ was replaced with Cs⁺ (4.6 ± 1.9 pA, n = 7, p = 0.06; Fig. 3B), suggesting the involvement of K⁺ channels. We further examined whether there is a role for cationic channels by performing the following experiments. First, we replaced extracellular NaCl with NMDG-Cl and Ca²⁺ with Mg²⁺ because, if cationic channels are involved, influx of extracellular Na⁺ and Ca²⁺ would be the ions to mediate the inward currents. However, in this condition, application of NT still induced a comparable inward HC (n = 8, p = 0.45 vs the inward HCs in control condition; Fig. 3B). Second, we examined the effects of other cationic channel blockers. Pretreatment of slices with and continuous bath application of SKF96365 (50 μm, an inhibitor of receptor-operated cationic channels, n = 6, p = 0.85), Gd³⁺ (10 μm, n = 7, p = 0.37), La³⁺ (10 μm, n = 7, p = 0.65), or ZD7288 (20 μm, an inhibitor of Ih channels, n = 6, p = 0.57), failed to exert significant effects on NT-induced increases in inward HCs, demonstrating that NT-induced depolarization is not mediated by activation of a cationic conductance (Fig. 3C).

NT has been reported to inhibit Na⁺/K⁺ pump (López Ordieres and Rodríguez de Lores Arnaiz, 2000), and this pump normally creates a concentration gradient by moving 3 Na⁺ out of the cell and 2 K⁺ into the cells. NT-mediated inhibition of this pump should create a net accumulation of Na⁺ inside the cells resulting in depolarization. Because extracellular K⁺ is necessary for the function of the Na⁺/K⁺ pump, we replaced the extracellular K⁺ (3.5 mm) with the same concentration of Na⁺. Instead of reducing the inward HCs, removal of extracellular K⁺ resulted in a significant increase in HCs (−83.5 ± 18.8 pA, n = 6, p = 0.012 vs NT alone, Fig. 3C), suggesting that NT-induced depolarization is unlikely the result of the inhibition of Na⁺/K⁺ pump but may be relevant to K⁺ movement across the membrane (see below).

NT generates membrane depolarization by inhibiting TREK-2 channels

We further probed the involvement of K⁺ conductance. We used a ramp protocol (from −140 to 0 mV) to construct the voltage-current relationship before and just after the application of NT. The intracellular solution contained K⁺-gluconate, and the extracellular K⁺ concentration was 3.5 mm. Under these circumstances, application of NT induced a current with a reversal potential of −86.7 ± 1.5 mV (n = 5; Fig. 3D). The measured reversal potential was close to the calculated K⁺ reversal potential (−85.4 mV) in our recording condition further confirming that NT increases neuronal excitability by inhibiting a background K⁺ conductance. We next characterized the properties of the involved K⁺ channels. NT-induced increases in inward HCs were not significantly changed by including tetraethylammonium (TEA, 10 mm, n = 9, p = 0.66), Cs⁺ (3 mm, n = 6, p = 0.51), or 4-aminopyridine (4-AP, 2 mm, n = 6, p = 0.78) in the extracellular solution (Fig. 3C). Application of the selective inward rectifier K⁺ channel inhibitor tertiapin-Q (100 nm) did not significantly alter NT-induced changes in HCs (n = 7, p = 0.97; Fig. 3C). Furthermore, NT-induced increases in inward HCs were not sensitive to extracellular application of Zn²⁺ (100 μm, n = 8, p = 0.63; Fig. 3C).

The above results prompted us to test the role of the two-pore domain K⁺ (K_2P) channels because K_2P channels are involved in controlling RMPs and usually insensitive to the classic K⁺ channel blockers (Lesage and Lazdunski, 2000; Bayliss et al., 2003; Lesage, 2003). Because some K_2P channels are sensitive to Ba²⁺, we tested the roles of Ba²⁺ in NT-induced increases in inward HCs. Application of Ba²⁺ (3 mm) alone generated an inward HC (−78.1 ± 18.2 pA, n = 9, p = 0.003; Fig. 3E), and it also significantly inhibited NT-induced inward HCs (−20.1 ± 2.1 pA, n = 9, p = 0.001 vs NT alone; Fig. 3E), suggesting the involvement of Ba²⁺-sensitive K_2P channels.

Among the K_2P channels, TASK-1 (Han et al., 2002), TASK-3 (Kim et al., 2000; Han et al., 2002), TREK-1 (Fink et al., 1996; Ma et al., 2011), TREK-2 (Bang et al., 2000; Han et al., 2002), and TWIK-1 (Lesage et al., 1996) and TRESK (Sano et al., 2003; Kang et al., 2004) are sensitive to Ba²⁺. TRESK channels are unlikely to be involved because they are expressed in the spinal cord and other organs but not in the brain (Sano et al., 2003; Kang et al., 2004). Among the remaining 5 K_2P channels, the EC expresses higher densities of TASK-1 and TREK-2 (Deng et al., 2009). Because TASK channel currents show GHK-type rectification (Lesage and Lazdunski, 2000; Bayliss et al., 2003) and fitting of the NT-induced net I-V curve by GHK equation demonstrated remarkable deviation (Fig. 3D, inset), suggesting that TASK-1 channels are unlikely to be the target for NT, we focused on TREK-2 channels by using TREK-2 KO mice as described previously (Guyon et al., 2009). Application of NT induced an inward HC (−38.8 ± 6.1 pA, n = 9 cells; data not shown) and robustly increased the firing frequency of APs (n = 9 cells, p < 0.001; Fig. 3F) in slices cut from 3 WT mice, whereas it failed to induce an inward HC (−3.7 ± 2.5 pA, n = 9 cells, p = 0.19; data not shown) and did not increase AP firing frequency (n = 9 cells, p = 0.32; Fig. 3F) in slices cut from 3 TREK-2 KO mice. However, the RMPs of stellate neurons in TREK-2 KO mice (−59.7 ± 1.1 mV, n = 9 cells from 3 mice) were not significantly different (p = 0.25) from those of WT mice (−61.4 ± 0.9 mV, n = 9 cells from 3 mice; data not shown). Together, these data indicate that NT facilitates neuronal excitability by inhibiting TREK-2 channels.

Activation of NTS1 inhibits TREK-2 currents via PLC/PKC pathway in HEK293 cells cotransfected with NTS1 and TREK-2 channels

We further probed the involvement of TREK-2 channels by using whole-cell recordings from the HEK293 cells expressing NTS1 and TREK-2 channels (Deng et al., 2007, 2009; Xiao et al., 2009). Bath application of NT (0.25 μm) induced an inward HC recorded at −55 mV in HEK293 cells cotransfected with NTS1 and TREK-2 channels (n = 8, p = 0.002; Fig. 4A,B). However, bath application of NT (0.25 μm) did not change the HCs recorded at −55 mV in HEK293 cells transfected with TREK-2 channels alone (n = 5, p = 0.82; Fig. 4B) or NTS1 alone (n = 6, p = 0.86; Fig. 4B). We further determined the roles of PLC, PKC, and intracellular Ca²⁺ release in NT-mediated depression of TREK-2 channels in HEK293 cells cotransfected with NTS1 and TREK-2 channels. NT-induced inward HCs were significantly reduced in the presence of PLC inhibitor (U73122, 10 μm, n = 8, p < 0.0001) compared with those in the presence of the inactive analog (U73343, 10 μm, n = 7; Fig. 4C). Similarly, treatment of the cells with another PLC inhibitor (edelfosine, 10 μm, n = 8, p = 0.001) significantly reduced NT-induced inward HCs compared with control (Fig. 4C). Furthermore, NT-induced inward HCs were significantly reduced by pretreatment of transfected cells with the selective PKC inhibitor (GF109203X, 1 μm, n = 5, p = 0.007) but were unaffected by intracellular application of xestospongin C (1 μm, IP₃ receptor inhibitor, n = 8, p = 0.08; Fig. 4C) or thapsigargin (10 μm, n = 5, p = 0.051; Fig. 4C). However, intracellular application of BAPTA (10 mm) significantly reduced NT-induced inward HCs (n = 6, p = 0.008; Fig. 4C). These results demonstrate that NT-induced depolarization requires the functions of PLC and PKC. One explanation for the effect of BAPTA is that intracellular Ca²⁺ may be required for the functions of the signaling molecules, such as PLC and PKC.

Application of NT to entorhinal slices increases serine phosphorylation of TREK-2 channels

The above experiments suggest that activation of NTS1 inhibits TREK-2 channels via PKC-dependent phosphorylation of TREK-2 channels. In accordance with our results, TREK-2 channels are phosphorylated and inhibited by PKC on serine 326 (S326) and serine 359 (S359) (Gu et al., 2002; Kang et al., 2006; Kréneisz et al., 2009). We next tested the roles of PKC-dependent phosphorylation of these two sites in NT-mediated depression of TREK-2 channels. Because mutation of S326 and S359 to alanine blocked PKC-mediated inhibition of TREK-2 channels (Kang et al., 2006; Kréneisz et al., 2009), we cotransfected HEK293 cells with NTS1 and the TREK-2 mutant in which both serine 326 and serine 359 were mutated to alanine (S326A/S359A) (Kang et al., 2006; Kréneisz et al., 2009). Application of NT to the HEK293 cells cotransfected with NTS1 and TREK-2 mutant induced significantly smaller HCs (−8.2 ± 1.4 pA, n = 7, p = 0.002 vs NT-induced HCs in the cells transfected with WT TREK-2; Fig. 4D,E). These data demonstrate that NT-induced depression of TREK-2 channels is mediated by PKC-dependent phosphorylation of TREK-2 channels on S326 and S359.

We further corroborated NT-induced phosphorylation of TREK-2 channels in the EC. We treated entorhinal slices for 5 min with or without NT in the absence or presence of the PKC inhibitor (GF109203X). Slices were then lysed, and the lysates were immunoprecipitated with TREK-2 antibody. The immunoprecipitates were resolved by SDS-PAGE and Western blotted with anti-phosphoserine antibody. As shown in Figure 4F, treatment of slices with NT for 5 min induced persistent increases in the phosphorylation of TREK-2 channels when the slices were kept in NT-free solution for ∼1 h (n = 3 experiments, p < 0.01 for each). Pretreatment of slices with GF109203X blocked NT-induced increases in the phosphorylation of TREK-2 channels (n = 3 experiments, p = 0.8). As a control, treatment of slices cut from WT mice with NT for 5 min induced phosphorylation of TREK-2 channels, whereas no band was detected in the samples prepared from the TREK-2 KO mice with the same experimental procedures (Fig. 4F). These results demonstrate that NT-induced depression of TREK-2 channels is mediated by PKC-dependent phosphorylation of TREK-2 channels.

NT enhances EPSP-AP coupling in layer II stellate neurons

All the above experiments regarding the effects of NT on AP firing were performed by injection of small positive current to elevate the RMP to just above threshold to induce AP firing. We next tested whether NT by itself could elevate the RMPs of stellate neurons to above the threshold to induce AP firing. In 10 stellate neurons, there was no spontaneous AP firing at rest. Application of NT induced AP firing in 6 of the 10 stellate neurons. The AP firing frequency reached maximal in ∼5 min after the beginning of NT application (141 ± 26 APs/min, n = 6, p = 0.003; Fig. 5A,B). Whereas the AP firing frequency was declined after wash in NT-free extracellular solution, there was still AP firing after wash for 1 h (34 ± 7 APs/min, n = 6, p = 0.004; Fig. 5A,B). Application of NT to the remaining 4 stellate neurons failed to induce AP firing but significantly elevated the RMPs (control: −58.7 ± 1.3 mV, NT: −55.2 ± 1.6 mV, n = 4, p = 0.012; data not shown). We then tested the effects of NT on EPSP-AP coupling in layer II stellate neurons by placing a stimulation electrode in layer I to stimulate the glutamatergic inputs. Application of NT induced the depolarization of the baseline of the EPSPs (control: −59.5 ± 1.3 mV, NT: −54.9 ± 1.6 mV, n = 5, p = 0.0009; Fig. 5C,D) and facilitated EPSP-AP coupling (n = 5, p = 0.0003; Fig. 5C,D). To differentiate the presynaptic and postsynaptic effects of NT, we replicated the experiment by using Cs⁺-gluconate-containing intracellular solution to block the depolarizing effect of NT on stellate neurons. Under these circumstances, NT failed to altered the slope (101 ± 2% of control, n = 9, p = 0.55; Fig. 5E,F) of the EPSPs, suggesting that NT-mediated facilitation of EPSP-AP coupling was not mediated by modulating presynaptic glutamate release, but because of its depolarizing effect on postsynaptic stellate neurons.

NT facilitates spatial learning and memory in Barnes Maze Test

We next probed the roles of NT in the modulation of spatial learning and memory because layer II of the medial EC is predominated by “grid” cells (Sargolini et al., 2006; Hafting et al., 2008; Brandon et al., 2011; Burgalossi et al., 2011), which are essential for spatial navigation, and NT-induced excitation of these neurons likely plays an important role in spatial learning and memory. We examined the roles of NT in spatial learning and memory using the Barnes Maze Test. Although we previously used Morris Water Maze and measured the spatial learning and memory of animals (Deng et al., 2009), the advantage of Barnes Maze Test is that it is less physically taxing and less stressful to animals. After ∼15 min after microinjection each day, rats were trained for 5 consecutive days (3 trials per day). On day 12, a probe trial was conducted without the escape box to measure the memory retention ability of the rats (Fig. 6A,B). Both the escape latency (F_(4,36) = 10.29, p < 0.001; Fig. 6C) and travel distance (F_(4,36) = 9.76, p < 0.001; Fig. 6D) of rats to locate the escape box were progressively reduced during the training sessions in control rats microinjected bilaterally with saline (0.9% NaCl, 1 μl/side) into the EC. Bilateral microinjection of NT at the dose of 0.01 nmol (1 μl/side) into the EC failed to significantly change the spatial learning and memory of the rats compared with that of the rats injected with saline (latency: F_(1,16) = 1.55, p = 0.23; Fig. 6C; distance: F_(1,16) = 1.21, p = 0.29; Figure 6D). However, bilateral microinjection of NT at doses of 0.1 nmol (1 μl/side) and 0.3 nmol (1 μl/side) into the EC significantly decreased the escape latency (0.1 nmol: F_(1,16) = 5.46, p = 0.03; 0.3 nmol: F_(1,18) = 26.06, p < 0.001; Figure 6C) and the distance traveled (0.1 nmol: F_(1,16) = 13.89, p = 0.002; 0.3 nmol: F_(1,18) = 37.16, p < 0.0001; Figure 6D) compared with those of the rats injected with saline demonstrating that NT enhances spatial learning and memory. Figure 6B displays the typical traces of rats injected with saline (upper) and NT (lower) on day 1 (left) and day 5 (middle) and day 12 (probe trial, right). Furthermore, NT significantly increased the percentage of time spent in the target quadrant in the probe trial measured on day 12 (Fig. 6E). Together, these data indicate that NT facilitates spatial learning and memory of rats.

We then tested the involvement of NT receptors in NT-induced enhancement of spatial learning and memory by using the NT receptor KO mice. We initially tested the effects of NT on facilitation of spatial learning and memory in WT mice. Bilateral microinjection of NT (0.3 nmol) into the EC of WT mice significantly reduced the escape latency (F_(1,17) = 17.17, p < 0.001; Fig. 7A,B₁) and the distance traveled (F_(1,17) = 14.88, p = 0.001; Fig. 7B₂) compared with those of the WT mice injected with saline. Moreover, NT also significantly increased the percentage of time spent in the target quadrant in the probe trial measured on day 12 (Fig. 7A–B₃), indicating that NT facilitates spatial learning and memory of mice as well. We also tested the roles of PD149163, the small-molecule NTS1 agonist that can cross the blood–brain barrier, in spatial learning and memory in WT mice. Bilateral microinjection of PD149163 (0.3 nmol) into the EC of WT mice significantly reduced the escape latency (F_(1,17) = 12.35, p = 0.003; Fig. 7B₁) and the distance traveled (F_(1,17) = 11.67, p = 0.003; Fig. 7B₂) but significantly increased the percentage of time spent in the target quadrant in the probe trial (Fig. 7B₃) compared with those of the WT mice injected with saline suggesting that activation of NTS1 facilitates spatial learning and memory. Likewise, microinjection of NT (0.3 nmol) into the EC of NTS2 KO mice significantly reduced the escape latency (F_(1,15) = 6.32, p = 0.02; Fig. 7C₁) and the distance traveled (F_(1,15) = 21.27, p < 0.001; Fig. 7C₂) compared with those of the NTS2 KO mice injected with saline. Similarly, NT also significantly increased the percentage of time spent in the target quadrant in the probe trial for NTS2 KO mice (Fig. 7C₃) indicating that NTS2 is not required in NT-induced facilitation of spatial learning and memory. However, microinjection of the same dose of NT into the EC of NTS1 KO mice failed to change significantly the escape latency (F_(1,16) = 0.004, p = 0.95; Fig. 7C₁) and the distance traveled (F_(1,16) = 0.38, p = 0.55; Fig. 7C₂) compared with those of the NTS1 KO mice injected with saline. NT failed to significantly alter the percentage of time spent in the target quadrant in the probe trial in NTS1 KO mice (Fig. 7C₃) demonstrating that NT facilitates spatial memory via activation of NTS1. Furthermore, compared with the NTS2 KO mice injected with saline, NTS1 KO mice injected with saline showed significantly increased latency (F_(1,15) = 13.33, p = 0.0024; Fig. 7C₁) and lengthened distance (F_(1,15) = 9.31, p = 0.008; Fig. 7C₂). However, there was a statistically insignificant difference for the time spent in the target quadrant measured at day 12 between the NTS1 KO mice injected with saline and the NTS2 KO mice injected with saline (p = 0.24; Fig. 7C₃). These results suggest that endogenously released NT also facilitates spatial learning and memory, although it may exert less effect on memory retention (see Discussion).

PKC-induced depression of TREK-2 channels is required for NT-mediated facilitation of spatial learning

Because our cellular physiology data demonstrate that PKC-induced depression of TREK-2 channels is responsible for NT-induced enhancement of neuronal excitability, we next tested whether the functions of PKC and TREK-2 channels are required for NT-induced facilitation of spatial learning. Microinjection of NT after the injection of the selective PKC inhibitor, GF109203X (50 pmol), into the EC of the mice significantly lengthened the escape latency (F_(1,16) = 5.97, p = 0.027; Fig. 8A₁) and travel distance (F_(1,16) = 19.37, p < 0.0001; Fig. 8A₂), compared with those of the mice microinjected with vehicle (DMSO) and NT. The percentage of time spent in the target quadrant in mice injected with GF109203X and NT was also significantly attenuated compared with that of the mice injected with vehicle and NT (Fig. 8A₃). These data together demonstrate that PKC is required for NT-induced increases in spatial learning. Moreover, neither the escape latency (F_(1,14) = 0.22, p = 0.65; Fig. 8B₁) nor the travel distance (F_(1,14) = 0.11, p = 0.75; Fig. 8B₂) in TREK-2 KO mice injected with NT was significantly different from those of the TREK-2 KO mice injected with saline, whereas these parameters were still significantly reduced in WT mice injected with NT compared with those of the WT mice injected with saline (latency: F_(1,13) = 9.92, p = 0.008; Fig. 8B₁; distance: F_(1,13) = 14.63, p = 0.002; Figure 8B₂). Likewise, the percentage of time spent in the target quadrant in TREK-2 KO mice injected with NT was not significantly different from that of TREK-2 KO mice injected with saline, whereas the WT mice injected with NT spent more time in the target quadrant than those injected with saline (Fig. 8B₃). These data together indicate that TREK-2 channels are required for NT-mediated enhancement of spatial learning. Furthermore, compared with WT mice injected with saline, TREK-2 KO mice injected with saline failed to show significant differences in latency (F_(1,13) = 0.0003, p = 0.98; Fig. 8B₁), distance traveled (F_(1,13) = 2.6, p = 0.131; Fig. 8B₂), and time spent in the target quadrant measured at day 12 (p = 0.4; Fig. 8B₃).

Activation of NTS1 increases neuronal excitability and facilitates spatial memory in a mouse model of AD

We then performed experiments by using APP/PS1 mice, a mouse model of AD (Garcia-Alloza et al., 2006). We have shown previously that these mice exhibited plaques when they were older than 6 months (Dhawan and Combs, 2012). We therefore used the APP/PS1 mice at the age of 8–10 months. We first tested the effects of NT on AP firing recorded from layer II stellate neurons in the EC. The RMPs recorded from layer II stellate neurons in slices cut from APP/PS1 mice (−59.7 ± 0.7 mV, n = 11 cells from 4 mice) were not significantly different from those of WT mice (−60.2 ± 0.5 mV, n = 12 cells from 4 mice; data not shown). Bath application of NT (0.25 μm) robustly increased the firing frequency of APs recorded from stellate neurons in slices cut from APP/PS1 mice (4.7 ± 0.8-fold increase, n = 11 cells from 4 mice, p < 0.001; Fig. 9A). NT-induced facilitation of AP firing frequency recorded from APP/PS1 mice was not significantly different from that recorded from the age-matched WT mice (5.3 ± 1.2-fold increase, n = 12 cells from 4 mice, p = 0.55; Fig. 9A). Furthermore, bath application of NTS1 agonist PD149163 (0.25 μm) also vigorously facilitated AP firing frequency in APP/PS1 mice (5.6 ± 1.2-fold increase, n = 14 cells from 4 mice, p < 0.001; Fig. 9A). These data together demonstrate that activation of NTS1 also increases neuronal excitability in APP/PS1 mice. We then compared the spatial learning and memory ability of APP/PS1 mice with the WT mice using the Barnes Maze Test. Compared with the age-matched WT mice, APP/PS1 mice exhibited significantly enhanced latency (F_(1,15) = 15.39, p = 0.001; Fig. 9B₁), lengthened travel distance (F_(1,15) = 6.42, p = 0.023; Fig. 9B₂), and decreased percentage of the time spent in the target quadrant (p = 0.02; Fig. 9B₃). These results indicate that APP/PS1 mice showed impaired cognitive functions. We further tested the effects of PD149163 on spatial learning and memory in APP/PS1 mice. Bilateral microinjection of PD149163 (0.3 nmol) significantly reduced the escape latency (F_(1,14) = 28.89, p < 0.0001; Fig. 9B₁) and the distance traveled (F_(1,14) = 23.47, p < 0.001; Fig. 9B₂) but significantly increased the percentage of time spent in the target quadrant (Fig. 9B₃) demonstrating that PD149163 improves spatial learning and memory in APP/PS1 mice.

Ionic mechanisms underlying NTS1-mediated increase in neuronal excitability and facilitation of spatial learning

Application of NT increases neuronal excitability by depolarizing entorhinal neurons suggesting that activation of NTS1 receptors modifies the movement of ions across the membrane, which is usually conducted by ion channels or transporters. NT has been shown to inhibit Na⁺/K⁺ pump (López Ordieres and Rodríguez de Lores Arnaiz, 2000). The Na⁺/K⁺ pump creates a concentration gradient by moving 3 Na⁺ out and 2 K⁺ into cells and inhibition of this pump, thereby, should create a net accumulation of positive ions inside cells leading to depolarization. Extracellular K⁺ is a requisite for the function of the Na⁺/K⁺ pump. Our result that removal of extracellular K⁺ did not block but significantly increased NT-induced inward HCs demonstrates that NT increases neuronal excitability unlikely by inhibition of Na⁺/K⁺ pump. One explanation for NT-induced higher level of inward HCs in the absence of extracellular K⁺ is that depletion of extracellular K⁺ increases the driving force for K⁺ efflux via K⁺ channels leading to stronger inhibition of K⁺ channels by NT.

Our results demonstrate that NT facilitates neuronal excitability by inhibition of background K⁺ channels based on the following lines of evidence. First, NT significantly increased the input resistance, suggesting that NT inhibits a membrane conductance. Second, replacing intracellular K⁺ with Cs⁺ blocked NT-induced inward HCs. Third, the reversal potential of NT-induced currents had a reversal potential close to the K⁺ reversal potential. Nonetheless, the involved K⁺ channels are unlikely to be the inwardly rectifier K⁺ channels because the I-V curve of NT-sensitive currents did not show inward rectification, but a little outward rectification and application of the inward rectifier K⁺ channel blocker tertiapin-Q failed to significantly alter NT-induced inward HCs. Moreover, SCH23390 is an antagonist of D₁-like receptors as well as a blocker for inward rectifier K⁺ channels (Kuzhikandathil and Oxford, 2002). Our result that application of SCH23390 failed to alter NT-induced facilitation of AP firing frequency (Fig. 1F) further excludes the participation of inward rectifier K⁺ channels. The results that NT-induced inward HCs were insensitive to the classical K⁺ channel blockers, such as TEA, 4-AP, and extracellular Cs⁺, suggest that NT inhibits K_2P channels. NT has been reported to inhibit a type of K_2P channels (possibly TRESK and/or TASK-3) in spinal cord neurons (Kadiri et al., 2011). However, the channels inhibited by NT in the spinal cord neurons are sensitive to both Ba²⁺ and Zn²⁺ (Kadiri et al., 2011), which is in contrast to our results that NT-induced inward HCs in the EC were sensitive to Ba²⁺ but insensitive to Zn²⁺. These results suggest that NT inhibits a different type of K_2P channels in the EC. We provide further evidence demonstrating that NT increases neuronal excitability by inhibition of TREK-2 channels. Application of NT failed to increase the AP firing frequency in slices cut from TREK-2 KO mice, whereas application of NT still robustly increased AP firing frequency in slices cut from WT mice. Bath application of NT to HEK293 cells coexpressing NTS1 and TREK-2 channels induced an inward HC, whereas NT failed to modify the HCs in HEK293 cells transfected with either NTS1 receptors or TREK-2 channels alone. These results together indicate that NT inhibits TREK-2 channels to increase the excitability of entorhinal neurons.

Although we have demonstrated that NT-induced facilitation of neuronal excitability is unlikely because of its indirect effects on the releases of glutamate, GABA, DA, and Ach, the potential indirect contributions of other factors, such as the serotoninergic, purinergic, glycinergic, and other peptidergic systems, have not been examined. However, it seems less likely that the effects of NT would be related to serotonin or glycine because both serotonin (Lei, 2012) and glycine (Breustedt et al., 2004) inhibit neuronal excitability in the EC. The potential interaction between NT and other neurotransmitters still needs further investigation.

Signaling mechanisms underlying NTS1-induced facilitation of neuronal excitability and augmentation of spatial learning in the EC

Whereas NT interacts with NTS1, NTS2, and NTS3 receptors, our results demonstrate that NT increases neuronal excitability and facilitates spatial learning via activation of NTS1 receptors based on the results from both pharmacological experiments and KO mice. Activation of NTS1 increases the activity of PLC leading to elevation of intracellular Ca²⁺ release and activation of PKC. Our results demonstrate that NT-induced facilitation of neuronal excitability is related to intracellular Ca²⁺ level because intracellular application of BAPTA attenuated the effects of NT. However, blockade of intracellular Ca²⁺ release or depletion of extracellular Ca²⁺ failed to alter significantly NT-mediated facilitation of neuronal excitability. Different from NT-induced facilitation of neuronal excitability, Ca²⁺ influx via nonspecific cation and L-type Ca²⁺ channels is required for the persistent firing induced by brief depolarization in layer V entorhinal neurons (Egorov et al., 2002). One plausible explanation for the role of Ca²⁺ in NT-induced facilitation of neuronal excitability is that intracellular Ca²⁺ is required for the activation of Ca²⁺-dependent intracellular signals. Indeed, our results demonstrate that PKC, possibly Ca²⁺-dependent PKC, is necessary for NT-induced facilitation of neuronal excitability. We further demonstrate that NT induces persistent phosphorylation of TREK-2 channels. In HEK293 cells coexpressing NTS1 and the TREK-2 mutant in which serine 326 and serine 359 were mutated to alanine, NT failed to induce an inward HC, suggesting that PKC phosphorylates TREK-2 channels on these two sites. Consistent with our results, PKC-dependent phosphorylation of TREK-2 channels on these two amino acids by activation of muscarinic receptors (Kang et al., 2006) and AMP-activated protein kinase (Kréneisz et al., 2009) also results in inhibition of TREK-2 channels. PKC-dependent phosphorylation and inhibition of TREK-2 channels likely contribute to NT-mediated facilitation of spatial learning because inhibition of PKC or deletion of TREK-2 gene nullified the facilitatory effects of NT on spatial learning. However, the specific PKC isoforms involved in NT-mediated facilitation of neuronal excitability and spatial learning still await further investigation. In line with our results, it is well established that both PKC (Bonini et al., 2007; Dash et al., 2007; Sun and Alkon, 2010) and TREK-2 channels (Pan et al., 2003; Huang and Yu, 2008; Deng et al., 2009) are substrates of learning and memory.

Whereas our results demonstrate that both NTS1 and TREK-2 channels are required for NT-induced facilitation of spatial learning, the roles of NTS1 and TREK-2 channels in the modulation of spatial learning by endogenously released NT are still elusive. NTS1 KO mice showed increased escape latency and travel distance, suggesting that NTS1 activated by endogenously released NT is involved in the processes of spatial learning. However, the percentage of time spent in the target quadrant in the probe test measured on day 12 was not altered in NTS1 KO mice, suggesting that NTS1 may play a less important role in memory retention. Moreover, TREK-2 KO mice showed no differences in spatial learning and memory, although acute microinjection of NT into the EC of TREK-2 KO mice failed to facilitate spatial learning and memory, demonstrating a role of TREK-2 channels in exogenously applied NT. One explanation for the discrepancy of these results is that the roles of TREK-2 channels may be replaced by other K_2P channels in TREK-2 KO mice because the electrophysiological properties, such as the RMPs, were not changed in these mice. Evidence supporting this possibility comes from the observations that the RMPs of neurons in TREK-1 (Heurteaux et al., 2004) or TASK-1 (Aller et al., 2005) KO mice were not changed either. Moreover, although our results demonstrate that TREK-2 channel is the major target of NTS1 receptors, the possibility that an alternative target for endogenously released NT has been developed in TREK-2 KO mice cannot be excluded.

Significance of NTS1-induced enhancement of neuronal excitability and spatial learning in AD

There is increasing evidence demonstrating that the functions of NT are closely associated with the memory status of both animals and humans. For example, in healthy adult humans, NTS1 polymorphisms are closely associated with variance in the performance of working memory (Li et al., 2011). NT binding sites are varied as a function of age and cognitive status with decreased NT binding sites in the brains of aged and cognitively impaired rats (Rowe et al., 2006). Application of NT or NT receptor agonists improves memory (Azmi et al., 2006; Ohinata et al., 2007; László et al., 2010), whereas administration of NT receptor antagonists impairs cognitive processes (Tirado-Santiago et al., 2006). Whereas the effects of NT on memory are usually considered to be mediated by NTS1 (Azmi et al., 2006; Tirado-Santiago et al., 2006; Ohinata et al., 2007; László et al., 2010), reduced fear memory has also been observed in NTS2 KO mice (Yamauchi et al., 2007). Although these data generally support a role for NT in learning and memory, the cellular and molecular mechanisms whereby NT facilitates memory have not been determined. In the present study, we took the advantage that high density of NT receptors is expressed in the superficial layers of the EC, an indispensable structure involved in learning and memory. We first examined the effects of NT on neuronal excitability, and our results demonstrated that brief (5 min) application of NT persistently increased the AP firing frequency. This phenomenon resembles the long-term potentiation. Whereas long-term potentiation is generally considered as a cellular and molecular substrate for learning and memory, the eventual outcome of long-term potentiation is to increase the excitability of the postsynaptic neurons. NT-induced persistent increase in neuronal excitability bypasses the processes of synaptic transmission and directly leads to increases in neuronal excitability. Thus, it is reasonable to speculate that NT-induced persistent increase in neuronal excitability may represent a novel cellular mechanism involved in learning and memory.

Memory impairment is a characteristic symptom observed in many diseases, such as AD. Up to date, there have been few effective drugs that can relieve the memory loss of these diseases. Grid cell firing correlates to forming an internal spatial map and is driven by the polymodal input to the EC that occurs as an animal explores its environment (Sargolini et al., 2006; Hafting et al., 2008; Brandon et al., 2011; Burgalossi et al., 2011). We propose that NT may increase the probability of grid cell firing in response to that input, which promotes a more active circuit in the encoding of information about one's environment. Our data clearly show that application of NT or NTS1 agonist PD149163 facilitates spatial learning. We further showed that microinjection of PD149163 into the EC of APP/PS1 mice also significantly improved the memory status of these mice, highlighting a therapeutic value of this type of drugs in AD. Therefore, one immediate implication of our results is that NTS1 agonists may have the potential to be developed as cognition-enhancing drugs to relieve the symptoms of AD.

In conclusion, we found that activation of NTS1 receptors in the EC induces a persistent increase in AP firing and facilitates spatial learning. NT-induced facilitation of neuronal excitability and spatial learning require the functions of PLC and PKC and are mediated by phosphorylation and inhibition of TREK-2 channels. NT and NTS1 agonist PD149163 also induced persistent AP firing and facilitated spatial memory in APP/PS1 mice, suggesting that NT-mediated augmentation of learning and memory may have therapeutic potentials to relieve memory loss exhibited in AD.