Abstract
Cholinergic innervation of the hippocampus has been implicated in memory formation and retrieval. Here we study cholinergic modulation of excitatory transmission in the CA3 area of the rat hippocampus. We used a combination of optical measurements of presynaptic calcium and electrophysiological measurements of synaptic currents to study associational–commissural (A/C) and mossy fiber (MF) synapses in brain slices. Direct synaptic modulation mediated by ACh receptors is only evident at the A/C synapse, where synaptic inhibition primarily reflects presynaptic calcium channel inhibition mediated by muscarinic receptors. MF synapses can, however, be indirectly modulated by muscarinic receptor activation. Muscarine elevates the firing rate of inhibitory cells, which increases GABA release and inhibits MF synapses by activating presynaptic GABAB receptors. Muscarine also depolarizes dentate granule cells and elevates their rate of firing. This leads to synaptic enhancement when combined with the use-dependent facilitation of MF synapses. In addition we were unable to evoke an increase in presynaptic calcium levels in MF boutons with local application of nicotinic receptor agonists. This finding does not support a leading hypothesis for MF modulation in which activation of presynaptic nicotinic receptors enhances transmission directly by elevating presynaptic calcium levels. However, indirect synaptic modulation could arise from nicotinic excitation of inhibitory neurons. Thus, to understand cholinergic modulation within the CA3 region, it is necessary to take into account secondary actions on synapses arising from other chemical messengers released by other cell types and to consider effects on firing patterns of presynaptic cells, which in turn influence release via use-dependent synaptic plasticity.
- acetylcholine
- hippocampus
- associational–commissural
- mossy fiber
- muscarine
- nicotine
- presynaptic
- calcium
- fura-2
- magnesium green
The hippocampal formation has been implicated in memory formation in many mammals including man (Squire, 1998). Substances that interact with receptors for the neurotransmitter acetylcholine have been shown to interfere with memory (Deutsch and Rocklin, 1967; Safer and Allen, 1971). Moreover, synaptic plasticity in the hippocampus, widely believed to underlie some forms of learning and memory, can be influenced by drugs that act at acetylcholine receptors (Huerta and Lisman, 1993; Hasselmo et al., 1995). Loss of cholinergic innervation may underlie some of the pathology in Alzheimer's disease (Bartus et al., 1982; Coyle et al., 1983).
Cholinergic effects are mediated by two broad classes of receptors for acetylcholine: muscarinic receptors, which are large transmembrane G-protein-coupled receptors (Hulme et al., 1990), and nicotinic receptors, which form nonselective cation channels (Albuquerque et al., 1995; Dani and Mayer, 1995; Boyd, 1997). Presynaptic inhibitory actions and postsynaptic excitatory actions have been documented for muscarinic receptors (Rouse et al., 1999). Activation of nicotinic receptors can excite cells by evoking inward currents (Albuquerque et al., 1995;Alkondon et al., 2000) and may also have modulatory presynaptic effects (McGehee et al., 1995; Gray et al., 1996).
Here we investigate the influence of cholinergic receptor activation on excitatory glutamatergic transmission in the CA3 area of the hippocampus. Cholinergic innervation of the hippocampus arrives mainly through the fornix/fimbria and originates in the medial septal nuclei and the vertical limb of the nucleus of the diagonal band of Broca (Gray and McNaughton, 1983). Muscarinic and nicotinic receptors are found on both principal cells and interneurons in the CA3 region (Levey et al., 1995; Albuquerque et al., 1997; McQuiston and Madison, 1999a).
We have focused our experiments on synaptic transmission via the mossy fiber (MF) system and the associational–commissural (A/C) fiber system. MF synapses provide glutamatergic input from dentate gyrus (DG) granule cells. MFs run in a well delineated tract within a distinct morphological layer in CA3, the stratum lucidum, and contact the proximal apical dendrites of CA3 pyramidal cells (Amaral and Dent, 1981). A/C synapses are a dense associative network of ipsilateral and contralateral excitatory connections between CA3 pyramidal cells that are found throughout the stratum radiatum of the CA3 (Amaral and Witter, 1989). Previous reports indicate that to some degree cholinergic modulation of excitatory synapses occurs throughout this region (Williams and Johnston, 1990; Hasselmo and Schnell, 1994), although the underlying mechanisms have not been investigated in detail. For example, it has been shown that the MF synapse is inhibited by muscarine (Williams and Johnston, 1990), and it has been reported that activation of calcium-permeable nicotinic receptors can modulate MF synapses by elevating presynaptic calcium levels (Gray et al., 1996). Thus an important part of understanding ACh control of hippocampal function is to elucidate more fully the effects of ACh on excitatory transmission in the CA3 region.
We used a combination of optical measurements of presynaptic calcium and electrophysiological measurements of synaptic currents to study A/C and MF synapses in brain slices from rats. Direct synaptic modulation mediated by ACh receptors is only evident at the A/C synapse, where synaptic inhibition reflects inhibition of presynaptic calcium channels mediated by muscarinic receptors. In addition, we found that MF synapses can be modulated by muscarine application by multiple indirect mechanisms. This study illustrates that in a highly interconnected region synaptic modulation cannot be understood solely by consideration of direct effects. No evidence of presynaptic nicotinic modulation of the MF synapse was apparent.
MATERIALS AND METHODS
Slices were cut from Sprague Dawley rats using standard procedures. Animals 15–21 d old were used, except in experiments on nicotinic receptor activation in which 18- to 28-d-old rats were used. Animals were decapitated under deep Halothane anesthesia, and their brains were rapidly removed and placed in ice-cold dissection solution equilibrated with 95% O2 and 5% CO2. The dissection solution contained (in mm): NaCl 87, sucrose 75, NaHCO3 26, glucose 25, KCl 2.5, NaH2PO4 1.25, MgCl2 7, and CaCl2 0.5. The hippocampi were dissected and placed onto agar blocks, and their longitudinal axis was straightened. Transverse slices 300 μm thick were cut on a vibrating microtome (Leica VT1000) in ice-cold dissection solution. They were transferred to a holding chamber at 32°C in the dissection solution. After 20 min the solution was changed to artificial CSF (ACSF) equilibrated with 95% O2 and 5% CO2. ACSF contained (in mm): NaCl 125, NaHCO326, glucose 25, KCl 2.5, NaH2PO4 1.25, MgCl2 2, and CaCl2 3. After a total of 1 hr at 32°C the slices were then held at room temperature. In cases in which the cell bodies of the dentate gyrus granule cells were separated from the MF synapses, slices were placed under a dissecting scope, and the MF tract was cut distal to the hilus of the dentate gyrus. Experiments were done at room temperature. 3-[(R)-2-Carboxy-piperazin-4-yl]-propyl-1-phosphonic acid [(R)-CPP], (2S,1′S,2′S)-2-(carboxy-cyclopropyl) (l-CCG-I), (+)-(2S)-5,5-dimethyl-2-morpholineatic acid (SCH50911), and 2,3-dioxo-6-nitro-1,2,3,4-tetrahydro-benzo(f)quinoxinaline-7-sulfonamide disodium (NBQX) were obtained from Tocris; bicuculline, baclofen, and muscarine were from Sigma (St. Louis, MO). Results are reported as the mean ± SEM.
Presynaptic labeling and Ca measurements. Presynaptic fibers were labeled with AM esters of either magnesium green or fura-2 (Molecular Probes, Eugene, OR) as described previously (Regehr and Tank, 1991; Regehr and Atluri, 1995). For MF labeling the injection pipette was placed onto the stratum lucidum, perpendicular to the orientation of the fibers. For A/C fiber labeling the injection pipette was placed midway between the stratum lucidum and stratum moleculare in CA3. In most cases the orientation in which the slices were cut preserved a several millimeter long MF tract that ran almost parallel to the surface of the slice. Individual MF boutons and axons were visible at high magnifications. After labeling in the stratum radiatum, a pattern of fine fibers interspersed with boutons ∼1 μm in diameter was observed 400–600 μm from the site of labeling. Slices were placed under an upright microscope (Olympus BX50WI) and continuously superfused with ACSF equilibrated with 95% O2 and 5% CO2. They were visualized using a 40× water-immersion lens. A small region of the labeled fibers was illuminated, and the fluorescence signal was measured with a photodiode (Regehr and Atluri, 1995). The signals from the photodiode were digitized (Instrutech) at 5 kHz and captured with a Macintosh personal computer using IGOR Pro (Wavemetrics) software (Herrington and Bookman, 1995). Action potential-induced signals were low-pass filtered at 500 Hz, and drug application-induced transients were low-pass filtered at 100 Hz with digital-filtering algorithms. Signals were transformed into ΔF/F measurements. With increasing calcium concentrations, magnesium green fluorescence increases, whereas fura-2 fluorescence decreases for 380 nm excitation and increases for 350 nm excitation.
Electrophysiological recordings. Whole-cell recordings of CA3 pyramidal cells were obtained under visual control. Glass pipettes (2–3 MΩ) were filled with an internal solution containing (in mm): CsCl 100, CsF 35, EGTA 10, and HEPES 10, pH 7.4. Access resistances ranged between 7 and 15 MΩand were continuously monitored. To measure NMDA currents, the cells were held at +30 mV.
Extracellular glass stimulus electrodes were filled with ACSF and placed into the stratum radiatum and stratum lucidum. Square pulses (1–5 μA) of 0.3 msec duration were used to evoke EPSCs of 100–500 pA. MF responses were identified because of their large amount of short-term facilitation. If identification was crucial, the presynaptic metabotropic glutamate receptor agonist l-CCG-I (10 μm) was used to identify positively the origin of MF EPSCs (Kamiya et al., 1996). For cell-attached recordings, patch pipettes with 3–5 MΩ resistance were filled with a solution containing (in mm): NaCl 165, HEPES 10, glucose 25, KCl 2.5, NaH2PO4 1.25, MgCl2 2, and CaCl23, pH 7.3. After establishing a cell-attached configuration, the viability of the cell was assessed using a brief focal KCl application. Only cells that exhibited clearly distinguishable action potentials were used for further studies. The whole-cell recordings in current-clamp mode were performed with an internal solution that contained (in mm): KMeSO4 130, EGTA 0.5, HEPES 10, Na2ATP 4, MgCl2 6, and NaGTP 0.3, pH 7.4.
Focal application to labeled MF tracts. Drugs were loaded into glass pipettes with a opening diameter of 2–4 μm attached to a pneumatic injection system (PV820; WPI). Pressure pulses of 1 sec duration and 1–5 psi pressure were used to eject drugs. The system was calibrated with a solution containing fast green before every experimental run, to prevent leaking of pipette solution or backfilling of pipettes with bath solution. Excitation filters with center frequencies of 350 and 380 nm were mounted on a filter wheel (Lambda 10; Sutter Instruments). Ten trials were performed for each excitation wavelength, and ΔF/F transients were calculated. Filters were switched two or three times for a total of 40–60 traces. Fura-2 fluorescence traces were not inverted for these experiments.
RESULTS
Fluorometric detection of calcium influx
Synaptic modulators in the vertebrate CNS often act by altering action potential-driven presynaptic calcium entry (Dittman and Regehr, 1996; Wu and Saggau, 1997). This makes it highly desirable to monitor presynaptic calcium influx in characterizing cholinergic modulation of synapses in the CA3 region. Such measurements are possible, provided presynaptic boutons can be selectively loaded with calcium-sensitive indicators. At many synapses, including the MF→CA3 pyramidal cell synapse (Regehr et al., 1994), this can be accomplished by locally applying a membrane-permeant form of a calcium indicator to a fiber tract (Fig.1A). The cell-permeant AM ester is hydrolyzed intracellularly into the membrane-impermeant fluorescent indicator dye. After allowing time for diffusion, a distinct tract of MFs is labeled with the indicator (Fig.1B). By illuminating a small area of the labeled fibers far from the site of application, action potential-driven calcium influx can be measured as the ΔF/F of the fluorescence transients evoked by extracellular stimulation of the MF bundle. This provides an aggregate measure of many activated MFs and contains signals from both synaptic boutons and axons (Regehr, 1999). These fluorescence transients are dominated by calcium transients in MF boutons onto CA3 pyramidal cells, which are large (3–6 μm in diameter) and have a volume much greater than that of the much thinner axons.
We tested two calcium-sensitive indicators: fura-2, which is a high-affinity dye with a dissociation constant of 200 nm(Grynkiewicz et al., 1985), and magnesium green (Zhao et al., 1996), which is a low-affinity indicator with a dissociation constant of 6 μm (Fig. 1C,D). The decay time course of the fluorescence transients after single stimuli was slower for fura-2 compared with that for magnesium green [t1/2, 175 ± 24 msec (N = 4) compared with 30 ± 3 msec (N = 6)] (Fig.1C1,D1 ). In addition, the second ΔF/F response to a pair of stimuli was decreased for fura-2 (0.61 ± 0.1; N = 4) but not for magnesium green (1.02 ± 0.02; N = 6) (Fig. 1C2,D2 ). This is consistent with previous studies at other presynaptic boutons (Regehr and Atluri, 1995; Feller et al., 1996; Regehr, 1999) and suggests that the calcium transients in MF boutons were sufficiently large that fura-2 could not faithfully report them but magnesium green could.
A/C fibers were labeled in a similar manner by local application of dye to the stratum radiatum (Fig.2A). The morphology of the A/C system (Amaral and Witter, 1989) is not as well suited to this labeling method as is that of the MF system. In contrast to the MF bundle, axons within the A/C system fibers are highly divergent. Consequently the density of labeled fibers decreases rapidly as a function of the distance from the site of dye application. Because unwanted nonspecific labeling is significant near the loading site, calcium must be monitored at some distance, resulting in dim fluorescence signals (Fig. 2B). To prevent a contribution from recurrent excitation, we blocked ionotropic glutamate receptors during recordings of presynaptic calcium transients from A/C fibers. As for MFs, calcium transients in the A/C system were sufficiently large that a low-affinity indicator such as magnesium green was needed to quantify changes in calcium influx. This is reflected in the decay time of fluorescence transients after single stimuli for fura-2 compared with that for magnesium green [t1/2, 204 ± 13 msec (N = 3) compared with 43 ± 2 msec (N = 4)] (Fig. 2C1, D1 ). Moreover, the second ΔF/Fresponse to a pair of stimuli was smaller than the first for fura-2 (0.51 ± 0.05; N = 3) but not for magnesium green (0.97 ± 0.02; N = 4) (Fig.2C2,D2 ).
To quantify changes in presynaptic calcium entry with optical methods, it is desirable that the indicator does not distort the calcium signal and that the signal arises from calcium entry through presynaptic voltage-gated calcium channels (Regehr and Atluri, 1995). For both MF and A/C fibers, magnesium green ΔF/F transients decayed fast and added linearly during repetitive stimuli in both fiber tracts, indicating that the calcium signals do not saturate the dye. The relative volume of these boutons compared with that of the axon and the greater density of calcium channels in boutons compared with that of the axons argue that most of this signal arises from presynaptic boutons. On the basis of the very rapid rise time of these calcium signals, most of this fluorescent signal arises from calcium influx in response to an action potential. It is therefore possible to quantify changes in action potential-driven calcium influx into both of these types of presynaptic terminals by measuring magnesium green fluorescence transients.
The relationship between calcium influx and release for MF and A/C synapses
As a first step in determining the mechanism of action of ACh in synaptic modulation, we determined the effect of modulating calcium entry on neurotransmitter release. The quantification of neurotransmitter release in the CA3 region is complicated by prominent recurrent excitation (Christian and Dudek, 1988) and inhibition (Miles and Wong, 1987). As a result, whole-cell measurements of AMPA receptor-mediated currents are contaminated by large recurrent inhibitory currents mediated by GABAA receptors. Removal of these inhibitory responses with GABAAreceptor antagonists is impractical because the slices become epileptic. We avoided such problems by including both AMPA (NBQX, 10 μm) and GABAA (bicuculline, 20 μm) receptor antagonists. Thus we were able to quantify synaptic strength, free from the complication of recurrent synaptic activity, by monitoring monosynaptic excitatory NMDA receptor (NMDAR)-mediated EPSCs from MF and A/C synapses.
Altering the external calcium concentration (Cae) is a convenient way of controlling calcium entry into presynaptic terminals to determine the relationship between calcium influx and release. To avoid changes in fiber excitability, the total amount of divalent cations was held constant by substituting Mg for Ca. We did not observe changes in the presynaptic fiber volley in the different calcium concentrations under these conditions (data not shown). Reducing Cae from 3 to 1 mm decreases presynaptic calcium entry to ∼50% into both MF and A/C synapses (Fig. 3A). Both A/C and MF synapses showed a sublinear relationship between extracellular calcium concentration and presynaptic calcium influx (Fig. 3B), consistent with saturation of influx through voltage-gated calcium channels (Mintz et al., 1995). The relationships between ΔF/F and Cae for both synapses are not significantly different.
Synaptic strength was more sensitive to changes in Cae than was calcium entry. Lowering Cae from 3 to 1 mm reduces NMDAR-mediated EPSCs to ∼10% for MF synapses and ∼20% for A/C synapses (Fig. 3C). For both A/C and MF synapses the relationship between calcium influx per action potential (Cai) and EPSC amplitude was well fit by the following equation: Equation 1where EPSC and Cai are normalized to values in 3 mm Cae, EPSCmax is the maximal EPSC amplitude,KD is the apparent affinity constant of the process, and n is the power law exponent (Fig.3D,E, solid lines). These fits approximate the relationship between calcium influx and release for these synapses when external calcium is systematically varied. The power law exponent is similar at both synapse types, 3.6 for the MF synapse and 3.9 for the A/C synapse. The A/C synapse shows clear signs of saturation of the release process, but the MF synapse does not.
The effects of presynaptic modulators on calcium entry and release
To examine direct presynaptic actions of muscarine, we cut the MF bundle to disrupt the connection between the dentate gyrus granule cells and the MF synapses and blocked GABABreceptors (except when we were testing the effects of activating GABAB receptors). We compared muscarinic effects with those of other neuromodulators: the metabotropic glutamate receptor II (mGluRII) agonist l-CCG-I (10 μm) and the GABAB receptor agonist baclofen (10 μm). The mGluRII agonists have become useful tools in distinguishing MF synapses, because they inhibit MF synapses and do not affect A/C synapses (Kamiya et al., 1996). Both MF and A/C fibers are thought to be inhibited by activation of GABABreceptors.
Examples illustrating the actions of different agonists on calcium influx and synaptic transmission are shown for the MF synapse (Fig.4A) and the A/C synapse (Fig. 4B). At the MF synapse, baclofen decreased presynaptic calcium entry to 76% of control and EPSC amplitude to 12% of control, l-CCG-I decreased calcium influx to 62% of control and synaptic release to 26% of control, and muscarine (10 μm) had no effect on either the presynaptic calcium signal or synaptic release. In the A/C system, baclofen also reduced calcium entry (64% of control) and EPSC amplitude (32% of control), muscarine (10 μm) decreased calcium influx to 74% of control and synaptic release to 52% of control, andl-CCG-I was without effect on A/C synapses.
The measurements of calcium influx and the relationships between calcium influx and EPSC amplitude of Figure 3 provided important insight into the mechanism of synaptic modulation. If the EPSC versus ΔF/F in the presence of the modulator falls on the line determined in Figure 3, it suggests that the neuromodulator affects release primarily by altering calcium influx. This appears to be the case for mGluRII inhibition of MF synapses and inhibition of A/C synapses by activation of either GABAB or muscarine receptors. For GABAB inhibition of MF synapses the point relating EPSC and ΔF/F is significantly below the line, which suggests that although the decrease in calcium influx contributes to synaptic inhibition, another mechanism, one that is independent of presynaptic calcium signaling, also participates.
Thus, these receptors have very different direct effects on these two types of synapses. Both MF and A/C synapses are inhibited by activation of GABAB receptors. In contrast there is a selective effect of mGluRII receptor agonists, which directly inhibit MF synapses but not A/C synapses. Muscarine is also selective but inhibits A/C synapses without directly affecting MF synapses.
Indirect effects of muscarine on MF synapses
Although muscarinic inhibition of MF synapses has been reported previously (Williams and Johnston, 1990), we observed no such inhibition for our experimental conditions. However, in our experiments we had severed connections from the dentate region and included GABAB receptor antagonists to allow us to study direct synaptic inhibition in isolation. Activation of muscarinic receptors can also affect neuronal excitability, which raises the possibility that activation of muscarinic receptors might have indirect effects on excitatory synapses in the CA3 region. Muscarine increases interneuronal firing (McQuiston and Madison, 1999b), and both A/C and MF synapses are sensitive to GABAB receptor activation (Fig. 4A,B). We therefore tested the possibility that muscarine could indirectly affect synapses in the CA3 region by causing interneurons to fire at high rates, thereby leading to a sufficient buildup of GABA to inhibit synapses.
We performed a number of experiments to examine the effects of muscarine on the firing of CA3 interneurons located in stratum lucidum and stratum radiatum. A whole-cell current-clamp recording of CA3 interneurons revealed that muscarine slowly depolarized the interneuron (McQuiston and Madison, 1999b) and led to sustained spiking (Fig.5A1 ). Cell-attached recordings were used subsequently to avoid disturbance of the ionic composition and excitability of the cell. Action potential frequencies were very low in the conditions used for MF NMDAR EPSC recordings (control) and were increased significantly by muscarine (Fig. 5A2 ). Muscarine increased the mean firing rate of CA3 interneurons from 0.03 ± 0.01 Hz (N = 4) for control conditions to 1.5 ± 0.9 Hz (N = 4) (Fig. 5A3 ).
We also measured the effect of muscarine on spontaneous IPSCs (sIPSCs) in CA3 pyramidal neurons. This provided for an average sample of interneuronal activity and thereby provided for a more realistic measure of the increase in GABA release than did measuring the activity of individual neurons. These experiments were performed in the presence of NBQX (10 μm) and (R)-CPP (5 μm) to block glutamatergic synaptic currents, and the GABAA receptor antagonist bicuculline (which was present in all other types of experiments) was omitted. The chloride reversal potential in our internal solution is close to 0 mV, and the cells were held at −60 mV; IPSCs are therefore detected as inward currents. Application of muscarine increased the sIPSC frequency in CA3 pyramidal cells (Fig. 5B1 ) from 0.3 ± 0.2 Hz (N = 5) in control to 2.0 ± 0.4 Hz (N = 5) (Fig.5B2 ).
To investigate whether this increase in inhibitory activity had an influence on excitatory transmission, we recorded NMDAR-mediated EPSCs onto CA3 pyramidal cells, but we initially did not include a GABAB antagonist in our external solution. The connection of the MF bundle to the dentate gyrus was severed in those experiments. Muscarine (10 μm) decreased both A/C and MF EPSC amplitudes (Fig. 5C1 ) for these conditions. However, the MF but not the A/C fiber depression was reversible after application of the GABABreceptor antagonist CGP 55845A (1 μm) or SCH50911 (20 μm). In control conditions MF synaptic release was on average reduced to 72% of control (±7%;N = 4) by muscarine. Application of the GABAB antagonist prevented this reduction (100 ± 8%; N = 10) (Fig.5C2 ). These results indicate that muscarine can indirectly inhibit MF synapses by activating inhibitory cells that release GABA and activate presynaptic GABAB receptors.
The properties of MF synapses make them potentially sensitive to another type of indirect modulation. The MF system is well known to exhibit prominent frequency facilitation even for frequencies as low as 0.1 Hz (Regehr et al., 1994; Salin et al., 1996). If muscarine alters the spontaneous activity of granule cells, synaptic strength would be indirectly affected. We performed a series of experiments to test this possibility. We found that muscarine increased the action potential frequency of granule cells (Misgeld et al., 1989) (Fig.6A1 ). The rate of spiking was highly variable. The average action potential frequency increased from 0.05 ± 0.05 Hz (N = 5) in control conditions to 1.6 ± 0.4 Hz (N = 5) (Fig. 6A2 ). We have therefore compared the effect of muscarine (10 μm) application on MF synaptic transmission in intact slices with preparations in which the cell bodies of the granule cells were separated from the MF bundle. GABAB receptors were blocked during these experiments. As shown in a representative case, muscarine increased the EPSC amplitude, and the extent of inhibition by l-CCG-I established that this was an MF input (Fig. 6B1 ). Muscarine increased the MF EPSC amplitude to 149 ± 11% (N= 4) of control in intact slices but had no effect on MF synaptic transmission (99 ± 7%; N = 10) (Fig.6B2 ) when the connection to the granule cell bodies was severed. Thus, muscarine can indirectly modulate MF synapses by increasing the frequency of granule cell firing, which in turn increases synaptic strength via frequency-dependent facilitation.
The effect of nicotinic receptor agonists on MF calcium levels
Another leading hypothesis for the way in which cholinergic signaling influences the CA3 region of the hippocampus is by activating presynaptic α7 nicotinic acetylcholine receptors (Gray et al., 1996). Activation of these calcium-permeable receptors (Seguela et al., 1993) is thought to increase residual calcium levels in MF presynaptic terminals, thereby influencing the probability of release. An important piece of evidence in support of this hypothesis is that application of nicotinic receptor agonists decreases the fluorescence of MF labeled with fura-2 for 380 nm excitation.
We examined this effect on presynaptic calcium signaling with an approach similar to that used previously (Gray et al., 1996). As in past studies, drugs were rapidly applied with pressure to minimize the extent of desensitization, which is prominent for α7 nicotinic receptors. MF tracts were labeled with fura-2, which has the advantage that its high affinity makes it sensitive to even small increases in presynaptic calcium. One difference in our study was that we took advantage of the ability to use fura-2 as a ratiometric indicator (Grynkiewicz et al., 1985) for which an increase in calcium decreases fluorescence for 380 nm excitation and increases fluorescence for 350 nm excitation. Excitation with both wavelengths is an important means of determining whether the fluorescence changes reflect alterations in calcium levels or movement. A fluorescence change arising from movement of fluorescent fibers would not show the wavelength dependence that is characteristic of a change in calcium.
The response to stimulation of the MF tract illustrates the dependence of the direction of the fluorescence signal on the excitation wavelength. Calcium transients evoked by extracellular stimulation gave rise to large decreasing fluorescence transients for 380 nm excitation (Fig. 7A). The smaller increasing signal observed for 350 nm excitation indicates that we are on the other side of the isosbestic point for excitation. Application of KCl (30 mm) by pressure ejection reliably evoked similar wavelength-dependent increases in fura-2 fluorescence (Fig. 7B), consistent with an increase in calcium (N = 5).
Application of nicotine (20 μm) in the same manner did not alter fura-2 fluorescence (Fig. 7C1 ). If the application pipette was brought close to the labeled tract (Fig.7C2 ), a decrease in fura-2 fluorescence at the 380 nm excitation wavelength could again be observed, which was accompanied by a similar decrease in fluorescence for 350 nm excitation. Injection of saline carrier alone had the same effect (data not shown).
We also tested the nicotinic receptor agonist choline, which has the advantage that receptor desensitization is much less pronounced compared with that with nicotine (Alkondon et al., 1997). The results were similar to those of nicotine. Choline (3 mm) did not alter fura-2 florescence (Fig. 7D1 ). Fluorescence changes were only observed if the application pipette was brought very close to the labeled tract (Fig.7D2 ). No changes in fura-2 fluorescence compatible with calcium increases were produced by either nicotine (N = 5) or choline (N = 5).
As a positive control, our ability to activate nicotinic receptors with our method of agonist application was assessed. We found that with the same method of choline application, we could reliably evoke inward currents in CA3 interneurons as assessed by whole-cell voltage-clamp recordings. For all cells tested (four of four) this inward current was blocked by the specific α7 nicotinic receptor antagonist methyllycaconitine (MLA) (10 nm) (Fig.7E1,E2 ). These results demonstrate that our method of agonist application can reliably activate α7 nicotinic receptors.
DISCUSSION
Comparison of MF and A/C synapses
Our findings extend the differences in the modulatory profile of MF and A/C synapses. It has long been known that mGluRII activation inhibits MF synapses without affecting A/C synapses (Manzoni et al., 1995; Kamiya et al., 1996), and this has been used to distinguish between these two inputs. We found muscarine to have complimentary actions in that it directly inhibits the A/C synapse without directly affecting the MF synapse. Thus, muscarine could be used to depress A/C fiber signaling selectively during MF investigation.
Direct effects on synaptic strength mediated by GABAB receptors, muscarine receptors, and mGluRII receptors were primarily a consequence of inhibition of presynaptic calcium entry. We found that mGluRII-mediated inhibition of the MF synapse could be explained entirely by inhibition of presynaptic calcium entry [but see Kamiya and Ozawa (1999)]. Presynaptic calcium influx was reduced in both synapses by GABABreceptor activation. In A/C synapses this inhibition of calcium influx was sufficient to account for all of the presynaptic inhibition, whereas at MF synapses additional mechanisms downstream of calcium influx may also be involved (Dittman and Regehr, 1996).
Muscarinic effects
Activation of muscarinic acetylcholine receptors affects A/C and MF synapses via multiple mechanisms. Direct presynaptic inhibition is only present at A/C terminals, whereas MF transmission was affected by two indirect mechanisms. Taken together these results establish that the actions of ACh on excitatory transmission in the CA3 region are complex, and they cannot be understood simply by considering direct effects on excitatory presynaptic terminals.
The depression of release from A/C synapses by muscarine can be fully explained by a reduction of presynaptic calcium entry. Muscarine reduces presynaptic calcium entry but does not alter the presynaptic action potential waveform (Hasselmo and Schnell, 1994). This suggests that muscarine inhibits A/C synapses by G-protein-mediated calcium channel inhibition, similar to the mechanism of inhibition by muscarine of the Schaffer collateral synapse in the CA1 region (Hasselmo and Schnell, 1994; Qian and Saggau, 1997).
One way that muscarine affects MF transmission is to excite inhibitory neurons, leading to the release of GABA, which in turn activates GABAB receptors on MF boutons and thereby indirectly inhibits synaptic transmission (Fig. 5). Several factors make the MF synapse particularly sensitive to this type of synaptic modulation. Muscarine induces massive increases both in the firing rate of interneurons in the hippocampus and in the release of GABA. There is even a type of inhibitory neuron with a morphology suited to elevating GABA levels near MFs (Vida and Frotscher, 2000). Moreover, MF synapses are extremely sensitive to GABAB receptor activation, and MF terminals are likely under a continuous dynamic GABAergic inhibition (Vogt and Nicoll, 1999).
Another way by which muscarine indirectly modulates MF synapses is to elevate the firing rate of DG cells, which in turn facilitates MF synapses. Again the MF synapse may be particularly sensitive to such modulation. The basal firing rate of DG cells is low in the slice as it is in vivo (Jung and McNaughton, 1993), and the increases in firing rate induced by muscarine are substantial. But it is the low threshold for frequency facilitation in the MF synapse that makes it particularly sensitive to changes in firing rates over the range produced by muscarine (Regehr et al., 1994; Salin et al., 1996).
Both of these indirect mechanisms of synaptic modulation are likely to be quite widespread (Frerking et al., 1999), although their contributions are certainly very synapse dependent. In the case of indirect GABAergic modulation, the lack of an indirect inhibition of the A/C synapse in our experimental conditions suggests that it is less susceptible to such modulation, despite the presence of GABAB receptors on A/C fibers (Fig. 4). Moreover, the manner in which alterations in the firing pattern of a presynaptic cell affect transmission depends on the use-dependent plasticity of the synapse. At facilitating synapses, such as the MF synapse, an elevation in firing enhances transmission. In contrast, for depressing synapses, increases in presynaptic firing rates depress synaptic transmission (Frerking et al., 1998).
The indirect effects of muscarine have important implications for the study of synapses in preparations with intact circuitry. Not surprisingly, because of indirect modulation, when DG connections are preserved and GABAB receptor antagonists are absent, the effects of muscarine on MF transmission are variable and difficult to interpret (data not shown). Enhancement by an indirect mechanism that requires intact connections between granule cell bodies and MF boutons is opposed by a depression that needs healthy inhibitory transmission. The extent to which either the DG connection or inhibitory neurons are preserved within a slice can be highly variable. Such complications are certainly not limited to the actions of muscarine at the MF synapse. More fundamentally, we see that to understand the effects of ACh on the CA3 region it is also necessary to consider how ACh changes the activity of cells capable of releasing modulatory chemical messengers and how ACh alters the activity in the presynaptic cell of interest.
Nicotinic effects
The possible presence of calcium-permeable nicotinic receptors on MF boutons is intriguing because elevations of presynaptic calcium greatly enhance MF synapses (Regehr et al., 1994). We were unable to detect such an increase in presynaptic calcium levels by locally applying nicotinic receptor agonists to MFs. By using two wavelengths to excite fura-2, we could distinguish between fluorescence changes caused by movement and those produced by elevations in presynaptic calcium. We found that the fluorescence changes produced by nicotine or choline reflected movement because of pressure ejection. Additional experiments suggested that our measurement approach is sufficiently sensitive to detect changes in presynaptic calcium levels that would be needed to affect neurotransmitter release. Finally, our method of agonist application could reliably evoke postsynaptic responses mediated by α7 nicotinic receptors in the slice. Thus, our findings do not support a prominent contribution of a mechanism of synaptic modulation in which activation of presynaptic nicotinic receptors enhances MF transmission by elevating presynaptic calcium levels.
Footnotes
This work was supported by National Institutes of Health PPG Grant P01-N538312. K.E.V. was supported by a grant from the Swiss National Science Foundation. We thank Bruce Peters for invaluable contributions in the initial stage of this project. We also thank Matthew Xu-Friedman, Chinfei Chen, Adam Carter, Anatol Kreitzer, and Kelly Foster for helpful comments on this manuscript.
Correspondence should be addressed to Dr. Wade Regehr, Department of Neurobiology, 220 Longwood Avenue, Boston, MA 02115. E-mail:wade_regehr{at}hms.harvard.edu.