Serotonin modulates the excitatory synaptic transmission in the dentate granule cells

Abstract

Serotonergic fibers from the raphe nuclei project to the hippocampal formation, the activity of which is known to modulate the inhibitory interneurons in the dentate gyrus. On the other hand, serotonergic modulation of the excitatory synapses in the dentate gyrus is not well examined. In the present study, we examined the effects of 5-HT on the excitatory postsynaptic potentials (EPSPs) in the dentate granule cells evoked by the selective stimulation of the lateral perforant path (LPP), the medial perforant path (MPP), or the mossy cell fibers (MCF). 5-HT depressed the amplitude of unitary EPSPs (uEPSPs) evoked by the stimulation of LPP or MPP, whereas uEPSPs evoked by MCF stimulation were little affected. The effect was partly explained by the decrease of the resting membrane resistance following the activation of 5-HT_1A receptors, which was confirmed by computer simulations. We also found that the probability of evoking uEPSP by LPP stimulation but not MPP or MCF stimulation was reduced by 5-HT and that the paired-pulse ratio of LPP-evoked EPSP but not that of MPP- or MCF-evoked ones was increased by 5-HT. These effects were blocked by 5-HT₂ antagonist, suggesting that the transmitter release in the LPP-granule cell synapse is inhibited by the activation of 5-HT₂ receptors. The present results suggest that 5-HT can modulate the EPSPs in the dentate granule cells by at least two distinct mechanisms

serotonin [5-hydroxytryptamine (5-HT)] is one of the most important neuromodulators in the central nervous system and is involved in the regulation of basic physiological functions such as sleep-wake cycle, thermoregulation, and food intake as well as emotion and cognition (Cools et al. 2008; Meneses 1999; Monti 2011; Morrison et al. 2008; Voigt and Fink 2015). Serotonergic fibers arising from the raphe nuclei project to many brain regions and one of the major targets is the dentate gyrus in the hippocampus. In the dentate gyrus, the serotonergic fibers distribute to the granular and molecular layers as well as hilus (Freund et al. 1990; Halasy et al. 1992; Oleskevich and Descarries 1990). Previous studies proposed that 5-HT mainly modulates the activities of inhibitory interneurons and therefore indirectly affects the activities of the granule cells (Gulyas et al. 1999; Levkovitz and Segal 1997; Piguet and Galvan 1994). The activities of the inhibitory interneurons are shown to be enhanced by the activation of 5-HT₃/5-HT₄ receptors (Bijak and Misgeld 1997; Piguet and Galvan 1994; Ropert and Guy 1991) or inhibited by the activation of 5-HT_1A receptors (Bijak and Misgeld 1997).

On the other hand, the effects of 5-HT on the excitatory synapses onto the dentate granule cells are not well resolved, and there are some confusing results in the literature. In in vivo studies of rats, the slope of field EPSPs (fEPSPs) in the granule cell layer evoked by the stimulation of the perforant path was not affected by a 5-HT releasing drug, fenfluramine, suggesting that 5-HT does not affect the fEPSPs (Levkovitz and Segal 1997; Richter-Levin and Segal 1990). In other in vivo studies, however, the slope of fEPSP was decreased by a 5-HT₄ agonist, methoxytryptamine, or a 5-HT_1A agonist, 8-hydroxy-2-di-n-propylaminotetralin (Kulla and Manahan-Vaughan 2002; Sanberg et al. 2006). In the rat hippocampal slice, Piguet and Galvan (1994) found that the amplitude of EPSPs evoked by subiculum stimulation in the presence of a GABA_A antagonist, bicuculline, was not affected by 5-HT.

There are multiple excitatory inputs to the granule cells. The main inputs are from the perforant path arising from the entorhinal cortex. The perforant path is separated into at least two pathways: the lateral perforant-path (LPP), which originates from the lateral entorhinal cortex and the medial perforant-path (MPP), which originates from the medial entorhinal cortex. The LPP projects to the granule cell dendrites in the outer region of the molecular layer whereas the MPP projects to the dendrites in the middle region of the molecular layer (Steward 1976; Tamamaki 1997; Wyss 1981). The hilar mossy cells are excitatory interneurons in the hippocampus and make synaptic contacts with the granule cells. The axonal fibers of the mossy cells (MCF) are rich in the inner one-third of the molecular layer and preferentially terminate on the proximal dendrites of the granule cells (Buckmaster et al. 1996; Scharfman 1995).

With such multiple excitatory synapses in mind, we revisited whether 5-HT has any direct modulatory actions on the excitatory synapses onto the granule cells. Specifically, we examined the effects of 5-HT on EPSPs evoked by the stimulation of LPP, MPP, or MCF. Here, we present evidences suggesting that the EPSPs in the dentate granule cells can be modulated differently by 5-HT.

MATERIALS AND METHODS

Animals.

Wistar rats from an in-house colony were used in the present study. All experimental protocols were approved by the Committee for Animal Experimentation of Hiroshima University and performed in accordance with the guidelines of the Japanese Association for Laboratory Animal Science.

Slice preparation.

Hippocampal slices were prepared from either sex of Wistar rats (17–31 days). The brain was quickly removed after decapitation and placed in ice-cold cutting solution (in mM: 110 sucrose, 63 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 12 Mg₂SO₄, 0.5 CaCl₂, 26 NaHCO₃, and 15 glucose) oxygenated with 95% O₂-5% CO₂. The brain was trimmed, glued on the stage of a microslicer (DTK-ZERO1; Dosaka EM), and immersed in the ice-cold oxygenated cutting solution. Horizontal slices (300 μm) were made and transferred to a holding chamber containing oxygenated artificial cerebrospinal fluid (aCSF; in mM: 126 NaCl, 2.5 KCl, 1.25 NaH₂PO₄, 2 Mg₂SO₄, 2 CaCl₂, 26 NaHCO₃, and 10 glucose, pH 7.4). The slices in the holding chamber were maintained at room temperature (23–26°C) for 45 min and then incubated at 33°C for 15 min. After that, the slices were kept at room temperature.

Electrophysiological recordings.

Hippocampal slices were transfered to a recording chamber on the stage of a microscope (ECLIPSE FN1; Nikon). The dentate granule cells were identified visually by using the infra-red differential interference contrast (IR-DIC) system (C10639-79; Hamamatsu Photonics). All recordings were performed using an Axoclamp 900A amplifier with a Digidata 1440A interface and AxoScope (Molecular Devices). Whole cell recordings were made from the dentate granule cells using patch pipettes filled with internal solution (in mM): 130 K gluconate, 10 NaCl, 2 MgCl₂, 10 HEPES, 1 EGTA, 2 Na₂ATP, and 0.2 NaGTP, pH 7.4. Experiments were performed at room temperature. To ensure that the recorded cells were not interneurons, the resting membrane potentials and firing properties of the cells were examined at first (see Scharfman 1992). The granule cells included in this study showed the resting membrane potentials of more than −70 mV and the input resistances were between 100 and 600 MΩ, suggesting that most recordings were made from mature granule cells (Ambrogini et al. 2004).

To stimulate the lateral perforant path (LPP) or the medial perforant path (MPP) selectively, we placed a stimulating electrode appropriately as described previously (Chiu and Castillo 2008; Petersen et al. 2013). Briefly, stimulating electrode made by a glass pipette (10–40 μm in diameter) filled with aCSF was positioned in the outer one third of the dentate molecular layer to stimulate LPP, whereas the electrode was in the middle one third of the molecular layer to stimulate MPP. To stimulate mossy cell fibers (MCF), we placed the stimulating electrode in the inner one third of the molecular layer. Usually, a brief electrical pulse (50–100 μs, 25–50 V) was applied every 1–5 s. In some experiments, we carried out so-called minimum stimulation to evoke unitary responses (Raastad et al. 1992; Stevens and Wang 1994).

We used 100 μM picrotoxin (PTX; Abcam Biochemicals, Cambridge, UK) to block GABA_A-dependent inhibitory synaptic potentials in all the experiments. In each experiment, control EPSPs were obtained for 5–10 min. 5-HT (Sigma-Aldrich, St. Louis, MO) was then applied by bath perfusion. We used WAY100635 (Tocris, Bristol, UK) as an antagonist of 5-HT_1A receptor, and ritanserin (Santa Cruz Biotechnology, Santa Cruz, CA) as an 5-HT₂ receptor antagonist. Although WAY100635 is also known to be a dopamine D4 receptor agonist (Chemel et al. 2006), the application of WAY100635 did not affect the membrane potential or input resistance of the granule cells in the present experiments. For example, the input resistance before and after application of 50 nM WAY100635 in nine experiments was 305.2 ± 129.1 and 307.2 ± 119.6 MΩ, respectively (P > 0.5, Wilcoxon signed rank test). To see the effect of 5-HT antagonist, we pretreated the slices with 5-HT antagonist for 5 min at least. After that, we took control EPSPs in the presence of the antagonist and then applied 5-HT.

Analysis of EPSPs.

We routinely analyzed EPSPs for 4 min in each condition. Control EPSPs were obtained just before the application of 5-HT. Usually, 5-HT was applied for more than 10 min to see the stationary effects, and EPSPs in the last 4 min were used to examine the effect of 5-HT. EPSPs were analyzed by Clampfit (ver. 10; Molecular Devices) and Origin (ver. 8; OriginLab). At first, events (100∼150 ms after the stimulus artifacts) were extracted by Clampfit. EPSPs were detected by visual inspection from the extracted events. The peak amplitude, the time to peak (the interval between the stimulus artifact and the peak of EPSP), the 10–90% rise time, and the decay time constant of detected EPSPs were then calculated by Clampfit. As reported previously in the central synapses (Bekkers et al. 1990; Malinow and Tsien 1990), we observed substantial failures of synaptic transmission as well as large variation of EPSP amplitude in the present study.

To examine the effects of 5-HT on the amplitude and the waveform of EPSPs, we averaged detected EPSPs and compared the peak amplitude, the time to peak, the 10–90% rise time, and the decay time constant of the averaged responses. We also determined success probability of evoking EPSP before and after 5-HT application in each experiment. The success probability was calculated by dividing the number of detected EPSPs by the number of stimulation. In some experiments, the paired-pulse ratio (PPR) was also examined before and after 5-HT application. Two sequential stimuli with 90- to 120-ms interval were applied every 2–5 s, and PPR was calculated by dividing the peak amplitude of second EPSP with that of first EPSP.

Data are shown as mean ± SD, and the statistical tests (one sample t-test, Wilcoxon signed-rank test, or Tukey-Kramer test) were carried out by using the open source statistical package R (R Development Core Team 2011). The difference of the data was concluded to be statistically significant if P < 0.05. In some figures, data are shown by a boxplot. Upper and lower ends of the box show 75th and 25th percentiles, respectively. A line in the box shows the median and a circle in the box shows the mean. Upper and lower error bars indicate 95th and 5th percentiles, respectively.

Computer simulations.

All simulations were carried out by the Surf-Hippo simulator (Borg-Graham 2000; Graham 2004). A passive multicompartment model for a granule cell was made based on a reconstructed morphological data (n212, NeuroMorpho.org ID:NMO 00137) in the Duke/Southampton database, NeuroMorpho.org (Ascoli et al. 2007; Cannon et al. 1998). We first used cvapp (http://neuron.duke.edu/cells/) to convert the file format of the morphological data (SWC-file) to a format that can be used by Neuron (http://www.neuron.yale.edu/; hoc-file). We next translated the hoc-file into a lisp script that can be used in the Surf-Hippo. The model consists of 619 compartments, and the passive electrical properties were as follows: membrane resistance: 20,000 Ω·cm²; membrane capacitance: 1 μF/cm²; and axial resistance: 200 Ω·cm. Resting potential was set to −80 mV. The input resistance of the model having the above-mentioned parameters was 333.7 MΩ. To change the input resistance of the model, we changed the membrane resistance appropriately. Synaptic conductance was approximated by an α-function, and the maximum conductance was fixed to 300 pS. Synaptic conductances were placed at all the compartments except for the soma. Synaptic conductance was activated one by one, and the EPSPs were recorded at the soma. Simulated EPSPs were analyzed by using R.

RESULTS

5-HT decreases the input resistance of the granule cells.

In the dentate granule cells, 5-HT is known to activate 5-HT_1A receptors, which then activate the G-protein activated inward rectifying K⁺ channels (GIRK channels). The activation of GIRK channels reduces the input resistance and hyperpolarizes the membrane potential of the granule cells (Baskys et al. 1989; Ghadimi et al. 1994; Piguet and Galvan 1994). In the present study, the resting potential of the granule cells was −89.0 ± 5.4 mV, which became −89.1 ± 4.2 mV after the application of 1 μM 5-HT (mean ± SD, n = 47). Because the resting potential was close to the reversal potential of K⁺, it is not surprising that the resting potential was little affected by the 5-HT induced activation of GIRK channels. By contrast, the input resistance was clearly reduced by 5-HT, and the effect was modestly concentration dependent. The input resistance of the granule cells in the control condition was 351.8 ± 116.4 MΩ (n = 69). The input resistance in the presence of 0.3, 1, 10 and 30 μM 5-HT were 285.2 ± 39.3 (n = 5), 226.1 ± 78.2 (n = 47), 183.6 ± 62.9 (n = 9), and 177.0 ± 83.2 (n = 8) MΩ, respectively. The effect of 5-HT on the input resistance was almost completely blocked by 50 nM WAY100635 (5-HT_1A receptor antagonist) in agreement with previous study (Piguet and Galvan 1994). In the following experiments, we routinely used 1 μM 5-HT, which decreased the input resistance of the granule cells to ∼60%.

Comparison of EPSPs evoked by the stimulation of LPP, MPP, or MCF.

We first compared some properties of EPSPs evoked by the stimulation of LPP, MPP, or MCF (Fig. 1). Examples of the EPSPs are shown in Fig. 1B. The amplitude of EPSP was quite variable from trial to trial irrespective of the stimulated sites. In the examples shown in Fig. 1B, means ± SD of the peak amplitude of LPP-, MPP-, and MCF-evoked EPSPs were 1.47 ± 0.48, 1.00 ± 0.34, and 1.07 ± 0.77 mV, respectively. The ranges of mean EPSP sizes in the present experiments were as follows (mV): LPP-evoked EPSPs, 1.5–3.5; MPP-evoked EPSPs, 1.0–3.4; and MCF-evoked EPSPs, 1.1–3.6.

Fig. 1.

Three types of excitatory synapses onto the granule cell. A: schematic drawing showing the position of stimulating electrode (stim). LPP, lateral perforant path; MPP, medial perforant path; MCF, mossy cell fiber; GC, granule cell; MC, mossy cell. B: examples of excitatory postsynaptic potentials (EPSPs) evoked by stimulation of LPP, MPP, or MCF. C: comparison of the kinetics of the averaged EPSPs. To compare the rising kinetics of the EPSPs evoked by LPP, MPP, and MCF stimulation, the EPSPs were averaged and their peak amplitudes were scaled to the same size. D: comparison of the time to peak (TP) of averaged EPSPs. LPP: n = 16; MPP: n = 16: MCF: n = 12. Multiple comparisons were made by Tukey-Kramer test; **P < 0.01, significantly different pairs. E: comparison of the 10–90% rise time (RT) of averaged EPSPs. F: comparison of the decay time constants (τ) of averaged EPSPs.

Figure 1C shows the rising phases of averaged EPSPs, which are scaled to make their amplitude the same size. In this and the following figures, the detected EPSPs without including failures were averaged. MCF-evoked EPSP was the fastest and sharply rose to its peak. The rising phases of MPP- and LPP-evoked EPSP were slower suggesting more dendritic filtering. Figure 1, D–F compares the time to peak, the 10–90% rise time, and the decay time constant of the averaged EPSPs. The time to peak of LPP-, MPP-, and MCF-evoked EPSPs was 15.5 ± 1.7, 15.4 ± 3.2, and 11.0 ± 2.1 ms, respectively (n = 12–16). The rise time of LPP-, MPP-, and MCF-evoked EPSPs was 6.7 ± 1.0, 6.3 ± 1.5, and 3.8 ± 0.8 ms, respectively (n = 12–16). On these parameters, MCF-evoked EPSP was significantly different from the other EPSPs (P < 0.0001, Tukey-Kramer test). On the other hand, the decay time constants of EPSPs were similar: the decay time constants of LPP-, MPP-, and MCF-evoked EPSPs were 31.2 ± 8.0, 35.6 ± 13.4, and 35.2 ± 7.2 ms, respectively (n = 12–16). Faster rising kinetics of MCF-evoked EPSPs suggests that MCF synapses are closer to the soma than the other two excitatory synapses. Although the axonal projections of LPP and MPP indicate LPP-granule cell synapses are more distal than MPP-granule cell synapses (Steward 1976; Tamamaki 1997; Wyss 1981), the kinetic difference between LPP-evoked EPSP and MPP-evoked one was not significant. The results may suggest that LPP- and MPP-granule cell synapses, which were detected electrophysiologically in the present experiments, were close. Alternatively, the kinetic difference may be obscured by rather strong dendritic filtering in the granule cells, which is known to attenuate EPSPs of synapses at >100 μm from the soma strongly and uniformly (Krueppel et al. 2011).

Effect of 5-HT on EPSPs.

To examine whether 5-HT has any direct action on the excitatory synaptic transmission in the granule cells, we compared EPSPs evoked by LPP, MPP, or MCF stimulation before and after the application of 5-HT in the presence of 100 μM PTX (Fig. 2). As shown in Fig. 2A, LPP-evoked EPSPs were clearly depressed by 5-HT, and the variance of EPSPs was also decreased. The effect of 5-HT was reversed by washing out (Fig. 2A2). The amplitude histograms of the EPSPs before and after 5-HT application are illustrated in Fig. 2A3. The distribution of EPSPs after 5-HT (the gray histogram in Fig. 2A3) was markedly shifted to lower amplitude range. In this example, means ± SD of the EPSPs before and after 5-HT were 1.82 ± 0.80 and 0.69 ± 0.35 mV. On the other hand, we did not observe such large inhibitory effect of 5-HT on the amplitude of MPP- or MCF-evoked EPSPs. Figure 2B shows a result of the effect of 5-HT on MPP-evoked EPSPs. In this example, the amplitude and the variance of EPSPs were not noticeably changed after the application of 5-HT, and the amplitude histograms before and after 5-HT overlapped well (Fig. 2B3). Means ± SD of the EPSPs before and after 5-HT were 1.93 ± 0.90 and 1.75 ± 0.84 mV.

Fig. 2.

Effects of 5-HT on LPP-, MPP-, and MCF-evoked EPSPs. A1: LPP-evoked EPSPs before and after application of 1 μM 5-HT. A2: effect of 5-HT on the amplitude of LPP-evoked EPSPs. A3: amplitude histograms of LPP-evoked EPSPs before (white) and after (gray) application of 5-HT. B1: MPP-evoked EPSPs before and after application of 1 μM 5-HT. B2: effect of 5-HT on the amplitude of MPP-evoked EPSPs. B3: amplitude histograms of MPP-evoked EPSPs before (white) and after (gray) application of 5-HT. C: Comparison of the effect of 5-HT on EPSP_5-HT/EPSP_before of the 3 excitatory synapses. Statistical significance of the results was examined by one sample t-test; * P < 0.05, ** P < 0.01. D: comparison of the effect of 5-HT on TP_5-HT/TP_before. E: comparison of the effect of 5-HT on RT_5-HT/RT_before. F: comparison of the effect of 5-HT on τ_5-HT/τ_before.

Figure 2C summarizes the effect of 5-HT on the EPSP amplitude. The amplitude ratio (EPSP_5-HT/EPSP_before) of the averaged LPP-evoked EPSPs was 0.51 ± 0.14 (n = 7, P < 0.01, one sample t-test), showing that the EPSP amplitude was reduced to ∼50% by 5-HT. By contrast, the amplitude of MPP- or MCF-evoked EPSPs was less affected. EPSP_5-HT/EPSP_before of the averaged MPP-evoked EPSPs was 0.92 ± 0.07 (n = 7, P < 0.05) and that of the averaged MCF-evoked EPSPs was 1.00 ± 0.15 (n = 6, P > 0.5).

Because 5-HT decreases the input resistance of the granule cells, it should affect the waveform of synaptic potentials. As shown in Fig. 2, D and E, the time to peak and the rise time of EPSPs were usually shortened by 5-HT. The decay time constants of EPSPs were more drastically changed by 5-HT and became 50–60% of the control following the application of 5-HT (Fig. 2F). Means ± SD of the decay time constants before and after 5-HT were as follows (in ms): LPP-evoked EPSPs, 31.8 ± 7.7 vs. 15.5 ± 7.3 (n = 7); MPP-evoked EPSPs, 36.9 ± 15.1 vs. 24.2 ± 8.7 (n = 7); and MCF-evoked EPSPs, 34.1 ± 7.1 vs. 16.2 ± 1.9 (n = 6). Such shortening of the time constants of the EPSPs is well explained by shunting effect due to the 5-HT-induced activation of GIRK channels (Baskys et al. 1989; Ghadimi et al. 1994; Piguet and Galvan 1994). The shunting effect was blocked by a 5-HT_1A receptor antagonist, WAY100635, as shown below (see Fig. 5).

Fig. 5.

Effects of 5-HT_1A and 5-HT₂ receptor antagonists on the inhibitory actions of 5-HT. 5-HT actions were examined in the absence of 5-HT antagonists, in the presence of 5-HT_1A receptor antagonist (50 nM WAY100635), or in the presence of both 5-HT_1A and 5-HT₂ receptor antagonists (50 nM WAY100635 and 10 μM ritanserin). A: effects of 5-HT receptor antagonists in the LPP-granule cell synapses. WAY, WAY100635; WAY,RS, WAY100635 and ritanserin. A1: examples of uEPSPs. The averaged uEPSPs before and after application of 1 μM 5-HT are shown by black and red traces, respectively. A2: effects of 5-HT receptor antagonists on EPSP_5-HT/EPSP_before. *P < 0.05, **P < 0.01 (n = 6–7, one-sample t-test). A3: effects of 5-HT receptor antagonists on τ_5-HT/τ_before. B: effects of 5-HT receptor antagonists in the MPP-granule cell synapses (n = 6–7). C: effects of 5-HT receptor antagonists in the MCF-granule cell synapses (n = 5–6). Control data in this figure are the same to those used in Fig. 3.

Effect of 5-HT on unitary EPSPs.

To restrict the number of presynaptic fibers being stimulated, we next employed a so-called minimum stimulation to evoke EPSPs. Under favorable minimum stimulation conditions, it is possible to stimulate even a single presynaptic axon (Raastad et al. 1992; Stevens and Wang 1994). We reduced the stimulus strength to the extent that it evokes EPSPs <1 mV in most trials. Figure 3A shows typical examples of EPSPs evoked by the minimum stimulation protocol. Means ± SD of detected EPSPs in the experiments shown in Fig. 3A were as follows (mV): LPP-evoked EPSPs, 0.43 ± 0.19; MPP-evoked EPSPs, 0.46 ± 0.30; and MCF-evoked EPSPs, 0.70 ± 0.30. In the present experiments, the amplitude of detected EPSPs by the minimum stimulation was mostly within 0.4–0.9 mV in spite of the fact that the success probability was quite different in each experiment (Fig. 3B). The result is consistent with a notion that the detected EPSPs are unitary responses. Our detection, however, may be skewed to large EPSPs because EPSP amplitude of less than ∼0.1 mV was not reliably detected in our recording conditions. Although we call the detected EPSPs as unitary EPSPs (uEPSPs) in the present study, some of the EPSPs may not be real uEPSPs but may result from coactivation of a few synapses.

Fig. 3.

5-HT reduces the amplitude of unitary (u)EPSPs. A: examples of the uEPSPs evoked by minimum stimulation protocol. B: relationship between the success probability and the amplitude of uEPSPs. Gray band shows the range of the median values of uEPSPs. C: examples of the averaged uEPSPs before (black) and after application of 1 μM 5-HT (red). D: comparison of the effect of 5-HT on EPSP_5-HT/EPSP_before of the averaged uEPSPs. LPP: n = 6; MPP: n = 6; MCF: n = 6. E: success probability of evoking uEPSP before (open circles) and after (closed circles) application of 5-HT. Mean values are indicated by horizontal bars. *P < 0.05 (Wilcoxon signed-rank test).

The averaged uEPSPs before and after 5-HT, which were obtained from the same cells shown in Fig. 3A, are illustrated in Fig. 3C. The amplitude of LPP- and MPP-evoked uEPSP was modestly depressed by 5-HT but that of MCF-evoked one was much less affected. Means ± SD of detected uEPSPs before and after 5-HT in the experiments shown in Fig. 3, A and C, were as follows (in mV): LPP-evoked uEPSPs, 0.43 ± 0.19 vs. 0.33 ± 0.13; MPP-evoked uEPSPs, 0.47 ± 0.18 vs. 0.37 ± 0.11; and MCF-evoked uEPSPs, 0.71 ± 0.30 vs. 0.64 ± 0.29. Figure 3D summarizes the effect of 5-HT on the amplitude of averaged uEPSPs. EPSP_5-HT/EPSP_before of LPP-, MPP-, and MCF-evoked uEPSPs was 0.78 ± 0.09 (n = 6), 0.76 ± 0.07 (n = 6), and 0.95 ± 0.12 (n = 6), respectively. EPSP_5-HT/EPSP_before of LPP- and MPP-evoked uEPSPs was significantly different from 1.0 (P < 0.01, one sample t-test), indicating that LPP- and MPP-evoked uEPSPs were depressed by 5-HT. Although MCF-evoked uEPSPs were also noticeably depressed in some preparations, EPSP_5-HT/EPSP_before of the pooled data was not significantly different from 1.0 (P > 0.05). By contrast, the decay time constants of the uEPSPs were shortened by 5-HT in all the cases (see Figs. 3C and 5). Means ± SD of the decay time constants of uEPSPs before and after 5-HT were as follows (in ms): LPP-evoked uEPSPs, 50.1 ± 19.8 vs. 19.8 ± 8.1; MPP-evoked uEPSPs, 47.9 ± 25.9 vs. 15.9 ± 2.9; and MCF-evoked uEPSPs, 31.0 ± 17.0 vs. 14.8 ± 3.7.

Figure 3E shows the effect of 5-HT on the success probability of evoking uEPSP. Although the success probability was quite variable among the experiments as described above, the success probability of evoking LPP-uEPSP was consistently decreased by 5-HT (n = 6, P < 0.05, Wilcoxon signed rank test). On the other hand, the success probability of evoking MPP- or MCF-uEPSP was not affected by 5-HT. The results may suggest that the transmitter release at LPP-granule cell synapses but not at MPP- and MCF-granule cell synapses is inhibited by 5-HT.

Simulations of the shunting effects on EPSPs.

To obtain some insights about to what extent the synaptic response is affected by the decrease of input resistance, we carried out simulations of a multicompartment model of the granule cell (see materials and methods). We subdivided the dendrites of the model into nine areas by physical distance from the soma (see Fig. 4A). Mean EPSPs that were evoked by the activation of synapses in the area 2, the area 5, or the area 8 are shown in Fig. 4A. As expected, the closer the activated synapses were to the soma, the larger and faster the EPSPs were.

Fig. 4.

Simulations showing the relationship among the EPSP, the input resistance, and the synaptic location. A: schematic drawings showing the position of activated synapses (shown by dots) and the averaged EPSPs. In the compartment model, 10 areas were assigned as follows: area 0 (soma), area 1 (within 5 μm from the soma), area 2 (5–50 μm away from the soma), area 3 (50–100 μm away from the soma), area 4 (100–150 μm away from the soma), area 5 (150–200 μm away from the soma), area 6 (200–250 μm away from the soma), area 7 (250–300 μm away from the soma), area 8 (300–350 μm away from the soma), and area 9 (>350 μm away from the soma). Approximate boundaries are shown by broken lines. A1: synaptic positions and the averaged EPSPs in the area 2. The EPSPs were from the different simulations in which the input resistance was changed by reducing the membrane resistance of the model appropriately. The input resistances were as follows (in MΩ): 333.7, 301.9, 269.7, 236.9, 203.3, 168.1, 130.1, and 84.9. The largest EPSP corresponds to the simulation of the model having the largest input resistance and vice versa. A2: synaptic positions and the averaged EPSPs in the area 5. A3: synaptic positions and the averaged EPSPs in the area 8. B1: 3-dimensional (3D) plot showing the relationship among the EPSP amplitude, the input resistance and the synaptic location (area). B2: 3D plot showing the relationship among the normalized EPSP amplitude, the input resistance, and the synaptic location (area). B3: relationship between the input resistance and the amplitude of EPSP. The normalized amplitudes of EPSPs are plotted against the normalized input resistances, and the relationships in 9 areas are superimposed. The number near each trace shows the area number. The experimental data of uEPSPs shown in Fig. 3D are also superimposed. LPP: black; MPP: red; MCF: blue. Each symbol shows the mean value and the error bar indicates SD. C1: 3D plot showing the relationship among the decay time constant (τ) of EPSP, the input resistance and the synaptic location (area). C2: 3D plot showing the relationship among the normalized τ of EPSP, the input resistance, and the synaptic location (area). C3: relationship between the input resistance and the τ of EPSP. The normalized τ of EPSPs are plotted against the normalized input resistances, and the relationships in 9 areas are superimposed. The uppermost trace is a relationship at area 1 and the lowermost one is that at area 9. The values calculated from the uEPSP data shown in Fig. 3 are also superimposed. LPP: black; MPP: red; MCF: blue. Each symbol shows the mean value and the error bar indicates SD.

Figure 4B1 illustrates a three-dimensional (3D) plot showing the relationship among the amplitude of EPSP, the input resistance, and the synaptic location (area) obtained from different simulations. In the model simulation, EPSPs evoked at distant sites were strongly and rather uniformly diminished. For example, the EPSPs evoked by the same synaptic conductance at different areas in the control model (the input resistance was 333.7 MΩ) were as follows (in mV): area 2, 2.49 ± 0.27; area 3, 1.83 ± 0.21; area 4, 1.45 ± 0.26; area 5, 1.28 ± 0.28; area 6, 1.05 ± 0.20; area 7, 0.94 ± 0.18; and area 8, 0.88 ± 0.19.

To see the relative change of synaptic responses by reducing the input resistance of the model, the EPSPs were normalized by their respective control values in Fig. 4B2. It is clear that the synaptic response at distal synapses are more strongly reduced by decreasing the input resistance. These results show that the dendritic filtering is substantial in the present passive model, which is consistent with the results of Krueppel et al. (2011).

Figure 4B3 shows the relationships between the normalized input resistance, and the normalized amplitude of EPSPs evoked at nine areas of the model neuron. In Fig. 4B3, the experimental data of uEPSPs shown in Fig. 3 are superimposed. Although there were large variations in the experimental results, the data of MCF-evoked uEPSPs were overlapped with the simulated data obtained at more proximal synapses in the model than the data of LPP- or MPP-evoked uEPSPs, consistent with the published results on the location of MCF-granule cell synapses (Buckmaster et al. 1996; Deller et al. 1995; Scharfman 1995).

The relationship among the decay time constant, the input resistance and the synaptic location is shown in Fig. 4C. The decay time constants of simulated EPSPs were dependent on the input resistance of the model as expected, but the dependency of the time constant on the synaptic location was little (Fig. 4, C1 and C2). The relationships between the normalized time constants and the input resistances in nine areas are illustrated in Fig. 4C3, showing that the relationship is largely unrelated to the synaptic locations. The mean decay time constants of uEPSPs are superimposed in Fig. 4C3. Although the time constants of uEPSPs are quite variable, the mean values are near the relationship obtained from the simulation. Overall, the properties of simulated EPSPs are consistent with the experimental results described above, suggesting that the inhibition of the EPSPs by 5-HT could be explained mainly by the shunting effect of 5-HT.

Pharmacology of the 5-HT actions.

We next examined some pharmacology of 5-HT induced inhibition of the EPSPs. At first, we examined the effect of a 5-HT_1A receptor antagonist, WAY100635, which completely blocks the effect of 5-HT on the input resistance as described above. In the presence of WAY100635, the inhibitory action of 5-HT on LPP-evoked uEPSP was still observed (Fig. 5A1, WAY). Although WAY100635 seemed to depress the 5-HT action partially in some preparations, the pooled data showed that the inhibitory effect on LPP-evoked uEPSPs was not significantly affected by WAY100635 (Fig. 5A2). By contrast, the effect of 5-HT on the decay time constant of the uEPSP was completely blocked by WAY100635 (Fig. 5A3). The inhibitory action of 5-HT on the peak amplitude of LPP-evoked uEPSPs was, however, completely blocked if the 5-HT₂ antagonist ritanserin was coapplied with WAY100635 (Fig. 5, A1 and A2, WAY,RS). On the other hand, the effects of 5-HT on both the amplitude and the decay time constant of MPP-evoked uEPSP were completely blocked by WAY100635 (Fig. 5B). As described above, the inhibitory action of 5-HT on the amplitude of MCF-evoked uEPSP was little, and none of the EPSP_5-HT/EPSP_before of MCF-evoked uEPSPs shown in Fig. 5C2 was significantly different from 1.0. The effect of 5-HT on the decay time constant of MCF-evoked uEPSP was completely blocked by WAY100635 (Fig. 5C3).

Because the success probability of evoking uEPSP by LPP stimulation but not by MPP or MCF stimulation was decreased by 5-HT (see Fig. 3E), we also analyzed the effect of 5-HT antagonists on this 5-HT action. As shown in Fig. 6A, WAY100635 alone did not affect the 5-HT action on the success probability at all: 5-HT decreased the success probability similarly either in the absence or the presence of WAY100635 (P < 0.05, Wilcoxon signed rank test). The addition of ritanserin, however, completely blocked the 5-HT action (WAY,RS in Fig. 6A), suggesting that 5-HT decreases the success probability of evoking LPP-uEPSP via the activation of 5-HT₂ receptor.

Fig. 6.

5-HT₂ receptor antagonist but not 5-HT_1A receptor antagonist blocks the actions of 5-HT on the success probability and the paired-pulse ratio (PPR) in the LPP-granule cell synapses. A: effects of 5-HT receptor antagonists on the 5-HT induced decrease of the success probability of evoking uEPSPs. WAY, 50 nM WAY100635; WAY,RS, 50 nM WAY100635 and 10 μM ritanserin. The data in control are the same to the ones shown in Fig. 3E. Mean values are indicated by horizontal bars. *P < 0.05 (Wilcoxon signed rank test). B: effects of 5-HT on the PPR of LPP-evoked EPSPs. B1: EPSPs evoked by the paired-pulse protocol before and after application of 5-HT. B2: effects of 5-HT antagonists on the 5-HT induced modulation of PPR. Mean values are indicated by horizontal bars. *P < 0.05 (Wilcoxon signed-rank test).

We also compared the effect of 5-HT on the paired-pulse ratio (PPR) in LPP-evoked EPSPs (Fig. 6B). In this experiment, we used our standard stimulation protocol and the pulse interval was 90–120 ms. Under the present experimental conditions, the PPR was usually within 0.9–1.2. An example of the PPR experiment is shown in Fig. 6B1. In this case, the PPR before and after application of 5-HT was 1.04 and 1.12, respectively. Figure 6B2 illustrates the effect of 5-HT on the PPR in control, in the presence of WAY100635, or in the presence of both WAY100635 and ritanserin. 5-HT increased the PPR of LPP-evoked EPSPs (control in Fig. 6B2), which is consistent with the decrease of the success probability of evoking uEPSP described above and suggests that 5-HT decreases the transmitter release from presynaptic terminals because the amplitude of PPR is inversely related to the release probability at first pulse (Dobrunz and Stevens 1997). The PPR was still increased by 5-HT in the presence of WAY100635 (WAY in Fig. 6B2), but the effect of 5-HT was completely blocked by the addition of ritanserin (WAY,RS in Fig. 6B2). The PPR of MPP- or MCF-evoked EPSP was not affected by 5-HT (data not shown).

DISCUSSION

In the present study, we found that the EPSPs in the dentate granule cells can be modulated by 5-HT. We examined three excitatory synapses onto the granule cells by stimulating the lateral perforant path (LPP), the medial perforant path (MPP), or the mossy cell fibers (MCF) separately. Following the application of 5-HT, LPP-evoked EPSP could be reduced more than 50%. By contrast, the inhibitory action of 5-HT on MPP- or MCF-evoked EPSPs was much less and may not be recognized in some cases. As discussed below, 5-HT reduces the input resistance of the granule cells, which in principle affects every synaptic potential more or less depending on the synaptic location. In the LPP-granule cell synapses, however, there seems to be other inhibitory mechanisms of 5-HT.

It is well established that 5-HT_1A receptor activates the GIRK channels (Andrade and Nicoll 1987; Luscher et al. 1997; Segal 1980). The dentate granule cells strongly express the mRNA of 5-HT_1A receptor (Tanaka et al. 2012), and the receptor proteins detected by immunohistochemistry are abundant in the molecular layer of the dentate gyrus (Kia et al. 1996; Zhou et al. 1999). The GIRK channels are homo- or heterotetramers of GIRK subunits (Krapivinsky et al. 1995; Lesage et al. 1995), and the GIRK subunits (GIRK1, GIRK2, GIRK3) are detected widely in the rat brain and are intensely expressed in the molecular layer of the dentate gyrus (Fernández-Alacid et al. 2011). Many neurons in the hippocampus including the dentate gyrus coexpress the 5-HT_1A receptor and GIRK subunits (Saenz del Burgo et al. 2008) consistent with the physiological action of 5-HT in the dentate granule cells (Baskys et al. 1989; Ghadimi et al. 1994; Piguet and Galvan 1994).

In accord with previous studies, we found that 5-HT decreases the input resistance of the dentate granule cells in the presence of GABA_A receptor blocker (PTX) and that the response was blocked by the 5-HT_1A receptor antagonist WAY100635. The increased resting conductance by 5-HT should in principle attenuate the synaptic potentials. Such a shunting effect on the synaptic potentials depends on the synaptic location on the dendrites as well as the dendritic area in which the resting conductance is changed. A similar shunting effect of the excitatory inputs by GABA_B receptors as well as more selective shunting effect by metabotropic glutamate receptors has recently been shown in the dentate granule cells (Brunner et al. 2013).

One micromolar 5-HT decreased the input resistance of granule cells to ∼60% in the present study. In our simulation, such reduction of the input resistance induces 10–40% reduction of EPSP size depending on the synaptic location. When the resting conductance of the model neuron is increased, EPSPs activated by proximal synapses are less affected. As described in results, the effect of 5-HT on the MCF-evoked EPSPs was little and at most less than 10% reduction of the EPSP was observed. Because the axons of mossy cells extend mostly into the inner one third of the molecular layer (Buckmaster et al. 1996; Scharfman 1995), the result is expected from the present simulation: EPSPs evoked at the synapses located within 50 μm from the soma were reduced to 89 ± 17% of the control when the input resistance of the model was reduced to 203.3 MΩ from 333.7 MΩ (∼60% of the control, see Fig. 4). On the other hand, 1 μM 5-HT reduced the LPP-evoked EPSPs to ∼50% of the control (see Fig. 2), which is much larger than the reduction expected from the simulation described above. Because 5-HT decreases the success probability and increases the PPR in the LPP-granule cell synapse, 5-HT also seems to reduce the transmitter release in this synapse (see below).

Although MPP-evoked EPSPs by our standard stimulation protocol were much less affected by 5-HT compared with LPP-evoked ones (see Fig. 2C), the inhibition of uEPSPs evoked by MPP stimulation by 5-HT was comparable to that observed in LPP-evoked uEPSPs (see Fig. 3D). The reduction of MPP-evoked uEPSPs by 5-HT (76.7 ± 7%) is within a range expected from our simulation in the passive model with a constant synaptic conductance. We also found that the success probability of MPP-evoked uEPSPs as well as that of MCF-evoked uEPSPs was not affected by 5-HT. More importantly, the 5-HT_1A antagonist WAY100635, which completely blocks the 5-HT-induced decrease of the input resistance, also blocked the inhibitory effect of 5-HT on MPP-evoked uEPSPs (Fig. 5B2). Thus the 5-HT action on MPP- as well as the MCF-evoked EPSPs is likely due to the shunting effect of 5-HT. Because the shunting effect is dependent on the synaptic location and the proximal synapses are less affected (Fig. 4B, see also Krueppel et al, 2011), the modest shunting effect of 5-HT in MPP-evoked EPSPs may be masked in many experiments by the intrinsic synaptic variability of MPP-granule cell synapses (see Fig. 2B). Piguet and Galvan (1994) also examined the effect of 5-HT on the perforant path-evoked EPSPs in the dentate granule cells. They found 5-HT shortens the EPSPs as described in the present study but did not observe the depression of EPSP amplitude (Piguet and Galvan 1994). Because they placed the stimulating electrode at the subiculum, which includes both MPP and LPP axons, it is likely that they examined mixed synaptic potentials in which the MPP-evoked ones might be dominant.

As described above, LPP-evoked EPSPs were depressed more than expected from the shunting effect of 5-HT and 5-HT decreased the success probability of evoking uEPSP by LPP stimulation. LPP originates from the lateral entorhinal cortex and 5-HT is known to inhibit lateral entorhinal cortex neurons by activating GIRK channels via 5-HT_1A receptors (Grunschlag et al. 1997). If the 5-HT_1A-GIRK mechanism also exists in the axonal membranes, 5-HT may reduce the excitability of LPP axons and may induce a failure of spike generation, which would decrease the success probability of LPP stimulation. Another possibility is that the transmitter release at presynaptic terminals of LPP is inhibited by 5-HT. The result of the paired-pulse experiments in which the PPR of the LPP-evoked EPSPs was increased by 5-HT favors the latter possibility. The depression of the amplitude of LPP-evoked uEPSPs by 5-HT was still observed in the presence of WAY100635 but completely blocked if ritanserin (5-HT₂ receptor antagonist) was coapplied. Moreover, the effects of 5-HT on the success probability of evoking uEPSP and the PPR in LPP-granule cell synapses were also blocked by coapplication of ritanserin with WAY100635 but not by WAY100635 alone. These results could be explained if the activation of 5-HT₂ receptor reduces the transmitter release from the LPP terminals. Indeed, the activation of 5-HT₂ receptor is known to reduce glutamate release from the isolated nerve terminals (Maura et al. 1991) and suppress the excitatory synaptic currents in the brainstem slices (Best and Regehr 2008). Recently, it has been shown that the application of a high concentration of 5-HT (100 μM) for 10 min induces a long-term depression of EPSPs in the granule cells evoked by the stimulation at the border between the subiculum/presubiculum and the dentate gyrus and that the depression is blocked by ritanserin (Gilling et al. 2013). We did not test such a high concentration of 5-HT but the inhibitory action of 5-HT in the present study was not long lasting and was reversed by washing out. Further quantitative experiments are required to see the mechanisms of 5-HT₂-dependent inhibition of the excitatory transmission in the dentate granule cells. In the dentate gyrus, major targets of serotonergic fibers from the raphe nuclei are considered to be hilar interneurons (Freund et al. 1990; Halasy et al. 1992; Oleskevich and Descarries 1990), and 5-HT is considered to affect the excitatory transmission between the perforant path and the granule cells indirectly by the activation of GABAergic inhibitory neurons (Gulyas et al. 1999; Levkovitz and Segal 1997; Piguet and Galvan 1994). However, the present experiments unequivocally showed that 5-HT can depress EPSPs in the dentate granule cells by reducing the resting membrane resistance and that more specific inhibition also exists in the LPP-granule cell synapses. The granule cells are principal neurons in the dentate gyrus which is a gateway for information processing in the hippocampus. The serotonergic modulation of excitatory synapses onto the granule cells should therefore have profound effects for the hippocampal functions.

GRANTS

This work was partly supported by Grant-In-Aid for Scientific Research (to Y. Furukawa; No. 23657058) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

K.N., R.K., and Y.F. performed experiments; K.N. and Y.F. analyzed data; K.N. and Y.F. interpreted results of experiments; K.N. and Y.F. prepared figures; K.N. and Y.F. drafted manuscript; K.N. and Y.F. edited and revised manuscript; K.N., R.K., and Y.F. approved final version of manuscript; Y.F. conception and design of research.